Search tips
Search criteria 


Logo of neuroncolAboutAuthor GuidelinesEditorial BoardNeuro-Oncology
Neuro-oncol. 2004 October; 6(4): 307–387.
PMCID: PMC1872004

Abstracts from the Ninth Annual Meeting of the Society for Neuro-Oncology

November 18–21, 2004



Manish Aghi, Kenneth Cohen, David Scadden, and E. Antonio Chiocca; Departments of Neurosurgery and Hematology-Oncology, Massachusetts General Hospital, Charlestown Massachusetts, USA

Because bone marrow cells can be readily obtained, transduced ex vivo, and reimplanted, they represent a potential vector for glioma gene delivery. However, the roles of angiogenesis, the incorporation of existing vessels into a growing tumor, versus vasculogenesis, the recruitment of bone marrow–derived endothelial progenitors from the bloodstream in order to assemble new vessels, have yet to be characterized for gliomas. We transplanted bone marrow from beta-galactosidase (B-GAL) transgenic mice into irradiated wild-type mice. Eight weeks later, KR158/EGFR or GL261 glioma cells were implanted intracranially. Four weeks later, tumors were stained for von Willebrand factor (vWF), B-GAL, leukocyte antigen CD45, or microglial antigen CD11b. Enzyme-linked immunoassays (ELISAs) were performed on KR158/EGFR and GL261 media for vascular endothelial growth factor (VEGF) or stromal derived factor-1 (SDF-1), hypothesized vasculogenesis mediators. NF1 −/− p53 −/− transgenic mice with spontaneously arising subcutaneous tumors received B-GAL transgenic bone marrow, with tumors stained eight weeks later. We found that 26.2% of vWF+ cells in KR158/EGFR tumors expressed B-GAL and 28.7% of B-GAL+ cells expressed vWF, compared to 1.7% and 2.1% in normal brain (P < 0.0005). KR158/EGFR gliomas therefore specifically stimulate bone marrow differentiation into endothelium, suggesting minimal toxicity would occur if therapeutic genes were expressed by endothelial promoters. Also, 34.7% of B-GAL+ cells in KR158/EGFR gliomas expressed CD45, but none expressed CD11b, which implied that marrow-derived intratumoral leukocytes were lymphocytes. Unlike KR158/EGFR tumors, 0.5% of vWF+ cells in intracranial GL261 tumors expressed B-GAL. KR158/EGFR media contained 3.8 ng/mL VEGF and 0.73 ng/mL SDF-1, significantly greater than the 0.36 ng/mL VEGF and 0.03 ng/mL SDF-1 secreted by GL261, a result that is consistent with a potential role for these agents in recruiting bone marrow–derived endothelial progenitors into intracranial tumors. NF1 −/− p53 −/− transgenic mice with pre-existing spontaneously formed subcutaneous tumors received B-GAL transgenic bone marrow, after which 19.3% of vWF+ cells in tumor expressed B-GAL. We conclude that transplanted bone marrow contributes significantly to the vessels of some experimental intracranial and spontaneously formed subcutaneous tumors. VEGF and SDF-1 may enable gliomas to recruit bone marrow–derived endothelial progenitors. Besides providing insight into glioma biology, identifying factors promoting angiogenesis or vasculogenesis may enable the use of bone marrow cells as a gene delivery vector for certain gliomas. Transplanting bone marrow transduced ex vivo with cytotoxic genes expressed by endothelial promoters may impede glioma growth.


Marie E. Beckner, Fatima Burovic, and Ian F. Pollack; University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA

This study was based on the following hypothesis: Anaerobic adaptation during invasive migration provides intermittent vascular independence; in PTEN-mutated astrocytoma cells, energy for adhesion involved in anomalous hypoxic migration is proposed to rely on loss of PTEN’s negative control of lactic acid removal resulting from changes in PI3K/Akt/GSK3/glycogen synthase (GS) interactions; and intracellular lactic acid removal as glycogen within detached, gluconeogenic glioma cells should prevent normal inhibition of anaerobic glycolysis that can aberrantly energize cell functions when mitochondrial respiration is suppressed. We tested the adhesion of U87 and LN229 cells that migrate under hypoxic conditions (inhibition of mitochondrial respiration with sodium azide) for reliance on anaerobic glycolysis with 2-deoxyglucose (2DG) inhibition. Adhesion of hypoxic U87 cells was further tested with wortmannin (Wm), a specific PI3K inhibitor. The status of GS activation in hypoxia was shown by phosphorylation changes on immunoblots. Earlier studies demonstrated U87 cells’ hypoxic migration and susceptibility to restoration of wild-type PTEN and downstream PI3K/Akt pathway inhibitors. In these studies, U87 cells’ adhesion and LN229 cells’ migration and adhesion were maintained in hypoxic conditions (inhibition of mitochondrial respiration) at levels 65% or greater than normal cells. Both cell types maintained adhesion with 2DG competition of an equal concentration of added glucose, 1 mg/ml, in normoxic conditions. However, with 2DG competition their hypoxic adhesion was suppressed to 1% ± 1.3% SD and 23% ± 16% SD of levels for normoxic control cells, U87 and LN229, respectively. Hypoxic U87 cell adhesion also showed inhibition by 30 nM Wm to levels that were 33% ± 8% SD of normoxic control cells. Phosphorylated GS, inactivated by GSK3, fell to negligible levels when mitochondria were inhibited, while total GS levels increased slightly compared to total protein and actin. We conclude that anaerobic adaptation permits anomalous energy production within vascular-independent glioma cells. In PTEN-mutated glioma cells, loss of mitochondrial respiration activated GS, a downstream effect of PI3K/Akt pathway signaling. Suppression of PI3K inhibited both adhesion and migration of these cells in hypoxia. Therefore, therapeutically targeting aberrant energy production within PTEN-mutated astrocytomas should inhibit adhesion involved in migration during tumor cell invasion. Supported by The Nick Eric Wichman Foundation and The Pittsburgh Foundation’s Walter L. Copeland Fund for Cranial Research.


Lisa Bernas,1,2 Paula Foster,1,2 Joseph Megyesi,3 Brian Rutt1,2;1Department of Medical Biophysics, University of Western Ontario;2Imaging Research Laboratories, John P. Robarts Research Institute;3Department of Neurosurgery, London Health Sciences Centre, London, Ontario, Canada

Gliomas are the most common primary tumours of the central nervous system and often present a very poor prognosis because of the aggressive invasion of glioma cells into the surrounding brain tissue. Invading cells make complete surgical resection of the disease impossible and are responsible for recurrence and progressive neurological dysfunction. A shortcoming of current in vivo studies of glioma invasion, which involve intracranial implantation of pre-labeled cells into rats, is the lack of ability to track the cells over time in a single animal. Cellular imaging, using MRI in particular, has been used to noninvasively visualize small groups of cells in vivo and is therefore being investigated as a potential tool for tracking invading glioma cells. Cellular MRI has been accomplished through the use of intracellular superparamagnetic contrast agents, which allow cells to be distinguished from background on MR images. In the current study, the contrast agent consisted of micron-sized polystyrene beads containing superparamagnetic iron oxide that are readily taken up by many cell types. Rat C6 glioma cells were incubated with different concentrations of beads, and the consequent effects on in vitro cell behaviour were assessed. No significant effects were observed on cell viability or proliferation, and experiments are ongoing to assess effects on plating efficiency and cell motility. The absence of detrimental in vitro effects will ensure the potential for clinically relevant in vivo studies using this model. Intracellular iron content was quantified using a magnetic susceptometry technique and was found to increase linearly with iron incubation concentration until reaching a plateau. Preliminary in vitro MRI studies have been conducted on a 1.5 Tesla clinical scanner with the addition of a custom-built gradient insert coil, enabling the acquisition of high-resolution images while maintaining an excellent signal-to-noise ratio. Labeled C6 cells were visible as areas of signal void on MR images, confirming that the bead uptake by the cells is sufficient for their detection. Results thus far suggest that our current approaches to cellular MRI have the potential to visualize glioma cells in vivo. Tracking invading glioma cells would provide a dynamic assessment of the invasion process and treatment response.


Kevin R. Bobbitt, Tyler T. Voss, Cathie G. Miller; Hermelin Brain Tumor Center, Department of Neurosurgery, Henry Ford Health System, Detroit, Michigan, USA

A significant problem with current available glioblastoma (GBM) therapies is the inability to target tumor cells that infiltrate normal brain tissue. Understanding the mechanism involved in GBM infiltration will lead to the development of novel, targeted, more effective therapies. One potential area for investigation is signaling through the chemokine stromal cell–derived factor 1 (SDF-1, now termed CXCL-12) and its sole ligand CXCR-4. Overexpression of the G-protein coupled receptor CXCR-4 is seen in many tumor types, including GBMs, where signaling upon binding of CXCL-12 to CXCR-4 has been demonstrated to contribute to cell survival. Migration through the chemotactic signaling of CXCL-12 has also been demonstrated to be important for the metastatic potential of many other tumors. Here we demonstrate that stable overexpression of CXCR-4 by human U87 GBM cells (U87-CD4-CXCR-4) results in a more infiltrative primary tumor growth and secondary tumor growth upon implantation into nude rats as compared to parental control implanted tumors (U87 and U87-CD4). Further characterization of these cell lines demonstrates that in addition to overexpression of CXCR-4, U87-CD4-CXCR-4 cells have decreased expression levels of CXCL-12. Thus, stable overexpression of CXCR-4 results in decreased CXCL-12 expression and increased infiltrative tumor growth. Further exploration into the role of CXCR-4 and CXCL-12 in GBM migration may lead to the development of therapeutic blocking agents that will prevent GBM migration or allow for the specific targeting of infiltrative GBM cells.


Christiana Charalambous,1 Florence M. Hofman,2 Thomas C. Chen2,3; Departments of 1Molecular Microbiology and Immunology, 2Pathology, and 3Neurosurgery, and USC Center for Brain Tumor Research, University of Southern California, Los Angeles, California, USA

There has been increasing evidence that tumor endothelial cells are not just normal endothelial cells in a tumor microenvironment. Instead, these cells may become phenotypically and genotypically transformed compared to their normal counterparts. Glioblastoma multiforme (GBMs) are hyper-vascular tumors characterized by endothelial cell proliferation, with a breakdown in the blood-brain barrier (BBB). The goal of this investigation is to characterize the functional and phenotypic differences between normal brain endothelial cells (BECs) and glioblastoma-associated endothelial cells (TuBECs). Both human endothelial cells were isolated from specimens obtained at surgery, BECs from trauma lobectomies and TuBECs from GBM specimens. Both primary cell cultures were purified by using flow cytometry and determined to be factor VIII positive by immunohistochemistry. Morphology of TuBECs was clearly different from that of BECs. The proliferation rate of TuBECs was lower than that of BECs, and growth factors such as interleukin-8 (IL-8) and endothelin-1 (Et-1) did not increase its proliferation rate as much as they increased BEC proliferation. However, TuBECs had a significantly higher baseline migration rate compared to BEC. Addition of growth factors (IL-8, Et-1, VEGF) led to an upregulation of migration rate for BEC but not for TuBEC. In order to determine the etiology of this differential migration response, we demonstrated that TuBECs constitutively produced IL-8, Et-1, and VEGF, compared to BECs, which had minimal production of these growth factors. TuBECs also expressed higher levels of the matrix metalloproteinases MMP-2 and MMP-9 compared to BECs. Our data demonstrates that TuBECs are phenotypically and functionally different from BECs. Further experiments will determine whether TuBECs are genotypically different from BECs.


T. Demuth,1 J. Rennert,1 D.B. Hoelzinger,1 A.E. Chiocca,2 T.S. Deisboeck,3 D. Weiz,4 and M.E. Berens1; 1Translation Genomics Research Institute, Phoenix, Arizona, 2Department of Neurosurgery, Ohio State University Medical Center, James Cancer Hospital and Solove Research Institute, Columbus, Ohio; 3Massachusetts General Hospital, Harvard Medical School, Charlestown, Massachusetts; 4Division of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts; USA

The early and pervasive tendency of glioma cells to invade into peritumoral normal brain underlies the poor prognosis for patients with glial tumors. This malignant diffusion of glioma prevents complete surgical resection, positions tumor cells behind an intact blood-brain barrier, and sequesters tumor cells outside the fields of focal radiation; each of these leads to heightened (almost certain) tumor recurrence. The aim of this study is to exploit a model of the biological events inherent to glioma cell egress from a multicellular tumor mass in the context of a controllable physical and biochemical space into which the glioma cells invade. In this 3D collagen I gel matrix, we determined the transcriptome of invasive glioma cells compared to their noninvading counterparts from the spheroid center. Gene products underlying the invasive behavior may serve as specific and potent targets for novel therapeutics or may reveal points of vulnerability in the invasive cells. Human glioma cell line U87 was studied in its constitutive state (U87WT) and as a subline stably transfected with the truncated form of EGFR (U87dEGFR). Multicellular spheroids were implanted into collagen I gels and then followed for 8 days. Gels were cryosectioned and mounted onto slides for laser capture microdissection. Cells from invasive rim and spheroid core were collected from three biological replicates representing each cell line. RNA was extracted, amplified, and labeled for oligonucleotide microarray analysis. Data was analyzed to identify differentially expressed genes that distinguish invasive rim and tumor core by using comparative statistics. Additionally, since U87dEGFR cells form secondary colonies after episodes of invasion (in contrast to U87WT), these cells serve as reporter systems to understand the gene expression changes that characterize glioma cells switching from an invasive to proliferative commitment. Findings of differential gene expression will be reported, comparing the in vitro 3D invasion model with 2D monolayer invasion and with in vivo invasive glioma transcriptomes from human biopsy specimen.


Per Øyvind Enger, Per Øystein Sakariassen, Kai Ove Skaftnesmo, Jian Wang, Eirik Sundlisæter, Lars Prestegarden, Anjan Misra, Berit Bølge Tysnes, Martha Chekenya, Burt G. Feuerstein, A. Jorge A. Terzis, Rolf Bjerkvig; NorLux Neuro Oncology, Department of Biomedicine, University of Bergen, Norway; NorLux Neuro Oncology, Centre Recherche de Public Santé, Luxembourg; Department of Neurological Surgery and Laboratory Medicine, University of California San Francisco, San Francisco, California, USA

Highly infiltrative brain tumor phenotypes were established by the xeno-transplantation of human primary glioblastomas (GBMs) in the brains of immunodeficient nude rats. These tumors co-opted the host vasculature and presented as an aggressive disease without signs of angiogenesis. The tumors expressed neural stem cell markers and showed a migratory behavior similar to that of normal human neural stem cells. Furthermore, the tumor cells exhibited self-renewal capacity and gave rise to tumors in vivo. Serial animal passages gradually transformed the stem cell tumors into angiogenesis-dependent phenotypes. These phenotypes were characterized by a reduction in stem cell markers. Pro-invasive genes were upregulated, and angiogenesis signaling genes were downregulated in the stem cell tumors. In contrast, pro-invasive genes were downregulated in the angiogenesis-dependent tumors, derived from the stem cell tumors. The described angiogenesis-independent tumor growth and the uncoupling of invasion and angiogenesis, represented by the cancer stem cells and the cells derived from them, respectively, points at two completely independent mechanisms that drive tumor progression. The presented results also indicate that there are cancer cells, even in primary GBMs, that do not fall into the definition of a GBM phenotype (i.e., necrosis and angiogenesis). These put a question mark on the de novo development of primary GBMs. The present work also underscores the need for developing therapies that specifically target the cancer stem cell pools in tumors.


Christopher S. Gondi,1 Sajani S. Lakka,1 Dzung H. Dinh,2 William C. Olivero,3 Meena Gujrati,4 and Jasti S. Rao; 1Cancer Biology Program and Departments of 2Neurosurgery, 3Pediatrics, and 4Neurology, University of Illinois, Peoria, Illinois, USA

The diffuse and extensive infiltration of malignant glioma into the surrounding normal brain is believed to rely on modification of the proteolysis of extracellular matrix components. Our previous results clearly demonstrated that uPA, uPAR, and MMP-9 molecules are significantly increased during tumor progression and that this tumor growth can be inhibited with antisense stable clones of these molecules. Since antisense-mediated gene silencing does not completely inhibit the translation of target mRNA and high concentrations of antisense molecules are required to achieve gene silencing, we used the RNAi approach to silence uPA, uPAR, and MMP-9 in this study. We examined a cytomegalovirus promoter-driven DNA template approach to induce hairpin RNA (hpRNA)-triggered RNAi to inhibit uPA, uPAR, and MMP-9 gene expression with a single construct. uPAR protein levels and enzymatic activity of uPA and MMP-9 were found to significantly decrease in cells transfected with a plasmid expressing hairpin siRNA for uPAR, uPA, and MMP-9. After transfection with a plasmid vector expressing dsRNA for uPA, uPAR, and MMP-9, glioma cell invasion and migration was retarded in in vitro models. Direct intratumoral injections of plasmid DNA expressing hpRNA for uPA, uPAR, and MMP-9 significantly regressed pre-established intracranial tumors in nude mice compared to controls and EV/SV treated groups. Downregulation of uPA, uPAR, and MMP-9 using RNAi-inhibited angiogenesis occurred in the co-culture system when compared to controls. Our results support the therapeutic potential of RNAi as a method for gene therapy in controlling glioma migration, invasion, and regression of pre-established tumor growth.


Dominique B. Hoelzinger,1 L. Mariani,2 M. Nakada,1 T.D. Demuth,1 C.E. Beaudry,1 S.W. Coons,3 and M.E. Berens1; 1TGen, The Translational Genomics Research Institute, Phoenix, Arizona, USA; 2Department of Neurosurgery, University Hospital, Inselspital, Bern, Switzerland; 3Department of Pathology, St. Joseph’s Hospital, Phoenix, Arizona, USA

One of the salient features of malignant astrocytomas is a propensity for diffusely invading the brain parenchyma. The molecular mechanisms driving this invasive behavior may be unique to brain tissue, or possibly specific to glial-lineage tumors. A gene expression profile of GBM invasion was created by comparing RNA from noninvading versus invasive GBM cells, isolated by laser capture microdissection of densely populated tumor core and from individually invading GMB cells, respectively. Analysis for differential overexpression revealed multiple candidate genes related to invasion and motility in the invading GBM cells. Autotaxin (ATX, also known as ectonucleotide pyrophosphastase/phosphodiesterase 2, ENPP2) was first described in invasive melanoma. It has since also been reported in neuroblastoma, metastatic breast carcinoma, and several other carcinomas. Recently, its role as a lysophospholipase D was identified, converting lysophosphatidyl choline into the bioactive lipid lysophosphatidic acid (LPA). LPA stimulates motility in various cell types, including glioma cells, which express high levels of LPA1 receptor. Measurement of ATX expression in a glioma-specific tissue microarray revealed high expression in the invading cells of all astrocytic tumor grades. This led us to investigate the role of ATX in astrocytoma invasion. Stable transfectants carrying wild-type ATX or a form of ATX whose catalytic site is rendered dominant negative (DN-ATX) were created by using the U251 glioma cell line. ATX-U251 migrated significantly faster than parental U251 in a two-dimensional radial migration assay. ATX-U251 also proved more invasive than U251 and DN-ATX in an ex vivo rat brain slice assay. It is anticipated that there will be interactions between ATX-mediated motility and other autocrine motility factors expressed in invading glioma cells. This work was supported by NIH NS042262.


Bo Hu,1,2 Michael J. Jarzynka,1,3 Ping Guo,1,3 and Shi-Yuan Cheng1,3; 1Cancer Institute and Departments of 2Medicine and 3Pathology, Hillman Cancer Center, University of Pittsburgh, Pittsburgh, Pennsylvania, USA

A hallmark of malignant human gliomas is their ability to diffusely invade into surrounding brain tissues. The rapid dissemination of single tumor cells throughout the brain renders these tumors incurable by surgical removal, even when combined with adjuvant radiation, chemotherapies, and/or immunotherapies, underlying their great propensity for recurrence. We have previously reported that angiopoietin-2 (Ang2), a known angiogenic factor, induces human glioma cell invasion through the activation of matrix metalloprotease-2 (MMP-2). Ang2 expression is strongly correlated with glioma invasiveness in primary human glioma specimens. Acquisition of invasiveness was evident in intracranial xenografts of glioma cells engineered to express Ang2. MMP-2 was upregulated in the Ang2-expressing invasive regions of human glioma specimens and mouse glioma xenografts, as well as in various glioma cell lines treated with Ang2. We have further investigated the mechanism by which Ang2 stimulates glioma cell invasion through the induction of MMP-2 expression. Ang2 directly binds to beta 1 integrin in Tie2-deficient U87MG cells inducing the activation of FAK, p130Cas, ERK1/2, and JNK. The beta 1 integrin/FAK-mediated intracellular signaling events substantially enhanced MMP-2 expression and secretion. Ang2-stimulated MMP-2 expression and secretion was attenuated by a functional neutralizing anti–beta 1 antibody, by an ERK1/2 inhibitor (U0126), and by a JNK inhibitor (SP600125). Stable expression of a specific negative regulator of FAK, FAK-related non-kinase (FRNK), but not its mutant FRNKS-1034, inhibited Ang2-stimulated tyrosine phosphorylation of FAK and p130Cas, blocked the activation of ERK and JNK kinases, and decreased Ang2-stimulated MMP-2 expression and secretion in U87MG cells. Inhibition of beta 1 integrin, FAK, p130Cas, ERK1/2, and JNK also attenuated Ang2-stimulated glioma cell invasion. Furthermore, expression of FRNK, but not FRNK-S1034, by intracranial glioma xenografts in the brain derived from U87MG Ang2-expressing cells suppressed Ang2-induced glioma cell infiltration and MMP-2 expression. Together, these data provide evidence of a functional link between Ang2 interaction with beta 1 integrin and cell invasion, demonstrating a mechanism whereby binding of Ang2 to glioma cells regulates subsequent MMP-2 expression and secretion through the integrin and FAK signaling pathways.


Jensen Randy; Huntsman Cancer Institute, University of Utah, Salt Lake City, Utah, USA

At present there are no proven adjuvant therapies for surgical or radiation therapy refractory meningiomas. Hydroxyurea and RU486 chemotherapies are the most promising, but clinical results have been disappointing for the most part. We have previously demonstrated that calcium channel antagonists can block meningioma growth in vitro and in vivo. There is evidence with other tumor types that calcium channel antagonists can potentiate the effectiveness of chemotherapeutic drugs. To test this, we examined primary meningioma cell cultures for growth studies, fluorescence-activated cell sorting for cell cycle determination, and apoptotic index after exposure to combinations of chemotherapy and various calcium channel antagonists. Similar meningioma cells were placed on the flanks of nude mice. These animals were treated with combinations of RU486 or hydroxyurea and calcium channel antagonists. Tumor growth rate, proliferative and apoptotic index, and tumor size and weight were assayed. Our results showed that in vitro meningioma growth inhibition by chemotherapeutic drugs was increased with combined calcium channel antagonist administration. This effect was correlated with cell cycle arrest and increased apoptotic rate. The xenograft animal studies showed a similar growth inhibition. However, the effect was more modest than effects found in the in vivo studies. This was partially due to limitations of the xenograft tumor model. These studies could result in potential clinical applications of treating recurrent or unresectable meningiomas. Further studies are warranted.


Balveen Kaur, Narra S. Devi, Daniel J. Brat, Erwin G. Van Meir; Laboratory of Molecular Neuro-Oncology, Departments of Neurosurgery, Hematology/Oncology, and Pathology and Winship Cancer Institute, Emory University, Atlanta, Georgia, USA

As a tumor progresses from a low-grade infiltrating astrocytoma (WHO grade II) to a glioblastoma multiforme (WHO Grade IV), a concurrent increase in its vascular density is observed, underscoring the importance of angiogenic dysregulation in brain tumors. A better understanding of which molecules prevent excessive vascular development in the brain and how a developing brain tumor overcomes these signals might lead to the development of novel therapeutics for cancer and other diseases related to aberrant angiogenesis. Brain angiogenesis inhibitor 1 (BAI1) is a transmembrane receptor that is expressed specifically in the human brain, but is absent in a majority of glioma cell lines and brain tumor samples (Kaur et al., 2003, Am. J. Pathol. 162:19). The extracellular domain of BAI1 contains an RGD integrin recognition motif and five antiangiogenic thrombospondin type 1 repeats (TSRs). We demonstrate that a G protein–coupled receptor proteolytic (GPS) cleavage site is present in BAI1 and results in the release of a secreted extra-cellular fragment containing five TSRs that we called vasculostatin. We further show that constitutive and tetracycline-inducible expression of vasculostatin in LN229 glioma cells is able to significantly suppress their in vivo tumorigenicity and vascularization in a mouse xenograft model in a dose-responsive fashion without affecting their in vitro growth. A similar effect was observed in U87MG glioma cells, either parental or stably transfected with a constitutively active EGFR (deltaEGFR). The delta EGFR form (deleted for exons 2–7) is a common alteration in GBM that confers strong protumorigenic and angiogenic effects in tumors, in part through upregulation of VEGF. Overall, our results show that BAI1 is proteolytically processed, leading to the release of vasculostatin, a cleaved, secreted fragment that can function as a potent inhibitor of angiogenesis and tumor growth in several human glioma models. It will be important to further study the regulation of the cleavage of this angiostatic fragment in the brain and its mechanism of action on brain endothelium with the hope of deriving novel anti-tumor strategies.


J.Y. Ljubimova,1,2 M. Fujita,2 N.M. Khazenzon, B.S. Lee,3 E. Holler3 and K.L. Black1,2; 1Maxine Dunitz Neurosurgical Institute, Cedars-Sinai Medical Center, Los Angeles, California, USA; 2Arrogene, Inc., Tarzana, California, USA; 3Institut für Biophysik und Physikalische Biochemie der Universität Regensburg, Regensburg, Germany

Laminin-8 chains overexpression is associated with glioma progression, and laminin-8 blocking inhibits glioma invasion in vitro. The effects of laminin-8 blocking on xenotransplanted human gliomas were examined here. A multifunctional drug delivery construct consists of modules attached to the pending carboxyl groups of polymalic acid (PMLA). The polymer is a natural product of Physarum polycephalum. The modules are (1) morpholino antisense oligonucleotides conjugated to the scaffold by disulfide bonds, which are cleaved in the cytoplasm to release the free drug, (2) antibody to transferrin receptor for cancer cell targeting and receptor-mediated endocytosis, (3) short-chain PEG-conjugated L-leucine and directly coupled L-valine, both by amide bonds, to provide pH-dependent lipophilicity to disrupt endosomal membranes, (4) long-chain PEG for protection, (5) fluorescent reporter (fluorescein or Cy5) to detect the construct in tissue/cell. Drug 1 had anti-sense oligonucleotides to laminin-8 -4 and -1 chains, and Drug 2 additionally carried a monoclonal anti-transferrin receptor antibody (antibody OX-26) conjugated to PMLA. Control drugs carried sense oligonucleotides to laminin chains. Human U-87MG glioblastoma cell line was injected intracranially into NIHRNU-M NIH nude rats. The drug was not toxic in three different concentrations in vivo. Drug 1 and 2 concentrations of 0.5 and 2.5 mg/kg were equal for the treatment in the survival study. After intracranial administration of four doses of Drug 2, the survival time of glioma-bearing animals was increased by 30%, P < 0.0074, compared to control groups. Drug 1 without transferrin receptor antibody did not affect survival. Therefore, the mechanism of drug cell delivery is transferrin receptor–mediated endocytosis. Intracarotid drug treatment was important to understand the ability of active Drug 2 to penetrate both the blood-brain barrier (BBB) and the blood-tumor barrier (BTB). Thirty minutes after infusion, the drug with rhodamine-labeled transferrin receptor antibody was visible in tumor vessels and in the cytoplasm of tumor parenchymal cells. The combination of anti-sense oligonucleotides to laminin-8 chains with a novel drug delivery vehicle, PMLA, was an efficient method to inhibit laminin-8 expression in a xenografted intracranial human glioma in rats and increase animal survival. The ability of the drug vehicle to penetrate BBB and BTB is important for potential intravenous patient treatment. The data hold promise for efficient brain tumor inhibition using laminin-8 as a therapeutic target.


Johanna Milan,1 Florence Hofman,2 Christiana Charalambous,2 Raphael Zidovetzki,1 and Thomas C. Chen3; 1Department of Cell Biology and Neuroscience, University of California, Riverside, California; Departments of 2Pathology and 3Neurosurgery, University of Southern California, Los Angeles, California; USA

In the present work we demonstrated that a vasoconstriction peptide, endothelin 1 (Et-1), induces migration of primary cultures of human brain–derived endothelial cells (BECs). A partial dissection of the signal transduction pathways involved in this process showed the requirement of multiple pathways. Both cell surface Et-1 receptors, ETA and ETB, are necessary to initiate Et-1-induced migration. Downstream of the receptors the following signal transduction enzymes are required: protein tyrosine kinase of the Src family, small GTP-binding proteins of the Ras family, phosphatidylinositol-3 kinase, protein kinase C (PK-C), and three mitogen-activated kinases (MAPK): ERK, p38, and JNK. Moreover, cAMP-dependent protein kinase is also partially required for the migration. We have further shown that ERK and p38 are positioned downstream of PK-C and PK-A in the signal transduction pathways. The participation of the MAPKs was further confirmed by transfecting BECs with dominant negative mutants of ERK, p38, and JNK, which completely blocked Et-1-induced migration. An important general finding of the present study is that each of multiple signal transduction pathways is required for BEC migration, which is an essential component of angiogenesis. Whereas it makes the system very complex for detailed investigation, it can also provide an opportunity for a therapeutic intervention at various regulation sites.


Dale Milfay,1,3 Scott VandenBerg,1,2,3 Christoph Wilhelm Turck,4 and Gabriele Bergers1,3; 1Departments of Neurosurgery and 2Pathology and 3Brain Tumor Research Center, University of California Comprehensive Cancer Center, University of California San Francisco, San Francisco, California, USA; 4Max Planck Institute of Psychiatry, Molecular, Cellular, Clinical Proteomics, Munich, Germany

Glioblastomas (GBMs) are invasive tumors infiltrating white matter and migrating along basement membranes. GBMs are also very angiogenic tumors. Interestingly, GBMs do not only initiate new blood vessel growth (angiogenesis) to retrieve sufficient oxygen and nutrients but they also use blood vessels as a “freeway system” to disperse significant distances into the brain parenchyma. We recently noted that genetically engineered GBMs deficient in the hypoxia-inducible factor HIF-1a became much more invasive than the wild-type control tumors, penetrating both hemispheres of the brain specifically by moving along the normal brain vasculature. In order to reveal the decisive factors that are implicated in perivascular migration, we will isolate noninvading and invading tumor cells (around normal blood vessels) from tumor sections by laser capture microdissection and identify differentially expressed and posttranslationally modified proteins using a proteomics approach (2D gel electrophoresis and mass spectrometry).


Melike Mut,1 Morry D. Brown,2 Joan E. Carpenter,3 Gerard T. Redpath,3 Scott R. VanderBerg,4 Maria-Beatriz S. Lopes,3 Isa M. Hussaini,3 and Mark E. Shaffrey1; Departments of 1Neurosurgery, 2Radiation Oncology, and 3Pathology, University of Virginia, Charlottesville, Virginia, USA; 4Department of Pathology, University of California San Francisco, San Francisco California; USA

Cellular properties that constitute invasive phenotype of gliomas are not well understood. Low-density lipoprotein receptor–related protein (LRP) mediates the endocytosis of a wide variety of ligands implicated in cellular growth, adhesion, and proteinase catabolism. Among LRP ligands, urokinase (uPA) and urokinase-type plasminogen activator receptors (uPARs) are upregulated in association with increasing anaplasia of astrocytomas. LRP appears to mediate the internalization of free uPA, uPA associated with proteoglycans, and the uPAR-uPA/PAI-1 (plasminogen activator inhibitor-1) complex. LRP expression is differentially regulated in neoplastic and non-neoplastic astrocytes. Epidermal growth factor (EGF) treatment in astrocytomas significantly reduces total and cell surface LRP. Given that uPA, uPA/PAI-1, or tPA/PAI-1 complexes are key ligands for LRP, EGF-mediated functional downregulation of LRP endocytotic activity in astrocytic tumor cells may affect levels of proteolytic activity at the cell surface and in the adjacent microenvironment of the extracellular matrix. Functional effects of LRP elimination and LRP overexpression in U-1242 glioblastoma multiforme (GBM) cells in vitro and in vivo were assessed. L -42 clone: LRP-deficient GBM cells were prepared from U-1242 human tumor cells, which were transfected with the antisense constructs. LRP in L -42 was completely eliminated as determined by Western blot analysis. L -32 clone: LRP-overexpressing U-1242 cells were transfected with the LRP gene, and 3-4 times more expression of LRP was confirmed by Western blot analysis. Nontransfected U-1242 cells (ntU-1242) were also used for the experiment. Western blot analysis, zymography, and transwell invasion assay were used for in vitro study. In an animal model, Gelfoam (cellulose matrix) three-dimension cultures of L -42, ntU-1242 and L -32 clones were implanted in the brains of mice, and invasive growth of the tumors was assessed at 5 and 9 weeks post-implantation. Five mice were included in each experimental group. We found that the L -42 clone invaded fibrin-coated transwell membrane more than non-transfected U-1242, and uPA neutralizing antibody reduced the invasiveness of the L -42 clone to the levels of control cells. In vivo, there was a significant difference in size of tumors between L -42, ntU-1242, and L -32 cells at 5 weeks postimplantation. L -42 clone presented a faster growth pattern nt, and U-1242 cells showed a medium growth without apparent invasion. In L -32 clones, only Gelfoam and scar formation was detected without apparent tumor growth. L -42 cells showed a substantial invasive pattern, comparably bigger, with significant mass effect and midline shift at 9 weeks postimplantation. LRP-expressing clones were also found to be invasive. However, it was noted that L -42 clones were diffusely infiltrating, while ntU-1242 and L -32 showed focal invasion with pushing borders and more prominent gliosis around tumors. These results suggest that the level of LRP expression may be critical in determining the invasive growth phenotype of glioblastomas.


Mitsutoshi Nakada,1,2 Jared A. Niska,1 Nhan L. Tran,1 Wendy S. McDonough,1 Anna M. Joy,1 and Michael E. Berens1; 1The Translational Genomics Research Institute, Phoenix, Arizona; 2Neuro-Oncology Research, Barrow Neurological Institute, Phoenix, Arizona; USA

Eph receptor tyrosine kinases and their ligands, ephrins, mediate neurodevelopmental processes such as boundary formation, axon guidance, vasculogenesis, and cell migration. Recently, we reported that EphB2 was involved in migration and invasion of glioma cells in vitro and in vivo (Nakada et al., 2004, Cancer Res., 64:3179). Here we analyze the signaling pathways that mediate EphB2 effects on glioma cell adhesion, proliferation, and migration. U251 cells overexpressing EphB2 (U251-EphB2) showed reduced cell adhesion, aggregation and proliferation, but enhanced migration. An interaction of R-Ras with EphB2 is indicated by the co-immunoprecipitation of R-Ras with EphB2, and co-immunoprecipitated R-Ras was phosphorylated. EphB2 effects on adhesion and proliferation may be mediated by R-Ras since both adhesion and proliferation in U251-EphB2 were recovered upon silencing of endogenous R-Ras with siRNA, while R-Ras siRNA did not affect aggregation or proliferation. Overexpression of EphB2 decreased phosphorylation of mitogen-activated protein kinase (MAPK), which was recovered upon treatment with R-Ras siRNA. Proliferation of U251-EphB2 treated with R-Ras siRNA was increased, and the R-Ras siRNA-induced increase in proliferation and MAPK phosphorylation was decreased by PD98059, which is a MAPK kinase inhibitor. R-Ras protein, as well as phosphorylation, increased with grade in human astrocytoma specimens. Laser capture microdissection of invading glioblastoma cells revealed elevated R-Ras mRNA (1.5–26-fold) in 8 out of 8 biopsy specimens. Immunohistochemistry portrayed R-Ras localization primarily in glioblastoma cells and not in normal brain. Aggregation of U251 stably transfected dominant negative form of EphB2 (U251-EphB2DN) was promoted compared with parental cells and was inhibited by GC-4, which is an N-cadherin blocking antibody. Increased expression level of N-cadherin in U251-EphB2DN cells was demonstrated by quantitative RT-PCR, Western analysis and immunocytochemistry. The PI3K inhibitor LY294002 inhibited EphB2-stimulated migration concomitant with phosphorylation of Akt. These results suggest that EphB2 decreases adhesion through R-Ras, diminishes proliferation by inhibition of R-Ras/MAPK pathway, reduces aggregation by decreasing expression of N-cadherin and increases migration by the PI3K/Akt pathway independently of R-Ras. EphB2 expression and signaling appear to play significant roles in glioma invasion. This work was supported by the American Brain Tumor Association (MN) and NS042262.


Jean L. Nakamura, Daphne A. Haas-Kogan, and Russell O. Pieper; Departments of Radiation Oncology and Neurological Surgery, University of California San Francisco, San Francisco, California, USA

Despite the standard role of radiotherapy in the management of malignant gliomas, there are conflicting data as to the effects of ionizing radiation on glioma migration. To address the role of irradiation in glioma invasion, EGFP-labeled U251 MG GBM cells were irradiated with sublethal doses of radiation (0.1, 0.3, 1, or 2 Gy) prior to plating on top of confluent control or sublethally irradiated normal human astrocytes. Ninety-six hours after plating, coverslips were fixed, and the number of invasive, EGFP-labeled glioma cells was scored at successive distances. Cell proliferation was quantitated by using FACS. Initial validation studies demonstrated that non-transformed human astrocytes failed to establish growth onto the confluent astrocytic bed, while transformed and malignant cell lines (glial derived and nonglial) demonstrated both growth and invasiveness. Using FACS to assess cellular proliferation, we found that while 1 Gy irradiation had no effect on glioma cell numbers, it reduced the number of migratory/invasive cells by 70% (P < 0.01 by one-way ANOVA and post hoc Neuman-Keuls). A dose response to the inhibition of invasion was observed, which began at 0.3 Gy and saturated at 1 Gy. Irradiation of the astrocytic bed alone with escalating single doses of radiation (up to 4 Gy) failed to inhibit glioma invasiveness. Concurrent irradiation of the astrocytic bed and glioma, however, led to a statistically significant further reduction in invasion compared to irradiation of glioma or astrocytic bed alone. These results show that although sublethal irradiation reduces the ability of malignant glioma to invade into established astrocytes, maximal inhibition of invasion requires concurrent irradiation of the tumor and tumor bed. These observations further suggest that glioma invasion is dependent on interactions between the tumor and tumor microenvironment, which can in turn be altered by ionizing radiation.


Elizabeth W. Newcomb, Tona Schnee, Yevgeniy Lukyanov and David Zagzag; New York University School of Medicine, New York, New York, USA

Cells have developed a highly efficient system for rapidly adapting to low oxygen levels. HIF-1 is the main transcription factor activated by hypoxia. HIF-1 is a heterodimeric transcription factor consisting of two subunits, HIF-1α and HIF-1α. The molecular pathway leading from the sensing of hypoxia to the activation of HIF-1 is critically dependent on the relative abundance and stabilization of the HIF-1α subunit. Under hypoxic conditions, HIF-1α accumulates in the cell in the nucleus. Since we are screening the potential of drugs to downregulate HIF-1α expression together with downregulation of its associated angiogenic activities, we wanted to determine whether stimulation of human U87MG glioma cells with the hypoxia mimetic CoCl2 would upregulate expression of HIF-1α, as we have described, and induce its accumulation in the nucleus. U87MG glioma cells were untreated or exposed to 125 μM CoCl2 for 5 h and processed for immunofluoresence with mouse anti-HIF-1α antibody. Similar to hypoxic stimulation, CoCl2 treatment upregulated HIF-1α in the nucleus shown by its colocalization with propidium iodide counterstaining of nuclei. Next, we wanted to determine whether noscapine, a novel tubulin-binding agent, could downregulate HIF-1α expression. We show that noscapine downregulates hypoxia-mediated HIF-1α expression in a dose-dependent manner using Western blot analysis. Treatment of U87MG cells simultaneously with 100 μM noscapine and 125 μM CoCl2 for 5 h reduced HIF-1αexpression by >50%. Treatment with a lower dose of 50 μM noscapine for 24 h reduced HIF-1α expression by >90%. We confirmed the Western blot data using immunofluorescence for HIF-1α at 5 h and after 24 h of treatment. We observed decreased accumulation of HIF-1α in nuclei of drug-treated cells. Blocking translocation to the nucleus would be expected to limit upregulation of HIF-1-regulated target genes, such as VEGF, a potent angiogenic factor. Our results show an association between disruption of the cytoskeleton and inhibition of HIF-1α expression and its accumulation in the nucleus that may explain the antiangiogenic activity associated with microtubule-binding chemotherapeutic agents in general. These findings have important implications for treatment of solid tumors and should be considered as a novel antiangiogenic treatment strategy for glioma.


J.C. Reijneveld,1,2 D. Brandsma,1,2 W. Boogerd,5 J.M. Bonfrer,6 S. Kalmijn,3 A. Geurts-Moespot,4 E.E. Voest,1 and M.J.B. Taphoorn2,7; 1Laboratory of Medical Oncology from the Division of Medical Oncology, 2Department of Neurology, and 3Julius Center for General Practice and Patient Oriented Research, University Medical Center Utrecht, Utrecht; 4Department of Chemical Endocrinology, University Medical Centre St. Radboud, Nijmegen; Departments of 5Neurology and 6Clinical Chemistry, Antoni van Leeuwenhoekhuis, Amsterdam; and 7Department of Neurology, Medical Center Haaglanden, Den Haag; The Netherlands

We have previously shown that angiogenesis is important in the progression of murine leptomeningeal metastases (LM). The goal of the present study was to determine whether angiogenesis-related factors are upregulated in the cerebrospinal fluid (CSF) of patients with LM and whether these proteins are correlated with clinical features and/or other CSF parameters. We investigated protein levels of vascular endothelial growth factor (VEGF) and urokinase-type plasminogen activator (uPA) in the CSF of 56 patients with proven LM, 25 patients without malignancy, and 18 cancer patients without LM. Patients with LM suffered from breast cancer (n = 38), non-small-cell lung cancer (n = 6), non-Hodgkin lymphoma (n = 5), melanoma (n = 2), and multiple myeloma, ovarian cancer, rectal carcinoma, acute lymphocytic leukemia, and melanoma or breast cancer (one patient each). Median VEGF and uPA levels were higher in LM patients than in subjects without malignancy and cancer patients without LM (VEGF median [range]: 1403 pg/ml [0–13,100], 102 pg/ml [29–262], and 92 pg/ml [0–245], respectively; uPA median [range]: 227 pg/ml [44–1375], 163 pg/ml [52–282], and 99 [41–226], respectively). VEGF and uPA were higher than the maximum levels found in the other two groups in 82% and 38% of patients with LM, respectively. Both VEGF and uPA CSF concentrations were higher in patients with breast cancer than in patients with other malignancies. This study shows that the angiogenesis-related factors VEGF and uPA are upregulated in a large proportion of patients with LM. This finding supports the hypothesis that angiogenesis and matrix remodeling contribute to the progression of LM. Apart from that, determination of intrathecal angiogenesis-related factors may also aid in the diagnosis of LM.


Radoslaw Rola, Scott R. VandenBerg, Isa M. Hussaini, Duncan Morhardt, and John R. Fike; Brain Tumor Research Center, Department of Neurological Surgery, Department of Pathology, University of California San Francisco, San Francisco, California; Department of Pathology (Neuropathology), University of Virginia Health Science Center, Charlottesville, Virginia; USA

The diffuse and aggressively invasive growth of glioblastomas into the surrounding brain significantly contributes to the short survival times and high morbidity associated with these tumors. Such invasion disperses tumor cells significant distances from the tumor mass, analogous to the migratory capacity of neuronal progenitor cells. Pilocytic astrocytomas, by contrast, have limited capacity for brain invasion while retaining a selective capacity for cellular motility. Doublecortin (DCx), a protein that is required for micro-tubule polymerization or stabilization and that is often coexpressed with class III α-tubulin, is considered a marker for these migrating neuroblasts. The role of DCx in the migration of neoplastic astrocytes is poorly understood. In the present study we show that DCx is expressed in highly motile fetal astrocytes, in transformed astrocytic cell lines, and in subpopulations of both types of astrocytic tumors. For cell cultures, we used fetal human astrocytes (NHA), cell lines derived from transformed fetal astrocytes (h-TERT/E6/E7, h-TERT/E6/E7/RAS, h-TERT/E6/E7/RAS/EGFR wt, h-TERT/E6/E7/RAS/AKT, and h-TERT/E6/E7/RAS/Beta-3), and two established glioblastoma cell lines (U1242MG, U251MG). For intact tumor tissue, 19 samples of glioblastomas were obtained from BTRC tissue bank, and seven samples of pilocytic astrocytomas were obtained from the University of Virginia. Western blot analysis, motility analysis, and immunohistochemistry for DCx were performed as previously described (Hussaini et al., 1999, Glia 25:71; Mizumatsu et al., 2003, Cancer Res. 63:4021; Hirose et al., 1997, Cancer 79:989). Fetal astrocytes, the most motile cells in monolayer culture expressed DCx at the highest levels, and DCx expression positively correlated with cellular motility in monolayer cultures for most cell lines studied. Subsequent analysis of intact astrocytic tumor tissue revealed the presence of distinct subpopulations of DCx-positive cells in both glioblastomas and pilocytic astrocytomas. Both DCx and TUJ1 (class III α-tubulin)-immunoreactive cells were more abundant in glioblastomas than in pilocytic astocytomas, and the cytoarchitecture of these cells varied from bipolar to multipolar cells with conspicuous processes. Neither tumor expressed other neuron-associated epitopes. We conclude that specific subpopulations of neoplastic astrocytes expressing DCx might be involved in a long distance migration from primary tumor site.


Susobhan Sarkar and V. Wee Yong; University of Calgary, Calgary, Alberta, Canada

Invasion of glioma cells involves the attachment of invading tumor cells to extracellular matrix (ECM), disruption of ECM components, and subsequent cell penetration into adjacent brain structures. These processes are accomplished in part by tumor-secreted matrix metalloproteinases (MMPs). Although ECM proteins play many important roles, a better understanding of the influence and mechanisms of ECM proteins on the invasive behavior of the glioma cells in three-dimensional (3D) barriers is yet to be clearly defined. The present study investigated the role of different ECM proteins, including tenascin-C (TN-C), collagen type IV (CL-IV), and vitronectin (VN) on glioma cell invasion and/or migration and the impact of ECM on MMP expression. We used a 3D matrix of collagen type 1 to determine the invasive behavior of glioma cells. Glioma cells were suspended into the 3D collagen gel impregnated with or without TN-C, CL-IV, or VN, which was then added to transwell inserts. Cells were allowed to invade through the matrix at 37° for 24 h, which was then quantitated. Gelatin zymography was performed with serum-free conditioned medium collected after 24 h of cell culture either from cells grown in 3D collagen gel or in monolayer treated or coated with ECM proteins for determining proteolytic activity of MMP-2 or MMP-9. Migration of glioma cells from the 3D collagen gel impregnated with ECM proteins was also monitored. Furthermore, immunohistochemical analysis of cytoskeletal proteins expressed during invasion/migration of glioma cells was performed. We found that 3D growth promoted MMP-2 and -9 generation from glioma cells compared to monolayer cultures. In the 3D-collagen matrix, the addition of TN-C and CL-IV increased the invasion of glioma cells; TN-C was found to be superior to CL-IV. The effect of TN-C or CL-IV was not attributed to MMP-2 or MMP-9, as no upregulation or activation of these proteases from controls was found. Current experiments using TaqMan PCR of all known MMPs seek to identify upregulated MMP members that might account for glioma invasiveness in the TN-C 3D matrix. With respect to migration out of a 3D matrix, TN-C and CL-IV were facilitatory compared to VN. Immunohistochemical analyses for actin filaments and vinculin demonstrated evidence of an increased invasive phenotype in the TN-C 3D matrix compared to VN. Our findings show that TN-C plays a facilitatory role in inducing glioma cell invasion in 3D growth and that this invasive phenotype is possibly mediated by activation of MMPs other than MMP-2 or MMP-9. The further investigations of permissive 3D substrates and MMPs would be useful in the development of effective treatment protocols to prevent glioma cell invasion.


Hanqiu Song,1 Pirjo Laakkonen2 and Gabriele Bergers1; 1Department of Neurosurgery and BTRC, University of California Comprehensive Cancer Center, University of California San Francisco, San Francisco, California, USA; 2Molecular/Cancer Biology Laboratory, Biomedicum Helsinki, University of Helsinki, Helsinki, Finland

Sufficient oxygen and nutrient delivery is a prerequisite not only for normal tissue but also for tumors. When oxygen levels drop to a critical point, tumors initiate new blood vessel growth in order to further propagate. While low-grade astrocytomas expand between existing blood vessels and co-opt them without the necessity to switch on angiogenesis, the fast and aggressively growing grade IV astrocytomas/glioblastomas need to initiate their own blood supply. To reveal the molecular differences evident in the “nonangiogenic” co-opted and angiogenic vasculature of astrocytomas, we use a peptide phage display profiling technology to identify phenotype-specific endothelial markers of co-opted and highly angiogenic blood vessel in different genetically engineered mouse models of astrocytomas. The goal is to identify peptides that specifically bind to either angiogenic or co-opted brain tumor vessels. Linking those peptides to cytotoxic agents can then help to specifically target angiogenic as well as nonangiogenic tumors and tumor lesions that disperse along blood vessels (perivascular invasion) in the brain parenchyma.


Virginia Sykes, Nermine Abd-Elfattah, Jessica McCready, William C. Broaddus, and Helen L. Fillmore; Department of Neurosurgery and Harold F. Young Neurosurgical Center, Virginia Commonwealth University, Richmond, Virginia, USA

A key feature of malignant gliomas is the ability of gliomas to aggressively infiltrate surrounding brain tissue. Certain members of the matrix metalloproteinase (MMP) family are thought to be involved in this process. Until recently it was thought that MMPs were responsible only for the degradation of extracellular matrix components. However, it has become clear that MMPs have a wide range of influences on biologic processes, including generation and inactivation of bioactive proteins such as latent growth factors, growth factor–binding proteins, chemotactic molecules, coagulation factors, and cell adhesion molecules. A functional single nucleotide polymorphism (SNP) in the MMP-1 promoter consists of either the presence (2G allele) or absence (1G allele) of a guanine nucleotide at position −1607. This creates a core-binding site for the Ets family of transcription factors adjacent to an AP-1 consensus site and can result in significantly higher transcriptional activity of MMP-1. We have recently reported a significant preponderance of the 2G allele in glioblastoma patients relative to the normal population (P < 0.02). There is also a significant increase in MMP-1 protein expression in glioma tissue compared to that in normal brain tissue (P < 0.02). We have also reported that the 2G genotype correlates with significant increases in MMP-1 transcriptional activity in human glioma cell lines when compared with the 1G allele. The tumor suppressor protein p53 has been shown by others to inhibit MMP-1 transcriptional activity, but is mutated in 30% to 40% of gliomas. We therefore wanted to determine if wild-type p53 was able to inhibit the increase in transcriptional activity due to the functional SNP at position −1607. Promoter/luciferase assays were performed with 1G and 2G promoter constructs in a p53 null cell line, LN-Z308, that are cotransfected with control vector, p53 wild-type, or p53 mutant expression constructs. MMP-1 transcriptional activity decreases in LN-Z208 cells expressing wild-type p53 but not mutant or vector control (P < 0.05). To examine the functional consequences of MMP-1 overexpression in human gliomas, stable cell lines were created that overexpress MMP-1. Four out of five clones demonstrated significant increases in invasion compared to vector control cells (P < 0.02). These studies suggest that the presence of the 2G allele, especially in the context of mutant p53, may favor a more invasive phenotype. This polymorphism may provide a mechanism for the increased expression of MMP-1 via the elevation of MMP-1 transcription, which may contribute to the extreme invasiveness of high-grade gliomas.


Nhan L. Tran, Benjamin Savitch, Wendy McDonough, Michael E. Berens; The Translational Genomics Research Institute, Phoenix, Arizona, USA

Glioblastoma multiforme (GBM) is the most common primary malignant brain tumor, biologically unique by virtue of the proclivity for local invasion into the adjacent normal brain tissue and the rarity of systemic metastasis. Genes overexpressed in migrating glioma cells have been identified by using DNA microarray analysis comparing migration-activated glioma cells to passive glioma cells. Several gene candidates potentially involved in cell migration or invasion, as well as anti-apoptotic related functions, have emerged. Among the migration-associated candidate genes is the transmembrane receptor fibroblast growth factor-inducible 14 (Fn14), whose level of expression correlates with glioma migration. The ligand for Fn14 is called TNF-like weak inducer of apoptosis (TWEAK). Fn14 is a growth-factor-inducible immediate-early-response gene encoding a 102-amino-acid type I trans-membrane protein. Fn14 is overexpressed in migrating astrocytoma cells in vitro and in invasive high-grade astrocytoma clinical specimens (Tran et al., 2003, Am. J. Pathol. 162:4). To elaborate the role of Fn14 in glioma pathobiology, we examined Fn14 activation as a potential mechanism by which cell survival is fostered. TWEAK treatment of glioma cells resulted in the activation of NFkappaB and subsequently the translocation of NFkappaB from the cytoplasm to the nucleus. In addition, Fn14 activation results in the induction of Bcl-xL mRNA, a potential candidate of NFkappaB-dependent transcriptional regulation. Interestingly, TWEAK pretreatment of glioma cells increases cellular resistance to TRAIL-induced apoptosis. Moreover, over-expression of Fn14 or activation of Fn14 by addition of TWEAK results in a 3- to 4-fold induction of astrocytoma cell motility. Overexpression of Fn14 in GBM cells induces the formation of filopodia, lamellapodia, and microspikes. Fn14 colocalizes with Rac1, a member of the Rho family of GTPases. Furthermore, immunoprecipitation studies show that Fn14 physically associates with Rac1 GTPase and enhances Rac1 activation. Thus, Fn14 overexpression triggers signaling pathways that dramatically influence tumor cell behavior, including activated motility functions and disengaged apoptotic activation. We hypothesize that the Fn14 protein functions, in part, to stimulate glioblastoma cell invasion into the normal brain parenchyma while conferring apoptosis resistance to these invasive cells. Targeted therapy against Fn14, or its signaling pathways, as an adjuvant to surgical extirpation may specifically treat invasive glioma cells and improve the outcome of this devastating cancer. This work was supported by NS042262 and NS043446.


Jiang Ling Wu, Tatsuya Abe, Yoshihira Kimba, Ryo Inoue, Minoru Fujiki, and Hidenori Kobayashi; Department of Neurosurgery, Oita University Faculty of Medicine, Oita, Japan

Our previous study demonstrated that IκBαM could inhibit glioma angiogenesis and tumorigenesis through downregulation of VEGF and IL-8. However, pathways involved in VEGF expression are not well understood. Growing evidence indicates that hypoxia-inducible factor-1α (HIF-1α) and cyclooxygenases-2 (COX-2) play important roles in this progression. In this study, we injected nude mice with human glioma cells (IN500ΔEGFR) that were transfected with mutant IκBαM (IκBαM) or control plasmid. After three weeks, the nude mice were moved to a hypoxic chamber (10% oxygen) for 3, 12, 24, 48, 96, or 144 h. The expression of HIF-1α, COX-2, and VEGF was then analyzed by Northern blot and immunohistochemistry. We found that hypoxic stress induced the expression of HIF-1α and COX-2 and that VEGF induction followed expression of these genes in control tumors. However, IκBαM suppressed these processes. These findings indicate that VEGF expression appears to be regulated through dual interdependent mechanisms involving HIF-1 and COX-2 genes, and IκBαM could inhibit VEGF expression through these two ways. Thus, IκBαM is identified as a pivotal factor in angiogenesis and a potential target for neoplasm therapy.


Gelareh Zadeh,1 Rob Reti,1 Baoping Qian,1 Patrick Shannon,2 and Abhijit Guha1; 1BTRC, Hospital for Sick Children, University of Toronto, Toronto, Ontario; 2Department of Pathology, University Health Network, Toronto, Ontario; Canada

VEGF and angiopoietins collaborate in a highly orchestrated manner to regulate the normal vasculature. This study focuses on investigating the collaborative role of VEGF and angiopoietins in astrocytoma angiogenesis. Established human astrocytoma cell lines were used to generate stable cell lines that were grown as subcutaneous xenograft models in NODSCID mice and studied for their growth and vascularity. Parental U87 cells express high levels of Ang1 and VEGF, while parental U373 cells express low amounts. Constitutive and tetracycline inducible stable U87 and U373 cell lines over-expressing Ang1 were generated (U87:Ang1 and U373:Ang1). To determine the comodulatory effects of VEGF and angiopoietins, we generated double stable cell lines to manipulate both VEGF and Ang1 levels. U87 cells were generated to have downregulation of VEGF using anti-sense strategy and overexpression of Ang1. Tumor vascularity was analyzed by determining microvascular density (MVD) and endothelial cell (EC) proliferation, with the vessel maturity index determined by degree of EC interaction with smooth muscle cells and vessel leakiness. We found that the effect of Ang1 overexpression on astrocytoma angiogenesis and growth depended on the endogenous levels of VEGF expression. In xenografts with high baseline VEGF levels, overexpression of Ang1 enhanced tumor angiogenesis and growth. In contrast, in xenografts with minimal baseline levels of VEGF, overexpression of Ang1 decreased the overall angiogenic growth of tumors. Additionally, Ang1 induced vascular architectural changes resembling “glomerular tufting” seen in human GBMs, characterized by piling of endothelial cells. The level of glomerular tufting in U87:Ang1 xenografts was directly dependent on Ang1 levels, as determined by tetracycline-regulated Ang1 expression. The glomeruloid tufts demonstrated effective luminal blood flow, with no altered vessel leakiness. Our findings indicate that Ang1 synergizes the potent tumor angiogenic response triggered by VEGF, increasing MVD and generating more stable vascular structures. However, in the absence of the initial angiogenic response triggered by VEGF, Ang1 does not have a pro-angiogenic effect. These findings have important therapeutic implications, supporting the need for combinatorial anti-angiogenic therapy. We have also found that Ang1 is a potential key molecular regulator of the characteristic glomeruloid tufting seen in human GBMs, and the U87:Ang1 xenografts can provide an ideal model in which to further investigate the significance and functional relevance of glomerular tufting in GBMs.


David Zagzag and Elizabeth W. Newcomb; New York University School of Medicine, New York, New York, USA

Glioblastoma multiforme (GBM) cells have a poor prognosis primarily because of their ability to invade the brain. Hypoxia is associated with adverse outcome for a number of solid tumors, including GBMs. We have previously reported the expression of hypoxia inducible factor 1α (HIF-1α) in invading tumor cells of GBMs. Recent studies suggest that CA9 is a useful intrinsic marker of tumor hypoxia. CA9 is a transmembrane enzyme with an extracellular active site overexpressed in a wide variety of tumor types. Carbonic anhydrases are proteins involved in the reversible catalytic hydration of carbon dioxide to carbonic acid. Here we use the hypoxia marker CA9 to investigate the relationship between hypoxia, expression of HIF-1α, and invasion of GBM cells into the brain adjacent to the tumor. To visualize invading tumor cells in the brain adjacent to tumor, we immunostained tissue sections from p53 positive GBMs with mouse anti-human p53 (clone DO7). Thirteen cases of untreated GBMs were selected to include tumor core and invasive edge. CA9 staining was present in 94% of GBMs in pseudopalisading cells around areas of necrosis. No staining for CA9 was found in invading gliomas cells immunopositive for p53. By contrast, HIF-1α was expressed in both pseudopalisading and invading GBM cells as previously described. On the basis of these results we want to propose that the expression and HIF-1α in invading tumor cells is controlled by mechanism(s) other than hypoxia.


Abdessamad Zerrouqi,1,2 Beata Pyrzynska,1,2 Daniel Brat1,4 and Erwin G. Van Meir1,2,3,5; 1Laboratory of Molecular Neuro-Oncology, Departments of 2Neurosurgery, 3Hematology/Oncology, and 4Pathology, and 5Winship Cancer Institute, Emory University School of Medicine, Georgia, USA

In a high proportion of malignant gliomas, homozygous deletion of the INK4A/ARF locus causes the loss of both INK4A and ARF genes. This co-deletion complicates our understanding of the individual contributions by these two genes in tumor suppression. How genetic inactivation of ARF, independently of p16INK4A, participates in the malignant process is poorly understood. The aim of the current study was to determine whether ARF functions as an anti-tumorigenic factor in human glioma cells and to examine mechanisms. To evaluate ARF function as a tumor suppressor, we engineered human glioma cells that were deficient in the INK4A/ARF locus to conditionally overexpress an ARF cDNA and that carried either a transcriptionally active p53 (LN229) or null allele (LNZ308). Our results showed that the re-expression of ARF in these cells inhibited both cell proliferation and plating efficiency. These effects were observed in a p53-dependent and independent manner. In p53-positive cells (LN229), these effects are due to p21 expression following p53 stabilization by p14Arf. We are further examining the mechanism in p53-null cells. Besides its function as a regulator of tumor cell growth, our preliminary results suggest that ARF negatively regulates pro-angiogenic activities of tumor cells. Conditioned culture media from glioma cells expressing p14Arf was able to inhibit the pro-angiogenic behavior of human microvascular endothelial cells in Boyden chamber and Matrigel plug assays. Furthermore, induction of p14ARF inhibited the expression of the key angiogenic factors: vascular endothelial growth factor and angiogenin. We are currently examining the causality of these two observations. In summary, our findings demonstrate that ARF controls both cell cycle and angiogenic activities in malignant glioma independently of p16INK4A and p53, and this work emphasizes the pathological significance of the loss of ARF per se during glioblastoma formation. This work was supported in part by the Pediatric Brain Tumor Foundation of the U.S.


S. Zhang, T. Burton, P. Baijal, E. Henson, N. Bristow, M. Fonseca, S.B. Gibson, and D.D Eisenstat; Manitoba Institute of Cell Biology, University of Manitoba, Winnipeg, Manitoba, Canada

Astrocytic tumors are the most common brain tumor. Glioblastoma multiforme (GBM, WHO grade IV) represents the most malignant form of astrocytoma with a time to progression of 12 weeks without intervention and survival of 12 to 15 months with multimodality therapy. Response to therapy may be due to tumor hypoxia facilitating resistance to radiation and chemotherapy. The BCL2 nineteen-kilodalton interacting protein, BNIP3, a Bcl-2 family member upregulated in hypoxic regions, is activated by HIF1 alpha and mediates cell death in a caspase-independent manner through interaction of its transmembrane (TM) domain with mitochondria. BNIP3 is expressed in malignant astrocytes and is upregulated in GBM, correlated with increased HIF1 alpha and glut-1 that indicate hypoxic regions within these tumors. In 16% of primary GBM we have detected mutations in the PEST domain of BNIP3, confirmed by SSCP, that result in a truncated protein lacking a functional TM domain. BNIP3 overexpression in glioma cells induces cell death, whereas treatment with antisense, dominant negative or mutant BNIP3 blocks hypoxia-induced cell death, due to failure of BNIP3 to localize to the mitochondria and inhibition of BNIP3-mediated mitochondrial dysfunction. We suggest that BNIP3 acts as a tumor suppressor and selective pressure within the tumor generates BNIP3 mutations providing a survival advantage. Our discovery could explain why treatments for malignant gliomas are often ineffective in hypoxic regions of these tumors.


Ping-Pin Zheng,1 M. van der Weiden,1 Peter A.E. Sillevis Smitt,2 T.M. Luider,2 and Johan M. Kros1; 1Department of Pathology, 2Neuro-Oncology and Neurology, Erasmus Medical Center, Rotterdam, The Netherlands

Gliomas are richly vascularized tumors displaying various patterns of angiogenesis. Tumor microvessels may arise from de novo vasculogenesis through single endothelial cells (ECs), endothelial precursor/progenitor cells (EPCs), or noncanalized endothelial cell chains (NCECCs). Furthermore, there is angiogenesis by sprouting or branching from pre-existing vessels, or co-option (i.e., taking over of host vessels by tumor tissue). In glioma microvasculature, not in the blood vessels of normal brain, we detected Hela-type caldesmon transcripts by RT-PCR on microdissected tumor vessels. Caldesmon (CaD) is an actomyosin regulatory protein found in smooth muscle (high molecular weight (h-CaD), 120–150kDa) and nonmuscle cells (low molecular weight (l-CaD), 70–80 kDa). The splicing variants for l-CaD are subdivided into inclusion (Hela-type) or exclusion (WI-38 type) of exon 1. In the present study we explored the actual Hela-type CaD protein expression in normal cerebral - and glioma vessels in a large variety of gliomas. To this aim, a Hela-isoform-specific antibody (corresponding to 18 residues of the N-terminal encoded by exon 1 of the CALD1 gene was synthesized. We found that the Hela-type protein isoforms are exclusively expressed in glioma vessels and are predominantly demonstrable in the early stage of neoplastic angiogenesis, but not in normal-looking coopted vessels within tumors or normal brain vasculature. The lineage of the ECs/EPCs and NCECCs was confirmed by the endothelial markers (CD31, CD34, glut-1, factor III). Furthermore, we found breakdown of the tight junctions between endothelial cells in tumor vessels expressing the Hela type isoforms. The results indicate that the Hela-type CaDs should be considered as a novel marker in early stages of neoplastic angiogenesis in glioma.



F. Balawi,2 C. Lorenzana,2 D. L’Heureux,2 K. Khalili,2 S. Amini2 and S. Croul1,2; 1Drexel University College of Medicine and 2Temple University, Philadelphia, Pennsylvania, USA

Medulloblastomas represent brain tumors that are one of the most common malignancies in children. Although much attention has been focused on mechanisms that would induce undifferentiated neuroectoderm to evolve into fully formed tumors, less work has focused on the role of growth factors and cytokines in the evolution of these tumors. One of the clearly important molecular pathways in this respect is TGF beta. The present study was undertaken to understand in greater detail the role of TGF beta in medulloblastomas both in vivo and in vitro. While RT PCR analysis demonstrated TGF beta 1 RNA in the human medulloblastoma cell lines D283 and Daoy, protein levels determined by quantitative ELISA were only in the picogram range. Western blots demonstrated both the TGF beta I and II receptors in these cells. Addition of recombinant TGF beta 1 to these cells in culture significantly decreased proliferation. Migration assays demonstrated a significant repulsive effect of TGF beta 1. In tissue samples of medulloblastoma, TGF beta RI and RII were detected both by Western blots and immunohistochemistry. These results demonstrate that the TGF beta system is expressed in medulloblastomas and offer some evidence that it may function as a growth regulator and/or modulator of migration in these tumors.


Eric C. Burton,1 Kathleen R. Lamborn,1 William Weiss,1 Arie Perry,2 and Scott VandenBerg1; University of California San Francisco, San Francisco, California; 2Washington University, St. Louis, Missouri (A.P.); USA

Platelet-derived growth factor (PDGF) is a potent mitogen that acts through autocrine and paracrine mechanisms in the development of glial neoplasms. It is one of the few oncoproteins involved early in gliomagenesis, as evidenced by aberrant expression in low-grade glial tumors. The PDGF family consists of at least four ligands PDGF A-D that dimerize and signal through two different PDGF receptor tyrosine kinases (PDGFR alpha/beta). The ligands and receptors each form homo- and heterodimeric combinations that have specific binding capacities. Although several studies have looked at the role of PDGF in glioma development, none have characterized PDGF ligands and receptors in surgically resected tissues and correlated them with clinical outcomes. We used real-time PCR (RT-PCR) to measure levels of PDGF A-D and PDGFR alpha/beta mRNA in 18 frozen tissue samples from untreated patients presenting with WHO grade II oligo-astrocytomas (OA2). The median age at diagnosis was 36 years (range, 23–50 years). The median Karnofsky performance score was 90 (range, 80–90). Twelve patients underwent a gross total resection before first progression. Five patients were treated with radiation, chemotherapy, or both. Ten of 17 patients have progressed. The median time to tumor progression (TTP) was 41 months (95% CI, 34–92, Kaplan-Meier). Follow-up for the remaining patients was 48 months (range, 3–84 months). We compared the median mRNA expression levels of each ligand and receptor isotype in each tumor. PDGF A-C were expressed at a >10-fold higher level than PDGF-D (P ≥ 0.0001, Wilcoxon signed-rank). PDGFR-alpha was expressed at levels 5-fold higher than PDGFR-beta (P ≥ 0.0001). By proportional hazards analysis elevated levels of the PDGF-B ligand and the PDGFR-beta receptor corresponded with delayed TTP, (P ≤ 0.02 and P ≤ 0.03, respectively). Including a variable for RT and/or chemotherapy (Y/N) did not change the conclusions. PDGF A-C and PDGFR-alpha were expressed at high levels in OA2 consistent with an autocrine mechanism for PDGF-driven glial development. Levels of the PDGF-D ligand were significantly lower than PDGF A-C, suggesting a lesser role for this ligand in autocrine activation of PDGFR. High expression of PDGF-B and PDGFR-beta receptor correlated with delayed TTP. Relative levels of the PDGF mRNAs have been found to correlate with phenotypic differences in cell lines. Therefore, a relatively higher expression of these genes may be a marker for a less aggressive phenotype. Larger cohorts will be tested in order to verify this favorable prognostic phenotype.


Waldemar Debinski and Denise M. Gibo; Brain Tumor Center of Excellence, Wake Forest University School of Medicine, Winston-Salem, North Carolina, USA

We found that patients with glioblastoma multiforme (GBM) overexpress a restricted receptor for IL13, IL13Rα2. We proposed that an increase in AP-1 activity might be responsible for an upregulation of IL13Rα2 in GBM and the promoter region of the IL13Rα2 gene contains a functional AP-1 binding site indeed. We also found that Fos-related antigen-1 (Fra-1), a member of the AP-1 family of transcription factors, is upregulated in GBM. AP-1 activity controls many factors that enable progression of solid tumors by extracellular matrix-remodeling and by neovascularization promotion. We thus hypothesize that an unopposed activated AP-1 axis is the cause of upregulation of molecular markers of the disease and factors that may be crucial in GBM progression. One of the ways to examine this hypothesis is to have an ability to measure AP-1 directly in GBM. In this study, we used an assay in which the binding of transcriptionally active complexes composed of Fos and Jun proteins to the AP-1 site on the DNA oligonucleotide is detected quantitatively in a TransAM AP-1 system (Active Motif, Carlsbad, Calif.). The TransAM AP-1 system is quantitative, nonradioactive, simpler, and of higher sensitivity than the mobility shift assay. It can be used for the detection of AP-1 activity in situ by using tumor-derived nuclear extracts. In the current study, nuclear extracts obtained from SNB-19 and A-172 GBM cells were assayed for c-Fos, FosB, Fra-1, Fra-2, c-Jun, JunB, and JunD. All these factors prominently and variably contributed to the active complexes, with the exception of c-Fos and FosB. In another set of experiments, nuclear extracts were obtained from a freshly resected human GBM and from G48a (our human GBM cell line isolate) tumors growing in immunocompromised mice, and assayed by TransAM for Fra-1, Fra-2, c-Jun, JunB, and Jun D. The majority of transcriptionally active complexes contained Fra-2, JunB, and JunD with less contribution from the Fra-1 and c-Jun factors, and the pattern observed in a surgically resected GBM specimen resembled the one observed in a human xenograft growing in mice. Furthermore, we used H4 glioma cells carrying Fra-1 transgene and clonally expanded and measured AP-1 activity in H4(Fra-1+) cells. The H4(Fra-1+) clone, a high expressor of Fra-1, exhibited 100% increase in an involvement of Fra-1 in forming active DNA binding complexes when compared with control cells. This indicated that an ectopic Fra-1 contributes to an increased number of productive AP-1 pairs with Juns. In addition, the A-172 GBM cells transfected with fra-1 demonstrated a similar pattern of changes as seen in H4 glioma cells, while cells transfected with an anti-sense fra-1 demonstrated a decrease in the JunB participation in the AP-1 site binding. We thus demonstrate the feasibility to measure AP-1 activity in human GBM directly and also in situ, and we identified AP-1 transcription factors that are primarily taking part in this activity. Being that several AP-1 factors are highly upregulated in GBM versus normal brain and AP-1 factors regulate multiple cellular processes important in cancer maintenance and progression, further exploration of the importance of AP-1 in GBM is warranted.


John F. deGroot,1 Ta Jen Liu,1 Jeffrey Rothstein,2 Gregory Fuller,1 and W.K. Alfred Yung1; 1The Brain Tumor Center, M. D. Anderson Cancer Center, Houston, Texas; 2Johns Hopkins Hospital, Baltimore, Maryland; USA

There is accumulating evidence that glutamate may play a key role in the proliferation and invasion of glioblastoma tumors. Glutamate is an important neuronal excitatory neurotransmitter in the CNS, where it acts on the ligand-gated N-methyl-D-aspartate (NMDA) channel, amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA), and G protein–coupled glutamate receptors (GPCRs). Surprisingly, astrocytic tumors have been shown to release glutamate at high levels, which can stimulate tumor cell proliferation and motility via activation of glutamate receptors. Excess glutamate has also been found to facilitate tumor invasion by causing excitotoxic damage to normal brain, thereby paving a pathway for tumor migration. Using a tissue microarray, we found that human glial tumors do not express the most abundant excitatory amino acid glutamate transporter-2 (EAAT-2), which may contribute to the increased extracellular glutamate observed in glioma. EAAT-2, which clears extracellular glutamate in the brain and prevents excitotoxic neuronal damage, is not expressed in human high-grade glioma biopsy specimens but is present in low-grade tumors. To test the hypothesis that lack of the EAAT-2 glutamate transporter results in increased extracellular glutamate, which can stimulate glutamate receptors, we determined the effect of adenoviral-mediated EAAT-2 (Ad-EAAT-2) expression on glioma cell proliferation and invasion in vitro. Ad-EAAT-2 infection resulted in surface membrane expression of a functional transporter at levels comparable to normal astrocytes as determined by immunohistochemistry and a [3H]glutamate uptake assay. EAAT-2 expression reduced cell proliferation in a dose-dependent fashion in LN229, U373, U251, and SNB-19 glioma cell lines compared to controls using the MTT assay. Cell viability was reduced by 50% after 4 days in SNB-19 and U373 cell lines and 7 days in U251 and LN229 cell lines. Seventy-two hours postinfection, 18% to 22% of U251, U373, and SNB19 cells but not U87 were Annexin V positive by flow cytometric analysis. Caspase 3 activation determined by fluorometric assay and Western blot showing PARP cleavage suggest that cell death occurred via apoptosis. EAAT-2 expression also significantly reduced glioma invasion compared to mock and adenoviral-infected controls in a Matrigel-coated transwell invasion assay. Invasion was decreased to 40%, 20%, and 25% percent of control in U251, U87, and U373 cell lines, respectively. Reductions in MMP-2 and MMP-9 were seen on zymography and may partly explain the decrease in invasion. EAAT-2 loss may provide a survival advantage for glioma. These studies improve our knowledge of the role of glutamate receptors in glioma growth and invasion and may provide useful prognostic information and provide alternative therapeutic targets for the treatment of glioma.


Qiang Ding,1 J. Robert Grammer,1 Mark A. Nelson,3 Jun-Lin Guan,4 Jerry E. Stewart, Jr.,1 and Candece L. Gladson1,2; Departments of 1Pathology and 2Cell Biology, The University of Alabama at Birmingham, Birmingham, Alabama; 3Department of Pathology, University of Arizona, Tucson, Arizona; 4Department of Molecular Medicine, College of Veterinary Medicine, Cornell University, Ithaca, New York; USA

We have shown previously that the overexpression of wild-type focal adhesion kinase (FAK) promotes the proliferation of glioblastoma cells propagated in vitro in soft agar and in vivo in the scid mouse brain. Here we examined the necessity of FAK for glioblastoma cell proliferation in the scid mouse brain by expressing TET-inducible mutant FAK(397F), and we found a 50% inhibition of glioblastoma cell proliferation with two different clones. We then determined the effect of wild-type FAK and mutant FAK(397F) on cell cycle progression. Propidium iodide labeling followed by FACS analysis showed that wild-type FAK promoted exit from G1 and mutant FAK(397F) inhibited exit from G1 of glioblastoma cells propagated both in vitro as a monolayer and in vivo in the scid mouse brain. BrdU labeling experiments showed a corresponding increase or decrease in DNA labeling with expression of wild-type or mutant FAK, respectively. Immunoblotting studies of tumor cells propagated in both growth conditions demonstrated that wild-type FAK promoted the expression of cyclins D1 and E and inhibited the expression of p27(Kip1) and p21(Waf1), whereas mutant FAK(397F) inhibited expression of cyclins D1 and E and promoted the expression of p27(Kip1) and p21(Waf1). Downregulation of p27(Kip1) with siRNA blocked the inhibition of cell cycle progression observed with the expression of the mutant FAK(397F), while downregulation of p21(Waf1) or p57(Kip2) had no effect. Downregulation of cyclin D1 with siRNA inhibited the cell cycle progression observed with the overexpression of wild-type FAK, while downregulation of cyclin E had no effect. These data indicate that FAK promotes exit from G1 in glioblastoma cells propagated in vitro and in vivo in the scid mouse brain and that this regulation requires p27(Kip1) and cyclin D1.


Mrinal K. Ghosh,1 Shaik Ohidar Rahaman1 and S. Jaharul Haque1,2; 1Department of Cancer Biology, Lerner Research Institute, The Cleveland Clinic Foundation, Cleveland, Ohio; 2Brain Tumor Institute, Cleveland Clinic Foundation, Cleveland, Ohio; USA

Constitutively activated Stat3 is present in more than 90% of primary glioblastoma multiforme (GBM) tumors and in all GBM cell lines that we have examined (Rahaman et al., Oncogene, 2002, 21:8404). GBM cells frequently contain amplification and/or gain-of-function mutation of the EGFR gene leading to an activation of multiple downstream signaling pathways. In U87 and D54 cells, EGFR persistently activates Stat3 and the PI3K-AKT pathway, as the EGFR-specific inhibitor PD153035 inhibits the activation of both AKT and Stat3 and induces spontaneous apoptosis. PD153035-mediated apoptosis of U87 cells is partly due to an inhibition of Stat3-dependent expression of Mcl-1. However, Mcl-1 expression is not significantly reduced in U87 cells by treatment with PI3K-inhibitor (LY294002 or wortmannin) that blocks AKT activation with a concomitant induction of spontaneous apoptosis. Surprisingly, inhibition of PI3K activity significantly increases the DNA-binding activity of Stat3 in U87 and D54 cells, and this is not due an increase in the level of tyrosine phosphorylation in Stat3. Conversely, ectopic expression of constitutively activated form of AKT significantly decreases the DNA-binding activity of Stat3 in 293T cells. Interestingly, blockade of protein phosphatase PP2A activity in these cells by calyculin A stabilizes the phosphorylation of multiple serine residues that are located in the carboxy-terminal transactivation domain of Stat3, and this in turn completely ablates the DNA-binding activity of Stat3. Collectively, these results suggest that serine phosphorylation of Stat3 by the PI3K-AKT pathway negatively controls the DNA-binding function of Stat3 and hence the expression of Stat3-targeted genes in GBM cells.


N. Gupta, S.R. Liu, and Y. Liang; Brain Tumor Research Center, Department of Neurological Surgery, University of California San Francisco, San Francisco, California, USA

Malignant astrocytic tumors are associated with a marked inflammatory response in the surrounding brain. The various components of this response include infiltrating lymphocytes and cells of the monocyte lineage, separation of endothelial tight junctions, and activation of normal astrocyte populations. This response is present in untreated tumors but is augmented following therapeutic interventions such as radiation or chemotherapy. A sustained inflammatory response worsens symptoms and often limits further therapy. The molecular mechanisms underlying tumor-associated inflammation in the central nervous system remain poorly defined. It is also unknown to what degree neoplastic and normal cells interact to create a robust pro-inflammatory signal. Initially, we used an in vitro cDNA expression screen to identify potential mediators of this response. mRNA levels for monocyte chemoattractant protein-1 (MCP-1) were consistently overexpressed in a number of glioma cell lines. MCP-1 is the dominant chemoattractant cytokine for monocytes and macrophages. Using a panel of protein kinase inhibitors, we examined the regulation of MCP-1 secretion following stimulation by TNF-alpha in a number of cell types: glioma cells, endothelial cells, untransformed astrocytes, and transformed monocytes. The glioma cell lines produced high levels of MCP-1 at baseline nonstressed culture conditions. MCP-1 secretion was, however, strongly increased in all cell types in response to exogenous TNF-alpha. Although some variability existed between cell types, p38 MAP kinase appeared to be the primary regulatory pathway for MCP-1 in these cell types. We extended these studies by examining the expression of MCP-1 and its cell surface receptor, CCR2, in approximately 30 newly diagnosed glioblastoma specimens. MCP-1 was overexpressed in a heterogeneous pattern in the majority of specimens. In the GBM specimens screened, CCR2A, an alternatively spliced isoform of CCR2, demonstrated consistently increased expression compared to normal brain. CCR2A was localized primarily in the intracytoplasmic localization in glioma cell lines. In normal tissues, CCR2 is primarily expressed in monocytes and endothelial cells. Finally, in preliminary experiments, we note that U87 glioma cells transfected with ectopic CCR2A demonstrate increased migration, suggesting that CCR2A contributes to glioma motility. In conclusion, our data suggest that MCP-1 is produced by both glioma and normal cells, is regulated by the p38 MAP kinase pathway, and may be an important tumor-associated pro-inflammatory cytokine in the central nervous system. Its receptor, CCR2, is also over-expressed in glioma cells, and this suggests that a positive feedback loop may exist that affects the phenotype of these cells in the setting of increased MCP-1 levels. This pathway is likely only one of many activated during glioma progression.


Terrance G. Johns, Rushika M. Perera, Angela A. Vitali, Sonja Vernes, and Andrew Scott; Ludwig Institute for Cancer Research, Melbourne, Australia

Mutations of the epidermal growth factor receptor (EGFR) gene are found at a relatively high frequency in glioma, with the most common being the de2-7 EGFR (or EGFRvIII). This mutation arises from an in-frame deletion of exons 2-7, which removes 267 amino acids from the extracellular domain of the receptor. Despite being unable to bind ligand, the de2-7 EGFR is constitutively active at a low level. Transfection of human glioma cells with the de2-7 EGFR has little effect in vitro, but when grown as tumor xenografts this mutated receptor imparts a dramatic growth advantage. We mapped the phosphorylation pattern of de2-7 EGFR, both in vivo and in vitro, using a panel of antibodies specific for different phosphorylated tyrosine residues. Phosphorylation of de2-7 EGFR was detected constitutively at all tyrosine sites surveyed in vitro and in vivo, including tyrosine 845, a known target in the wild-type EGFR for src kinase. There was a substantial upregulation of phosphorylation at every tyrosine residue of the de2-7 EGFR when cells were grown in vivo compared to the receptor isolated from cells cultured in vitro. Upregulation of phosphorylation at tyrosine 845 could be stimulated in vitro by the addition of specific components of the ECM via an integrin-dependent mechanism. These observations may partially explain why the growth enhancement mediated by de2-7 EGFR is largely restricted to the in vivo environment.


Masayuki Kanamori, Janice Nigro, Burt Feuerstein, Daphne Haas-Kogan, David Stokoe, Mitchel S. Berger, Lucio Miele, and Russell O. Pieper; Department of Neurological Surgery and the Brain Tumor Research Center, University of California San Francisco, San Francisco, California, USA

The four human Notch genes (1–4) encode heterodimeric transmembrane receptors. Activation of Notch receptors by ligand binding leads to cleavage and release of an intracellular fragment that translocates to the nucleus and stimulates transcription of downstream target genes. Notch activation plays a key role in proliferation and differentiation, and it recently was shown to be critical for Ras-mediated transformation of fibroblasts. Because many gliomas exhibit deregulated Ras signaling, we measured Notch levels and activation in primary samples and cell lines derived from glioblastoma multiforme as well as the contribution of Notch pathway activation to astrocytic transformation and growth. Notch-1 protein was overexpressed and/or activated in Ras-transformed astrocytes, in 3 of 4 GBM cell lines (U373, A172, U251) and in 4 of 5 primary GBM samples. Expression array analysis showed that cDNAs encoding Notch ligand (JAG1), Notch-3, and the downstream targets of Notch (Hes1, Hes2) were also overexpressed relative to non-neoplastic brain controls in 23%, 71%, and 51% of tumors in a panel of 35 primary human GBM, respectively. Pharmacologic inhibition of Notch cleavage/activation had a significantly greater suppressive effect on the colony-forming ability of cells overexpressing Notch-1 (U251, Ras-transformed astrocytes) than on that of cells that did not overexpress Notch-1 (U87). Furthermore, antisense RNA targeting Notch-1 significantly reduced cleaved/activated Notch-1 levels and suppressed the soft agar growth of cells exhibiting Notch-1 activation without altering the soft agar growth of cells in which Notch-1 was not activated. These results suggest that Notch activation contributes to Ras-induced transformation of glial cells and may represent a new target for glioma therapy.


Dmitri Kapitonov, John Boockvar, and Donald M. O’Rourke; Department of Neurosurgery, University of Pennsylvania, Philadelphia, Pennsylvania, USA

Glioblastoma multiforme (GBM) is the most common primary malignant neoplasm of adult central nervous system. Because of its invasive phenotype, treatment outcomes after extensive surgical resection, radiotherapy, and chemotherapy remain poor, with average survival of less then a year. EGFRvIII is expressed in about 50% of primary GBMs. Since neural progenitor cells are known to be more permissive for GBM formation than differentiated astrocytes, we decided to investigate EGFRvIII signaling in neural stem cells. Previously, our lab established a neuronal stem cell line overexpressing EGFRvIII and showed that it acquired an invasive, but not tumorigenic, phenotype when injected into rat brain. This observation suggested that EGFRvIII initiates or modifies signal transduction pathways to confer invasive phenotype. The aim of present work was to investigate EGFRvIII signaling in a C17.2-derived neural stem cell line. C17.2 EGFRvIII cells exhibited decreased spreading, attachment, and survival in serum-free media. These clones also exhibited increased reactive oxygen species (ROS) production when compared to controls. When serum-starved, EGFRvIII was not phosphorylated in these cells, but was phosphorylated on tyrosine residues corresponding to Y845 and Y1173 in a time- and concentration-dependent manner upon treatment with either EGFR ligands (EGF, HB-EGF), or other RTK ligands (HGF), cytokines (IL-6), GPCR ligands-chemokines (CCL2, CXCL8), sphingosine phosphocholine (SPC), lysophosphatidyl choline (LPC), ceramide-1-phosphate (C1P), lysophosphatidic acid (LPA), phosphatidic acid (PA), PKC activator PMA, and low serum (0.5%). No endogenous wild-type EGFR transactivation was observed in the parental cell line when treated with the same ligands. All ligands fall into several groups based on the dynamics of EGFRvIII phosphorylation. Downstream targets (Erk1/2, Erk5, JNK, p38, Akt, and IkB) were activated in EGFRvIII cells with different dynamics in a ligand-dependent manner and to a significantly higher degree than in parental cells. IkB phosphorylation by HB-EGF, CCL2, SPC, LPC, C1P, PA, and low serum was observed in EGFRvIII cells, but not in parental cells. Thus, it appears, EGFRvIII served as an amplifier of signals coupled to several pathways, which may explain the invasive phenotype of C17.2-EGFRvIII cells. Using pharmacological inhibitors, we investigated signal transduction pathways mediating EGFRvIII transactivation and signaling to downstream effectors, and the data suggest a pivotal role for Src family kinases, PI3K, and Ca2+/PLC pathways in EGFRvIII signaling and neural stem cell motility.


Gurpreet Singh Kapoor, and Donald M. O’Rourke; Department of Neurosurgery, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA

Recently, we showed that EGFR-mediated transcription in glioblastoma cells is regulated via a critical association between the Gab1 (Grb2-associated binder-1) adapter and the SHP-2 tyrosine phosphatase (Kapoor et al., 2004, Mol. Cell. Biol. 24:823). Our previous work has shown that inhibitory signal regulatory protein alpha1 (SIRPalpha1) receptors interact with SHP-2 to inhibit EGFR-mediated tumor migration, survival, and cell transformation (Wu et al., 2000, Oncogene, 19:3999). In this study, we present a model for the regulation of EGFR transcription by activating SHP-2 and inhibitory SIRPalpha1 proteins. SHP-2 significantly upregulates Akt kinase and NFkB activities in glioblastoma cells by associating with the Gab1 adapter protein, indicating that SHP-2 positively upregulates EGFR signaling and transcription. Conversely, SIRPalpha1 significantly reduces PI-3K/Akt, p42/44 kinase, and NFkB activities when compared to either empty vector controls or mutant SIRPalpha1-expressing cells, which indicates that SIRPalpha adhesion receptors negatively regulate EGFR-mediated signaling and transcription. Interestingly, Northern blot analyses reveal increased SIRPalpha receptor expression in EGFR-inhibited glioblastoma cells, suggesting that erbB complex kinase activity negatively regulates SIRPalpha transcriptional activation and expression. Consistent with this, immunoprecipitation assays using an anti-SHP-2 antibody show increased SIRPalpha expression, but decreased total Gab1 protein, in erbB-inhibited glioblastoma cells. Similarly, SHP-2 immunocomplexes from SIRPalpha cells show decreased total Gab1 when compared to controls and mutant SIRPalpha-expressing cells, indicating that functional SIRPalpha receptors interfere with Gab1/SHP-2 complex formation to inhibit EGFR-mediated signaling and transcription. Importantly, modulation of EGFR-mediated transcriptional programs by SHP-2 and SIRP correlates with transforming potency of glioblastoma cells. Collectively, our results indicate that there is balance between activating phosphatases and inhibitory receptors in regulating EGFR signaling and transcription and that perturbation of this balance may result in phenotypes with distinct transforming efficiencies.


Adel Kardosh,1 Nathaniel Soriano,3 Wei Jun Wang,3 Axel H. Schönthal,1 and Thomas C. Chen2,3; Departments of 1Molecular Microbiology and Immunology and 2Neurosurgery, Keck Medical School, and 3Department of Pathology, USC Center for Brain Tumor Research, University of Southern California, Los Angeles, California, USA

Primary central nervous system lymphoma (PCNSL), an aggressive form of non-Hodgkin’s lymphoma, is the number one intracranial lesion in patients with advanced HIV infection (AIDS) and is increasingly seen in immune competent elderly patients. We set out to investigate whether selective cyclooxygenase-2 (COX-2) inhibitors would affect the growth of intracranial lymphomas. As no known intracranial lymphoma cell line exists, we used the Raji B cell lymphoma line as a model for PCNSL. Raji cells were injected intracranially into nude mice. An intracranial tumor formed and localized in the brain parenchyma and leptomeninges in an indistinguishable manner from what is commonly observed in PCNSL The antiproliferation effect of various COX-2 inhibitors (Celebrex, Vioxx, Bextra) on Raji cells was compared in vitro by using an MTT assay, Celecoxib was found to have highest potency. In addition, the inhibitory effect of celecoxib was further demonstrated in vivo in nude mice injected with Raji cells subcutaneously. To determine the mechanism underlying the growth-inhibitory effects, we examined the activity of cyclin-dependent kinases (CDKs), vital regulators of cell cycle and proliferation. Celecoxib was found to downregulate CDK activity because of the transcriptional inhibition of cyclins, which are the necessary regulatory subunits of CDKs. In addition, DMC, a derivative of cele-coxib that lacks COX-2 function, exerted antiproliferative effects similar to those of celecoxib, indicating that the antiproliferative effect is independent of COX-2. Instead of COX-2, we found that two other major growth regulators, the MAP kinase pathway and the Akt/PKB pathway, were profoundly inhibited by celecoxib and DMC. Thus, our results indicate that celecoxib and/or DMC might be useful for the therapeutic management of intracranial lymphomas and that this potential is independent of celecoxib’s established role as a COX-2 inhibitor.


A. Lal, A.L. Slocum, P. Jun, J.F. Costello, R.O. Pieper, A.W. Bollen, and M.W. McDermott; Brain Tumor Research Center, Department of Neurological Surgery, University of California San Francisco, San Francisco, California, USA

Atypical and malignant (high-grade) meningiomas are aggressive tumors that are associated with a poor clinical outcome and have higher rates of recurrence when compared to benign meningiomas. Virtually nothing is known about the signaling pathways that are disrupted in these tumors. We used serial analysis of gene expression (SAGE) to identify molecular changes associated with meningiomas and focused our initial analysis on the 165 genes that are induced in high-grade meningiomas because this population is expected to contain components of activated signal transduction pathways. A novel finding from this screen is the induction of three downstream components of the Notch signaling pathway: the transcription factor, Hairy and Enhancer of Split 1 (HES1), and two members of the Groucho/Transducin-like enhancer of split (Gro/TLE) family of co-repressors, TLE2 and TLE3. Gro/TLE co-repressors interact and modulate the activity of a wide range of transcriptional regulatory systems, one of which is HES1. Using quantitative PCR on a larger, independent set of meningiomas, we confirm that HES1 transcript levels are induced in meningiomas of all three grades while induction of TLE2 and TLE3 is specific to high-grade meningiomas. In particular, overexpression of TLE3 is restricted to 50% of malignant meningiomas with the amplitude of induction ranging from 4-fold to 65-fold. We also show that protein levels of TLE3 are induced in malignant meningiomas and that TLE3 is correctly localized to the nucleus where it would carry out its transcriptional co-repressor functions. It is our hypothesis that Gro/TLEs are important for the malignant progression of meningiomas and that one of the mechanisms by which this occurs is by modulation of the Notch signaling pathway. The Notch 2 receptor and the Jagged 1 ligand are the primary homologs expressed in meninges and we also find that these upstream components are induced in a subset of meningiomas. We are currently evaluating the functional consequences of aberrant Notch signaling and TLE induction on meningioma growth and tumorigenesis.


J.-F. Lavoie,1 L. LeSauteur,2 J. Wong,2 C.J. Thiele,3 and D.R. Kaplan4; 1University of Toronto, Hospital for Sick Children, Toronto, Ontario, Canada; 2Montréal Neurological Institute, Montréal, Quebec, Canada; 3National Cancer Institute, Bethesda, Maryland, USA; 4Hospital for Sick Children, Toronto, Ontario, Canada

Neuroblastoma is the most frequent extracranial solid tumor of children. High expression of the TrkA/nerve growth factor receptor is a favorable prognosis marker of this tumor. To explore the molecular basis of why TrkA is a good prognosis marker, we examined the effect of expressing this receptor in human neuroblastoma cells that had little or no endogenous TrkA. TrkA, transduced via a recombinant adenovirus, induced the apoptosis of all tested neuroblastoma cells lines, while expression of kinase-inactive TrkA or of the TrkB/brain-derived neurotrophin receptor, did not. TrkA expression increased the levels of nuclear p53, p21Waf-1, and of cleaved caspase-3. Expression of the anti-apoptotic protein Bcl-XL via recombinant adenovirus or treatment with a broad-spectrum caspase inhibitor prevented TrkA-induced cell death. The apoptosis-inducing activity of TrkA required wild-type p53, as the neuroblastoma cell line SK-N-BE(2), which expresses a nonfunctional mutant p53, was resistant to TrkA-mediated death. These results suggest that TrkA induces the apoptosis of neuroblastoma cells by a p53-dependent mechanism and that one of the reasons that TrkA is a favorable prognosis marker for neuroblastoma is that high expression of TrkA is deleterious to this tumor’s survival. We propose that TrkA, which is a survival protein for many normal neural cells, may be an ideal gene therapy agent for the treatment of neuroblastoma.


Christopher A. Lipinski,1 Nhan L. Tran,2 Emmanuel Menashi,1 Kim Perkins,1 Carole Rohl,1 Jean Kloss,1 Michael E. Berens,2 and Joseph C. Loftus1; 1Mayo Clinic, Scottsdale, Arizona; 2Translational Genomics Research Institute, Phoenix, Arizona; USA

Invasion of malignant gliomas cells out of the tumor mass into surrounding normal brain tissue precludes effective tumor resection and reduces the efficacy of radiation treatment. In addition, invasive glioma cells are highly resistant to apoptosis induced by chemotherapeutic agents. This problem is not unique to gliomas, as invasion and metastasis contribute substantially to nearly every malignancy. Integrin-mediated interactions with extra-cellular matrix plays an important role in the process of glioma cell migration both mechanically and by activation of signaling networks that affect gene expression and cell growth. The related non-receptor tyrosine kinases FAK and Pyk2 function as integrin effectors in the regulation of cell growth and migration in many cell types. We have recently reported differential roles for FAK and Pyk2 in the proliferation and migration of high-grade glioma. Glioma cell in vitro migration correlated positively with Pyk2 activity while FAK activity stimulated cell cycle progression. To more precisely define the contribution of Pyk2 and FAK to glioma migration and proliferation, we generated human glioma cell lines stably expressing siRNAs targeting endogenous Pyk2 or FAK. The interfering RNAs used are highly effective (>95% target protein knockdown) and specifically knock down expression of the targeted kinase without altering expression of the corresponding kinase. Migration of SF767 glioma cells expressing RNAi targeting Pyk2 (SF Pyki) was significantly reduced compared to wild-type SF767 cells. Migration of cells expressing RNAi targeting FAK (SF FAKi) was also reduced compared to wild-type SF767 cells, but the inhibition was less than that observed in SF Pyki cells. To reconstitute Pyk2 or FAK expression, knockdown cells were infected with recombinant adenoviruses expressing rat Pyk2 or murine FAK, which differ from the respective targeted human sequence by a single nucleotide. Reconstitution of Pyk2, but not FAK, restored migration in both SF Pyki and SF FAKi cells in a dose-dependent manner. G112 glioma cells expressing RNAi targeting FAK were less proliferative than wild-type G112 cells, an effect rescued by reconstitution of FAK expression but not by re-expression of Pyk2. These data indicate that FAK preferentially stimulates glioma cell proliferation, but also contributes to cell migration while Pyk2 appears to have only promigratory effects.


Juinn-Lin Liu,1 Xiaoyang Sheng,1 Zhenyu Mao,1 Gary E. Gallick,2 and W.K. Alfred Yung1; 1Brain Tumor Center, Department of Neuro-Oncology and 2Department of Cancer Biology, University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA

Phosphoinositide 3-kinases (PI3K)/PDK1/AKT/mTOR/p70S6K signaling cascades are pivotal in regulation of cell survival, proliferation, differentiation, invasion, and angiogenesis. Conversely, PTEN, a tumor suppressor gene, has been identified to be a phosphoinositide phosphatase that specifically antagonizes the PI3K-mediated signaling pathways. Extensive deletions or mutations of PTEN are frequently encountered in various advanced tumors including glioblastomas. Wild-type PTEN has been found to be expressed both in the cytoplasm and nucleus in normal cells, with a preferential nuclear localization in differentiated or resting cells. However, the biologic function of PTEN in the nucleus remains poorly characterized. Previously we showed that nuclear PTEN alone is capable of suppressing anchorage-independent growth and facilitating G1 arrest in U251MG cells without affecting their cell invasiveness. Moreover, the nuclear PTEN-induced growth suppression is mediated, at least in part, through down-regulation of p70S6K phosphorylation/activation in an AKT-independent fashion. In this report, we demonstrate that the down-regulation of p70S6K could be alternatively mediated by AMPK-induced mTOR inhibition bypassing AKT/TSC/Rheb pathway. In addition, a functional lipid phosphatase domain is indispensable for nuclear PTEN’s growth-suppressing activities, as both nuclear PTEN (G129R) and PTEN (G129E) mutants are defective in suppressing cell growth of U251MG cells. Furthermore, using an ecdysone-inducible expression system in conjunction with Affymetrix oligonucleotide microarray analysis, we are able to identify several downstream target genes specifically modulated by nuclear PTEN that may contribute to cell growth suppression. Ongoing experiments are under way to verify the Affymetrix microarray results. Our goal is to investigate the biologic function of PTEN in the nucleus to fully understand how PTEN is involved in regulating cell growth and survival besides its signature activity to antagonize PI3K pathways through dephosphorylating PIP3 on the plasma membrane. Ultimately, our identification of nuclear PTEN-specific downstream genes will lead to the discovery of novel signaling molecules for GBM target therapy.


J. Micallef,1 J. Caldwell,2 M. Moran,2 and A. Guha1,3; 1Arthur & Sonia Labatt Brain Tumor Centre, Hospital for Sick Children, Toronto, Ontario; 2MDS Proteomics, Toronto, Ontario; 3Division of Neurosurgery, University of Toronto, Toronto, Ontario; Canada

The most common gain of function mutations in GBMs are amplification/overexpression of the wild-type (wt) and mutated epidermal growth factor receptors (EGFRs), the most common of which is EGFRvIII. The differential signaling pathways utilized by EGFRvIII vs. wt-EGFR, which may contribute to its more aggressive behavior, are not well known. A novel differential mass spectral analysis, termed ICAT, was used to examine the differential protein profile of GBMs harboring EGFRvIII vs. GBMs that only express wt-EGFR. From the ICAT analysis, MARCKS, a protein previously implicated in breast cancer, was identified to be markedly increased in EGFRvIII vs. wt-EGFR-expressing GBMs. To verify this ICAT finding, standard Western immunoblot analysis using MARCKS antibody was undertaken on a larger panel of GBM cell lines (Tet-Off expression of EGFRvIII or wt-EGFR), GBM xenografts (±EGFRvIII) and GBM operative specimens (±EGFRvIII). This study demonstrates proteomic-based technologies’ ability to identify differences at the level of the proteome, between different subtypes of human diseases, such as GBMs. Current work involves assessing the functional role of MARCKS in GBMs. MARCKS is being down-regulated in GBMs expressing EGFRvIII and increased in GBMs expressing only wt-EGFR. If the in vitro/in vivo functional assays demonstrate MARCKS involvement in promoting the aggressive biology of EGFRvIII-expressing GBMs, then it may be an additional GBM-specific biological target.


Neelanjan Mukherjee,1 Qinhui Song, Gary G. Zhai,1 Min Zhang,1 Rajeev Malhotra,1 Douglas E. Latham,1 Meaghan A. Delaney,1 Pierre Robe,2 Ulf Nestler,1,3 Jay Loeffler,1 and Arnab Chakravarti1; Departments of 1Radiation Oncology and 3Biostatistics, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts; 2Department of Neurosurgery, Brigham and Women Hospital/Harvard Medical School, Boston, Massachusetts; USA

The observed radioresistance of human glioblastoma multiforme poses a major challenge, which, if overcome, may lead to significant advances in the management of this patient population. There is accumulating evidence from correlative studies that survivin expression is associated with increased malignant potential of human gliomas. The purpose of this study was to investigate whether survivin plays a direct role in mediating radiation resistance in primary human glioma cell lines, and, if so, to investigate the underlying mechanisms. Our panel of GBM cell lines included two that were relatively radiation resistant (GM 20 and GM21) and two that were more radiation sensitive (GM 22 and GM23), which demonstrated differential levels of survivin expression between the two groups. Through the use of adenoviral vectors containing either dominant-negative [pAd-S(T34A)] or wild-type survivin [pAd-S(WT)], we were able to inactivate or overexpress survivin, respectively. Our findings suggest that survivin plays a critical role in mediating radiation resistance in primary GBM cells, in part through suppression of apoptotic cell death via a caspase-independent manner. We have identified novel mechanisms by which survivin may enhance tumor cell survival upon radiation exposure such as regulation of double-strand DNA break repair and tumor cell metabolism, which were most evident in the radiation resistant cell lines. These differences in survivin function both in radiation resistant vs. radiation sensitive cell lines and in the presence vs. absence of radiation exposure warrant further investigation and highlight potentially important mechanisms of radiation resistance in these tumors.


Sarah M. Myers,1,3 Joseph Najbaur,2 Elizabeth Garcia,2 Marianne Z. Metz,2 Michael E. Barish,1 and Karen S. Aboody1,2; Divisions of 1Neurosciences and 2Hematology/HCT, City of Hope National Medical Center and Beckman Research Institute, Duarte, California; 3Pomona College, Claremont, California; USA

Endogenous neural progenitor cells (NPCs) exist in mammals through adulthood, resident in the dentate gyrus of the hippocampus, the subventricular zone (SVZ) of the lateral ventricles, and the olfactory bulbs. These NPCs respond to many types of brain injury by proliferation and migration to the injured areas, but how endogenous progenitors will respond to tumors has not been clear. Here we examined the responses of endogenous NPCs to U251 human glioma in a mouse model. Animals were sacrificed at various time points following establishment of experimental glioma in frontal cortex of 8-week-old nude mice. Immunofluorescent techniques were used to detect markers of endogenous progenitor cells (mouse nestin), glioma (human nestin), astrocytes (glial fibrillary acidic protein, GFAP), and proliferation (proliferating cell nuclear antigen, PCNA) on tissue sections. On the ipsilateral (glioma-injected) side, we observed a stream of ms-nestin immunoreactive cells extending from the SVZ along the wall of the ventricle to the tumor, where they were found surrounding the tumor border and appearing to infiltrate the tumor mass. On the contralateral side (no glioma), ms-nestin immunoreactive cells were restricted to the SVZ, as is seen in control non-tumor brains. Near the tumor, PCNA immunoreactivity was most evident in reactive astrocytes (high-GFAP-expressing), as well as tumor cells, suggesting that endogenous NPCs may not be proliferating in the tumor microenvironment at the time points examined. Proliferation of NPCs in the SVZ is under investigation. Our results suggest that endogenous NPCs are attracted to, but not necessarily “recruited by” the tumor. Rather, NPCs may have a bystander role more suitable for potential repair of tumor-induced injury. Future experiments will examine the signal(s) to which the NPCs respond and the origins of these signals, as well as the precise nature of NPC responses over time.


Jennifer L. Nelms and Michael A. Vogelbaum; Center for Translational Therapeutics, Cleveland Clinic Brain Tumor Institute and Department of Neurosurgery, Cleveland, Ohio, USA

Targeted therapy of signal transduction pathways is a recently employed strategy for treating malignant gliomas. Most clinical investigation to date has had a single agent/single target focus. It has been suggested that combined targeting of multiple points in a signal transduction pathway would lead to synergistic effects. We have evaluated, in vitro, combined targeting of the epidermal growth factor receptor (EGFR) signaling pathway in human malignant glioma cell lines. We evaluated various combinations of the mTOR inhibitor rapamycin and the EGFR tyrosine kinase inhibitor AG1478 in U251, D54, U87, U87wtEGFR (overexpressing the wild-type EGF receptor), U87vIII (overexpressing the vIII mutant EGF receptor), and U87kd (over-expressing a kinase-dead form of the EGF receptor) human glioma cell lines. We examined the effects of these combinations on cell proliferation and apoptosis and we evaluated the relationships between these measures of cell function and the activation status of various components of the EGFR signaling pathway. Cell proliferation was reduced by about 10% to 20% following treatment with rapamycin (0.1 or 1 nM) and about 5% to 10% by AG1478 (5 μM). Combined treatment with rapamycin and AG1478 produced a range of 15% to 55% reduction in cell proliferation. The greatest benefit was observed in the U87vIII cell line. There was no significant increase in apoptosis seen when the two drugs were combined. Western blot analysis of the EGFR signal transduction pathway revealed a decrease in phospho-AKT and phospho-p70S6 kinase levels, without a change in the total protein levels, when the two drugs were combined compared to either drug alone. These data demonstrate that, compared to single target inhibition, combined inhibition of EGFR and mTOR reduces cell proliferation and increases the inhibition of downstream signaling proteins in human malignant glioma cell lines, regardless of EGFR phenotype. Further studies are under way to evaluate the efficacy of this combination in vivo.


Janet L. Oblinger, Ola M. Omran, James W. Waldman, Hany Saqr, and Allan J. Yates; The Ohio State University College of Medicine, Columbus, Ohio, USA

Glioblastoma multiforme (GBM) is the most common and aggressive primary human brain tumor in adults. The average survival is 8 to 10 months after diagnosis and 2-year survival is less than 10%. However, the occasional patient with this tumor lives somewhat longer. The prognosis of GBMs has been little changed in the past decades despite improvements in imaging and treatment strategies. Thus the development of new treatment modalities for GBM patients is of high importance. The disialoganglioside GD3 has been previously reported to induce apoptosis in several tumor cell lines. The U-1242 MG glioblastoma cell line has GM2 and GM3, but no GD3. Using this as the parental cell line, we transfected it with the GD3 synthase gene under the control of a modified Tet-On system. Our results show that in response to doxycycline this cell line expresses considerable amounts of GD3 and undergoes apoptosis. This involves the activation of the TRAIL receptor DR5 and subsequent cleavage of caspase-8. Pretreatment of GRD2 cells with the specific caspase-8 inhibitor zIETD-fmk prevented GD3-induced apoptosis, indicating that this probably occurs through the activation of a death receptor. We also investigated the effects of GD3 on radiosensitivity and found that GD3 doubled the sensitivity of GRD2 cells to irradiation by γ-rays as measured by clonogenic assays. Thus, our data indicate that GD3 induction may have beneficial effects in GBM patients by activating apoptosis via the TRAIL-DR5 pathway and also by sensitizing cells to radiotherapy. This work was supported by NCI grant PO1 CA 85799.


Jeffrey J. Olson, Zhoabin Zhang, and Douglas Mattox; Emory University, Atlanta, Georgia, USA

Surgical and radiation-based therapy of spinal and intracranial schwannomas is associated with risk that might be foregone if a medical therapy were available. Based upon an anecdotal clinical report of a vestibular schwannoma becoming rapidly symptomatic after treatment with synthetic erythropoetin (EPO), a series of similar lesions were analyzed for EPO and EPO receptor (EPOR) by immunohistochemistry. Ninety-three percent of tumors demonstrated positive EPO expression. For EPOR this was 64%. Additionally, hypoxia is an important property of a variety of tumors. EPO is known to be an important hypoxia responsive element. The purpose of this work was to assess in vitro TM-31 human schwannoma cell’s (1) EPO, EPOR, and hypoxia-inducible factor 1alpha (HIF-alpha) response to hypoxia and (2) proliferation in response to EPO. Subconfluent TM-31 cells were cultured in 1% oxygen (hypoxia) or with room air (normoxia) for 5 and 24 h to assess EPO, EPOR, HIF1alpha protein, and mRNA expression. TM-31 cells were grown in normoxic conditions with escalating concentrations of human recombinant EPO (rhEPO) for 7 days. In hypoxia, the quantity of EPO message expressed increased from 5 to 24 h. Alternatively, a slight decrease in EPOR message was expressed over the same interval with hypoxia. HIF1alpha ·message in hypoxic conditions increased compared to baseline over 5 h but was little different from baseline after 24 h. EPOR protein demonstrated a possible increase in its phosphorylated form expression after 24 h of hypoxia. A marked and progressive increase in HIF1alpha protein expression during hypoxia from 5 to 24 h was noted. Proliferation as assessed by an MTT assay showed 10%, 15%, and 20% increase (all P < 0.05) over control circumstances with 0.5 U/ml of rhEPO after 1, 5, and 7 days, respectively. At higher concentrations (1–32 U/ml rhEPO), lesser degrees of proliferation stimulation were seen. We conclude that TM-31 cells accelerate proliferation in response to low doses of EPO. HIF1alpha expression increases with hypoxia as expected. The decrease in EPOR message with time of hypoxia may be correlated with increasing EPOR phosphorylation, suggesting a feedback system between these events. These in vitro observations suggest an opportunity to develop and assess a series of possible schwannoma therapies based on inhibition of EPO and EPOR function and alteration of responses to hypoxia.


Amith Panner, Mitchel Berger, and Russell O. Pieper; Department of Neurosurgery and The Brain Tumor Research Center, University of California San Francisco, San Francisco, California, USA

The tumor necrosis factor–related apoptosis-inducing ligand (TRAIL) protein is an attractive therapeutic molecule because it induces TRAIL receptor–mediated apoptosis in tumor cells, but not in normal cells. While most tumor types are TRAIL sensitive, however, GBM cell lines are largely TRAIL resistant. To address the mechanism of TRAIL resistance we examined components of the TRAIL-induced apoptotic signaling pathway in TRAIL-sensitive (U87 and U251) and TRAIL-resistant (U373 and A172) GBM cell lines. Western blot analysis of the expression of pro-apoptotic TRAIL receptors DR4 and DR5 showed no difference among the TRAIL sensitive and resistant cell lines. However, levels of the caspase-8 inhibitory, anti-apoptotic protein FLIPS were twofold higher in TRAIL-resistant A172 and U373 cells than in the TRAIL-sensitive U87 and U251 cells. In addition, retroviral infection of a construct encoding FLIPS reduced TRAIL-induced apoptosis in U87 and U251 cells by 70% compared to cultures receiving an empty vector. FLIPS levels appeared to be dependent on the Akt pathway because activation of a 4HT-inducible Akt-encoding retroviral construct resulted in a 3-fold increase in FLIPS expression in U87 cells and decreased TRAIL-induced apoptosis by 90% compared to cells not overexpressing Akt. Akt-mediated control of FLIPS levels was in turn dependent on the downstream target of Akt, mTOR. Western blot analysis showed that inhibition of mTOR by 100 nanomolar rapamycin resulted in a 30% to 45% reduction in FLIPS protein expression in A172 and U373 cells and increased TRAIL-induced apoptosis by 40% in both cell lines. Although the mechanism whereby Akt regulates FLIPS is unclear, sucrose gradient analysis of ribosomal mRNA distribution showed a 3-fold higher association of FLIPS mRNA with the polyribosomal (translated) fraction in TRAIL-resistant U373 and A172 cells compared to TRAIL-sensitive cells and showed that this association was decreased by 60% following the addition of 100 nanomolar rapamycin. These results show that the Akt pathway activation commonly seen in GBMs activates mTOR and facilitates recruitment of FLIPS mRNA to the polyribosomal fraction, which in turn increases FLIPS mRNA translation and FLIPS protein levels, resulting in a block of TRAIL-induced apoptosis. These studies suggest that mTOR activation may predict glioma sensitivity to TRAIL and that mTOR inhibition by CCI779 or equivalent may be an effective means of sensitizing even resistant GBM to TRAIL-induced apoptosis.


C. Damodar Reddy, Asha Guttapalli, Mohan Vemuri, Peter C. Adamson, Leslie N. Sutton, and Peter C. Phillips; Division of Neuro-Oncology and Neurosurgery, Department of Surgery, and Division of Pharmacology and Experimental Therapeutics, The Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania; Department of Neurosurgery, University of Pennsylvania, Philadelphia, Pennsylvania; USA

N-(4-Hydroxyphenyl) retinamide (4-HPR) a synthetic derivative of retinoic acid (RA) is a preferred anti-cancer agent in many cancers. However, its biological effects and therapeutic value in childhood brain tumor medulloblastoma (MB) has not been investigated. In this study, we report for the first time that pharmacologically achievable doses of 4-HPR induce apoptotic cell death in human MB cells. In a panel of four MB cell lines (D425, D458, D283, and D341) we show that 4-HPR (2.5–10 μM) significantly reduces the cell survival as determined by MTT assays. Further, 4-HPR at these doses inhibited anchorage-independent colony formation in soft agar, suggesting its antineoplastic activity. Caspase-3 activation in 4-HPR-treated cells suggests that the decreased cell viability was in part due to activation of caspase-dependent cell death programs. This observation is further supported by cleavage of poly (ADP-ribose) polymerase-1 (PARP-1), a caspase-3 substrate. In addition, 4-HPR-induced cell death was significantly prevented by the antioxidant L-ascorbic acid, suggesting that reactive oxygen species (ROS) in part may mediate the cell killing effects of 4-HPR. In summary, we demonstrate for the first time that pharmacologically relevant doses of 4-HPR is an effective apoptotic agent in MB cells, which suggests the potential therapeutic value in human MB.


Y. Rong, M. Pecot, J. Kadire, S.N. Devi, D. Post, D. Durden, E.G. Van Meir, and D.J. Brat; Emory University School of Medicine, Atlanta, Georgia, USA

Intravascular thrombosis and subsequent vaso-occlusion are thought to contribute to pseudopalisading necrosis during the transition from anaplastic astrocytoma (AA) to glioblastoma (GBM). Tissue factor (TF), a trans-membrane protein that forms a complex with factor VII/VIIa to initiate blood coagulation, is upregulated in malignant gliomas and could promote thrombosis. We hypothesized that PTEN loss and tumor hypoxia, both of which occur during the transition from AA to GBM, could upregulate TF expression and promote intratumoral thrombosis. To investigate this hypothesis, we examined the effect of PTEN restoration and hypoxia on TF expression and plasma coagulation by a PTEN null human glioma cell line containing a muristirone-inducible PTEN cDNA (U87MG clone 23-11). Cells exposed to hypoxia (1% O2) showed a marked increase in TF expression compared to normoxia (21% O2) by Western blot, Northern blot, and immunocytochemistry. Restoration of PTEN expression led to modest decreases in TF expression under both normoxia and hypoxia. Using tilt-tube assays of plasma coagulation, hypoxic 23-11 cells in PBS induced clotting at 52 ± 1 s (ave ± SEM), significantly faster than normoxic cells (92 ± 4 s; P < 0.001) (thromboplastin caused clotting in 18 s). A monoclonal inhibitory antibody against TF greatly prolonged clotting time induced by normoxic (432 ± 8 s) and hypoxic (284 ± 4 s) cells (P < 0.05). Clotting was similarly prolonged when we used plasma that lacked Factor VII, implicating TF-dependent mechanisms. Conditioned media from hypoxic 23-11 cells also caused shortened clotting times and contained increased TF by ELISA compared to media from normoxic cells (P < 0.05). PTEN expression led to decreased TF concentrations in conditioned media and prolonged clotting times. In 23-11 cells lacking PTEN expression, the PI3 kinase inhibitor LY294002 and the mTOR inhibitor rapamycin decreased hypoxia-induced TF expression, which suggests that PI3 kinase/Akt/mTOR signaling may facilitate TF expression and be the pathway antagonized by PTEN phosphatase activity. To examine whether TF is upregulated during the malignant progression of GBM, we investigated its expression in human astrocyte cells sequentially transformed with E6/E7/hTERT, Akt, and Ras which recapitulates astrocytoma progression. Cells transformed with Akt showed the greatest increase in hypoxia-induced TF expression and secretion. Overall, these results indicated that both PTEN loss and hypoxia are associated with increased TF expression and enhanced plasma clotting by GBM cells in vitro. Downregulation of TF expression by PTEN may be partially due to reduced Akt activation. Both hypoxia and loss of PTEN could contribute to TF-dependent intravascular thrombosis and necrosis in GBM.


Per Øystein Sakariassen,1 Lars Prestegarden,1 Jian Wang,1 A.J.A Terzis,2 Rolf Bjerkvig,1,2 and Per Øyvind Enger1; 1NorLux Neuro Oncology, Department of Biomedicine, University of Bergen, Norway; 2NorLux Neuro Oncology, Centre Recherche de Public Santé, Luxembourg

Recent work has demonstrated that many tumours contain a subpopulation of cancer cells with stem cell–like properties. These cancer stem cells have the ability to self-renew, give rise to new tumours, and also differentiate into cells with different phenotypes. In this work, we grafted biopsies from patients with GBM into the nude rat brain. Five months after implantation, sections of the rat brain showed a highly invasive tumour that displayed an angiogenesis-independent growth pattern. The malignant cells expressed the neural stem cell markers nestin and vimentin, and they showed a migratory behavior similar to that of normal human neural stem cells. Furthermore, this tumour population proliferated and could be expanded as spheroids in stem cell medium containing only basic fibroblast growth factor (bFGF) and epidermal growth factor (EGF). After implantation, these cells developed new tumours. When the cells were exposed to differentiating factors such as ciliary neurotrophic factor (CNTF) or brain-derived neurotrophic factor (BDNF) in vitro, they changed their phenotype and expressed markers associated with mature neurons and glial cells. Furthermore, the proliferation rate in these cells was lower than in the stem cell–like cells. The present findings suggest that differentiation may provide a therapeutic strategy, targeting the cancer stem cell pools in tumours.


Bodour Salhia, Frederieke Rutten, and James Rutka; Division of Neurosurgery, The Hospital for Sick Children, Toronto, Ontario, Canada

Tumour progression is the net result of an accumulation of multiple genetic lesions that affect the tumour cell phenotype. During this process, invasive and migratory properties are acquired, enabling neoplastic cells to travel significant distances away from the original tumour location. The presence of invasive cells in the tumour margin after surgical resection of malignant astrocytomas will invariably result in tumour recurrence and is a major obstacle to effective treatment. The intracellular mechanisms governing motility, while poorly understood, require modifications of the cytoskeleton. The Rho-GTPases (Rac, Cdc42, and Rho) are pivotal regulators of cytoskeletal organization and cell motility. We have shown that inhibition of ROCK, a serine/threonine effector kinase of Rho, with Y27632 or the dominant-negative mutant results in dramatic morphological and cytoskeletal alterations in astrocytoma cells. These changes are characterized by stellation, an increase in the number and length of cell processes, increased membrane ruffling, inhibition of actin polymerization, longitudinal reorientation of micro-tubules, and aggregation of intermediate filaments. In contrast to several studies demonstrating that inhibition of ROCK leads to decreased tumour cell invasiveness and motility, we found that ROCK inhibition stimulated the motility of astrocytoma cells. The mechanisms by which this occurs are currently under investigation, but the formation of membrane ruffling following inhibition suggests a possible role for Rac1 activation. Rac activation following treatment with Y27632 has also been previously reported in Swiss 3T3 cells. Whether knock-down of Rac1 using siRNA is effective in overcoming ROCK inhibition–induced motility of astrocytomas cells is currently being investigated. Elucidating the mechanisms by which ROCK inhibition alters astrocytoma morphology and motility will be useful to identifying other targets that may be of therapeutic benefit to managing glioma invasion.


H.E. Saqr, O. Omran, J. Oblinger, and A.J. Yates; Department of Pathology, The Ohio State University, Columbus, Ohio, USA

An increasing amount of evidence indicates that the disialoganglioside GD3 is critically involved in apoptosis in many cell lines. Our previous studies demonstrated that endogenous GD3 expression induced apoptosis in U1242MG glioma cells transfected with the GD3 synthase gene (U42-GRD2 cells). In this study we present further investigations on the molecular mechanisms of GD3-induced apoptosis in this cell line. We found that endogenously synthesized GD3 localizes to the caveolae of U42-GRD2, where it induced aggregation of Fas receptors. We also found that endogenous GD3 promotes the localization of death receptor 5 (DR5), Fas receptor, and tumor necrosis factor receptor-1 (TNF-R1) to the caveolae. In addition, caspase-8 was translocated to the caveolar fraction, where it was cleaved, and then the cleaved proteins were relocated into the heavy fractions. However, GD3 had no effect on the distribution of the adapter protein Fas-associated death domain (FADD). The subcellular location of cytochrome c was unaffected by GD3 expression, which suggests that GD3-mediated apoptosis is independent of the mitochondrial pathway. In conclusion, we found that GD3 increases the expression and induces the translocation of several death receptors to the caveolae. Furthermore, caspase-8 is translocated to the caveolae and cleaved as part of the molecular mechanisms of GD3-inducing apoptosis. This work was supported by NCI grant CA85799.


Mirko H.H. Schmidt,1,2 Jiuhong Yu,1 Frank B. Furnari,3 Webster K. Cavenee,3,4 Ivan Dikic,2 and Oliver Bögler1; 1William & Karen Davidson Laboratory of Brain Tumor Biology, Hermelin Brain Tumor Center, Neurosurgery, Henry Ford Hospital, Detroit, Michigan, USA; 2Institute of Biochemistry II, Goethe University Medical School, Frankfurt am Main, Germany; 3Ludwig Institute for Cancer Research, San Diego Branch, and 4Center for Molecular Genetics, Department of Medicine and Cancer Center, University of California at San Diego, La Jolla, California, USA

Abnormal receptor tyrosine kinase (RTK) signaling is a hallmark of many cancers, including gliomas. One mechanism of deregulation is loss of interaction of the receptor with the machinery that ordinarily results in its internalization and downregulation. An important component of this is the Cbl-SETA/CIN85-endophilin complex, which forms at the C-terminus of the EGFR following ligand activation and mediates internalization and ubiquitination. The glioma-associated deleted (2–7) EGFR mutant (deltaEGFR or EGFRvIII), which is phosphorylated in a ligand-independent fashion and confers enhanced tumorigenicity on glioma cells, is characterized by low intensity of signaling, which fails to promote interaction with the Cbl-SETA/CIN85-endophilin complex. We have identified a new modulator of RTK function: the SETA/CIN85-interacting protein Alix/AIP1, which also binds endophilins and is involved in membrane trafficking. Alix was found to associate with EGFR regardless of its activation state, and with delta-EGFR, and this binding did not occur via SETA/CIN85. Increasing the level of Alix expression weakened the interaction between SETA/CIN85 and Cbl proteins, reduced the tyrosine phosphorylation of c-Cbl, and diminished the ubiquitination of EGFR, SETA/CIN85, and Cbls after EGFR stimulation. This antagonism of the activity of the Cbl-SETA/CIN85 complex by Alix was reflected in its diminution of the internalization of ligand-stimulated EGFR. However, Alix was capable of increasing the level of endophilin associated with EGFR, implying that increased endophilin binding is not in itself sufficient to promote increased receptor internalization. We propose that Alix inhibits EGFR internalization by attenuating the interaction between Cbl and SETA/CIN85 and by inhibiting Cbl-mediated ubiquitination of the EGFR, and so may promote signaling through this RTK pathway.


D.L. Senger,2 A.L. Johnston,1 X.Q. Lun,2 Z.Q. Shi,2 H. Muzik,2 D. Johnson,2 S.M. Robbins,1 and P.A. Forsyth2; Departments of 1Biochemistry & Molecular Biology and 2Oncology & Clinical Neurosciences, University of Calgary, Calgary, Alberta, Canada

Malignant gliomas are highly invasive tumors that extend isolated tumor cells several centimetres away from the main tumor mass rendering them surgically incurable. The mechanisms of glioma invasion are unknown. We developed an unbiased strategy to identify novel genes involved in glioma invasion by using serial in vivo selection, isolating highly invasive and highly noninvasive subpopulations from the noninvasive human malignant glioma cell line U87. We then compared “Parental” noninvasive cells to the highly invasive “remote” cells by using cDNA microarrays. To do this we used U87 cells transfected with GFP which we implanted into the right putamen of SCID mice and allowed to grow for 35 days. Cells found within the tumor (T cells) and cells found in the contralateral hemisphere (where no tumor was grossly visible but contained a few scattered invasive/remote (R) cells) were grown in vitro under neomycin selection. After sufficient cells were obtained in culture, this procedure was repeated, with both the T and R populations, to select for increasingly noninvasive or invasive populations. The resulting noninvasive T cells, upon reimplantation into SCID mice, formed large, well-circumscribed tumors, while the highly invasive R cells formed smaller tumors with very diffuse, infiltrative edges and a highly invasive phenotype. Tumor size and invasive properties were studied. We found a remarkable change in glioma phenotype after a single passage in a SCID mouse. The well-circumscribed, noninvasive U87 remained noninvasive in vivo, but the tumors derived from “remote” cells were highly invasive and formed multiple tumor foci in vivo. Tumors from remote cells grew more slowly in vivo and showed slower generation and doubling times with increased motility in vitro. Microarray analysis identified several candidate genes including the neurotrophin receptor p75 which was expressed 7- to 12-fold higher in the R cells; differential expression was confirmed using RT-PCR and Western Blot. In addition, NGF and BDNF increased motility of p75 positive R cells (≈2.5 and >4 fold, respectively) but not p75 negative tumor cells. A strong correlation between levels of p75 (RNA and protein) and NGF-induced motility was also seen in a panel of human glioma cell lines with p75 positive U251 and SF767 migrating 2-fold faster in response to NGF. Correspondingly, overexpression of p75 in noninvasive T cells (U87p75+) resulted in an 8-fold increase in migration compared to noninvasive controls. Immunohistochemistry on seven GBM patients found p75 positive glioma cells to be present in five of the seven patients. Other genes upregulated in the highly invasive R cells include G-CSF, Aquaporin, and RhoE. We are further testing the hypothesis that these novel candidates mediate glioma invasion using siRNA. This approach identifies candidates not previously considered in this context that may mediate glioma invasion and represent potential therapeutic targets.


James R. Van Brocklyn,1 Catherine Jackson,1 Dennis K. Pearl,2 Pamela J. Snyder,1 and Thomas W. Prior1; Departments 1Pathology and 2Statistics, The Ohio State University, Columbus, Ohio, USA

The lipid kinase Sphingosine Kinase (SphK)-1 creates the bioactive lipid sphingosine-1-phosphate (S1P), which signals through G protein-coupled receptors to regulate proliferation and migration of a wide variety of cells. SphK-1 overexpression has been shown to transform fibroblasts in culture, as well as enhance tumorigenicity of breast cancer cells. We have previously shown that S1P potently stimulates mitogenesis, motility, and invasiveness of glioblastoma cells in culture. We have also shown that S1P receptor expression is common in human glioblastoma tissue. In this study, we have investigated the role of SphK-1 in these tumors. High SphK-1 expression in human glioblastomas was found to correlate with poor prognosis. Examination of survival, accounting for patient age, in 48 glioblastoma patients revealed that those whose tumors expressed high levels of SphK-1 had a median survival of 102 days, while those with low levels of SphK-1 expression had a median survival of 352 days. To examine the biological role played by SphK-1 in glioblastoma cells, we used RNA interference to knock down SphK-1 expression in glioma cell lines expressing high basal levels of SphK-1. U-87 MG cells were found to express high levels of SphK-1, while M059K cells expressed intermediate levels, and U-373 MG cells had very low expression. SphK-1 knockdown decreased growth rate in U-87 MG cells, while almost completely blocking growth of M059K cells. In addition, motility of M059K cells was inhibited by knockdown of SphK-1; however, motility of U-87 MG cells was not affected. Interestingly, RNA interference for SphK-1 in U-87 MG cells decreased expression by approximately 50%, to about the same level as that seen in untransfected M059K cells. Thus, SphK-1 expression is important for maximum proliferation of glioma cells, while moderate levels of SphK-1 expression are sufficient to enhance motility of glioma cells. Combined with the prognostic significance of SphK-1 expression, these data indicate that SphK-1 may be an important factor controlling malignant behavior of glioblastoma cells. This work was supported by grant #R01 NS41517 from NINDS.


John Wiencke,1 Shichun Zheng,1 Margaret Wrensch,1 Mitchel Berger,1 Daphne Haas-Kogan,2 and David Stokoe2; 1Department of Neurological Surgery and 2Comprehensive Cancer Center, University of California San Francisco, San Francisco, California, USA

PTEN is a tumor suppressor gene whose activity is commonly disrupted in gliomas; loss of expression is sometimes but not consistently associated with deletion or mutation. Epigenetic mechanisms may suppress PTEN expression; however, a highly homologous transcribed pseudogene can easily be confused with the authentic PTEN and also contains a CpG rich 5′-region that can be aberrantly methylated. We used methylation-specific PCR methods to assess PTEN and psi-PTEN methylation within a series of 52 fresh frozen gliomas (23 de novo glioblastoma [GBM], 18 oligoastrocytoma [OA], 11 astrocytoma) and in 13 normal blood lymphocyte controls. Another locus (FHIT; 3p14) was also examined for comparison. AKT phosphorylation was also assessed. Only two GBMs contained PTEN methylation (8.7%), compared with 11 (61%) within the oligoastrocytoma group and five of the astrocytoma (45.5%), and none of the control lymphocytes were methylated. The methylation of psi-PTEN did not correlate with tumor grade. In contrast, FHIT was found commonly methylated in GBMs but not in OAs or astrocytomas. AKT phosphorylation was lower among the lower grade tumors; we are currently assessing the relationship of PTEN methylation with phosphorylation. PTEN methylation is common among OA and astrocytoma but not among high-grade glioma and should be investigated as a marker of secondary GBM development. This work was supported by P50CA097257.


B. Mark Woerner, Nicole Warrington, Biplab Dasgupta, Wen Li, Robyn S. Klein, Arie Perry, David H. Gutmann, and Joshua B. Rubin; Washington University School of Medicine, St. Louis, Missouri, USA

The molecular basis for the curious predilection of glioma formation along the optic pathway in children with neurofibromatosis 1 (NF1) is unclear. In this study, we demonstrate that the chemokine receptor CXCR4 is expressed, along with its ligand CXCL12, in sporadic and NF1-associated optic pathway gliomas. Three of four NF1-associated optic pathway gliomas demonstrated punctate cytoplasmic and membranous CXCR4 staining. In each of these tumors, CXCL12 was expressed in tumor-associated blood vessels as well as in scattered tumor cells. Two sporadic optic pathway gliomas from patients without NF1 also exhibited this pattern of staining. In contrast, two sporadic posterior fossa pilocytic astrocytomas were positive for CXCR4, but negative for CXCL12 staining. The differences in CXCL12 staining in pilocytic astrocytomas of the optic pathway were compared to those of the cerebellum parallel differences in CXCL12 expression encountered in the normal adult brain. Beyond the post-natal period, CXCL12 is expressed in multiple regions of the brain including the retina, the suprachiasmatic hypothalamus, and the tegmentum, but not in the cerebellum. To determine the relevance of CXCR4 activation to NF1-associated tumor formation, we studied the growth of Nf1−/− mouse astrocytes in response to CXCL12 treatment in vitro. Exposure to CXCL12 resulted in a significant growth advantage for Nf1−/− astrocytes compared to wild-type control astrocytes. In these experiments, wild-type astrocytes grown in glial-defined media exhibit a decrease in cell proliferation in response to CXCL12, while Nf1/−astrocytes responded to CXCL12 with an increase in cell number as a function of both increased proliferation and decreased apoptosis. This growth effect correlated with changes in intracellular signaling in Nf1/− astrocytes in response to CXCL12 that were characterized by a sustained intracellular calcium flux and decreased cAMP generation compared to wild-type astrocytes. Collectively, these studies suggest that CXCR4 activation could provide an anatomically restricted, growth-promoting signal that potentially explains the predilection of NF1 patients to form gliomas along the optic pathway. This is reminiscent of CXCR4 function during brain development and highlights the importance of stromal signals in determining the pattern of tumor formation in the brain. Finally, the effect of CXCR4 activation on Nf1−/− astrocyte proliferation and survival may provide novel insights into the molecular pathogenesis of this common childhood tumor in individuals with NF1.


S. Xia and J. Laterra; The Johns Hopkins School of Medicine, Baltimore, Maryland; The Kennedy Krieger Research Institute, Baltimore, Maryland, USA

A prominent feature of glioblastoma is its relative resistance to cytotoxic chemotherapy and ionizing radiation. Elucidating mechanisms by which glioblastoma resists cytotoxic death and developing strategies that enhance glioblastoma apoptosis could have a major impact on therapeutic outcomes. In this study, we explored the possibility of modulating the susceptibility of glioblastoma cells to death receptor activation with chemotherapeutic agents. Sublethal concentrations of topoisomerase inhibitor camptothecin (CPT) reduced the IC50 of agonist anti-Fas antibody 10-fold, from 500 ng/ml to 50 ng/ml, in human U87 glioblastoma cells (p53-wt). Cell death quantified by MTT in response to CPT alone (50 μM), anti-Fas alone (50 ng/ml), and combination CPT+anti-Fas was found to be 16%, 15%, and 53%, respectively. A similar pattern of relative cytotoxicity was found in U373 cells (p53-mutant). Thus, treatment with CPT significantly increased the glioblastoma cell death response to Fas receptor activation regardless of p53 status. We examined the death pathways and mechanisms involved in this apparent synergistic cytotoxic response. Cell death was found to be predominantly apoptotic involving both extrinsic and intrinsic pathways as evidenced by annexin V staining, cleavage of multiple caspases (3, 8, and 9), increased caspase 3 and 8 enzyme activity, Smac release, and cytoprotection by inhibitors of caspase-8 and caspase-9 (53% and 45% protection, respectively). The expression of FADD, and not Fas, FasL, or caspase 8 protein, increased following cell treatment with both CPT and anti-Fas but not in response to either agent alone. C-Jun-NH2-terminal kinase (JNK) was activated from 1 to 9 h, with maximal activation (3-fold) at 6 h specifically in response to combination CPT and anti-Fas. Pretreating glioblastoma cells with the JNK agonist anisomycin (1 μM, 30 min) in the absence of CPT significantly enhanced the cytotoxic effect of anti-Fas antibody (50 ng/ml) from 11% to 48% cell death. These findings demonstrate that CPT and possibly other topoisomerase inhibitors synergise with Fas activation to induced glioblastoma apoptosis and implicate a mechanism involving the JNK/stress-activated protein kinase (JNK/SAPK) pathway.


Nicholas Young and James R. Van Brocklyn; Integrated Biomedical Science Graduate Program and Department of Pathology, The Ohio State University, Columbus, Ohio, USA

The bioactive lipid signaling molecule sphingosine-1-phosphate (S1P) has previously been found to regulate cellular proliferation, migration, survival, and angiogenesis, through acting on its target G-protein coupled receptors S1P1-5. We recently showed that S1P potently stimulates proliferation and invasiveness of glioblastoma cells. In addition, high expression of the enzyme which forms S1P, sphingosine kinase-1, correlates with poor prognosis in glioblastoma patients. In this study, we explored the receptors responsible for these S1P-induced, cancer-associated effects on glioma cells in vitro. The glioma cell lines U-373 MG and U-118 MG are well-established and characterized models that express endogenously higher and lower levels of S1P receptors, respectively. These cell lines were used to explore the effects of S1P1, S1P2, S1P3, and S1P5 on glioma cell growth, migration, and invasion. The high levels of S1P receptor in U-373 MG cells were decreased through the use of siRNA technology, while overexpression of S1P receptors in U-118 MG cells was achieved through gene transfection. S1P1 and S1P3 were found to be critical in mediating S1P-induced proliferation, migration, and invasion. S1P2 positively influenced invasion and growth, but negatively regulated migration. S1P5 overexpression decreased cellular activity in all of the areas investigated independent of S1P presence, which could ultimately be due to constitutive receptor activation upon overexpression. These receptors, which are commonly expressed in human gliomas, signal through various G-proteins to regulate diverse signaling pathways and biological responses. By examining the downstream cellular effects individually regulated by each of the endogenously expressed S1P receptors in glioma cell lines, this study begins to define the roles of S1P and its individual receptors in glioma pathobiology. This work was supported by grant #R01 NS41517 from NINDS.


Gary Zhai,1 Yoshiyuki Suzuki,1 Sormeh Sarkesh,1 Peter McL. Black,2 Alona Muzikansky,3 Jay S. Loeffler,1 and Arnab Chakravarti1; Departments of 1Radiation Oncology and 3Biostatistics, Massachusetts General Hospital/Harvard Medical School, Boston, Massachusetts; 2Department of Neurosurgery, Brigham and Women Hospital/Harvard Medical School, Boston, Massachusetts; USA

Preclinical data suggests that activation of the phosphatidylinositol 3-kinase (PI3K) pathway is a common event in glioblastoma and is associated with resistance to radiation and chemotherapy-induced cell death. Data from clinical specimens is, however, more limited in this regard. The key objectives of this study were to examine activation patterns of the PI3K pathway in gliomas of various grades and to examine the prognostic significance of PI3K pathway activation using snap-frozen clinical specimens. Ninety-two frozen glioma tissues of various histologies were prospectively collected through an IRB-approved protocol at Brigham and Women’s Hospital. Levels of expression of PI3K pathway members were assessed through quantitative Western analysis by using total and phospho-specific antibodies for PI3K, AKT, and p70s6k. Both expression and expression levels of total and activated isoforms of these PI3K pathway members were correlated with histology, markers of apoptosis (cleaved caspase 3), and clinical outcome (e.g., overall survival). It was found that activation of all three PI3K pathway members was significantly more frequent in GBM than in non-GBM tumors. Levels of phospho-PI3K, phospho-AKT, and phospho-p70s6k were all found to be inversely associated with cleaved caspase 3 levels, suggesting PI3K pathway activation is associated with reduced levels of apoptosis in clinical specimens, consistent with data reported from preclinical studies. Perhaps most importantly, activation of PI3K pathway members was found to be significantly associated with reduced survival times when all glioma cases were considered in aggregate. When glioblastoma cases were considered separately, the prognostic value of PI3K activation remained significant, suggesting that PI3K activation may directly be associated with radiation resistance, given that this was the only adjuvant therapy administered to this subset of patients. Activation of the PI3K pathway is significantly associated with increasing tumor grade, decreased levels of apoptosis, and adverse clinical outcome in human gliomas. Molecular pathways regulating PI3K activation would appear to be promising targets in the clinical management of glioma patients.


Yi Zhan and Donald M. O’Rourke; Department of Neurosurgery, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA

Since glioblastoma cells and primary glioblastoma tumors often contain mutations in EGFR and other signaling molecules, it is important to determine mechanisms of inhibition of EGFR signaling for the design of improved molecular therapeutic approaches. Our focus has been to independently characterize the effects of EGFR kinase inhibition and inhibition of the Shp2 protein tyrosine phosphatase (PTP) function on glioblastoma growth and transformation. EGFR has been shown in many systems to require Shp2 for downstream signaling to induce cell growth, motility, and transformation. In our work, we have shown that two distinct inhibitory strategies both yield inhibition of glioblastoma transformation by PTEN-independent mechanisms. Expression of kinase deficient erbB mutants in wild-type p53 U87MG and mutated p53 U373MG cells resulted in the formation of inactive erbB heterodimers and reduced cell growth and transformation. Further studies demonstrated that erbB-inhibited glioblastoma cells expressing the kinase-impaired erbB mutants exhibited a higher G0/G1 population than parental cells. Biochemical experiments showed that the tyrosine phosphorylation of the Shp2 substrate, Gab1, and Akt were decreased in response to EGF ligand treatment. Interestingly, glioblastoma cells expressing disabled erbB receptors were more sensitive to radiation treatment, independent of p53 or PTEN status. Phenotypic inhibition of glioblastoma growth more strongly correlated with a diminution of PI3K/Akt activation than Erk/MAPK activation. Conversely, inhibition of the PTPase activity of Shp2 (Shp2C459S mutant) inhibited cell growth and transformation mediated by oncogenic activation of EGFR in soft agar and nude mice. Consistent with these results, Shp2C459S cells displayed a reduction in Erk MAPK activity. However, in contrast to EGFR kinase inhibition, inhibition of growth resulting from Shp2 PTPase inhibition caused an increase in the G2/M population. Moreover, Gab1 and Akt tyrosine phosphorylation was upregulated in growth-inhibited Shp2C459S glioblastoma clones. Shp2C459S cells also showed decreased cell death in response to cisplatin treatment. Our results suggest that perturbation of EGFR signaling pathways at different levels may result in similar biological effects, but through different mechanisms. Collectively, two distinct mechanisms of EGFR signal attenuation are revealed by these data, and both mechanisms are independent of PTEN function. Molecular therapies that modulate both signaling mechanisms may result in enhanced inhibition of glioblastoma growth and transformation.


Jing Zhang and V. Wee Yong; Departments of Oncology and Clinical Neuroscience, University of Calgary, Calgary, Alberta, Canada

Chemokines have been found to alter tumor growth and metastasis. We have previously described that a particular chemokine receptor, CXCR4, was predominantly expressed in various glioma cell lines and in resected glioblastoma specimens (Zhou et al., 2002, J. Biol. Chem. 277:49481). As the only known ligand of CXCR4 is stromal cell–derived factor-1 (SDF-1, CXCL12), we tested the response of two CXCR4-bearing human glioma cell lines, LN827 and U373, to SDF-1alpha. We found, by real-time PCR, that 100 ng/ml SDF-1alpha increased the expression of transcripts encoding membrane type-2 matrix metalloproteinase (MT2-MMP) (P < 0.01) in both cell lines. This was selective in that the expression level of MT1-, MT3-, and MT5-MMP did not change in response to SDF-1alpha. Similarly, two growth factors of relevance to glioma cells, VEGF and HGF, did not vary when treated with 100 ng/ml SDF-1alpha. Flow cytometry and Western blot analyses confirmed that SDF-1alpha upregulated MT2-MMP protein expression. Functionally, invasion assays indicated that SDF-1alpha stimulated LN827 cells to transmigrate through Matrigel-coated chambers and that this effect was inhibited in glioma cells by the stable downregulation of MT2-MMP expression using an RNA interference (RNAi) approach. Ongoing experiments address the requirement of MT2-MMP in glioma growth in vivo by implanting MT2-MMP downregulated clones into mouse brain. Our results have revealed an important downstream effector of SDF-1-CXCR4 interaction in glioma cells that regulates invasiveness: MT2-MMP. We propose that therapies aimed at CXCR4 and MT2-MMP may prove to be of utility for this incurable cancer.


Leying Zhang, Jill M. Schartner, Aaron Hagar, and Behnam Badie; Department of Neurological Surgery, University of Wisconsin Hospital and Clinics, Madison, Wisconsin, USA

Stimulation of tumor-associated microglia (MG) and macrophages (MP) may be an important strategy in enhancing the CNS immune response against malignant brain tumors. Using animal glioma models, we recently reported that MHC-II induction in response to activators such as CpG, LPS and IFN-γ to be muted in tumor MG/MP as compared to normal MG/MP. This observation suggested that suppressive factors intrinsic to the tumor environment may regulate the expression of MHC-II in MG/MP. To better understand this process, we investigated possible changes in transcription factors involved in IFN-γ-induced regulation of MHC class II expression in tumor MG/MP. Nuclear extracts from C6 gliomas, C6-associated MG/MP, normal brain, and normal MG/MP were obtained and studied using an electrophoretic mobility shift assay (EMSA). Among nuclear factors studied (AP-1, IRF, USF-1 and Stat-1) only USF-1, which is constitutively expressed in most cells, was down-regulated in tumor-associated MG/MP, but not normal MG/MP. Interestingly, when tumor MG/MP were isolated and cultured in the absence of tumor cells, USF-1 expression returned to baseline. Considering that USF-1/Stat-1 interaction has been shown to be important in the activation of MHC class II transactivator CIITA, these observations imply possible mechanism by which MHC II is regulated in glioma-associated MG/MP. Understanding the mechanism by which USF-1 is downregulated may help design strategies to further enhance the immune activity of MG/MP in brain tumors.


Yi-Hong Zhou,1 Yue Teng,1 Xiaosong Wu,1 T.J. Liu,2 Frederick Lang,3 and W.K. Alfred Yung2; 1Department of Neurobiology and Developmental Science, University of Arkansas for Medical Science, Little Rock, Arkansas; Departments of 2Neuro-Oncology and 3Neurosurgery, The University of Texas M.D. Anderson Cancer Center, Houston, Texas; USA

Glioblastoma multiforme (GBM) is the most malignant primary brain tumor. The complete or partial loss of chromosome 10 is found in as many as 80% of GBMs, and the loss of genes on chromosome 10 is a critical step in the malignant progression typical of gliomas. We demonstrated that PAX6, a developmental gene with a putative suppressive function of malignant progression in gliomas, was significantly downregulated in GBM and that a low level of PAX6 expression is a harbinger of an unfavorable prognosis for patients with malignant astrocytic glioma. Our preliminary data showed that the downregulated expression of PAX6 in GBM could result from the loss of unidentified tumor suppressor gene(s) on chromosome 10. Using Affymetrix U133A chip hybridization, real-time qRT-PCR, Western hybridization, ELISA, and a zymography assay, we screened and verified alternations in gene expression resulting from the expression of exogenous PAX6 in the U251HF glioma cell line in vitro and in vivo induced either by stable transfection or infection with a PAX6-expressing adenovirus. We found that PAX6 suppressed the expression of multiple genes known to be involved in the malignant progression of brain tumors, such as IGFBP2, MMP2, MCP-1, PDGFRα, and VEGF. Additionally, PAX6 may directly upregulate the expression of EFEMP1, which encodes an EFG-containing, fibulin-like extra-cellular matrix protein. We observed a 3-fold increase and decrease (77%) of EFEMP1 expression at the RNA level in U251HF GBM cells 24 h after infection with a recombinant adenovirus expressing PAX6 and a dominant negative form of PAX6 (−344), respectively. The function of EFEMP1 in glioma is under investigation. The lower expression of PAX6 in five GBMs compared with adjacent normal tissues correlates with higher expressions of IGFBP2, MMP2, and VEGF, and a lower expression of EFEMP1. In contrast, no marked difference at the RNA level for the expression of PDGFRα in GBM was observed, and the expression of MCP-1 was changed in different directions in the GBMs versus adjacent normal tissues. In conclusion, our data indicates that the mechanism for suppression function of PAX6 may be suppressing invasion and angiogenesis processes involving expression of IGFBP2, MMP2, and VEGF.



Jill S. Barnholtz-Sloan, Richard K. Severson, Bonita Stanton, Merlin Hamre, and Andrew E. Sloan; Departments of Internal Medicine, Neurosurgery, and Family Medicine and Pediatrics, Wayne State University School of Medicine, Detroit, Michigan; Karmanos Cancer Institute, Detroit, Michigan; USA

The object of this study was to assess racial differences in incidence rates and survival of children with brain tumors. Subjects identified through the SEER Program were 3326 children, ≤19 years old at diagnosis, newly diagnosed between 1973 and 1995 with primary, malignant brain tumors. Chi-square tests were used to evaluate prognostic variables by race. Age-adjusted incidence rates were calculated by race. Kaplan-Meier models were used to assess racial differences in overall survival, and Cox proportion hazards models were used to examine racial differences by interval of survival, adjusting for prognostic factors. The distribution of gender and age at diagnosis was similar by race. The majority of children had infratentorial tumors regardless of race. Children of Other race had the highest proportions of medulloblastoma/PNET tumors and both surgery and radiation treatment. African-American children had the highest proportions of astrocytoma tumors, grade 3 tumors, and no treatment or radiation-only treatment. Caucasian children had a significantly higher incidence of brain tumors. There were no significant differences in crude survival by race. However, for children surviving at least 24 months, African-Americans had an increased risk of death compared to Caucasians. For children surviving at least 60 months, African-Americans and Others had decreased risks of death compared to Caucasians. For children surviving beyond 60 months, African-Americans had a significantly decreased risk of death compared to Caucasians. We conclude that important racial differences in incidence and survival of children with brain tumors existed, particularly when focusing on specific time intervals after diagnosis.


Svetlana Blitshteyn, Julia Crook, and Kurt A. Jaeckle; Department of Neurology, Mayo Clinic Jacksonville, Mayo College of Medicine, Jacksonville, Florida, USA

Meningioma has been established as a possible hormone-sensitive tumor, with approximately 70% of meningiomas expressing progesterone receptors and 30% expressing estrogen receptors. In vitro, proliferation of meningioma cell lines in culture after exposure to estrogens has been observed. There has been much speculation by physicians and patients that hormone replacement therapy (HRT) may be a risk factor for meningioma. However, the relationship between HRT administration and meningioma has not been well studied. The goal of the study is to determine whether an association exists between a diagnosis of meningioma and either current or past use of HRT in female patients. A retrospective study using the Mayo Clinic Jacksonville patient database between 1993 and 2003 identified a total of 398,889 female patients (age range, 26–86), 17,907 of whom had either current or past use of HRT. There were a total of 1410 women with a diagnosis of symptomatic or incidentally discovered meningiomas, and 156 of these women were confirmed to have either current or past use of HRT. A multivariate analysis, adjusted for age, demonstrated a positive association between a diagnosis of meningioma and the use of HRT (odds ratio, 2.2; 95% CI, 1.9–2.6, P < 0.0001). The frequency of meningiomas among women with the use of HRT was 156/17,907, corresponding to a calculated prevalence of 870/100,000, and the frequency of meningiomas in women without the use of HRT was 1,254/380,982, corresponding to a calculated prevalence of 330/100,000. The higher frequency of meningiomas among women with either current or past use of HRT compared to women who were not on HRT suggests that the use of HRT may be a risk factor for meningioma.


D.T. Blumenthal, D.S. Keith, and L.A. Cannon-Albright; Neuro-Oncology, Genetic Epidemiology, and Department of Bioengineering, University of Utah and Huntsman Cancer Institute, Salt Lake City, Utah, USA

The familiality of brain tumors has not been well documented. The Utah Population Database (UPDB) is a unique resource which contains pedigrees of over 2.5 million Utah individuals linked to over 80,000 cancer registry records for the entire state, including over 1500 individuals with brain tumors. We have previously shown that first-degree relatives of individuals with brain tumor are at increased risk of brain tumor, and that individuals with brain tumor show a higher average relatedness than expected in this population. Additionally, subgroup analysis of familiality of the 571 glioblastoma cases and the 364 astrocytoma cases shows significantly elevated relative risk and familiality for the astrocytoma cases, but not for the glioblastoma cases. Multiple descending pedigrees have been identified with an excess of brain tumor cases among the descendants of founder pairs. The genetic component to brain tumors is described, and a set of unique lineages informative for predisposition gene identification will be presented.


Robert Cavaliere and David Schiff; University of Virginia, Charlottesville, Virginia, USA

At present, the growing demand for organ donation greatly exceeds supply. The resulting shortage has led to the use of “marginal donors.” Although cancer patients in general are excluded as organ donors, patients with primary brain tumors (PBTs) are permitted to donate. The use of donor organs from patients with PBTs is controversial. Although extraneural metastases (ENMs) from PBTs are rare, donor transmission of these malignancies to organ recipients have been reported. The rarity of this phenomenon, the incomplete reporting of donor history, and the limited understanding of brain tumor classification and management have confounded the transplantation literature. Recent recommendations (Buell et al., 2003, Transplantation 76:340) were that “Organs from donors with CNS tumors can be used with a low risk of donor tumor transmission in the absence of the following risk factors: high-grade tumors, ventriculoperitoneal or ventriculoatrial shunts, prior craniotomy, and systemic chemotherapy.” We critically review the literature on ENMs of PBTs and donor transmission of PBTs. In some cases, surgery and cerebrospinal fluid shunting appear to contribute to systemic dissemination. Often, however, the coexistence of ENMs and these interventions appears coincidental. Contrary to some published recommendations, in the absence of adenopathy or masses in the scalp, abdomen, or pelvis, previous craniotomy or shunt is insufficient to exclude a patient from organ donation. Nor does the duration of PBT correlate with increased risk of ENM or transmission of malignancy to organ recipient. No evidence suggests patients receiving chemotherapy for PBT carry increased risk of developing ENM or transmitting tumor in a donated organ. Medulloblastoma, ependymoma, hemangiopericytoma, and anaplastic meningioma have sufficiently increased metastatic potential to warrant patient exclusion from donation. In contrast, glioblastomas and other tumors of astrocytic and oligodendroglial origin, subependymomas, and low-grade meningiomas rarely metastasize. At the time of harvesting, the donor’s head, chest, abdomen, pelvis, and shunt tubing should be meticulously explored. In the absence of systemic pathological findings at the time of organ harvesting, donors with these tumors represent a low risk of transferring malignancy. Potential recipients of organs from such donors should be notified preoperatively of the small but serious risk of PBT transmission.


Susan M. Chang1 and Fred G. Barker2; 1University of California at San Francisco, San Francisco, California; 2Massachusetts General Hospital, Boston, Massachusetts; USA

Social factors influence treatment choices in cancer, often with consequences for patient survival. We studied the interrelations between marital status, received treatment, and survival in patients with glioblastoma multiforme (GM) using population-based data. The data source was the SEER Public-Use Database, 1988–2000, 2003 Release, all registries. Multivariate logistic and ordinal logistic regression and Cox proportional hazard regression were used to study risk factors. There were 10,088 adults (age > 17) with new histologically confirmed diagnoses of supratentorial GM who were actively followed. Of these, 68% were married (M), 30% not-married (NM; included single, divorced, and widowed), and 2% unknown. Tumor size at diagnosis was slightly larger in unmarried patients (51% < 4.5 cm NM, vs. 55% M), even after adjustment for age, sex, race, poverty level in county of residence, year of treatment, and tumor site (P = 0.001). Unmarried patients more frequently had biopsy rather than surgical resection (26% NM vs. 23% M) and were less likely to receive radiation treatment after surgery (26% NM vs. 19% M received no radiation). In a multivariate analysis adjusted for age, sex, race, poverty level in county of residence, tumor site, and year of treatment, the odds ratio (OR) for biopsy rather than resection in unmarried patients was 1.19 (95% CI, 1.07–1.33, P = 0.002), with little change in OR with further adjustment for tumor size in a subset of 5500 patients in whom this was recorded (OR 1.15, P = 0.07). Adjusted for age, sex, race, poverty level in county of residence, tumor site, year of treatment, and biopsy rather than resection, the OR for receiving no postoperative radiation was 1.45 for unmarried patients (95% CI, 1.30 to 1.62, P < 0.001). Unmarried patients were more likely than married patients to refuse both resective surgery (0.7% NM refused vs. 0.4% M) and postoperative radiation (2.7% NM refused vs. 1.5% M). Unmarried patients had shorter survival than married patients (median 6 months NM, 7 months M). In multivariate analysis corrected for age, sex, race, poverty level in county of residence, tumor site and year of treatment, the hazard ratio for unmarried status was 1.14 (95% CI, 1.09–1.19, P < 0.001). In a subgroup of 2496 patients with lobar tumors, all of whom received postoperative radiation, in multivariate analysis adjusted for age, tumor site and size, and extent of resection, the hazard ratio for unmarried status was 1.18 (95% CI, 1.07–1.29: P < 0.001). Marital status is correlated with both received treatment and survival in adult patients with supratentorial GM. Unmarried GM patients present with larger tumors and are less likely to receive both surgical resection and postoperative radiation. More frequent refusal of recommended therapies by unmarried patients may contribute to the observed disparities in treatment. Survival after diagnosis is shorter in unmarried GM patients than in married patients who receive equivalent surgical and radiation treatment, even after adjustment for other known prognostic factors.


Prabal Deb,1 Chitra Sarkar,1 Mehar Chand Sharma,1 Rajbir Singh,2 Vani Santosh,3 and S.K. Shankar3; Department of 1Pathology and 2Biostatistics, All India Institute of Medical Sciences, New Delhi; 3Department of Neuropathology, National Institute of Mental Health and Neuro Sciences, Bangalore; India

Over the last two decades, an increase in the incidence of primary central nervous system lymphoma (PCNSL) cases has been reported in the West, both among immunosuppressed and immunocompetent patients. No such reports are available from India. Hence, this multicentric study was undertaking involving two large neurosurgical referral centres, one in Northern India (AIIMS, New Delhi) and another in Southern India (NIMHANS, Bangalore), to evaluate the incidence of PCNSL cases in India. All biopsy-proven PCNSL cases obtained from the neurosurgical databases of these two hospitals from the period 1980 to 2003 were reviewed. Immunophenotyping was done, and relevant clinical details were collected. Appropriate statistical analysis was done to assess any change in trend of incidence or age at presentation. The total number of PCNSL cases in this study was 116 at AIIMS and 70 at NIMHANS, which constituted 0.95% and 0.92% of the total intracranial neoplasms, respectively. Frontal lobe was the most common site of involvement. A majority of the cases were diffuse, high grade, large cell lymphoma, B-cell immunophenotype. Unlike in the West, statistical analysis to evaluate any change in trend, either of incidence or of age at presentation, over the study period did not reveal any significant change. The mean age for cases diagnosed at AIIMS and NIMHANS was 44.4 years and 39.5 years, respectively, which was a decade younger than in the West. Only about 15.1% of all PCNSL cases in our study were in the elderly age group (>60 years). There was no statistically significant increase in incidence in this age group as well, which was unlike reports from the West. Further, in contrast to the West, all cases in this study occurred in immunocompetent patients, except one case of HIV positive at NIMHANS and one case of renal transplant at AIIMS. Thus, this multicentric hospital-based study did not reveal any increase in incidence of PCNSL cases in India over the past 24 years. Although not a population-based study, it reports hospital-based data from two of the largest neurosciences referral centres of the Indian subcontinent and further reconfirms reports that the rise of PCNSL is not a universal phenomenon, but possibly confined to certain geographical areas only. The differences in age distribution of PCNSL cases between India and the West are probably related to the population demographics. The life expectancy and >60-year population in the Indian subcontinent is around 62.5 years and 7%, respectively, as compared to 77 years and 16% in the United States. The low association of PCNSL with HIV/AIDS in India is possibly due to early death in AIDS on account of opportunistic infections.


Herbert H. Engelhard III, Kimberly Porter, and Frederick G. Barker II; University of Illinois at Chicago, Chicago, Ill.; American College of Surgeons, Chicago, Illinois; Harvard University, Boston, Massachusetts; USA

Extensive data on 9241 patients diagnosed to have a primary brain tumor during the year 2000 have been collected in a comprehensive study conducted by the Commission on Cancer of the American College of Surgeons. The original objectives of this study included determining the demographics, symptoms and signs, pathological distribution, and methods of diagnosis and treatment for these patients. Data from the study are still being analyzed. This presentation summarizes results related to patient clinical presentation and treatment. Questionnaires were sent to all U.S. hospitals having a cancer registry. Six hundred four hospitals voluntarily contributed data, which were collected via electronic record submissions directly to the National Cancer Data Base and/or on scannable paper forms. Eligible CNS cases were selected from those assigned ICDO-2 topography codes C70.0–C70.9, C71.0–C71.9, C72.0–C72.9, and C75.1–C75.3. Brain tumor patients were extracted from this larger (CNS) data set. Data were analyzed with SPSS, a standard statistical software package. For the 9241 brain tumor patients, headache was the most common symptom, followed by weakness, difficulty with coordination and/or balance, visual and/or speech disturbance, altered alertness, memory loss, and many other symptoms. The frontal lobe was the most common tumor location (18.9% of patients). Multiple primary CNS tumors were present in 8.9%. The initial Karnofsky rating was ≥80 in 52.8%. Glioblastoma was the most common histological type, followed by meningioma, and then other astrocytomas. Only 6.5% of patients participated in clinical protocols. A total of 5031 patients underwent craniotomy for resection, with or without image guidance, BCNU wafers were implanted 227 times, 3270 patients had open or stereotactic biopsy, and 541 patients had ventriculostomy or a ventricular shunt. Hemorrhage at the operative site (with or without clinical significance) occurred in 3.4% of patients; infection was noted in 1.9%. A total of 452 patients were treated with radiosurgery, 4234 received radiation therapy, and 1963 received chemotherapy (with either single or multiple agents). This study is the largest comprehensive single-year study of brain tumor patients ever conducted. Data from the study are deemed suitable for benchmarking, describing prevailing patterns of care, and generating additional hypotheses for future studies. Further analysis of this data should prove useful, as it is broken down according to factors such as age, ethnicity, tumor histology, residual tumor size, and hospital size. Information related to tumor recurrence and patient survival will continue to be tracked.


Serap Erdal,1 Darell D. Bigner,2 and Faith G. Davis3; 1Environmental and Occupational Health Sciences, School of Public Health, University of Illinois at Chicago, Chicago, Illinois; 2Duke Comprehensive Cancer Center, Duke University Medical Center, Durham, North Carolina; 3Biostatistics and Epidemiology, School of Public Health, University of Illinois at Chicago, Chicago, Illinois; USA

Absence of associations and conflicting results in epidemiological investigations of brain cancer to date may be due to inadequate exposure assessment tools and poor characterization of human exposures to animal neurocarcinogens in occupational and environmental settings. In an effort to reduce this significant data gap, we have undertaken a comprehensive effort to better understand significance and extent of dermal exposure to known and suspected animal neurocarcinogens. Humans, in general, come into contact with chemicals via inhalation, ingestion, and skin absorption pathways. Although our understanding of inhalation and ingestion exposures to many compounds present in environmental and occupational settings have improved over time, there is often scant data available for dermal exposure of chemicals. The proper exposure metric to be used in deriving or exploring exposure-response relationships in epidemiological investigations should be the cumulative exposure (or dose). This requires that information on dermal exposure should be sought. In our attempts to improve the human exposure to known and suspected animal neurocarcinogens, we have theoretically estimated dermal permeability coefficient (Kp) and the dermally absorbed dose per event (DA-event) for about 50 animal neurocarcinogens. This was accomplished using the DERMWIN software program, which is one of the components of the Estimation Program Interface (EPI) suite of physical property and environmental fate estimation models developed by the U.S. Environmental Protection Agency’s Office of Pollution Prevention and Toxics and Syracuse Research Corporation. The estimated dermally absorbed dose in the units of mg/cm2-event varies from 10−1 to 10−13. While propylene imine, acrylamide, acrylonitrile, ethylene oxide, glycidol, diethyl hydrazine, chloroethane, methyl methane sulfonate, and vinyl chloride have more ability to penetrate the skin upon dermal contact, the skin is quite impermeable to other animal neurocarcinogenic compounds. These include azodyes (i.e., CI Direct Blue 15, 3,3,-dimethylbenzidine dihydrochloride, and 3,3-dimethoxybenzidine dihydrochloride), polynuclear aromatic hydrocarbons (benzo[a]pyrene, 3-methyl-choanthrene, and 7,12 dimethylbenz[a]anthracene), and fluorenylacetamides. In addition, some of the animal neurocarcinogens used as the pharmaceutical drugs (i.e., furosemide and procarbazine hydrochloride) are found to be not of concern for dermal exposure. The results show a wide spectrum of dermal permeability to animal neurocarcinogens, from very permeable to impermeable, indicating the importance of dermal exposure to specific animal neurocarcinogens by humans.


Maria Teresa Giordana,1 Paolo Ghiglione,1 Piero Meineri,2 Cristina Mocellini,2 Pier Paolo Panciani,1 Antonella Tribolo,3 and Adriano Chiò1; 1Department of Neuroscience, University of Turin; 2Neurology Unit, S. Croce Hospital, Cuneo; 3Neurologic Clinic, University of Novara; Italy

There is a considerable controversy regarding whether incidence of primary cerebral lymphomas (PCNSL) is increasing in immunocompetent patients. The present study was aimed at ascertaining the epidemiological characteristics of PCNSLs in non-HIV-positive patients in Piedmont, Italy, in 1984–2003, and at determining the trend in incidence by analysing a period subsequent to that so far generally reported, that is, 1999. Piedmont is a region in Northwest Italy, which had a mean population of 4.3 million/year during the study period. From the files of clinical records of patients hospitalized in neurological and neurosurgical departments, 127 cases (50.8% males) of histologically verified PCNSLs were recorded. Mean age was 59 years (range 21–83; one case 8 years). The incidence rate per million in the whole period studied was 1.47 (95% confidence interval, 1.25–1.73). The highest incidence rate was in the age group 60–69. Incidence rates were steady over the whole period. Histology was reviewed in 110 cases: 62 were diffuse large cells, 39 immunoblastic, and 9 were unclassifiable malignant lymphomas. A comparison was made with incidence data concerning all intracranial tumors and noncerebral non-Hodgkin lymphomas (NHL) reported in the Cancer Registry of Piedmont for the same period: PCNSLs were 3.97% of brain tumors and 7.25% of NHL. In the study period, incidence rates for cerebral tumors increased from 33.7 to 43.7 per million; those of NHL (20.3 per million) did not change. Our data are within the ranges reported in the literature and do not show a trend toward an increasing incidence of non-HIV-related PCNSL. In particular, the incidence in the late 1990s up to 2003 is stable in comparison with previous years; the data are consistent with the reported leveling off of PCNSL in the late 1990s in areas where an excessive increase was found.


Thomas M. Kollmeyer, Ping Yang, Kristin Buckner, William Bamlet, Karla V. Ballman, and Robert B. Jenkins; Department of Laboratory Medicine and Pathology, Mayo Clinic and Foundation, Rochester, Minnesota; Health Sciences Research, Mayo Clinic and Foundation, Rochester, Minnesota; USA

Deletions of 19q have been associated with gliomas, especially oligodendrogliomas. In addition, oligodendrogliomas with 19q deletion have been observed to have a better survival. We have previously described a 150-kb minimal deletion region in gliomas that maps to 19q13.33 and contains three novel candidate genes (GLTSCR1, EHD2 and GLTSCR2) (Smith et al., 2002, Genes Chromosomes Cancer 29:16). Recently, polymorphisms in loci near this deletion region (ERCC2, RAI, and D19S246) have been associated with basal cell, breast, and lung carcinoma. A polymorphism within GLTSCR1 has recently been associated with prostate cancer aggressiveness. We carried out a pilot SNP association study using 141 glioma cases (61 astrocytomas, 40 oligodendrogliomas, 40 mixed oligoastrocytomas) and 108 normal controls. We evaluated 7 SNPs in 6 genes within and nearby the minimal 19q deletion region (ERCC1, RAI, ASE-1, ERCC2, GLTSCR1, and LIG1). The presence of germline GLTSCR1-exon-1 T and ERCC2-exon-22 C alleles was significantly associated with the development of oligodendroglioma (P = 0.029 and 0.043, respectively). Very similar and statistically significant associations were observed when the GLTSCR1 exon 1 polymorphism data was analyzed using genotype- and haplotype-based methods. No other associations between the tested SNP alleles and oligodendroglioma or other glioma types were observed. One high-risk and one low-risk haplotype were identified to be associated with oligodendroglioma development (P = 0.003 and 0.026, respectively). The haplotypes differed only by their GLTSCR1-exon-1 and ASE-1-exon-3 SNP alleles. Oligodendrogliomas with 19q deletion had a significantly higher frequency of the GLTSCR1-exon-1 T allele compared to oligodendrogliomas without 19q deletion (P = 0.019). Interestingly, glioma patients who were homozygous for the GLTSCR1-exon-1 T allele had a significantly better survival: 77% and 68% survival at 2 and 5 years, respectively, compared to 54% and 34% for other genotypes (P = 0.02). Multivariate analysis identified grade, age, GLTSCR1 genotype, and ERCC2-exon-22 genotype as the variables that best predicted survival. While these results must be validated by other studies, they suggest that alterations of GLTSCR1 (or a closely linked gene) are associated with the development and progression of glioma, especially oligodendroglioma.


E. Lee, J. Grutsch, V. Persky, R. Glick, and F. Davis; University of Illinois at Chicago, Chicago, Illinois; Cancer Treatment Centers of America, Arlington Heights, Illinois; Stroger Hospital, Chicago, Illinois; USA

Meningiomas are benign brain tumors for which high-dose ionizing radiation is an established risk factor. These tumors occur more commonly in females, particularly in the reproductive years. The coincidence between meningioma and breast cancer, and case reports of tumor growth during pregnancy, may point to a hormonal role. A case control study was conducted to investigate the relationship between hormonal and reproductive factors and meningioma risk. Female subjects treated between 1987 and 1992 were identified from three Chicago area medical institutions. Female spouses of male black patients were recruited through their partners as controls. This spouse control shares the advantage of general population controls, although there is some potential for overmatching on some lifestyle and environmental factors. A self-administered mail questionnaire focused on exogenous, endogenous, and other hormonal factors; personal and family medical history; radiation exposures; and other potential risk factors. Odds ratios (ORs) and 95% confidence intervals were estimated by using crude, stratified, and multivariable logistic models including 219 cases and 260 controls. The response rates were 86% among cases and 75% among controls. An elevated OR was suggested comparing blacks to whites (OR = 2.4, 95% CI, 1.0–6.1), with 8% of the cases self reporting as black. A protective effect was observed for pregnancy, which increased with number of pregnancies and age at first pregnancy. The ORs for three or more pregnancies compared to none was 0.3 (95% CI, 0.2–0.6). Age at menarche or total period of hormonal activity did not seem to have protective effects on meningioma development. Menopause, either naturally induced (OR = 1.1; 95% CI, 0.7–1.7) or surgically induced (OR = 1.1; 95% CI, 0.7–1.9), did not show any effect on meningioma on crude analysis. However, when other potential factors including age were controlled simultaneously, menopause showed an increased risk (OR = 2.0; 95% CI, 1.0–4.0). Use of oral contraceptives (OCs) was protective, with current use showing a stronger effect (OR = 0.2; 95% CI, 0.0–0.8) than previous use (OR = 0.5; 95% CI, 0.4–0.8). Hormonal replacement therapy (HRT) showed a modest, nonstatistically significant protective effect. Ever-smokers also showed a decreased risk for meningioma (OR = 0.6; 95% CI, 0.4–0.9). Stratifying by menopausal status, the OR associated with black race was retained in postmenopausal women only, while the protective ORs for pregnancy, oral contraceptive use, and smoking became stronger in premenopausal women. The pattern by duration and timing of use does not suggest a strong etiologic role for OCs and HRT. Smoking has not shown consistent protective effects across studies, although an anti-estrogenic effect of smoking is known. While preliminary, these protective effects with endogenous and exogenous hormonal factors suggest that factors other than these need to be considered as potential explanations for the female preponderance of these tumors. Further elucidation of hormonal etiologic factors should consider menopausal status and location of tumors.


Longjian Liu and Yi-Hong Zhou; Department of Neurobiology and Developmental Sciences, University of Arkansas for Medical Sciences, Little Rock, Arkansas, USA

Polymorphisms in glutathione S-transferase (GST) genes, a super-family that plays a key role in carcinogen metabolism, have been associated with an increased susceptibility to several types of cancers. However, only until recent decades, few studies examined the association between GST polymorphisms and risk of brain tumors. To assess whether GST polymorphisms were associated with an increased susceptibility for brain tumors, we performed a literature review and meta-analysis. All relevant English-language articles were retrieved by keyword searches of Medline (1966 to April 2004). References cited in related studies were also reviewed. Criteria of the selection of studies included in the present meta-analysis are case-control study and data presented to be possible for further calculations of odds ratio (OR) and its 95% confidential intervals (95% CI). In the pooled analyses, we fitted random effects models, which estimate summary OR using each individual-study results weighted by a factor of within- and between-study variance. Homogeneity among different groups was assessed via chi-square statistic, with P-values of <0.05, indicating lack of homogeneity. Six studies (a total of sample of 1747 cases and 2217 controls) met the selection criteria. The majority of study subjects were Caucasians. Three main families of GST genes (μ [GSTM], [GSTT], and [GSTP]) related to the risk of brain tumor were examined. Pooled analyses indicate no overall association between total brain tumors and the GSTM1 null, GSTT1 null, GSTP1 105-val/val, or GSTPI 114-Ala/val genotypes. However, after stratification by histopathologic subtypes, GSTT1 null and GSTP1 105-val/val were risk factors for glioblastoma multiforme (OR = 1.41 [95% CI, 1.06–1.87] and 1.77 [95% CI, 1.21–2.59], respectively). GSTT1 null was a risk factor for meningioma (OR = 1.98; 95% CI, 1.35–2.90). No significant association was observed between GSTM1 null and risk of brain tumors, although subjects with GSTM1 null appeared to be at higher risk for brain tumors than those who had GSTM1. One study tested gene-gene interactions, but no statistically significant was observed. None of these studies tested gene-environmental interactions. The present meta-analyses suggest that certain types of GST polymorphisms may play a role in the risk of a subset of brain tumors, which encourages further, detailed study.


Katie L. Miller,1,2 Karl T. Kelsey,3 John K. Wiencke,4 Michelle Moghadassi,4 Rei Miike,4 Mei Liu,3 and Margaret Wrensch4; Departments of 1Environmental Health, 2Epidemiology, and 3Genetics and Complex Diseases, Harvard School of Public Health, Boston, Massachusetts; 4Departments of Neurological Surgery and Epidemiology and Biostatistics, University of California San Francisco, San Francisco, California; USA

The multi-drug resistance-1 gene (MDR1) encodes an ATP-dependent pump that prevents potentially carcinogenic substances from crossing the blood-brain barrier. We compare adult glioma cases and controls for the C3435T polymorphism that has been associated with reduced MDR1 expression. Data for adults with glioma in the San Francisco Bay area were obtained through rapid case ascertainment, and population-based controls were identified through random-digit dialing between 1991 and 1994 (series 1) and 1997 and 1999 (series 2). MDR1 genotype was determined for 52% of cases (458) and 61% of controls (528) who provided blood or buccal samples. Logistic regression controlling for age, gender, and ethnicity compared genotype frequencies for all cases and controls, with later stratification by ethnicity and gender, and three histologic subtypes. With CC as the referent, the TT genotype was nonsignificantly less frequent among cases compared to controls (OR = 0.87; 95% CI, 0.6–1.2), and there was no difference between the CT and CC genotypes (OR = 1.01; 95% CI, 0.7–1.4). Because of the small numbers of other ethnicities, remaining analyses were restricted to 846 white subjects. Among six categories of histology and gender, only male glioblastoma was associated with MDR1 TT genotype (OR = 0.51; 95% CI, 0.3–1.0), and these results were nearly statistically different from those for women (P for interaction = 0.05). Although the C3435T polymorphism does not appear to be associated with other types of glioma, we cannot rule out that this MDR1 polymorphism may be associated with glioblastoma among men.


Jennifer M. Propp1 and Bridget J. McCarthy2; 1Central Brain Tumor Registry of the United States, Chicago, Illinois; 2University of Illinois at Chicago, Chicago, Illinois; USA

Numerous mortality data and statistics are available in the population-based epidemiologic literature regarding primary malignant brain/CNS tumors, but the impact of all primary (nonmalignant and malignant) brain/CNS tumors on mortality in the United States (US) has not been elucidated. Person-years of potential life lost (PYPLL) and average years of potential life lost (AYPLL) are two indicators of premature mortality that were utilized in this study to describe the burden of all primary brain/CNS tumors in the US. Mortality data and life tables for the US for 2001 (the most recent year for which complete data were available) were obtained from the National Center for Health Statistics. Deaths due to primary brain/CNS tumors were identified by using the data item underlying cause of death (coded and classified according to the International Classification of Diseases). PYPLL was calculated by summing the number of years of expected additional life for an average person of the same age, race, and sex for each individual who died of brain and other CNS tumors in the US in 2001. AYPLL was calculated by dividing the PYPLL by the number of deaths from brain and other CNS tumors in the US in 2001. PYPLL and AYPLL were computed for all brain and CNS tumors and for selected tumor sites by tumor behavior, sex, race, Hispanic ethnicity, and/or age. Analyses were conducted with SAS software (Release 8.1). A total of 16,819 deaths (males: 8903; females: 7916) were recorded with underlying cause of death indicated as brain and other CNS tumors in the US in 2001. Twenty-five percent (n = 4150) of the brain/CNS tumor deaths recorded in the US in 2001 were due to tumors of nonmalignant behavior. A total of 357,483 PYPLL (males: 188,047; females: 169,436) were attributed to brain and other CNS tumors in the US in 2001. AYPLL per person dying from brain/CNS tumors was 21.3 years (males: 21.4 years; females: 21.1 years). Data will be presented in graphic and tabular formats to best reflect meaningful and statistically significant variations in PYPLL and AYPLL by behavior, sex, race, ethnicity, and age that emerged for all brain/CNS tumors and tumor sites. This initial look into the burden of premature death due to all (nonmalignant and malignant) primary brain/CNS tumors in the US will aid in the identification of groups disproportionately impacted by these tumors. In conjunction with other measures of mortality for these tumors (e.g., mortality rates), these PYPLL and AYPLL data have public health applications such as quantifying social and economic loss due to premature death and informing health policy planning, resource allocation, and public health priorities.


Aimee L. Quan,1 Jeffrey S. Ross,2 Shih-Yuan Lee,1 Glen H.J. Stevens,3 David M. Peereboom,4 Michael A. Vogelbaum5, Steven A. Toms5, Gene H. Barnett,5 and John H. Suh1; 1Departments of Radiation Oncology, 2Radiology, 3Neurology, 4Medical Oncology, and 5Neurosurgery, Brain Tumor Institute, The Cleveland Clinic Foundation, Cleveland, Ohio, USA

Very little data has been published regarding the outcome of glioblastoma multiforme (GBM) patients with multicentric or multifocal disease. The purpose of this study is to evaluate the prognostic significance of these two relatively rare subtypes of GBM. A retrospective review of all multicentric and multifocal GBM patients treated with surgery and/or radiation therapy at the Cleveland Clinic Foundation from July 1997 to December 2003 was performed. The imaging studies of all patients identified as having multi-centric or multifocal disease in an IRB-approved database were reviewed by a neuroradiologist to confirm the MRI findings. After the scans were reviewed, a total of 40 patients were felt to have true multicentric or multifocal disease. Multicentric disease was defined as two or more distinct enhancing lesions within the brain parenchyma without evidence of contiguous T2 changes, and no evidence of cerebrospinal fluid (CSF) spread. Multifocal disease was subdivided into multifocal disease and multifocal disease in close proximity. Multifocal disease was defined as two or more enhancing lesions located greater than 1 cm apart within the brain parenchyma with contiguous T2 changes, or with evidence of CSF spread. Patients with multifocal lesions located 1 cm or less from one another were defined as having multifocal disease in close proximity. In addition to GBM subtype, various prognostic factors were evaluated to determine the factors that influenced survival and radiographic local control. Survival was measured from the initial date of diagnosis of GBM. Radiographic local control was defined as the absence of progression on follow-up CT or MRI scans. The overall median survival was 4.5 months, with a 1-year survival of 8%. Ten of the 40 patients had multicentric disease, 13 patients had multifocal disease, and 17 patients had multifocal disease in close proximity. On univariate analysis, the factors that significantly influenced survival were mental status at the time of diagnosis (normal vs. abnormal, P = 0.03), treatment with radiation (>60 Gy vs. <60 Gy vs. no radiation, P = 0.0003), and treatment with chemotherapy (yes vs. no, P = 0.0006). On multivariate analysis, median survival was significantly improved for patients with normal mental status (hazard ratio [HR] = 0.4, P = 0.044) and patients who received chemotherapy (HR = 0.3, P = 0.001). There was no difference in survival between patients with multicentric disease, multifocal disease, and multifocal disease in close proximity. The 6-month overall local control was 16%. On univariate analysis, only GBM subtype (multifocal in close proximity vs. multifocal vs. multicentric, P = 0.01) was found to predict for improved local control. GBM patients with multicentric or multifocal disease have a particularly poor prognosis with a median survival of 4.5 months. Although there was no difference in survival between the three groups, patients with multifocal disease in close proximity had a longer median local control than patients with multicentric or multifocal disease. Further studies are needed to improve on these results.


Tatjana Seute,1 Pieter Leffers,2 Guul P.M. ten Velde,3 and Albert Twijnstra1; Departments of 1Neurology, 2Epidemiology, and 3Pulmonology, University Hospital of Maastricht, Maastricht, The Netherlands

Our objective was to study the frequency of leptomeningeal metastases (LMMs) in small-cell lung cancer (SCLC) patients, the effect of LMM on survival, whether location of LMM influences survival, and whether patients with brain metastases (BMs) located in the posterior fossa are at higher risk of developing LMM. LMMs are a devastating complication of malignancy. The solid tumors in which LMM are most frequently found are breast cancer, melanoma, and lung cancer. Whereas previous studies amalgamated data from patients with various primary malignancies, the present study prospectively followed a homogeneous group of SCLC patients. Data from 1980 to 2003 for 458 consecutive patients with microscopically proven SCLC were included. All patients were seen by neurologist on a regular basis. Routine imaging of the brain was performed (CT or MRI) before and after systemic therapy. Clinical diagnosis of LMM was confirmed by malignant cells in the cerebrospinal fluid, MR imaging of brain and/or spine, or autopsy. In this study including 458 patients, LMMs were found in 45 patients. The prevalence of LMM was 2%, and the incidence was 10%. Median survival after the diagnosis of LMM was 1.3 months (range, 0.2–25.3 months). In 21 patients (46%) LMMs were located only in the leptomeninges surrounding the brain (cranial), in 14 patients (31%) LMMs were located surrounding the spine, and in the remaining 10 patients (22%) LMMs were diagnosed in both locations. Median survival for these groups was respectively 0.7, 2.4, and 1.1 months. BMs were diagnosed in 67% of all patients with LMM. In 67% BMs were located (also) in the posterior fossa. We conclude that patients with BMs located in the posterior fossa are at higher risk of developing LMM compared to patients with BMs in the cerebrum.


Dima Suki, Zhong Xie, Myrna Turner, MaryJo Gleason, Judith McConathy, Olubunmi Owolabi, Weiming Shi, W.K. Alfred Yung, and Raymond Sawaya; The University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA

The Brain and Spine Center at The University of Texas M.D. Anderson Cancer Center is a leading multidisciplinary referral center for patients with CNS malignancies. The large volume of patients visiting the Center has created the need for a data repository that: (1) is easily accessible by health care providers at the point-of-care, (2) provides a useful census of various entities for administrative purposes, and (3) can be used in research endeavors to advance knowledge on the characteristics of patients with CNS malignancies, their treatments, and outcomes. The Brain and Spine Center Clinical Information System (BASCIS) integrates individual databases that already existed across the numerous medical specialties represented in the Center. These include Neurosurgery, Neuro-Oncology, Radiation Oncology, Neuroradiology, and Neuropathology. The system provides an efficient and effective approach to handling the data management needs of a sizable multispecialty treatment center. Using the latest in informatics technology, the database allows access to comprehensive high-quality data for administrative, research, and patient care purposes while fostering collaboration among the various disciplines. The database is a web-based, relational, ORACLE database. Its many strengths include but are not limited to the following: (1) It applies the knowledge accumulated through the creation and utilization of the individual departmental databases. (2) It eliminates the duplication of data in multiple databases, while allowing for more details on the specific treatments/procedures. (3) It ensures highest data quality through a large number of external and internal consistency checks, complex built-in logic checks and validation rules, dedicated “to be resolved” fields on every screen, an audit trail, a detailed and embedded data dictionary, and extensive training of data management staff. (4) It allows for automated data transfer from institutional sources. (5) It features a user-centered, screen-driven, structured, and efficient system for data entry, review, and mining, and a role-based system security. (6) It ensures compatibility with commonly used statistical software, thereby allowing the quick export of pertinent data in a standard format. (7) It is compliant with the Health Insurance Portability and Accountability Act (HIPAA). Details of the system’s features and components will be presented.


Dima Suki, Zhong Xie, MaryJo Gleason, Judith McConathy, Myrna Turner, Weiming Shi, Olubunmi Owolabi, Raymond Sawaya, and W.K. Alfred Yung; The University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA

A clinical information system (CIS) is a special type of electronic health record (EHR) that has the potential to translate clinical and laboratory discoveries into new and better treatments and therapies, thereby improving the safety and quality of health care. A comprehensive CIS system should meet the needs of the users. Some current EHR and CIS systems have problems that prevent them from being universally accepted by healthcare professionals and widely adopted in clinical practice and research. These problems may have arisen from a limited understanding of the issues involved and a limited attention to the users’ needs of the programmers who developed many of the systems. The Brain and Spine Center Clinical Information System (BASCIS) is a multidisciplinary, multipurpose CIS for patients with central nervous system malignancies. Integrating existing individual databases from Neurosurgery, Neuro-Oncology, Radiation Oncology, Neuroradiology, and Neuropathology, we faced challenges like development of a simple and structured data entry system with large storage capacity and adequate security safeguards. We also faced unique challenges stemming from the database’s unique history and aims. These include (1) integration of pre-existing specialty-specific databases with large data sets, established fields, field definitions, structures, and procedures, (2) structuring the database to meet the research, administrative, and patient care needs of multiple and varied users, and (3) designing a system to integrate function for the large number of departments and staff involved at all levels. The main goal was to ensure a system that meets everyone’s needs and is simple to use for various purposes. With the support of the Center’s Executive committee, as well as the faculty and clinical staff from the various departments, we developed a multidisciplinary, multilevel database task force. For this, we identified individuals with extensive experience at the clinical, research, programming, data management, and statistical levels to be part of the database development team. A strong line of communication was secured between various team members. We applied a user-centered method to analyze tasks, workflow, and optimal interfaces for data entry, review, and mining. We then designed prototypes to map the results of user, task, and representation (interface) analysis and evaluated these prototypes. These steps are critical and fundamental to wide adoption and use of the final product. Details of the user-centered approach followed will be presented.


Margaret Wrensch,1 Karl T. Kelsey,2 Mei Liu,2 Rei Miike,1 Michelle Moghadassi,1 Kenneth Aldape,3 Alex McMillan,1 Joseph Wiemels,1 and John K. Wiencke1; 1University of California San Francisco, San Francisco, California; 2Harvard School of Public Health, Boston, Massachusetts; 3M.D. Anderson Cancer Center, Houston, Texas; USA

Few studies have examined the role of constitutive variation in DNA repair genes in relation to adult glioma. We genotyped population-based cases ascertained through a rapid case ascertainment program and controls identified through random digit dialing in the San Francisco Bay Area between 1991 and 1994 (series 1) and 1997 and 2000 (series 2) for one variant in ERCC1 and two variants each in ERCC2, XRCC1, and MGMT. A single neuropathologist for each series determined histologic type. Blood or buccal swabs were obtained from about 54% of cases and 65% of controls. Case-control genotype frequencies were compared overall and by histologic type and by age group (≤40, 41–60, and >60), ethnicity, gender, and series. Approximately 500 controls and 450 cases were genotyped for the DNA repair variants. There were no notable or significant differences between cases and controls overall for any variant. Compared to controls, variants in ERCC1 and XRCC1 were significantly elevated in glioblastoma cases (ERCC1 odds ratio [OR] = 1.7 [95% CI, 1.0–3.1]; XRCC1 399 OR = 1.5 [95% CI, 1.0–2.4]), and a variant in MGMT 84 was elevated in cases with other histologies (OR = 1.5; 95% CI, 1.1–2.2). In comparisons restricted to white cases, DNA repair variant frequencies differed significantly across histologic groups only for XRCC1 399; wt/var and var/var genotype frequencies were 59%, 71%, 49%, and 44%, respectively, for subjects with glioblastoma, other astrocytic tumors, oligodendroglial and mixed tumors, and other histologies (P = 0.02). Also of interest, genotype frequencies for all variants except MGMT 84 varied significantly by ethnic group. Certain DNA repair polymorphisms may influence gliomagenesis or progression, but given the large numbers of comparisons, some results could be due to chance. Funding for this study was provided by NCI RO1CA52689 and P50CA097257.



Christine Billecke,1 Simona Cazacu,1 Ashley Movsisyan,1 Nick Farrell,2 and Oliver Bogler1; 1William & Karen Davidson Laboratory of Brain Tumor Biology, Hermelin Brain Tumor Center, Neurosurgery, Henry Ford Hospital; 2Department of Chemistry, Virginia Commonwealth University, Richmond, Virginia; USA

A new class of chemotherapeutics, the BBRs, with two platinum groups separated by different backbone spacers (platinum, spermine or spermidine), have greater efficacy against glioma cells in vitro and in xenograft models than conventional platinums, prompting further preclinical development and an analysis of their mechanism of action. It had previously been demonstrated that these BBR compounds form longer and predominantly interstrand DNA crosslinks in contrast to cisplatin’s short intrastrand adducts, show little cross-resistance with previous platinum compounds, and have a different mechanism of action. Using in vitro clonogenic assays we have found that the IC90 for BBR compounds is from 20- to 250-fold lower than that for cisplatin. BBR compounds induce ERK phosphorylation as does cisplatin, but do so more effectively at isotoxic doses. In addition, the response of glioma cells to BBR compounds is modulated by the mutational status of p53. A P53 mutant at position R248, with a near-wild-type structure and residual DNA binding activity, sensitized the cells to BBRs compounds, and we are determining whether these cells mounted a partial wild-type p53 response to drug treatment, using gene arrays. A complementary approach to determining the basis for the greater efficacy of BBR compounds is the use of proteomic profiling, with an all-liquid-phase chromatographic platform (ProteomeLab PF2D, BeckmanCoulter), to examine the response of cells at the protein level and to generate profiles indicative of responsiveness. Protein peaks that are differentially expressed are readily isolated for mass spectrometric identification using this system. Our initial focus is to profile cell-line-derived tumors with differential sensitivity to these compounds to identify characteristics of sensitive and resistant tumors that could eventually be used to stratify patients in the clinic. The low toxicity and promising, if limited outcome data obtained for BBR3464 in early phase clinical trials for solid non-CNS tumors, together with our preclinical data in gliomas, suggest that the BBRs represent a promising new class of drugs for the treatment of brain tumors and understanding their mechanism of action, and the identification of proteome-level profiles will allow them to be applied in the appropriate molecular context for glioma treatment.


Jennifer K. Chen, Lily J. Hu, DongFang Wang, Jingli Wang, Kathleen R. Lamborn, and Dennis F. Deen; Brain Tumor Research Center, University of California San Francisco, San Francisco, California, USA

Current therapy for glioblastoma relies on a multimodal approach. However, efficacy of treatment is limited by therapeutic ratio: Doses sufficient to kill tumor cells often prove toxic to normal cells as well. Therefore, it is especially important to develop therapies that are specifically cytotoxic to cancer cells, either directly or through bystander or sensitizing effects. Hypoxic cells would be ideal targets for such an approach since they are specific to diseased tissue and often comprise the most treatment-resistant subpopulation of a tumor. Cystosine deaminase (CD), which has been widely researched as a form of suicide gene therapy, acts by deaminating the nontoxic pro-drug 5-fluorocytosine (5-FC) to form the highly cytotoxic 5-fluorouracil (5-FU). Previous studies have found that this method of treatment can induce a bystander effect and radiosensitization in cancer cells. However, none of these studies were conducted under hypoxic conditions, which themselves select for cells with anomalous behavior. Therefore, in this study, we hoped to find whether CD/5-FC enzyme/pro-drug therapy could successfully induce a bystander effect in hypoxic glioblastoma cells. We used a previously made gene construct consisting of the SV40 minimal promoter under the control of 9 copies of hypoxia responsive elements (HREs). Under hypoxia, hypoxia-inducible factor-1 (HIF-1) becomes activated and binds to HRE sequences, facilitating transcription of the yeast CD gene downstream. We performed colony formation efficiency assays to assay for clonogenic cell survival and found strong evidence that a bystander effect occurred under hypoxic conditions. This work was supported by CA-85356 and NS-42927.


J.L. Eller,1 S.L. Longo,1 M.M. Kyle,1 D.J. Hicklin2, and G.W. Canute1; 1Department of Neurosurgery, SUNY Upstate Medical University, Syracuse, New York; 2ImClone Systems Incorporated, New York, New York; USA

We have previously demonstrated that the anti-EGFR monoclonal antibody cetuximab (C225) was effective against EGFR-amplified glioblastoma multiforme (GBM) in vitro and in vivo. Cetuximab was also effective against amplified lines when combined with either radiation or chemotherapy. The purpose of this study was to compare the effectiveness of cetuximab, as well as other EGFR tyrosine kinase inhibitors, against EGFR-amplified and unamplified GBM. EGFR-amplified and unamplified human GBM cell lines were exposed to cetuximab alone and in combination with radiation or chemotherapeutic agents. The cell lines were also exposed to the tyrosine kinase inhibitors OSI-774 and EKB-569. Cytotoxicity and apoptosis were assessed in vitro. Two EGFR-amplified GBMs were implanted intracranially or in flanks of nude mice, while the unamplified line U87 was implanted in the flank only. Animals received cetuximab twice a week intraperitoneally for 5 weeks, and median survivals were obtained. Cytotoxicity assays demonstrated a much greater sensitivity of the EGFR-amplified versus unamplified lines when using cetuximab, OSI-774, or EKB-569 as single agents. For the amplified lines, cetuximab-treated mice had very significant increases in median survival for both intracranial and flank tumor models. There was no significant difference in survival when mice harboring the unamplified U87 line were treated with cetuximab. These results demonstrate that the effectiveness of cetuximab, OSI-774, and EKB-569 as solo agents is dependent upon EGFR amplification. The molecular basis for the differences in sensitivity of GBM to these anti-EGFR treatments is currently being explored and will be presented.


Roberta Glick, Terry Lichtor, Henry Lim, Katherine Tarlock, Alessandra Spagnolo, Cinzia Dello Russo, and Douglas L. Feinstein; Department of Neurosurgery, Rush University/John H. Stroger Hospital of Cook County; Chicago, Illinois, Department of Anesthesiology, University of Illinois at Chicago, Chicago, Illinois; USA

Perioxisome proliferator-activated receptors (PPARs) belong to the nuclear hormone superfamily that form heterodimers with retinoic acid receptors and activate transcription. PPARs regulate normal glucose metabolism and mitochondrial function and inhibit proliferation while inducing apoptosis in a variety of malignant cells. We investigated the effects of two of these agents, Pioglitazone (Pio) or Troglitazone (Trog), both in vitro and in vivo, as a new treatment for malignant gliomas. In vitro, mouse GL261 glioma cells were grown in complete media until confluent. Normal astrocytes were used for comparison. Cells were incubated in fresh media with either low- or high-dose glucose and varying amounts of Pio, Trog, or media. Cell viability was assessed by measuring intracellular LDH release and mitochondrial reductive capacity using an MTS kit. Cellular metabolism was assessed by glucose loss and lactate release. In vivo, both oral and intra-cerebral (i.c.) PPAR treatments were investigated. C57Bl/6 mice were fed regular diet or Pio before and after i.c. implantation of Gl261 cells. Subsequently, mice with an i.c. glioma received an i.c. injection of Pio. In vitro, PPAR agonists stimulated lactate release and increased MTS reduction in tumor cells but not astrocytes, in a dose-dependent fashion. In contrast, PPAs stimulated increases in lactate production in astrocytes but not glioma cells. In vivo, oral PPARs protected against the development of a glioma, and i.c. injection was effective in prolonging survival of mice with an i.c. glioma. These results suggest that PPAR agonists can selectively induce cell death in tumor cells versus normal cells in vitro and prolong survival in vivo. These effects may be due to differences in the metabolic capacity of cells to compensate via anaerobic glycolysis and lactate synthesis for impaired respiratory function, or the induction of apoptosis. PPAR agonists may have therapeutic value in the treatment of malignant brain tumors.


Vinay Gupta1,3 and Thomas C. Chen1,2,3; Departments of 1Pathology and 2Neurosurgery and 3USC Center for Brain Tumor Research, University of Southern California, Los Angeles, California, USA

Pituitary tumors constitute 10% of all primary brain tumors with an incidence of 1 in 10,000 people. Not all pituitary tumors can be cured by conventional surgical and radiotherapeutic means, especially the malignant ones. There are no effective chemotherapeutic options available for nonfunctioning pituitary tumors. We looked at the feasibility of using Gleevec, a new selective tyrosine kinase inhibitor, currently being used for chronic myelogenous leukemia. The drug is known to exert its effects by inhibiting protein tyrosine kinase pathways via the PDGF receptor and C-kit receptor. Protein tyrosine kinases are involved in cellular differentiation, proliferation, and migration. Deregulation of these protein pathways is commonly observed in malignant tumors. We sought to examine the expression of PDGF ligand A and receptor alpha in pituitary tumors. Upon immunohistochemical analysis of pituitary tumor samples, expression of PDGF ligand A (34 of 48 cases) and PDGF receptor alpha (35 of 48 cases) was observed, which provides the rationale for use of Gleevec. The in vitro efficacy of the drug in a rat pituitary tumor cell line GH3 and pituitary primary human tumor cultures was demonstrated by MTT assays. The IC50 values ranged from 14.6 μM in primary pituitary cultures to 21 μM in the GH3 cell line. The mechanism of cytotoxicity was determined to be apoptosis by the TUNEL assay and cell cycle analysis. We plan on testing the efficacy of Gleevec in an in vivo animal model. These results indicate that Gleevec may be an important drug in the treatment of pituitary tumors


Walter A. Hall, Deborah A. Todhunter, Edward Rustamzadeh, Yanquin Shu, Sekou O. Doumbia, and Daniel A. Vallera; Minnesota Medical School, Minneapolis, Minnesota, USA

A bispecific immunotoxin (IT) DTAT13 was synthesized in order to target simultaneously the urokinase-type plasminogen activator receptor (uPAR)-expressing tumor neovasculature and IL-13 receptor expressing glioblastoma cells with the goal of intratumoral administration for brain tumors. The recombinant hybrid was created by using the noninternalizing N-terminal fragment of uPA (ATF) and the IL-13 molecule for binding plus the catalytic and translocation portion of diphtheria toxin (DT) for killing. The 71-kDa protein was highly selective for human glioblastoma in vitro, showing no loss on binding compared with DTAT and DTIL13 controls. In vivo, DTAT13 caused the regression of small tumors when administered at 10 μg/day given on a five-dose schedule every other day. DTAT13 was able to target both overexpressed uPAR and the vasculature, as demonstrated by its ability to kill HUVEC cells. Also, mortality studies indicated that DTAT13 was less toxic than DTAT or DTAT13. These findings indicate that bispecific IT may allow treatment of a broader subset of antigenically diverse patients while simultaneously reducing the exposure to toxin that would be required if two separate agents were employed.


F. He, L. Li, C.C. Ling, P.H. Gutin, and G.C. Li; Memorial Sloan-Kettering Cancer Center, New York, New York, USA

Ku70 is one component of a protein complex, Ku70 and Ku80, that functions as a heterodimer to bind DNA double-strand breaks and activates DNA-dependent protein kinase. Generation of Ku70/− and Ku80/− mice and cell lines has confirmed that Ku70 and Ku80 play an essential role in DNA damage repair. Specifically, Ku70 and Ku80 deficiency compromises cells’ ability to repair DNA double-strand breaks, increases cells’ radiosensitivity, and enhances radiation-induced apoptosis. Because of the extreme radioresistance of glioblastoma cells, we examined the feasibility of using adenovirus-mediated expression of a dominant negative Ku70 fragment as a gene therapy paradigm to sensitize glioma cells to ionizing radiation. Based on the analysis of the structure-function of Ku70 and the crystal structure of Ku70/Ku80 heterodimer, we designed and identified a candidate dominant negative fragment (designated DNKu70). We generated this mutant construct (involving a N-terminal deletion), stably overexpressed it in Rat-1 cells, and showed that it has a dominant negative effect. That is, overexpression of DNKu70 in Rat-1 cells resulted in decreased Ku-DNA end-binding activity and increased radiosensitivity. Furthermore, our pilot study shows that over-expression of DNKu70 in Rat-1 cells increases the radiosensitivity of these cells under both aerobic and hypoxic conditions. Taken together, these data suggest that this Ku70 fragment (DNKu70) confers a dominant negative effect, providing a novel approach to sensitize radioresistant, including hypoxia-mediated radioresistant, tumor cells. We then constructed and generated prototype recombinant replication-defective virus rAd(CMV-DNKu70) in which the DNKu70 gene fragment is placed under the control of the constitutively active CMV promoter. Human glioma U-87MG cells were infected with these viruses at a multiplicity of infection of 30. Our data show that after infection, the DNKu70 is significantly expressed in the U-87MG cells and the radiosensitivity of these cells is greatly increased. If tumor cells in general and radiation-resistant tumor cells in particular could be sensitized by downregulating the cellular level/activity of Ku, new approaches could be developed as adjuvant to radiation therapy.


S. Idema,1 M.L. Lamfers,1,2 V.W. van Beusechem,2 W.R. Gerritsen,2 W.P. Vandertop,1 and C.M. Dirven1; 1Departments of Neurosurgery and 2Medical Oncology, Division of Gene Therapy, VU University Medical Center, Amsterdam, the Netherlands

Conditionally replicative adenoviruses (CRAds) represent a promising class of agents used against therapy-resistant brain tumors such as glioblastoma multiforme (GBM). The oncolytic effect of these viruses may be enhanced by the expression of a transgene. The adenovirus AdD24-p53 which has a 24-bp deletion in the E1 region limiting replication to Rb mutant cells and is encoding the p53 tumor suppressor protein, has shown superior oncolytic activity compared to the parental control AdD24, which lacks the p53 expression cassette, in glioblastoma cell lines. As both the adenovirus E1A protein and p53 have been shown to enhance radiosensitivity, the current experiments were performed to assess the effect of AdD24-p53 in combination with radiotherapy in GBM monolayer cultures and multicellular spheroids. U-87 (p53 wild type) and U-251 (p53 mutant) glioma cells growing in monolayers were irradiated with 3, 6, and 9 Gy. After 24 h, cells were infected at a multiplicity of infection (MOI) of 0.1 or 0.01 with AdD24-p53 or the control AdD24. Eleven days after treatment, viability was assessed by using a quantitative crystal violet assay. U-87 spheroids were irradiated at 4 and 9 Gy and infected with 5 × 104 PFU of AdD24-p53 or AdD24. Viability was measured 12 days after infection using the tetrazolium salt–based WST-1 assay. Infection of U-87 cells at an MOI of 0.1 with AdD24-p53 or AdD24 resulted in 82% and 89% viability compared to uninfected controls, respectively. Irradiation with 3 Gy alone had no effect on viability, while 6 and 9 Gy reduced viability to 86% and 59%. The combination of AdD24-p53 and irradiation synergistically reduced viability to 39% (3 Gy), 19% (6 Gy), and 5% (9 Gy), whereas combination treatment with the control virus AdD24 resulted in 74%, 59%, and 12% viability, respectively. Combination of radiotherapy and viral treatment in U-251 cells at 6 Gy and MOI of 0.01 even resulted in a decrease in viability from 91% and 87% for the single treatments to 6% of untreated control when combined, compared to 81% viability when AdD24 was combined with radiotherapy. U-87 spheroids showed no reduction in viability when treated with irradiation up to 9 Gy. Combining AdD24 with radiotherapy did not significantly improve the oncolytic potency of the virus (80% vs. 75%). However, combination treatment with AdD24-p53 and 9 Gy significantly reduced viability from 76% (infection alone) to 50% (combination). Combining the radiosensitizing effects of a conditionally replicative adenovirus and p53 expression has the potential to greatly enhance the effect of radiotherapy in GBM. Here we show in vitro that combining the CRAD AdD24-p53 with irradiation reduces viability in glioma monolayer cultures and radioresistant spheroids from >75% for any single treatment down to 6% for the combination. These results show great promise for the treatment of human GBM in vivo combining radiotherapy with AdD24-p53.


Hirokazu Kambara,1 Hideyuki Okano,2 E. Antonio Chiocca,1 and Yoshinaga Saeki1; 1Molecular Neuro-Oncology Laboratory, Massachusetts General Hospital and Harvard Medical School, Charlestown, Massachusetts, USA; 2Department of Physiology, Keio University School of Medicine, Tokyo, Japan

HSV mutants, possessing deletions in both copies of the ICP34.5 gene and an insertional mutation in the ICP6 gene (ribonucleotide reductase), have proven safe through a number of animal experiments and clinical trials. Although these double mutants clearly demonstrate oncolytic effects as well as antitumor immunization, their oncolytic effects are markedly reduced. ICP34.5-null viruses fail to grow in a variety of cultured cells because of their inability to prevent RNA-dependent protein kinase (PKR)-mediated inhibition of protein synthesis. In HSV-infected cells, viral double-stranded RNA will activate PKR by autophosphorylation, and activated PKR phosphorylates eIF-2a, which inhibits initiation of protein translation within the cells and in turn inhibits viral replication. The ICP34.5 dephosphorylates eIF-2a, thus allowing continued protein synthesis and viral replication. In order to increase therapeutic efficacy and tumor selectivity, glioma selective expression of ICP34.5 in oncolytic HSV has been explored. As a candidate for glioma selective promoters, we examined a synthetic nestin promoter, which consists of the nestin enhancer and hsp68 minimum promoter. We first performed RT-PCR to examine nestin expression in various types of cultured cells. All six glioma cell lines tested were confirmed to be positive for nestin expression. Astrocytes, on the other hand, were negative. We, therefore, constructed a novel HSV mutant, rQNestin34.5, which expresses ICP34.5 under control of the synthetic nestin promoter. In the experiments using cultured human glioma cell lines in vitro, rQNestin34.5 showed remarkable enhancements in both viral propagation and oncolysis compared with the ICP34.5-null virus (rHsvQ1). In four primary glioblastoma cells, rQNestin34.5 also showed remarkable enhancement in both viral propagation and oncolysis compared to control rHsvQ1. In normal human astrocytes, rQNestin34.5 showed equivalent toxicity to control rHsvQ1. We also performed Western blotting to examine the phosphorylation state of eIF2a. In glioma cell lines infected with rQNestin34.5, the level of phospho-eIF2a was lower than that of control rHsvQ1, confirming that the ICP34.5 is expressed selectively in glioma cells and resulted in a decrease of phospho-eIF2a level. This was responsible for its efficient propagation in glioma cells. We further characterized the anticancer efficacy of this new virus in vivo using an athymic mouse subcutaneous tumor model and a brain tumor model (U87dEGFR cells). The rQNestin34.5 showed significantly more potent inhibition of tumor growth compared with control rHsvQ1. The mice with brain tumor treated by rQNestin34.5 survived longer than those treated by rHsvQ1.These results demonstrated that the oncolytic HSV expressing ICP34.5 under control of the synthetic nestin promoter is a promising therapeutic reagent for treating human glioma.


Dimpy Koul, Sherry Bergh, Ruijun Shen, and W.K. Alfred Yung; Brain Tumor Center, The University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA

The phosphatidylinositol 3-kinase (PI3K) pathway is an important regulator in a wide spectrum of tumor-related biological processes including cell proliferation, survival, motility, and neovascularization. The activation of PKB/Akt is regulated in a complex manner via phosphorylation of PKB/Akt on Thr-308 and Ser-473. Although PDK-1 has been shown to phosphorylate Akt at Thr-308, it is not clear whether there is a distinct kinase that exclusively phosphorylates Akt at Ser-473. Integrin-linked kinase (ILK) has been shown to phosphorylate PKB/Akt on Ser-473 in vitro. The activity of ILK is sensitive to levels of PI(3,4,5)P3, and, therefore, inactivation of a PI(3,4,5)P3 phosphatase, such as PTEN, would be expected to constitutively activate ILK leading to PKB/Akt activation. We have previously demonstrated that re-introduction of PTEN in cells with mutant PTEN (or amplification of PI3K) results in growth arrest, apoptosis, and anoikis and inhibition of invasion in vitro and in vivo. We demonstrate herein that a newly developed ILK inhibitor 74728 (by Kinetek Pharmaceutics, Inc., Vancouver, Canada) effectively inhibits signaling through the ILK-AKT cascade in U87 cells. Also, 74728 blocks both basal and EGF-induced phosphorylation of AKT downstream targets intracellularly, including p70S6K and GSK3a/b. Treatment of U87 glioma cells with 74728 showed 100% growth inhibition at 25 μM drug concentration at the 48-h time point. Inhibition of cell growth by this compound was confirmed by inhibition of anchorage-dependent growth as marked by decreased colony number and size in a colony-forming assay. The decrease in cell growth induced by 74728 is directly correlated with a dramatic accumulation of cells in the G2/M phase of the cell cycle. In addition, 74728 reduces the invasive capability of U87 glioma cells concomitant with reduction in MMP-2 activity. Taken together, these data demonstrate ILK inhibitor 74728 inhibits growth and invasion in glioma cells and therefore the AKT-ILK pathway is a high-quality target for molecular therapeutics for cancer.


Tiffany LaFortune, Ta-Jen Liu, and W.K. Alfred Yung; The Brain Tumor Center, Department of Neuro-Oncology, M.D. Anderson Cancer Center, Houston, Texas, USA

A novel anticancer agent, rapamycin, antagonizes mammalian target of rapamycin (mTOR) by complexing with FKBP12 and subsequently inhibiting mTOR’s kinase activity. RAD001, a derivative of rapamycin, has been shown to effectively inhibit mTOR-associated activation of downstream targets, such as the 40S ribosomal protein S6 kinase (p70s6k) and the eukaryotic initiation factor 4E-binding protein 1. Inhibiting mTOR results in cell cycle arrest. Our studies with glioma cell lines have been consistent with previous findings, in that treatment with RAD001 for 72 h leads to an accumulation in the G1 phase of the cell cycle. Western blot analysis confirms a decrease in G1-associated protein cyclin D1 in treated samples. While the effect of RAD001 on cell cycle progression has been well documented, little is known about the effect of targeting mTOR on other cellular functions. Since epidermal growth factor receptor (EGFR)-dependent signaling contributes to cellular invasion and angiogenesis through the PI3/AKT pathway, we hypothesize that disruption of mTOR activity, which is one target of EGFR signaling, may decrease glioma invasion as well as VEGF secretion. In this study, we tested the effect of RAD001 on the invasion of four glioma cell lines with varying levels of EGFR expression. The inhibitory concentration (IC50) of RAD001 (1–5 μM), determined by the MTT assay, was used to treat cells. Western blot analysis confirmed a decrease in phosphorylation of downstream targets S6, 4EBP1, and P70 S6 kinase, indicating an inhibition of mTOR activity. A Matrigel invasion assay was used to measure cellular invasion of U87, U87/EGFRVIII, U251, and U251/EGFR glioma cell lines. Cells were treated with 1, 2, and 4 μM RAD001 for 72 h prior to placement in transwells for an additional 24 h. Results demonstrate a decrease in cellular invasion of all four cell lines after treatment when compared with control. U87/EGFRVIII cells had a 75% decrease in cellular invasion following treatment with RAD001 compared with a 25% decrease of the U87 parental cell line. Similarly, U251/EGFR had a 70% decrease, with a 45% decrease found in U251 parental cells. VEGF secretion was measured by using the ELISA kit. All four cell lines were treated with 0.5, 1, and 2 μM of RAD001 for 72 h. Our results show a dose-dependent reduction of secreted VEGF. At 2 μM RAD001, there is a 60% reduction in U87, a 70% reduction in U87/EGFRVIII, an 80% reduction in U251, and a 60% reduction in U251/EGFR. Taken together, our results demonstrate that antagonizing mTOR activity by RAD001 decreases VEGF secretion and tumor cell invasion. This study indicates that mTOR plays a role in cellular invasion and angiogenesis of glioma cell lines, possibly through the EGFR signaling pathway. Results of this study further our understanding of additional anticancer effects of RAD001 on glioma cell lines.


B. Lal, S. Xia, R. Abounader, and J. Laterra; Johns Hopkins University and Kennedy Krieger Institute, Baltimore, Maryland, USA

Scatter factor/hepatocyte growth factor (SF/HGF) and its receptor tyrosine kinase c-Met are linked to malignancy of numerous tumors including gliomas. We have shown that activation of the SF/HGF:c-Met pathway enhances experimental glioma malignancy by stimulating angiogenic, proliferative, and cytoprotective pathways. In this study we asked if SF/HGF and c-Met gene targeting enhances therapeutic responses to gamma radiation in pre-established wild-type human glioma xenografts. Wild-type U87 human glioblastoma cells that express both SF/HGF and c-Met were implanted subcutaneously (SQ) in Nu/nu mice. On postimplantation day (PID) 5, xenografts (n = 8) were 24 ± 2 mm3 in size. On PIDs 5, 15, and 25 animals received intravenous liposomes complexed with endotoxin-free plasmid expression vectors (100 mg DNA in 50 μl) coding for either U1/control or U1/ribozyme (anti-SF/HGF + anti-Met) transgenes. SQ tumors received 300 cGy radiation (or mark radiation) on PIDs 7, 12, 18, 22, and 27 (1500 cGy total). SQ tumor growth was monitored by caliper measurements. All control-treated SQ tumors grew rapidly requiring animal euthanization on PID 25 (tumor volume 195 ± 23 mm3). Monotherapy with either radiation or U1/ribozymes alone resulted in modest tumor growth retardation, with tumors reaching 166 + 88 mm3 and 120 + 39 mm3, respectively, by PID 35. Combining radiation with U1/ribozymes induced tumor regression in all animals (mean size 6 ± 1.7 mm3 at PID 35, P < 0.005), 40% of which were tumor free by pathological examination on PID 60. In parallel experiments, Nu/nu mice received intracranial (IC) U87 glioma cells via fixed transcranial cannulae that were used later for intratumoral delivery of adenoviral vectors coding for U1/ribozymes. Pre-established IC tumors (n = 10) received adenoviruses coding for control U1 or U1/ribozymes (anti-SF/HGF + anti-Met) (5 × 107 pfu) on PIDs 5 and 15. Animals received 300 cGy gamma radiation (or mark radiation) on PIDs 7, 12, and 18 (900 cGy total). Median survival of control animals was 29 days, and all died from tumor by PID 50. Radiation alone insignificantly improved survival (P < 0.05) with no long-term survivors. U1/ribozymes alone modestly improved median survival, and 20% survived to PID 90. Combining radiation with U1/ribozymes dramatically improved long-term survival to 80% (P < 0.005), none of which had histologically detectable tumor at PID 90. A subset of animals bearing intracranial tumors was sacrificed on PID 21 for histological analyses. Immunohistochemical analyses revealed a substantial increase in TUNEL staining and caspase-3 activation in tumors treated with both radiation and U1/ribozymes compared to those treated with either modality alone. Pretreating cultured wild-type U87 cells with Ad-U1/ribozymes (anti-SF/HGF + anti-Met) enhanced 3-fold their death response to radiation by a combination of apoptosis and necrosis. The findings show that targeting SF/HGF:c-Met can strongly enhance antiglioma responses to radiation therapy via multiple mechanisms including apoptosis and necrosis.


M.L.M. Lamfers,1,2 D. Gianni,3 Y. Tang,4 J. Carette,2 G. Fulci,3 E.K. Hoebe,2 Y. Saeki, 3 R. Weissleder,4 W.P. Vandertop,1 W.R. Gerritsen,2 C.M.F. Dirven,1 and E.A. Chiocca3; 1Department of Neurosurgery and 2Division of Gene Therapy, Department of Medical Oncology, VU University Medical Center, Amsterdam, The Netherlands; 3Molecular Neuro-Oncology Laboratories, Neurosurgery Service, and 4Center for Molecular Imaging Research, Massachusetts General Hospital, Harvard Medical School, Charlestown, Massachusetts, USA

Replication-competent adenoviruses are currently under investigation as novel anticancer agents for various types of cancer including malignant glioma. Approaches to improve their oncolytic potency include the insertion of therapeutic transgenes into the viral genome. Such “armed” replication-competent viruses allow the action of therapeutic proteins to be combined with the antitumor properties of the viral infection. However, to date, little is known about the levels and duration of transgene expression by cells infected by such armed viruses. Using a tumor-selective adenovirus encoding luciferase (Ad.D24Luc), we investigated these questions in an intracranial mouse model for malignant glioma. Transgene expression was detected by using a charged-coupled device camera and was followed over a period of 15 days post intratumoral injection of the virus. In addition, the effects of cyclophosphamide (CPA) and etoposide on levels of transgene expression were studied. Both agents have been reported to enhance transgene expression by suppressing the antiviral immune response. Mice treated with Ad.D24Luc demonstrated a gradual decrease in intracranial luciferase expression during the course of 15 days. By day 15, luciferase expression had decreased a mean of 33-fold relative to the first measurements taken at 24 h postinfection. In mice treated with etoposide or CPA in combination with Ad.D24Luc, transgene expression followed a similar pattern as the Ad.d24Luc group for the first 7 days. Thereafter, levels stayed higher in both combination-treated groups, with etoposide-treated mice decreasing a mean of 10-fold by day 15 and CPA-treated mice decreasing 6-fold by day 15. Our results demonstrate that detectable levels of transgene expression are maintained for at least 15 days after infection with the oncolytic virus. Moreover, combination treatment with agents known to suppress the immune response can prolong high levels of expression.


Yael Mardor, Ofer Rahav, Zvi Lidar, Aharon Ocherashvilli, Dianne Daniels, Yiftach Roth, Jacob Zauberman, and Zvi Ram; The Advanced Technology Center and the Department of Neurosurgery, Sheba Medical Center, Tel-Hashomer; The Department of Neurosurgery, Sourasky Medical Center, Tel-Aviv, Israel

Convection-enhanced delivery (CED) is a promising technique for distribution of drugs into brain and tumor tissue. We have previously presented the clinical application of diffusion-weighted MRI (DWMRI) for monitoring CED of Taxol in patients with malignant brain tumors. In addition, we have recently demonstrated the application of Gd MRI for the immediate assessment of the formation of convection and its extent. Convection is known to depend on parameters such as catheter size, flow rate, tissue consistency, catheter localization, infusate concentration, and molecular weight. After accounting for these variables, from our clinical experience in treating GBM patients with various convection-based protocols, we have observed significant variability in the extent of convection among patients, as well as differential tumor response to the therapeutic drugs. This has prompted us to evaluate the impact of additional infusate characteristics (viscosity, density, and capillarity) on convection using drugs and solvents that are either used or in consideration for CED-based therapies. Combinations of ethanol, Cremaphore, Taxol, sucrose, human serum albumin, and soap at different concentrations were mixed with Gd-DTPA and infused into the rat striatum (1 mcl/min, 40 min). T1-weighted MRI scans were acquired immediately post-treatment. The efficacy and extent of convection, as determined by MRI, were found to depend strongly on infusate viscosity. While CED of low-viscosity solutions was associated with significant backflow along the catheter followed by leakage into the ventricles, CED of high-viscosity solutions resulted in efficient convection and minimal backflow and leakage into the ventricles. Increasing the viscosity of a carboplatin solution with sucrose led to a significant increase of infusate distribution within the striatum, reflected by increased volume of enhancement in the striatum depicted on the immediate T1-weighted MRI scans. This effect was accompanied by a corresponding increase in toxic tissue changes, clearly depicted by DWMRI 24 h post-treatment. These results suggest that increasing the viscosity of solvents (by increasing sugars or albumin concentrations, for example) may be a simple way to significantly enhance the efficacy of CED treatments, thus increasing their antitumor effect.


J. I. Martin, W.C. Broaddus, and H.L. Fillmore; Department of Neurosurgery and Harold F. Young Neurosurgical Center, Virginia Commonwealth University, Richmond, Virginia, USA

A functional single nucleotide polymorphism (SNP) at position -1607 in the matrix metalloproteinase-1 (MMP-1) promoter creates an Ets-1 consensus binding site which can result in significantly higher transcriptional activity of MMP-1. The incidence of the 2G allele has been found to be significantly higher in certain aggressive and metastatic nonglioma tumors. We have recently reported a significant preponderance of the 2G MMP-1 promoter genotype in glioblastomas (P < 0.02) when compared with normal distributions. Additionally, presence of the 2G SNP correlates with increased MMP-1 transcriptional activity in glioma cells. We therefore wanted to determine if a novel SNP decoy oligonucleotide could inhibit the increase in MMP-1 transcriptional activity due to SNP site activation in the MMP-1 promoter. Decoys were designed to mimic position −1622 to −1603 of the MMP-1 promoter. The decoy contains ETS and AP1 DNA consensus sites, as well as MMP-1 flanking sequences. Electromobility shift assays (EMSAs) were used to identify binding characteristics of the decoy. The effect of the decoy oligonucleotides on glioma cell transcriptional activity was assessed by using a dual-luciferase reporter assay, and on protein expression using a sandwich enzyme-linked immunoassay (ELISA). Statistical analysis was carried out by using a two-way ANOVA to evaluate the effects of the decoy oligonucleotides on MMP-1 activity. EMSAs indicated that Ets-1 and AP1 probes and MMP-1 promoter probes effectively bound proteins from glioma cell nuclear extracts. Excess decoy oligonucleotides were able to competitively inhibit protein interactions with the 2G MMP-1 promoter probe but not the 1G promoter probe, whereas the scrambled decoy oligonucleotides had no effect. Promoter studies showed a significant increase in transcriptional activity in cells transfected with a luciferase reporter coupled to the 2G promoter, and addition of 5 mM decoy oligonucleotides could effectively inhibit the increased activity (P = 0.0001). The 2G SNP decoy oligonucleotide at 5 mM resulted in reduction of MMP-1 protein levels by 30.5% in U87 (2G/2G) glioma cells and by only 10.3% in T98 (1G/1G) glioma cells. In summary, we have designed and validated a transcription factor decoy oligonucleotide specific to the MMP-1 promoter 2G SNP that can inhibit nuclear extract protein/MMP-1 promoter DNA binding, MMP-1 (2G) transcriptional activity, and MMP-1 protein expression in glioma cells containing the 2G allele. This decoy oligonucleotide will be a valuable tool for investigating the role of the 2G SNP in glioma cell biology and has potential for therapeutic applications in glioma therapy.


Susan Murphy-Poulton, Lauren McCann, Miriam Haywood, Ross Davey, and Frances Boyle; Bill Walsh Cancer Research Laboratory, Department of Medical Oncology, Royal North Shore Hospital, Sydney, Australia

The median survival of patients with high-grade gliomas (GBMs) after surgery and radiotherapy is 10 months, and strategies to enhance the effect of chemotherapy are urgently required. A recent clinical trial at the Royal North Shore Hospital of the anti-angiogenic agent thalidomide as treatment for relapsed GBM revealed a greater than 40% stabilization of disease. The purpose of this study was to investigate the potential of thalidomide to synergise with standard chemotherapy in a rat glioma model system in order to better design future clinical trials. Eighty female F344 rats were implanted SC with 9L tumors. They were divided into 10 treatment arms: (1) control, (2 a and b) BCNU 10 mg/kg IP on days 7 and 26, ± thalidomide, (3 a and b) BCNU 5 mg/kg IP on days 7 and 26, ± thalidomide, (4 a and b) BCNU 2.5 mg/kg IP on days 7 and 26, ± thalidomide, (5 a and b) cisplatin 1 mg/kg ×5 IP starting on day 6, then ×4 starting on day 25 ± thalidomide, (6) thalidomide (1% in food) starting on day 6. Treatment started when all animals had a palpable tumor. This experiment demonstrated synergy when thalidomide was combined with nontherapeutic doses of cisplatin or BCNU. Growth curves show no response when chemo or thalidomide are given alone. Both cisplatin and BCNU 5 mg/kg when combined with thalidomide induce SC tumor response and improve survival (31 vs. 38 days P < 0.05). The combination groups did not experience increased toxicity as compared with the single treatment groups. This experiment was repeated with 9L tumors implanted intracranially. The treatment groups were: (1) control, (2) thalidomide (1% in food) starting on day 2, (3 a and b) cisplatin 1 mg/kg IP ×4 starting on day 6, and ×2 on day 14, ± thalidomide. This experiment resulted in a similar survival benefit (18 vs. 24 days P < 0.05). Therefore, thalidomide can be combined with cisplatin or BCNU to improve the response of glioma to standard treatment but not at the cost of increased toxicity. Cisplatin levels were measured in tumor, heart, lung, kidney, liver, and plasma. These results show a trend to higher levels of cisplatin that were measured in the tumors of the rats receiving the combination treatment compared to the cisplatin level in the tumors of rats treated with cisplatin alone. However, lower levels of cisplatin in the normal tissues of rats receiving the combination treatment vs. rats treated with cisplatin alone. Thalidomide levels were also measured in the tumor, brain, and plasma. These results revealed higher levels of thalidomide in all tissue types from the rats treated with the combination treatment vs. thalidomide alone. These results suggest that thalidomide may be acting to increase cisplatin uptake or retention in the tumor but not in normal tissue.


Shigeo Ohba, Yuichi Hirose, and Takeshi Kawase; Department of Neurosurgery, Keio University, Tokyo, Japan

Recent studies on biological responses of tumor cells to DNA damage suggest that targeting stress-related proteins might enhance the tumoricidal effect of chemotherapeutic agents. The stress-associated, 90-kD heat shock protein HSP90 is one of the cellular molecular chaperones, which interact with client proteins and are associated with stability and function of those proteins. Many of the client proteins are involved in cell cycle regulation, cell survival, and oncogenesis and are associated with a cytoprotective mechanism against biological stress such as DNA damage. Therefore HSP90 inhibitors may act as antitumor agents in a variety of tumors and potentiate the cytotoxicity of other antitumor agents. We tested the cytotoxicity of the HSP90 inhibitor geldanamycin (GA) alone and GA in combination with BCNU or temozolomide (TMZ) on the human glioma cell line U87MG using a colony formation efficiency assay. GA itself showed remarkable cytotoxicity at 3 nM or higher concentration. When GA was combined with BCNU or TMZ, the cytotoxicity of these DNA-alkylating agents was potentiated by GA at 0.5–1 nM. FACS analyses showed that DNA-alkylating-agent-induced G2 arrest of U87MG cells was not affected by GA, which revealed that GA-induced potentiation of DNA-alkylating agents was not a consequence of cell cycle checkpoint abrogation. Immunoblot analyses showed that GA treatment did not affect expression of anti-apoptotic protein Akt, which is associated with HSP90. However, single cell electrophoresis (comet assay) of the drug-treated cells showed that GA treatment increased early phase DNA damage created by the alkylating agents. These data suggest that GA has anti-tumor effect at low concentration and, even at the concentration where GA does not show cytotoxicity, potentiates the cytotoxicity of clinically relevant DNA damaging agents. An in vivo study to evaluate the effectiveness of HSP90 inhibitor is scheduled.


R.M. Perera, Y. Narita, F.B. Furnari, M.L. Tavarnesi, R.B. Luwor, A.W. Burgess, E. Stockert, A.A. Jungbluth, L.J. Old, W.K. Cavenee, A.M. Scott, and T.G. Johns; Ludwig Institute for Cancer Research, Melbourne, Australia; San Diego, California, USA; New York, New York, USA

Monoclonal antibody (mAb) 806 is a novel epidermal growth factor receptor (EGFR) antibody that mediates significant antitumor activity in vivo. This antibody recognizes a mutant EGFR commonly expressed in glioma known as delta2-7 EGFR (de2-7 EGFR, or EGFRvIII) and a subset of the wild-type (wt) EGFR found in cells that overexpress the receptor. It does not bind normal tissue such as the liver. Nude mice bearing subcutaneous or intracranial glioma xenografts expressing the de2-7 or wt EGFR were treated with mAb 806 and 528 as single agents or in combination. MAb 528 is a prototypical EGFR antibody that antagonizes ligand binding in a manner analogous to C225 (Erbitux). Tumor growth rates were determined for each treatment group and sample xenografts collected and analyzed for proliferation, apoptosis, and EGFR expression. Combination therapy with mAb 806 and 528 resulted in synergistic antitumor activity when used to treat xenografts expressing the de2-7 EGFR or overexpressing the wt EGFR. While neither antibody alone induced downregulation of the de2-7 EGFR, when combined, the antibodies caused a rapid and dramatic decrease in the total cell surface de2-7 EGFR. This receptor downregulation was associated with a significant decrease in tumor cell proliferation as measured by Ki67 immunostaining. Additional studies also supported the notion that the antitumor efficacy of combination antibody therapy was mediated via modulation of EGFR signaling and not antibody effector function. Interestingly, mAb 528 was also effective against xenografts expressing the ligand independent de2-7 EGFR when used alone, demonstrating that its antitumor activity is not merely mediated through inhibition of ligand binding. We conclude that mAb 806 can synergize with other EGFR-specific antibodies by enhancing receptor downregulation leading to a reduction in EGFR signaling, thereby providing a further rationale for its translation into the clinic.


Dawn E. Post and Erwin G. Van Meir; Departments of Neurosurgery and Hematology/Oncology and Winship Cancer Institute, Emory University, Atlanta, Georgia, USA

Hypoxia plays a critical role in driving tumor malignancy and is associated with poor survival in cancer patients. Hypoxia-inducible factor (HIF) is a heterodimeric transcription factor that functions as the master regulator of the physiological reaction to hypoxia by directly binding to hypoxia-responsive elements (HREs) within target genes. GBMs contain extensive areas of hypoxia, and the HIF protein is expressed in all grades of gliomas, with the strongest expression in GBMs. To exploit HIF activation in gliomas for therapeutic purposes, we first generated novel bidirectional hypoxia/HIF-responsive promoters. We then used one of these promoters to control the expression of the adenoviral E1A replication gene, thereby creating a hypoxia/HIF-dependent replicative adenovirus (HYPR-Ad#1). In preclinical studies using human glioma models, we found that HYPR-Ad#1 replicates in and kills hypoxic/HIF-active cells in culture and in tumor xenografts and exhibits anti-tumor activity, including the prevention of tumor formation and growth inhibition of established tumors. Moreover, combining HYPR-Ad#1 with BCNU, a mainstream treatment for GBMs in the clinic that is most effective against normoxic tumor cells, resulted in dramatic antitumor effects compared to the individual monotherapies. To augment the antitumor abilities of HYPR-Ad#1, beyond the killing of infected hypoxic/HIF-active tumor cells, we have generated a novel second-generation HYPR-Ad that delivers the anti-tumorigenic interleukin-4 (IL-4) cytokine to the tumor microenvironment (HYPR-IL4-Ad). We utilized the unique bidirectional hypoxia/HIF-responsive promoters present within HYPR-Ad#1 to conditionally coregulate the expression of the Ad E1A replication and IL-4 genes. IL-4 was chosen because it exhibits multimodal antitumor activity, including the induction of a host immune response against the tumor and inhibition of tumor angiogenesis, and therefore should be particularly effective against GBMs which have immunosuppressive activity and are highly vascularized. We found that in culture HYPR-IL4-Ad expresses IL-4 and displays oncolytic activity in a hypoxia/HIF-dependent manner. Preclinical studies are under way to evaluate the antitumor abilities of HYPR-IL4-Ad compared to HYPR-Ad#1 against brain tumor xenografts. This anticancer strategy based on intratumoral delivery of a novel type of oncolytic adenovirus that delivers IL-4 therapy to the tumor microenvironment is innovative and is expected to result in enhanced efficacy, as these treatments have nonoverlapping antitumor activities. Most importantly, HYPR-IL4-Ad can be used as a therapy for a broad range of tumors that develop hypoxia or activated HIF regardless of their tissue origin or genetic makeup.


H. Ren1, A. Giese2, C. Walker3, L. Pazmany4, S. Pelech,5 and N.G. Rainov1,6; 1University of Liverpool, Department of Neurological Sciences, Liverpool, UK; 2Department of Neurosurgery, Faculty of Medicine, University of Lubeck, Germany; 3Clatterbridge Cancer Research Trust, J.K. Douglas Laboratories, Clatterbridge Hospital, Wirral, UK; 4Academic Rheumatology Unit, University of Liverpool, Liverpool, UK; 5Kinexus Bioinformatics Corporation, Vancouver, British Columbia, Canada; 6The Walton Centre for Neurology and Neurosurgery NHS Trust, Liverpool, UK

Imatinib mesylate (STI571, Gleevec), a novel receptor tyrosine kinase signal transduction inhibitor, is currently in clinical trials for treatment of leukemia and solid cancers, including malignant gliomas. This study investigated the biological activity and effector mechanisms of imatinib in human malignant glioma cell lines. Expression of platelet-derived growth factor receptors (PDGFRa and PDGFRb) and ligands (PDGFA and PDGFB) in malignant glioma cell lines was investigated by RT-PCR. Real-time PCR was carried out to quantitate the expression of PDGF receptors after imatinib treatment. Tumor cells were treated with imatinib and proliferation was measured by MTT assays. Cell cycle and apoptosis were analyzed by flow cytometry, and migration was quantified by monolayer migration assays. Multi-immunoblot based differential protein kinase profiling was performed on imatinib-treated and control tumor cells. All tested glioma cell lines expressed PDGFA and PDGFB and, with the exception of T98G, PDGFRa and PDGFRb were also detectable. Treatment with Imatinib caused dose-dependent downregulation of PDGFRb, while PDGFRa expression was less affected. There was no correlation between the presence of PDGFRa or PDGFRb and the biological effects of imatinib. The PDGFR-negative T98G glioma cells were the most sensitive, growth-inhibited, and unable to form colonies at > 5 μM imatinib. Other receptor positive glioma cell lines were inhibited at higher concentrations. The ability of some glioma cell lines to migrate was significantly reduced by imatinib in a dose-dependent fashion. FACS analysis demonstrated that after imatinib treatment glioma cells arrested in the G0–G1 phase of the cell cycle and higher concentrations induced cell death by apoptosis and necrosis. Protein kinase profiling indicated that treatment with imatinib resulted in downregulation (69%–44% of control) of casein kinase 2, protein tyrosine kinase 2, Janus kinases 1 and 2, and extracellular-regulated kinase 3, and in upregulation (113%–303% of control) of cyclin-dependent kinases 1 and 4, S6 kinase p70, and protein kinase Ca. In conclusion, imatinib blocks the proliferation and migration of malignant glioma cells and can downregulate some protein kinases, including PDGFRa and PDGFRb. Imatinib exhibits inhibitory activity even in the absence of PDGF receptors, indicating further molecular targets for the drug.


Ryuta Saito,1 John R. Bringas,1 Michal Krauze,1 Tracy R. McKnight,3 Pamela Jackson,3 Michael F. Wendland,3 Charles Noble,2 Christoph Mamot,2 Daryl C. Drummond,4 Dmitri B. Kirpotin,4 John W. Park,2,4 Mitchel S. Berger,1 and Krys S. Bankiewicz1; 1Department of Neurological Surgery, Brain Tumor Research Center, 2Division of Hematology-Oncology, and 3Department of Radiology, University of California San Francisco, San Francisco, California; and 4Hermes Biosciences, Inc., South San Francisco, California; USA

Many systemically delivered therapeutic agents in patients with brain tumors have resulted in suboptimal tissue penetration, systemic toxicity, and limited efficacy. Current studies involving the local delivery of therapeutic liposomal agents using convection-enhanced delivery (CED) have shown significant advantages over conventional systemic therapies in targeting regions of solid tissue. This unique infusion technique provides an attractive process for drug delivery into targeted regions of the brain while minimizing systemic exposure. Based on the hypothesis that visualizing drug distribution will be a first step toward developing an effective therapeutic strategy, our group has combined CED with real-time MRI to track liposomal delivery into targeted regions of the brain in nonhuman primates. Real-time MRI monitoring of liposomal distribution into the corona radiata, putamen, and brainstem was performed using a mixture of Gd-encapsulated liposomes and fluorescence-labeled liposomes. Following completion of the study, the animal was euthanized and the brain harvested and processed for histology. Coronal slices containing fluorescently labeled liposomes were coregistered with MRI of the same regions and confirmed the accuracy of our real-time monitoring. The application of these techniques in locally directing anticancer agents into affected areas of the brain containing the tumor may considerably improve the margin of safety for the patient and dramatically improve the therapeutic index of anti-cancer medication in therapy for gliomas.


Hidemitsu Sato, Naruo Kuwashima, Tsukasa Sakaida, Takahiko Tsugawa, Jill E. Dusak, Wendy K. Fellows-Mayle, Glenn D. Papworth, Simon C. Watkins, Andrea Gambotto, Ian F. Pollack, and Hideho Okada; Departments of Neurological Surgery and Surgery, University of Pittsburgh Cancer Institute, Pittsburgh, Pennsylvania, USA

We have created a novel cellular vehicle for gene therapy of malignant gliomas by transfection of murine bone marrow stroma cells (MSCs) with a cDNA encoding epidermal growth factor receptor (EGFR). These cells (EGFR-MSCs) demonstrate enhanced migratory responses toward glioma conditioned media in comparison to primary MSCs in vitro. Enhanced migration of EGFR-MSC was at least partially dependent on EGF-EGFR, PI3-K, MEK, and MAP kinases, PKC, and actin polymerization. Unlike primary MSCs, EGFR-MSCs were resistant to FasL-mediated cytotoxicity and were capable of stimulating allogeneic mixed lymphocyte reaction, suggesting EGFR-MSCs possess suitable characteristics as vehicles for brain tumor immunogene therapy. Following injection at various sites, including the contralateral hemisphere in the brain of syngeneic mice, EGFR-MSCs were able to migrate toward GL261 gliomas or B16 melanoma in vivo. Finally, intratumoral injection with EGFR-MSCs adenovirally engineered to secrete murine interferon-alpha to intracranial GL261 resulted in long-term survival of host animals at 60%. These data indicate that EGFR-MSCs may serve as attractive vehicles for infiltrating brain malignancies such as malignant gliomas.


Chalet Tan,1 Rita Noronha,3 Anthony J. Roecker,3 Paraskevi Giannakakou,2 Kyriacos C. Nicolaou,3 and Erwin G. Van Meir1,2; Departments of 1Neurosurgery and 2Hematology-Oncology, Winship Cancer Institute Brain Tumor Program, Emory University School of Medicine, Atlanta, Georgia; 3Department of Chemistry and The Skaggs Institute for Chemical Biology, The Scripps Research Institute, La Jolla, California; USA

Hypoxia-inducible factor-1 alpha (HIF-1 alpha), the oxygen-regulated subunit of the heterodimeric transcription factor HIF, is overexpressed in various types of human cancers, including glioblastoma, as a result of intratumoral hypoxia and genetic alterations. HIF activates the expression of angiogenic factors and glycolytic enzymes that promote tumor cell survival. Inhibition of HIF-1 activity is therefore a promising molecular target for cancer therapy. To find HIF inhibitors we have screened a 10,000-membered library of natural-product like compounds based on a benzopyran structure. We have identified a novel small molecule 103D5R which decreases HIF-1 alpha levels in glioma cells in a dose- and time-dependent manner with an IC50 value of 35 μM. Following 24-h incubation, 103D5R (50 μM) abrogates HIF-1 alpha protein expression induced by cobalt or hypoxia. The cytotoxicity of 103D5R in glioma cells is minimal at concentrations up to 100 μM. The HIF-1 inhibitory activity of 103D5R is also observed in prostate and breast cancer cell lines. 103D5R markedly decreases the protein synthesis rate of HIF-1 alpha within 1 h of treatment, while HIF-1 beta is not affected. Further study to elucidate the mechanism of action and potential antitumor activities of 103D5R are currently ongoing. 103D5R could serve as a lead compound for the development of clinically useful agents for antitumor therapy.


J. R. Tauro1 and R. A. Gemeinhart2; Departments of 1Biopharmaceutical Sciences and 2Bioengineering, The University of Illinois, Chicago, Illinois, USA

Glioblastoma multiforme accounts for 50% of all gliomas and has poor prognosis. One of the main reasons for the poor prognosis with current treatments is recurrence even after surgical resection. Cisplatin is an antineoplastic agent that has been shown to be effective against gliomas. An effective targeted release system to deliver cisplatin may help improve survival time for patients with gliomas. Matrix metalloproteases (MMPs) are a family of proteolytic enzymes that are involved in wound healing, angiogenesis, tumor invasion, and tissue remodeling. These enzymes, MMP-2 and MMP-9 in particular, have been shown to be overexpressed in malignant gliomas. The current study explores a potential system for the activation of cisplatin from a polymeric matrix using peptide substrates for MMPs. We hypothesized that peptide complexes of platinates can be used for targeted release of the drug without completely diminishing activity. Fluorescently labeled peptides were incubated with MMP-2 or MMP-9. Cleavage products of a series of peptides were studied by using reverse phase HPLC with fluorescence detection. Different conjugation bonds between the platinum and the peptides were evaluated for in vitro cytotoxicity with a malignant glioma cell line, including Pt complexed to the amino terminal (-NH-Pt) and Pt complexed to the carboxy terminal (-COO-Pt). To assess the in vitro effectiveness, platinates, platinum bound to the primary cleavage fragments, the targeting peptides and the primary cleavage fragments were incubated for 24 h with U-87MG cell line. Cell viability was measured by using a modified MTS assay. Cells were observed under a fluorescence microscope after staining with ethidium homodimer (stains dead cells) and the fluorescently labeled peptides. HPLC results prove that the peptides were cleaved at the expected location within the peptides. The most intriguing peptide was cleaved (100%) within 2.5 h by MMP-2. Maximum cleavage of this peptide by MMP-9 was 70% in 4 h. Cis-platin complexed fragments of this peptide (-COO-Pt) had EC50 of 87.66 μM, which is comparable to the effectiveness of cisplatin (EC50 = 37.39 μM). The other peptides (-NH-Pt conjugation) showed almost 4 times higher EC50 values. This may indicate that the -NH-Pt bond is much stronger than the -COO-Pt bond, which may limit the ability of the drug to crosslink DNA. However, in both cases the platinates were not completely inactivated. We conclude that the selected peptides were substrates for MMP-2 and MMP-9, although some selectivity was observed. The cytotoxicity of cisplatin is somewhat retained following cleavage of the peptide. This system may be further explored for the controlled activation of cisplatin by MMPs.


Steven A. Toms,1 Osman Muhammad,1 Jason Tasch,1 and Wei-Chiang Lin2; 1Brain Tumor Institute, The Cleveland Clinic Foundation, Cleveland, Ohio; 2Department of Biomedical Engineering, Vanderbilt University, Nashville, Tennessee; USA

Optical spectroscopic tools exist that allow open surgical and minimally invasive assays of intrinsic tissue optics. Optical detection of cellular and tissue viability may offer a minimally invasive way to assess tumor responsiveness to chemotherapies. We report on an optical spectroscopic change that precedes apoptotic cell death and appears related to NAD(P)H autofluorescence. The cell lines SW 480 and U87-MG were grown in culture and treated with cisplatin 100 μg/ml and tamoxifen 10 μM, respectively. Fluorescence spectroscopy at 355 nm excitation and 460 nm emission were collected. MTS assays were used to determine cell viability. Cell lysates were analyzed for NAD(P)H concentrations by mass spectroscopy. Autofluorescence at 355 nm excitation and 460 nm emission declines markedly after treatment with tamoxifen or cisplatin prior to the loss of cell viability as measured by MTS assay. For example, the relative viability of the U87-MG cell treated with tamoxifen at hours 0, 8, 12 and 24 of treatment was 100 ± 6, 85 ± 6, 53 ± 9, and 0 ± 3, respectively. The relative fluorescence at the same time points were 100 ± 2, 57 ± 6, 47 ± 3, and 0 ± 1. TUNNEL assays confirm that cell death is via apoptosis. The key cellular fluorophore at these wavelengths is NAD(P)H. Mass spectroscopic analysis of cell lysates at these time points reveals a drop in NAD(P)H concentrations that is parallel to the loss of fluorescence signal. NAD(P)H autofluorescence decline precedes apoptotic cell death. This may allow the design of minimally invasive spectroscopic tools to monitor chemotherapeutic response.


Bakhtiar Yamini, Xiaohong Yu, and Ralph Weichselbaum; Section of Neurosurgery, Department of Surgery, and Department of Radiation and Cellular Oncology, University of Chicago Hospitals, Chicago, Illinois, USA

Temozolomide (TMZ) is an alkylating agent with modest efficacy in the treatment of glioblastoma multiforme (GBM). Tumor necrosis factor alpha (TNFalpha) has shown synergistic antitumor effects in regional perfusion strategies with alkylating drugs. We investigated the combined use of Ad.Egr-TNF, a replication defective adenoviral vector encoding the cDNA for TNFalpha under the control DNA sequences of the chemo-inducible egr-1 promoter, and TMZ in a nude mouse experimental glioma model. TNFalpha induction was measured via ELISA in vitro and in vivo in U87MG cells and hind-limb tumors, respectively, following Ad.Egr-TNF ± TMZ administration. Intracranial U87 xenografts were grown and treated with intratumoral injection of Ad.Egr-TNF and intraperitoneal TMZ alone and in combination. Animal toxicity and survival was recorded, and Kaplan-Meier survival curves were plotted. Tumor sections were evaluated histopathologically, and TUNEL assessment of apoptosis was made. U87 cell cytotoxicity following treatment with 10 ng/ml TNFalpha and 100 μM TMZ was evaluated. Flow cytometric (FACS) assessment of fractional DNA content and annexin V binding was performed in U87 cells following treatment with TNFalpha and TMZ to evaluate for induction of apoptosis. We found that 20 mg/kg TMZ produced a 6.4-fold greater induction of TNFalpha from Ad.Egr-TNF than baseline Ad.Egr-TNF alone (P < 0.02). Median survival for animals treated with Ad.Egr-TNF alone, TMZ alone, and Ad.Egr-TNF/TMZ combination was 21, 28, and 76 days, respectively (P < 0.001 Ad.Egr-TNF/TMZ compared to TMZ alone by log-rank). There was no increase in necrosis following combination treatment. However, there were significantly more apoptotic tumor cells (TUNEL positive) following combination treatment than following Ad.Egr-TNF or TMZ alone (P < 0.05). TNFalpha/TMZ combination treatment led to a synergistic increase in U87 cell cytotoxicity when compared to either treatment alone (P < 0.001 by ANOVA). In vitro FACS analysis showed significantly increased apoptosis following combined treatment in comparison with either TNFalpha or TMZ alone (P < 0.05 by ANOVA). Combination treatment with transcriptionally activated intratumoral TNFalpha and systemic TMZ significantly prolongs survival in a nude mouse intracranial GBM model without increased toxicity compared with either treatment alone. TNFalpha and TMZ appear to interact synergistically to induce tumor cell cytotoxicity and apoptosis.



Lorena Bissola,1 Marica Eoli,2 Donata Bianchessi,1 Bianca Pollo,2 Antonio Silvani,2 Giovanni Broggi,3 Amerigo Boiardi,2 and Gaetano Finocchiaro1; Departments of 1Experimental Neurology and Diagnostics, 2Clinical Neurology, and 3Neurosurgery, Istituto Nazionale Neurologico Besta, Milan, Italy

To identify different genetic profiles and evaluate their prognostic significance, we analyzed 93 patients who underwent surgery for oligodendrogliomas and oligoastrocytomas in the Istituto Nazionale Neurologico Besta in Milan: 25 oligoastrocytomas (OA, grade II), 37 anaplastic oligoastrocytomas (AnOA, grade III), 18 oligodendrogliomas (O, grade II), and 13 anaplastic oligodendrogliomas (AnO, grade III). Loss of heterozygosity (LOH) was investigated with different microsatellites on 1p, 10q, 17p, and 19q. LOH on 1p was found in 58% of O and in 46% of OA, LOH on 19q in 61% of O and 41% of OA, LOH on 17p in 16% of O and 20% of OA, LOH on 10q in 12% O and 17% OA. LOH on 1p and 19q was combined in 37% (51% O and 30% OA) and mutually exclusive with LOH on 10q (P < 0.0001). LOH on 10q was associated with anaplastic features (P < 0.0001), while LOH on 17p was present both in grade II and grade III tumors. Both in O and OA, LOH on 1p was associated with prolonged overall survival (P = 0.003 and P = 0.01, respectively) and LOH on 10q with short overall survival (P < 0.0001 and P = 0.02, respectively). LOH on 1p was present in 18 of 37 AnOA and was associated with a better prognosis independently from the histological grade (P = 0.006). Only in O was LOH on 17p associated with a poor prognosis (P = 0.01).


Waldemar Debinski, Jill E. Floyd, Denise L. Harbaugh, and Denise M. Gibo; Brain Tumor Center of Excellence, Wake Forest University School of Medicine, Winston-Salem, North Carolina, USA

We have uncovered a receptor for interleukin 13 (IL13), IL13Rα2, which is an attractive marker/target for therapeutic interventions in patients with GBM. In search for novel markers/targets, we used cDNA microarrays that detect genes encoding for a variety of cytokines, their receptors and for oncogene/tumor suppressor genes. This search has identified patterns of various genes expression in both established GBM cell lines and GBM tumor specimens that are highly reproducible and compatible with several now published reports on gene/protein expression profiling in this disease. One of the several frequently upregulated genes in both GBM specimens and cell lines was EphA2. EphA2 belongs to the Eph family of receptor tyrosine kinases, and it has been implicated in embryogenesis and blood vessel formation. The presence of EphA2 has been previously documented in several peripheral solid tumors, such as skin, breast, and prostate. Also, the EphA2 function has been attributed to cancer cell invasiveness and tumor neovascularization. We have begun to examine the presence of a 130-kDa protein of EphA2 in GBM, since its presence might mark the existence of a powerful system important to GBM etiopathology. We employed immunohistochemistry and Western blots to determine the presence of immunoreactive EphA2 in both cell lines and tumor specimens. Seven different established GBM cell lines, including U-87, U-251, U-373, and A-172, eight GBM specimens, and four normal brain tissues were used. The assays were performed with fixed cells or snap-frozen tumor specimens using C-20 polyclonal antibody against EphA2 (Santa Cruz Biotechnology, Calif.). We found that EphA2 is overexpressed in all studied human GBM cell lines, suggesting a truly widespread expression of this receptor in GBM. The cells demonstrated clear-cut cell plasma membrane staining; some additional staining in the cytoplasm was also seen. When grown to confluence, these cells exhibited a characteristic honeycomb pattern of staining for EphA2 in vitro. Normal cells, such as human endothelial cells (HUVEC), did not exhibit immunoreactivity for EphA2. However, NIH 3T3 cells demonstrated a weak staining, as expected. In Western blots, cell lysates of GBM cells demonstrated an immunoreactive band of ≈130 kDa that was not present in normal cells or it was weakly present in NIH 3T3 cells. Six out of eight GBM specimens contained the immunoreactive band, while normal brain tissue did not exhibit any measurable EphA2 immunoreactivity. Thus, EphA2 receptor appears to be widely overexpressed in human GBM. In accordance with our expectation based on the gene expression profiling, EphA2 represents a novel, attractive molecular target for diagnostic, imaging, and therapeutic applications in patients with GBM. EphA2 may also be an important participant in GBM progression/maintenance. All these new investigative opportunities are currently being explored in our Center.


William P. Hendricks, Wolfgang K. Pfisterer, Ronald A. Nieman, Stephen W. Coons, Mark C. Preul, and Adrienne C. Scheck; Barrow Neurological Institute and Arizona State University, Phoenix, Arizona, USA

Meningiomas account for about 20% of all primary intracranial tumors. Although the majority of these are benign (WHO grade I), grade II atypical and grade III anaplastic meningiomas exhibit aggressive behavior such as infiltration of bone, muscle, dura, or brain. Molecular, molecular genetic, and biochemical analyses of meningiomas across meningioma grade and between primary and recurrent tumors of the same grade may provide clues to the molecular basis of progression and recurrence in meningiomas, ultimately leading to the discovery of novel therapeutic targets and treatments. In this study, differential gene expression was analyzed across 40 meningiomas (19 grade I primary, 5 grade I recurrent, 12 grade II primary, 3 grade III primary, and 1 grade III recurrent) with oligonucleotide microarrays (Affymetrix U133 gene chips). These analyses provided large lists of differentially expressed genes with a P-value < 0.05 and >2-fold change (700–1500 genes). These lists were further narrowed and prioritized based on molecular genetic and biochemical characteristics of these tumors that were discovered with fluorescent in situ hybridization (FISH) and 1H nuclear magnetic resonance (1H-NMR) spectroscopy. The lists were first prioritized according to molecular genetic analyses in which aberrations of chromosomes 1, 14, and 22 were studied with FISH. The aberrations found included deletion of chromosomes 1p and 14q in grades I and II, deletion of 1p and 14q in grade III, monosomy of chromosome 22 in grades I and II, and trisomy of chromosome 22 in grade III. These data will filter differential gene expression analysis by genes on specific chromosomes. To further extract biologically relevant genes differentially expressed across tumor categories, data on the biochemical content of meningioma tissue extracts were obtained with 1H NMR of ex vivo samples. These data assist in selecting genes based on their involvement in metabolic pathways. Spectral analyses for specific differences in the concentrations of various metabolites showed elevated levels of the metabolites creatine, glutamine, alanine, and lactate in primary tumors relative to recurrent tumors. Additionally, the glycine-to-alanine ratio is elevated in recurrent tumors versus primary tumors and grade II tumors versus grade I. These data then allow for further gene expression analyses focusing based on a gene’s involvement in a corresponding metabolic pathway. The expression of individual genes found to be both statistically and biologically significant is being further examined with real-time quantitative reverse transcription polymerase chain reaction (QRT-PCR). Molecular, molecular genetic, and biochemical data may additionally provide clinically useful profiles for the diagnosis and prognostication of meningiomas. These profiles will enhance the ability to recognize more aggressive meningioma phenotypes early in their clinical course, allowing improved diagnosis and prognostication of meningiomas.


Chibo Hong,1 Alika Maunakea,1 Peter Jun,1 Andrew W. Bollen,2 William A. Weiss,1,3 and Joseph F. Costello1; Departments of 1Neurological Surgery, 2Pathology, and 3Neurology, University of California San Francisco, San Francisco, California, USA

Tumors arise from the combined effects of genetic and epigenetic mechanisms on gene expression. Tumorigenic epigenetic changes such as aberrant methylation of CpG islands lead to widespread inactivation of genes, including tumor suppressors. However, epigenetic defects are generally undetectable with conventional genetic approaches. Using an integrated approach for genome-wide methylation (restriction landmark genome scanning, RLGS) and copy number (array CGH) analyses of human gliomas, we identified a gene, SLC5A8 on chromosome 12q23.1, that was affected exclusively by aberrant methylation in human astrocytomas and oligodendrogliomas. SLC5A8 encodes a sodium/monocarboxylate cotransporter that was highly expressed in normal brain but was significant downregulated in primary gliomas (13 of 13 gliomas). Bisulfite sequencing analysis showed that the CpG island was unmethylated in normal brain, but extensively methylated in brain tumors, consistent with the tumor-specific loss of gene expression. SLC5A8 expression was suppressed in glioma cell lines and could be reactivated with a demethylating agent. Overexpression of SLC5A8 in LN443 glioma cells inhibited colony formation, suggesting it may function as a growth suppressor in normal brain cells. Furthermore, 7 of 8 murine oligodendroglial tumors (from animals transgenic for S100β-v-erbB and p53+/−) demonstrated a similar tumor-specific downregulation of AIT, the highly conserved mouse homologue of SLC5A8. AIT downregulation was not strictly model dependent, as it also occurred in mouse gliomas induced by S100β-v-erbB and p16+/−. Taken together, these data suggest that SLC5A8 functions as a tumor suppressor gene in vitro, and that it is silenced exclusively by aberrant methylation in primary gliomas. The epigenetic inactivation of AIT in mouse gliomas indicates an additional degree of commonality in the origin and/or pathway to tumorigenesis between primary human tumors and these mouse models of gliomas.


Peter Jun,1 Anita Lal,1 Michael W. McDermott,1 Andrew W. Bollen,2 and Joseph F. Costello1; 1The Brain Tumor Research Center and Departments of Neurological Surgery and 2Pathology, University of California San Francisco, San Francisco, California, USA

Meningiomas are the second most common sporadic neoplasm affecting the central nervous system. In spite of their frequency, their biology remains enigmatic with only several recognized genetic alterations. Epigenetic changes, namely CpG island methylation, also inactivate tumor suppressor gene function through transcriptional silencing. By integrating genetic and epigenetic analyses, we show that aberrant methylation predominates in WHO grade I and II tumors, while convergence of methylation and deletion is a rare phenomenon that increases with tumor grade. In contrast, both methylation and deletion are common in atypical and malignant tumors, but their convergence remains rare. This suggests that the pathway to formation of benign tumors is largely epigenetic, while higher grade tumors arise from the accumulation of both epigenetic and increased genetic defects that act predominantly on different sets of genes. One gene that exhibits specificity of tumor grade and mechanism of inactivation is PRKWNK2, a putative protein kinase located on chromosome 9q22.31. PRKWNK2 was found to be homozygously methylated in a large proportion of grade II and III tumors, but not in benign meningiomas. PRKWNK2 expression was specifically decreased in these primary tumors and upregulated in a meningioma cell line treated with the demethylating agent 5-aza-2′-deoxycytidine, which suggests that methylation has a functional consequence for PRKWNK2 expression. Potential effects of decreased PRKWNK2 expression on growth properties of meningioma cells are being investigated. Thus, aberrant methylation events are commonplace in all grades of meningiomas, whereas atypical and malignant tumors are more likely to follow a combined genetic and epigenetic pathway to tumor formation. For specific genes such as PRK-WNK2, aberrant methylation appears to be the predominant mechanism of inactivation. Despite the widespread defects in methylation in all grades of meningiomas, the tumor grade–specific methylation and silencing of PRK-WNK2 suggest its involvement in a more malignant tumor pathway.


D. Kamnasaran1 and A. Guha1,2; 1Arthur & Sonia Labatt Brain Tumor Centre, Hospital for Sick Children Research Institute, Toronto, Ontario; 2Division of Neurosurgery, and Western Hospital, University of Toronto, Toronto, Ontario; Canada

We developed an astrocytoma mouse model with GFAP-regulated expression of 12V-Ha-Ras, using embryonic stem cell transgenesis. In one line (RasD7), multiple integration sites of the 12V-Ha-Ras transgene led to death of the mice within ≈3 to 4 weeks, while the RasB8 line with a single copy integration of the transgene developed multifocal low- and high-grade astrocytomas, with death over a period of ≈3 to 6 months. The RasB8 mice are not born with astrocytomas, but progressively develop an increased number of them that are of higher grade. We therefore hypothesize that embryonic activated Ras creates genomic instability in the astrocytes, inducing genetic alterations leading to transformation. Taking advantage of this susceptibility, we are undertaking random mutagenesis by gene trapping and cDNA microarray analysis to identify early onset genes that contribute to astrocytoma development and progression in this mouse model. Since we have demonstrated the model to have many pathological and molecular similarities to human astrocytomas, some of these early onset genes may be similar, while others may be novel and yet unrecognized from our studies of human astrocytomas. Astrocyte cultures were developed from P0-RasB8 (prior to development of astrocytomas), 3mth-RasB8 (harboring low- and high-grade astrocytomas), 2wk-RasD7 (harboring multiple high-grade astrocytomas) and wild-type (wt) mice. The P0-RasB8 and P0-wt astrocytes did not grow colonies in soft-agar assays, keeping with their nontransformed phenotype, while 5% to 10% of the 3mth-RasB8 and >90% of 2wk-RasD7 astrocytes did, keeping with their transformation. Retroviruses were engineered encoding a Splice-acceptor type of gene trap vector, to infect ≈1.2 million of the P0-RasB8 and P0-wt astrocytes, at a multiplicity of infection of 1.5. Of the gene-trapped P0-RasB8 astrocytes, 0.002% became transformed as per growth in soft-agar, compared to none of the retroviral infected P0-wt astrocytes. This is consistent with our hypothesis that early expression of activated Ras makes the astrocytes susceptible to transformation. Currently, these gene-trapped P0-RasB8 transformed astrocytes are being evaluated for growth in Nod-Scid mice, with characterization of the trapped genes in those clones that demonstrate this second level of transformation. Furthermore, to investigate additional susceptibility genes, cDNA microarray expression experiments on 15,000 murine genes have been undertaken on all of the astrocyte lines. Several genes were found to be differentially expressed, results of which will be discussed in detail. Therefore, this study illustrates how mouse models based on genetic alterations characterized in human tumors, such as astrocytomas, can themselves be used to identify potentially novel and unrecognized genetic alterations in the human tumors.


F.W. Khwaja,1,2,5 J.J. Olson,2,6 D.J. Brat,3,4 S. Mendrinos,3,4 and E.G. Van Meir1,2,3,6; 1Laboratory of Molecular Neuro-Oncology and Departments of 2Neurosurgery, 3Hematology/Oncology, and 4Pathology, 5Genetics and Molecular Biology Graduate Program, and 6Winship Cancer Institute, Emory University School of Medicine, Georgia, USA

Glioma formation and malignant progression to high grade is driven by sequential accumulation of genetic alterations. These alterations and their downstream signaling events determine individual tumor properties. Genome alterations and the resulting changes in tumor cell transcriptome and proteome can all be used to make diagnosis and therapeutic decisions. Because of recent advances in analytical proteomic technology, it is now possible to study the changes in protein expression profiles induced by malignant transformation. The identification of individual proteins or groups of proteins expressed in neoplastic tissue has the potential to serve as diagnostic, prognostic, and treatment outcome markers. Proteome profiles better reflect the biological phenotype of each individual tumor than genetic and gene expression analyses because they can also detect posttranslational protein modifications, which may also be critical in malignant transformation. Cerebrospinal fluid (CSF) has been used as a source of potential protein biomarkers for various diseases. It allows rapid screening and accurate protein identification by proteomic technology. The goals of this study were to (1) examine whether CSF of patients of a given brain tumor type and malignancy grade have a characteristic proteomic profile and (2) use this information to understand the mechanism of gliomagenesis and malignant progression. Therefore, we studied differentially expressed proteins in CSF samples from patients with low- and high-grade astrocytomas as well as nontumoral controls using two-dimensional gel electrophoresis. These analyses showed consistent differences in the expression profiles present in CSF samples of the same tumor type that appeared to be correlated with malignancy grade. The differentially expressed proteins were identified by using matrix-assisted laser desorption/ionization–time-of-flight mass spectrometry (MALDI-TOF-MS) analysis. This allowed us to identify a spectrum of signature proteins for different glioma types that may constitute new therapeutic markers. Some of these markers are known tumor-related proteins, for example, N-myc oncoprotein, Beta-defensin, and VEGF. We have also identified several immune response proteins that are expressed at a high level in lower grade gliomas, yet were decreased in higher grade tumors, perhaps indicating a lowering of the immune response upon tumor progression. We are currently investigating the functions of the differentially expressed proteins.


Roger A. Kroes, Lori Crosson, Mary Nuzzo, Nigel Otto, Verne Hulce, and Joseph R. Moskal; The Falk Center for Molecular Therapeutics, McCormick School of Engineering, Department of Biomedical Engineering, Northwestern University, Evanston, Illinois; Field Neurosciences Institute, Saginaw, Michigan; USA

Altered glycosylation has been linked to cancer cell metastasis and invasivity. We have previously shown that expressing a specific sialyltransferase gene in glioma cells inhibited tumor formation in vivo. In order to identify other “glyco-gene” targets with therapeutic potential, oligonucleotide micro-arrays were constructed containing 366 human glyco-genes including glycosidases, glycosyltransferases, polysaccharide lyases, carbohydrate esterases, and carbohydrate-binding proteins. Gene expression profiles of normal human brain and malignant gliomas were compared. We designed 45-mer oligonucleotides using Array Designer [v 2.03] software (Premier Biosoft International, Palo Alto, Calif.). Selected oligos were synthesized by using a Poly-Plex 96-well oligonucleotide synthesizer (GeneMachines, Palo Alto, Calif.), and microarrays were manufactured with an OmniGrid robotic microarrayer (GeneMachines, Palo Alto Calif.). For microarray hybridization, a reference design was used allowing comparisons to be made across multiple samples. Both tissue RNA and Universal Human Reference RNA (Stratagene, LaJolla, Calif.) were amplified to produce antisense RNA (aRNA) by using an Amino Allyl MessageAmp aRNA kit (Ambion, Austin, Tex.) and postlabeled with Cy5 and Cy3 monofunctional dyes, respectively. We analyzed the microarray data sets using the Significance Analysis of Microarrays (SAM) algorithm. There were 11 significant genes more highly expressed in gliomas compared to normal brain and 25 genes more highly expressed in normal brain compared to gliomas. The differential expression of a select subset of these genes was then validated by quantitative PCR (qRT-PCR) using clinical specimens along with a panel of human glioma cell lines. Among the most noteworthy of these were the high levels of expression of two fucoslytransferase genes and the extracellular matrix protein CHI3L1 in gliomas and the alpha-mannosidase and sialyltransferase 7E genes in normal brain tissue. Historically, the identification of changes in glycoconjugate expression associated with cancer was obtained by measuring individual enzyme activities or structural changes of specific molecules. With microarray technology, it is possible to measure all genes associated with glycoconjugate biosynthesis and degradation simultaneously. Clearly, from the data presented here, there are many significant differences in expression of these genes, many of them novel and all of them potential targets for the development of therapeutics for the treatment of brain tumors.


Michael Lim,1 Griffith Harsh,1 Yingyun Wang,2 Mark Bednarski,2 and Samira Guccione2; Departments of 1Neurosurgery and 2Radiology, Stanford University School of Medicine, Stanford, California, USA

The management of patients with glioblastoma multiforme (GBM) would be greatly facilitated by a noninvasive means of making the diagnosis and monitoring response to therapy. Although GBM has a highly characteristic ring contrasting enhancing appearance on Gd(DTPA) MRI, pathologic analysis following biopsy or craniotomy is usually required for definitive diagnosis. The need for frequent MRI scans and their inability to clearly distinguish recurrent tumor from radiation change plague follow-up of patients with GBM. The goal of this study is to identify specific, sensitive blood markers for GBM diagnosis and monitoring. Initially, a genomic array comparison was performed between the contrast-enhancing and nonenhancing portions of 14 newly diagnosed GBMs. Gene products known to be extracellular or membrane bound were specifically compared. Serum levels of the proteins corresponding to genes upregulated in the contrast-enhancing regions were then measured in these 14 patients. Then, serum was obtained prior to surgery from 30 subsequent patients and was screened with the ELISA assay against the candidate proteins. IGFBP-2, aFGF, Galectin-3, and myeloperoxidase all were elevated in the sera of patients with GBM compared to levels in normal controls. We have identified potential marker proteins for the diagnosis and monitoring of GBM. These include proteins involved in cell structure, angiogenesis, and the immune system. We are currently studying whether these same, or different, proteins are upregulated in the sera of patients with other types of CNS tumors.


Eriks A. Lusis, Mark A. Watson, Michael R. Chicoine, Guido Reifenberger, David H. Gutmann, and Arie Perry; Departments of Neuropathology, Pathology and Immunology, Neurosurgery, and Neurology, Washington University School of Medicine, St. Louis, Missouri, USA; Heinrich-Heine-University, Düsseldorf, Germany

Meningiomas are the second most common CNS neoplasm, and while the majority are benign, up to 20% may display clinically aggressive features causing patient morbidity and mortality. Two early changes that occur are loss of the NF2 and 4.1B genes, but other molecular events that lead to tumor progression are poorly understood. Accordingly, we analyzed the transcriptomes of 5 benign and 5 anaplastic meningiomas using GeneChip microarrays representing over 37,000 unique human transcripts. We identified a set of 119 genes whose expression level was attenuated in anaplastic tumors as compared to benign lesions. When the genes corresponding to these transcripts were mapped to human chromosomes, 38 (32%) of the 119 differentially expressed genes were located on either chromosomal arms 1p or 14q. Overrepresentation of probe sets at these loci was statically significant (P = 0.0205, P < 0.0001, respectively). One transcript, NDRG2, at 14q11.2, demonstrated complete loss of expression in all five anaplastic meningiomas examined by gene expression microarray. Quantitative RT-PCR was used to validate NDRG2 expression levels in a larger, independent group (n = 34) of meningiomas. NDRG2 expression was significantly attenuated in WHO III meningiomas compared to the WHO I meningiomas (P = 0.0015). Gene expression was also examined in an independent set of 16 atypical (WHO II) meningiomas stratified according to clinical behavior. NDRG2 expression in clinically aggressive meningiomas was significantly decreased relative to that in more indolent cases (P = 0.0379). We conclude that progression toward higher meningioma grades involves significant alterations in gene expression profiles, especially along chromosomes 1p and 14q. NDRG2 shows uniform and consistent loss of expression in anaplastic as well as clinically aggressive atypical meningiomas and therefore represents an attractive candidate bio-marker for meningioma tumor progression.


Gayatry Mohapatra, Bjorn Carey, Leah Gaumont, Jim Earel, Rebecca A. Betensky, Tracy T. Batchelor, and David N. Louis; Molecular Pathology Unit and Molecular Neuro-Oncology Laboratory, Massachusetts General Hospital, Harvard Medical School and Harvard School of Public Health, Boston, Massachusetts, USA

Array-based comparative genomic hybridization (aCGH) is a powerful, high-throughput tool for whole-genome analysis. Until recently, however, aCGH could be reproducibly performed only on fresh or frozen tissue samples and required significant amounts of tissue. For brain tumors, paraffin-embedded tissue blocks may be the only tissue routinely available. The development of methods to analyze archival material therefore has the potential to impact molecular diagnosis in a significant way. We have constructed a BAC array representing chromosomes 1, 7, 19 and X. We have also optimized a labeling procedure for generating probe for aCGH using DNA from formalin-fixed, paraffin-embedded tissues: a two-step labeling procedure using an amine-modified nucleotide (amino-allyl dUTP). In step 1, amino-allyl dUTP is enzymatically incorporated into DNA fragments generated by DOP-PCR. In step 2, amine-modified DNA is chemically labeled with an amine reactive, fluorescent dye. Prior to hybridization, unincorporated nucleotides are removed from the labeled DNA by using Sephadex G-50 columns. The test and reference probes are then hybridized to the BAC array along with cot-1 DNA. Using this procedure, we have analyzed a small group of oligodendroglioma DNAs obtained from both frozen and formalin-fixed, paraffin-embedded tissues that were independently assayed for 1p/19q deletion either by FISH or by PCR-based LOH analysis. There was good concordance between the frozen and formalin-fixed DNAs as well as between aCGH and other results in this small series. Currently, we are analyzing a larger series of tumors utilizing this procedure to extend these results.


Wolf Mueller, Catherine L. Nutt, and David N. Louis; Department of Pathology, Cancer Center and Neurosurgical Service, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts, USA

Epigenetic alteration of DNA methylation is a hallmark of human malignancy. The promoter regions of many genes are populated by CpG dinucleotide islands, and methylation of these islands is associated with loss of gene expression. In carcinogenesis, global hypomethylation is often accompanied by dense hypermethylation of specific promoters. Pharmacologic reversal of epigenetic silencing by 5′-aza-2-deoxycytidine (5′Aza-dC) combined with expression profiling is a powerful tool for comprehensive analysis of methylation-related gene regulation and possibly for identification of tumor-related genes. Triplicates of three primary short-term glioblastoma cell cultures were exposed to 5μM 5-aza-2-deoxycytidine (5-Aza-dC) for 96 h followed by cRNA hybridization to an oligonucleotide microarray chip (Affymetrix U133A chip; 22,284 probe sets). Probe sets were selected for an average raw expression of greater than 20 in the treated samples and a greater than threefold increase in expression following 5′Aza-dC treatment; 110 probe sets met these criteria. Fifty of these probe sets, representing 42 genes, were statistically significant at p < 0.05. The chromosomal locations of the affected genes mapped to 1p, 1q, 3p, 3q, 4p, 4q, 6p, 7q, 9p, 12q, 15q, 17p, 17q, 18q, 19p, 19q, 22q, and Xp. Some of these chromosomal regions (4q, 15q, 18q, and 22q) are frequently lost in glioblastomas, indicating potential tumor suppressor gene loci. Bioinformatics revealed CpG islands in the promoter region of 17 of these 42 genes. Since demethylation of the whole genome does not distinguish between cell type–specific methylated genes and epigenetically silenced genes in cancer, we used gene expression data from RT-PCR in nontumorous adult and fetal brain tissue to prioritize potentially interesting candidate genes. In addition, genes counteracting the cytotoxic and antitumor activity of the 5′Aza-dC itself will be induced by the treatment; in this regard, it is noteworthy that cytidine deaminase (CDA), the enzyme that catalyzes the deamination of cytosine nucleoside analogs such as 5′Aza-dC, was among the 42 genes. These experiments provide a glimpse of the glioblastoma methylome, and it is expected that further work will identify specific genes that are regulated by methylation during glioblastoma tumorigenesis.


A. Pandita,1 L. Bereskin,1 J. Wiley,1 Z. Karim,1 A. Okhawat,1 P. Shannon,2 and A. Guha1,2; 1Brain Tumor Research Centre, The Hospital for Sick Children, Toronto, Ontario; 2The Toronto Western Hospital, Toronto, Ontario; Canada

Neurofibromas are the most common benign PNST, further classified as superficial dermal or deep, large plexiform neurofibromas. Dermal neurofibromas remain benign, while there is an ≈10% risk of plexiform neurofibromas becoming malignant (MPNST). Neurofibromas can arise sporadically, or when they are multiple in the context of neurofibromatosis-1 (NF-1), with the pathology and malignant predisposition almost identical. Common to all these subtypes is the transformation of Schwann cells, resulting from alterations at the level of DNA, RNA, and ultimately protein, in conjunction with the primary bi-allelic inactivation of Nf1 and neurofibromin expression. A limited number of low-resolution conventional CGH studies (≈10 MB) have been undertaken on entire neurofibroma specimens, which comprise transformed Schwann cell and several other cell types. Overall, these studies have not yet led to a thorough understanding of the molecular alterations from the varying neurofibroma subtypes that underlie their biologic/clinical differences. In this study, we are isolating Schwann cells from PNST specimens, resected from patients by using laser capture microdissection (LCM) and creating a high-resolution genetic alteration map (DNA and RNA). For DNA analysis, Schwann cells and normal endothelial cell controls are being isolated. Array CGH analysis is performed using BAC-arrays (Spectral Genomics) with an ≈1-MB resolution. Preliminary results reveals plexiform neurofibromas have more losses in comparison with MPNSTs, indicating alternate genetic pathways. Losses were mainly restricted to chromosomes 6, 9, 11, and 15. For MPNSTs, alterations involving chromosomes 4, 6, 8, 12, 17, and 20 were detected in addition to high copy gains on chromosomes 2q33-34 and 3q22-25. A complete analysis of these neurofibroma subtypes is currently under way. In parallel, we are also using matched frozen specimens to isolate RNA from the Schwann cells using LCM. The RNA from these cells is being tested for specificity and quality before being used for microarray analysis. These results from the array analysis will be confirmed using FISH and/or real time PCR on tumor specimens. These experiments will improve our understanding of the genetic differences in Schwann cells that explain differing biological/clinical behaviors of different types of neurofibromas in both sporadic and NF-1 patients, findings that may be exploited as molecular prognostic markers and targets for novel therapies.


Michael Taylor, Amar Gajjar, Twala Hogg, James Rutka, Abijit Guha, Tobey McDonald, Brian Rood, Christine Fuller, Deanna Naeve, Xioping Su, Yongxing Liu, and Richard Gilbertson; St. Jude Children’s Research Hospital, Memphis, Tennessee, USA

Central nervous system ependymomas comprise a group of histologically similar, but clinically heterogeneous tumors. We have characterized genome-wide molecular alterations in a group of 40 ependymomas using a combination of array comparative genomic hybridization (aCGH, 4000 Bacterial Artificial Chromosome array) and U133A Affymetrix RNA expression array analysis. aCGH identified many areas of genomic gain and loss in the ependymoma samples, including small, recurrent areas of amplification at chromosome 1q42 (25%), 8p12 (25%), 16p11.2 (20%), 11q13 (18%), and 19p13.3 (15%). In some cases, these areas of gain or loss were confirmed by genome-wide profiling of the same tumors by using the Affymetrix 10,000 Single Nucleotide Polymorphism (SNP) array. aCGH also identified homozygous deletion of the INK4A/ARF locus in 3 of 5 supratentorial (ST) ependymomas, but in no cases of posterior fossa (PF) or spinal (SP) ependymoma. Homozygous deletion of INK4A/ARF was confirmed by PCR and SNP array analysis. Unsupervised, hierarchical clustering of the aCGH data also separated the tumors into PF, ST, SP intramedullary, and SP myxopapillary groups. Thus we have identified a series of novel chromosome alterations that characterize ependymomas derived from different anatomical sites. We next used the Affymetrix U133A expression array to determine the expression status of over 20,000 transcripts in the same tumor samples analyzed by aCGH. Similar to analysis of aCGH data, unsupervised hierarchical clustering of the Affymetrix expression data divided the tumors into PF, ST, and spinal groups. To determine the relationship between chromosomal alterations and underlying gene expression, we correlated expression levels of genes (by Affymetrix RNA array) with aCGH data of underlying gene copy number change. This analysis demonstrated that most amplicons detected by aCGH included genes that were statistically overexpressed at the RNA level as measured by Affymetrix array analysis. We report, for the first time, a comprehensive characterization of genome-wide chromosome abnormalities and gene expression patterns in ependymoma. We show that pediatric ependymomas display anatomical site-specific gene expression patterns that correlate with underlying chromosome imbalance. We also show that homozygous deletions of the INK4A/ARF locus affect the majority of ST ependymomas and may represent the first tumor suppressor gene described for this subset of tumors.


K.L. Templeton1, M.E. Law1, A. Misra2, B.G. Feuerstein2, and R.B. Jenkins1; 1Department of Laboratory Medicine and Pathology, Division of Laboratory Genetics, Mayo Clinic and Foundation, Rochester, Minnesota; 2Department of Laboratory Medicine, University of California San Francisco, San Francisco, California, USA

Deletions of the chromosome 1p arm are frequent genetic abnormalities in primary human gliomas. However, the chromosome 1p status of many glioma cell lines has not been established. Using homozygosity mapping, fluorescent in situ hybridization (FISH), and comparative genomic hybridization to arrayed BACs (CGHa), we screened 17 glioma cell lines for chromosome 1 deletions. STS polymorphisms were used to evaluate the cell lines for regions of chromosome 1p homozygosity. Cell lines A172, U87, TP483, D37, U118, MO67, and TP265 contained significant regions of 1p homozygosity. FISH probes localized to 1p36.33 and CGHa were used to determine which cell lines had deletions 1p. Cell lines A172, U87, TP483, TP265, H4, U251, and D37 were deleted for portions of 1p. A combination of CGHa, FISH mapping, and homozygosity mapping of these cell lines defined a 700-kb common deletion region. This common deletion region is encompassed by a larger deletion region previously mapped in sporadic gliomas. This common deletion region is localized at 1p36.31 and includes CHD5, a putative tumor suppressor gene that is highly expressed in normal adult brain and has low expression in neuroblastoma cell lines with 1p deletion. These cell lines with deletions are a resource to study the function of candidate genes mapped to 1p36.


Timothy Van Meter,1 Catherine Dumur,2 Naiel Hafez,2 Carleton Garrett,2 Helen Fillmore,1 and William C. Broaddus1; 1Harold F. Young Neurosurgical Center and Departments of Neurosurgery and 2Pathology, Virginia Commonwealth University, Medical College of Virginia Campus, Richmond, Virginia, USA

Glioblastoma multiforme is the most aggressive and treatment-resistant adult primary brain tumor. The molecular changes that underlie tumor heterogeneity in glioblastoma have not been studied in detail. We have undertaken a detailed analysis of tissue prospectively collected intraoperatively from multiple intratumoral regions in glioblastoma, corresponding to the enhancing periphery and tumor core. We have performed a genome-wide expression analysis of these tissues using the high-density oligonucleotide HG-U133A arrays manufactured by Affymetrix. Our approach used an RNA extraction protocol paired with in-process histological scoring of regional samples using H&E staining of frozen sections. Gene expression profiles from different tumor regions were compared and correlated with percent tumor, percent necrosis, and other histological features. As expected, samples from the tumor core region generally had a greater extent of necrosis than periphery samples, but revealed striking differences in gene expression patterns. Analysis of very high quality RNA from perinecrotic tumor core tissue versus tissue from a similar area of tumor core with minimal necrosis revealed the greatest gene expression difference. A total of 778 probe sets, corresponding to 618 genes, showed increased expression in perinecrotic tissue, with >3-fold difference. Among these upregulated transcripts, we found several invasion-related genes, such as MMP-1, MMP-7, MMP-11, MMP-12, and MMP-23, and cell adhesion molecules such as CD44. Also striking was the overall comparison of data from periphery and core regions, in which 380 genes were found to have significantly different expression by univariate analysis (P < 0.01), although only six genes were differentially expressed over twofold. Interestingly, only a single gene was found to have significantly higher expression in periphery versus core samples, HMGCSI (3-hydroxy-3-methylglutaryl-Coenzyme A synthase 1, locus Chr. 5p14; P = 0.0089), with the remaining five genes showing higher expression in the tumor core regions. We are presently validating these results in a larger sample size of GBM, by measuring the observed changes at the RNA level by quantitative PCR and at the protein level by immunohistochemistry and Western blot. The results of this and similar studies may identify important regional differences in the molecular phenotype of individual GBMs, which may underlie regional differences in tumor behavior and treatment response.


D.S. Wang,1 S. Cha,3 C. Nardini,4 M. Diehn,2 B.K. Chan,1 and M.D. Kuo5; Departments of 1Radiology and 2Biochemistry, Stanford University School of Medicine, Stanford, California, USA; 3Department of Radiology, UCSF School of Medicine, San Francisco, California, USA; 4Department of Computer Science, University of Bologna, Bologna, Italy; 5Department of Radiology, UCSD School of Medicine, San Diego, California, USA

Genomic studies using cDNA microarray analysis have revealed global gene expression patterns specific for a number of cancers. This study sought to determine whether imaging features of glioblastoma multiforme (GBM) found on conventional contrast enhanced MRI may reflect underlying molecular oncologic biological processes revealed by microarray analysis. GBM tissue from 20 patients was analyzed by using cDNA microarrays (23,000 elements, representing over 17,000 different Unigene clusters). Corresponding MRI scans (standard T1, T2, and contrast enhanced) were retrospectively evaluated by two radiologists blinded to the genomic information. Four tumor imaging parameters were scored across all patients: mass effect, contrast enhancement, T2 heterogeneity, and edema. Statistical analysis was then performed to determine whether MRI findings correlated with the corresponding GBM cDNA microarray gene expression profile data. Two of four imaging parameters showed statistically significant correlation with large-scale gene expression changes: contrast enhancement and mass effect. Contrast enhancement correlated (P < 0.05) positively with upregulation of angiogenesis and endothelial-specific genes. It was also found to be positively correlated (P < 0.05) with upregulation of B cell and T cell–specific genes, as well as immune cell signaling genes such as those involved in chemokine and interferon expression. Mass effect correlated (P < 0.01) positively with upregulation of cell cycle progression and proliferation genes. Certain MRI features of GBM correlate with alterations in gene expression determined through microarray analysis. MRI may provide noninvasive means to derive genomic insights into GBM.


Weijun Wang,1 Chien-Kuo Tai,4 Allan D. Kershaw,3 Sounkery K. Solly,5 David Klatzmann,5 Noriyuki Kasahara,4 and Thomas C. Chen,1,2; Departments of 1Neurosurgery and 2Pathology, Keck School of Medicine, and 3Department of Chemistry, University of Southern California, Los Angeles, California; 4Department of Medicine, University of California at Los Angeles, Los Angeles, California; USA; 5Hôpital Pitié-Salpêtrière, Paris, France

We have recently demonstrated that a modified replication competent retrovirus (RCR) vector containing either the marker gene for green fluorescent protein (RCR-GFP) or the suicide gene E. coli cytosine deaminase (RCR-CD) was capable of achieving efficient replicative spreading and gene transfer in human gliomas in vitro and in vivo using an immune-deficient murine glioma model. The survival time of mice containing gliomas transduced with RCR-CD, followed by treatment with 5-fluorocytosine, resulted in 100% survival over 60 days, compared to 0% survival in the control group. RCR env gene expression, however, was detected in the bone marrow, spleen, and liver using real-time PCR, suggesting transduction of some of the normal replicating cells. The effect of this virus in an immune-competent model is especially important to answer the following questions: (1) Does RCR-CD elicit a cytotoxic immune response in the normal brain as in adenovirus? (2) Will the immune system eliminate RCR-CD systemically? The syngeneic RG2 rat glioma model was used in Fischer 344 rats. RG2 cells (5000 cells/10 μl) were stereotactically implanted into the rat right-frontal lobe; three days later, RCR-GFP (1 × 104 IU/ml) was injected intratumorally. Flow cytometry demonstrated that 70% of the glioma was transduced by three weeks; no transduction was found in the normal brain cells. No evidence of severe inflammation (CD4, CD8, CD11b) was observed in the adjacent normal parenchyma in the RCR-GFP transduced gliomas compared to PBS-injected controls. Rats with RCR-CD transduced gliomas treated with 5-FC survived approximately two weeks longer than control animals (P < 0.001). Importantly, no RCR env gene expression was detected systemically in the immune-competent rat, suggesting that the host immune system was capable of preventing inadvertent RCR transduction of normal host tissue. Our present data suggests that RCR-mediated gene therapy may be a safe and potent strategy for treating human gliomas.



Yasuharu Akasaki, Gentao Liu, Moneeb Ehtesham, Keith L. Black, and John S. Yu; Maxine Dunitz Neurosurgical Institute, Cedars-Sinai Medical Center, Los Angeles, California, USA

Prostaglandin E2 (PGE2), synthesized by cyclooxygenase-2 (COX-2) overexpressing glioma, is known to contribute to cellular immune suppression in glioma patients, but the mechanism remains unclear. We report a mechanism of a T regulatory type-1 (Tr1) induction by mature dendritic cells (DCs) that phagocytose allogeneic and autologous COX-2 overexpressing glioma. Human glioma cell line U-87MG and primary cultured glioblastoma cells (MG-377) overexpressed COX-2. Exposure to these gliomas on DCs induced potent secretion of IL-10 in DCs and impaired their ability to produce IL-12p70. These DCs induced a Tr1 response, which is characterized by robust secretion of IL-10 and TGF-beta with negligible IL-4 secretion by CD4+ T cells. Peripheral CD4+ T cell populations isolated from an MG-377 patient also predominantly demonstrated a Tr1 response against MG-377 cells. Selective COX-2 inhibition in COX-2 overexpressing gliomas at the time of phagocytic uptake by DCs abrogated this regulatory response and instead elicited a Th1 activity. COX-2 stable transfectants in LN-18 (LN-18-COX2) also induced a Tr1 response. The beneficial effect of a COX-2 inhibition in LN-18-COX2 is reversible following administration of PGE2. Taken together, robust levels of PGE2 from COX-2 overexpressing glioma cause DCs to secrete high levels of IL-10, and these DCs induce a Tr1 response.


Gary E. Archer, Darell Bigner, Allan Friedman, Henry Friedman, Kara Penne, Denise Lally-Batts, Alison Paolino, Sandra Tourt-Uhlig, David Reardon, John Sampson; Duke University Medical Center, Durham, North Carolina, USA

The use of immunotherapy for the treatment of malignant gliomas requires the generation of a strong immune response in what has traditionally been considered an immunologically privileged site. Most importantly, the immune response must be specific for the tumor so as not to cross-react with normal brain, which could result in a devastating autoimmune encephalomyelitis. The deletion mutation of the epidermal growth factor receptor (EGFRvIII) is a tumor-specific antigen, which is expressed on approximately 47% of all malignant gliomas (MGs). The in-frame EGFRvIII deletion combines distant parts of the molecule producing a novel glycine at the fusion junction. We have initiated a phase I clinical trial for patients with newly diagnosed MGs to determine the safety of vaccinating with mature DCs loaded with a peptide spanning the fusion junction (PEPvIII) conjugated to keyhole limpet hemocyanin (KLH). Vaccinations consist of 3 vaccines 2 weeks apart of PEPvIII-KLH loaded, mature DCs beginning 2 weeks following completion of post-resection radiotherapy. To date, 19 patients have been enrolled, with 17 completing their vaccinations. No patient showed a positive delayed-type hypersensitivity reaction to KLH or PEPvIII before vaccination, but after vaccination, 16/16 patients reacted to KLH, and 10/16 (62.5%) reacted to PEPvIII. All patients tested also had significant increases in antigen-specific T-cell proliferation in vitro after vaccination in response to PEPvIII (P = 0.01) and KLH (P ≤ 0.03). No humoral responses to PEPvIII were identified, however. Of two patients with residual radiographic disease after resection and radiation, one had a nearly complete response, and the other has not progressed for >653 days. Of the 17 patients without radiographically evident disease, 4/15 (26.7%) have not progressed and the median overall time to progression of was 288 days. For patients with glioblastoma multiforme (GBM), the median survival time was 596 days, which compares favorably with recently published trials evaluating newly diagnosed patients with GBM treated with Gliadel (417 days) (Westphal et al., 2003, Neuro-Oncol. 5:79); radiation and concurrent Temodar (480 days) (Stupp et al., 2002, J. Clin. Oncol. 20:1375); or radio-labeled anti-tenascin monoclonal antibodies (556 days) (Reardon et al., 2002, J. Clin. Oncol. 20:1389). These findings suggest that autologous mature DCs loaded with the tumor-specific antigen, PEPvIII, are safe and may induce a beneficial immunologic response in patients with MGs.


Michael J. Ciesielski, Tina C. Weiss, James Toldi, Lisa Apfel, Tara A. Barone, and Robert A. Fenstermaker; Departments of Neurosurgery and Immunology, Roswell Park Cancer Institute, Buffalo, New York; Department of Neurosurgery, School of Medicine and Biological Sciences, State University of New York at Buffalo, Buffalo, New York; USA

Malignant gliomas are susceptible to immunologically mediated attacks. Intracellular proteins can serve as targets for antiglioma vaccines. Epitopes of intracellular proteins are found at the cell surface of antigen-presenting cells (APC) in association with MHC-I molecules. If recognized by T cells, these molecules can induce effective responses against tumor cells bearing the same MHC-I-associated epitopes. The cellular protein survivin is a member of the family of proteins known as inhibitor of apoptosis proteins (IAP). Human gliomas express survivin at high levels, whereas normal cells do not. Therefore, survivin represents a tumor-specific target for immunologically based therapy. Since survivin is an intracellular protein that is expressed only in tumor cells, it may be possible to produce MHC-I-restricted cellular immunological attack in response to a survivin vaccine. F98 glioma cells, which express survivin at high levels, were injected into syngeneic Fischer rats. Four days after tumor implantation, rats were injected with either vaccine vector alone, or with vector containing a truncated xenogeneic (human) survivin gene. Each vaccine included GM-CSF as a vaccine adjuvant. The DNA vaccine vector without survivin had no effect on tumor growth; however, a single dose of the survivin-containing vaccine vector significantly attenuated tumor growth. Large numbers of CD8+ cells were detected in the tumors of rats vaccinated with the survivin vector, but not in the tumors of rats injected with the vaccine vector alone. These findings are consistent with the production of a specific cellular antitumor immune response against F98 gliomas and suggest that “molecular mimics” of survivin may stimulate tumor-specific immune responses to the endogenous survivin protein in gliomas in vivo. Xenogeneic anti-survivin vaccines may be of use in the treatment of malignant gliomas.


G. Fulci, D. Gianni, L.M. Breymann, E.J. Fontana, H. Wakimoto, and E.A. Chiocca; Department of Molecular Neuro-Oncology, Massachusetts General Hospital, Charlestown, Massachusetts, USA

Oncolytic viruses (OVs) generate intratumoral progeny viruses that propagate the oncolytic effect throughout the neoplasm. However, when compared to in vitro growth curves, the generation of progeny viruses in vivo appears relatively suppressed, and clinical trials for OV have yielded poor efficacy. We have previously demonstrated that, in rats harboring gliomas, pretreatment with cyclophosphamide (CPA) enhances OV therapeutic efficiency. Our goal now is to identify specific factors responsible for limiting OV survival in tumors in order to design novel pharmacologic means that enhance intratumoral OV replication in a selective fashion. The experimental model we use is a singenic tumor in rats treated with either OV or with a combined therapy CPA+OV. Our data show that pretreatment of rats with CPA allows in vivo propagation of OV within tumor tissue by preventing clearance of viral infected cells. This indicates a role for the host immune response in preventing OV intratumoral spread. We have thus characterized the inflammation response elicited by OV in rats pretreated with CPA and in those that did not receive CPA. Our data show that OV induces a three-step cascade of inflammation responses. The first step starts within 6 h following OV injection and consists of activation and proliferation of local microglial cells. At 76 h following OV injection, an angiogenic switch and strong induction of tumor blood vessels are observed. Finally, angiogenesis will lead to tumor infiltration of peripheral macrophages and other cells of the immune system. OV-induced proliferation of inflammatory cells increases their sensitivity to CPA, thus preventing this inflammation cascade in rats pretreated with CPA and allowing viral spread. Analysis of gene expression was performed at different time points, in peripheral blood lymphocytes (PBL) and tumor tissue of 4 treatment groups of rats: CPA/OV, CPA+/OV, CPA/OV+ and CPA+/OV+. In absence of CPA, OV induced the expression of IFNα, β, andγ, TNFα, IL15, and IL18 in PBL, but combined treatment OV+CPA showed a decrease to baseline of all those cytokines. In the tumor tissue IFNγ was the only cytokine among the above that changed in expression. It is induced within 6 h of OV injection and then remains constant until 76 h after OV injection, when its levels are dramatically increased. In CPA-pretreated rats, expression of IFNγ remains steady. The expression pattern of IFNγ is coherent with the model of three-step cascade of inflammation, suggesting that IFNγ has the major role in activating the host immune system in response to OV. These data uncover molecular pathways regulating OV interactions with the host defense system, providing an explanation for the enhanced therapeutic effect observed in tumor treatment using OV with CPA and suggesting new molecular targets for improving the delivery of oncolytic viruses and the outcome of this novel therapy for tumors.


Manabu Hatano, Tomohide Tatsumi, Naruo Kuwashima, Jill E. Dusak, Ian F. Pollack, Walter J. Storkus, and Hideho Okada; Departments of Neurological Surgery and Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania, USA

EphA2 is a receptor tyrosine-kinase that is frequently overexpressed in a wide array of advanced cancers, in which it contributes to many different aspects of malignant character, such as development of disseminated disease and metastatic lesions. High levels of EphA2 expression have also been observed in tumor-vascular endothelial cells. Our previous studies identified cytotoxic and helper T lymphocyte (CTL and HTL, respectively) epitopes in human EphA2 (hEphA2) in the context of HLA-A2 and DR4 as well as in mouse EhpA2 (mEphA2) in H-2Db, Kb, or I-Ab restriction. Immunohistochemical analyses with primary human glioma tissues demonstrated that a majority (9 of 14) of glioblastoma multiforme and 3 of 9 anaplastic astrocytoma cases expressed high levels of EphA2 in the tumor cells. In contrast, neither low-grade astrocytomas (4 cases) nor normal brain specimens (4 cases) expressed detectable levels of EphA2. Interestingly, endothelial cells within high-grade gliomas expressed high levels of EphA2, even in cases in which glioma cells did not express detectable levels of EphA2. To address whether vaccinations with mEphA2-derived epitopes induce anti-tumor and/or anti-tumor-angiogenic effects in vivo, C57BL/6 mice were immunized with syngeneic bone marrow-derived dendritic cells (DCs) loaded with H-2Kb-presented mEphA2(671–679) CTL and I-Ab-binding mEphA2(30–4) HTL epitopes. These mice mounted potent and antigen-specific CTL responses in their splenocytes and rejected a challenge with syngeneic, mEphA2-positive MCA205 sarcoma cells. Interestingly, therapeutic vaccinations with DCs loaded with the mEphA2-derived peptides in mice bearing lung metastases of mEphA2-negative B16-F10 melanoma resulted in dramatic reduction of metastatic tumors in comparison to the mock-vaccinations with irrelevant peptides. Subsequent in vivo Matrigel assays revealed that vaccinations with the mEphA2-derived CTL and HTL epitopes inhibited vascular endothelial growth factor (VEGF)-induced neoangiogenesis. We also examined whether vaccinations with autologous glioma cells elaborating transgene-derived interleukin-4 (IL-4) could elicit a specific response against hEphA2 in an HLA-A2+ patient who demonstrated a temporary clinical response to the vaccinations. Enzyme-linked ImmunoSPOT (ELISPOT) assays displayed elevated, specific IFN-gamma responses against HLA-A2-restricted hEphA2(883–891) following the vaccinations. Taken together, these data suggest that EphA2-derived T cell epitopes may serve as attractive targets as well as surrogate markers in glioma vaccine strategies by inhibiting both tumor growth and tumor neovascularization.


Michael C. Jensen,1 Adam Mamelak,2 Julie Ressler,3 and Jana Portnow4; 1Departments of Pediatric Hematology/Oncology, 2Neurosurgery, 3Radiology, and 4Medical Oncology, City of Hope National Medical Center, Duarte, California, USA

The genetic engineering of cytolytic T lymphocytes (CTLs) for redirected glioma recognition is a conceptually attractive strategy for acquiring effector cells for adoptive therapy. To this end, we have designed the IL13-zetadine chimeric antigen receptor composed of an extracellular IL 13 (E13Y) mutein targeting domain and a cytoplasmic signaling domain of CD3-zeta. Our pre-clinical studies demonstrate that human CTLs genetically modified to express the IL13-zetakine display IL13Ralpha2-specific redirected glioma cell cytolysis, zetakine-regulated Tc1 (IFN-g, TNF-alpha, IL-2) cytokine production, and helper-independent proliferation. In vivo models demonstrate the capacity of these CTLs to home to glioma xenografts and mediate the regression of established intracranial GBM xenografts. Based on these findings, a pilot feasibility study (CPHNMC IRB#01020/BB-IND#10109) was initiated. Patients enrolled following frontline initial resection and radiation therapy are leukapheresed, and autologous IL 13-zetakine/HyTK+CD8+CTL clones are isolated, expanded, and cryopreserved. Patients with localized tumor at first recurrence undergo repeat cytoreductive surgery and intracavitary Rickham shunt reservoir placement. After recovery and tapering of decadron, patients begin adoptive therapy. Patients receive 12 cell doses over 5 weeks escalating from an initial cell dose of 10 × 10e6 to the target cell dose of 100 × 10e6. Patients are studied by serial MRI/PET scans, while resection cavity CSF is analyzed for molecular markers of T cell activation, apoptosis, and tumor response. To date, T cell products making all release criteria have been generated on each of the five study participants. Adoptive therapy of the first relapsing patient has commenced. The tolerability of T cell dosing to the target of 100 × 10e6 cells was excellent. Correlative CSF and imaging studies are in progress, and the results of these studies will be presented.


Terry Lichtor,1 Roberta Glick,1 Elika Derek,1 InSug O-Sullivan,2 and Edward Cohen2; 1Department of Neurosurgery, Rush University/John H. Stroger Hospital of Cook County, Chicago, Illinois; 2Department of Microbiology and Immunology, University of Illinois at Chicago, Chicago, Illinois; USA

Previous studies have revealed that transfection of genomic DNA can stably alter both the genotype and phenotype of the cells that take up the exogenous DNA. The goal of this study was to determine the effectiveness of a DNA-based vaccine on the treatment of mice with an intracerebral (i.c.) malignant tumor. To investigate this question, we transfected LM mouse fibroblasts (H-2k) modified to secrete either IL-2, IL-18, or GMCSF with genomic DNA from a breast adenocarcinoma (SB5b) that arose spontaneously in a C3H/He mouse (H-2k). To increase their nonspecific immunogenic properties, the fibroblasts were also modified before transfection to express allogeneic MHC determinants (H-2Kb). These fibroblasts were then tested in C3H/He mice for their immunotherapeutic properties against the growth of an i.c. breast neoplasm (SB5b). Secretion of cytokines was confirmed by an ELISA assay. The results demonstrate that mice injected i.c. with a mixture of breast carcinoma and IL-2-secreting semi-allogeneic fibroblasts transfected with tumor DNA survived significantly longer (P < 0.005) than control animals injected with nonsecreting fibroblasts. No prolongation in survival was noted in the animals treated with IL-18- or GMCSF-secreting fibroblasts transfected with tumor DNA. An MTS proliferation assay using spleen cells taken from the animals two weeks following the i.c. injection of tumor and treatment cells revealed that a systemic antitumor response did develop in the animals treated i.c. with either IL-2 or GMCSF-secreting semi-allogeneic fibroblasts. An ELISPOT assay detecting INF-γ-producing cells done at six weeks demonstrated the presence of peripheral spleen cells active against SB5b breast carcinoma in only the animals treated with IL-2 secreting fibroblasts. The effect of monoclonal antibodies for T-cell subsets or NK/LAK cells on the anti-tumor response detected by the ELISPOT assay was studied to identify the predominant cell types activated, and the results indicate that the cellular anti-breast carcinoma immune response was mediated by CD4+, CD8+, and NK/LAK cells. In summary, mice bearing an i.c. breast cancer immunized with IL-2-secreting semi-allogeneic fibroblasts transfected with tumor DNA survived significantly longer than mice in control groups. Furthermore a systemic anti-tumor immune response was documented after two weeks in animals treated with fibroblasts that secrete either IL-2 or GMCSF and at six weeks only in the animals treated with IL-2-secreting cells.


Gentao Liu, Yasuharu Akasaki, Christopher J. Wheeler, Asha Das, Keith L. Black, and John S. Yu; Maxine Dunitz Neurosurgical Institute, Cedars-Sinai Medical Center, Los Angeles, California, USA

Tyrosinase-related protein (TRP)-2 is an immunogenic antigen in melanoma. Our previous study has demonstrated that TRP-2 was not only expressed on glioma cells, but also naturally processed and presented by their surface MHC molecules. TRP-2 specific CTL clone could recognize GBM cells in an HLA class I manner, suggesting that TRP-2 is an immunotherapeutic target in human malignant glioma. In this study, stable overexpressing TRP-2 GBM cell lines were established after TRP-2 expression plasmid was successfully transfected into the GBM cell line U-373. TRP-2 transfected clones had a significantly increased drug resistance to carboplatin and Temodar compared to wild-type U-373. However, there was no significant difference in expression of other common drug resistance genes after TRP-2 transfection, such as MDR-1, MRP-1, MRP-2, MRP-3, BCRP-1, and MGMT. TRP-2-specific cytotoxic T cell activity was detected in patients’ PBMC after DC-pulsed autologous tumor lysate vaccinations. Tumor cells derived from postvaccination primary cell cultures showed significantly lower TRP-2 expression and higher drug sensitivity (i.e., lower drug resistance) to carboplatin and Temodar than those autologous cell lines derived from pre-vaccination surgical specimens of two patients. Clinical responses were also observed in patients after dendritic cell–pulsed tumor lysate vaccinations followed by chemotherapy. Our study demonstrated that TRP-2 is a novel drug resistance–related protein in glioma. For the first time, we provide evidence that targeting tumor-associated antigen TRP-2 by immunotherapy could significantly increase the chemotherapeutic sensitivity and shed light on the molecular mechanism of synergic effects of immunotherapy on cancer chemotherapy.


Neal Luther, Erika B. Mark, Will J. Ryder, Justin F. Fraser, Nai-Kong V. Cheung, Philip H. Gutin, and Mark M. Souweidane; Departments of Neurosurgery and Pediatrics, Memorial Sloan-Kettering Cancer Center, New York, New York, USA; Department of Neurosurgery, Weill Medical College of Cornell University, New York, New York, USA

Monoclonal antibody (MAb)-mediated immunotherapy has potential in the treatment of gliomas. The MAb 8H9 (Modak et al., 2001, Cancer Res. 61:4048) has demonstrated reactivity with a number of primary human brain tumors and minimal cross-reactivity with normal brain cells. Immunotherapy of invasive brain tumors is contingent on a mode of delivery that bypasses the blood brain barrier to achieve high local antibody concentration and optimal distribution in tumor and normal brain parenchyma. Interstitial infusion is a method of local drug administration that utilizes a pressure-dependent gradient to enhance distribution and uniformity of therapeutic agents in the brain. We use the preclinical rat model to optimize the distribution of MAb 8H9 following interstitial infusion. Stereotactically guided interstitial infusion of biotinylated 8H9 into the rat striatum was performed. Dose, volume, and rate of 8H9 infusion were titrated individually. Animals were examined for evidence of clinical neurotoxicity and were subsequently sacrificed at 1 h, 24 h, or 72 h following 8H9 infusion to determine kinetics of MAb distribution. Following sacrifice, brains were sectioned on a cryostat, reacted with streptavidin-peroxidase, and visualized after incubation in diaminobenzidine. Areas of 8H9 distribution were digitized and measured using ImageJ software (NIH). No rats experienced significant clinical neurotoxicity following interstitial infusion of MAb 8H9. Preliminary results suggest 8H9 dose of infusion is the key variable that determines volume of distribution. Even at medium dose of 8H9 (1.9 μg), favorable volumes of distribution were achieved (mean volume of 54.3 mm3), peaking at 24 h following infusion. Significant volumes of distribution of 8H9 can be safely achieved in the rat brain via interstitial infusion. Preliminary results further suggest that dose of antibody is most predictive of volume of distribution. The distributive properties of 8H9 within naive brain parenchyma support the potential utility of this MAb as a conjugate for therapeutic purposes.


A.B. Madhankumar and Waldemar Debinski; Brain Tumor Center of Excellence, Wake Forest University School of Medicine, Winston-Salem, North Carolina; Department of Neurosurgery, Penn State College of Medicine, Hershey, Pennsylvania, USA

We have found that a vast majority of patients with high-grade astrocytoma (HGA) overexpress IL13Rα2, a receptor (R) for an immune regulatory cytokine, interleukin 13 (IL13). The other receptor for IL13, IL13/4R, is shared with IL4, transduces intracellular signaling, and is expressed comparably in both vital organs and human malignancies. In addition, IL13Rα2 can be induced in normal cells otherwise not expressing this tumor-restricted receptor, in response to IL13 itself. The cytokine stimulates a signaling cascade in cancer cell that leads to further activation of oncogenic pathways. Thus, in the development of diagnostics/therapeutics using IL13Rα2 as a target, IL13 should be deprived of an ability to bind and activate IL13/4R and preferably bind tighter the targeted tumor-associated receptor. Our laboratory documented that residues at positions 105, 106, and 109 in alpha-helix D of IL13 contribute to the binding of IL13 to IL13Rα2. We have begun to determine the structural requirements of IL13 to interact with IL13Rα2. In Conolly’s algorithm surface analysis, we found Lys-105 and Arg-109 lay one above the other with a depression in between the residues while Lys-106 lies farther away. It is plausible that the receptor-binding pocket for IL13Rα2 encompasses this specific region. Next, we analyzed the structural side-chain changes at position 105 upon substitution with various amino acids. We noted a proximity between the side chain –NH2 group of Lys-105 and –C (NH2)=NH- group of Lys-109, and we suggest that such orientation may in fact favor the residues to anchor the receptor site coordinately. All position 105 mutants of IL13 studied, namely IL13.K105E, IL13.K105D, IL13.K105A, and IL13.K105L, do not have the extended side-chain and none of them competes for the IL13Rα2 binding. Accordingly, a mutant of IL13, IL13K105R, has the arginine side chain protruding in parallel to that of Lys-109 and competes for the IL13Rα2 sites even better than the wild-type IL13. Another possible explanation for Arg-105 substitution enabling preserved or even better binding to IL13Rα2 is that the epsilon-N-group in Arg has a restricted rotation because of the partial double bond character due to resonance, and Arg is able to interact better with the backbone carbonyl oxygens, providing structural rigidity to the protein. We suggest that in IL13.K105R, Arg increases the rigidity of the protein and influences the overall electrostatic potential of the protein surface. Alternatively, the arginine residue may participate in the production of hydrogen bonds with the receptor’s residues. We thus propose that a positively charged polar residue with an extended side-chain orientation at position 105 in the alpha-helix D region in IL13 may be crucial for its anchoring property toward the HGA-associated receptor. This knowledge will allow us to generate other IL13 mutants/peptidomimetics of desired characteristics for future development of specific tumor targeting molecules.


Duane A. Mitchell, Chris Learn, Robert Schmittling, Allan Friedman, Roger E. McLendon, Darell D. Bigner, and John H. Sampson; Division of Neurosurgery, Departments of Surgery and Pathology, Duke University Medical Center, Durham, North Carolina, USA

A major impediment to the development of effective cancer immunotherapy is the lack of the identification of strong tumor rejection antigens present within human cancers. The recent finding that human cytomegalovirus (HCMV) antigens are expressed within a high proportion of malignant gliomas, but not within the surrounding normal brain, offers an unparalleled opportunity to direct brain tumor immunotherapy against well-characterized and highly immunogenic viral targets. We have confirmed and extended the identification of HCMV antigens in malignant glioma specimens using immunohistochemistry, in situ hybridization, direct immunofluorescence assay, and PCR on frozen and paraffin-embedded tissues, freshly resected glioma specimens, and tissue microarrays. Furthermore, we have demonstrated that tumor infiltrating lymphocytes cultured from glioma specimens, in some cases, contain a high proportion of CMV-specific T cells (up to 40% of infiltrating CD8+ T cells) suggesting an ongoing, although insufficient immunologic response against viral antigens expressed within the tumor cells. Detection of the immunodominant CMV antigens pp65 and IE1 is widespread throughout malignant astrocytes, with little to no detection in surrounding normal brain, suggesting that HCMV antigens may provide novel and potent targets for immunotherapeutic intervention. Ongoing efforts in the laboratory are focused on the development of dendritic cell vaccination and adoptive T cell therapy protocols targeting HCMV antigens expressed within malignant gliomas.


Shaik Ohidar Rahaman,1 Michael A. Vogelbaum,1,2,3 and S. Jaharul Haque1,2; 1Department of Cancer Biology, Lerner Research Institute, 2Brain Tumor Institute, and 3Department of Neurosurgery, Cleveland Clinic Foundation, Cleveland, Ohio, USA

IL-4 gene therapy induces a significant anti-tumor immune response in animals bearing experimental gliomas. This is mediated by the action of IL-4 on various immune effector cells present in and around the tumor mass. Glioblastoma multiforme (GBM) cells, like normal human astrocytes (NHAs) express the type II IL-4 receptor that is activated by both IL-4 and IL-13. Therefore, we wish to study the direct action of IL-4 on GBM cells in vitro. IL-4 normally activates Stat6 in most cell types including NHA but not in GBM cells (Rahaman et al., 2002, Cancer Res. 62:1103). GBM tumors and cell lines are found to contain persistently activated Stat3 (Rahaman et al., 2002, Oncogene 21:8404). Here we report that IL-4 further increases the level of activated Stat3 in GBM cells. IL-4 does not activate Stat3 in NHA. An optimal level of Stat3 activation is observed in GBM cells within 30 min of IL-4 treatment (10–25 ng/ml), which sustains for 48 h. Furthermore, overexpression of the IL-4Ra transgene in T98G cells further increases the Stat3 activation in response to IL-4. Previously, we have shown that overexpression of IL-13Ra2, a decoy receptor for IL-13, inhibits IL-4/IL-13-dependent activation of Stat6 in GBM cells. Now we have found that inhibition of IL-13Ra2 expression by siRNA markedly reduces the IL-4-mediated activation of Stat3 in U251 cells. Conversely, transient expression of the IL-13Ra2 transgene in 293T cells increases IL-4-mediated Stat3 activation and subsequent gene expression. Chromosomal immunoprecipitation assay using anti-Stat3 antibody shows the in vivo binding of activated Stat3 to the promoter regions of genes that encode anti-apoptotic proteins Bcl-2, Bcl-XL, and Mcl-1. It is noted that long-term treatment of IL-4 causes significant upregulation of Bcl-2, Bcl-XL, and Mcl-1 in GBM cells. These results suggest that IL-4 increases the level of Stat3 activation and thus contributes to the pathogenesis of GBM cells by upregulating the Bcl-2-family of anti-apoptotic proteins expressed in GBM cells.


L.G. Salford,1 P. Siesjö,1 G. Skagerberg,1 E. Visse,1 A. Darabi,1 Å. Lilja,1 C. Blennow,1 S. Strömblad,1 E. Ask,2 A. Rydelius,2 B.R.R. Persson,3 C. Bauréus Koch,3 E. Englund,4 E-M. Larsson,5 N. Mandahl,6 and B. Widegren7; 1Departments of Neurosurgery (The Rausing Lab), 2Neurology, 3Radiation Physics, 4Neuropathology, 5Neuroradiology, 6Clinical Genetics, and 7Tumour Immunology, Lund University, Lund, Sweden

In our project to treat malignant gliomas by immunisation with autologous glioma cells, transduced with interferon-gamma by adenoviral vector, we have completed the treatment of eight patients and have immunised the ninth patient six times. Our goal is to ascertain whether this immunisation is safe for the patients, gives rise to an immunological response, and adds any beneficial effect to conventional therapy. The study is approved by the Swedish Medical Products Agency for inclusion of a maximum of 20 patients. Inclusion criteria are as follows: glioblastoma multiforme, age 50–69 years, tumour operable, chance for >80% reduction of tumour burden, patient’s consent, Karnofsky >70 preoperatively, and RT the only other treatment after surgery. The glioma cells grown in our Transduction Laboratory are analysed for their karyotypes 14 days after the operation and also at later stages. When a stable growth of malignant cells is established, the cells are harvested. Every immunisation utilises 2 million cells. Each patient is immunised with three-week intervals on at least at four and at most at ten occasions. The transduction of the human IFN-gamma genes to the tumour cells is performed the day before immunisation. At the day of immunisation, the cells are irradiated with X rays, 100 Gy. Seven days after each immunisation, a skin biopsy is taken from the centre of one of the injection sites. Peripheral blood is sampled both before and after operation and also after each immunisation event. Co-culture of this blood with tumour cells from the patient allows for a selection of T cells that can recognise tumour-specific antigens. The patients are followed regularly with neurological and neuropsychological examinations, as well as with MRI examinations including perfusion-weighted MRI, measuring cerebral blood volume. Results from the first human treatments show that the method is safe for the patients and that it gives rise to positive DTH reactions and an increase of infiltrative CD8+ and CD4+ T cells at the site of immunisation. Further, it seems as if the immune response is biphasic, with a maximum at the third to fourth immunisation, a minimum at the fifth to sixth, and another maximum at the seventh to ninth immunisation. It is not possible to conclude from this small material whether the method adds any beneficial effect. It may, though, be noted that two patients who received the full treatment with 10 immunisations survived for 27 months and another for 22 months after the pathological-anatomical diagnosis. One patient has no MR signs of remnant tumour 15 months after diagnosis.


Jill M. Schartner, Aaron Hagar, Michelle Van Handel, and Behnam Badie; University of Wisconsin Madison, Madison, Wisconsin, USA

Understanding the local CNS immune response to neoplasms is essential in the development of novel immunotherapeutic treatments for malignant brain tumors. Using animal glioma models, we have recently found that tumor-associated microglia/macrophages (MG/MPs) are less responsive to activators than normal brain MG/MPs. To determine the mechanism for this impairment, real-time quantitative PCR was used to evaluate the expression of both pro- and anti-inflammatory cytokines in normal and RG2-bearing rat brains. Message for several pro-inflammatory cytokines, such as TNF-α, interleukin-12, interferon-γ, and interleukin-1, were significantly upregulated in intracranial RG2 tumors as compared to normal brains (5-fold, 50-fold, 100-fold, and 23-fold increases, respectively). Interestingly, the expression of mRNA for anti-inflammatory cytokines was also increased in RG2 tumors. Message for interleukin-10 was increased by more than 1500-fold while TGF-B1 signal was increased by more than 50-fold in tumors as compared to normal brains. Further analysis of sorted MG/MPs from RG2 tumors revealed significant mRNA expression of interleukin-10, but no expression of interferon-γor TNF-αby these cells. These observations suggest that tumor-associated MG/MPs may contribute to the local immune suppression in malignant gliomas.


Robert Schmittling, Duane A. Mitchell, Gary Archer, and John H. Sampson; Departments of Surgery and Pathology, Duke University Medical Center, Durham, North Carolina, USA

Glioblastomas are typically fatal, and conventional therapies lack efficacy. It was recently discovered that malignant gliomas (MGs), but not surrounding normal brain tissue, serve as a refuge for Cytomegalovirus (CMV) and express the immunodominant CMV-specific antigen pp65. Given that adoptive transfer of pp65-specific CD8+ T cells has previously been shown to be safe and efficacious in controlling CMV disease in bone marrow transplant patients, we sought to evaluate pp65 as an immunotherapeutic target in MGs. Using autologous dendritic cells (DCs) electroporated with pp65mRNA or pulsed with pp65 peptide, T cells could be enriched such that on average 75.6% (n = 10; range, 36.4%–98.3%) of the CD8+ T cell population became pp65-specific, and the total number of pp65-specific CD8+ T cells increased on average 3.1 log (range, 1.7–4.0 log; n = 10) within only 2 weeks. Despite downregulation of antigen presentation by CMV, expanded T cells from patients with MG were capable of specific lysis of CMV-infected MG cells as well as pp65-pulsed T2 cells. In addition, when these T cells were reacted against autologous pp65-pulsed DC or autologous tumor, they specifically released IFN-γ as detected by ELISPOT and intracellular flow cytometry assays. Our data indicated that pp65 is potential target antigen for immunotherapy of MGs. We demonstrate the ability to activate and expand CMV-specific T cells from the patients with MGs using autologous DCs electroporated with pp65 mRNA or pulsed with pp65 peptide. Furthermore, CMV-specific T cells can be expanded to clinically relevant levels for adoptive immunotherapy. These studies provide a rationale for the clinical development of an adoptive T cell therapy targeting CMV antigens expressed in MGs.


Stephanie Shoaf, Elizabeth Grimm, Raymond Sawaya, and Amy Heimberger; University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA

Microglia are bone marrow–derived immune cells in the CNS that can be activated by various inflammatory and pathological conditions into macrophage-like phagocytes. During the growth of brain tumors, microglia infiltrate the tumor, but despite the increased number of microglia present in CNS gliomas, these cells do not eradicate the tumor or suppress growth. The goal of this study was to determine the immune phenotype and function of microglia within human gliomas and to determine the role microglia may play in perturbing the mechanisms of immune responses against gliomas. Microglia have been previously characterized in murine model systems and perpetuated cell cultures. However, these model systems may not appropriately recapitulate the immune phenotype and function of microglia within human gliomas. Therefore, we have immediately isolated glioma-infiltrating microglia (GIM) from human anaplastic astrocytomas (WHO grade III) (n = 2) and glioblastoma multiforme (WHO grade IV) (n = 6) by sequential Percoll gradient purification and determined their immunological phenotype and functional capabilities. The GIM were purified to at least 80% based on established parameters (CD45low CD11b/c+). Lymphocytes, NK cells, and dendritic cells were not found to be substantially contributing contaminating cell populations. An average of 2.6 × 106 GIM (range, 1.3 × 105–8.2 × 106) were obtained per gram of tumor. The GIM in the subG0/G1 (apoptosis) compartment were in the range of 21%–46%. Of the viable cells, usually >77% were in the G1 compartment, with a range of 4%–11% and 3%–12% in the G2 and S compartment, respectively. In comparison, microglia isolated from “normal” tissue had <13% in the subG0/G1 compartment. By flow analysis cytometry, greater than 88% of microglia express MHC class II. However, the microglia have either negligible or low levels of the co-stimulatory molecules CD80, CD83, and CD86. The co-stimulatory inhibitor molecule CTLA-4 did not appear to be upregulated on GIM. Cumulatively, this data would suggest that GIM could induce anergy to tumor infiltrating lymphocytes. In contrast to previous reports, the expression of FasL was low, but variable, indicating that inducing apoptosis to the incoming lymphocyte population may not be a predominant mechanism of immune resistance. The GIM express FcR I and FcR II, which is necessary for antibody-dependent cellular cytotoxicity (ADCC), and we are currently evaluating whether the GIM can functionally mediate ADCC immediately ex vivo. Intracellular cytokine staining for IL-1, IL-2, IL-4, IL-6, IL-10, IL-12, gamma-interferon, and TNF-alpha was performed, and IL-12 most consistently produced cytokine. Studies addressing whether GIM can mediate tumor cytotoxicity and phagocytosis are ongoing. Ultimately, by understanding the immunological function of GIM, or the lack thereof, rational, viable therapeutic strategies can be designed that may be beneficial for glioma patients. Future studies will examine the effects of radiation and chemotherapy on GIM.


Andrew Edward Sloan1 and Prahlad Parajuli2; 1Departments of Neurosurgery and Interdisciplinary Oncology, Moffitt Cancer Center, Tampa, Florida; 2Department of Neurosurgery, Wayne State University, Detroit, Michigan; USA

The failure of conventional chemoradiotherapy to improve survival of patients with malignant glioma has led to renewed interest in alternative therapeutic modalities including cellular and humoral immunotherapy. Most pre-clinical and clinical studies of cellular immunotherapy have endeavored to activate CD3+CD8+ cytolytic T cells (CTLs) as effector cells. However, CTLs fail to destroy tumor cells that have lost MHC class I expression, as do most gliomas and other tumors. Recently, T cells bearing the NK phenotype (CD3+CD56+), known as NK-like T cells (NK-T), have been identified with the potential to induce non-MHC I–restricted, tumor-specific cytolysis. Here we describe the induction and in vitro efficacy of NK-T for immunotherapy of human malignant glioma. Dendritic cells (DCs) prepared from autologous or MHC-matched PBL were fused with tumor (GBM) or pulsed with whole-tumor RNA, DNA, apoptotic bodies, or freeze-thaw lysate. After in vitro maturation, phenotype and cytokine profile and induction of CTL and NKT and their anti-tumor efficacy were analyzed with FACS, ELISA, and cytotoxicity assays. DC/Apo induced optimal stimulation of NK-T cells compared to the other methodologies. NK-T stimulated by DC/Apo or DC/RNA were equal in numbers and had significantly greater tumor-specific cytotoxicity compared to CTL induced by these stimulation methodologies and were non-MHC restricted. Glioma-specific NK-T cells induced by stimulation with DC/Apo are equal in number and greater in anti-tumor cytotoxicity than CTL. These non-MHC I–restricted effector cells may represent the most effective strategy for induction of a glioma-specific cellular immune response.


Omar N. Syed,1 Christopher Mandigo,1 Danila Valmori,2 Achim Jungbluth,2 Peter Canoll,3 and Jeffrey N. Bruce1; 1The Bartoli Brain Tumor Laboratory, Neurological Institute of New York, Department of Neurosurgery, and Departments of 2Medical Oncology and 3Neuropathology, Columbia University College of Physicians and Surgeons, New York, New York, USA

Immune-based treatment paradigms rely on the identification of tumor-specific antigens as a crucial and necessary step in developing future immunotherapeutic strategies for both benign and malignant brain tumors. We analyzed the expression of cancer-testis antigens (CTAs) and melanocyte-differentiation antigens (MDAs) in malignant gliomas, meningiomas, and normal brain, as there are a number of antigens from both groups that are well understood and have been clinically effective in anti-cancer vaccination protocols against non-CNS tumors. cDNA was prepared by reverse transcription of total RNA extracted from 44 frozen GBMs, 20 frozen meningiomas, and 9 normal brain frozen tissue specimens. The following primers were analyzed for CTAs by PCR amplification: LAGE-1, NY-ESO-1, MAGE-1, MAGE-3, MAGE-4, MAGE-10, CT-7, CT-10, HOM-MEL 40, BAGE, and SCP-1; and MDAs: tyrosinase, gp100, MELAN-A/MART-1, and TRP-2. The expression level was determined by ethidium bromide-stained agarose gel. CTA expression in GBMs was as follows (positive/total tumors tested): 2/41 for LAGE-1, 2/41 for NYESO-1, 7/44 for MAGE-1, 10/44 for MAGE-3, 4/43 for MAGE-4, 2/43 for MAGE-10, 5/44 for CT-7, 4/44 for CT-10, 1/41 for HOM-MEL 40, 1/42 for BAGE, and 0/42 for SCP-1. MDA expression in GBMs: 0/24 for tyrosinase, 12/24 for gp100, 1/24 for MELAN-A/MART-1, and 10/24 for TRP-2. CTA expression in meningioma tissue was as follows: 3/20 for NY-ESO-1; all of the other CTAs were undetectable in meningioma frozen tissue. MDA expression in meningiomas was as follows: 0/20 for MELAN-A/MART-1, 8/20 for gp100, 0/20 for tyrosinase, and 10/20 for TRP-2. Of the 9 normal brain specimens, 9/9 samples tested positive for TRP-2. All other CTAs and MDAs were undetectable in normal brain. The CTAs and MDAs demonstrate low and variable levels of expression within GBMs. However, the CTAs, MAGE-1, MAGE-3, and gp100, are demonstrated in a significant percentage of GBMs, and may be considered candidate antigens based on their restricted expression in GBMs. With the exception of low levels of expression of NY-ESO-1 and gp100, the CTAs and MDAs are not expressed in meningiomas. These specific antigens are very well defined tumor antigens, and existing vaccination protocols for these antigens may be applicable to a subset of glioma and meningioma patients.


Reena P. Vishwanath,1 Christine E. Brown,1 Brenda Aguilar,1 Joseph Najbauer,2 Karen S. Aboody,2 and Michael C. Jensen1,3; 1Division of Molecular Medicine; 2Division of Hematology and Hematopoietic Cell Transplantation; 3Department of Pediatrics Hematology-Oncology, Beckman Research Institute, City of Hope National Medical Center, Duarte, California, USA

Initial clinical trials for malignant glioma employing adoptive transfer of glioma-specific cytolytic T lymphocytes (CTLs) have commenced at City of Hope. Tumor homing is a critical parameter that will impact efficacy. Here we report our findings of studies evaluating T-cell trafficking and tropism to tumor cells of malignant glioma. In vitro chemotaxis assays employing firefly luciferase positive (ffLuc+) CTLs reveal chemotactic activity to conditioned media from glioma cell lines and resection cavity CSF. By comparison, ex vivo expanded CTLs did not migrate in response to conditioned media from human 10HTB neuroblastoma or human Daudi lymphoma cell lines. Chemokine expression profiles and chemokine immunodepletion analysis of these three tumor types identified several chemokines produced in high levels by glioma, but not 10HTB or Daudi. We identify MCP-1 (CCL2), a CC-chemokine that is highly expressed by all glioma tumors thus far tested, as playing the predominant role in chemoattracting effector CTLs, with an ED50 of about 10 ng/ml. Using bioluminescent human lymphocytes to track T-cell migration in vivo with the Xenogen imaging system, we demonstrate that adoptively transferred CTLs home to flank and intracranial human glioma xenografts. Within 72 h, intravenously administered ffLuc+ T-cells preferentially co-localized to palpable U251T flank tumors with a 7- to 8-fold increase in luminescent intensity, which persisted for greater than 14 days when compared to the control flank. Similarly, trafficking of T cells to intracranial U251T tumor was observed 48 to 72 h post intravenous injection with a co-localization that persisted for greater than 5 days. Daudi tumors engineered to express MCP-1 were proficient at chemoattracting CTLs both in vitro and in vivo, demonstrating that MCP-1 is sufficient to modulate T-cell trafficking to the tumor microenvironment. T-cell infiltration into the intracranial tumor mass in vivo was verified by immunohistochemistry. Our studies highlight cytokine profile differences between tumor types that influence the innate ability of T-cells to infiltrate a tumor microenvironment, a critical parameter influencing adoptive immunotherapeutic success.


Friedrich Weber (for the TP-38 study group); Department of Neurosurgery, Klinikum Saarbrucken, Saarbrucken, Germany

TP-38 is a recombinant chimeric protein composed of the epidermal growth factor receptor (EGFR) binding ligand (TGF-α) and a genetically engineered form of the Pseudomonas exotoxin PE-38. We report preliminary results of a randomized, phase II study conducted at multiple European centers. Either of two dose levels of TP-38 (50 ng/ml or 100 ng/ml) was administered to patients with recurrent glioblastoma in a single treatment consisting of a continuous, intratumoral infusion. This was a nonresection study; patients did not undergo tumor resection immediately prior to treatment. Three catheters were stereotactically placed in investigator-determined locations within the enhancing tumor area. The infusion rate was 0.2 ml/min per catheter. Each catheter delivered 13.4 ml over 67 h. The total volume infused was approximately 40 ml, and the total dose of TP-38 infused was approximately 2 μg or 4 μg. Patients were followed until death. Tumor responses were assessed by MRI at every 8 weeks, beginning 4 weeks after treatment. Safety was closely evaluated. Time to progression, progression-free survival, and overall survival were measured end points. Thirty-four of the planned 38 patients have been treated thus far. Safety and tolerability has been excellent. The 4-week and 12-week post-infusion MRI scans often showed treatment-related changes that make response assessment difficult. These changes usually resolved by the 20-week post-treatment MRI. Preliminary data show that one patient has a near-complete response 12 weeks after infusion, and another maintains stable disease 28 weeks after infusion. Efficacy parameters such as survival are still under analysis and will be available in autumn 2004.


Christopher J. Wheeler, Cedars-Sinai Medical Center, Maxine Dunitz Neurosurgical Institute, Los Angeles, California, USA

Conventional therapies are ineffective at improving clinical outcome for the overwhelming majority of glioblastoma multiforme (GBM) patients despite their ability to confer significant benefits to patients with nonglioma tumors. Thus, novel therapies or means of substantially enhancing conventional therapies for GBM are needed. Cancer vaccines represent a novel GBM therapy whose clinical efficacy remains controversial. Evidence consistent with vaccine-mediated enhancement of chemotherapeutic responsiveness, however, has recently been presented. Such putative enhancement appeared to be confined to the younger cohort of GBM patients. Since chemotherapeutic responsiveness is associated with discrete genetic alterations in gliomas, we examined the potential of vaccine therapy to reproducibly induce relevant genetic changes in recurrent GBM tumors. Analysis of tumor RNA obtained pre- and post-vaccine from the same six patients revealed a highly convergent pattern of gene expression upon vaccination that was evident in whole-genome analyses. Specifically, post-vaccine tumor samples exhibited higher similarity to gene expression profiles in pre-vaccine samples than all pre-vaccine samples did among themselves, and exhibited highest similarity to other post-vaccine samples. This suggests that vaccination elicits a stereotypical pattern of expression across a large set of genes. Similarity to a stereotypical post-vaccine pattern of expression in 1701 genes predicted overall patient survival in 59 non-vaccinated GBM samples from an independent gene expression database (UCLA Microarray Core; P < 0.02, Wilcoxon), and correlated (r = 0.67; P = 0.05) with levels of a T cell subset (CD8+ recent thymic emigrants) associated with both GBM destruction and chemotherapeutic responsiveness in the vaccinated patients analyzed. Chromosomal analysis using LOH revealed a similarly convergent post-vaccine pattern. In both microarray and LOH analyses, unique age-dependent alterations that paralleled chemotherapeutic responsiveness were evident in post-vaccine samples. For example, tumors from all young patients (the group predominantly responding to chemotherapy post-vaccine) exhibited LOH at 1p36 post-vaccine, a genetic abnormality strongly associated with chemotherapeutic responsiveness in oligodendrocytic and astrocytic gliomas. In contrast, <25% of samples from older and/or pre-vaccine patients exhibited this event. We conclude that vaccination can lead to stereotypical genetic changes in GBM tumors that may be exploited by subsequent conventional therapeutic approaches.



Christopher Calabrese, Waleed Gaber, John Killmar, Christine Fuller, Meredith Allen, and Richard Gilbertson; Department of Developmental Neurobiology, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA

We have developed a new in vivo model of medulloblastoma that recapitulates the behavior of classic and large-cell anaplastic (LCA) forms of the disease and allows for serial measurement of tumor growth. To do this we generated green fluorescent human medulloblastoma cells (Daoy, MEB-MED-8A and MHH-MED-1) by in vitro infection using an MSCV-Green Fluorescent Protein (GFP) retrovirus. Under stereotactic control, 1 × 106 GFP-tagged medulloblastoma cells were inoculated into the superficial cerebral cortex of CD1 nude mice, and the cranium was sealed with a glass plate. Serial tumor measurements can be made through the cranial window by using intravital fluorescence microscopy. By comparing in vivo total fluorescence measurements with 3D tumor reconstruction, we show that the measurement of in vivo fluorescence provides a highly accurate measure of tumor burden. MEB-MED-8A tumors grow very rapidly and result in the death of animals within two weeks: Histopathologically, these tumors are LCA, invasive, and contain an isochromosome of 17q and amplification of MYCC. In contrast, MHH-MED-1 and Daoy tumors display classic morphology with focal anaplasia and are much less aggressive in behavior, taking more than 7 weeks to become symptomatic. We are now using these models to test a number of molecular targeted therapies for medulloblastoma. We previously reported ERBB2 to be an independent marker of poor prognosis in medulloblastoma. Daoy.2 and MEB-MED-8A tumors express high levels of the ERBB2 receptor, while MHH-MED-1 cells express ERBB4 and low levels of ERBB2. Using our models we now show that well-tolerated doses (50 mg/kg/bd) of the oral ERBB1 and ERBB2 kinase inhibitor Erlotinib (Tarceva, OSI774) abolish ERBB2 signaling, induce cell cycle arrest, and generate marked inhibition of medulloblastoma growth in the CNS. Furthermore, continuous oral administration of Erlotinib significantly prolonged the survival of tumor-bearing mice relative to those treated with vehicle alone. This model system provides a useful new tool for studying the biology of signal transduction systems in medulloblastomas in the brain, and affords an efficient system for preclinical testing of novel therapies for this disease.


Christopher Calabrese, Roberto Hernan, Charles Sherr, Martine Roussel, and Richard Gilbertson; Department of Developmental Neurobiology, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA

ERBB2 is undetectable in developing and adult human cerebellum (Gilbertson et al., Cancer Res., 1998) but is expressed in 40% of medulloblastomas and is associated with a poor clinical outcome (Gajjar et al., J. Clin. Oncol. 2004). Therefore, aberrant expression of ERBB2 in granule neuron precursor cells (GNPCs) might play a role in the initiation and/or progression of medulloblastoma. To test this hypothesis we have generated transgenic mice in which ERBB2, placed under the control of a 1.7-Kb enhancer element of the Math1 promoter, is expressed at high levels in mitotic GNPC. Mice bearing the Math1-ERBB2 transgene display high-level expression of ERBB2 throughout the neural tube from embryonic (E) day 14.5. High levels of membrane-associated ERBB2 are also detected in the rhombic lip and throughout the mitotic zone of the cerebellar external germinal layer (EGL) from mid-gestation through to postnatal day 20. Western blot analysis of whole postnatal (P) day 5 cerebellum taken from transgenic mice identified high-level expression of phospho-Y1248 ERBB2 and phospho-Ser473 AKT1 relative to transgene negative littermate controls, indicating ERBB2 is actively signaling in mice bearing the transgene. Further, using Affymetrix expression array analysis of P5 cerebellum, we have identified a group of genes that are significantly and differentially expressed in transgene-positive compared to transgene-negative littermate controls. We are now using this model to study the role of ERBB2 in tumorigenesis in the cerebellum. To do this we are expanding our colonies of Math1-ERBB2 animals for tumor surveillance studies. Further, we recently reported that the TP53-ARF pathway is disrupted in large-cell anaplastic medulloblastoma, a tumor subtype that frequently contains high levels of ERBB2. Therefore, to determine whether ERBB2 expression and TP53-ARF defects are interdependent during medulloblastoma formation, we are breeding Math1-ERBB2 mice with p53 null and Arf null animals. Finally, studies of human medulloblastoma indicate ERBB2 may play a role in tumor progression since ERBB2 is often expressed in metastatic forms of the disease. Therefore, to investigate if ERBB2 expression may accelerate medulloblastoma development, we are also breeding Math1-ERBB2 mice with the Ptc1+/− mice that are prone to medulloblastoma. In addition to reporting the characterization of this new mouse model at the meeting, we will also provide a full update on tumor surveillance data.


Sharon Chiappa and Corey Raffel; Neuro-Oncology Research, Department of Neurosurgery, Mayo Clinic College of Medicine, Rochester, Minnesota, USA

CD133 positive neural stem cells have been isolated by FACS from Lig4/p53 heterozygous mice cerebella. Embryonic days 18 and 20, and postnatal days 1, 3, 5, 10, and 20 were included. Neurospheres developed from positive sorted cells from both embryonic samples and from postnatal days 1, 3, and 5. By postnatal day 10, less neurospheres were seen, and they did not proliferate as robustly as those obtained from earlier time points. No neurospheres developed from the postnatal day 20 sort. CD133+ stem cells from the cerebellum of early postnatal C57BL6 (normal) mice and CD133 positive and negative cells from a murine ptc tumor have also been isolated. Following differentiation, stem cells were examined by immunohistochemistry in order to identify the various neuronal cell types. The antibodies used for this study included nestin, for undifferentiated embryonic central nervous system progenitor cells; MAP-2, for neurons; GFAP, for astrocytes; and MBP, for oligodendrocytes. The proportions of neuronal cell types observed in differentiated postnatal day 1 cerebellar stem cells from Lig4/p53 mice were 5% to 10% astrocytes, 60% to 70% neurons, and 10% to 20% oligodendrocytes. Differentiated CD133- tumor cells produced more oligodendrocytes (up to 50%) with a concomitant loss of neurons. The CD133+ tumor cells appear to be composed mostly of neural precursors (up to 80%). However, the CD133+ tumor cells do not adhere well to the poly-l-ornithine substrate used for all differentiation studies and, as a consequence, many cells do not extend substantial neurites. The gene expression profile of normal murine external granule cell layer cells, of normal CD133+ cerebellar stem cells, of CD 133+ ptc tumor stem cells, and of CD133- ptc tumor cells are being compared by using the Affymetrix high-density mouse genome 430 2.0 array. Differential gene expression results will be presented. A comparison of the tumorigenic properties of normal CD133+ cerebellar stem cells and ptc tumor CD133+ stem cells and CD133-negative tumor cells will also be presented.


Daniel A. Lim,1 Mayte Suárez-Fariñas,2 Felix Naef,2 Benedicte Menn,1 Coleen Hacker,3 Marcelo Magnasco,2 Nila Patil,3 and Arturo Alvarez-Buylla1; 1Department of Neurological Surgery, University of California San Francisco, San Francisco, California; 2The Rockefeller University, New York, New York; 3Perlegen Sciences Inc., Santa Clara, California; USA

The adult human brain contains neural stem cells in the subventricular zone (SVZ) of the lateral ventricles, and certain brain tumors may arise from these stem cells. The gene expression controlling adult neural stem cells and their progeny is poorly understood. In the adult mouse, cells born in the SVZ migrate to the olfactory bulb (Ob) where they differentiate into interneurons. To identify gene expression important for SVZ neurogenesis, we performed three complementary sets of transcriptional analysis experiments using Affymetrix GeneChips: (1) comparison of gene expression profiles of the adult mouse SVZ and Ob with those of the striatum, cerebral cortex, and hippocampus; (2) profiling of different subpopulations of SVZ cells isolated by fluorescent-activated cell sorting (FACS); and (3) analysis of gene expression changes during in vivo regeneration of the SVZ after anti-mitotic treatment. We used hierarchical clustering, gene ontology, and principal component analysis to integrate the information from these three approaches to identify specific genes related to SVZ neurogenesis, and to infer cellular processes and signaling pathways that regulate the precursor cell populations. In addition to anticipated cell cycle components, we identified multiple Sox and Dlx genes among a host of other transcription factors with known developmental, stem cell, or oncogenic roles. Genes important for the neurogenic and migratory SVZ-Ob also emerged from the data; for instance, components of the wingless and semaphorin signaling pathways were found. Interestingly, many mRNA splicing factors were present in our analysis, suggesting this as an important process in adult neurogenesis. These transcriptional portraits of the SVZ neurogenic cells merged with the known cellular architecture of this region are an important resource for the molecular characterization of adult neurogenesis and provide an important point of departure for expression studies comparing brain tumors to neural stem cells, their potential origin.


Duane A. Mitchell, Hidenobu Ochiai, Stanley C. Henry, John D. Hamilton, and John H. Sampson; Divisions of Neurosurgery and Infectious Disease, Departments of Surgery, Pathology, and Medicine, Duke University Medical Center, Durham, North Carolina, USA

Human cytomegalovirus antigens were recently reported to be expressed in a high proportion of malignant astrocytomas, offering an unparalleled opportunity to focus brain tumor immunotherapy against highly immunogenic and well-characterized viral targets. In order to investigate the safety and potential efficacy of immunotherapeutic targeting of CMV antigens in malignant brain tumors, we sought to develop a murine model of CMV-associated astrocytoma. Our laboratory has characterized a spontaneously arising murine astrocytoma line (SMA560) that is tumorigenic when implanted intracranially into syngeneic (VMDk) mice. Using a murine CMV recombinant virus expressing GFP under control of an immediate early gene promoter, we have established a persistently infected astrocytoma line (SMA560-CMV) that forms CMV-associated tumors in vivo and is lethal in immunocompetent mice. Tumor growth and viral replication could be visualized in situ using fluorescent microscopic detection of GFP expression, allowing for a novel mechanism of assessing viral growth within intracranial astrocytomas, and providing a simplified platform for analyzing the impact of immunotherapeutic interventions on viral and tumor growth. Mice rendered immune to mCMV by recent viral inoculation were protected from tumor challenge through a T cell dependent mechanism, indicating that malignant astrocytes expressing CMV antigens can be effectively targeted in the CNS by the immune system. The development of this novel model of CMV-associated astrocytoma will be critical in the preclinical evaluation of the safety and efficacy of immunotherapy directed against CMV antigens expressed in the brain, as well as for studies investigating the effects of CMV expression on the malignant properties of brain tumor cells.


Yot Navalitloha,1,2 Cathleen V. Allen,2 Limin Li,3 Alice M. Wyrwicz,1,3 Robert M. Levy,1,4 and Dennis R. Groothuis1,2; 1Northwestern University Institute for Neuroscience, Northwestern University, Evanston, Illinois; 2Department of Neurology, Northwestern University Medical School, Evanston Northwestern Healthcare, Evanston, Illinois; 3Center for MR Research, ENH Research Institute, Evanston, Illinois; 4Department of Neurological Surgery, Feinberg School of Medicine, Northwestern University, Chicago, Illinois; USA

Convection-enhanced delivery (CED) is a local drug delivery method for treatment of brain tumors and other diseases. A noninvasive method for monitoring VD of CED would help clinicians adjust delivery parameters to ensure an adequate distribution of drug in the target area. To develop such a method, gadolinium-diethylenetriamine pentaacetic acid (Gd-DTPA) was tested in in vitro and in vivo models and monitored with MRI. Since the solute size or biochemical properties of the solute are not significant in CED, the distribution of Gd-DTPA and drug molecules should be the same. Following determination of the optimum concentration Gd-DTPA without neurotoxicity, Gd-DTPA in normal saline was infused into Fischer 344 rats to evaluate this monitoring method by different methods: (1) into caudate nucleus with an Alzet minipump (10 μl/h) or an infusion pump (1 μl/min for 60 min), (2) into intracerebral RG-2 tumor with the Alzet minipump or the infusion pump, and (3) into subcutaneous RG-2 tumor with an infusion pump (12 μl/min for 30 min). All were monitored with a T1-weighted gradient–echo sequence. In rats implanted with minipumps, MR images were acquired 72 h after implantation of the pump. In rats infused by infusion pump, MR images were acquired every 5 min during the infusion and every 5 min for 1 h after stopping the infusion. In the first group, T1-weighted images showed a VD of CED in the striatal region comparable with images of VD of CED acquired by quantitative autoradiography. In the infusion pump group, the signal from Gd-DTPA was detected in the striatal region as early as 5 min after the start of infusion; the signal slowly increased in the external capsule and ventricles over time. In the second group, the distribution of Gd-DTPA was mostly confined to tumor capsule and surrounding normal brain. The signal from Gd-DTPA was inhomogeneous in the tumor. With the infusion pump, the signal was detected in tumor as early as 15 min after the beginning of the infusion. The signal slowly increased in the region of the tumor capsule, the surrounding normal brain, the ventricle, and only some parts of the tumor. The signal was never detected in some parts of the tumor during the infusion. In the final group, the signal was detected inside the tumor within 1 min after starting the infusion. Again, the signal was detected only in some parts of the tumor during the infusion and was never detected in other parts. Only in the surrounding connective tissue was the signal increasing over the infusion period. This noninvasive method of monitoring VD of CED provides useful information about local drug delivery and a basis for clinical application in the treatment of many neurological diseases.


G. Trudy Oliver, Anriada Mehmeti, Jonathan F. Wells, Simon M. Lin, and Robert J. Wechsler-Reya; Departments of Pharmacology and Cancer Biology and Biostatistics and Bioinformatics, Duke University Medical Center, Durham, North Carolina, USA

Identifying early stages of cancer is critical because it allows intervention at a time when treatment is most likely to succeed. Medulloblastoma is the most common malignant brain tumor in children, but little is known about its early stages. Mice heterozygous for mutations in patched develop tumors that resemble human medulloblastoma, and can be used to study the mechanisms of tumorigenesis. Here we identify a preneoplastic stage of medulloblastoma in patched mutant mice. We show that preneoplastic cells are present in the majority of patched mutants, although only 16% of these mice develop tumors. Preneoplastic cells, like tumor cells, exhibit activation of the Sonic hedgehog pathway and constitutive proliferation. These cells also lack expression of the wild-type patched allele, suggesting that loss of patched is an early event in tumorigenesis. While preneoplastic cells resemble tumor cells in many respects, they have a distinct molecular signature. Genes that mark the preneoplastic stage include regulators of migration, apoptosis, and differentiation, processes critical for normal development but previously unrecognized for their role in medulloblastoma. The molecular characterization of preneoplastic cells provides insight into the early steps in medulloblastoma formation and may yield important markers for early detection and therapy of this disease.


Alan Shih1 and Eric Holland1,2; 1Departments of Cancer Biology & Genetics, 2Surgery (Neurosurgery), and Neurology, Memorial Sloan-Kettering Cancer Center, New York, New York, USA

Notch is in a family of single-pass transmembrane receptors that is activated by ligand binding and intramembrane proteolysis. This step releases an activated transcription factor that enters the nucleus and turns on target genes. The Notch pathway is critical in regulating multiple steps in neural and glial development. Specifically, the Notch pathway has been demonstrated to regulate the maintenance of neural progenitor cells, specification of oligodendrocytes, and differentiation of other glial types including radial glia and astrocytes. Notch has also been implicated in cancer, particularly in leukemia and breast carcinoma. We demonstrate that Notch is activated in human samples of brain tumors. Using mouse models of brain tumor formation, we find that Notch mRNA expression is elevated in tumors compared to normal brain and that it may be a target of PDGF signaling. Activation of Notch and Ras pathways in neural progenitor cells appears to arrest these cells in a progenitor state and generate small proliferative lesions. In the setting of p19Arf deficiency, Notch pathway activation may act to determine the type of tumors formed from GFAP positive precursor cells. Notch activation may promote an oligodendrocytic tumor compared to other tumor types. Thus, the Notch pathway may play a part in brain tumor formation by affecting normal developmental pathways by arresting glial development or altering glial cell specification.


Sheila K. Singh,1,2 Cynthia Hawkins,3 Ian D. Clarke,1 Jeremy Squire,5 Jane Bayani,5 Takuchiro Hide,1 Michael D. Cusimano,4 and Peter B. Dirks1,2; 1Program in Developmental Biology, 2Division of Neurosurgery, and 3Department of Laboratory Medicine, The Hospital for Sick Children, 4Division of Neurosurgery, St. Michael’s Hospital, 5Ontario Cancer Institute, University of Toronto, The Arthur and Sonia Labatt Brain Tumor Research Centre, Toronto, Ontario, Canada

We recently reported the identification of a cancer stem cell from pediatric brain tumors of different phenotypes. Less obvious evidence exists for a hierarchy of proliferation, self-renewal, and tumorigenicity in adult brain tumors. Here, we report the prospective purification of a cancer stem cell from primary human adult gliomas through cell sorting for the neural stem cell surface marker CD133. CD133+ cells represented 10% to 30% of the tumor cells, with a higher proportion in higher grade gliomas. In vitro, these CD133+ cells exclusively generated clonal tumor spheres, exhibited increased self-renewal and proliferation capacity, and could differentiate to phenotypically resemble the lineage of cells in the patient’s tumor. When injected into NOD-SCID mouse forebrains, as few as 1000 CD133+ cells generated a tumor resembling the original patient’s tumor, whereas injection of 100,000 CD133- cells never formed a tumor. These data demonstrate that CD133 can prospectively identify human glioma stem cells that exclusively maintain glioma growth in vitro and in vivo, reinforcing the hypothesis that solid tumor growth is determined by small numbers of cancer stem cells. The cancer stem cell hypothesis as applied to human gliomas suggests that research and new potential therapies must focus on these key brain tumor–initiating stem cells.


Nicolas Spaeth,1 Matthias T. Wyss,1,6 Bruno Weber,1 Stephan Scheidegger,2 Amelie Lutz,3 Jorn Verwey,7 Ivan Radovanovic,4 Jens Pahnke,4 Damian Wild,1 Gerrit Westera,6 Dominik Weishaupt,3 Dirk M. Hermann,5Barbara Kaser-Hotz,2 Adriano Aguzzi,4 and Alfred Buck1; 1PET Center, Division of Nuclear Medicine, 2Section of Diagnostic Imaging and Radio-Oncology, Veterinary Hospital, 3Institute of Diagnostic Radiology, 4Institute of Neuropathology, Department of Pathology, and 5Department of Neurology, University Hospital Zurich; 6Center for Radiopharmaceutical Science of ETH, PSI and USZ and 7Division of Radiation Medicine, Paul Scherrer Institute, Villigen; Switzerland

The differentiation between posttherapy radiation necrosis and recurrent tumor in human brain tumor patients still poses a difficult diagnostic task. The new PET tracers 18F-fluoro-ethyl-l-tyrosine (FET) and 18F-fluorocholine (N,N-dimethyl-N-[18F]fluoromethyl-2-hydroxyethylammonium [FCH]) showed promising features to improve the diagnostic accuracy. The purpose of this study was to assess the uptake of these tracers in experimental radiation injury and F98 rat gliomas. In a first model circumscribed lesions were induced in the cortex of 15 rats using proton irradiation of 150 or 250 Gy. Following the development of a radiation lesion the uptake of FET and FCH was measured in eight and seven rats, respectively, using autoradiography. In a second model, the uptake of both tracers was assessed in seven (FET: 4; FCH: 3) intracerebrally inoculated F98 glioma rats. The mean FET and FCH standardized uptake values (SUVs) in the acute radiation injury and the contralateral normal cortex (in parentheses) were 1.63 ± 0.34 (1.42 ± 0.23) and 1.50 ± 0.33 (0.61 ± 0.12), respectively. In the rat glioma, mean FET and FCH SUVs were 3.90 ± 0.90 (1.21 ± 0.16) and 3.02 ± 0.19 (0.46 ± 0.05). The uptake of both tracers in the radiation injuries is generally lower than the uptake in the F98 gliomas, suggesting that FET and FCH are promising tracers to separate radiation necrosis from tumor recurrence. However, a final conclusion still warrants a higher number of animals with tumors.


Q. Wei,1 L. Clarke,1 N. Sabha, B. Qian,1 J. Lavoie,1 P. Shannon,2 D. Kaplan,1 D. Gutmann,3 A. Guha1,2; 1The Hospital for Sick Children, Toronto, Ontario, Canada; 2University of Toronto, Toronto, Ontario, Canada; 3Washington University, St. Louis, Missouri, USA

Amplification and mutations of EGFR, such as the constitutively activated EGFRvIII mutant, are common in high-grade but not lower grade gliomas. Transgenic mouse models also support this role of EGFR in glioma progression, with our previous work demonstrating that GFAP-regulated expression of wtEGFR and EGFRvIII in glial progenitor cells with embryonic stem cell transgenesis failed to initiate gliomas. However, in a background of mice susceptible to develop gliomas, such as our GFAP-regulated V12Ha-Ras transgenics, which develop and die from diffuse astrocytomas by ≈12 to 16 weeks, embryonic expression of EGFRvIII and not wt-EGFR potentiated glioma formation. In addition to dying at ≈3 to 4 weeks, these double transgenic mice also developed mainly malignant oligodendroglial or oligoastrocytoma (OA) lineage gliomas. In this study, we investigated the effects of postnatal expression of EGFRvIII in normal and glioma susceptible mice and compared our results with those found with embryonic expression in the double transgenics. One-month-old GFAP-V12Ha-Ras transgenics and normal littermates were injected with AdEGFRvIII or control AdGFP virus into the frontal lobe. The mice were observed for neurological symptoms, with all mice sacrificed after BrDU injection by 4 weeks post virus injection and the brains analyzed for glioma development. AdEGFRvIII or AdGFP injection into normal mice did not result in glioma formation. However, GFAP-V12Ha-Ras transgenics injected with AdEGFRvIII developed large, confluent low-grade and many high-grade gliomas, which were mainly astrocytic in nature, with a few being malignant OAs. Of interest, EGFRvIII expression was detected in the astrocytes and/or oligodendrocytes in these high-grade astrocytomas or OAs and not in their lower grade counterparts. AdGFP-injected mice did have a few low-grade multifocal astrocytomas, similar to our prior observations in noninjected GFAP-V12Ha-Ras mice. Postnatal expression of EGFRvIII failed to initiate glioma formation, similar to prior findings with embryonic expression. However, in the glioma predisposed GFAP-V12Ha-Ras transgenics, postnatal EGFRvIII expression did potentiate gliomagenesis, with EGFRvIII expression associated with mainly the high-grade gliomas. Of interest, the gliomas with postnatal EGFRvIII were mainly astrocytomas, compared with the predominance of oligodendrogliomas with embryonal expression of EGFRvIII reported in our prior study. These observations are consistent with human gliomas, where EGFRvIII is mainly but not exclusively associated with high-grade astrocytomas or OAs. Furthermore, this data is another illustration of glioma lineage being dependent on the nature of the transforming genetic alterations and the differential potential of the transformed cell. Ongoing experiments are directed toward understanding the role of EGFRvIII in gliomagenesis in the background of other relevant glioma-associated genetic alterations, such as loss of Ink4a/Arf and Pten.


Q. Wei,1 L. Clarke,1 B. Qian,1 C. Zhang,1 P. Shannon,2 D. Gutmann,3 and A. Guha 1,2; The Hospital for Sick Children, Toronto, Ontario, Canada; University of Toronto, Toronto, Ontario, Canada; 3Washington University, St. Louis, Missouri, USA

Somatic inactivation of Pten is prevalent in high-grade gliomas, with germline Pten mutations implicated in a variety of clinical syndromes. Previous reports of mainly neuronal deletion of Pten described seizures, ataxia, and increased neuronal size, resembling Lhermitte-Duclos syndrome, without glioma development. Here, a more astrocyte-specific GFAP-Cre mouse was created to examine the developmental and tumorigenic role of Pten in the CNS. Pro-nuclear injection was used to create a GFAP-Cre mouse, which demonstrated astrocyte-specific cre-expression. Cre-excision was characterized by creating double transgenics with Z/AP reporter mice, demonstrating AP expression mainly in the astrocytes and not neurons or other glial cells. Double transgenics with the GFAP-Cre and PTENflox/flox mouse (from Dr. Tak Mak) were created. Cre-recombinase expression and function in GFAP-Cre mice were mainly in astrocytes of the hippocampus, cortex, and cerebellum. GFAP-Cre X PTENf/f double transgenics died from seizures with enlarged brains and failure to thrive by 2 to 6 weeks. Loss of Pten resulted in activated P-Akt, mainly in astrocytes and not neurons, Purkinje cells, or oligodendrocytes. The Pten- astrocytes were increased in size and numbers, which resulted from increased proliferation, though no gliomas were noted. Multiple and reproducible structural CNS abnormalities were noted. The cerebellum was enlarged with abnormal radial glia and disorganized but normal Purkinje cells. The hippocampal neurons were increased in size, with a gradient of increasing Pten expression, inversely correlated to P-Akt expression, toward the cortical surface. Similarities and differences were present compared to prior reports of GFAP-regulated cre-excision of Pten, most likely based on the cell specificity and level of cre-excision, since the same PTENflox/flox mice were used. Our GFAP-Cre mouse mainly induced cre-excision in astrocytes, compared to the mainly neuronal cre-excision in the prior reports. The CNS developmental alterations resulted in even earlier death, without any glioma formation in both experiments. Loss of Pten was associated with increased cell size, which was mainly in astrocytes versus the previously reported neurons. Our main interest is in the role/interactions of Pten in gliomagenesis, which we are pursuing with somatic and inducible cre-excision, in the context of additional relevant genetic alterations associated with human gliomas.



J.M. Baehring, J. Longtine, J. Sklar, and F.H. Hochberg; Yale University School of Medicine, New Haven, Connecticut; Brigham and Women’s Hospital, Boston, Massachusetts; Massachusetts General Hospital, Boston, Massachusetts, USA

The diagnosis of lymphoma affecting the nervous system and the eye (NSL) is based upon morphological analysis of tissue, vitreous body material, or cerebrospinal fluid. Analysis of immunoglobulin heavy-chain gene rearrangement (AIGHR) complements classical methods of diagnosis in B-cell lymphoma outside the nervous system. However, this modality has not commonly been applied to paucicelluar specimens retrieved from patients with NSL, and thus its sensitivity and specificity in this setting are unknown. Patients in whom NSL was suspected and AIGHR of CSF or vitreous fluid had been performed were included in this retrospective series. The protocol for AIGHR was as follows: Cells from 2- to 8-ml CSF samples or up to 0.5-ml vitreous fluid or vitreous wash samples were separated by centrifugation. The cell pellet was resuspended in 100 mcl of the supernatant and boiled for 10 min. Ten mcl of the lysate served as the template for semi-nested polymerase chain reaction using consensus primers for complementarity-determining region III (CDRIII). PCR products were separated on a 10% polyacrylamide gel, which was then stained in ethidium bromide and photographed under ultraviolet light. Clonal rearrangement was defined as the occurrence of one or two identical prominent bands from parallel PCR assays of at least two aliquot of the same specimen. Thirty-nine CSF specimens obtained from 35 patients and 60 vitreous biopsy specimens (47 patients) were included in the analysis. Thirty-four patients had lymphoma affecting the nervous system (17 patients) or eye (17 patients). There were 48 cases (CSF, 18; vitreous, 30 patients) with benign lymphoid processes or infectious meningitis, meningoencephalitis, or vitritis. Sensitivity of AIGHR in CSF was 0.62, and 0.5 in vitreous fluid. Specificity in this cohort was 0.88 in CSF and 1.0 in vitreous. Of particular interest are three patients with clonal AIGHR whose clinical course up to the last day of follow-up suggested a “benign” process (neuroborreliosis, granulomatous angiitis). AIGHR appears to be a useful addition to morphological and flow cytometry analysis in lymphoproliferative diseases. Sensitivity determined in our cohort is well within the range that has been reported for AIGHR using CDRIII consensus primers in systemic lymphoma. Factors that interfere with sensitivity, amongst others, are primer-binding competition from a reactive polyclonal infiltrate and partial deletion or replacement of 3’ segments of VH genes. As clonal rearrangement is not a marker of malignancy per se, clonal lymphoproliferative diseases may be identified at a stage when transformation into a malignant process may still be reversible.


Jaclyn A. Biegel, Peter C. Burger, Lucy B. Rorke, and Alexander R. Judkins; Departments of Pediatrics and Pathology, Children’s Hospital of Philadelphia and The University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania; Johns Hopkins University, Baltimore, Maryland; USA

Since the initial description of atypical teratoid/rhabdoid tumor in 1987, cytogenetic and molecular genetic studies of chromosome 22 and the INI1 gene have been used to distinguish this entity from other malignant brain tumors considered in the differential diagnosis. Monosomy 22 or a smaller deletion involving chromosome band 22q11.2 results in loss of one copy of the INI1 tumor suppressor gene. The second copy of INI1 is subject to mutation or deletion, thus resulting in homozygous inactivation of the gene. Patients with germline mutations of INI1, most often characterized by mutation in one copy of INI1, are predisposed to development of rhabdoid tumors in the brain, kidney, and soft tissues. We have recently shown that immunohistochemical staining for INI1 protein may be used to identify tumors with inactivation of INI1 due to alterations in the coding sequence, or changes in the expression level of INI1 mRNA. Moreover, loss of expression of INI1, as detected by negative staining, can be observed in tumors with or without chromosomal deletions of 22q11.2, even in the absence of detection of alterations at the DNA or RNA level. Conversely, retained expression of INI1 has been used to rule out AT/RT, even in tumors with monosomy 22 and histologic features that would suggest such a diagnosis. Report of INI1 mutations in choroid plexus carcinoma (CPC) has suggested that CPC and AT/RT share a common molecular etiology. We analyzed 29 tumors diagnosed as choroid plexus carcinoma and found loss of INI1 expression by immunohistochemistry only in tumors that were independently reclassified as AT/RT. Loss of chromosome 22q11.2 is not specific for rhabdoid tumor, and classical histopathologic analysis and immunophenotyping may be too subjective in a routine diagnostic setting to reliably make a diagnosis of AT/RT. In contrast, loss of staining of INI1 by immunohistochemistry may reflect inactivation of INI1 through upstream effects that may not be tumor specific. A summary of our genetic and immunohistochemical analyses of a series of pediatric rhabdoid and other malignancies will be presented.


D.T. Blumenthal,1 M. Quezado,2 M. Santi,3 A. Rapkiewicz,2 R. Ronchetti,2 and E.J. Rushing 4; 1University of Utah/Huntsman Cancer Institute, Salt Lake City, Utah; 2Laboratory of Pathology, National Cancer Institute, Bethesda, Maryland; 3Department of Pathology, Children’s Hospital National Medical Center, Washington, D.C.; 4Armed Forces Institute of Pathology, Washington, D.C.; Southwest Oncology Group (SWOG) Brain Committee, San Antonio, Texas1,4; USA

A retrospective study was performed to correlate the expression of epidermal growth factor receptor (EGFR) and small-cell (SCGBM) phenotype with outcome in glioblastoma. The study population comprised 56 patients with newly diagnosed GBM (treated on SWOG protocol S0001) with sufficient pretreatment tumor biopsy specimens for study. Thirty cases of SCGBM and 26 with non-small-cell glioblastoma (NSCGBM) were identified. EGFR amplification was assessed by chromogenic in situ hybridization (CISH) by using formalin-fixed paraffin-embedded tissue sections. All patients were treated with radiation and carmustine, and patients on the experimental arm were treated with additional 0–6 benzylguanine. Twenty-two SCGBM cases (22/30) showed EGFR amplification, while only five NSCGBM cases (5/26) showed amplification. In cases with no amplification (29), chromosome 7 CISH evaluation showed polysomy (28/56). Survival data as correlated to histologic phenotype and EGFR amplification will be presented. The small-cell phenotype corresponds to EGFR amplification in GBM. Identification of this histologic subtype may have implications for prognosis and for targeted therapy for GBM.


Dieta Brandsma,1,2 Willem Boogerd,3 Wilco de Jager,5 Tiny Korse,4 Hans Bonfrere,4 Marcel Verbeek,6 Jaap C. Reijneveld,1 Emile E. Voest,2 and Martin J.B. Taphoorn1,7; 1Department of Neurology and 2Laboratory of Medical Oncology of the Department of Medical Oncology, University Medical Center Utrecht, Utrecht; Departments of 3Neurology and 4Clinical Chemistry, The Netherlands Cancer Institute, Amsterdam; 5Department of Pediatric Immunology, Wilhelmina Children’s Hospital, University Medical Center Utrecht/IACOPO Institute for Translational Medicine, Utrecht; 6Department of Neurology, University Medical Center Nijmegen, Nijmegen; 7Department of Neurology, Medical Center Haaglanden, The Hague; The Netherlands

Leptomeningeal metastases (LM) can cause severe neurological deficit and are associated with a dismal prognosis. The identification of tumor cells in CSF is diagnostic for LM. However, a positive CSF cytology is found on the initial lumbar puncture in approximately 50% of patients with LM, and its yield further increases to 90% if a total of three lumbar punctures are obtained. (Pre-) clinical studies have shown an important role for both tumor cell adhesion and angiogenesis in the progression of LM. We measured the levels of several soluble adhesion molecules, chemo- and cytokines in the CSF of patients with cytologically proven LM (n = 59), patients with a primary extracranial tumor but no LM (n = 21), and patients having no infection nor tumor (n = 20), using the Luminex fluorescent-bead based technology. We found significantly increased levels of Pulmonary and Activation Regulated Chemokine (PARC), Interferon-g inducible protein (IP-10), Interleukin-8 (IL-8), Interleukin-18 (IL-18), soluble Vascular Cell Adhesion Molecule-1 (sVCAM-1), and soluble Intercellular Adhesion Molecule-1 (sICAM-1) in the CSF of patients with LM as compared to the other two patient groups. Tumor necrosis factor-α (TNF-α), Interleukin-1β (IL-1β) and Interferon-γ (IFN-γ) levels were not present in the CSF of all three patient groups. Our data indicate that specific chemo-, cytokines and adhesion molecules are elevated in the CSF of patients with LM, reflecting the pathophysiological processes involved in LM. These findings justify a prospective study in patients with suspected LM to determine whether a profile of chemo-, cytokines and adhesion molecules in the CSF can discriminate between patients with or without LM.


Alberto Broniscer, Christine Fuller, Melissa Fraser, Suzanne Baker, Larry Kun, Mehmet Kocak, Amar Gajjar, and Richard Gilbertson; St. Jude Children’s Research Hospital, Memphis, Tennessee, USA

Anaplastic progression of LGG in children is rare; therefore, the clinical characteristics and the molecular abnormalities associated with this transformation are unknown. In adult patients, progression of LGG to secondary glioblastoma (GBM) is associated with a well-defined series of molecular alterations: These include mutation of TP53, overexpression/amplification of PDGFRA, mutation of PTEN, and deletion of 19q and 10q. We identified 15 patients aged <21 years treated at St. Jude Children’s Research Hospital whose primary LGG underwent progression to high-grade glioma (HGG). In all instances, anaplastic progression was confirmed by serial imaging and histologic review. The median age at the diagnosis of the LGG was 12.3 years (range, 0.8–19.8 years). The initial histologic diagnoses included fibrillary/low-grade astrocytoma (n = 7), ganglioglioma (n = 2), PXA (n = 2), pilocytic astrocytoma (n = 2), oligodendroglioma (n = 1), and mixed oligoastrocytoma (n = 1). Initial therapy for LGG included radiotherapy for six patients and chemotherapy for five. Seven patients underwent surgical resection only for the LGG. Eleven patients died of their disease after a median survival of 0.6 years from the time of progression. The median interval for anaplastic progression was 2.5 years (range, 0.3–13.4 years). Tumor biopsies taken during progression from LGG to HGG were available for all patients. Final histologic diagnoses at progression included GBM (n = 10) and other high-grade gliomas (n = 5). To investigate whether the molecular alterations involved in the progression of pediatric LGG are similar to those that occur in adults, we generated a tissue microarray that included all available LGG and subsequent progression samples from the current study group (n = 9 patients; 23 samples). We then determined the changes that occurred during anaplastic progression in the expression of ERBB1, ERBB2, ERBB3, and ERBB4, pAKT, PTEN, pS6, and TP53 protein (immunohistochemistry) and ERBB1, ERBB2, PTEN, RB, P16, 1p, 19q, 17p, CDK4, PDGFRA, and PDGFRB copy number (dual color fluorescence in situ hybridization). Alterations identified included co-overexpression of ERBB1 and ERBB4, overexpression of TP53, amplification of PDGFRA, and deletion of P16, 10q, and 19q. Our data provide the first detailed characterization of anaplastic progression in pediatric LGG and show that these tumors accrue a series of molecular alterations similar to those associated with adult secondary GBM.


Asha Das,1 Carole Hurvitz,2 Charles Simmons,2 Asha Puri,2 Xuemo Fan,3 and Moise Danielpour1; 1Maxine Dunitz Neurosurgical Institute, 2Department of Pediatrics, and 3Department of Pathology, Cedars-Sinai Medical Center, Los Angeles, California, USA

Neonatal brain tumors occur in 0.34 per million live births and account for 0.5 to 1.5% of detected childhood tumors. Many of the phases of assisted reproduction, including the use of fertility drugs to stimulate multiple folliculogenesis, the process of oocyte retrieval and preparation of spermatozoa, the growth of embryos in culture medium, and the freezing and thawing of embryos, have a potential harmful effect on the developing embryo. Given the limited environmental influences on tumorigenesis, there may be a relationship between neonatal brain tumors in the setting of assisted reproduction. This report describes the clinical features of a rare neonatal anaplastic astrocytoma in the setting of in vitro fertilization. The baby is a full-term product of a 29-year-old prima gravida. After approximately 5 years of attempts to conceive, conception was achieved via intrauterine insemination. Ovulation was induced by clomiphene and gonadotropin, and the transferred embryos were fresh. At birth the baby boy was found to have irritability and poor feeding. Cranial ultrasound and MRI revealed an echogenic mass in the left hemisphere with midline shift and hydrocephalus. On day 4 of birth the baby underwent craniotomy with resection of an astrocytic neoplasm with anaplastic features. Subsequently, chemotherapy with temozolomide and vincristine was administered with no evidence of recurrent tumor in the baby currently aged 20 months. A possible association between assisted reproduction and neonatal brain tumor is highlighted. A review of the reported clinical features of congenital brain tumors suggests that there is a slight male predominance. While intracranial hypertension and hydrocephalus are the most common presentations, behavioral disturbances and loss of previously acquired motor skills also occur. Among the tumors encountered, astrocytomas, medulloblastomas, ependymomas, choroid plexus tumors, embryonal tumors, teratomas, and meningiomas have been reported. Several studies have observed a significantly increased risk of neuroblastomas and of embryonal tumors in children born to women who conceived as a result of fertility treatment. The etiology of these tumors remains uncertain. They may be incidental, causally related to in vitro fertilization, or there may be a genetic predisposition for tumors in this subgroup of relatively infertile patients. Further large studies are required to determine if there is a small increase in the overall incidence of these congenital brain tumors in babies conceived via in vitro fertilization.


Sydney M. Evans, Cameron J. Koch, Kevin Judy, W. Tim Jenkins, and Kevin Jenkins; University of Pennsylvania, School of Medicine, Philadelphia, Pennsylvania, USA

New hypotheses regarding the development of pseudopalisades and glomeruloid blood vessels in GBM tumors place hypoxia in a central role for the progression and aggression of these tumors (Brat and Van Meier, multiple references). Our studies extend this work and evaluate the binding of the 2-nitroimidazole hypoxia imaging agent, EF5, as a quantitative measure of hypoxia in Grade 2–4 glial neoplasms. We examined the relationships between hypoxia and tumor aggression. In addition, the relationship between blood vessel and hypoxia was studied to determine the mechanism(s) by which hypoxia develops. Patients signed informed consent and received 21 mg/kg EF5 intravenously 24 h preceding surgical biopsy or excision of their brain tumor. Tissues were frozen, sectioned, and stained using previously described techniques for EF5 and PECAM (blood vessels). Methods for the image analysis, quantification of EF5 binding, and production of oxygen maps are published. Patients were followed with serial MRI studies for time to tumor recurrence. Hypoxia measurements were performed by using EF5 in 18 patients with supratentorial glial neoplasms. Time to recurrence was used as an indicator of tumor aggression and was analyzed as a function of EF5 binding. A correlation between more rapid tumor recurrence and hypoxia was demonstrated, suggesting that hypoxia is associated with tumor aggression. The two hypotheses put forth by Brat et al for the development of hypoxia were identified: diffusion-limited, “chronic” hypoxia and perfusion-limited, “acute” hypoxia. However, in some tumors, a discordance between vessel and hypoxia patterns, incompatible with either of the previously reported processes, was seen. The level and degree of hypoxia increase with increasing tumor aggressiveness. Detailed examination of the patterns of EF5 binding support Brat and Van Meier’s findings, but provide evidence for additional mechanisms. A better understanding of the etiology of these varied regions is important in terms of the molecular signaling as well as choice of therapeutic intervention.


S.D. Finkelstein, D. Mohan, P.A. Swalsky, F.S. Lieberman; Redpath Integrated Pathology, Pittsburgh, Pennsylvania; Department Pathology and Departments of Neurology and Medicine (Hematology/Oncology), Pittsburgh Cancer Institute, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA

While the presence and cumulative amount of mutational damage is predictive of glioma biological aggressiveness, the timeline of mutation acquisition has not been evaluated as an independent factor. Recently, it has become feasible to define the unique time course of specific mutation occurrence in routine tissue specimens of gliomas. We undertook to evaluate the significance of mutational time course as a prognostic indicator in gliomas. Thirty-six anaplastic gliomas (AGs) were microdissected at multiple sites using standard 4-μm-thick fixed tissue sections. Allelic imbalance (LOH) was determined quantitatively using a broad panel of 20 microsatellite markers (1p, 3p, 5q, 9p, 9q, 10q, 17p, 17q, 19q, 21q, 22q) and PCR/automated capillary electrophoresis. Individual patient mutation time course was established based on a clonal expansion model using topographic distribution and extent of individual marker imbalance at each microdissected site. Fluorescent in-situ hybridization (FISH) was performed on serial sections for determination of 1p loss. Mutational timeline was correlated with histologic, FISH, and outcome parameters. We found that 1p/19q allelic imbalance was present in 21 AGs. However, it was the first detectable mutation in only 8 of these cases. Oligodendroglial differentiation (OD), focal or diffuse, correlated with 1p/19q imbalance independent of its acquisition timing. Similarly, FISH detection of 1p loss closely correlated with that found by microdissection-based analysis. However, it did not delineate timing of 1p imbalance occurrence. Cumulative mutations of remaining markers were present in all AG (range, 2–14 mutations/case; average, 6.2). Major therapeutic responses occurred significantly more often in 1p/19q first event AG (6/8) compared with 1p/19q later event AG (2/13) or no 1p/19q detectable imbalance AG (1/15) (P < 0.001). Mutational damage including 1p/19q loss can occur at any time during glioma development and progression in a temporal pattern that is unique to an individual patient. Loss of 1p/19q is associated with OD in those cells having acquired this mutational alteration. Early acquisition of 1p/19q increases its likelihood of being present in all AG cells, potentially rendering them more susceptible to treatment. Reliance on histology alone using OD will not effectively delineate the treatment responsive subset of AG patients. While FISH is capable of detecting 1p loss, it did not define its timing when it was performed in the simple manner used in this study. Delineation of the time course of mutation acquisition provides useful information for prognostication of glioma behavior on a per case basis. Mutational timing should be considered for inclusion in the molecular pathology characterization of gliomas. Therapeutic strategies should target those mutational events occurring early in glioma tumorigenesis at the individual patient level.


William A. Freije, F. Edmundo Castro-Vargas, Zeke Fang, Steve Horvath, Timothy Cloughesy, Linda M. Liau, Paul S. Mischel, and Stanley F. Nelson; University of California at Los Angeles, Los Angeles, California, USA

In current clinical practice, histology-based grading of diffuse infiltrative gliomas is the best predictor of patient survival time. Yet histology provides little insight into the underlying biology of gliomas and is limited in its ability to identify and guide new molecularly targeted therapies. We have performed large-scale gene expression analysis using the Affymetrix HG U133 oligonucleotide arrays on diffuse infiltrating gliomas of all major histologic types from 74 patients to assess if a gene expression–based, histology-independent classifier is predictive of survival and to determine if gene expression signatures provide insight into the biology of gliomas. We found that gene expression–based grouping of tumors is a more powerful survival predictor than histologic grade or age. The poor prognosis group could be grouped into two different poor prognosis groups, each with distinct molecular signatures. We further describe a list of 44 genes whose expression patterns reliably classify gliomas into previously unrecognized biological and prognostic groups: These genes are outstanding candidates for use in histology independent classification of high-grade gliomas. The ability of the large-scale and 44 gene set expression signatures to group tumors into strong survival groups was validated with an additional external and independent data set from another institutions comprising 50 additional gliomas. This demonstrates that large-scale gene expression analysis and subset analysis of gliomas reveals unrecognized heterogeneity of tumors and is efficient at selecting prognosis-related gene expression differences that can be applied across institutions.


Christine Fuller, Maryam Fouladi, James Dalton, Larry Kun, Amar Gajjar, and Alberto Broniscer; St. Jude Children’s Research Hospital, Memphis, Tennessee, USA

Malignant astrocytomas in adults generally arise via two distinct molecular pathways; primary glioblastomas (GBMs) frequently harbor amplification of epidermal growth factor receptor (EGFR), whereas secondary GBMs that evolve from low-grade lesions often have alterations of p53. The molecular events involved in malignant progression in oligodendroglial tumors are less well defined. Information regarding these molecular abnormalities in pediatric gliomagenesis is limited, particularly concerning oligodendroglial and low-grade astrocytic lesions. We investigated 57 glioma specimens from 36 patients treated at St. Jude Children’s Research Hospital, including 29 low-grade and 28 high-grade glial tumors. The patients included 15 females and 21 males with a mean age of 9 years (range, 0.75–19 years). Archival formalin-fixed paraffin-embedded tissue blocks were used to generate tissue microarrays containing the following: 33 astrocytic tumors (grades I = 7, II = 5, III = 2, IV = 19), 17 oligodendrogliomas (grades II = 13, III = 4), and 7 oligoastrocytomas (grades II = 4, III = 3). The p53 expression status was assessed by immunohistochemistry, while fluorescence in situ hybridization (FISH) was used to determine EGFR copy number. Overexpression of p53 was encountered in 72% of cases, most frequently in astrocytic tumors and oligoastrocytomas (90% and 86%, respectively) and less often in oligodendrogliomas (37%). Positive p53 staining was more diffuse in the astrocytic lesions; oligodendroglial and oligoastrocytic tumors tended to have far fewer cells staining with anti-p53. Approximately half as many low-grade fibrillary astrocytomas and oligodendrogliomas had p53 overexpression as their high-grade counterparts. Interestingly, all primary specimens from two patients with pilocytic astrocytoma (5 samples) showed diffuse p53 overexpression, and subsequently they developed malignant transformation to glioblastomas. In one third of patients with astrocytic tumors for which multiple samples documented malignant progression, a sequential elevation in the p53 labeling index was encountered. In contrast, EGFR amplification was detected in only four tumors (2 high-grade astrocytomas and 1 each oligodendroglioma and anaplastic oligoastrocytoma); three of these showed concomitant p53 overexpression. These findings indicate that p53 overexpression is much more frequently encountered in pediatric astrocytomas and oligoastrocytomas than in oligodendrogliomas; p53 labeling is likewise more diffuse in astrocytomas compared to oligodendroglial tumors. Additionally, alterations in the p53-related pathways may play a role in malignant progression in not only fibrillary astrocytomas and oligodendrogliomas, but also a subset of pilocytic astrocytomas. Amplification of EGFR is an infrequent finding in pediatric gliomas, and it appears to be most often encountered in high-grade lesions.


M.D. Groves, T.J. Liu, K.R. Hess, A.D. Forman, S.H. Hsu, M.R. Gilbert, H. Colman, V.K. Puduvallia, C.A. Conrad, and W.K.A. Yung; The University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA

Neoplastic meningitis (NM) occurs in up to 8% of cancer patients, shortens survival (8–16 weeks), and causes neurological deterioration. Early detection/intervention may improve outcomes. A single cytological analysis of CSF in patients with known NM is ≈50% sensitive for detecting malignant cells. Sensitive and specific biomarkers that predict the presence or development of NM are needed. A prior study demonstrated elevated CSF VEGF levels in patients with NM from breast, ovarian, and lung cancer. Here, we confirm and expand this work. The objectives of this study were as follows: (1) Correlate CSF VEGF levels with the presence or absence of malignant cells in the CSF. (2) Ascertain whether elevated CSF VEGF levels can predict the later development of NM. We measured VEGF levels in CSF and serum (when available) using an ELISA assay. After each patient signed an IRB-approved consent form, samples were collected from patients undergoing CSF sampling for diagnostic or therapeutic purposes. Statistical modeling estimated the need for accrual of a total of 24 breast and lung cancer patients with NM to achieve a lower 95% confidence interval limit of no less than 85% for an estimated sensitivity of 100% for these two histologic groups. Data are available for 59 patients, 22 with NM. Nine of 24 of the needed breast and lung cancer patients with NM have been accrued. Overall, CSF VEGF levels ranged from undetectable (U) to 4324 pg/ml. All breast, lung, and other solid tumor patients (excluding brain tumors) with elevated CSF VEGF levels have so far had malignant cells in the CSF at the time of testing. For patients with breast, lung, and other solid tumors (non primary brain), elevated CSF VEGF levels (cut point 20 pg/ml) perfectly correlate with the presence of malignant cells in the CSF. With additional patient accrual, if such high degrees of sensitivity and specificity are maintained, and if elevated levels are demonstrated in some patients who later develop NM, CSF VEGF may serve as an excellent biomarker for NM. Updated data will be presented.

No. of patients (range CSF VEGF pg/ml)
Histology (n = total no. patients)+ Neoplastic Meningitis−Neoplastic Meningitis
Breast (n = 16)n = 5 (33–597)n = 11 (U-10)
Lung (n = 10)n = 4 (24–464)n = 6 (U-17)
Melanoma (n = 2)n = 1, 204n = 1, U
Other solid tumors (n = 5)n = 2, 53, 4324n = 3, all U
Primary brain tumors (n = 14)n = 5 (U-21)n = 9 (U-3125)
Leukemia/lymphoma (n = 10)n = 5 (U-16)n = 5 (U-15)
Normal controls (n = 2)n = 2 (4, 14)


Amy Heimberger, Roman Hlatky, Dima Suki, David Yang, Mark Gilbert, Raymond Sawaya, and Kenneth Aldape; University of Texas M.D. Anderson Cancer Center, Houston Texas, USA

The epidermal growth factor receptor (EGFR) is overexpressed in approximately 50% to 60% of glioblastoma multiforme (GBM) cases and the most common EGFR mutant, EGFRvIII, is expressed in 24% to 67% of GBM cases. The impact of the amplified wild-type EGFR and the mutant EGFRvIII as an independent prognostic indicator of overall survival (OS) has not been definitively determined. Previous studies suggested a trend toward a poor prognosis in patients with GBM with positive EGFRvIII expression. One hundred-ninety six patients with GBM underwent a 95% or greater volumetric resection (based on T1-weighted Gd-enhanced images) followed by conformal radiation. By immunohistochemistry, the EGFR and EGFRvIII status was determined, Cox regression analysis corrected for age and Karnofsky status, and OS was plotted on Kaplan-Meier curves. The amplified wild-type EGFR and EGFRvIII were present in 51% and 30% of GBM cases, respectively. There was no significant difference in age, sex, Karnofsky status, volume of tumor resection, or postoperative treatment between the non-expressing, amplified EGFR or EGFRvIII expression groups. The overall median survival for nonexpressing, amplified and EGFRvIII patients was 0.99 (95% CI, 0.82–1.16), 0.98 (95% CI, 0.69–1.27), and 1.06 (95% CI, 1.00–1.12) years, respectively. In contrast to previously published reports, amplified wild-type EGFR or EGFRvIII are not independent predictors of overall survival and did not confer a worse prognosis. However, in those patients surviving greater than one year, the expression of EGFRvIII was a negative prognostic indicator. The median survival in this subcategory of patients for nonexpressing, amplified and EGFRvIII patients was 2.03 (95% CI, 0.47–3.59), 1.95 (95% CI, 1.58–2.32), and 1.21 (95% CI, 1.12–1.30; P < 0.001 comparing EGFRvIII to all others) years, respectively. Similar to findings in previous studies, KPS score, age, and necrosis were found to be valid prognostic factors independent of amplified wild-type EGFR or EGFRvIII status. Furthermore, patients with GBMs expressing amplified wild-type EGFR and EGFRvIII were not more likely to be diagnosed with multifocal disease or to have a leptomeningeal involvement on radiographic imaging. Finally, tumor subtype (EGFR negative, amplified wild-type EGFR, and EGFRvIII) does not impact overall survival as an independent prognostic indicator from the time of the initial surgery in patients who undergo re-resection.


Adília Hormigo,1 Bin Gu,2 Martin Fleisher,2 Lisa M. DeAngelis,1 and Eric C.Holland1,3; 1Departments of Neurology, 2Clinical Laboratories, and 3Neurosurgical Service of the Department of Surgery, Memorial Sloan-Kettering Cancer Center, New York, New York, USA

YKL-40 is an extracellular matrix glycoprotein highly expressed in glioblastoma multiforme (GBM) which is present in the serum of patients. We are conducting a prospective study in patients with brain tumors using multiple longitudinal serum samples from each patient to determine if YKL-40 levels correlate with tumor type, grade, active or recurrent disease, and response to treatment. YKL-40 protein levels are measured by ELISA assay and are correlated with disease status assessed by MRI. The serum samples are usually obtained on the same day that the MRI is received. We have analyzed from one to eight serum samples on each GBM patient for a total of 25 patients. The mean of YKL-40 in these patients was 99.6 mcg/ml. In six of the patients, the determinations were done peri-operatively to evaluate the impact of surgery on the YKL-40 level. The YKL-40 level increased in five patients between 24 to 96 h after the operation and rapidly dropped. In the other 19 patients, 97% of the YKL-40 levels correlated with the corresponding MRI determination of stable disease, complete or partial response, and progression of disease. In four patients, the YKL-40 level anticipated disease status on the subsequent MRI. In 11 patients with primary central nervous system lymphoma, the YKL-40 levels were elevated (mean 149 mcg/ml), but they did not correlate with disease status on the corresponding MRI. YKL-40 levels may supplement neuroimaging in determination of disease status in patients with GBM. Further longitudinal evaluation is necessary to ascertain whether YKL-40 is a reliable serum marker for patients with malignant glioma.


Annie Huang, Daniel Picard, Cynthia Ho, Romina Ponzielli, Dalia Bareste-Lovejoy, and Linda Z. Penn; Hospital for Sick Children, Ontario Cancer Institute and the University of Toronto, Toronto, Ontario, Canada

c-Myc is a potent oncoprotein that is widely implicated in advanced and metastatic cancers. Recent clinical studies correlate c-myc mRNA expression with poorer outcomes in medulloblastoma, and c-myc genomic amplification with particularly aggressive and invasive medulloblastoma tumour phenotypes. These observations suggest an important role for Myc dysregulation in medulloblastoma development. Indeed, both the c- and N-myc genes are now known to be downstream targets of two major medulloblastoma signaling pathways. Studies of proteins that interact with c-Myc to modulate its activity in medulloblastoma cells will be important for delineating the role of Myc proteins in medulloblastoma transformation. We used a modified Yeast 2 hybrid system to screen a medulloblastoma cell line library for proteins that interact with the N-terminal transformation domain of c-Myc. Characterization of clones from the modified 2-hybrid screen resulted in the identification of two highly related and evolutionarily conserved proteins as novel Myc protein interactors. The proteins share 40% overall amino acid homology with greatest sequence conservation within a putative ring finger domain. Both proteins of this novel, ring finger Jpo family, interact with Myc in vitro and in vivo. Immunofluorescence studies indicate co-localization of the Jpo proteins with c-Myc in the nucleus. In cell fractionation experiments, both proteins are enriched with c-Myc in nuclear chromatin, suggesting potential roles for these novel interactors in Myc gene regulation. Constitutive expression of Jpo proteins with Myc enhances Myc-mediated colony formation in rat fibroblasts and medulloblastoma cells. RT-PCR analyses show broad expression of genes encoding both Jpo proteins in medulloblastoma cell lines and primary tumours. Collectively, these observations suggest a role for these novel Myc interactors in medulloblastoma transformation. Experiments are ongoing to more directly evaluate the role of these proteins in medulloblastoma cell phenotypes by stable RNAi knock down, and to identify target genes of Myc and the Jpo proteins in medulloblastoma cells by chromatin immunoprecipitations:promoter microarray analyses.


Iris Lavon, Bracha Zelikovitsh, Yigal Shoshan, and Tali Siegal; Leslie and Michael Gaffin Center for Neuro-Oncology, Department of Neurosurgery, Hadassah Hebrew University Hospital, Jerusalem, Israel

In oligodendroglial neoplasms, chromosomal deletions at 1p and 19q are associated with chemosensitivity and increased overall survival. Thus, 1p/19q testing becomes an increasingly important adjunct tool for diagnosis and assessment of prognosis of these neoplasms. The frequent requirements for the test from clinicians led us to look for a reliable, rapid, inexpensive and safe assay. Currently, there are two main techniques that are used to assess loss of heterozygosity (LOH) status of tumors: the time-consuming FISH analysis and the radioactive-based assay of microsatellite repeat marker analysis. To avoid some of the disadvantages of these methods we have established an assay that is both safe and easy. The described method allows us to provide results within less then 48 h. We used modified techniques for the three steps required for the assay as follows: (1) extraction of DNA from whole blood by salting-out, (2) extraction of DNA from paraffin-embedded tissues using microwave, instead of xylene and ethanol extraction, (3) PCR for microsatellites on 1p, 19q, and 17p and capillary electrophoresis of the PCR products. Blood and paraffin-embedded tumor sections were obtained from 40 patients that were histologically diagnosed as having oligodendroglial neoplasms. There were 17 oligodendrogliomas (O), 16 anaplastic oligodendrogliomas (AO) and 7 mixed oligoastrocytomas (OA). DNA was extracted from both the blood and the tumor, and LOH was studied by capillary electrophoresis following PCR amplification of microsatellite markers on chromosomes 1p, 19q, and 17p. Overall, LOH was detected in 64% of AO and 71% of O. The loss was either a combined deletion of 1p and 19q or an isolated one. In contrast, no aberrations in these regions were found in tumors diagnosed as OA. LOH in 17p at the p53 region was detected in relatively few tumors from the three histological groups. The described technique of our assay overcomes some of the disadvantages associated with the methods that are currently in common use. The salting-out method that we use for DNA extraction from whole blood is rapid and inexpensive. Extraction of DNA from paraffin-embedded tissues by microwave yielded enough DNA even from small biopsy samples. This method is very quick and avoids the use of toxic materials like xylene and ethanol. The last step of capillary electrophoresis is also a safe and fast technique that has a significantly better size resolution compared to the traditional solid gel-based methods. This leads to an improved identification of the allelic pattern. The disadvantage of this method is that samples that contain only infiltrative tumors with extensive normal brain portions are not suitable for analysis by this assay. In our experience, this is an uncomplicated and safe LOH assay that yields rapid results that facilitate the inclusion of the test as an essential routine part of the evaluation of patients with oligodendroglial neoplasms.


P.S. Mischel,1,6 M. Carlson,2 Z. Fang,2,5 W. Freije,2 E. Castro,2 L.M. Liau,3,6 D.H. Geschwind,4,7 T.F. Cloughesy,4,6 S. Horvath,2,5 and S.F. Nelson2,6; Departments of 1Pathology and Laboratory Medicine, 2Human Genetics, 3Neurosurgery, 4Neurology, and 5Biostatistics; 6Henry E. Singleton Brain Cancer Research Program; and 7Neurogenetics Research Program; David Geffen School of Medicine, University of California Los Angeles, Los Angeles, California, USA

Gene expression is organized into modular networks that reflect key cellular functions. It is currently unknown whether disease states such as cancer alter this global network architecture. Using DNA microarray data from 63 glioblastomas, 92 other cancers, and 105 normal tissues, we identify disease-specific networks of co-expressed genes including a module of cell cycle-related genes whose co-expression is preserved across multiple cancers but not normal proliferating tissue. To demonstrate the biological importance of these co-expression interactions, we show a highly significant relationship between the connectedness of a gene within the module (but not its expression level) and its ability to predict glioblastoma patient survival. Thus, analysis of gene expression network topology can potentially identify altered networks and biologically important genes that are not readily detected by traditional methods.


Anjan Misra,1,3 Janice Nigro,1,3 Ivan Smirnov,1,3 Kathleen Lamborn,1,3 Scott Vandenberg,1,3 Ken Aldape,4 Burt G. Feuerstein1,2,3; Brain Tumor Research Center, Departments of 1Neurosurgery and 2Laboratory Medicine and 3Cancer Center, University of California San Francisco, San Francisco, California; 4Department of Pathology, M.D. Anderson Cancer Center, Houston, Texas; USA

We used a DNA microarray to map DNA copy number aberrations (CAN) in >70 grade 3 (AA) and >150 grade 4 primary human astrocytomas (GM) and Affymetrix U95 chips to evaluate the expression profiles in a subset of the GM. The BAC array detects genome-wide DNA copy number at ≈1.4-Mb resolution, and the expression array detects expression of ≈12000 genes/ESTs. We analyzed the data with unsupervised clustering to visualize possible genetic subgroups and supervised clustering methods (Significance Analysis of Microarrays [SAM], shrunken centroid analysis [PAM], linear discriminant analysis) to evaluate differences between histology, genetic subgroups, and survival. Our experiments confirmed previously known amplifications and deletions and identified infrequent amplifications and homozygous deletions. Unsupervised cluster analysis of AA suggests there are several genetic subgroups of AA. Unsupervised cluster analysis of GM suggests there are two major genetic subgroups. One group sustains loss of chromosome 10 and/or gain of chromosome 7 (7/10 group), and the other does not (intact 7/10). Most of the 7/10 group sustain CNAs of both chromosomes, but a few do not sustain the CAN on chromosome 7. Long term GM survivors were concentrated in the intact 7/10 group. Furthermore, these CNAs were associated with global alterations in gene expression, especially chromosome 10. The data suggest that genetic subgroups based on CAN and on gene expression are similar and that they are associated with tumor behavior. We are currently identifying genes that characterize these groups and plan to evaluate their function in tumor models. This work was supported by NIH (NS42927 and CA85799) and the National Brain Tumor Foundation.


C.E. Pelloski, K. Hess, W.K.A. Yung, J.L. Liu, H. Yang, A. Ledoux, and K.D. Aldape; Departments of Radiation Oncology, Biostatistics, and Neuro-Oncology, Pathology and Brain Tumor Center, The University of Texas, M.D. Anderson Cancer Center, Houston, Texas, USA

The AKT-PI3K pathway represents an important target for therapeutic intervention in glioblastoma. The relationship of activation of downstream intermediates with the response to radiation and overall survival in GBM is not known. To address this, we tested newly diagnosed GBMs from patients (n = 141) who underwent a subtotal resection (STR) with residual enhancing disease. Following surgery, pre-radiation and post-radiation images were compared and were scored for radiation response on a 5-point scale (ranging from .2 [progression] to +2 [complete response]). Response score was able to be determined for 134/141 cases and was highly correlated with survival (HR = 0.7, P < 0.01) providing validation that imaging-assessed scoring represented a meaningful end point. Tumor tissue was stained for PTEN, the phosphorylated forms of AKT (Ser473), mTOR (Ser2448), and p70S6K (Thr389) and scored semiquantitatively. Spearman correlations indicated that the expression of p-AKT, p-mTOR and p-p70S6K were all highly correlated with each other (all P <0.01), suggesting concordant activation of these pathway members. While a previous study identified a significant negative correlation of PTEN expression with p-AKT, a significant correlation between PTEN and p-AKT was not identified (P = 0.56) in these data. In fact, PTEN expression was positively correlated with p-mTOR and p-p70S6K (both P < 0.03), suggesting a complex relationship between AKT activation and expression of PTEN protein. With respect to clinical end points, higher levels of p-AKT were associated with worse radiation response score (P = 0.03, Spearman) and shorter time to progression (P = 0.01, Cox). While none of these markers were significantly associated with overall survival, we noticed a trend toward poor survival in cases with high p-mTOR expression in the patients (n = 43) who were less than 50 years of age. To address this, we tested an additional 30 GBM cases from patients who were under age 50 who had received a gross total resection. Analysis of the combined data set (n = 73) confirmed that p-mTOR was a significant negative prognostic factor in GBM patients under age 50 (P = 0.03). Taken together, the data indicate co-activation of AKT pathway members in GBM tumors and suggest an association of pAKT with diminished radiosensitivity and p-mTOR with worse survival in younger patients. The expression of PTEN protein may not adequately represent its functional status in tumor samples.


C.E. Pelloski, M.H. Maor, E.L. Chang, A. Mahajan, H. Colman, H. Yang, A. Ledoux, and K.D. Aldape; Departments of Radiation Oncology, Neuro-Oncology, and Pathology and Brain Tumor Center, The University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA

Response to radiation is an important prognostic factor in GBM, and efforts to better understand the molecular mechanisms underlying responsiveness/resistance may lead to novel targeted agents. We identified YKL-40, an extracellular matrix protein of unknown function, as a potential oncogenic factor in GBM in a genome-wide expression microarray screen. To explore this finding further, we tested newly diagnosed GBMs from patients (n = 139) who underwent a subtotal resection (STR) who had residual enhancing disease. Following surgery, pre- and post-radiotherapy MRI images were compared and scored for radiation response based on a 5-point scale (ranging from .2 [progression] to +2 [complete response]). Response score was able to be determined for 133/139 cases and was highly correlated with survival (HR = 0.72, P < 0.01); providing validation that imaging-assessed scoring represented a meaningful end point. Poorer response was also associated with advanced age (<50 vs. ≥50, P = 0.03, Spearman) as has been previously described. Tumor tissue was stained for YKL-40, scored semi-quantitatively and compared with response and overall survival. Increased expression of YKL-40 was significantly associated with poorer radiation response (P < 0.01), decreased overall survival (P = 0.03) and with older age (P = 0.02). To confirm an association of YKL-40 expression with outcome, we tested an additional set of newly diagnosed GBM patients (n = 140) who underwent a gross total resection (GTR). YKL-40 was significantly associated with overall survival in the GTR group as well (P < 0.01). Patients with intermediate to high YKL-40 staining had a median overall survival of 49 weeks versus 116 weeks for those with no YKL-40 expression. When combined with the STR group (total n = 279), multivariate analysis indicated that YKL-40 was an independent prognostic factor (HR = 1.3, P = 0.01), after adjustment for age. Overall, the data indicate that YKL-40 overexpression may promote radioresistance and confer a worse prognosis in GBM.


B. Pollo, E. Maderna, and A. Salmaggi; Istituto Neurologico C. Besta, Milano, Italy

Recently, cancer stem cells are reported in brain tumors and in other malignancies as cells that have a role in proliferation and differentiation and that have self-renewal capacity. In order to identify the presence and the distribution of multipotential neuroepithelial stem cells that could include cancer stem cells in brain tumors, we investigated the immunohistochemical expression of nestin and alpha-internexin on 100 CNS tumors in children and in adults. The aim of our study was also to recognize early stages of differentiation in tumoral cells and to investigate endothelial proliferations. Nestin and alpha-internexin are intermediate filaments (IFs) of the cytoskeleton of cells during the development of CNS and may be expressed in tumors. In embryonic brain, IFs are involved in early stages of lineage commitment, in proliferation, and in differentiation. Nestin is a class VI IF, highly expressed in embryonic progenitor cells, and is a marker for multipotential neuroepithelial stem cells, detected in neuroepithelial tumors and in endothelial cells in active proliferation. Alpha-internexin is an IF protein of class IV, abundantly expressed in the early neuronal development and plays a role in neuronal differentiation. Our cases were 60 gliomas of different phenotype and grade of malignancy, 15 ependymomas, 20 medulloblastomas, and 5 PNETs. Our data were related to the presence of other IFs of the cytoskeleton (neurofilaments, GFAP, vimentin) markers for differentiation and to the proliferative potential (evaluated with Ki67/MIB1). Higher immunostaining for nestin and alpha-internexin was found in medulloblastomas and PNETs with partial correlation to neuronal differentiation. Any clinical correlation was found in these tumors. In gliomas alpha-internexin was absent, while nestin-positive cells were detected in all glioblastomas, variably expressed in neoplastic cells and in endothelial proliferations, and in other gliomas, focal or weak immunoreactivity in cells and in vessels was found. Nestin was immunodetected also in the major part of ependymomas, mostly in anaplastic cases. Our data suggest that nestin and alpha-internexin in embryonal tumors may indicate early neuronal differentiation, nestin in gliomas and ependymomas seems to be related to malignancy grade, and especially in low-grade gliomas, might provide useful information related to angiogenesis. Our results might add prognostic information for patients with variable life expectancy.


A. Ray, M. Ho, J. Ma, R. Parkes, T.G. Mainprize, S. Ueda, J. McLaughlin, R. Bouffet, J.T. Rutka, and C.E. Hawkins; Divisions of Neurosurgery, Oncology, and Pathology, The Hospital for Sick Children, Toronto, Ontario; Department of Epidemiology and Biostatistics, The Mount Sinai Hospital, Toronto, Ontario; Canada

The objective of this study was to determine the accuracy with which biological markers influence survival in infant medulloblastoma. We obtained clinical presentation and survival information for 19 patients who had surgery at The Hospital for Sick Children. All patients were less than 36 months old, had had no radiotherapy for primary disease, and had received chemotherapy. A tissue microarray was constructed from their tumour samples. The arrays were immunohistochemically assayed for expression of P53 and ErbB2, and their ability to influence survival was tested. The age of presentation varied from 1 to 31 months. The average follow-up was 152 weeks (range, 13 days–514 weeks). The overall mortality rate was 55%. P53 was positive in four patients, all of whom were dead (mortality rate 100%), and ErbB2 was present in five patients, four of whom were dead (mortality rate 80%). Of the eight patients with metastatic disease, six were dead (mortality rate 75%). Of the 11 patients who were p53 and Erb B2 negative, only four patients were dead (mortality rate 36%). Though considered to be clinically high risk, the survival in the infant medulloblastomas is influenced by the presence of molecular markers that have been established in the average risk population.


Adrienne C. Scheck, Jeanette K. Pueschel, Lynn S. Ashby, Stephen W. Coons, and William R. Shapiro; Barrow Neurological Institute, Phoenix, Arizona, USA

Oligodendrogliomas carry a more favorable prognosis than astrocytic tumors, and anaplastic oligodendrogliomas with deletions of 1p and 19q are typically more sensitive to radiation and chemotherapy. The presence of 1p/19q deletions generally predicts a favorable prognosis; however, some patients with these tumors rapidly succumb to their disease, suggesting a need for additional genetic markers of prognosis. In addition, the contribution of cellular heterogeneity to the interpretation of molecular genetic results has not been addressed, and it is not clear whether the percentage of cells in a tumor with 1p/19q deletions is clinically relevant. These issues are likely to be contributing factors to the confounding results reported in the literature. Furthermore, while molecular genetics are emerging as predictors of prognosis in patients with high-grade oligodendrogliomas, the utility of these tests in determining the prognosis of patients with low-grade oligodendrogliomas or mixed tumors is uncertain. To address these issues, we are using fluorescent in situ hybridization (FISH) to analyze 150 anaplastic oligodendrogliomas, low-grade oligodendrogliomas, and mixed oligoastrocytomas for the presence, prevalence, and regional distribution of cells with deletion of 1p and 19q. Results to date demonstrate that many tumors are heterogeneous with respect to the distribution of cells with 1p/19q deletions. This heterogeneity does not appear to be correlated with tumor grade, nor is it more common in mixed tumors compared with pure oligodendrogliomas. Statistical analyses of 1p/19q deletion, heterogeneity, and prevalence (%) of cells with deletions compared to survival and time to progression are under way. To further define patients with poor prognosis, we have begun an analysis of PTEN deletion in these same tumors. Our results to date demonstrate that PTEN deletion is rare when defined as the loss of PTEN without loss of chromosome 10; however, in some tumors, as many as 50% of the cells had only a single copy of chromosome 10. Our preliminary analyses suggest that loss of PTEN does not correlate to tumor grade or type. Multivariate statistical analyses are under way to determine if 1p/19q and/or PTEN status will provide additional information to define subsets of patients with a poor prognosis despite diagnoses of low-grade oligodendroglioma or mixed oligoastrocytoma. In addition, further analyses will determine if PTEN status will help identify a more aggressive subset of anaplastic oligodendrogliomas with 1p/19q deletions.


Ezequiel I. Surace,1 Eriks Lusis,2 Erin Winkeler,1 Yoshinori Murakami,3 Bernd W. Scheithauer,4 Arie Perry,2 and David H. Gutmann1; 1Department of Neurology and 2Division of Neuropathology, Washington University School of Medicine, St. Louis, Missouri, USA; 3Tumor Suppression & Functional Genomics Project, National Cancer Center Research Institute, Tokyo, Japan; 4Division of Neuropathology, Mayo Clinic, Rochester Minnesota, USA

Meningiomas constitute the second most common central nervous system tumor affecting adults. Compared to the glioma, relatively less is known about the molecular pathogenesis of meningiomas. Previously, we and others have shown that the two most frequent early genetic changes in meningioma tumorigenesis involve inactivation of the neurofibromatosis 2 (NF2) and 4.1B genes. Recently, 4.1B was shown to interact with the Tumor Suppressor in Lung Cancer-1 (TSLC1) protein, prompting us to examine the expression of TSLC1 in meningiomas. We developed specific anti-TSLC1 antibodies to examine TSLC1 expression in normal human leptomeninges, human meningioma cell lines, and human meningiomas of different pathological grades by Western blot and immunohistochemistry. Whereas TSLC1 was expressed in normal human leptomeninges by immunohistochemistry as well as in normal mouse brain and lung, TSLC1 expression was absent in three human malignant meningioma cell lines and in 30% of benign meningiomas by Western blot. To determine whether TSLC1 functioned as a tumor suppressor for meningioma, we re-expressed TSLC1 in TSLC1-deficient human meningioma cell lines. In keeping with its function as a meningioma negative growth regulator, restoration of TSLC1 expression resulted in significantly reduced cell proliferation. Lastly, we examined a large series of 123 meningiomas representing all malignancy grades. Loss of TSLC1 expression was frequently observed in anaplastic (grade III) meningiomas (82%) and the subset of WHO grade II meningiomas with high proliferative indices (87%). Moreover, loss of TSLC1 expression was associated with decreased overall patient survival, both in the entire cohort and in the atypical WHO grade II meningiomas, which exhibit the greatest variability in clinical behavior. Collectively, these results suggest that TSLC1 plays an important role in meningioma growth regulation and that TSLC1 loss may identify a subset of meningiomas with aggressive clinical behavior and reduced patient survival.


C.J. Watling; Departments of Clinical Neurological Sciences and Oncology, University of Western Ontario, London, Ontario, Canada

Angiocentric immunoproliferative lesions (lymphomatoid granulomatosis) represent an unusual angiocentric and angiodestructive lymphoproliferative process predominantly affecting the lungs, but often also involving extrapulmonary sites including the skin, kidneys, upper respiratory tract, and the peripheral and central nervous systems. CNS involvement occurs in 20% of affected individuals and is a negative prognostic feature. Two cases with prominent involvement of the central nervous system are described. Case 1 was a 54-year-old woman who presented with headache, cognitive changes, and left hemianopsia, and cranial MRI revealed a large ring-enhancing lesion in the right temporal lobe with mass effect. Two brain biopsies were nondiagnostic, revealing only an atypical lymphoid infiltrate without clearly neoplastic features. She responded clinically and radiologically to corticosteroid treatment, only to present with recurrence of her neurologic symptoms six months later, when repeat imaging revealed enlargement of the right temporal mass and a new asymptomatic lesion in the left lung. Biopsy of the lung lesion confirmed the diagnosis of lymphomatoid granulomatosis, but she died of rapidly progressive neurologic disease despite reinstitution of high-dose corticosteroid treatment. Case 2 was a 65-year-old man who presented with ascending numbness in the legs followed by progressive paraparesis. MRI showed areas of high T2 signal within the thoracic spinal cord as well as the brainstem and subcortical white matter. Initial improvement with corti-costeroid treatment was not sustained, and he developed increasing weakness, along with dysphagia and dysarthria, and repeat imaging showed more extensive signal change within the cerebral white matter. Brain biopsy showed changes consistent with a diagnosis of angiocentric immunoproliferative lesion. Although vasculitis was considered in the differential diagnosis, cerebral angiography was normal. Treatment with corticosteroids and monthly pulses of intravenous cyclophosphamide has resulted in clinical and radiologic improvement, over a 9-month follow-up period. These two cases illustrate that angiocentric immunoproliferative lesions may affect the central nervous system predominantly or exclusively. Central nervous system involvement is variable and may involve a tumor-like lesion mimicking malignant glioma (Case 1) or diffuse involvement mimicking acute disseminated encephalomyelitis (Case 2). Diagnosis may be difficult, even with biopsy, and a high index of suspicion is required. Relapse seems likely after corticosteroid therapy alone, and early aggressive intervention should be considered.



Deborah H. Allen, Karen Carter, Susanne Jackson, Karen Ziegler, Mary Lou Affronti, Renee Dunn, and Bebe Guill; Brain Tumor Center, Duke University Medical Center, Durham, North Carolina, USA

Clinically significant neurological deficits are frequently observed in patients with gliomas. This health state is often reported by patients and families to impact perceived quality of life (QOL). Therefore, the individual’s perceived QOL should be considered in discussions regarding therapy and potential outcome. Quality of life, defined as the physical, psychological, and social elements that are involved in the individual’s ability to perform daily activities, is associated with patient and family satisfaction of their current health state and treatment modality. As we previously reported at the SNO 8th Annual Meeting, only 64 of 488 patients (13%) had moderate to severe depression as reported on the Beck Depression Inventory–II (BDI-II). However, the majority of patients reported symptoms of fatigue (n = 375), appetite disturbances (n = 265), sleep disturbances (n = 349), or sexual dysfunction (n = 253) affecting their perceived QOL. In an attempt to better understand the impact of symptoms and symptom management on QOL, we are retrospectively reviewing these specific items (appetite or sleep disturbances, fatigue, sexual dysfunction) as reported by patients on the BDI-II over multiple clinic encounters. The BDI-II is a 23-item, four-choice statement inventory that contains specific questions relating to change in appetite, sleep, fatigue, and sexual desire/function; the total score indicates the severity of depression. The BDI-II was completed by the patient/proxy as an adjunctive screening tool in our adult clinic from 09/01 to 03/02. All interventions initiated for symptom management were recorded. Since many patients had multiple clinic encounters over these six months, the impact of symptom management on their perceived QOL can be assessed. Analyses pertain to frequency, the Fisher exact tests, and chi-squares for evaluation of significance. Over the six months, 488 patients (males = 276, female = 212) completed the inventory (95% completion rate) for a total of 969 clinic visits. There were 184 patients evaluated twice, 97 patients evaluated in three visits, 9 patients evaluated four times, 7 patients had five visits, and 3 patients had six monthly evaluations. Glioma grade and treatment modality did not predict severity of symptoms. However, level of depression and severity of symptom change (score greater than 1) are clearly associated. For those patients who had multiple clinic encounters, the preliminary analyses of the symptom management have suggested clinical improvement of these symptoms. Statistical significance and further analyses of these encounters in association of tumor progression, prior and current tumor treatment modalities, and steroid use are ongoing and will be presented. As anticipated, patients with glioma report many associated symptoms that impact their perceived QOL. Early evaluation and intervention may improve their QOL, alter level of depression, and improve overall clinical outcome. Assessment of these symptoms by the use of an easily administered self-report tool is a valuable adjunct for the clinical setting.


Terri S. Armstrong,1,2 Tito Mendoza,1 Marlene Z. Cohen,1,2 Marilyn Morrissey,1 Mark R. Gilbert,1 and Charles Cleeland1; 1M.D. Anderson Cancer Center, Houston, Texas; 2University of Texas Health Science Center Houston, School of Nursing, Houston, Texas; USA

The aim of this study is to validate the M.D. Anderson Symptom Inventory–Brain Tumor Module (MDASI-BT) in patients who have primary brain tumors and to evaluate the occurrence and severity of selected symptoms in this patient population. The occurrence of multiple symptoms has been shown to impact patient outcomes, such as quality of life, mood, disease progression, and survival in other solid tumor patients. The occurrence of multiple symptoms in patients with primary brain tumors has not been routinely measured. Tools specifically designed to measure multiple symptoms in persons with cancer have been developed for use in patients with other solid tumors. No instrument has been validated that includes both neurologic symptoms and symptoms commonly associated with cancer and its treatment. The study sample included 200 adult patients with PBT presenting to the clinic or inpatient unit at MDACC. It is anticipated that at the time of presentation all data will be collected and analysis will be completed. The study will evaluate the reliability and validity of the MDASI-BT in primary brain tumor patients. Demographic and disease related data will be collected at the time the MDASI-BT is completed. Reliability will be assessed by calculation of Cronbach’s alpha. Construct validity will be determined by using confirmatory factor analysis. In addition, descriptive statistics will be used to describe how patients rate symptom severity and interference with function. Finally, we will examine the relationship of symptom severity to diagnosis, concomitant medications, Karnofsky performance status, and where the patient is in treatment. The development of an instrument to measure multiple symptoms in the primary brain tumor population is the first step in a program of research aimed at improving symptom management in this patient population.


Beverly Barkon,1 Susan O’Rourke, 1 Robert Reed, 1 Lorraine Reiser1 and Regina I. Jakacki,2 and Student Contributors: Leslie D’Antonio,1 Sherri Scheboth,1 Lori Dunlay,1 and Louise Waszak1; 1Carlow College, Pittsburgh, Pennsylvania; 2Pediatric Hematology/Oncology, Children’s Hospital of Pittsburgh, Pittsburgh, Pennsylvania; USA

As the number of survivors of pediatric brain tumors continues to grow, a new challenge has arisen: how to increase the likelihood of success in school for the children. School success is a major component of quality of life for survivors, and it requires that families and schools possess a flexible manner of addressing the academic needs of these students. Often the children have not been identified for special education services prior to the diagnosis and treatment of the tumor. The school may offer home instruction during the period of time the child is under active treatment. Likewise, the school may offer the opportunity to reintegrate the child into his “appropriate” grade and class when he or she is ready to return to school and be reluctant to see the child any differently than they did before the diagnosis. Schools may take a “wait and see” perspective and deal with problems as they arise. The decision to evaluate a child for special education services may be far removed from the diagnosis and treatment stage of the tumor. The relationship between current school problems and the disease may not be seen or if seen may be viewed as unrelated to the current problems. These are but of few of the school issues families face. Another problem is that the knowledge base regarding the effects of tumor and its treatment are not part of teacher education, and most teachers and schools are forced to get the knowledge they need only when the need arises. If they do not see the connection between the problems the children demonstrate and the disease, they may not look in the right places to get that information. It is important that the information schools need to serve survivors will be available in a timely manner so that the possibility of school success is maximized. This poster will present an innovative program housed in a division of teacher education, composed of a cross-discipline team of faculty and graduate students with advisement from a pediatric neurooncologist. The disciplines currently represented on the team include special education, nursing, counseling psychology, and instructional technology. There is one faculty member and one student from each discipline focused on identifying the educational and social emotional needs of a child and his family and supporting the school and teacher as they serve the child in the classroom. There are four goals of this project: to develop a knowledge base among school professionals who will serve more survivors as the numbers continue to increase; to identify the needs of each child in a sophisticated, integrated manner; to support the social and emotional needs of the family and child; and to develop a distance technology model for the support of the teacher and the school.


Nicole M. Beadles and Elana Farace; Department of Neurosurgery, University of Virginia, Charlottesville, Virginia, USA

Depression is a common affective reaction in patients with brain tumors. Many studies have examined the risk factors associated with developing depression in this population, as well as in populations with other types of cancers, and these factors often include but are not limited to financial troubles, relationship difficulties, incapacity in daily living skills, inactivity, and premorbid psychiatric history. Rather than focus on what contributes to depression, this study focuses on the factors that are associated with psychological well-being, or absence of depression, in brain tumor patients. This study uses the positive aspects of the previously identified variables, as well as others, as predictors of freedom from depression. Using a sample of 75 patients that were seen at the Departments of Neurosurgery Division of Neuropsychology at the UVa Health System over a four-year period, we reviewed charts for the following information on each subject: age, gender, race, educational level, marital status, malignancy of tumor, tumor location (L/R hemisphere), extent of family/community support other than caregiver, financial security (as measured by the hospital’s income scale), activity or energy level (high or low), independence in activities of daily living (ADLs; e.g., driving, cooking, caring for oneself), family psychiatric history, and personal experience with other medical conditions or surgery. The Beck Depression Inventory–II (BDI-II), which is a measure of depression given to each patient, was the dependent measure. All charts were coded prior to the BDI-II scores being recorded so as to reduce bias. For data analysis, first correlations are run to determine which variables did not correlate with the BDI-II. Of the remaining variables, a factor analysis will be conducted to determine inter-correlation and to reduce the number of constructs in the study. When clear factors emerge, these factors will be used as predictors of freedom from depression in stepwise, multivariate regression equations. If each variable is used separately, co-variates will include demographic information including age, gender, race, and education. It is predicted that tumor location, ADL independence, activity or energy level, and existence of family and community support will be significant predictors of freedom from depression. Results will aid the medical community in creating appropriate interventions to improve the affective status of patients with brain tumors.


Raylene Coleman1 and Leslie D. McAllister2; 1The Neurological Clinic, Portland, Oregon; 2Legacy Brain and Spinal Cord Tumor Program, Good Samaritan Hospital, Division of Hematology and Medical Oncology, Oregon Health Sciences University, Portland, Oregon; USA

Seizures remain an important quality of life issue for patients with primary and metastatic brain tumors (BTs). Approximately 40% to 60% of BT patients will experience at least one seizure during their illness. However, primary prevention is limited by anticonvulsant drug side effects and drug interactions. Levetiracetam (LVT) is an attractive therapeutic option because of established efficacy for lesion-related epilepsy, excellent patient tolerance, and lack of drug interactions. The objective of this study was to conduct a preliminary assessment of effectiveness and tolerability of LVT in adult patients with primary brain tumors. We retrospectively reviewed charts for 28 patients with regard to tumor type and location, seizure type and frequency, prior antiepileptic drug (AED) use, and dose and tolerance of LVT. The study population includes 28 patients, ages 27 to 69 years (median 43), 12 men and 16 women. Twenty (71%) patients had been on a prior AED, and 8 patients received LVT as their initial AED. LVT was started because of inadequate seizure control on current AED-21, allergic reaction to current AED-4, intolerable side effects from the current AED-2, and as primary prevention-1. Daily LVT doses ranged from 1000 to 4000 mg. Overall, 22 patients (79%) had a significant reduction (>50% decrease) in seizure frequency, and 64% of patients became seizure free. Of 16 patients, 13 (81%) who achieved successful transition to monotherapy became seizure free. No patient experienced status epilepticus or allergic reaction while on LVT. Eight patients (29%) reported side effects while on LVT, including dizziness (4), personality changes (2), imbalance (2), anorexia (1), mild cognitive difficulties (2), mild sedation (1), and slurred speech (1). Side effects were more common when LVT was combined with another AED and decreased with time. No patient discontinued LVT therapy because of side effects. LVT appears to be effective and well tolerated in primary brain tumor patients. LVT should be considered for initial therapy in selected patients or as add-on therapy when the initial AED is insufficient to control seizures. A prospective, open-label trial using LVT as primary prevention for BT-related seizures is planned.


T. Kaleita, T. Cloughesy, J. Ford, W. Steh, L. Nghiemphu, C. Graham, S. Segal, and D. Wellisch; Departments of Psychiatry, Neurology, and Radiation Oncology, the UCLA Neuro-Oncology Program, and the Henry E. Singleton Brain Cancer Research Program, David Geffen School of Medicine at UCLA, University of California at Los Angeles, Los Angeles, California, USA

This abstract is a description of an ongoing clinical trial to evaluate safety and efficacy of Modafinil for treatment of two key clinical outcomes in adult brain tumor patients, namely (1) fatigue and (2) neurobehavioral dysfunctions. Inclusion criteria for patients were as follows: primary malignant brain disease; 21 to 65 years of age; neurosurgery, chemotherapy, and/or radiation; mild to severe fatigue and/or attention/memory impairment. Thus far, the study has enrolled 15 of 30 patients. Phase 1 of the trial (3 weeks) involves double-blinded randomization between two dosage levels of Modafinil (200 mg vs. 400 mg/day in divided doses). Following a one-week washout, an extended open-label treatment phase is 8 weeks in duration. Measures include physical/neurologic examinations, fatigue questionnaires, cognitive tests, and a depression scale. Longitudinal evaluation of each patient involves a combination of measures to evaluate the primary end points, which include (1) fatigue, (2) attention/concentration, (3) psychomotor speed, and (4) depression. Patient-, disease-, and therapy-specific variables will be considered when analyzing results. This is the first study to assess the effects of Modafinil in the adult brain tumor patient population. The initial report of findings from this clinical trial will be presented.


Raja B. Khan, Ramana M. Chitti, Frederick A. Boop, and Thomas E. Merchant; St. Jude Children’s Research Hospital, Memphis, Tennessee, USA

We conducted a prospective neurologic evaluation of children after ependymoma surgery to establish patterns of neurologic recovery. Eligibility criteria included diagnosis of posterior fossa ependymoma, maximal surgical resection, conformal radiation treatment (RT) on an institutional protocol, and at least 2-year follow-up after RT without recurrent tumor. All patients were periodically assessed by the study neurologist. When present, neurologic deficits were defined as “hard signs” when they involved cranial nerves, sensory-motor function, or axial dysmetria and as “soft signs” when present in fine finger movement, alternating hand movements (dysdiadokinesia), gait, static balance, extraocular movements, speech (dysarthria), and nystagmus. Soft signs were graded as described by Stokman et al. (1986, Dev. Med. Child Neurol. 28:428), with 1 as normal and 4 as inability to perform the task. Performance status was assessed with Karnofsky performance status (KPS). Of the 66 treated on the protocol, 55 children met the study’s eligibility requirement. Of these 55, 50 had gross total tumor resection, 3 near total, and 2 subtotal. Median age at RT was 2.6 years (range, 1.1–22.9) and median follow-up is 48 months (range, 24–60). No soft signs were present at last follow-up in 12 (22%), 3 or less in 30 (55%), and 4 or more soft signs in 25 (45%) children. At the last follow-up, individual soft signs were not assessable in 9%–14%, normal throughout the study period in 6%–35%, present without improvement in 4%–27%, present with partial improvement in 1%–24%, and present initially with complete recovery in 4%–11%. A majority of children with complete recovery had a low-grade deficit. Mean KPS at study onset and last follow-up were 79 and 92 (P ≤ 0.0001). Children with 3 or fewer soft signs at last follow-up had better functional status compared to those with 4 or more at study onset (85 vs. 71, P = 0.0002) and at the last follow-up (96 vs. 87, P = 0.002). Except for one child who suffered post-RT meningitis, all children were functionally independent at the last follow-up. Children with 4 or more soft signs at the last follow-up were more likely to have received more than 1 surgical resection, or had CSF infection, or a ventriculoperitoneal shunt (P ≤ 0.0001) and were more likely to have hard neurologic signs present (80% vs. 30%, P = 0.003). We conclude that neurologic soft signs are common in children treated for posterior fossa ependymoma and improve with time in most. Higher numbers of soft neurologic signs are associated with >1 surgical resection, CSF infection, hydrocephalus requiring shunt, presence of hard neurologic signs, and worse functional status. Operated by experienced pediatric neurosurgeons, gross total tumor resection was achieved in most patients with retention of functional independence.


Sur Ja Min and John Slopis; Department Of Neuro-Oncology, The University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA

Most health-care providers encounter problems with the management of patients with neurofibromatosis type 1 (NF1). NF1 is a dominantly inherited neurocutaneous syndrome resulting from a mutation of chromosome 17 (at 17q 11.2). Most patients present with multiple medical and psychosocial problems and are diagnosed by clinical features. The incidence of NF1 is 1:4000 people. NF1 is the most common form of NF and is 10-fold more frequent than NF2. The clinical features of NF1 are café au lait spots, freckling in the axillaries or groin areas, optic gliomas, Lisch nodules in the iris, plexiform fibromas, and cutaneous neurofibromas. Other accompanying symptoms are learning disabilities, speech problem, attention deficit hyper-activity disorder in children, and attention deficit in adults. Most patients with NF1 seek health care for problems concerning their physical disability, pain secondary to a bulky tumor, or psychosocial issues. Although there is extensive literature and information about NF1, we are continuing to learn and develop methods to effectively manage this group of patients. A team composed of a physician, an advanced practice nurse, a clinic nurse, a social worker, and a patient service coordinator supports the Neurofibromatosis Clinic in the University of Texas M.D. Anderson Cancer Center Brain and Spine Center. NF1 and NF2 patients have multiple medical and psychosocial complications, and they needed long-term care plans to manage the individuals appropriately. The patients have a vast diversity of symptoms to manage. There are no clear guidelines for the diagnostic tests and treatment plans for the patients. Currently, the team has developed an algorithm to help guide diagnosis and management of these patients using a consistent pattern. This will make it easier for oncology nurses to develop plans for long-term care for their NF patients.


Stephen R. Rapp, Robin Rosdhal, Ralph B. D’Agostino Jr., James Lovato, Michelle Naughton, Michael E. Robbins, and Edward G. Shaw; Departments of Psychiatry and Behavioral Medicine, Radiation Oncology, and Public Health Sciences, Wake Forest University School of Medicine, and the Community Clinical Oncology Program, Comprehensive Cancer Center of Wake Forest University, Wake Forest University Baptist Medical Center, Winston-Salem, North Carolina, USA

Patients who have undergone cranial radiation for primary or metastatic brain tumors often suffer neurocognitive effects that interfere with daily activities, mood, and quality of life (QOL). Donepezil (Aricept) is an acetyl-cholinesterase inhibitor that has been effective in improving cognitive functioning in patients with dementia. It is unknown whether donepezil could also be useful in this population. This study is an open-label, Phase II clinical trial of donepezil. Patients were recruited if they were >18 years of age had a life expectancy >30 weeks, had completed partial or whole-brain irradiation of at least 3000 cGy >6 months prior to enrollment, had had no evidence of progressive brain disease for 3 months, were on a stable/decreasing steroid dose, had a Karnofsky score >70, and had no planned brain tumor treatment therapy during study. Each patient received 5 mg/day of donepezil for 6 weeks, which was increased to 10 mg/day for 18 weeks, followed by a 6-week wash-out period. Neurocognitive measures were taken at baseline, 6 weeks, 12 weeks, 24 weeks, and 30 weeks. Neurocognitive functions assessed (and measures) included global cognitive function (Mini Mental State Exam), verbal fluency (F-A-S), attention/concentration (Trail Making Test-Part A, Digit Span Test), verbal memory (California Verbal Learning Test-II), figural memory (modified Rey-Osterrieth Complex Figure), and executive function (Trail Making Test-Part B). Health-related QOL and mood were also assessed and are reported elsewhere. All measures were administered and scored by a trained and certified research nurse. Changes in neurocognitive scores from pre-treatment baseline to week 24 are presented. Data from 24 (mean age = 43 years; female = 47%; white = 91%) of 34 enrolled patients who completed all assessments were analyzed. Scores significantly improved between baseline and week 24 on verbal fluency (FAS P = 0.02), verbal memory (CVLT-II: short-delay cued verbal recall, P < 0.0001; long-delay cued recall, P = 0.004; long-delay recall, P = 0.02), attention/concentration (DST: backward, P = 0.03; total, P = 0.007; TMT-A: errors, P = 0.03), and figural memory (ROCF: immediate recall P < 0.0001; delayed recall, P < 0.0001). There were no significant changes for global cognitive function (MMSE, P = 0.11) or executive function (TMT-B, P = 0.64). No significant worsening of performance was noted on any measures. A 24-week course of donepezil was associated with improved performance in several important cognitive functions in middle aged adults. Donepezil may be useful in reducing the adverse late effects of partial or whole-brain irradiation and improve cognitive function of patients. Research was supported by NCI grant 1 U10 CA 85851.


L.R. Rogers,1 B.A. Given,2 C.W. Given,2 S.R. Remer,1 and P.R. Sherwood2; 1Henry Ford Hospital, Detroit, Michigan; 2Michigan State University, Lansing, Michigan; USA

Caregivers (CGs) play an important role for brain tumor patients (BTPs) because of the cognitive and physical limitations induced by this disease. Since the survival of most BTPs is short and treatments are often associated with toxicity, we questioned CGs after the BTP died in order to determine if the CGs perceived the treatments to be beneficial. We devised and mailed a 34-item questionnaire regarding treatments the BTP received to 60 CGs after the BTP died. Of these, 70% responded. CG age was 26 to 63 years, median 47. In 62%, the CG was the spouse; 78% of BTPs were male. The majority of tumors were glioblastoma. Survival from tumor diagnosis was 1 to 108 months, median 17. Respondents were the primary CG for 1 to 84 months, median 13. Treatments included brain radiation (41), surgical resection (35), and chemotherapy (33). Sixteen BTPs were on a clinical trial. CGs perceived the majority of treatments (surgery 80%, radiation 62%, and chemotherapy 77%) as beneficial, and 64% of clinical trials were perceived as beneficial. However, treatment was perceived to reduce the BTP’s quality of life in 48% of patients receiving radiation and 44% receiving chemotherapy. Unexpected complications were encountered from surgery (54%), radiation (49%), and chemotherapy (44%). Common themes identified in self-proffered comments include inadequate brain tumor treatment information provided to the BTP and CG, disappointment that brain tumor treatments did not improve the BTP’s quality of life, and a regret that the BTP had not consulted a major medical center. The majority of CGs perceived that treatments benefited the BTP, but at the risk of reduced quality of life. Health care providers for BTPs should be aware of this, as well as the CG perceptions regarding difficulty in obtaining information about treatments and unexpected complications from all types of therapy. A prospective analysis is indicated to determine the adequacy of treatment information (including side effects) and quality of life expectations for BTPs and CGs while the patient is alive.


Hadas Rosenne and Tali Siegal; Leslie and Michael Gaffin Center for Neuro-Oncology, Hadassah Hebrew University Hospital, Jerusalem, Israel

Patients diagnosed with malignant glioma are facing a threat on their physical, emotional, and cognitive skills, and they also have to cope with the severity of their prognosis. This affects both the patient and the whole family. Comprehending the ways patients and families understand their condition can help improve multi-disciplinary intervention. The objective of this study was mapping ways by which both patients and families grasp and cope with the severity of diagnosis and prognosis. This study was a retrospective content analysis of social work service files of 40 patients with malignant gliomas, treated by the same social worker from diagnosis until death. Two main categories were evaluated and coded based on the reports in the files: (a) level of understanding the prognosis and diagnosis and (b) expressions of optimism. The level of understanding was classified as follows: (1) full understanding: accepting the fact that death is inevitable as reflected in talking about death and/or making preparations for it; (2) partial understanding: comprehending a threat to life while not accepting death as a necessary consequence; (3) lacking understanding: realizing that there is some illness that requires treatment and follow-up without understanding its implications. Expressions of optimism were classified as either existing optimism or as the lack of it based on verbal expressions of planning the future and maintaining hope. These variables were coded both for patients and their families. Level of understanding was measured for two points in time: shortly after diagnosis and later with progression of symptoms. Chi-square tests were conducted to evaluate the association between these categorical variables. At diagnosis the rate of patients with full comprehension was higher among patients whose families had higher levels of understanding as compared to patients whose families had lower level of comprehension (P < 0.01). At progression of symptoms the rate of patients expressing full understanding was higher among those who expressed full understanding soon after diagnosis, in comparison to patients who expressed lower levels of comprehension (P < 0.01). For family members the findings were similar to those of patients regarding the rate of full understanding at the two points of evaluation (P < 0.05). The rate of optimism expressions was higher among families with lower level of understanding only at diagnosis (P < 0.01), but no difference was found at clinical deterioration. No significant association was found between patients’ expression of optimism and their level of understanding at the 2 points of evaluation. Understanding the diagnosis of malignant gliomas is a process that evolves and changes over time, parallel to the development of new symptoms. There is often a gap between the severity of symptoms and the gravity of prognosis. This gap requires emotional coping that is processed differentially by patients and families and has implications on levels of understanding and optimism.


Edward G. Shaw, Robin Rosdhal, Ralph B. D’Agostino Jr., James Lovato, Michelle Naughton, and Stephen Rapp; Departments of Radiation Oncology, Public Health Sciences, and Psychiatry and Behavioral Medicine, Wake Forest University School of Medicine, and the Community Clinical Oncology Program, Comprehensive Cancer Center of Wake Forest University, Wake Forest University Baptist Medical Center, Winston-Salem, North Carolina, USA

Long-term survivors of partial or whole-brain irradiation (RT) for primary or metastatic brain tumors often experience adverse sequelae of treatment that impact health-related quality of life (HRQL), mood, and cognitive function. We report the results of a prospective, open-label Phase II study of donepezil, an FDA approved acetylcholinesterase inhibitor used to treat mild to moderate dementia of the Alzheimer’s type. Major eligibility criteria included age >18, life expectancy >30 weeks, prior partial or whole-brain RT >6 months prior to enrollment, no imaging evidence of progressive brain disease in 3 months prior to enrollment, stable/decreasing steroid dose, Karnofsky Performance Status (KPS) >70, and no planned brain tumor therapy during the 30-week study period. Patients received donepezil 5 mg/day for 6 weeks, then 10 mg/day for 18 weeks, followed by 6 weeks observation off drug. HRQL tests included KPS, Functional Assessment of Cancer Therapy (FACT) with the Brain (BR) Subscale, and the Profile of Mood States (POMS). Cognitive function was also assessed and is reported separately. HRQL assessments were made at baseline and at weeks 6, 12, 24, and 30. Results will be presented comparing baseline to 24-week HRQL data. Of 34 patients who initiated the study, 24 remained on study for 24 weeks and completed all QOL testing. All 24 had primary brain tumors, mostly glial in origin. Overall HRQL as measured by the FACT-BR improved (increased) from baseline to 24 weeks (mean baseline score 124 [SD = 24] vs. mean 24-week score 134 [24], P = 0.0065). Several of the FACT-BR subscales also showed significant improvements in the mean scores from baseline to 24 weeks: emotional functioning (18 [4] vs. 20 [3], P = 0.0383), social/family (20 [5] vs. 22 [5], P = 0.0165), and the brain subscale (48 [11] vs. 54 [11], P = 0.003). Total score on the POMS also improved (decreased) from baseline to 24 weeks, indicating an improved mood (mean baseline score 47 [38] vs. mean 24 week score 30 [30], P = 0.0272). Significant improvements in the mean scores from baseline to 24 weeks of several domains also occurred: confusion (12 [5] vs. 8 [4], P = 0.002) and fatigue (12 [7] vs. 10 [6], P = 0.0383). There was no difference between the baseline and 24 week KPS values (88 [9] vs. 90 [8], P = 0.13). The most common toxicities were fatigue, insomnia, and diarrhea, most of which were mild. Health-related quality of life and mood, as measured by the FACT-BR and POMS, were significantly improved with a 24-week course of the acetylcholinesterase inhibitor donepezil. A similar study utilizing ginkgo biloba, 120 mg/day, is ongoing. This research was supported by NCI grant 1 U10 CA81851.


Martin J.B. Taphoorn, Roger Stupp, David Osoba, Jürgen Curschmann, Rolf Kortmann, Martin van den Bent, Warren Mason, Corneel Coens, Elizabeth Eisenhauer, and Andrew Bottomley; EORTC Brain Tumor Group and Radiotherapy Group, European Organization for Research and Treatment of Cancer, Brussels, Belgium; National Cancer Institute of Canada Clinical Trials Group, Kingston, Ontario, Canada

In a randomised trial radiotherapy with concomitant and adjuvant temozolomide, chemotherapy has been shown to prolong survival in newly diagnosed glioblastoma multiforme (GBM) compared to standard radiotherapy alone. This analysis evaluates health-related quality of life (HRQOL) of the patients included in the pivotal phase III trial. Five hundred seventy-three newly diagnosed GBM patients were randomized to radiotherapy alone or radiotherapy and concomitant and adjuvant temozolomide. The primary end point was survival, with HRQOL being a secondary end point. HRQOL was assessed at baseline, and during treatment at three-month intervals until progression using the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire C30 and the Brain Cancer Module (BN 20). Changes from baseline in the scores of seven preselected HRQOL domains (fatigue, overall health, social functioning, emotional functioning, future uncertainty, insomnia, and communication deficit) were calculated for both groups, as well as differences on each time point between patient groups. Statistical significance and proportions of patients with improved HRQOL scores (changes of ≥10 points) were calculated. Baseline questionnaires were available in 85.7% (n = 490) of patients. There were no significant differences in baseline scores for either treatment group on any HRQOL scales. During radiotherapy, a significant difference (P < 0.05) was observed for only one of the seven HRQOL domains (social functioning) in favor of standard treatment with radiotherapy alone. Over the following assessments, the HRQOL was comparable between treatment arms. The addition of temozolomide during and following external beam radiotherapy for newly diagnosed GBM significantly improved survival without any (long-term) negative impact on HRQOL. This novel treatment should be considered as clinically relevant progress for the treatment of GBM patients.



L. Berk, W. Seiferheld, T. Rich, W. Hrushesky, D. Blask, M. Gallagher, M. Kudrimoti, R. McGarry, J. Suh, and M. Mehta; Columbus CCOP, Columbus, Ohio; Radiation Therapy Oncology Group, Philadelphia, Pennsylvania; University of Virginia, Charlottesville, Virginia; VA Research and Development, Columbia, South Carolina; Bassett Research Institute, Cooperstown, New York; Scranton Rad. Med. Assoc., Scranton Pennsylvania; University of Kentucky, Lexington, Kentucky; Indiana University, Indianapolis, Indiana; Cleveland Clinic, Cleveland Ohio; University of Wisconsin, Madison, Wisconsin; USA

Previous research showed that the addition of melatonin to radiation therapy for glioblastoma multiforme prolonged 1-year survival (Lissoni et al., Oncology, 1996, 53:43) and given for progression of brain metastases after radiotherapy prolonged survival (Lissoni et al., Cancer, 1994, 73:699). In vivo research suggested that the timing of the melatonin relative to the radiation therapy may change the efficacy of the treatment. Therefore, we performed a Phase II randomized trial, testing the combination of these approaches for patients with brain metastases. RTOG RPA Class II brain metastasis patients (Gaspar et al., 1997, Int. J. Radiat. Oncol. Biol. Phys. 37:745) were eligible for this trial. These patients had Zubrod Performance status of 0–1 and any of the following: ≥65 years of age, extracranial metastases, or uncontrolled primary malignancy. The patients were treated with whole-brain radiotherapy (30 Gy in 10 fractions over two weeks) and 20 mg of melatonin daily, from the beginning of radiation therapy to death or progression. The patients were randomized to receive the melatonin in the morning (8–9 a.m.) or evening (8–9 p.m.), and the radiation was always given between 2 and 6 p.m. The study was designed to compare survival of each arm separately to the historical database of 1070 RPA Class II brain metastases. Between May 2002 and July 2003, 130 patients were accrued, and 124 were analyzable (61 a.m. and 63 p.m.). Of 124 patients, 95 had died at the time of this analysis. The median FU time at analysis of the living patients was 5.9 months. There was no statistically significant difference in survival between either arm and the historical controls. The preliminary data indicate melatonin does not increase survival over historical controls. Survival results at one year are still pending.


Stuart H. Burri,1 Anthony L. Asher,2 H. James Norton,3 and Robert W. Fraser1; 1Radiation Oncology, 2Neurosurgery, 3Statistics, Carolinas Medical Center, Charlotte, North Carolina, USA

Catheter-based brachytherapy has produced encouraging results in the treatment of high-grade gliomas, but a high risk of radiation necrosis (RN). The GliaSite balloon (GS)-based brachytherapy system has been investigated in the setting of recurrent high-grade gliomas with very low risk of RN. GS has also been utilized in the treatment of brain metastases, with preliminary data showing a higher risk of RN. The purpose of this study was to review the rate of RN in patients with longer follow-up and examine risk factors. A retrospective review was conducted of 21 patients with a single GS catheter implanted more than one year ago as initial therapy for high-grade gliomas (5 patients), recurrent high-grade gliomas (5 patients), and brain metastases (11 patients). The median follow-up was 12 months, with a median follow-up of 20 months for patients still alive. The median age was 55 years. Symptomatic RN was defined as RN that required intervention with steroids, biopsy, or surgical resection. If the tissue specimen was noted to have >90% necrotic tissue, it was scored as RN rather than local failure. Actuarial results were obtained from Kaplan-Meier curves generated in SAS with statistical significance determined by the log-rank test. All patients were treated with doses ranging from 50 to 60 Gy dosed to a depth of 0.7 to 1.2 cm. The 12-month survival was 50.4%, with an 18 month survival of 30.0%. Actuarial freedom from tumor progression was 56.6% at 12 months and 45.3% at 18 months. RN as defined by radiographic changes was noted in 5.6% of patients at 6 months, 31.1% at 12 months, and 72.5% at 18 months. Symptomatic RN (SRN) was noted in no patients at 6 months, 9.0% at 12 months, and 63.6% at 18 months. Overall survival, RN, and SRN were all evaluated by diagnosis, balloon size, balloon fill volume, treatment radius, volume treated, treated brain volume, dose, and prescription depth. Of all of the variables reviewed statistical significance was found only for diagnosis group. Patients with brain metastases were the least likely to develop SRN, followed by patients with initial therapy for high-grade gliomas, followed by recurrent high-grade gliomas (P = 0.013). Trends were noted for patients treated with a balloon volume of <8 cm3 for a decreased risk of RN (P = 0.086) as well as a decreased risk of SRN (P = 0.112). Similarly, larger balloon size (P = 0.107) and treatment radius >2.05 cm (P = 0.123) were associated with increased risk of RN. Patients treated to a dose ≥60 Gy had a decreased risk of SRN (P = 0.063). The risk of RN and SRN in patients treated with balloon-based brachytherapy (GS) rises with longer follow-up. The overall risk of SRN was 28.6%. The actuarial risk of symptomatic RN was only 9.0% at 12 months and had risen to 63.6% by 18 months. Only one patient has failed locally beyond 12 months of follow-up. Overall, the GS brachytherapy system allows the delivery of high doses of radiation with a significant risk of long-term symptomatic RN at the described dose levels.


Thomas C. Chen,1 Sara Kim,2 Axel Schonthal,3 Weijun Wang,1 and Colin K. Hill2; Departments of 1Neurosurgery, 2Radiation Oncology, and 3Microbiology, University of Southern California, Los Angeles, California

We have previously demonstrated that upregulation of the cAMP/PKA pathway using cyclic AMP (cAMP) analogues (8-Br-cAMP, 8-Cl-cAMP) and the phosphodiesterase inhibitor rolipram induce a G1 block in human glioma cells, by upregulating the cell cycle inhibitors p21 and p27. Because of the G1 block, we hypothesized that these agents could be used as radiation sensitizers. To test this hypothesis, two human glioma cell lines with different p53 status (A-172:p53 mutant, U-87:p53 wild type), but the same PTEN status (both PTEN mutants), were used. Each cell line was divided into five treatment groups: radiation alone (6 Gy), cAMP analogue alone for 24 h, cAMP analogue plus rolipram for 24 h, cAMP analogue for 24 h plus gamma radiation (6 Gy), and rolipram for 24 h plus gamma radiation (6 Gy). After 24 h of treatment, the cells were examined for survival by using a colony-forming assay (CFA); cell cycle analysis was performed by using flow cytometry immediately after radiation and at 6 h post radiation treatment. A-172 cells pretreated with rolipram (1 mM) for 24 h or with 8-Cl-cAMP, followed by radiation, demonstrated an increase in G1, decrease in S phase, and slight increase in G2 phase. Combination of these two drugs resulted in an accentuation of the cell cycle kinetic changes compared to the individual drugs alone. U-87 cells had a similar effect on treatment on cell cycle analysis. Survival studies were performed using rolipram with increasing doses of radiation (0–10 Gy). CFA demonstrated that U-87 cells were more radiosensitive than A-172 cells. However, synergistic cell killing was demonstrated for the A-172 cells; U-87 cells had an additive cell killing with rolipram and radiation. Preliminary in vivo experiments were performed by injecting U-87 glioma cells in the flank of nude mice (2 mice/treatment group), followed by rolipram alone (1 mM), rolipram and radiation (600 cGy), or no treatment. Over a two-week time period, control mice had an increase in tumor volume (1000–5500 mm3); whereas tumors in mice treated with rolipram alone or with rolipram and radiation were stable in tumor size for up to 2 weeks. At that time, the rolipramalone mice demonstrated an increase in tumor size compared to mice treated with rolipram and radiation. Our results suggest that upregulation of the cAMP/PKA pathway leads to radiation sensitization in human gliomas via a cell-cycle-dependent mechanism.


Wagner G. Dos Santos,1 Adrian Edwards,1 Dong Kim,1 Helen Fillmore,1 Howard L. Elford,2 and William C. Broaddus1; 1Department of Neurosurgery, Medical College of Virginia, Virginia Commonwealth University, Richmond, Virginia; 2Molecules For Health, Inc., Virginia Biotechnology Research Park, Richmond, Virginia; USA

One of the hallmarks of cancer cells is the uncontrolled and high rate of proliferation which requires production of precursors for DNA synthesis. Ribonucleotide reductase (RNR) is a key enzyme for converting ribonucleotides into deoxyribonucleotides, an activity crucial for DNA replication and repair. This enzyme contains two different homodimeric subunits consisting of two protein components, R1 and R2. Recently, a new gene with high similarity to the R2 subunit (p53R2) was identified that is induced by the tumor suppressor protein p53 following treatment that results in DNA damage. Approximately 30% of malignant gliomas exhibit mutant forms of p53, and 80% express elevated levels of p53 protein, suggesting that the majority of gliomas exhibit some degree of functional loss in p53 function. Here we have investigated the pattern of expression of RNR subunits in glioma cells with different p53 status. Also, in order to evaluate the role of RNR in the survival of gliomas we have analyzed the effect of radiation and inhibitors of RNR on cell growth and on the level of expression of the subunits. Cell viability assays demonstrated that the RNR inhibitor trimidox (Tx), a compound belonging to a new class of polyhydroxy-substituted benzohydroxamates, inhibits the growth of human glioma cells: U87 (wt p53), T98 (mutant p53), and LN-Z308 (p53-null), as well as the rat glioma cell line RT2 (wt p53), at concentrations as low as 4 μM. On the other hand, only concentrations higher than 50 μM produced a significant decrease in clonogenic survival. Treatment of cells with Tx following IR was more efficient in decreasing clonogenicity than treatment with drug or IR alone. Tx was found to potentiate the effect of IR on a variety of glioma cell lines. Western blot analysis with specific antibodies directed to the R1, R2, and p53R2 subunits demonstrated that IR decreased the expression of R1 and R2 subunits in a dose-dependent manner in U87 cells. However, no alteration in expression was observed in T98 cells. In contrast, IR increased the expression of the p53R2 subunit in U87 cells. Together, our results suggest that inhibition of RNR could be a useful strategy to hamper DNA repair and therefore increase cell death of tumor cells in response to IR. Further studies are under way to better characterize the mechanisms of the effects of this drug in glioma cell lines, and to assess its potential as a radiosensitizer for the treatment of malignant gliomas.


Cole A. Giller, Brian D. Berger, David A. Pistenmaa, Frederick Sklar, Bradley Weprin, Kenneth Shapiro, Naomi Winick, Arlynn F. Mulne, Janice L. Delp, Joseph P. Gilio, Kenneth P. Gall, Karel A. Dicke, Dale Swift, David Sacco, Keisha Henderson and Daniel C. Bowers; Departments of Neurological Surgery and Radiation Oncology, UT Southwestern Medical Center, Dallas, Texas; Children’s Medical Center, Dallas Texas; Arlington Cancer Center, Arlington, Texas; USA

A robotically guided linear accelerator has recently been developed that provides frameless radiosurgery with high precision. Potential advantages for the pediatric population include the avoidance of the cognitive decline associated with whole-brain radiotherapy, the ability to treat young children with thin skulls unsuitable for frame-based methods, and the avoidance of general anesthesia. We report our experience with this system (the “Cyberknife”) in the treatment of 24 children. Forty-one treatments were given to 24 patients of age 7.8 ± 5.3 years (range, 8 months–16 years). Three had pilocytic astrocytomas, two had anaplastic astrocytomas, three had ependymomas (two anaplastic), four had medulloblastomas, three had atypical teratoid/rhabdoid tumors, three had craniopharyngiomas, and six had other tumor pathologies. Mean target volume was 10.7 ± 20 cm3, mean marginal dose was 18.7 ± 7.8 Gy, and mean follow-up was 16 ± 11.4 months. Twenty-nine (71%) of the treatments were single-shot, and 10 (42%) patients did not require general anesthesia. Local control was achieved in the patients with pilocytic and anaplastic astrocytomas, three of patients with medulloblastomas, and all of the patients with craniopharyngiomas but not for those with ependymomas. Two of the patients with atypical teratoid/rhabdoid tumors are alive 16 to 35 months after initial diagnosis. Cyberknife radiosurgery can be meaningfully used to achieve local control for children with CNS tumors.


Costas G. Hadjipanayis1,2 and Neal A. DeLuca2; Departments of 1Neurological Surgery and 2Molecular Genetics and Biochemistry, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA

The catalytic subunit of DNA-dependent protein kinase (DNA-PKcs) is a key player in the repair of DNA double-strand breaks (DSBs) produced by ionizing radiation (IR). The herpes simplex virus (HSV) immediate-early (IE) protein ICP0 has been shown to induce the degradation of DNA-PKcs. ICP0 degradation of DNA-PKcs may disrupt the repair of DNA DSBs and allow for radiosensitivity enhancement of human GBM. Two replication-defective HSV-1 vectors were used to infect radioresistant T98 and U87-MG human GBM cells. The HSV-1 vector, d106, solely expresses ICP0, while d109 is defective for all IE proteins. After cell infection (MOI of 10 PFU/cell) for 24 h, cells were treated with escalating single doses of IR (0, 5, 10, and 20 Gy). Cell viability and proliferation after IR treatment were assessed with an MTT cell proliferation assay at 0, 2, 4, and 6 days. Cells infected with d106 showed a dose-dependent decrease in cell survival and proliferation after IR treatment compared to d109 and mock infected cells. A replication-defective E1– E3 – E4 - adenovirus, which expresses ICP0 at a reduced level relative to d106, confirmed the effect of ICP0 on U87-MG cell survival after infection and IR treatment. Clonogenic survival assays revealed a lower surviving fraction of U87-MG cells after d106 infection and IR treatment (0.001) compared to cells treated with IR (0.01). Western blot analysis revealed DNA-PKcs degradation between 6 and 24 h after d106 infection in both cell lines. Persistent DNA DSBs, or H2AX foci, were found with indirect immunofluorescence in U87-MG cells infected with d106 and treated with IR. Apoptosis was determined to be the mode of cell death in T98 cells as levels of cleaved caspase-3 were detected in d106-infected T98 cells 72 h after IR treatment. Our results suggest that degradation of DNA-PKcs by ICP0 in human GBM cells inhibits the repair of DNA DSBs after IR treatment, decreasing the survival of these cells by induction of apoptosis.


Joseph M. Hoxworth, Soonmee Cha, Nicholas Butowski, Susan Chang, Jamie Chang, Daniel B. Vigneron, Xiaojuan Li, Mitchel S. Berger, and Sarah J. Nelson; Departments of Radiology and Neurosurgery, University of California San Francisco, San Francisco, California. USA

Identification of factors predictive of early treatment failure could aid in patient counseling and potentially warrant more aggressive measures. Lactate metabolite within tumor has been associated with hypoxia and potentially radioresistance of gliomas. The purpose of our study was to determine whether quantitative measurement of lactate derived from proton magnetic spectroscopy in tumor before radiotherapy correlates with clinical assessment of early treatment failure. Eleven patients without prior treatment or biopsy were diagnosed with gliomas (n = 2, Grade III; n = 9, Grade IV) and were included in the study. In addition to preoperative imaging, the patients were evaluated immediately prior to and following postoperative radiotherapy. The amount of residual enhancing tumor was determined pre- and post-radiotherapy, and patients were grouped as treatment responders (stable or decreased tumor) or treatment failures (increased tumor). The MRSI protocol included 3D J-difference lactate-edited MRSI using PRESS (Point REsolved Spectral Selection) volume selection that incorporated a two-cycle BASING (Band Selective Inversion with Gradient dephasing) pulse. Lactate was quantified only within the solid tissue and not within the resection cavity. Comparison of lactate on the basis of treatment response was undertaken with the Mann-Whitney U test. All patients underwent attempted gross total tumor resection. The intervals (mean ± standard deviation) between preoperative and pre-radiotherapy imaging and between pre- and post-radiotherapy imaging were 31 ± 15 days and 62 ± 12 days, respectively. Five patients were treatment responders (3 stable, 2 decreased residual tumor), and six patients represented treatment failures. Preoperatively, the maximum lactate (median) ranged from 4.73 to 19.37 (10.82) in treatment failures and 6.22 to 51.54 (13.59) in responders (P = 0.273). Prior to radiotherapy, the maximum lactate ranged from 4.55 to 13.01 (6.68) in treatment failures and 1.49 to 5.70 (4.97) in responders (P = 0.045). Lactate-edited MRSI is pertinent to the serial study of high-grade gliomas. Patients with disease progression during treatment demonstrated increased tumor lactate immediately prior to radiation therapy. In contrast, treatment response was independent of preoperative lactate levels. These preliminary results warrant further study in larger populations with inclusion of additional measures of clinical outcome. This study was supported by grant NS45013 and an Accelerate Brain Cancer Cure grant.


Ashley W. Jensen1 and Paul D. Brown2; 1Division of Radiation Oncology, Mayo Clinic College of Medicine, Rochester, Minnesota; 2Division of Radiation Oncology, Mayo Clinic College of Medicine, Rochester, Minnesota, USA

The objective of this study was to determine local control (LC) and complication rates for patients with radiation-induced intracranial tumors who underwent Gamma Knife radiosurgery (GKR). Review of a prospectively maintained database (2714 patients) treated with GKR between 1990 and 2004 identified 16 patients (20 tumors) who met the Cahan criteria for radiation-induced tumors. The median dose of previous external beam radiation therapy (EBRT) was 49.5 Gy (range, 20–80 Gy) and was given at the median patient age of 11.8 years (range 1.3–55.2 years). The median time from EBRT to GKR was 28.4 years (range, 12.4–55.2 years). Thirteen patients had 1 tumor treated with GKR, 2 patients had 2, and one patient had 3. The tumor types were typical meningioma (n = 17), atypical meningioma (n =2), and schwannoma (n = 1). Median patient age at the time of GKR was 47.5 years (range, 27–70 years). The median prescription isodose volume was 5800 mm3 (range, 594–42,600 mm3); the median tumor margin dose was 16 Gy (range, 12–30 Gy). The median follow-up period was 37.4 months (range, 6.7–135.9 months). Kaplan-Meier estimates were used to calculate time-to-event outcomes. The 3- and 5-year LC rates were 100%. The 3- and 5-year survival rates were 92% and 80%, respectively. The cause-specific survival (CSS) rates at 3 and 5 years were 100%. Three patients died: One patient had in-field tumor progression noted at 65.1 months after radio-surgery and later succumbed to the tumor, one patient died of progression of a pre-existing brain malignancy, and the third patient died of unrelated causes. One patient (6%) had an increase in seizure activity that correlated with the development of edema on MRI. GKR is a safe and effective management strategy for radiation-induced intracranial tumors, the vast majority of which are typical meningiomas. The local control, survival, and complication rates in this series are comparable to previous reports of radio-surgery for intracranial meningiomas.


Timothy Korytko, Valdir Colussi, Barry Wessels, Kunjan Pillai, Robert Maciunas, and Douglas Einstein; Case Western Reserve University School of Medicine, Case Comprehensive Cancer Center, Cleveland, Ohio, USA

Despite the widespread use of radiosurgery in the treatment of a variety of intracranial tumors, there is a paucity of data about its toxicity. The Arteriovenous Malformation (AVM) Radiosurgery Study Group recently found that the volume of the AVM and surrounding brain receiving a 12-Gy Gamma Knife radiosurgical dose (12-GyV) was highly predictive of the development of symptomatic radiation necrosis. It is unknown whether this 12-GyV can predict the risk of radiation necrosis for non-AVM tumors treated with Gamma Knife Radiosurgery (GKSRS). The aim of this study was to describe the correlation, if any, between the 12-GyV and the development of both symptomatic and asymptomatic radiation necrosis for patients with a variety of non-AVM tumors treated with GKSRS. We performed a retrospective review of 198 separate common non-AVM lesions (105 brain metastases, 33 gliomas, 29 meningiomas, 14 acoustic schwannomas, 7 pituitary tumors, and 10 others) from 129 patients treated with GKSRS at a single institution from January 2000 to February 2003. Patients were followed with MRI and clinical evaluation at regular intervals depending upon pathology for an average of 10.7 months. (Patients followed for less than 3 months were excluded.) Lesions were categorized into three groups: symptomatic necrosis (S-NEC), asymptomatic necrosis (A-NEC), or no necrosis (N-NEC) based on the interpretations of clinical and MRI data by the attending radiation oncologist and neurosurgeon. When all lesions were analyzed together, the risk of any necrosis as a function of 12-GyV was as follows: 42% (0–5 cm3), 38% (5–10 cm3), 71% (10–15 cm3), and 78% (>15 cm3). The risk of A-NEC was 19% (0–5 cm3), 18% (5–10 cm3), 17% (10–15 cm3), and 22% (>15 cm3). The risk of S-NEC was 23% (0–5 cm3), 20% (5–10 cm3), 54% (10–15 cm3), and 57% (>15 cm3). The ratio of symptomatic to asymptomatic necrosis (S/A-NEC ratio) for 12-GyV < 10 cm3 was 1.2:1, and for 12-GyV >10 cm3 it was 2.9:1. For patients with single lesions (100), the risk of any necrosis as a function of 12-GyV was as follows: 34% (0–5 cm3), 35% (5–10 cm3), 69% (10–15 cm3), and 75% (>15 cm3). The risk of A-NEC was 16% (0–5 cm3), 19% (5–10 cm3), 31% (10–15 cm3), and 17% (>15 cm3). The risk of S-NEC was 18% (0–5 cm3), 15% (5–10 cm3), 38% (10–15 cm3), and 58% (>15 cm3). The S/A-NEC ratio for 12 GyV <10 cm3 was 1.0:1 and for 12 GyV <10 cm3 was 2.0:1. The higher rate of necrosis for lesions with 12 GyV >10 cm3 was predominantly attributable to an increase in S-NEC rates, as the A-NEC rates remained relatively constant with increasing 12-GyV. In conclusion, this study extends the current knowledge of the dose-volume toxicity of Gamma Knife Radiosurgery to the treatment of non-AVM intracranial tumors. For single non-AVM tumors treated with radiosurgery, the 12-GyV highly correlates with the development of radiation necrosis, with the majority of necrosis risk occurring in lesions with 12 GyV >10 cm3. Radio-surgery-induced necrosis is equally symptomatic and asymptomatic for 12 GyV <10 cm3. The vast majority of radiosurgery-induced necrosis for 12 GyV >10 cm3 is symptomatic.


N.N. Laack, K.V. Ballman, P.D. Brown, and B.P. O’Neill; Mayo Clinic and Foundation, Rochester, Minnesota, USA

PCNSL is a disease of older adults that is invariably lethal without treatment. Current treatments are extending survival but are not increasing the number of cured patients. Furthermore, the quality of such survival can be poor. Elderly PCNSL patients are especially susceptible to neurotoxicity. NCCTG 967351 was designed to evaluate the efficacy, toxicity, and survival of WBRT and HDMP in newly diagnosed PCNSL patients 70 years of age and older. HDMP was chosen because it has cytotoxic activity against lymphoma cell lines and may decrease toxicity by direct and indirect modulation of excitatory CNS cytokines. The current study reports toxicity of this regimen; efficacy and survival data is reported separately. Patients received 1 g of methylprednisolone once-daily for 5 days starting 30 days after completing radiotherapy (WBRT 4140 cGy/23 fractions with 900-cGy boost to contrast-enhancing disease). Patients then received 1 g of methylprednisolone every 28 days until progression. Baseline evaluation including physical, neurologic exam, and Mini-Mental Status Exam (MMSE) was performed within 14 days of registration and at day 1 of WBRT. Neurologic, gastrointestinal, and dermatologic toxicity was monitored regularly throughout the protocol and scored with the Common Toxicity Criteria (CTC) version 2.0. Nineteen patients were accrued between 1998 and 2003. Median age of patients was 76 years (range, 70–83). Median ECOG performance status was 1.5. Toxicity data was available in 16 of the 19 patients entered on the study. Thirteen patients had grade 3 or higher toxicity at one or more evaluation. No treatment-related deaths were documented; 3 grade 4 toxicities were observed. Of patients with grade 3 or higher toxicity, 7 patients developed neuromotor toxicity, 2 patients developed infections, 3 patients exhibited neurocortical toxicity, 2 patients exhibited seizures, 2 patients experienced lethargy, and 1 patient experienced headache. Toxicity results were compared to patients 70 years or older (median 73.5) in a previous NCCTG trial (867252) that utilized CHOP/AraC chemotherapy and WBRT. No statistically significant differences in maximal toxicity were observed (n = 10, 5 grade 3 or higher toxicities). Treatment of elderly patients with WBRT and HDMP is tolerated with acceptable toxicity. No statistical difference is noted between patients receiving WBRT and HDMP and historical patients treated with CHOP/AraC chemotherapy and WBRT. A replacement study utilizing HDMP at WBRT initiation is under review.


Eriks Lusis, David Limbrick, Michael Chicoine, Joseph Simpson, Keith Rich, Ralph Dacey Jr., Joshua Dowling, Robert Grubb Jr., Eric Filiput, Robert Dryzmala, and Robert Malyapa; Departments of Neurosurgery and Radiation Oncology, Washington University School of Medicine, St. Louis, Missouri, USA

Surgical resection combined with gamma knife radiosurgery is a treatment option for patients with cerebral metastases from systemic cancers. This might avoid toxicities of whole-brain radiation (WBXRT), but this strategy remains controversial. This study analyzes the effectiveness of such a treatment paradigm. We conducted a retrospective analysis of 16 patients with 1 to 2 (11 with 1, five with 2) brain metastases treated with surgical resection and gamma knife radiosurgery (GKS). Medical records were reviewed to determine overall survival and freedom from progression (FFP) measured from date of surgery. Mean patient age was 57 years. Histopathologically, these tumors were classified as lung (31%), breast (25%), renal cell (19%), esophageal (6%), ovarian (6%), and unknown (13%). Fifty six percent, 38%, and 6% of patients, respectively, were classified in recursive partition analysis (RPA) classes 1–3. Seventy one percent of patients had a gross total resection, 7% had near-total resection, and 21% had STR. The median dose for GKS was 19 Gy (range, 16–24 Gy) at a median volume of 3.5 cm3 (range 184 mm3–16 cm3) at the 50% isodose line. Median survival time was 13.7 months (mean, 16.7; range, 5.3–34.4), with median survival of 14.6, 15.1, and 8.4 months, respectively, for RPA classes 1–3. Median FFP after receiving both treatments was 9.9 months (mean, 11.7; range, 4.2– 28.9). Three patients subsequently went on to whole-brain radiation at a mean of 7.6 months from initial presentation. Eight of 16 patients are in good condition and continue clinical follow-up. Surgical resection combined with gamma knife radiosurgery is a rational treatment option for some patients with brain metastases. This may enable patients to avoid or at least delay exposure to the toxicities of WBRT. Survival rates are equal to or better than those reported by other investigators using up-front whole-brain radiation strategies.


Amit Maity,1 Hui-Kuo G. Shu,1 Jacqueline E. Tan,2 James Ruffer,3 Leslie N. Sutton,4 Zelig Tochner,1 and Robert Lustig,1; 1Department of Radiation Oncology, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania; 2Department of Radiation Oncology, Albany Medical Center, Albany, New York; 3Good Shepherd Radiation Oncology Center, Barrington, Illinois; 4Department of Neurosurgery, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania; USA

There are relatively few reports detailing the outcome of children and adolescents with arteriovenous malformations treated with stereotactic radio-surgery (SRS). We reviewed our experience over the past decade to determine whether the outcomes and toxicity were similar to those reported in adults. Seventeen patients 18 years of age or younger underwent linear accelerator–based stereotactic radiosurgery. The targeted volume was greater than 3 cm3 in 65% of cases (range, 0.7–25 cm3; median volume, 6.9 cm3). The prescribed radiation doses varied from 16 to 18 Gy, with 70% receiving the highest dose. When only angiographic data were used, the obliteration rate was 78% (7/9 patients), but when both MRA/MRI and angiographic data were used, it was 59% (10/17 patients). Four patients developed late effects potentially attributable to the radiation between 2 months and 3 years following SRS. One of these was transient and disappeared completely within a few days, but in the other three patients, the neurologic sequelae were permanent. Two of the four complications appeared to be due to radiation necrosis/edema, whereas the remaining two may have been due to vasculopathy. All four patients with complications had treatment volumes greater than 5 cm3 (5.4, 6.9, 11.1, 16.4 cm3), all had a prescribed dose of 18 Gy and all had initially presented with an AVM hemorrhage. Linear accelerator–based SRS is effective in obliterating most AVMs in children; however, the potential for late effects exists, especially for those patients with larger target volumes.


Ashwatha Narayana, Kamil Yenice, Jenghwa Chang, Priti Shah, Linda Hong, Margie Hunt, and Philip Gutin; Department of Radiation Oncology, Medical Physics and Neurosurgery, Memorial Sloan Kettering Cancer Center, New York, New York, USA

Intensity-modulated radiotherapy (IMRT) is an emerging treatment for brain tumors, but there is very little data on doses and outcomes from IMRT treatment for brain tumors as compared to conventional 3-dimensional (3D) treatment planning. This study compares the potential benefits and disadvantages of IMRT to 3D planning for brain tumors. Twenty brain tumor patients treated with dynamic multileaf collimator IMRT with inverse planning were evaluated for dose coverage to the planned target volume (PTV), the critical structures, and the normal brain. Ten of these patients with larger volumes (range, 81–411 cm3) were treated with 3–5 non-coplanar beams to a plan that covered at least 95% of the target volume (D95) with the prescription isodose line. The remaining 10 patients with smaller volumes (range, 14.5–27.4 cm3) were treated with fractionated stereotactic radio-therapy using IMRT with a similar plan using an average of 13 beams. A separate 3D plan was developed with similar beam arrangement, and comparison was made using the conformity (the ratio of the treated volume to tumor volume) and uniformity (the ratio of the maximum dose to the minimum dose received by the target volume) indices, as well as dose-volume histograms for both target and normal structures. Regardless of tumor location or the volume of the lesion, IMRT did not improve the target coverage compared to 3D treatment. The mean dose, D95, and the maximum doses were similar for both IMRT and 3D plans for the larger tumors. For the smaller tumors, IMRT plans consistently delivered more uniform dose to the target volume than 3D, with a uniformity index of 1.03 and 1.11, respectively. However, the use of IMRT resulted in a decreased dose to the critical structures like optic chiasm and brain stem by 4.5% and 8%, respectively. The mean normal brain radiation dose and the volume of brain receiving 18 Gy and 24 Gy were lower with IMRT, with a benefit of 7%, 10%, and 14% respectively. The biggest overall benefit of IMRT was noted in smaller volume lesions and in irregular shaped lesions located close to the critical structures. It is unlikely that IMRT will increase local control in larger brain tumors like gliomas without dose escalation since the target coverage is unchanged. However, it may result in decreased late toxicities by improving the uniformity in target coverage in smaller tumors like acoustic neuroma or by decreasing the dose to critical structures and the normal brain tissue in larger tumors.


B.P. O’Neill, N.N. Laack, K.V. Ballman, and P.D. Brown; Mayo Clinic and Foundation, Rochester, Minnesota, USA

PCNSL targets especially vulnerable populations and is invariably lethal without treatment. Current treatments are extending survival but are not increasing the number of cured patients. Furthermore, the quality of such survival is poor. Elderly PCNSL patients are even more susceptible to neurotoxicity. NCCTG 967351 was designed to evaluate the efficacy, toxicity, and survival of WBRT and HDMP in newly diagnosed PCNSL patients 70 years of age and older. HDMP was chosen because it has cytotoxic activity against lymphoma cell lines, may decrease toxicity by direct and indirect modulation of excitatory CNS cytokines, and effectively “closes” the blood-brain barrier for 30 days. All patients in this study had a negative HIV status. WBRT was administered as 4140 cGy at 180 cGy per fraction with 900-cGy boost to contrast-enhancing disease, and 1 g HDMP was given once daily for 5 days starting 30 days after completing radiotherapy and then 1 g HDMP every 28 days until progression. The primary end point was survival at six months. Interim analysis was performed after the twelfth eligible patient was followed for six months. Survival results were compared to age-matched patients on NCCTG trial 86-72-52. Nineteen patients were accrued, with a median age of 76 years (range, 70–83) and median ECOG performance status of 1.5. Interim analysis revealed 6-month survival of 33%. (By the protocol-decision rule, the regimen was considered inactive and the trial closed.) There were no differences in OS or PFS, or toxicity between age-matched patients on NCCTG 967351 and 867252. The 10 patients who received HDMP appeared to have longer OS (14.3 months vs. 6.5 months, P = 0.54) and PFS (14.3 months vs. 3.1 months, P = 0.032) compared to the 10 NCCTG 867252 patients alive 30 days after WBRT. We conclude that the subset of patients on NCCTG 967351 who received HDMP had prolongation of OS and PFS, but the numbers of patients are too small for statistical conclusions. However, given that the comparison cohort from NCCTG 867252 had received WBRT and chemotherapy, the HDMP regimen deserves further study. A replacement study utilizing HDMP at WBRT initiation is under review.


Ania G. Pollack, James P. Chandler, Robert M. Levy, MaryAnne H. Marymont, John A. Kalapurakal, Sathiaseelan Vythialingam, Alan Kepka, and Amy Stevenson; Departments of Neurosurgery and Radiation Oncology, Northwestern Memorial Hospital, Chicago, Illinois, USA

The value of stereotactic radiosurgery for glomus jagulare tumors of the temporal bone remains controversial. The literature reports provide results that vary from excellent to disappointing. We evaluated toxicity and long-term outcome of symptomatic patients with glomus jagulare tumors of the temporal bone treated with stereotactic radiosurgery at Northwestern Memorial Hospital. Eight consecutive patients with glomus jagulare tumors of the temporal bone were treated with Gamma Knife, Model B, between March 1999 and June 2003 at our institution. Two patients were African-American and six Caucasian, with mean age of 60.9 (range, 25–77 years). Four patients were treated at the initial presentation and four at the recurrent or residual disease following previous surgery. All patients were symptomatic at the time of treatment with gamma knife radiosurgery, with a mean Karnofsky score of 84.3 (range, 70–90). Six tumors involved the left and two the right temporal bones. Matrix volume receiving 50% isodose ranged from 2.3 to 8.1 cm3 (mean 5.0 cm3). The treatment dose was specified as 50% iso-dose shell in all cases and varied from 14 Gy (6 cases) to 15 Gy (2 cases) (mean dose, 14.25 Gy). Treatment planning involved 2 to 10 isocenters (mean, 4.75 isocenters per treatment.) All patients were followed prospectively, with the mean clinical and radiological follow up of 23.1 months and 22.4 months, respectively (range, 10–55 months.) Tumor control was achieved in 100%. Four tumors decreased in size, and four remained stable. One of the tumors was treated only partially. The untreated portion of the tumor continued to grow while the treated part decreased in size. Symptom control was documented in 87.5%, with only one patient developing worsening headache, most probably due to growing portion of his untreated tumor. No neurotoxicity related to stereotactic radiosurgery was reported. We conclude that Gamma Knife radiosurgery is an effective and relatively risk-free option of treatment for glomus jagulare tumors of the temporal bone.


Ania G. Pollack, James P. Chandler, Robert M. Levy, MaryAnne H. Marymont, John A. Kalapurakal, Sathiaseelan Vythialingam, Alan Kepka, and Amy Stevenson; Departments of Neurosurgery and Radiation Oncology, Northwestern Memorial Hospital, Chicago, Illinois, USA

In recent years, stereotactic radiosurgery has become an accepted modality for treatment of cerebral metastases. However, there are relatively few reports dedicated solely to the outcome of gamma knife radiosurgery in treating historically radioresistant tumors such as melanoma and renal cell carcinoma (RCC). The purpose of our study was to evaluate the efficacy of gamma knife (GK) in management of brain metastases from melanoma and RCC. Thirty-one patients with brain metastases from melanoma and RCC were treated with Leksell Gamma Knife, Model B, at our institution between December 1998 and March 2003 (21 melanomas, 10 RCCs). They presented with a total of 139 lesions, 109 melanomas, 30 RCCs, with the mean 5.2 and 3.0 lesions per patient, respectively. Fifteen patients (48.34%) had solitary lesions, and 16 patients (51.61%) had multiple lesions. Seventeen were female and 14 were male, with a mean age of 61.2 (range, 21–82 years). Pathology was confirmed by tissue analysis in 85.2% (data unobtainable in 4 cases). A total of 37 treatment sessions were performed, with four patients receiving two treatments and one patient three treatments. Planning consisted of a total of 260 isocenters (mean, 6.5 isocenters per treatment; range, 1–20). Eighty-six matrices were used (mean 2.5 matrices per treatment), with a mean dose volume of 3.79 cm3 (range, 0.09–31.5 cm3). The mean treatment dose per matrix while treating melanoma was 16.8 Gy (median 20 Gy), and that of RCC was 14.83 Gy (median 12 Gy), both ranging from 10 to 20 Gy. Systemic disease was documented in 92.5%, with two patients having brain metastases as the only manifestation of their metastatic disease. We followed all patients until death (66.7% melanoma, 100% RCC) or the most recent follow-up. The mean survival from GK to death was 6.24 months (range, 1–14 months) for melanoma and 8.75 months (range, 0.5–33 months) for RCC. The mean survival from diagnosis to death was 75.9 months (range, 1–288 months) and 36 months (range, 8–117 months), respectively. Tumor control was obtained in 65.9% of cases (46.15% for melanoma, 85.7% for RCC). The lesions increased in size in 34.6% of the melanomas and 14.3% of the RCCs in a mean time of 2.8 months. In addition, 34% of the patients developed distant metastasis in a mean time of 7.15 months (2.9 months for melanoma, 12 months for RCC). Nine patients required subsequent WBRT. The patients were followed up clinically and radiologically for a mean of 12.75 months (range, 2–47 months) for melanoma and 7.9 months (range 0.5–47 months) for RCC. Complications included one posterior fossa hemorrhage requiring surgery and one headache treated with steroids. We conclude that gamma knife radiosurgery provides reasonable control of cerebral metastases for melanoma and renal cell carcinoma.


Ania G. Pollack, James P. Chandler, MaryAnne H. Marymont, and John A. Kalapurakal; Departments of Neurosurgery and Radiation Oncology, Northwestern Memorial Hospital, Chicago, Illinois, USA

In this report we describe an acute facial and acoustic neuropathy following gamma knife radiosurgery for a vestibular schwannoma. A 39-year-old female presenting with tinnitus underwent gamma knife radiosurgery for a small right-sided intracanalicular vestibular schwannoma. She received a maximum dose of 26 Gy and marginal dose of 13 Gy to the 50% isodose line. She presented 36 h following treatment with the complaint of nausea, vomiting, vertigo, diminished hearing, and a House-Brackmann 3 facial palsy. She was started on intravenous glucocorticosteroids, and over the course of the subsequent two weeks, her facial function returned to House-Grade 1. However, her hearing loss persisted. An MRI done at the time of the initial deterioration demonstrated a significant decrease in tumor enhancement but no changes in tumor size or peritumoral edema. The patient then went on to develop severe hemifacial spasms, which persisted for a period of three weeks and then subsequently developed House-Grade 5 facial palsy. Over the subsequent three months the patient was treated with steroids, and in time her facial function and hearing returned to baseline. Subsequent MRI imaging demonstrated transient enlargement of her tumor by 3 mm and then subsequently it returned to its baseline size. At 10 months post treatment, the patient remains radiologically and neurologically stable. This is the first detailed report of acute facial and acoustic neurotoxicity after gamma knife radiosurgery for vestibular schwannoma that improved with time. In addition, MRI findings are suggestive of early neurotoxic changes. A review of possible risk factors and causative mechanisms will be provided.


H. Robins, M. Won, W. Seiferheld, C. Schultz, A. Choucair, D. Brachman, W. Demas, and M.P. Mehta, University of Wisconsin, Madison, Wisconsin, USA

Preclinical studies support the concept that inhibition of protein kinase C (PKC) by TAM should provide both antineoplastic effects and radiosensitization. High-dose TAM (80 mg/m2 PO daily in divided doses) was given with and after conventional XRT to inhibit PKC-mediated signaling, which is known to be enhanced in GBM. Patients (N = 77) were accrued between 12/00 and 12/01; 75 patients were analyzable. Pretreatment characteristics included the following: 52% were <60 years of age, 39% had a Zubrod score of 0, 70% had minor or no neurological symptoms, and 65% were RTOG-recursive partition analysis (RPA) class III and IV. A sample of 46 patients was reviewed for TAM delivery, and 78% achieved acceptable dosing. Notable toxicity included late radiation grade (G) G3 (n = 2) and thromboembolic disease G2 (n = 2), G3 (n = 8), G4 (n = 3), and G5 (n = 1), for an incidence of 18.7% (which is lower than expected, based on the literature for DVT in GBM patients not receiving TAM). Median survival time (MST) was 9.7 months. This result was compared (with three different statistical methodologies) to the historical GBM control database of 1443 drug/XRT-treated patients RPA class III, IV, and V. After controlling for RPA class IV, the MST was 11.3 months, which compares to the historical RPA control of 11.2 months (P = 0.38). The results obtained do not exhibit a substantial advance over previous studies with various XRT/drug doublets including BCNU. As TAM does not have significant overlapping toxicities with most other drugs, its testing in a combined modality approach with other medications may be justified in future clinical trials. Historically, the incidence of thromboembolic events in GBM patients is ≈30%. The lower than expected incidence seen here has also been observed in other high-dose TAM GBM studies. We speculate that TAM inhibited the PKC-mediated phosphorylation of coagulation factors.


L.R. Rogers,1 J.R. Rock,1 A. Sills,2 M. Vogelbaum,3 M. Ewend,4 and E. Shaw5; 1Henry Ford Hospital, Detroit, Michigan; 2Semmes-Murphey Clinic, Memphis, Tennessee; 3Cleveland Clinic, Cleveland, Ohio; 4University of North Carolina at Chapel Hill, Chapel Hill, North Carolina; 5Wake Forest University, Winston-Salem, North Carolina; USA

Surgical resection with WBRT is superior to WBRT alone for a single brain metastasis. There is controversy regarding the overall benefits and risks of WBRT after resection. We designed a Phase 2 trial of resection and Glia-Site brachytherapy without WBRT for patients with a single supratentorial brain metastasis. The primary objective is to determine the local control rate at one year. Secondary objectives are survival, distant recurrence, quality of life, and toxicity. We designed a multicenter, prospective, single-arm study of surgical resection of a single brain metastasis followed by GliaSite balloon implantation and Iotrex brachytherapy. Eligibility criteria includes a supratentorial location, histology other than SCLC or lymphoma, KPS ≥70, age >18, expected survival ≥3 months. Planned brachytherapy dose was 60 Gy at 1 cm, given within 21 days of implant. Assessments of neurological exam, KPS, MMSE, Fact-BR, and MRI were obtained at baseline, 1 and 3 months, and q 3 months afterward to 12 months. Seventy-one patients were entered. Fifty-four treated patients comprise the study group, the last patient having been entered April 2003. Age ranged from 29 to 86 years, median 61. There were 33 men, 21 women. The most common histology (29) was NSCLC. Tumors were most commonly frontal or parietal. Maximal cross-sectional tumor diameter ranged from 1 to 5.9 cm, median 2.5. Implanted balloon sizes were 2 cm in 61%, 3 cm in 32%, and 4 cm in 7%. More than one implant was attempted in three patients. A balloon malfunction occurred in only one patient. There were no complications resulting from the balloon implantation. Patients were hospitalized 4 to 7 days for brachytherapy. The median dwell time was 4.75 days (3.86–6.77). The median final dose delivered was 60 Gy (55.6–63.9). The local control at one year is 81.5%. No decline in KPS, MMSE, or Fact-BR was detected at one year. Serious device-related adverse events occurred in 11 patients. The balloon was explanted in all but five patients. GliaSite implantation and brachytherapy were well tolerated in this trial of patients with a single brain metastasis. The local control and quality of life are encouraging. Final results of survival, toxicity, and distant recurrence will be presented.


Ratna Sajja,1 Gene H. Barnett,2 Shih-Yuan Lee,1 Glen H.J. Stevens,2 Joung Lee,2 and John H. Suh2; 1Department of Radiation Oncology, Cleveland Clinic Foundation, Cleveland, Ohio; 2Brain Tumor Institute, Cleveland, Ohio; USA

The purpose of this study was to evaluate tumor control, complications, and outcome from intensity-modulated radiation therapy (IMRT) for intracranial meningiomas. Between July 1997 and November 2003, 61 patients with intracranial meningiomas were treated at our institution with the NOMOS Peacock system utilizing the Multileaf Intensity Modulating Collimator (MIMiC). Thirty-five patients with 37 lesions were identified, with a minimum of 6 months of radiologic follow-up for this retrospective review. The median age of the patients was 65 years (range 24–89 years), with a median Karnofsky performance score (KPS) of 90 (range, 50–100) prior to treatment with IMRT. The median MRI/CT follow-up for the 37 treated lesions was 19.1 months (range, 6.4–62.4 months). Twenty meningiomas (54%) were previously treated with surgery prior to IMRT, and 17 meningiomas (46%) were treated with IMRT primarily after diagnosis was confirmed by MRI/CT. The median time from previous surgery to treatment with IMRT was 18.1 months. Thirty-five lesions were of benign histology, and only two lesions were atypical meningiomas. The median tumor dose was 50.4 Gy (range, 27–57.6 Gy) prescribed to the 87% isodose line (median) providing a median target coverage of 95%. The median maximum dose was 56.67 Gy. The median tumor volume was 13.2 cm3. The median dose heterogeneity within the tumor volume (maximum dose/peripheral dose) was 1.14. The median dose conformity of the treatment relative to the target (prescription isodose volume/tumor volume) was 1.95. Local control was found to be excellent at 97% three years after treatment with IMRT. Only three patients exhibited local failure after treatment. Although local control was slightly better in the up-front-IMRT lesions as compared to the lesions treated with prior surgery (100% vs. 95%), this difference was not statistically significant. Analysis of various patient and treatment characteristics revealed no significant differences between the up-front-IMRT lesions and the lesions previously treated with surgery. On univariate analysis, the IMRT prescription dose and maximum dose were found to be predictors for local control (P = 0.05). Time from prior surgery to treatment with IMRT was not a significant independent predictor for local control. On multivariate analysis, these factors did not remain significant for influencing local control. Of the three patients who displayed radiologic progression of their tumor, two patients received an inadequate IMRT prescription dose (28 Gy/7 fx and 28 Gy/8 fx) due to medical comorbidities, and the third patient had a meningioma of atypical histology. There were no long-term complications from IMRT documented among the 35 patients. KPS among the treated patients at the time of last follow-up visit was improved in 19% and remained stable in 57% of the patients when compared to their pretreatment KPS. The median time from IMRT treatment to last follow-up KPS was 19.1 months. We conclude that intensity-modulated radiation therapy is a safe and effective treatment for intracranial meningiomas. A greater number of patients with longer follow-up after treatment may be needed to determine treatment variables predicting for local control.


J.H. Suh, A.P. Boyd, B. Stea, A. Nabid, J.O. Hackman, J. May, E. Shaw, and P.J. Cagnoni; The Cleveland Clinic Foundation, Cleveland, Ohio; Wake Forest University School of Medicine, Winston-Salem, North Carolina; USA

The purpose of this study was to assess the validity of using response at 3 months in the brain after whole-brain radiation therapy (WBRT) to predict subsequent 6-month survival in brain metastases patients. A randomized open-label phase 3 study was designed to compare the efficacy and safety of RSR13 (efaproxiral) combined with WBRT and supplemental O2 versus WBRT and supplemental O2 alone in patients with brain metastases. Patients with KPS ≥70, at least one measurable lesion, and no prior WBRT or stereotactic radiotherapy were eligible. In addition to survival, response rate at 1-month post-WBRT, 3-months post-WBRT, and every 3 months thereafter using consistent MRI or CT studies was evaluated until progression or death. Response evaluations were made by independent central review relative to the baseline scan on up to 3 indicator lesions. Progressive disease was defined as an increase in the bi-dimensional product of 25% in at least one target lesion, or the progression of any treated lesion not enumerated as a target lesion, or progression of any non-assessable but treated lesion. The appearance of new lesions (reseeding) was not considered a progression. Imaging studies that did not meet the definition of progressive disease were evaluated as complete response (CR), partial response (PR), or stable disease (SD). Complete disappearance of all target lesions was required for CR. A PR was defined as at least a 50% reduction in the bi-dimensional products of all indicator lesions. All other evaluations were SD. For the purposes of this analysis, CRs and PRs were combined into one group (“Responders”). All other responses were defined as “Non-responders”. Survival time was measured from date of randomization until date of death or the last date known alive. We analyzed data with a median potential follow-up of 26.9 months to assess the utility of using the 3-month post WBRT scan to predict survival status. Between 2/2000 and 7/2002, 538 patients were randomized in 82 sites in 12 countries to receive RSR13 plus WBRT and supplemental O2 (n = 271) or WBRT plus supplemental O2 without a placebo (n = 267). The majority of patients (88.7%) were randomized within 4 days of WBRT day 1. Twenty-three patients were in violation of eligibility criteria and were therefore excluded from analysis. Of the 515 eligible patients, 510 patients had a baseline scan, 399 patients had a 1-month scan, and 236 patients had a 3-month scan. Of these 236 patients, 124 patients alive and with a scan at the 3-month follow-up visit (53%) were responders. Responders at 3 months had a higher likelihood of living an additional 6 months compared to non-responders (odds ratio 1.82 [95% CI, 1.08–3.07], P = 0.023). Median survival time beyond this scan was 8.0 months for responders versus 5.2 months for non-responders. We conclude that response in the brain at 3 months post-WBRT is a clinically and statistically valid method to predict subsequent 6-month survival. These results should be taken into account when designing future studies in patients with brain metastases.


H.K. Tsai,1 L.C. Goumnerova,2 S.L. Pomeroy,3 R.M. Scott,2 T.Y. Poussaint,4 R. Xu,5 M.W. Kieran,6 N.J. Tarbell,7 and K.J. Marcus1; Departments of 1Radiation Oncology, 2Neurosurgery, 3Neurology, and 4Radiology, Children’s Hospital Boston; Departments of 5Biostatistical Science and 6Pediatric Oncology, Dana-Farber Cancer Institute; 7Department of Radiation Oncology, Massachusetts General Hospital, Boston, Massachusetts; USA

The records of all patients ≤21 years old with craniopharyngioma who underwent consultation for surgery and/or radiation therapy (RT) from 1960 through 2003 at Children’s Hospital Boston were reviewed. Patients who had surgical resection or RT at other institutions and patients who had bleomycin or radioactive yttrium instillations as part of their initial management were excluded from analysis. Radiotherapy consisted of fractionated, megavoltage radiation delivered to a median dose of 54 Gy. A total of 91 patients were included in the final analysis. The median age was 8.5 years, and the median follow-up was 7.3 years. There were 46 males and 45 females. The 10-year overall survival (OS) rate for all patients was 93%. Thirty-seven patients were managed initially with surgery alone, either with gross total resection (n = 30) or subtotal resection (n = 7). The 10-year local control (LC) and OS rates for this subgroup of patients were 43% and 97%, respectively. Thirty-six patients underwent subtotal resection (n = 23) or cyst aspiration (n = 13), followed by adjuvant RT at a median of 15 days after surgery. The 10-year LC and OS rates were 75% and 83%, respectively. Eighteen patients received RT after shunt placement (n = 8), biopsy only (n = 2), or no surgery (n = 8). The 10-year LC and OS rates for this subgroup of patients were both 100%. There was no significant difference in the OS rates among the three subgroups of patients (P = 0.26). However, LC was significantly worse in the group of patients treated initially with surgery alone (P = 0.0004). Long-term morbidities included panhypopituitarism (63%), obesity (49%), and visual deficits (44%). The rates of panhypopituitarism were 78% for patients treated initially with surgery alone, 67% for those treated with surgery and adjuvant RT, and 22% for those treated with RT alone (P = 0.0002). The rates of visual deficits were 57%, 44%, and 17%, respectively (P = 0.02). Despite excellent overall survival, craniopharyngioma remains a tumor associated with significant long-term morbidity due to the tumor and treatment. More than half of the children who undergo surgery alone for initial management of craniopharyngioma will experience a local tumor relapse; radio-therapy appears to be beneficial in controlling disease in patients who have undergone subtotal or minimal tumor resections. These results must be interpreted with recognition of the intrinsic selection biases associated with a retrospective study.


M.J. van den Bent, D. Afra, O. de Witte, M. Ben Hassel, S. Schraub, J.Y. Delattre, P.O. Malmstrom, L. Collette, M. Piérart, R. Mirimanoff, and A.B.D.F. Karim; EORTC Radiotherapy and Brain Tumor Groups, European Organization for Research and Treatment of Cancer, Brussels, Belgium; UK Medical Research Council, London, UK

Trial EORTC 22845/MRC BR4 is a prospective randomized trial comparing 54 Gy RT to no RT until the time of progression in low-grade glioma. In 1998, an interim analysis showed an increase in progression-free survival (PFS) but no difference in overall survival (OS) after early RT (Karim et al., 2002, Int. J. Radiat. Oncol. Biol. Phys. 52:316]. We now present the long-term results of this study. To be eligible, patients had to have a histologically proven low-grade glioma, 16 < age < 65 years, and WHO PS 0–2. After stratification for institution, histology, and extent of resection, patients were randomized to either early RT of 54 Gy in fractions of 1,8 Gy delivered to involved fields or to the control arm that received no RT. Further treatment at progression was left to the discretion of the physician, with a strong recommendation to give RT to patients in the control arm. The primary end points were PFS and OS. Three hundred eleven patients entered the trial between March 1986 and September 1997. Over 80% of the patients had a WHO PS of 0 or 1, and 35% of patients had an oligodendroglioma or a mixed oligoastrocytoma. Known prognostic factors were well balanced between the two treatment arms. At the time of this intent-to-treat analysis, the median follow-up is 7.75 years, 156 patients (50%) have died, and 217 (70%) have progressed. PFS was significantly longer in the early RT arm (log-rank P < 0.0001; Table 1). In the control arm, 64% of the patients received RT at the time of progression, versus 4% in the early RT arm. Survival after progression was longer in the control arm (median survival after progression in the early RT arm, 1.0 year [95% CI, 0.8–1.5] versus 3.4 years [1.8–3.9] in the control arm). Consequently, OS survival did not differ between both treatment groups (P = 0.873). Early RT in low-grade glioma prolongs progression free survival without affecting OS. We conclude that the absence of an effect on OS seems related to effective salvage RT in the control arm. This trial establishes the clear efficacy of RT in low-grade glioma, but the best timing of RT remains to be determined.

Table 1
Survival data


Genevieve N. Wu, Jeffry R. Alger, and Judith M. Ford; Biomedical Physics Interdepartmental Graduate Program and Department of Radiation Oncology, University of California at Los Angeles, Los Angeles, California, USA

The purpose of this study was to use MRI as a surrogate marker to investigate whether a loading plus maintenance dosing regimen for Motexafin gadolinium, used over 6.0 to 6.5 weeks during standard radiation therapy for glioblastoma multiforme, maintained levels of drug in tumor. The study group consisted of 33 patients in a phase I clinical trial in which MGd was administered during standard radiation therapy. MGd was given for 5 or 10 doses on a daily (Monday–Friday) schedule as a loading regimen, followed by 3 times per week dosing as a maintenance schedule. Each patient received 10 to 22 total MGd doses. Since MGd is MRI detectable, T1-weighted MRI imaging was undertaken 2 h after the first, fifth, and last doses of drug. Depending on cohorts, certain patients also had MR scans prior to the first dose, or on days when no MGd was given, or a week after the last dose. For patients with a 10-day loading regimen, the scans were done on both day 5 and day 10. All scans included standardization tubes of known concentrations of MGd in brain tissue equivalent agar. Twenty-one patients had images with adequate residual tumor for measurement. MRI signals from tumor were normalized to the signal from tube 1 for each individual scan to correct for global between-study variation in the MRI intensity. Taking maintenance of MRI signal as a surrogate for drug retention, each patient’s data was then expressed as ratios to their load end (day 5) signal intensity. The analysis of results indicates (1) with both 5-day and 10-day loading, the load end signal intensity is consistently higher than pre-drug, (2) the 10-day loading regimen (4 patients) gave higher load end signal intensity than at 5 days loading, with no detectable signal loss on the subsequent non-dose day, and (3) 12 patients in the 5-day loading group had MRI 7 days after their last MGd dose without detectable signal loss in the tumor, suggesting drug is retained in tumor with a half life of at least 7 days. Using MRI signal as a surrogate marker for MGd in tissue and a dosing regimen of 5 to 10 days loading daily Monday–Friday and 3 times per week thereafter shows increase in signal during loading and maintenance without significant fall off on non-dosage days. Although numbers are small, the 10 days loading regimen appeared to show superior drug loading and retention compared with the 5-day loading regimen.



Ashley Aiken, Nicholas Butowski, Jamie Chang, Susan Chang, David Larson, Mitchel Berger, and Soonmee Cha; Departments of Radiology and Neurosurgery, University of California San Francisco, San Francisco, California, USA

Serial MR imaging is commonly used in monitoring patients with glioblastoma multiforme (GBM) following standard external beam radiation therapy (EBRT). Increasing numbers of GBM patients are now being treated with higher dose, local irradiation such as brachytherapy (Brachy). The purpose of this study was to determine imaging patterns of new contrast enhancement (CE) in patients treated with EBRT alone versus EBRT combined with Brachy. Serial MRI scans at two-month intervals over the first year after gross total resection (GTR) and radiation therapy (RT) were available to review for 10 patients treated with EBRT and 10 with both EBRT and Brachy. Imaging features were analyzed for location and characteristics of CE (linear, nodular, feathery, or solid). Radiologic recurrence was correlated with histopathologic data in three of the EBRT and seven of the EBRT/Brachy patients. Half of the EBRT patients demonstrated focal nodular CE along the resection cavity by four months. This entire half progressed to recurrence, confirmed by histopathology in three of the five patients. The other half of the EBRT patients who did not progress developed either transient linear CE along the resection cavity or no CE. Two of 10 Brachy patients with early focal nodular CE along the resection cavity progressed to pathology-proven recurrence. Eight of 10 Brachy patients showed linear CE along the resection cavity at three months after GTR and RT. On subsequent imaging, three of the eight patients developed focal nodular CE and progressed to recurrence. Five of these eight developed more diffuse, feathery CE around the resection cavity. While this was initially interpreted as radiographic recurrence, four had only radiation necrosis by histopathology. Three Brachy patients eventually developed new CE remote from the original resection cavity. Our preliminary data suggest that serial MRI features differ between GBM patients treated with EBRT versus EBRT/Brachy. In both groups, nodular enhancement strongly suggested recurrence. Extensive feathery contrast enhancement seen only in Brachy patients, however, was more indicative of radiation necrosis rather than recurrence. Remote satellite contrast enhancement was seen only in Brachy patients. This study was supported by grant NS45013 and an Accelerate Brain Cancer Cure grant.


D.T. Blumenthal, D.W. Litzau, K.L. Salzman, A.G. Osborn, and K. Koeller; University of Utah, Huntsman Cancer Institute and Armed Forces Institute of Pathology, Washington, District of Columbia, USA

The World Health Organization (WHO) classification of brain tumors recognizes two types of oligodendroglial tumors: oligodendroglioma (WHO grade II) and anaplastic oligodendroglioma (AO) (WHO grade III). These tumors differ genetically and in treatment response from diffuse astrocytomas. The imaging findings of oligodendroglioma are well characterized. However, only a few scattered case reports and small series have reported imaging findings of AOs. We examined the imaging features and clinical presentation in patients with documented AOs in an attempt to characterize patterns that might define accurate preoperative diagnosis. Data for 36 patients with pathologically proven AOs from two institutions were retrospectively reviewed. MRI studies were evaluated for tumor location, margins, signal characteristics, and hemorrhagic components, as well as pattern, intensity, and location of contrast enhancement. CT studies were evaluated for calcification, hemorrhage, bone changes, and tumor density. Clinical information reviewed included age at presentation, gender, presenting symptoms, and survival data. Histopathologic correlation was reviewed in all cases. There were 21 female and 15 male patients. The age at presentation ranged from 8 years to 73 years, with mean age at initial diagnosis of 38.5 years. The most common presenting features were headache (52%), focal neurologic deficit (38%), and seizure (35%). AOs are predominantly infiltrating tumors (32/36) and show hypointense T1 and hyperintense T2 signal in all MRI cases. Of the enhancing tumors (30/36), the most common contrast enhancement pattern was punctate/patchy (50%). Enhancement intensity was typically minimal (57%) and central (41%). Six cases (17%) showed no enhancement. The most common tumor location was the frontal lobe (64%), with unilateral frontal lobe involvement seen in 47%. CT revealed calcification in 10 tumors and calvarial erosion in 3 patients. Hemorrhage was seen in 7 tumors. The clinical and imaging spectrum of AOs is much broader than expected. Although a characteristic imaging appearance is seen in the majority of cases, 17% showed no enhancement, similar to low-grade gliomas. Sixteen percent had a necrotic-appearing center with intense rim enhancement, indistinguishable from classic primary glioblastoma multiforme. We conclude that while there are no specific clinical or imaging criteria that identify AOs preoperatively, an infiltrating mass in the frontal lobe that demonstrates minimal patchy/punctate central enhancement strongly suggests the diagnosis.


Nicholas A. Bock,1 Gelareh Zadeh,2 Lori M. Davidson,1 Baoping Qian,2 John G. Sled,1 Abhijit Guha,2 and R. Mark Henkelman1; 1Mouse Imaging Center (MICe) and 2Brain Tumor Research Center, Hospital for Sick Children, Toronto, Ontario, Canada

One limitation of intracranial mouse models of disease is the inability to monitor and evaluate the intracranial compartment non-invasively over time. In this study, we have established protocols for multiple-mouse MRI to follow the growth and behavior of intracranial xenografts of gliomas longitudinally. We successfully obtained weekly 3D images on 15 mice bearing human U87 xenograft gliomas for a total of 5 weeks on a multiple-mouse MRI. Four different imaging sequences (T1, T1+gadolinium, T2, and diffusion-weighted images) were analyzed for best definition of tumor margin, tumor necrosis, and tumor cell invasion. Additionally, sequence characteristics were compared with MRI characteristics of human astrocytomas. The overall survival of tumor-bearing mice included in the MRI screen was (33 ± 1 days) the same as the survival mice not included in the MRI screen (36 ± 5 days). These results confirm that MRI for follow-up of tumor growth of intracranial tumors is feasible and not detrimental to the outcome of the mice. The lesions detected on the MRI images were confirmed to be tumors by both Evans blue and histology. T1 plus gadolinium best outlined the tumor cell borders, while T2 best defined the associated area of tumor cell invasion. It is generally accepted that the level of enhancement is a reflection of tumor angiogenesis. We found that the extent of enhancement with gadolinium in T1 sequences closely corresponded to the microvascular density of this U87 intracranial model. An interesting and unexpected finding was high variability in tumor growth pattern, which provides invaluable insight into intracranial tumor behavior and has significant therapeutic relevance. This study establishes a protocol for high-throughput whole-brain MRI to be used as a research tool for investigating the biology of both xenograft and transgenic mouse models of intracranial tumor, identifying sequences that can best determine characteristics of interest, such as tumor necrosis, edema, angiogenesis, and invasion.


Bradley J. Erickson, Christopher Wood, Norbert Campeau, E. Paul Lindell, Vladimir Savcenko, Norman Arslanlar, and Liqin Wang; Department of Radiology, Mayo Clinic, Rochester, Minnesota, USA

After IRB approval, we collected a series of 100 MRI examination pairs from patients with a variety of brain tumors. We then presented pairs of examinations either with or without automated image alignment (“image registration”). Six neuroradiologists then reviewed the images and determined if they thought the examination showed progression or stability, and their degree of confidence. There were 2 settings, where half the exam pairs were registered, and half were without registration. The other setting presented the converse—registered if first setting was without registration. Later, the same group met, and a consensus opinion was reached as to whether there was progression or not. Agreement was reached in 97 of the 100 cases. In some cases, there were areas of both progression and regression, but any study showing progression in any area was considered progression. We found that image registration showed a trend to improved accuracy. There was also a higher confidence in ratings, as indicated by a higher area under ROC curve (Az .87 vs .78). Anecdotally, there was also consensus that image registration made comparison more efficient and introduced no significant image degradation. Future clinical studies should consider using image registration to improve accuracy and efficiency of image interpretation.


A.J. Frew, P.B. Tseng, D.M. Bier, A. Batra, R. Pagare, and J.R. Alger; Biomedical Physics, Neurology and Radiology, University of California at Los Angeles, Los Angeles, California, USA

Brain tumor patients undergo multiple MRI studies for evaluating changes in morphology and image characteristics (e.g., enhancement). It is frequently the case that MRI studies obtained at different times can not be evaluated for subtle changes because of differing slicing and between-study intensity variations. Standalone JAVA software was developed to address this problem. JAVA was used to develop user-friendly software that (1) reads digital MRI data from multiple time points, (2) automatically co-registers all MRI data, (3) normalizes the MRI intensity across time, and (4) animates the resultant registered and re-normalized images. Three-dimensional volume images from each time point are registered to a user-defined target volume. The registration process aligns and reslices the volumes so that each time point has a constant number of slices having the same orientation. Registration is performed by AIR 5.0 (Automatic Image Registration) programs using a 12-affine algorithm that allows for shift, rotation, rescaling, and shearing of the registrant volume. To standardize the intensity across time, the program allows the normalization of signal intensity to white matter intensity. The animation feature allows two “movement” axes. The first axis (slice changer) implements the transition among sequential slices in a given volume, much like in any common DICOM-viewing application. The second axis (longitudinal) implements the transition among corresponding slices of volumes in the longitudinal study. The software was tested on personal computers using the Windows operating system. The software was used to visualize changes occurring in clinically acquired MRI from patients with brain tumors. Longitudinal inspections can be made by scrolling through the different volumes and examining the temporal changes that have occurred in a particular slice. The software allows inspection of spatially registered, multiple slice images across time so that changes in tumor morphology and image characteristics can be readily appreciated. The JAVA MRI animator provides a user-friendly interface that facilitates quick and qualitative inspection of brain tumor changes over time. The software is structured so that it could be used on computers that are commonly available in offices and clinics.


Vinay Gupta,1,4 Weijun Wang,2,4 Chien-Kuo Tai,5 James R Bading,3 Noriyuki Kasahara,5 and Thomas C Chen1,2,4; Departments of 1Pathology, 2Neurosurgery, and 3Radiology and 4USC Center for Brain Tumor Research, University of Southern California, Los Angeles, California; 5Department of Medicine, University of California at Los Angeles, Los Angeles, California; USA

Molecular imaging enables real-time analysis of in vivo tumors transfected with reporter genes of interest to non-invasively study molecular events and pathways of interest. One such paradigm is the use of herpes simplex virus thymidine kinase (HSV-tk) gene as the reporter gene and radiolabeled nucleotide analogues as probes. We have used a viral vector that is a modified replication competent retrovirus carrying the HSV-tk gene (RCR-tk). This vector serves a double purpose, as the transfected gene “thymidine kinase” can be used for imaging as well as for treating the tumor in the traditional suicide gene therapy. In vitro experiments show that RCR-tk is stable and is able to transduce fresh U87 cell cultures over eight transfection cycles. Cytospins obtained from these cultures showed positive immunohistochemical staining for the thymidine kinase protein. The functional activity is retained as the prodrug ganciclovir was cytotoxic to the transduced cells using MTT assay. In vivo experiments demonstrate the ability of RCR-tk to efficiently transduce subcutaneous U87 tumors and tumor-wide expression of the thymidine kinase protein as early as 7 days after vector inoculation. Since the timing of prodrug delivery is important to enable optimal therapeutic response, PET imaging for the tk gene product will determine the level of transduction and extent of transgene expression. We plan to use [C11] FMAU as the radiolabeled probe in PET imaging of animals carrying intracranial and subcutaneous tumors in monitoring tumor growth and response to therapy.


David C. Hockaday, Sui Shen, Andrew Raubitschek, David Colcher, An Liu, John Fiveash, and Adam N. Mamelak; Departments of Radioimmunotherapy and Neurosurgery, City of Hope Cancer Center and Beckman Research Institute, Duarte, California; Department of Radiation Oncology, University of Alabama at Birmingham, Birmingham, Alabama; USA

Traditional imaging methods such as MRI, MRS, and 18FDG PET poorly define glioma extent and often fail to differentiate viable tumor from necrosis. TM-601, a 36 amino-acid peptide, selectively binds to glioma cells but not normal brain parenchyma and as such may be an ideal imaging agent to determine glioma extent. A Phase I clinical trial of intracavitary 131I-TM-601 in adult patients with recurrent high-grade glioma was completed. We evaluated images and biodistribution data from this trial to assess if 131I-TM-601 is useful in determining tumor extent. Adult patients with recurrent high-grade glioma underwent tumor resection, implantation of an intracavitary reservoir, and a single-dose injection of 10 mCi 131I-TM-601 (0.25–1.0 mg) 2 to 4 weeks after surgery as part of a Phase I/II clinical trial. Total body planar images and brain SPECT scans were obtained on days 0, 1, 2, 3, and 6 to 8 after injection. Postoperative MRI was also obtained and co-registered to the SPECT scans using Analyze 5.0 software. Analysis of the rate of radio decay and biological elimination in the body and at the cavity site was performed. T1-Wc, T2, and SPECT volumes were estimated by stereological Cavalieri sections and compared for overlap. Non-bound 131I-TM-601 was eliminated by 48-h postinjection with the remaining radiolabeled peptide bound to tumor for at least 6 to 8 days. Biological decay rates from 24 to 168 h postinjection mimicked the physical decay of 131I. 131I-TM-601 determined tumor volumes more closely paralleled T2 volumes than T1-Wc, suggesting the radiolabeled peptide diffused into the brain parenchyma to the approximate volume of glioma extent. Overlap between co-registered MRI and SPECT scans corroborated the presence of radiolabeled peptide in the vicinity of tumor invasion at 24 and 168 h postinjection. Even with the poor spatial resolution of 131I, reliable tumor extent measures could be obtained. We conclude that 131I-TM-601 reliably estimates primary tumor extent. Further development of TM-601 with better imaging isotopes such as 123I or 124I may provide an important tool for determining tumor extent and differentiating regions of viable tumor from necrosis.


P. Jun, S. Cha, T. Tihan, J. Chang, S. Chang, M.D. Prados, W.P. Dillon, and M.S. Berger; Departments of Neurosurgery, Radiology and Pathology, University of California San Francisco, San Francisco, California, USA

Differentiating recurrent high-grade glioma from radiation necrosis is clinically important for guiding therapeutic decision-making, but determining this distinction has often proved challenging because of the limited specificity of conventional anatomic imaging. Novel physiologic MR imaging methods, such as dynamic susceptibility contrast-enhanced (DSC) MR imaging, promises to offer better distinction between these lesions by providing pertinent underlying microvasculature and permeability information. The purpose of this study was to investigate the ability of conventional anatomic and DSC MR imaging methods in the differentiation of recurrent high-grade glioma from radiation necrosis with histopathologic correlation. We retrospectively reviewed data for 22 patients with previously treated glioblastoma multiforme (GBM) who presented with clinical and radiographic evidence of recurrence. All patients underwent anatomic and DSC MR imaging immediately before their second resection, followed by histologic confirmation. Based on histologic evaluation of the resected tissue, patients were grouped into those with predominantly tumor (n = 16) or those with predominantly necrosis (n = 6, defined as >90% necrosis on histology). From anatomic MR images, contrast-enhancing lesion (CEL) and non-enhancing surrounding T2 lesion (T2L) volumes were determined. From DSC MR images, quantitative cerebral blood volume (CBV) measurements were derived. Anatomic imaging displayed no significant difference in the volumes of CEL and T2L in the two groups. DSC MR imaging, however, revealed that the maximum relative CBV (rCBVmax) in the recurrent tumor group ranged from 2.16 to 7.95 (3.60 ± 1.64, median ± SD), whereas the rCBVmax in the predominantly necrosis group ranged from 0.73 to 3.8 (1.84 ± 1.13). The difference in median rCBVmax between the two groups was significant (P < 0.01). Anatomic MR imaging is commonly used to differentiate tumor recurrence from therapy-related necrosis but is limited by its nonspecificity. Histologic evaluation remains the reference standard for distinguishing between the two but is subject to tissue sampling bias and radiation-induced tissue changes. The results of DSC MR imaging suggests that there is significant difference in blood volume between histologic evidence of active tumor and necrosis in patients with previously treated GBM. Further investigations to directly correlate imaging with histologic standards may strengthen the validity of DSC MR as a noninvasive method to differentiate active tumor from therapy-related necrosis. This study was supported by grant NS45013 and an Accelerate Brain Cancer Cure grant.


N. Kamdar, W. Chen, T. Cloughesy, L. Liau, M. Bergsneider, N. Satyamurthy, J. Barrio, J. Czernin, M. E. Phelps, and D. H. Silverman; University of California Los Angeles, David Geffen School of Medicine, Los Angeles, California, USA

The object of this study was to compare the utility of FDG, FLT, and FDOPA-PET tracer uptake in predicting survival of patients with brain gliomas. Twenty-three adult patients (mean age 49 ± 15 years) underwent brain PET scans with FDG, FLT, and FDOPA PET tracers. Ten patients had newly diagnosed gliomas (two with grade II and eight with grade III/IV gliomas), and 13 patients had residual/recurrent tumor (one with grade II and 12 with grade III/IV gliomas). The maximum SUV (SUVmax) of each tracer in each tumor was measured. At the time of this writing, 10 patients had died within 10 months of the time of the PET scan, and 10 patients have survived for greater than 10 months. Three patients who have been followed for less than 10 months are still alive. Descriptive SUVmax statistics of patients who survived and those who died were obtained for each of the tracers and compared by a 2-tailed, Student t test. ROC analysis identified threshold PET tracer uptake values that distinguished those who died less than 10 months from the PET scan from those who did not. (The three patients who remain alive but have been followed less than 10 months were excluded from this ROC analysis). Survival of patients above the threshold value for each tracer was compared to survival of those below the threshold value using Kaplan-Meier analyses and log-rank statistical tests. FLT uptake for those who died (1.36 ± 0.70) was twice as great as that for patients who survived (0.69 ± 0.27, P = 0.01). Patients who died had a mean FDG uptake of 6.38 ± 3.87 (mean ± SD) compared to 3.93 ± 1.86 for those who survived (P = 0.11). Patients who died had a mean FDOPA uptake of 4.52 ± 2.48 compared to 3.10 ± 1.38 for those who survived (P = 0.13). ROC analysis showed optimal separation with respect to survival using threshold SUVmax values of 4.1, 1.0, and 3.2 for FDG, FLT, and FDOPA, respectively. The FDOPA uptake threshold did not predict patient survival (P = 0.29). Patients with FDG and FLT uptake above the threshold levels showed decreased survival compared to the “below-threshold” group, with significance levels of P = 0.06, and P = 0.03, respectively. Among the tracers examined, FLT uptake was the best predictor of survival for patients with brain tumors. Those patients having tumors with FLT SUVmax ≥1.0 had higher mortality (70% died within the first year of follow-up) than those having tumors with SUVmax < 1.0 (23% died within 14 months of follow-up). These results suggest that FLT has prognostic value for patients with gliomas. FDG uptake has less predictive value of survival than FLT, whereas FDOPA uptake does not appear to predict patient survival in this study.


M. Law, E. Knopp, S. Oh, E .Wang, J. Babb, M. Inglese, D. Zagzag, and G. Johnson; NYU Medical Center, New York, New York, USA

The purpose of this study was to determine if cerebral blood volume (CBV) measurements can help predict tumor behavior in low-grade gliomas (LGGs) and specifically whether LGGs with high initial CBV show more rapid progression than those with low CBV. Thirty-five patients with histologically diagnosed LGGs, comprising low-grade astrocytomas (n = 21), low-grade oligodendroglioma, and low-grade mixed oligoastrocytomas (n = 14), were studied with dynamic susceptibility contrast-enhanced perfusion MRI (DSC MRI). Patients were divided into two groups, low and high CBV, based on a threshold value of 1.75. This threshold had previously been found to give the optimal sensitivity and specificity for differentiating LGGs from HGGs in a logistic regression analysis of 120 high-grade and 40 low-grade gliomas at our institution. Tumor volumes (enhancing T1 and T2 volumes) were measured at the initial examination and again at 6, 12, 18 and >18 months. The patient’s clinical status was also followed over time. LGGs with low CBV (n = 16) demonstrated a CBV of 1.24 ± 0.42 (mean ± SD) initially and 1.38 ± 0.42 at follow-up (P = 0.6). LGGs with high CBV (n = 19) demonstrated CBV of 3.50 ± 1.44 initially and 5.31 ± 3.03 at follow-up (P < 0.05). Tumor volumes decreased over time (follow-up to 900 days) for the low-CBV group and remained stable for the high-CBV group. The clinical follow-up results for the low-CBV versus high-CBV groups were as follows: complete response, 4/16 versus 0/19; stable disease, 9/16 versus 6/19; progressive disease, 2/16 versus 11/19; death, 1/16 versus 2/19. Complete response and stable were assigned a value of “0” and progressive and death a value of “1,” and the two groups were compared, yielding a P-value = 0.000025. DSC MRI can be used to identify two groups of LGGs which appear to have different biologic behavior on follow-up studies. LGGs with low initial CBV tend to have low CBV on follow-up, decreasing tumor volumes, and a favorable clinical course. LGGs with high initial CBV tend to have progressively increasing CBV and a significantly poorer clinical course. This suggests either that high-grade gliomas are misdiagnosed as LGGs because of sampling error at biopsy or that DSC MRI can predict the subset of LGGs which may have a propensity for malignant transformation.


Anastasios N. Mavrakis, Elkan F. Halpern, R. Gilberto Gonzalez, and John W. Henson; Brain Tumor Center, Division of Neuroradiology, Department of Radiology, Institute for Technology Assessment, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, USA

We sought an approach to optimize rapid selection of a biopsy site in patients with a newly detected brain mass in the absence of a history of cancer. Many patients with a newly detected brain mass undergo extensive diagnostic testing in search of a systemic primary neoplasm prior to selection of a biopsy site. This testing is expensive and may delay biopsy site selection, leading to further cost increases. We performed a retrospective review of the diagnostic workup of patients whose systemic cancer presented as brain metastasis and who had their radiologic evaluation and diagnostic biopsy at our institution over a 6-year interval. For a comparison group, we reviewed the diagnostic evaluation of patients with newly diagnosed primary brain tumors. Symptoms of brain metastasis were the presenting manifestation of systemic cancer in 168 (14%) of 1194 patients with brain metastasis. Eighty-eight patients with brain metastasis met study selection criteria. Lung cancer was over-represented as a primary tumor, occurring in 82% of patients presenting with brain metastasis, compared to a 47% frequency of lung cancer in brain metastasis patients who did not present with neurological symptoms (P < 0.01). Chest CT revealed evidence of tumor in 83/86 (97%) of patients with a systemic primary. Brain MRI and chest CT together identified the site ultimately chosen for diagnostic biopsy in all except two patients (one with cutaneous melanoma and one with carcinoma of the uterine cervix). One half of the patients underwent brain biopsy as the primary diagnostic procedure, with the majority undergoing a craniotomy. The initial biopsy site decision in the majority of patients was whether to perform a neurosurgical procedure. The average interval between first detection of a brain mass on neuroimaging and a diagnostic biopsy of any lesion was 4.7 days. We estimated that an optimized approach focusing on brain MRI and CT chest, followed by an early neurosurgical decision, could reduce the interval to diagnosis by at least 10%. We conclude that chest CT and MRI of the brain can be used together to rapidly identify the optimal biopsy site in virtually all patients with a newly detected brain mass.


K. Mayo, J. Lupo, N. Butowski, S. Chang, M. Prados, D. Larson, S. Nelson, M.S. Berger, and S. Cha; Departments of Radiology, Neurosurgery, and Radiation Oncology, University of California San Francisco, San Francisco, California

The current standard of neuroimaging to monitor tumor response to radiation therapy is limited by non-specificity and its inability to depict beyond anatomic alterations. We investigated the morphologic and hemodynamic parameters of treatment-naive glioblastoma multiforme (GBM) using anatomic and T2* bolus tracking perfusion MR imaging (pMRI), before and after external beam radiation therapy (EBXRT). We hypothesized that changes in peak height (PH) and abnormal recovery (aRec) indirectly correlate with alterations in tumor angiogenesis and capillary permeability, respectively, following EBXRT in patients with GBM. Eighteen newly diagnosed GBM patients prior to surgery were enrolled in the study. These patients were imaged at four time points: preoperative, pre-EBXRT, post-EBXRT, and at a six-month follow-up. The patients were grouped according to the initial surgical treatment: (1) gross total resection (GTR; n = 11) or (2) subtotal resection or biopsy (STR/Bx; n = 7). Serial anatomic volumetric changes were measured in two distinct image-defined regions: contrast enhancing lesion (CEL) and non-enhancing T2 lesion (T2L). In addition, serial microvascular changes were assessed within these regions by measuring PH and aRec from the analysis of T2* signal intensity-time curves obtained from pMRI. The changes in T2L volume were interpreted in the context of any alteration in steroid dosage. In the GTR group, the greatest decrease in anatomic volumes after surgery was noted between pre- and post-EBXRT. As indicated by pMRI, the PH and aREC mildly decreased after EBXRT within both the CEL and T2L regions. None of the changes were statistically significant. In the STR/Bx group, a dramatic increase in T2L volume was noted between pre- and post-EBXRT, with a modest increase in CEL volume. Perfusion parameters mirrored these anatomical changes with an increase in PH and a decrease in aRec in both enhancing and non-enhancing regions of the tumor. Our preliminary data suggest that quantitative hemodynamic assessment of tumor within anatomically defined regions adds more specific information about tumor activity during and after radiation therapy. Our results suggest that the tumor burden immediately prior to starting radiation therapy has a profound effect on imaging of GBM following radiotherapy. We observed marked changes in both morphologic and hemodynamic parameters in the subtotal/biopsy group but not in the gross total resection group. Quantitative assessment of perfusion MR imaging may be able to provide more specific information on vascular changes in GBM following irradiation. This study was supported by grant NS45013 and an Accelerate Brain Cancer Cure grant.


Edward A. Neuwelt, Tulio P. Murillo, Gary Nesbit, and Sandor Manninger; Oregon Health and Science University, Portland, Oregon, USA

Ultrasmall super paramagnetic iron oxide (USPIO) particles are MR contrast agents that produce signal changes comparable to or greater than gadolinium (Gd). However, unlike gadolinium, these agents are trapped by reactive astrocytes and macrophages and are thereby helpful in functional identification of lesions in the central nervous system. The MR images can be correlated with tissue localization using iron stains. Seventy-one intracranial lesions in 66 patients were prospectively studied with IRB approval. Ferumoxtran-10 (Combidex) at a dose of 2.6 mg/kg in 100 cm3 of normal saline was infused at 4 ml/min 24 h prior to MRI - T1, T2, T2*, and DWI. This MR was compared to a Gd-contrasted study done within 28 days prior to the ferumoxtran-10 study. No adverse events were observed. In 9/28 cases in the glioma group, ferumoxtran-10 showed more (4 cases) or much more (5 cases) enhancement than Gd. In non-glial tumors, more (3/22) or much more (2/22) enhancement was seen with ferumoxtran-10. In multiple sclerosis (n = 8) the enhancement was less with ferumoxtran-10 compared to Gd, but more in 2/5 cases of acute disseminated encephalomyelitis. One patient with an acutely partially thrombosed cavernoma and another patient with an acute middle cerebral artery stroke showed much more enhancement with ferumoxtran-10 than with Gd. Ferumoxtran-10 can assist in delineation of lesions including brain tumors by MR, especially when the BBB is compromised and inflammatory cells are involved in the lesion.


Justin S. Smith, 1,4 Soonmee Cha,2,3 Katie Mayo,2 Susan M. Chang, William P. Dillon,2 Michael D. Prados,3 and Mitchel S. Berger1,4; Departments of 1Neurological Surgery, 2Radiology, and 3Neuro-Oncology and 4Brain Tumor Research Center, University of California San Francisco School of Medicine, San Francisco, California, USA

Diffusion-weighted MRI (DWI) is a physiology-based MR imaging method that measures the degree of water diffusion in vivo. Areas of reduced diffusion are most notably present in acute infarction of the brain but can also be seen in any type of brain injury that alters extracellular water motion. Immediately following brain surgery for glioma resection, we have observed areas of reduced diffusion around the resection cavity. The purpose of this study was to investigate the incidence, clinical significance, and course of the reduced diffusion abnormality. Forty-four consecutive patients with newly diagnosed glioma (WHO grade II to IV) were prospectively followed with serial MR imaging, including DWI. Clinical and surgical data were also collected for these patients. Post-operative imaging demonstrated 28 (64%) patients with areas of reduced diffusion adjacent to the resection cavity with a mean volume of 8.2 ± 1.5 mm3. Corresponding apparent diffusion coefficient (ADC) maps were used to distinguish reduced diffusion from T2 shine through. Areas of reduced diffusion resolved and subsequently demonstrated gadolinium enhancement in 25 cases (89%), with one case having persistent reduced diffusion at last imaging on post-operative day (POD) 244. Enhancement was noted as early as POD 15 and was noted as late as POD 198. Enhancing regions cavitated in 24 cases (96%). The occurrence of postoperative reduced diffusion was not predicted by any of the clinical or surgical parameters assessed. There were no discernable clinical deficits attributable to the areas of reduced diffusion. DWI abnormality on the immediate postoperative MR imaging was not infrequent following surgical resection of newly diagnosed diffuse gliomas. The abnormality typically resolved, replaced with contrast enhancement on one-month follow up, and ultimately resulted in a cavity on long-term follow up. Although there were no apparent clinical consequences directly attributable to these abnormalities, imaging during the period of enhancement could be confused with recurrent tumor and interpreted as early treatment failure. The baseline MR imaging just prior to the start of irradiation was invaluable in documenting the emergence of contrast enhancement within the prior DWI abnormality on the immediate post-operative scan and in accurate assessment of early response to irradiation based on comparison with the post-irradiation scan. Our study strongly suggests that immediate post-operative MR imaging with DWI is essential, as is the immediate pre-irradiation MR imaging, for monitoring disease progression. Any new enhancement following glioma surgery should be interpreted in the context of immediate post-operative DWI abnormalities.


K.R. Swanson,1 M. Muzi,2 A.M. Spence,3 J.G. Rajendran,2 J.R. Grierson,2 and K.A. Krohn2; 1Departments of Pathology, 2Radiology, and 3Neurology, University of Washington, Seattle, Washington, USA

[18F]-Fluoromisonidazole (FMISO) has been demonstrated as a potential quantitative imaging agent for hypoxia in high-grade brain tumors. Imaging hypoxia in brain tumors is of particular interest, as the persistence of the hypoxic component of gliomas has been implicated in the failure of radio-therapies. To assess the difference in the information provided by hypoxia imaging versus anatomical imaging of gliomas during radiation therapy, high-grade glioma patients were imaged with FMISO and with standard contrast-enhanced MRI. Sixteen adult patients with gliomas were imaged with FMIS0 before, during, or after radiation therapy: 14 before, 2 during, and 5 after radiation therapy These patients were imaged 120 to 140 min after injection of 7 mCi FMISO to generate a static image. The venous blood concentration during the imaging period was used to create tissue to blood ratio (TB) images. Using previous estimates that TB values above 1.2 represent hypoxia, we determine hypoxic volumes for each of these patients. These patients also received MR imaging, gadolinium-enhanced T1-weighted and T2-weighted, of their brains before and after treatment. Co-registration of the FMISO and MR images as well as segmentation of these images provided MR-defined tumor volumes at each time point. FMISO-defined hypoxic volumes ranged from 0 to 33.1 cm3 while the MRI-defined tumor volumes ranged from 20.6 to 148.7 cm3 (correlation r = 0.77). Although FMISO-defined hypoxic volumes are smaller than MR-defined tumor volumes, the hypoxic volume typically overlaps (just inside or outside) the periphery of the gadolinium-enhanced MR signal while completely within the limits of the T2 volume. Although MR volumes typically decrease (–28%) during the course of radiation therapy, the average increase in hypoxic volume during radiation therapy is 10%. This difference suggests that the radiation therapy administered to these patients was insufficient to treat the resistant hypoxic volume. These results are consistent with the hypothesis that FMISO provides different information about the extent of the disease as well as the efficacy of radiation therapy. This work was supported by NIH grant P01-CA42045.


K.R. Swanson,1,2 H. Harpold1, S. Nissen1, C. Bridge1, A. Lai,3,4 T. Cloughesy,3,4 and E.C. Alvord Jr.1; 1Department of Pathology, University of Washington School of Medicine, Seattle, Washington; 2Department of Applied Mathematics, University of Washington, Seattle, Washington; 3Department of Neurology, University of California, Los Angeles, California; 4The Henry E Singleton Brain Cancer Research Program, UCLA School of Medicine, Los Angeles, California; USA

We have developed a mathematical model describing the growth and invasion of glioma cells throughout the complex structure of the brain (Swanson et al., 2002, Br. J. Cancer 86:14; Swanson et al., 2002, Acta Biother. 50:223; Swanson et al, 2003, Math. Comp. Modelling 37:1177; Swanson et al., 2003, J. Neuro. Sci. 216:1). The model integrates the two key fundamental processes of net proliferation (r) and migration (D) rates of individual glioma cells and firmly supports the concept that gliomas infiltrate so extensively that they cannot be cured by resection alone (Swanson et al., 2002, Br. J. Cancer 86:14; Swanson et al, 2003, Math. Comp. Modelling 37:1177). We have also developed a formula to quantify the model parameters, r and D, in enhancing gliomas of each individual patient through analysis of standard gadolinium-enhanced T1-weighted and T2-weighted MRI sequences. This allows us to dynamically assess r and D for the malignant glioma cells during the course of treatment. We present the results of the application of this technique to 7 retrospectively obtained glioblastoma patients with sufficient MRI follow-up to allow for detailed analysis. Each patient received at least 2 MRIs before surgical treatment and was followed with serial MRIs approximately every 2 months (range, 4–26, mean: 11 available MRIs per patient). The mathematical model’s ability to quantify the ever-changing kinetics of the glioma cells sheds light on the in vivo effects of standard radiotherapy as well as chemotherapies including Temodar and Accutane. Our results suggest the differential propensity for treatments to target the central bulk of the lesion while neglecting the critically invasive cells at the periphery.


M.J. Vos,1 J. Berkhof,2 T.J. Postma,1 O.S. Hoekstra,2,3 F. Barkhof,4 and J.J. Heimans1; Departments of 1Neurology, 2Clinical Epidemiology and Biostatistics, 3Nuclear Medicine and PET Research, and 4Radiology, VU University Medical Center, Amsterdam, the Netherlands

In an earlier study, we demonstrated that thallium-201 single-photon emission computed tomography (201Tl SPECT) is superior to conventional CT and MRI in the early prediction of overall survival (OS) in patients treated with chemotherapy for recurrent glioma (Vos et al., 2003, J. Clin. Oncol. 21:3559). The aim of the present study was to estimate 201Tl SPECT and CT-MRI cut-off values that lead to a prognostic classification with an optimal predictive validity for OS, in order to discriminate glioma patients with good and poor prognosis at an early stage during chemotherapeutic treatment. We studied 44 patients who underwent 201Tl SPECT and CT-MRI before and after two courses of chemotherapy. Cut-off values were retrieved from the Cox model. Patients were classified according to the computed cutoff values, creating subgroups of patients with different prognosis in terms of survival (tumor regression (TR), stable disease (SD), and tumor progression (TP). The differences between the subgroups were assessed by Cox regression and Kaplan-Meier analyses. We found that 201Tl SPECT discriminated 25% of patients as TR, and 34% as TP, with corresponding good, respectively poor prognosis (median OS 31, respectively 7 months). With CT-MRI, the vast majority (82%) of patients were classified as SD, and only 4% and 14% as TR and TP, respectively. We constructed a 201Tl SPECT model that enables the identification of glioma patients with good and poor prognosis already at an early stage during chemotherapeutic treatment. This model offers the treating physician the possibility to select responding patients for further treatment, and to avoid ineffective, potentially toxic treatment in non-responding patients.


Yigal Shoshan, Laboratory for Cellular and Molecular Immunology, Department of Medicine, Hadassah University Hospital, Jerusalem, Israel

Intraoperative MRI (iMRI) has gained broader acceptance and use in recent years, particularly with the third generation of intraoperative MRI devices, such as the Odin PoleStar N-10. The scanner is a portable unit on wheels that is stored in a shielded iron cage located in the neurosurgical operation room when not in use. Prior to surgery, the unit is brought out of the cage, positioned by one or two people, and docked under a regular operating table. A shielded cabling system permanently connects the scanner to the electrical cabinets located in an adjacent room. Our PoleStar N-10 system is delivered with a mobile shielding system (RF shield), which represents an alternative to a complete OR shielding and covers only the scanner and the OR table. Between January 2003 and April 2004, 25 procedures were performed with the second worldwide PoleStar N-10 installation in a conventional, non-RF-shielded neurosurgical suite. The procedures included 22 craniotomies for the resection of hemispheric tumors (17 gliomas, 1 lymphoma, 2 meningiomas, and 2 metastases) and three transsphenoidal surgery pituitary adenomas. The surgeries were performed with the use of conventional neurosurgical equipment including surgical microscope and ultrasonic aspirator, with the PoleStar N-10 system in a standard non-RF-shielded neurosurgical suite. Average length of surgery was 7.2 (6.1–11.5) h compared to 4.8 (2.2–7.3) h with standard neuronavigation procedures. Based on our initial clinical experience, we conclude that the quality of the images obtained during surgery with restricted field of view was very good compared to pre-operative diagnostic MR images. Magnet integration with standard neuro-surgical environment was good. The accuracy and precision of the integrated navigation system is very good compared to a standard neuronavigation system. The capital investment for the system is relatively small compared to other available iMRI systems. However, the system’s user interface is less intuitive, and not easy to use. The logistics associated with using the RF shield are somewhat cumbersome and adds 60 to 90 min to a typical procedure. Head positioning and fixation with the PoleStar N-10 limits the surgical approach in some patient. Some of these limitations will be overcome with further innovations of the system and improved operator efficiency.



M.A. Badruddoja, D.A. Reardon, J.J. Vredenburgh, J.A. Quinn, J.N. Rich, A. Desjardins, A. Friedman, A. Walker, P. Bailey, R. Jex, and H.S. Friedman; Brain Tumor Center, Duke University Medical Center, Durham, North Carolina; Pfizer Pharmaceuticals, Kalamazoo, Michigan; USA

CPT-11 is a semisynthetic derivative of camptothecin (camptotecin accuminata) that preferentially binds and inhibits DNA topoisomerase I, an enzyme necessary for DNA replication. Celecoxib is a selective inhibitor of cyclooxygenase-2, an important enzyme in the formation of prostaglandins (PGs) from arachidonic acid. Increased PG levels due to increased COX-2 expression have been demonstrated in a variety of malignancies including malignant glioma, pancreatic, gastric, esophageal, prostate, lung, head and neck, and colon cancers. COX-2 has also been implicated in several tumor-related processes including angiogenesis, invasion, and apoptosis. The primary objective of this study was to evaluate the objective response rate of CPT-11 and celecoxib in patients who had newly diagnosed and previously untreated glioblastoma multiforme. A secondary objective was to evaluate the toxicity of this regimen. Eligibility criteria included the following: histologic diagnosis of glioblastoma multiforme; age >18 years; KPS >60%; adequate hepatic, renal, and bone marrow function; and measurable tumor (>1.5 cm residual tumor on postoperative imaging). Patients were stratified according to concurrent use of enzyme-inducing anti-convulsants (EIACs). CPT-11 was administered on days 1, 8, 22, and 29 of each 42-day cycle at 350 mg/m2 for patients receiving EIACs (Stratum A) and at 125 mg/m2 for those not on EIAC (stratum B). All patients received celecoxib 200 mg po BID for 7 days, then increased to 400 mg po BID thereafter. Physical exams and MRI evaluations were performed after every cycle. Patients received radiation therapy after completion of three cycles or if there was evidence of clinical and radiographic progression of disease. A total of 36 patients were enrolled, including 26 on Stratum A and 6 on Stratum B. The mean age was 57 years (range, 41–74). Eighteen (50%) were male. Eight patients had only biopsy performed, while the remaining patients had partial resections performed with residual tumor on postoperative imaging. Overall the regimen was well tolerated. Three patients experienced diarrhea (grade 3, n = 2; grade 1, n = 1), and one patient experienced grade 5 sepsis during cycle one. Among 18 assessable patients, 2 patients achieved partial responses, 6 patients had stable disease and 11 had progressive disease. It is too early to assess six patients. The combination of CPT-11 with celecoxib is well tolerated and may have modest activity in the neoadjuvant setting. Further outcome analysis will be presented pending further accrual and follow-up.


Julette F. Batara, Stuart A. Grossman, and Ziya Gokaslan; The Johns Hopkins Medical Institutions, Baltimore, Maryland, USA

Metastatic epidural cord compressions (ECCs) are common and often result in irreversible neurologic damage and pain. A recent randomized trial documented that surgery plus radiation therapy is superior to radiation in patients with ECC. However, many patients with ECC are not optimal candidates for surgery. This retrospective review was conducted to estimate the proportion of patients with ECC who would qualify as surgical candidates on the Patchell study. All spinal MRIs performed at Johns Hopkins over a five-year period ending in December 2002 where the requisition form or radiologist’s final impression contained the words “spinal cord compression” were identified by electronically searching the institutional radiology database. The eligibility criteria used in the randomized Patchell study were applied to the identified cases by using MRI reports and medical records. MRIs of all cases meeting the Patchell criteria were reviewed by a spine surgeon. A total of 1592 MRIs were identified from the database by using the search criteria defined above. Of these, 80 met the initial Patchell criteria. However, 51 (64%) were excluded because of involvement of multiple spinal levels (39), death in less than 3 months (8), myeloma or lymphoma histology (2), complete paraplegia (2), or young age (1). MRIs from four patients could not be found. Neurosurgical review of MRIs from the 25 potentially eligible cases confirmed that 18 (22%) were suitable for surgery assuming that there were no clinical contraindications. Although ideally managed with surgery plus radiation, approximately 75% of patients with ECC will not ful-fill the criteria used in the Patchell study which documented the advantage of this combined modality approach. Careful multidisciplinary evaluations are required to make optimal decisions for patients with ECC.


Peter M. Black, Elizabeth B. Claus, Ronghui Xu, Andres Horlacher, Liangge Hsu, Richard B. Schwartz, Florin Talos, and Ferenc A. Jolesz; Brigham and Women’s Hospital, Boston, Massachusetts, USA

There is suggestive evidence that resection of low-grade gliomas increases patient survival. Intraoperative MR imaging and intravenous sedation anesthesia make gross total resection increasingly possible. We report our early experience in adult patients with supratentorial low-grade gliomas resected by one surgeon (PB) in the magnetic resonance operating suite at Brigham and Women’s Hospital. To provide a preoperative assessment of resectability, we have begun to develop a mathematical model of resectability from 101 patients. This model uses proximity to eloquent cortex, deep white matter tracts, and cerebral vessels. To estimate the effect of resection on survival, we used Kaplan-Meier methodology and a Cox proportional hazards model in 156 patients. We compared death rates to the expected number of deaths from age- and histologic-specific survival rates obtained from the Surveillance, Epidemiology, and End Results Registry (SEER). Patients who underwent subtotal resection were at 1.4 (95% confidence interval [95% CI], 0.7–3.1) times the risk of recurrence and 4.9 (95% CI, 0.61–40.0) times the risk of death relative to patients undergoing gross total resection. We will review the techniques that allow more aggressive resection. These data suggest a reconsideration of therapy for low-grade gliomas, with aggressive resection using intraoperative imaging playing a prominent role.


D.T. Blumenthal,1 F. Bokstein,2 J.W. Conlee,1 J.A. Batten,1 R.L. Jensen,1 K.L. McNeill,1 E. Jacques,1 J. Baram,3 and Z. Ram2; 1University of Utah/Huntsman Cancer Institute Neuro-Oncology Program, Salt Lake City, Utah, USA; 2University of Tel Aviv/Sourasky Medical Center, Tel Aviv, Israel; 3Chaim Sheba Medical Center, Tel Hashomer, Israel

The use of adjuvant chemotherapy for the treatment of glioblastoma multiforme (GBM) varies between differing health care systems. Controversy exists regarding the role of chemotherapy concurrent with or immediately following radiation for these tumors. In the United States, common practice for treatment of GBM typically includes early use of adjuvant chemotherapy. This approach is not used routinely in Israeli centers, where chemotherapy is often reserved for documented tumor progression only. We examined retrospective data from a subset of 200 patients with de novo GBM treated from 1997 to 2003 at Tel-Aviv University-affiliated Medical Centers in Israel and 90 patients with de novo GBM treated during the same time period at the University of Utah/Huntsman Cancer Institute. All patients were treated with surgical resection or biopsy and external beam conformal radiation (RT). Demographic information of gender, age, degree of surgical resection, and performance score was obtained. Following RT, the patients from Israel were observed without chemotherapy until the time of documented progression, while the Utah patients received adjuvant chemotherapy immediately following (or in some cases concurrent with) RT. The mean overall survival for the group from Israel was 41.1 weeks. Mean survival in the U.S. group was 57.7 weeks. Overall survival and progression-free survival of the two sets are compared with demographic differences noted. Significant variations in outcome between the patients treated in the two systems may be attributed to different modes of practice regarding adjuvant chemotherapy use.


A.A. Brandes,1 G. Cavallo,3 M. Reni,1 A. Tosoni,1 L. Nicolardi,1 E. Scelzi,1 A. Morabito,1 L. Scopece,3 E. Franceschi,3 and M. Ermani2; 1Departments of Medical Oncology and 2Neurological Sciences, Padova University, Padova; 3Bellaria Hospital, University of Bologna, Bologna; Italy

The role of chemotherapy in recurrent ependymoma in children is poorly defined, and no data are available in literature on response to chemotherapy in adults. The aim of the present study was therefore to investigate outcomes following chemotherapy in this setting. A retrospective review was made of the charts of 28 adults with ependymomas treated between 1993 and 2003 in five institutions of the GICNO network in order to analyze the response rate and TTP following chemotherapy delivered for recurrence after surgery and radiotherapy. Thirteen patients (46.3%) were treated with cisplatin-based chemotherapy and 15 with a nitrosourea-based regimen. Platinum-based chemotherapy was followed by two CRs (15.4%) and two PRs (15.4%), while seven patients (53.8%) remained stable. Nitrosourea-based regimens resulted in no CR and two PRs (13.3%), while 11 patients (73.3%) remained stable. The overall median time to progression was 9.9 months (CI 95%, 7.5–21.7), and 71% (CI 95%, 56%–90%) and 44.3% (CI 95%, 29%–68%) of the patients were progression free at 6 and at 12 months, respectively. There were no differences between platinum-based and nitrosourea-based regimens for progression-free survival. The overall median survival was 40.7 months (CI 95%, 16–NA); 57% (CI 95%, 40.5%–80.1%) of the patients were alive at two years. Cisplatin-based chemotherapy is followed by a higher response rate, although this does not translate into a prolonged time to progression or survival, and the toxicity is greater. More active regimens still need to be found for the treatment of progressive ependymomas.


Steven Brem, Aileen Staller, Marguerite Wotoczek-Obadia, John Robb, Frank D. Vrionis, James L. Pearlman, Amyn Rojiani, and Susan Snodgrass; Neuro-Oncology Program, H. Lee Moffitt Cancer Center & Research Institute, Department of Interdisciplinary Oncology, University of South Florida, College of Medicine, Tampa, Florida, USA

Local control and tumor recurrences remain clinical problems despite dramatic improvements in image-guided surgery/radiosurgery and radiation therapy for the treatment of cerebral metastases. We report our experience in a series of 42 consecutive patients with metastases to the brain treated with implantable BCNU wafers. Patients treated from 1998 to 2004 were treated at the time of presentation with CNS metastases using surgical removal and WBRT for metastases to the brain. Patients had stable systemic disease and a life expectancy of at least 12 weeks. We used minimally invasive, image-guided, surgical neuronavigation. The primary pathology was non-small cell lung cancer (n = 20, 47.6%), melanoma (n = 11, 26.2%), renal (n = 4, 9.5%), breast (n = 3, 7.1%), sarcoma (n = 2, 4.8%), colon (n = 1, 2.4%), and ovarian (n = 1, 2.4%). Patients were followed every 2–3 months with serial MRIs. The average age was 55.4 years (range, 29–74). There were 23 males and 19 females. Thirty-four patients were newly diagnosed, and eight patients were treated at time of recurrence. Newly diagnosed patients were treated with 3000 to 4400 cGy to the whole brain. The treatment was well tolerated. Follow-up ranged from 2 to 67 months. The median length of stay after surgery was one day. There were no local recurrences in any newly diagnosed patients. Three patients developed CNS recurrence at a distant site. The mean survival for 41 patients (1 lost to follow up) was 16.8 (± 2.8) months. Survival was higher for newly diagnosed patients (17.8 ± 3.4 months) compared to patients treated at recurrence (12.9 ± 2.9 months). Twenty-three of the 41 patients died due to systemic disease, 1 due to leptomeningeal carcinomatosis, and 2 due to neurological deterioration. Six patients with early deaths (<3 months) were due to systemic (non-neurological) medical illness. Nine patients are still alive with no evidence of recurrent CNS disease. We conclude that the combination of surgery, BCNU wafers, and WBRT provided local control of newly diagnosed metastatic brain cancer. These results represent an improvement from prior published experience with surgery, radiation therapy, or the combination of the two modalities.


Hannah R. Briemberg,1 Paul G. Richardson,2 Patrick Wen,2 Dixie-Lee Esseltine,3 Ken Anderson,2 and Anthony A. Amato1; 1Brigham and Women’s Hospital, Boston, Massachusetts; 2Dana Farber Cancer Institute, Boston, Massachusetts; 3Millenium Pharmaceuticals, Boston, Massachusetts, USA

Bortezomib (PS-341 or VELCADE) has demonstrated efficacy in the treatment of refractory multiple myeloma in phase II trials and recently received FDA approval for use in the treatment of this population of patients. It acts primarily through proteasome inhibition, although it is believed to have other associated mechanisms of action that may play a role in its efficacy in multiple myeloma as well as its associated side effect profile. It appears to be fairly well-tolerated with dose-limiting neurotoxicity as its most significant side-effect. Development or progression of peripheral neuropathy was the most common neurologic event noted in clinical trials. To determine the incidence, severity, and character of the neuropathy, 193 of 202 patients enrolled in a Phase II clinical trial underwent neurological evaluation at baseline and at repeat visits throughout the duration of the trial. This included a neurological exam performed by a neurologist and/or a patient-completed questionnaire designed to assess symptoms of neurological toxicity completed by 191 patients. A subgroup of patients (12 total) also underwent nerve conduction studies and quantitative sensory testing. During the course of the trial, 63 patients developed symptomatic new onset peripheral neuropathy or worsening of existing peripheral neuropathy. Symptoms were primarily sensory, consisting of pain, burning, paresthesias, and sensory loss. Of these, 10 patients (5% of the total population) discontinued the study drug because of peripheral neuropathy and 30 patients (15% of the total population) required at least one dose adjustment of bortezomib because of symptoms of peripheral neuropathy. An additional 23 patients (12%) required treatment for neuropathy symptoms but did not require changes in bortezomib dosing. The incidence of symptomatic treatment-emergent neuropathy was increased in patients who had neuropathy at baseline compared with those who did not. Of those who required drug discontinuation, dose-adjustment and/or treatment for neuropathy symptoms, 4 (13% of patients with no neuropathy at study entry) had new-onset neuropathy and 59 (35% of patients with neuropathy at study entry) had worsening of pre-existing neuropathy. All of the patients who required discontinuation of bortezomib had clinical evidence of neuropathy at study entry. The clinical features of bortezomib-induced neuropathy are strongly suggestive of a length-dependent sensory axonal polyneuropathy with predominant small fiber involvement.


Nicholas Butowski, Kathleen Lamborn, Mitchel Berger, Michael McDermott, Sandeep Kunwar, Michael Prados, and Susan Chang; University of California at San Francisco, San Francisco, California, USA

Efforts are ongoing to develop more effective treatments for patients with GBM to improve overall survival or prolong time to progression. Such efforts include surgically based investigational approaches that require GTR for optimization of treatment. Examples are placement of chemotherapy wafers, convection-enhanced delivery of agents, gene therapy, and immunotherapy. For appropriate comparison in single arm phase II studies, a historical control group is required. Recursive partitioning analysis of large numbers of patients with GBM have created defined risk groups, and results from these analyses have often been used as the historical comparator. These comparisons include all patients irrespective of the initial extent of resection. Since patients who are able to undergo a greater extent of resection may have a better prognosis, this comparison may not be accurate. Therefore, to provide a more appropriate comparator, we sought to determine the survival outcome of a homogenous group of patients with newly diagnosed GBM who were treated uniformly on prospective clinical trials and had GTR of the enhancing disease. Of 893 patients with GBM, 153 had a GTR, defined as >90% of the enhancing mass on the basis of the postoperative images that were reviewed by the Neuro-Oncology Group. All protocols included provision for external-beam radiotherapy. Some included a temporary radioactive seed boost and/or adjuvant chemotherapy. Of the GTR group, there were 59% males. Median age was 54 years (range, 25–77). Median KPS was 90 (range, 60–100). KPS was ≥90 for 82% of the patients. Of the 153 GTR patients, 137 have died. Median survival was 71 weeks (95% CI, 65–76). After adjustment for age and KPS, those who had a GTR had an improved survival compared to those with less than a GTR (P = 0.01; hazard ratio = 0.78; CI, 0.65–0.94). Considering only patients with GTR, age at diagnosis continued to be a statistically significant prognostic factor. However, KPS was not, probably because most patients with GTR had high KPS. Analysis is ongoing to determine if there are other patient characteristics that can better divide these patient groups in terms of predicting survival time. This information may be helpful in future designs of phase II surgically based studies for newly diagnosed glioblastoma multiforme.


Gregory Cairncross, Wendy Seiferheld, Edward Shaw, Robert Jenkins, Bernd Scheithauer, David Brachman, Jan Buckner, Karen Fink, Luis Souhami, and Walter Curran; Radiation Therapy Oncology Group, Philadelphia, Pennsylvania, USA; Southwest Oncology Group, San Antonio, Texas, USA; North Central Cancer Treatment Group, Rochester, Minnesota, USA; Eastern Cooperative Oncology Group, Philadelphia, Pennsylvania, USA; and the National Cancer Institute of Canada’s Clinical Trials Group, Toronto, Ontario, Canada

Anaplastic oligodendrogliomas (AOs) and anaplastic oligoastrocytomas (AOAs) are treated with surgery and RT at diagnosis. They also respond to procarbazine, lomustine, and vincristine (PCV), raising the possibility that PCV plus RT at diagnosis may improve outcome. Furthermore, for AOs, response to PCV and long survival have been associated with 1p and 19q allelic loss. An RCT was conducted to test whether dose-intense, pre-RT PCV prolongs overall survival (primary end point) or progression-free survival (secondary end point) versus RT alone. Serious toxicity rates and quality of life were other end points. Patients with AOs or AOAs confirmed by central review who were age >18 years, had a Karnofsky performance score (KPS) >60- and consented were study-eligible. Tumor sections and peripheral blood were also collected. Two hundred ninety-one eligible patients were randomized; 60% were male; 68% were <age 50, 88% had a resection, 90% had a KPS >80, and 70% had an AO. One hundred forty-eight patients had PCV plus RT, and 143 had RT alone; the study arms were balanced for prognostic factors. Median survival times were similar for both groups; 4.8 years for PCV plus RT and 4.5 years for RT alone (HR 1.04; 95% CI, 0.74–1.45; P = 0.830). Progression-free survival tended to be longer after combined treatment, 2.6 years for PCV plus RT versus 1.9 years for RT alone (HR 1.34; 95% CI, 1.00–1.80; P = 0.053). During PCV, 95 patients experienced grade 3 or 4 toxicity, and one died. Grade 3 and 4 RT toxicities were similarly infrequent in both groups. Tissue for molecular analysis was available on 206 tumors; 92 (46%) had combined 1p and 19q loss. Irrespective of treatment, patients whose tumors lacked 1p and 19q lived longer than other patients (median survival not reached vs. 2.8 years [HR 0.31; 95% CI, 0.20–0.48; P < 0.001). Longer follow-up is needed to ascertain treatment-specific outcomes for cases with 1p and 19q loss. Pre-RT PCV does not impart a survival advantage for histologically defined AOs and AOAs, but may prolong progression-free survival at the expense of greater acute toxicity. Allelic loss of 1p and 19q is a powerful prognostic marker in oligodendroglial tumors.


H. Carraway, L.J. Swinnen, C. Rankin, K.S. Albain, J. Townsend, and D. Blumenthal; Johns Hopkins Cancer Center, Baltimore, Maryland; Southwest Oncology Group, San Antonio, Texas; USA

Drug resistance is a major problem in the management of gliomas. Pre-clinical studies showed that inhibition of the DNA excision-repair system modulates resistance to platinum compounds in vitro. Cytosine arabinoside (Ara-C) and hydroxyurea (HU), in combination, inhibit the removal of platinum DNA adducts, and marked cytotoxic synergy has been demonstrated in very platinum-resistant HT29 colon carcinoma cells. The HU (1 mM) and Ara-C (1 μM) drug levels needed to achieve this have been shown to be achievable in clinical pilot studies of the three-drug regimen. Therapeutic options for patients with relapsed or progressive grade 3 or 4 glioma are limited. The primary end point of the current study was six-month survival, historically about 42% in such patients. Patients had histologically confirmed Grade 3 (anaplastic astrocytoma) or 4 (glioblastoma multiforme), relapsed or progressive, as documented by CT or MRI after prior surgery and/or radiotherapy. Up to four cycles of one prior regimen not containing cisplatin were allowed. Stable steroid dose prior to imaging was required. Patients were stratified by histology: AA or GBM. Patients received a loading dose of HU 1260 mg/m2 IV over 1 h followed by Ara-C 1200 mg/m2 plus HU 5040 mg/m2 IV over 12 h, followed by cisplatin 100 mg/m2 IV over 1 h. Allopurinol, mannitol, and aggressive hydration were used, as well as dexamethasone to prevent cerebral edema. A total of 76 patients were registered. The AA stratum was closed early after 20 patients because of slow accrual. The GBM stratum registered 56 patients in a two-stage accrual. Anaplastic Astrocytoma: Among the 19 eligible patients, the six-month survival probability was 58% (95% CI, 36%–80%). No responses were seen. Six patients (32%) had stable disease (95% CI, 13%–57%). Eleven patients progressed, and two were not assessable for response. All AA patients have progressed or died, with a median progression-free survival of 3 months (95% CI, 2–5 months). The six-month progression-free survival probability was 26% (95% CI, 6%–46%). Of the 19 eligible patients, 18 have died, with a median overall survival of 7 months (95% CI, 6–14 months). Glioblastoma Multiforme: The six-month survival probability was 41% (95% CI, 28%–55%). Of the 51 eligible GBM patients, one achieved a partial response (2%; 95% CI, 0%–10%). Thirteen patients (25%) had stable disease (95% CI, 14%–39%). Twenty-two patients progressed, and 14 were not assessable for response. Fifty of the 51 patients have died, with a median overall survival of 5 months (95% CI, 4–6 months). All GBM patients have progressed or died, with a median progression-free survival of 2 months (95% CI, 2–4 months). The six-month progression-free survival probability was 25% (95% CI, 14%–37%). Of the 57 eligible patients evaluated for toxicity, two died from infection. Twenty-one additional patients (37%) have experienced Grade 4 toxicities, primarily hematologic. Five of these patients (9%) had Grade 4 non-hematologic toxicities. There was one major protocol deviation due to lack of required dose reduction. Cisplatin, when combined with HU and Ara-C as DNA excision-repair inhibitors, did not improve the six-month survival rate in patients with relapsed or progressive AA or GBM. Significantly more hematologic toxicity was seen than expected from cisplatin alone.


Robert Cavaliere,1 Elana Farace,1 Patrick Wen,2 and David Schiff1; 1Neuro-Oncology Center, University of Virginia, Charlottesville, Virginia; 2Department of Medical Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts; USA

The incidence of thromboembolic events (TEEs), including deep vein thrombosis (DVT) and pulmonary embolus (PE), is increased in patients with high-grade gliomas (HGGs) compared to the general population. Thalidomide, a drug with presumed anti-angiogenic properties, is increasingly being used to treat multiple cancer types including HGGs. Recent reports suggest that thalidomide is associated with a higher rate of TEE, most notably among patients with multiple myeloma. The purpose of this report is to review a prospectively maintained neuro-oncology database and explore the relationship between thalidomide and TEE in patients with HGG. The Neuro-oncology database at the University of Virginia was queried for patients with HGGs and TEE. Basic patient demographics and data regarding survival, tumor pathology, and chemotherapeutics were abstracted. A TEE was defined as a radiographically confirmed DVT or PE. Only patients with at least one outpatient clinic visit were included in the analysis. Cox proportion hazard model was used to compare the incidence of TEE among patients treated with chemotherapy, either alone or with thalidomide, and patients not treated with any chemotherapy after adjusting for age and tumor grade. Sixty-three patients with HGGs were treated with chemotherapy, including 25 with thalidomide and 38 with regimens that did not include thalidomide. An additional 27 patients with HGG did not receive any chemotherapy. Twenty-three TEEs were recorded, nine in patients while receiving thalidomide for a median of 43 days (7 to 147) at the time of their event, six in patients while receiving chemotherapy that did not include thalidomide, and eight patients who did not receive any chemotherapy. Among patients treated with chemotherapy, the use of thalidomide was associated with a significantly higher risk of TEE (P = 0.045). On multivariate analysis, patients receiving chemotherapy, with or without thalidomide, were more likely to have had a TEE than patients who did not receive chemotherapy (P < 0.0001). However, among those who received chemotherapy, the addition of thalidomide was not associated with a higher risk of TEE (P = 0.312). Tumor grade was a significant predictor of TEE (P = 0.007). Patients with grade IV lesions (N = 58) were more likely to have an event than patients with a grade III tumor (N = 28). Age was not associated with a higher risk of TEE (P = 484). We conclude that the use of thalidomide may be associated with a higher risk of TEE. However, when accounting for tumor grade, the addition of thalidomide to other therapies does not appear to increase the risk of TEE in patients with HGG above the heightened TEE risk conveyed by treatment with chemotherapy.


A. Chakravarti, W. Seiferheld, H.I. Robins, A. Guha, D. Brachman, W. Curran, A. Choucair, and M. Mehta; Massachusetts General Hospital, Boston, Massachusetts, USA; American College of Radiology, Philadelphia, Pennsylvania, USA; University of Wisconsin, Madison, Wisconsin, USA; University of Toronto, Toronto, Ontario, Canada; Foundation for Cancer Research, Phoenix, Arizona, USA; Thomas Jefferson University, Philadelphia, Pennsylvania, USA; LDS Hospital, Salt Lake City, Utah, USA; University of Wisconsin, Madison, Wisconsin, USA

The Radiation Therapy Oncology Group (RTOG) initiated a Phase I/II study of the EGFR tyrosine kinase inhibitor gefitinib, in combination with radiation for newly diagnosed GBM patients. This report is an update of Phase I toxicity data. Patients entered on RTOG 0211 were stratified according to whether they were on enzyme-inducing anticonvulsant drugs (EIACDs: Group I) or not (non-EIACDs: Group II). Gefitinib dose escalation during radiation was from 250 to 750 mg in Group I and 250 to 500 mg in Group II in 250-mg increments. A standard 3+3 design was used to evaluate dose-limiting toxicity (DLTs), which were defined as designated acute (< 90 days) events. A total of 18 patients in Group I (16 eligible and analyzable) and 13 patients in Group II (12 eligible and analyzable) were included in this update. In Group I, doses up to 750 mg were not associated with DLTs. Two patients experienced late (>90 days) grade 3 DLTs in Group I (diarrhea and elevated SGOT values). In Group II, one patient experienced a Grade 3 elevation of SGOT at the 250-mg dose level within 90 days. At the 500-mg dose level in Group II, one patient had late Grade 3 elevation of SGOT and one patient had acute Grade 4 elevation of SGOT. Other observed side effects did not fit the definition of DLTs. Gefitinib (ZD1839) appears to be well tolerated when combined with radiation at doses up to 500 mg in the non-EIACD group and doses up to 750 mg in the EIACD group. RTOG 0211 has met its accrual objectives and has closed. Further information about the protocol can be viewed at


Marc C. Chamberlain1 and Michael J. Glantz2; 1University of Southern California, Norris Cancer Center, Los Angeles, California; 2University of Massachusetts, Worchester, Massachusetts; USA

Intracranial meningiomas are common and comprise 20% of all primary brain tumors. Meningiomas infrequently metastasize. However, there is limited data regarding CSF metastasis. Patient and methods information for this study is as follows: Eight of 200 (4%) consecutively seen meningiomas manifested CSF dissemination. CSF cytology was positive in eight, and neuro-radiography was consistent with CSF dissemination in seven. Patients (6 women, 2 men) ranged in age from 24 to 87 years (mean 52). All patients had prior surgery (range, 1–5; median 2), radiotherapy (involved field in 7, stereotactic in 6), and chemotherapy (hydroxyurea in 8). Concomitant metastases were seen in all patients and were both intraneural (subarachnoid or ventricular tumor; intracranial in 8; spinal cord in 4) and extraneural (subcutaneous, cervical lymph nodes, orbit or pulmonary in 5). Treatment utilized both systemic (temozolomide in 4, CPT-11 in 3, hydroxyurea in 3, a-interferon in 2) and intraventricular (liposomal ara-C in 7, thiotepa in 1, busulfan in 1) chemotherapy. We observed treatment-related toxicity in 8 patients, including chemical meningitis (8, grade 2). Best response comprised stable disease in 7 and progressive disease in 1. Response duration ranged from 0.5 to 31 months (median, 3.5 months). Median survival was 5.5 months with 3 patients alive and with disease. We conclude that treatment of CSF-disseminated meningioma although feasible and comparatively non-toxic is associated with modest outcomes despite combined systemic and intraventricular chemotherapy.


Asha Das,1 Brian Pikul,1 Xuemo Fan,2 Marcel Maya,3 and Keith L Black1; 1Maxine Dunitz Neurosurgical Institute and Departments of 2Pathology and 3Radiology, Cedars-Sinai Medical Center, Los Angeles, California, USA

Multiple adjacent primary intracranial tumors of different histological subtypes are rare. Most commonly, multiple primary tumors of the central nervous system occur in the setting of previous cranial radiation or neuro-cutaneous syndromes, such as neurofibromatosis type 1. We describe the clinical features and radiological findings of an extremely rare case of meningioma with concurrently occurring adjacent anaplastic astrocytoma. The patient is a 53-year-old woman who presented with a generalized tonic-clonic seizure. Her past medical history was unremarkable. Neuroimaging studies including an MRI scan of the brain revealed a left frontoparietal enhancing dural-based 1-cm mass with marked surrounding edema measuring approximately 8 cm. The patient underwent a left frontal craniotomy with resection of a left frontal WHO grade I meningioma. A left peritumoral brain biopsy revealed a WHO grade III anaplastic astrocytoma. Moderate cellularity and nuclear pleomorphism with scattered mitoses were observed. The difficulty of managing patients with multiple primary brain tumors begins with obtaining an early diagnosis. Multiple primary intracranial tumors must be considered in the setting of unexplained signs and symptoms, unexpected radiological findings, or intraoperative observations. Careful attention should be paid to peritumoral edema surrounding a meningioma because if edema increases a definitive diagnosis is imperative. The origin of multiple primary intracranial tumors of different histological types is unclear. Genetic factors or locally acting oncogenic factors may play a role. In addition, either the meningioma or anaplastic astrocytoma may stimulate the adjacent brain parenchyma or arachnoid cells into neoplastic proliferation. This stimulation may be caused by a paracrine mechanism that secretes growth factors into the surrounding medium and transforms adjacent cells.


Asha Das, Jennifer Learn, Minh-Thu Le, John Yu, Brian Pikul, and Keith Black; Maxine Dunitz Neurosurgical Institute, Cedars-Sinai Medical Center, Los Angeles, California, USA

Gliomas release excessive amounts of glutamate, leading to aggressive tumor growth. In animal models, gliomas that actively released glutamate were larger and were associated with a shorter survival than those that showed enhanced uptake of glutamate. As a neurotoxin, glutamate destroys neighboring normal cells and thereby facilitates tumor growth. In these animal models, treatment with N-methyl-d-aspartate (NMDA) receptor antagonists slowed the growth of glutamate-secreting tumors in situ. We extended these observations to the clinical setting by using an NMDA antagonist, riluzole, to treat patients with recurrent or progressive AA or GBM. Riluzole mediates its effects by blocking glutamate transmission through NMDA receptor-linked processes, stabilizes sodium channels, and blocks gamma-aminobutyric acid reuptake. Our primary objective was to establish the safety and toxicity of riluzole in patients with recurrent or progressive AA and GBM. The secondary objectives were to obtain preliminary evidence of anti-tumor activity of riluzole in patients with recurrent or progressive AA and GBM based on neurological exam, Karnofsky performance score, mini-mental status exam, neuropsychologic evaluation, and MRI scans of the brain with and without gadolinium. In available tumor samples, assays were performed to detect glutamate release in brain tumor samples. Patients older than 18 years with histologically confirmed AA or GBM with recurrent or progressive disease and Karnofsky performance status >40 were included. Baseline laboratory parameters, neurologic exam, MRI scans of the brain, neuropsychologic evaluation, and MR spectroscopy were performed. In those patients with available tumor tissue, glutamate release was determined by using imaging with bioluminescence. The treatment protocol included riluzole 50 mg twice daily. Patients were reassessed monthly, and toxicities to treatment were recorded by using the NCI common toxicity criteria. If toxicity was experienced, the dose was modified accordingly. If tumor progression was observed, the patient was removed from the study. Those patients with stable disease or partial or complete responses continued the study medications until tumor progression or dose-limiting toxicities occurred. Our preliminary data of eight assessable patients included three patients with recurrent AA and five with recurrent GBM. Their median age was 51.3 years (range, 24–79 years). No grade 3 or 4 toxicities were observed. The median time to progression in patients and overall survival requires further monitoring. Preliminary results suggest that riluzole is well tolerated in this patient population and may potentially have a role as an adjuvant to other therapies, including chemotherapy. Further enrollment in this phase I study continues.


Annick Desjardins, Jennifer Quinn, David Reardon, Jeremy Rich, James Vredenburgh, Allan Friedman, Sandra Tourt-Uhlig, Michael Badruddoja, Chris Efird, and Henry Friedman; Duke University Medical Center, Durham, North Carolina, USA

With therapy, the prognosis for malignant gliomas remains poor. We performed a phase II study to determine the activity and toxicity of thalidomide plus daily cyclophosphamide in patients with recurrent anaplastic astrocytomas (AA) and anaplastic oligodendrogliomas (AO) (Grade 3 malignant gliomas). Eligibility included adult patients with recurrent AA or AO in first relapse with measurable disease, interval of at least 3 weeks between surgical resection or 6 weeks between radiotherapy or chemotherapy, Karnofsky score ≥60%, negative pregnancy test, and adequate marrow, renal, and hepatic function. Thalidomide was begun at 100 mg qHS and increased by 100 mg weekly to a maximum dose of 400 mg. Cyclophosphamide was administered at 100 mg daily. Responses were assessed every eight weeks, and patients remained on study for up to one year or until disease progression or unacceptable toxicity. To date, 16 patients with recurrent AA (n = 15) or AO (n = 1) have been enrolled, of a planned accrual of 30 patients, with a median age of 38 years (range, 25–72). All patients were previously treated with radiation therapy and chemotherapy (median prior chemotherapy regime, 1; range, 1–3). Grade 3 or greater toxicities included only neutropenia (grade 3, n = 3; grade 4, n = 2) and hyponatremia (grade 3, n = 1). Responses include one complete and two partial responses. Two responding patients completed one year of treatment and remain off therapy with no evidence of disease progression, while one additional responding patient continues on week 41 of therapy. Eleven patients progressed after cycle 1. It is too early to evaluate one patient, and one patient was not assessable. We conclude that thalidomide plus daily cyclophosphamide demonstrates encouraging activity, is of convenient administration, and is well tolerated among patients with recurrent AA or AO.


Annick Desjardins, David Reardon, James Vredenburgh, Jeremy Rich, Sridharan Gururangan, Allan Friedman, John Sampson, James Provenzale, Roger McLendon, James Herndon II, Michael Badruddoja, Julie Amara, Sandra Tourt-Uhlig, Herbert Newton, Henry S. Friedman, and Jennifer A. Quinn; Ohio State University Medical Center, Columbus, Ohio; Duke University Medical Center, Durham, North Carolina; USA

Preclinical studies have demonstrated that the combination of Temodar plus CPT-11 displays a schedule-dependent enhancement of antitumor activity secondary to the trapping of topoisomerase I by O6-methylguanine residues in DNA. We are conducting a phase I trial of Temodar plus CPT-11 to define the maximum tolerated dose (MTD) of CPT-11 and toxicity of this drug combination in the treatment of patients with recurrent or progressive malignant gliomas. Patients are administered oral Temodar at a dose of 200 mg/m2 on days 1–5, and intravenous CPT-11 over 90 min on days 1, 8, 22, and 29 of a 6-week cycle. On day 1, Temodar is given prior to the CPT-11. The CPT-11 dose is escalated in cohorts of 3 to 6 patients with two independent dosing schedules based on concurrent enzyme-inducing antiepileptic drug (EIAED) use. Radiographic response criteria are utilized to evaluate activity using T1-weighted, enhanced MRI images. One hundred three patients have been treated to date, 84 with glioblastoma multiforme (GBM), 16 with anaplastic astrocytoma (AA), 1 with anaplastic oligoden-droglioma, and 2 with high-grade glioma NOS. Sixty-four patients have been accrued to the EIAED stratum at CPT-11 doses of 40, 60, 80, 100, 125, 150, 175, 200, 225, 275, 325, and 375 mg/m2. Thirty-nine patients have been accrued to the non-EAIED stratum at CPT-11 doses of 40, 60, 80, 100, 125, and 150 mg/m2. Dose-limiting toxicities observed thus far have been limited to three episodes of grade 4 thrombocytopenia and one episode of grade 3 thrombocytopenia, three episodes of grade 4 leukopenia, three episodes of grade 4 neutropenia, one episode of grade 4 anemia, one episode of grade 3 ALT, and two episodes of grade 3 dehydration. Responses include three complete and nine partial responses, and stable disease in 45 patients. Twenty-eight patients progressed after cycle 1. Five patients are not assessable for response, and it is too early for an evaluation of 13 patients. The MTD of CPT-11 for this drug combination has been determined for the non-EIAED stratum at 125 mg/m2 and has yet to be identified for the EIAED stratum.


C.E. Fadul, A.L. Kominsky, L.P. Meyer, L.S. Kingman, W.B. Kinlaw, C.H. Rhodes, C.J. Eskey, S.E. Pepin, and N.E. Simmons; Neuro-Oncology Program, Norris Cotton Cancer Center, Dartmouth Hitchcock Medical Center, Lebanon, New Hampshire, USA

The hallmark of pituitary carcinoma is the presence of nervous system or extraneural non-contiguous tissue metastases from an invasive pituitary adenoma. Unlike typical metastatic malignancy, the secondary seeding frequently maintains the cytological characteristics of the original tissue. Although the latency between the appearance of invasive adenoma and the development of pituitary carcinoma is on average several years, once carcinoma is diagnosed, average survival is extremely short despite aggressive therapy. We report two patients with pituitary carcinoma who responded to temozolomide following disease progression after multi-modality treatment. The first patient developed visual symptoms and headaches after 5 years of receiving dopamine-agonists for a prolactin-secreting adenoma. After surgical excision, the patient experienced symptomatic metastases, confirmed by biopsy to be of pituitary origin, to the cervical and thoracic spine. Despite surgical excision, radiation therapy, octreotide, bromocriptine, multiple-drug chemotherapy, and metaiodobenzylguanidine (MIBG) I-131, his disease progressed with neck pain requiring high dose of opioids and increasing levels of prolactin. The second patient was diagnosed with having a non-hormone-secreting pituitary adenoma after the onset of headaches and visual field loss. Over the next 6 years, multiple recurrences led to therapy with several surgeries, fractionated external beam radiation, and gamma knife therapy. He then experienced progressive localized neck pain with radicular extension into the upper extremities, as well as progression of bilateral temporal hemianopsia. MRI revealed progression of pituitary mass and located multiple spinal drop metastases in the cervical and thoracic spine. Visual fields confirmed the visual defect. Biopsy of cervical spine mass revealed pituitary carcinoma. In spite of radiation, cervical pain and radicular symptoms persisted. A bone scan revealed lesions in the ribs and the right tibia. Both patients were treated with monthly temozolomide at a dose of 200 mg/m2/d administered orally on the first 5 days of a 28-day cycle. They were followed for side effects and response to therapy. The first patient received 10 cycles, and therapy was stopped because of severe persistent fatigue despite dose reduction. Pain disappeared after several cycles and the opioids were tapered off. His pre-treatment prolactin level was 694 ng/ml (normal range, 4.1 to 18.4 ng/ml), and after completion of treatment it decreased to 52 ng/ml. The second patient received 12 cycles without any significant side effects except for grade 2 lymphopenia. Neck pain disappeared, visual fields improved, and MRI revealed decreased size of both the pituitary tumors and the spine lesions, although they persist. Both patients continue to have sustained partial responses and having full-time employment for more than a year after temozolomide was stopped. We report a rare example of persistent response to cytotoxic chemotherapy in two patients with pituitary carcinoma. Pituitary carcinoma may be a chemosensitive tumor.


J.B. Fiveash, B. Guthrie, J.M. Markert, S.A. Spencer, S.R. Rosenfeld, S.M. Russo, and L.B. Nabors; University of Alabama-Birmingham, Birmingham, Alabama, USA

Because of potential toxicity concerns associated with adjuvant whole-brain radiation therapy (WBRT), we have explored temozolomide (Tem) chemotherapy as an adjuvant to gamma knife radiosurgery (GKRS) without WBRT in order to prevent or delay the development of additional (new) brain metastases. Since no prior prospective study has reported the toxicity of combined radiosurgery and Tem, this initial analysis was performed to confirm the safety of the combined regimen. Thirteen patients with one to four brain metastases have been enrolled (August 2002–April 2004) in a phase II clinical trial. Two patients received radiosurgery but no temozolomide (financial issues and CNS progression at the time of radiosurgery MRI requiring WBRT). Patients received 15 to 24 Gy (median 18 Gy) generally to the 50% isodose line followed by monthly cycles of Tem given 200 mg/m2/day × 5 days (no prior chemotherapy) or 150 mg/m2/day × 5 days (prior chemotherapy). Tem was started within two weeks of GKRS. Twelve cycles of Tem were planned. Other cytotoxic systemic therapy was not administered concurrently with Tem. Patients were followed with clinical and radiographic examination at one month, three months, and every three months. Tem was discontinued for new brain tumors, systemic tumor progression requiring other chemotherapy, or toxicity. Of the 11 patients that actually received Tem, the mean number of tumors treated per patient with radiosurgery was 2.1 (range, 1–4). Follow-up was for a mean 6.6 months (range, 1–15 months). Forty-three cycles have been completed to date, including one patient that received 10 cycles. The mean number of cycles completed by patients that had discontinued Tem was 3.5. Tem was discontinued because of systemic progression in five patients and toxicity (fatigue) in one patient. Grade 3 or greater hematologic toxicity occurred in one patient. Two patients died of causes thought to be unrelated to therapy (acute MI and ruptured diverticulum). Five patients required hospitalization (thromboembolism in two patients, locally recurrent brain tumor in two patients, and ruptured diverticula in one patient). One patient required resection of recurrent local tumor that had been treated with radiosurgery, but no patient required hospitalization for radiation necrosis. The combination treatment of GKRS and Tem has not resulted in unacceptable toxicity. Accrual to this clinical trial continues to determine if Tem can delay progression of potential microscopic foci to macroscopic brain metastases in patients receiving GKRS without WBRT. This work was supported in part by a research grant from Integrated Therapeutics–Schering-Plough


A. Flowers and S. Nerenstone; Hartford Hospital, Hartford, Connecticut, USA

Prognosis for patients with breast carcinoma and multiple brain metastases (BMs) is poor, with an average survival of 6 months. Palliative whole-brain radiation therapy (WBRT) remains the standard treatment for these patients. Short time interval from diagnosis to BM, presence of systemic metastases, and leptomeningeal involvement are poor prognostic indicators. We present the case of a patient with multiple BM and poor prognostic factors who was treated with multimodality therapy and has prolonged survival. Our patient is a 40-year-old lady, diagnosed in 2000 with breast carcinoma, treated with chemotherapy, mastectomy, and local radiation therapy (RT). In 2001 she developed bone and liver metastases and was treated with RT to the lumbar spine and hips, and Taxol and Herceptin. In July 2003 she presented with vertigo, decreased vision in right eye, right-hand weakness, word-finding difficulty, memory problems, and unsteady gait. CT and MRI of the brain showed multiple (>140) supra and infratentorial BM “too numerous to count,” ranging in size from punctate to 1 cm. There was some leptomeningeal involvement in the brain but not the spine. She received WBRT and improved clinically. She received 9 cycles of chemotherapy with cyclophosphamide, methotrexate and 5-FU for the BM and systemic metastases and showed a response. She was followed up on Herceptin until March 2003, when she developed seizures, and MRI showed progression of BM. The patient received a WBRT boost and then started chemotherapy with capecytabine. She improved clinically and radiologically, was able to taper off steroids, and resumed a normal life. She is now 22 months from the diagnosis of BM. Multimodality therapy may benefit patients with multiple BM from breast carcinoma and may help prolong survival in this group of patients with poor prognosis.


E. Franceschi, A.M.P. Omuro, C. Nolan, A. Lassman, A. Demopoulos, and L.E. Abrey; Memorial Sloan-Kettering Cancer Center, New York, New York, USA

TMZ is often used as adjuvant therapy for newly diagnosed patients with glioma. Because of potential hematologic complications, it is usually discontinued after 12 to 18 cycles, even in responders. Subsequent salvage therapies have limited efficacy at recurrence. We assessed the outcome and complications of reusing TMZ at relapse in prior responders. We conducted a retrospective review of patients with recurrent/progressive glioma who had a history of response to TMZ and were re-treated with the same agent at recurrence. Nine patients were identified (5 men, 4 women). Median age at diagnosis was 57 (25–66). Eight patients had glioblastoma (GBM) and one had low-grade oligodendroglioma. All patients with GBM were initially treated with surgery and radiotherapy. TMZ was given as adjuvant therapy for six patients and for first recurrence after BCNU to two. The patient with low-grade oligodendroglioma was treated with surgery and received TMZ for first recurrence. For the initial exposure to TMZ, eight patients received the conventional regimen with 150–200 mg/m2/d for 5 days, every 28 days for a median of 13 (11–28) cycles; one patient received 10 cycles of 75 mg/m2/d for 6 weeks every 8 weeks. No patient progressed on initial TMZ treatment, and median progression-free survival was 21 (13–45) months. At recurrence, all patients were restarted on the conventional schedule of TMZ. Median interval between last day of initial treatment with TMZ and start of retreatment was 7 months (4–26). One patient had grade 3 neutropenia during retreatment. Six patients are assessable for response; a CR was achieved in two patients, PR in one, SD in two, and one PD. Five patients continue on TMZ at last follow-up. In this small series of patients, retreatment with TMZ was an effective and safe option for patients with recurrent gliomas who previously responded to TMZ.


A. Freeman, J.M. Gomori, E. Linetsky, Z. Zakay-Rones, A. Panet, E. Libson, C.S. Irving, E. Galun, and T. Siegal; Goldyne Savad Institute of Gene Therapy and the Leslie and Michael Gaffin Center for Neuro-Oncology, Hadassah Hebrew University Hospital, Jerusalem; Department of Virology, Hebrew University, Jerusalem; OVCure (Israel) Ltd., Jerusalem; Israel

Newcastle disease virus (NDV) is a negative-sense, single stranded, RNA virus belonging to the Paramyxoviridae family. It produces disease in some species of birds, but rarely causes minor illness in humans. NDV has been evaluated as a potential oncolytic therapeutic agent for a variety of tumors, but its effect in patients with GBM has not been systematically studied. OV001 (NDV HUJ) is an oncolytic and lentogenic, wild-type strain of NDV. NDV is cytopathic on several tumor cell lines and is suitable for intravenous administration because of a lack of pre-existing immunity, relative ease to manufacture, and high level of environmental safety. Following preclinical repetitive dosing toxicity studies, a Phase I/II trial was initiated in GBM. Our objectives were to determine the toxicity and the MTD of intravenously administered OV001 and to obtain initial efficacy information based on tumor imaging. Fourteen patients (age, 11–58 years, KPS, 50%–90%) with GBM were enrolled after they failed conventional therapy. In the first part of the study an intra-patient dose escalation method was used with one cycle dosage steps of 0.1, 0.5, 1, 5, and 10 billion infectious units (BIU) of OV001 (1 BIU = 1 × 109 EID50, 50% egg infectious dose). One cycle consisted of 15-min infusions on each of 5 consecutive days over a period of 7 days (<1 BIU) or 14 d (≥1 BIU). Patients reaching 10 BIU received 3 additional cycles of 50 BIU. Patients enrolled in the second part received three cycles of 10 BIU. Patients with stable disease were maintained with 2 doses of 10 BIU weekly. Six patients were enrolled in the dose escalation part and eight in the constant dosing part. Three patients were withdrawn during screening and prior to treatment. Grade I/II fever was seen in 5 patients and was the only adverse event related to OV001 administration. Four patients completed at least 2 cycles of 50 BIU OV001. MTD was not reached. Patients that did not complete dosing were withdrawn due to disease progression. Infectious OV001 particles were recovered from blood, saliva, and urine samples and from the tumor biopsy of one patient. Anti-NDV hemagglutinin antibodies were detected after 7–14 days. Of the 6 patients that completed study dosing, one achieved a complete response and another had stable disease. We conclude that intravenous OV001 is well tolerated. There were no significant side effects and the MTD was not reached. The proof-of-concept provided by the complete response seen in one patient warrants the continued evaluation of OV001 (NDV-HUJ) in GBM and other gliomas.


M.R. Gilbert,1 K. Hess,1 P. Gaupp,1 M. Groves,1 V. Puduvalli,1 V. Levin,1 C. Conrad,1 H. Colman,1 S. Hsu,1 P. De Porre,2 and W.K.A. Yung1; 1The University of Texas M.D. Anderson Cancer Center, Houston Texas, USA; 2Johnson & Johnson Pharmaceutical Research & Development, Beerse, Belgium

Farnesylation is an essential step in the post-translational modification of several proteins that play a role in cell proliferation and growth, including Ras, RhoB, centromere binding proteins (CNPE/CNPF), lamin B, and protein tyrosine phosphatase (PTP). Inhibition of farnesylation may inhibit tumor cell proliferation. Preclinical testing demonstrates antiproliferative effects of FTIs in a variety of tumor cell lines and xenograft models. Preliminary data from trials evaluating the efficacy of the FTI tipifarnib in recurrent glioblastoma (GBM) demonstrates modest activity. However, combining a targeted therapeutic with a cytotoxic drug, such as TMZ with isotretinoin or with marimastat, shows an increased response rate in malignant gliomas over temozolomide alone. Additionally, studies demonstrate that TMZ administration could be altered to increase dose intensity to extend beyond the conventional 5-day dosing on a 28-day cycle. Our objective was to determine the MTDs, as measured by clinical and laboratory adverse events of temozolomide and tipifarnib when administered on an alternating week, continuous dosing schedule. Eligibility criteria were as follows: Histologically proven GBM; unequivocal evidence of tumor relapse or progression after radiation therapy; up to 2 prior chemotherapy regimens for recurrent disease; ability to provide informed consent; KPS ≥ 60. Patients could not be on cytochrome P450-inducing anticonvulsants. The treatment plan entailed escalating the doses of TMZ and tipifarnib in an alternating schedule as outlined in the table below. In our statistical design, a conventional 3 + 3 dose escalation scheme was used. An additional three patients were treated at the dose level below the toxic dose to confirm the MTD. Dose escalation and toxicities are reported in the table. On the basis of these data, cohort 3 (TMZ 150 mg/m2 and tipifarnib 500 mg bid) will be the phase II dose. Standard response criteria established by MacDonald were used. To date, 12 patients were assessable for response: 1 CR (10+ months), 1 PR (18+ months), 4 SD (5, 8, 6+, 8+ months), and 6 PD. It is too early to evaluate three of the patients. We were able to safely dose intensify this combination regimen as well as provide continuous treatment with a cytotoxic and a targeted agent on a weekly alternating basis. The treatment has been well tolerated, even over a prolonged time period. Preliminary evaluation suggests that this combination has activity and efficacy is currently being evaluated in a phase II trial.

Treatment Plan

CohortTemozolomide qd, Days 1–7, 15–21Tipifarnib bid Days 8–14, 22–28Toxicity (DLT)
1100 mg/m2400 mg0/3
2100 mg/m2500 mg0/3
3150 mg/m2500 mg0/6
4150 mg/m2600 mg2/3*
5150 mg/m2800 mgNot tested
*Grade 3 fatigue, Grade 4 thrombocytopenia and leukopenia


Gil A. Golden, Matt Meldorf, and the PROLONG Study Group; Guilford Pharmaceuticals, Baltimore, Maryland, USA

Wafer usage is a unique and important aspect of Gliadel therapy. PROLONG is a patient registry designed to capture data on routine clinical use of Gliadel. We report on wafer utilization in patients who were implanted with Gliadel for the treatment of brain cancer. Data were collected prospectively from 322 patients at more than 100 centers from December 2001 to December 2003. Patients were followed at 3-month intervals until death or loss to follow-up. Intraoperative histology was based on frozen sections or tissue smears, and final pathology was determined by examination of fixed tissue samples. In the PROLONG registry, 2168 wafers were implanted in 322 patients for an average of 6.8 ± 2.2 wafers per patient. The median number of wafers per patient was 8, and accounted for 50% of the treated patients. Ninety-five percent of patients received between 1 and 8 wafers.

No. of Wafers12*345678 9101112131416
No. of Patients3151032243622162 3712121
*Or 2.5
Or 4.5

By final tumor histology, of the 304 patients who received 8 or fewer wafers, most (149 patients, 49.0%) were diagnosed with initial malignant glioma (MG), followed by recurrent MG (95 patients, 31.3%), and metastatic lung cancer (21 patients, 14.8%). More than 8 wafers were used in ≈6% of initial and recurrent MG patients.

Wafers usedInitial MGRecurrent MGLung METSOther METSOtherBreast METS
≤8 wafers149952114142
>8 wafers961

Abbreviation: METS, metastases.

Wafer number varied from 1 to 16 per patient in this registry of 322 patients. Patients with metastatic disease who have smaller tumors routinely receive 8 wafers, while patients with larger resection cavities often receive >8 wafers (17 patients) to ensure the cavity is completely lined with wafers.


Gil A. Golden, Matt Meldorf, and the PROLONG Study Group; Guilford Pharmaceuticals, Baltimore, Maryland, USA

PROLONG is a patient registry designed to capture “real-world” Gliadel wafer use. Intracranial metastases occur in >40% of cancer patients, and ≈10–30% of those will recur locally, even after surgery and radiation therapy. Because chemotherapy often fails to achieve sufficient intracerebral levels due to the blood-brain barrier, local therapies such as Gliadel wafer are being examined. Data were collected prospectively from December 2001 to December 2003. Patients were followed at 3-month intervals until death or loss to follow-up. Intraoperative histology was based on frozen sections or tissue smears, and final pathology was determined by fixed tissue samples. Of 322 patients enrolled by December 31, 2003, 36 (11.2%) received Gliadel for treatment of metastatic cancer to the brain. Thirty-one patients (86.1%) underwent surgical resection for the first time, four patients underwent resection for a first recurrence, and one patient had two prior resections. Typically, eight wafers are used following high-grade glioma resection; however, resection cavities left by metastatic tumors tend to be smaller. In this cohort, the mean number of wafers was 5.3 ± 2.6 (median 5.0, range 1–10 wafers). Metastatic tumors were from lung (19, 52.8%), other metastatic (9, 25%), unknown metastatic (7, 19.4%), 1 breast (2.8%), and 1 patient who did not have metastatic brain cancer at final pathology (2.8%). In 27 patients for whom follow-up data were available, none (0/27) experienced local recurrence, and no recurrence was observed in 20 (74%) patients, while there was distal recurrence in 7 patients (26%). This is the largest data set available describing the use of Gliadel in patients undergoing resection for metastatic cancer to the brain. These data suggest that Gliadel is used on a regular basis as part of a multimodality approach to the treatment of metastatic brain cancer with no evidence of local recurrence in those patients for whom follow-up data were available.


K. Jaeckle, K. Ballman, J. O’Fallon , B. Scheithauer, C. Giannini, T. Moynihan, and J. Buckner; Mayo Clinic, Jacksonville, Florida; Mayo Clinic, Rochester, Minnesota, USA

A paradigm for the evaluation of new therapeutic agents in malignant glioma trials involves the assessment of drug activity prior to radiation therapy (RT). However, very few studies have attempted to correlate pre-radiation chemotherapy response with patient outcome. Three recent NCCTG trials (NCCTG 93-72-52; 98-72-51; 98-72-52) involving 186 newly diagnosed malignant astrocytoma patients (GBM-170; AA-16) included pre-radiation chemotherapy and served as the database for this analysis. The study population was restricted to patients with radiographic evidence of measurable disease following initial biopsy or subtotal resection whose response (or lack thereof) was confirmed. Patients were defined as responders if there was a pre-RT radiographic response of CR, PR, or REGR; or, as non-responders if they had a pre-RT response of SD or PROG. A by-study and combined all-patient analysis did not show a statistically significant difference in best pre-RT response as a function of biopsy or subtotal resection (all patients, P = 0.084). There was no significant difference in median survival between non-responders (SD, PROG) and responders (CR, PR, REGR) in the by-study or combined analysis (all patients, 11.2 vs. 14.4 months, log-rank P = 0.087). Overall, survival rates (95% CI) for non-responders versus responders, respectively, at 6 months from study entry was 0.827 (0.771, 0.887) versus 0.958 (0.882, 1.00); at 12 months, the proportion surviving was 0.423 (0.353, 0.506) versus 0.708 (0.548, 0.916), respectively. There was a statistically significant difference in the median survival between pre-RT non-progressors and progressors in 2 of the 3 studies, and for all patients combined (12.1 vs. 6.9 months, log-rank P < 0.0001). Comparison of 12-month outcome (alive vs. dead) by response status showed a statistically significant, albeit modest agreement between the pre-RT response and vital status at 12 months (Kappa statistic = 0.12; 95% CI, 0.03–0.22). Similarly, agreement between pre-RT progressor status and vital status at 12 months was statistically significant, albeit modest (Kappa statistic = 0.16; 95% CI, 0.06–0.26). The median survival of patients with pre-RT response to chemotherapy did not differ significantly from non-responders, suggesting that this parameter may not necessarily predict survival in patients with malignant astrocytoma. However, these data confirm that progression during pre-RT chemotherapy is associated with a significantly shorter survival.


K. Jaeckle, K. Ballman, J. Uhm, J. O’Fallon, P. Schomberg, B. Scheithauer, C. Giannini, P. Flynn, and J. Buckner; Mayo Clinic, Jacksonville, Florida; Mayo Clinic, Rochester, Minnesota; USA

Enzyme-inducing anticonvulsants (EIACs) reduce serum levels of some chemotherapeutic agents that are metabolized via the P450 microsomal enzymes CYP3A4 and CYP2A6. Observations of reduced toxicity of agents metabolized by CYP 3A4 in patients receiving EIACs has resulted in Phase I–II trials which include separate strata for determination of MTD for EIAC and non-EIAC patients. Prior studies of patients with brain metastases receiving EIAC have shown shortened survival, but little data is available in GBM patients in this regard. Three recent prospective NCCTG trials involving 642 patients (93-72-51, 98-72-52, and N0074) with newly diagnosed glioblastoma (GBM) provided the database on which univariable and multivariable analyses were performed, in order to assess relationships between AC/EIAC treatment and overall survival (OS) and progression-free survival (PFS). Baseline seizure history was available on 75 patients. PFS was measured from the time of study enrollment, and OS measured from surgery until death or last follow-up. Of the 642 patients, there were 447 patients receiving EIAC, and a total of 463 patients receiving any type of anticonvulsant (AC). AC use was univariately positively associated with steroid use (P = 0.003), age (P = 0.005) and MMSE (P = 0.028); extent of resection distributions also differed between patients on AC and those not (P = 0.0001). In a multivariable Cox model, OS and PFS were independently associated with AC use (HR = 0.75, P = 0.0029; and HR = 0.80, P = 0.0023; respectively), when adjusting for age, steroid use, extent of resection, and MMSE. EIAC use (N = 447) was also independently associated with OS and PFS (HR = 0.75, P = 0.0029; and HR = 0.81, P = 0.024; respectively). EIAC use was more common in patients with a history of seizures versus no seizures at study entry (100% vs. 66%, P = 0.0005), but there was no significant difference in OS (18.4 vs. 11.2 months, respectively, P = 0.16) or median PFS (10.8 vs. 6.8 months, respectively, P = 0.053). We originally speculated that patients on AC (and especially EIAC) might have shorter survivals because of decreased chemotherapy exposure, but in contrast we observed longer OS and PFS in patients receiving AC and EIAC, a finding which remained significant on multivariable analysis. A history of seizures, a positive prognostic variable in prior reported studies, was associated with improved survival but did not achieve statistical significance. Although more studies are needed, it is possible that EIACs exert a positive effect on survival of unclear etiology, and possibly unrelated to the EIAC-related reduction in chemotherapy levels. These data suggest that AC and EIAC status at study entry should be considered for stratification variables in prospective clinical trials with survival end points.


David A. Jellinek1 and Levy David2; 1Consultant Neurosurgeon, Sheffield Teaching Hospitals NHS Trust, Sheffield; 2Consultant Oncologist, Weston Park Hospital, Sheffield; UK

As part of internal quality control in neuro-oncology management, the local brain tumour database is analysed repeatedly to audit patient outcome by disease type, treating practitioner, and modalities of treatment. In our institution patients with a diagnosis of glioblastoma multiforme are channeled through a multidisciplinary team at each critical stage of their disease progression. All cases of radiologically proven recurrent GBM are discussed in this forum, and a decision is made by a single surgeon with a specialist interest in neuro-oncology about the suitability of further surgical therapy aimed to radically reduce radiologically detectable disease (Gd MRI). It is understood that all patients who are to be offered second time surgery will also be offered post operative chemotherapy with PCV or temozolomide. A small number of patients may also be eligible for intraoperative placement of BCNU wafers. To date 25 patients have fulfilled these criteria. Mean age was 44.5 years. Median postoperative survival was 39.8 weeks. Thirty-day mortality was zero, permanent morbidity was under 5%. These results compare favourably with previous published series which report median postoperative survival times of 14 to 36 weeks. The role of second time surgery will be discussed and the relative contributions of patient selection bias, surgery, and chemotherapy considered.


David A. Jellinek1 and Levy David2; 1Consultant Neurosurgeon, Sheffield Teaching Hospitals NHS Trust, Sheffield; 2Consultant Oncologist, Weston Park Hospital, Sheffield; UK

As part of internal quality control in neuro-oncology management, the local brain tumour database is analysed repeatedly to audit patient outcome by disease type, treating practitioner, and modalities of treatment. This database has now accumulated over 1000 consecutive patients with primary CNS malignant disease, of whom 420 had histologically-proven glioblastoma multiforme. Median survival of the glioblastoma multiforme group as a whole was 34 weeks. When the database was interrogated to address the question of whether there is a survival benefit of radical debulking surgery over biopsy. The result was statistically in favour of radical surgery (median survival prolonged by 34.3 weeks), with a P-value of 0.038. When the database was interrogated to address the question of whether it is important that the operating surgeon who performs the debulking procedure is one who has a specialist interest in neuro-oncology, there was no significant difference in survival time after surgery. Allowing for its inevitable scientific weakness as a retrospective audit, this study provides further supportive evidence for the role of radical surgery over biopsy in the initial management of newly diagnosed glioblastoma multiforme. The lack of difference in survival by operating surgeon may reflect UK philosophy in the choice of patient for radical first time surgery in GBM.


G. Evren Keles, Edward Chang, Tarik Tihan, Kathleen R. Lamborn, and Mitchel S. Berger; Department of Neurological Surgery, University of California San Francisco, San Francisco, California, USA

The available literature evaluating extent of resection for anaplastic astrocytomas as a distinct histological group is limited, and assessment of residual disease is not quantitative. To investigate the prognostic significance of extent of resection on time to tumor progression (TTP), overall survival (OS), and recurrence patterns, we retrospectively analyzed preoperative and postoperative tumor volumes on 102 adult patients who underwent surgery for hemispheric anaplastic astrocytoma. Histological diagnosis of anaplastic astrocytoma was confirmed for all patients using the current World Health Organization criteria. Patients with recurrent anaplastic astrocytoma were not included in this study. Quantification of tumor volumes was based on a previously described method involving computerized image analysis of MRI scans. Volumetric analysis was conducted on contrast-enhancing tumor volumes on T1-weighted MR images for 67 patients who had contrast-enhancing tumors, in addition to measurements of T2 hypointensity for all 102 patients. The variables analyzed included age, Karnofsky performance status (KPS), preoperative tumor volume (T1 enhancement and T2 hypointensity), percent of resection (POR), and volume of residual disease (VRD) (T1 enhancement and T2 hypointensity). All patients had postoperative radio-therapy and 94% (96/102) of the patients received chemotherapy. When extent of resection was analyzed using hypointensity on T2-weighted images for all patients (n = 102), age and VRD were statistically significant factors for both TTP (P < 0.001 and P = 0.0001, respectively), and OS (P = 0.001 and P = 0.012, respectively). Although at a lesser degree of significance (P = 0.015), POR was a prognostic factor for TTP (but not for OS) when VRD was removed from the statistical model. For contrast-enhancing tumors, age and POR were the most significant predictors of TTP (P = 0.03 and P < 0.0001, respectively) and OS (P = 0.008 and P = 0.005, respectively). Unlike low-grade gliomas, there was no statistically significant relationship between extent of resection, that is, POR and VRD, and histology at the time of recurrence. This retrospective analysis of a histologically uniform group of hemispheric anaplastic astrocytomas treated in the MR era suggests that volumetrically assessed extent of resection, as documented on postoperative imaging studies, may be a prognostic factor for time to progression and survival for this patient population.


S. Kesari,1,2,3 G. Shah,2 R. Xu,1 J. Henson,3 T.T. Batchelor,3 A. O’Neill,3 F.H. Hochberg,3 B. Levy,1 J. Bradshaw,1 and P.Y. Wen1,2; 1Dana Farber Cancer Institute, Boston, Massachusetts; 2Brigham and Women’s Hospital, Boston, Massachusetts; 3Massachusettes General Hospital, Boston, Massachusetts; USA

The RECIST criteria (1D measurement) are widely used to measure response in solid tumors, but there are few studies validating these criteria in brain tumors. In a small study in childhood brain tumors, Warren et al. (2001) found a high concordance between 1D, 2D, and volume measurements in detecting partial responses but not disease progression. More recently, Galanis et al. (2003) found that tumor responses assessed by RECIST and 2D measurements were equally associated with survival improvement. In this study we compared the 1D, 2D, 3D, computer-assessed total tumor volume (TV) and enhancing volume (EV) measurements in adult high-grade supratentorial gliomas. Tumor size on MRI scans in 104 patients with high-grade gliomas treated on clinical trial protocols were measured by using 1D, 2D (product of perpendicular diameters), 3D (products of the longest axial perpendicular diameters and the longest vertical diameter), TV and EV. A total of 388 T1 post-gadolinium MRI scans (104 baseline and 284 follow-up scans) were evaluated. Volumetric analysis was performed using VITREA 2 software. Correlation was excellent (r > 0.9) between all the modalities except for 3D; 1D versus 2D (r = 0.94; 95% CI, 0.92–0.95), 1D versus 3D (0.57; 0.49–0.64), 1D versus TV (0.93; 0.91–0.94), 1D versus EV (0.95; 0.94–0.96), 2D versus 3D (0.44; 0.34–0.53), 2D versus TV (0.95; 0.94–0.96), 2D versus EV (0.96; 0.95–0.97), 3D versus TV (0.48; 0.39–0.56), 3D versus EV (0.5; 0.41–0.58), and TV versus EV (0.99; 0.98–0.99). Intra-observer (2 measurements separated in time, n = 50) and inter-observer (2 different observers, n = 50) correlations were high for all modalities (r > 0.87) with the exception of 3D inter-observer correlation (0.78, CI 0.65–0.87). Median progression-free survival (mPFS) and 6-month progression free survival (6m-PFS) were 5.9 months (95% CI, 5.1–8.3) and 48% (SD 5%) for 1D, 6.0 m (5.1–8.3), and 49% (5%) for 2D, 5.6 m (4.8–7.4) and 45% (5%) for 3D, 5.7 m (5.0–6.9) and 45% (5%) for TV, 6.0 m (5.1–7.2) and 48% (5%) for EV, and 6.7 m (5.1–7.5) and 54% (5%) based on clinical criteria. These results suggest that there is high concordance among 1D, 2D, TV, and EV, but not 3D, methods in assessing tumor progression and in estimating mPFS and 6m-PFS in adult brain tumor patients. These findings are likely to be of importance in the design of phase II trials.


E. Linetsky, A.A. Hercbergs, S. Dotan, E. Shalom, and T. Siegal; Leslie and Michael Gaffin Center for Neuro-Oncology, Hadassah Hebrew University Hospital, Jerusalem, Israel; Department of Radiation Oncology, Cleveland Clinic Foundation, Cleveland Ohio, USA

Under hypothyroid condition, gene expression of several growth factors and their receptors is downregulated. Experimental and clinical studies demonstrated that hypothyroidism (HT) is correlated with decreased tumor growth rate and improved survival in various types of solid tumors. A previous Phase I/II clinical study suggested that high-dose tamoxifen given in association with chemical induction of HT prolongs survival of patients with recurrent malignant gliomas. The objective of this study was to evaluate the toxicity and effect of chemically induced HT on time to tumor progression (TTP) of patients with malignant gliomas who failed conventional therapies. Propylthiouracil was used to induce HT in 20 patients (median age 49) with failed malignant gliomas (8 glioblastomas, 7 anaplastic astrocytomas, 5 anaplastic oligoastrocytomas). Tamoxifen (240 mg/kg/d) was given only to patients who became hypothyroid. Clinical evaluation was performed monthly and MRI every 8 weeks. Thyroid function tests were done weekly until HT was induced (TSH > 8) and then repeated monthly. Quality of life was assessed monthly by using a standard questionnaire. Twelve patients (60%) became hypothyroid (median TSH 9.8) within a median time of 3 months (range, 1.5–6 months). KPS and age did not differ between the two groups of patients stratified as achieving or not achieving HT. Allergic skin reaction to PTU and excessive fatigue required early withdrawal of two patients. After induction of HT, mild fatigue was the main adverse effect, with no other clinical symptoms of HT. No ophthalmologic side effects were noted for patients on tamoxifen. Median TTP was significantly longer in the HT group (5 months vs. 2.7 months, P = 0.002), with 6-month PFS of 33% versus 0%. Clinical improvement was noted in 8/12 (66%) patients who became hypothyroid and led to withdrawal of steroid therapy in two patients and to reduction of the dose in 6 others. Marked decrease in seizures’ activity was noted and two patients became seizure free. MRI showed objective response in 25% of the HT group (maintained at 6, 18 and 26 months) and stable disease in 75% (median duration 6 months). Quality of life of the HT group was improved after 3 months and differed significantly from the group without HT (P = 0.02). We conclude that induction of HT is associated with significantly longer median TTP in patients with failed malignant gliomas. Clinical and objective responses improve quality of life despite the induction of HT. Further studies are warranted to evaluate the benefit of early induction of HT shortly after diagnosis and prior to tumor recurrence.


Michael Linnebank,1 Horst Urbach,2 Klaus Fließbach,1 Natalia Kleczar,1 Hendrik Pels,1 Susan Farmand,1 Ingo GH Schmidt-Wolf,3 and Uwe Schlegel1; Department of 1Neurology, 2Neuro-Radiology, and 3Internal Medicine, University Hospital Bonn, Bonn, Germany

Methotrexate (MTX), a folate antagonist, is the most efficient drug in primary CNS lymphoma (PCNSL). Periventricular leukoencephalopathy, a common neurotoxic effect of high-dose MTX in PCNSL, may be related to CNS demyelination. The capacity of (re)myelination in the CNS depends on S-adenosylmethionine-levels, which vary according to genetic polymorphisms in key enzymes of the methionine-metabolism. In the present study, 38 patients with PCNSL who had been treated with a chemotherapy regimen including high-dose MTX were investigated for MRI evidence of leukoencephalopathy and for the presence of polymorphisms affecting the methionine synthesis. A genetic risk profile was defined as presence of at least one of the following synergistic polymorphic genotypes: methylenetetrahydrofolate reductase c.677C>T (A222V): TT, transcobalamin c.776C>G (P259R): GG, methionine synthase c.2756A>G (D919G): AG or GG. Thirteen patients (34%) developed leukoencephalopathy during treatment. Of 13 patients with LEP, 11 (85%) showed this genetic risk profile, but only 9 of 25 (36%) patients without LEP (two-sided χ2 [Pearson] = 8.1 and P = 0.004, OR = 9.8 [95% CI, 1.5–82], relative risk = 5.0). Our data suggest that the methionine metabolism may play a crucial role in the development of MTX-induced leukoencephalopathy.


David R. Macdonald; London Regional Cancer Centre, University of Western Ontario, London, Ontario, Canada

Pilocytic astrocytomas (PAs) are WHO grade I tumors that usually occur in children. Surgical resection is often curative. Incompletely resected PAs are often treated with radiotherapy (RT). Hypothalamic and optic pathway PAs in children often respond to carboplatin- or lomustine-based chemotherapy, but side effects are common and sometimes serious. There is little information on chemotherapy in adult PA. Two adults with recurrent PA responded to temozolomide (TMZ). A man, age 54, had a cerebellar PA partly resected in 1995 and then received RT. Multifocal recurrence was biopsy-confirmed in 2/2002. He received 12 cycles of standard oral TMZ (200 mg/m2/day × 5 days, every 28 days). A partial response (PR) was seen after 6 cycles, and a complete response (CR) after 12 cycles. He remains in CR over 15 months off treatment. A woman, age 27, had partial resection of a right temporal PA in 1987. MRI progression was biopsy confirmed in 1/2000, and treated with RT. Symptomatic MRI progression was treated with standard TMZ starting 11/2001. A minor response was seen after 9 cycles and a PR after 12 cycles. She received 24 cycles of TMZ with continuing PR more than 6 months off treatment. TMZ was well tolerated in both patients. TMZ is active in PA, producing sustained responses, but prolonged treatment may be needed.


M. Mehta,1 J.M. Ford,2 J. Suh,3 and S. Phan4; 1University of Wisconsin, Madison, Wisconsin; 2University of California at Los Angeles, Los Angeles, California; 3Cleveland Clinic, Cleveland, Ohio; 4Pharmacyclics, Inc., Sunnyvale, California; USA

The prognosis for newly diagnosed GBM remains poor. Pre-clinical studies with the tumor selective anti-cancer agent, motexafin gadolinium (MGd, Xcytrin) in glioma cell lines have shown increased cytotoxicity after MGd and radiation compared to radiation alone. Two trials combining MGd with cranial irradiation have been completed in patients with newly diagnosed GBM. The first, a dose escalation trial (NCI-T97-0108) using increasing cumulative doses of MGd in successive cohorts, resulted in improved survival compared to matched historical controls from the RTOG database (median survival, 16.1 months vs. 11.8 months, HR 2.3, P = 0.03 [Ford et al., 2003, Proc. Am. Soc. Clin. Oncol. 22:106]). The second trial (PCYC-0206) evaluated a total MGd dose of 110 mg/kg. Data sets from these trials were combined to evaluate for possible dose-dependent effects on outcomes. Trial NCI-T97-0108 for newly diagnosed untreated GBM used 58.4 Gy of cranial irradiation combined with MGd over 6.5 weeks. Eligible patients had histologically confirmed GBM, and KPS ≥60. Cumulative dose of MGd was escalated from 40 mg/kg to 117 mg/kg. Trial PCYC-0206 with similar eligibility criteria treated patients uniformly with 60 Gy of cranial irradiation combined with 110 mg/kg of MGd over 6 weeks. The two data sets were merged for analysis. A total of 56 GBM patients were treated with MGd and cranial irradiation (31 on NCI-T97-0108; 25 on PCYC-0206). Median age was 56.5 years (28–80). Resection status included 20 complete, 23 partial, and 13 biopsy only. Eleven patients received <60 mg/kg of MGd; 45 patients received ≥60 mg/kg of MGd. Median dose of radiation delivered was 58.4 Gy (2–60). The median survival of the combined patients treated with MGd and cranial irradiation was 14.7 months. Patients receiving <60 mg/kg of MGd had a shorter median survival (11.5 months) and fewer patients alive at 6 and 12 months (82% and 45%) compared to those receiving ≥60 mg/kg (14.7 months, 93% and 75%, respectively). Total radiation delivered to each dose group was similar (<60 mg/kg, median 58.4 Gy; ≥60 mg/kg, median 60 Gy). We conclude that newly diagnosed GBM patients treated with greater cumulative doses of MGd had longer survival despite similar exposure to total cranial radiation than those treated with lower MGd doses. These results suggest a possible dose-dependent effect of MGd on survival.


C. Mihalcioiu, S. Huszar, A. Demers, G. Schroeder, K. Jones, K. Vijay, R. Rhodes, and D.D. Eisenstat; Brain Tumour Disease Site Group, CancerCare Manitoba, Winnipeg, Manitoba, Canada

A recent phase II study reported activity of TMZ and cRA for recurrent/progressive malignant gliomas with 43% six-month progression-free survival (PFS) and 32% for glioblastoma multiforme (GBM) patients. From 2000 to 2003, we retrospectively analyzed adults with malignant glioma who received, following surgery and radiation therapy (RT) ± low dose TMZ (75 mg/m2/d × 42 d), TMZ (150–200 mg/m2/d, d 1–5) and cRA (100 mg/m2/d, d 1–21) every 28 d for up to 24 cycles or until progression on neuroimaging. Time to progression (TTP) was the primary end point, and overall survival (OS) was the secondary end point. Of 30 patients (26 GBM, 2 anaplastic astrocytoma [AA], 2 malignant glioma [MG]), 15 remain alive. Thirty patients received 238 cycles of TMZ/cRA; seven patients received greater than 10 cycles, but only two completed more than 15 cycles. Responses to TMZ/cRA were documented in 12/30 (40%) patients (5 stable disease/7 partial responses). For all patients, 6- and 12-month PFS was 57%/35% (53%/25% for RT alone and 60%/45% for RT/low dose TMZ, NS). OS at 6, 12, and 23 months was 93%, 67%, and 42%, respectively, for all 30 patients (87%/50%/33% for RT alone and 100%/83%/55% for RT/low dose TMZ, P = 0.09). The chemoradiation strategy using concurrent low-dose temozolomide appears to provide a survival advantage when compared to radiation alone. Our study demonstrates that TMZ/cRA prolongs TTP for patients with malignant gliomas when compared to historical controls and supports use of this regimen in a Phase III clinical trial.


T. Mikkelsen,1 T. Doyle,1 J. Margolis,2 N. Vick ,3 N. Paleologos,3 G. Divine,1 and D. Croteau4; 1Hermelin Brain Tumor Center, Josephine Ford Cancer Center, Henry Ford Hospital, Detroit, Michigan; 2Rose Cancer Treatment Center, William Beaumont Hospital, Royal Oak, Michigan; 3Northwestern University, Evanston, Illinois; 4NeoPharm, Inc., Lake Forest, Illinois; USA

Recent description of the favorable prognostic significance of chromosome 1p/19q loss of heterozygosity (LOH) in patients with anaplastic oligodendroglioma has allowed the implementation of genotyping as part of clinical management at some centers. Owing to the favorable long-term prognosis in this population and the potential for long-term radiation toxicity in this patient population when used in an adjuvant setting, we elected to perform a pilot study to examine the safety and efficacy of chemotherapy alone in a group of patients with newly diagnosed anaplastic oligodendroglioma and mixed oligoastrocytoma. We intended to test the hypothesis that, in those patients harboring the most favorable histologic profile with complete 1p/19q LOH, single agent chemotherapy with temozolomide could produce sustained objective responses and time-to-tumor progression (TTP) which was not worse than historical control, and which was safe. The long-term goal will be to assess our ability to withhold radiation therapy and hopefully postpone long-term radiation toxicity in those patients who are likely to survive long enough to sustain clinically significant radiation injury, while maintaining long-term response and prolonging overall survival. We have undertaken a clinical trial to evaluate the clinical efficacy and toxicity of temozolomide single-agent chemotherapy for patients with newly diagnosed anaplastic oligodendrogliomas or mixed anaplastic oligoastrocytomas (AOAs). Treatment assignment is based on the identification of chromosome 1p/19q LOH, allowing those patients demonstrating this favorable prognostic factor to continue with chemotherapy alone until progression, whereas those patients not showing LOH are treated with concurrent chemo-radiation and adjuvant chemotherapy until progression. Primary end points for this trial are clinical efficacy, defined as radiographic tumor responses by MRI and TTP, in addition to overall survival and adverse events. We will correlate chemoresponsiveness with specific molecular genetic alterations (1p/19q LOH) and histopathological features for patients with newly diagnosed anaplastic oligodendrogliomas and mixed AOAs. To date, 28 of a planned 58 patients have been accrued, and planned interim analysis of outcome data will be presented for response, TTP, and adverse events. Survival data to date will also be presented.


Melike Mut,1 James Larner,2 Chris I. Shaffrey,3 Bradley Miller,4 Mark E. Shaffrey,5 and David Schiff,6; Departments of 1Neurosurgery, 2Radiation Oncology, 3Neurosurgery, 4Pathology, 5Neurosurgery, and 6Neurology/Neuro-oncology, Charlottesville Virginia, USA

We present a case with spinal cord compression from osteoradionecrosis. A 45-year-old man presented with lower cranial nerve dysfunction and myelopathy with asymmetrical quadriparesis. He had a history of squamous cell carcinoma of the floor of the mouth treated with surgery and 6000 cGy of radiotherapy to the tumor area, posterior neck, and skull base five years ago. Radiological studies revealed a marrow-replacing process involving the clivus, C1, and odontoid process with a 9-mm enhancing mass resulting in compression of the medulla oblongata and spinal cord. He underwent surgery via a posterior approach to relieve spinal cord compression, perform instrumentation and fusion to correct the instability at the occipitocervical junction, and permit tissue sampling of the anteriorly located mass. Histopathological examination of the enhancing nodule as well as the destroyed anterior arch of C1 showed fibrosis and bony destruction consistent with osteoradionecrosis (ORN). ORN is one of the most serious complications in the treatment of head and neck malignancies. Spinal cord and medullary compression from ORN producing an epidural fibrotic mass has not been previously reported. The present report discusses the differential diagnosis of enhancing epidural masses arising in a previously irradiated port in the presence of cancer history and therapeutic approaches for ORN.


L.B. Nabors, S.S. Rosenfeld, T. Mikkelsen, T. Batchelor, J. Olson, S. Grossman, S. Piantadosi, and J.D. Fisher; The New Approaches to Brain Tumor Therapy (NABTT) CNS Consortium, Baltimore, Maryland, USA

New Approaches to Brain Tumor Therapy (NABTT) CNS Consortium EMD 121974 is a cyclic peptide containing the RGD (Arg-Gly-Asp) sequence. This peptide has demonstrated in vivo and in vitro anti-angiogenic activity by interfering with alpha v beta 3 and alpha v beta 5 integrin signaling in proliferating and migrating endothelial cells. In this multi-institutional trial, we have conducted a phase I dose-escalation study to determine the maximum tolerated dose (MTD) of EMD 121974 in patients with recurrent malignant glioma. Cohorts of six patients have completed enrollment through the following dose levels: 120 mg/m2, 240 mg/m2, 360 mg/m2, 480 mg/m2, 600 mg/m2, 1200 mg/m2, 1800 mg/m2, and 2400 mg/m2. Therapy was given as a twice a week infusion with cycles of 30-day duration. A total of 51 patients were enrolled and assessable for toxicity. Treatment-related dose-limiting toxicity (DLT) was defined as any grade 3 or 4 non-hematological toxicity or grade 4 hematological toxicity of any duration. The following DLTs were observed: one grade 4 joint and bone pain at the 480-mg/m2dose, one thrombocytopenia at 600 mg/m2, and an anorexia, hypoglycemia, hyponatremia at 1800 mg/m2. A MTD was not defined. Of the assessable patients, 37 are with glioblastoma multiforme (GBM), 11 with anaplastic astrocytoma (AA), 2 with anaplastic mixed glioma, and 1 with anaplastic oligodendroglioma (AO). There were two patients that demonstrated complete response (CR), three patients with partial response (PR), and four patients with stable disease (SD). Correlative biology studies included perfusion MRI and plasma analysis for angiogenic factors. A statistically significant relationship was observed between tumor CBF (cerebral blood flow) and clinical response of any kind. Our conclusions are as follows: (1) EMD121974 (Cilengitide) is well tolerated with limited toxicities to doses of 2400 mg/m2, (2) perfusion MRI is feasible in a multi-institutional setting, and CBF changes correlate with clinical response; (3) plasma assays for angiogenic factors are not reliable indicators of clinical response. A randomized phase II trial of this agent has been initiated in patients with newly diagnosed glioblastoma multiforme concurrent with radiation therapy.


A.M.P. Omuro, E. Franceschi, M.G. Kris, V.A. Miller, N. Shah, and L.E. Abrey; Memorial Sloan-Kettering Cancer Center, New York, New York, USA

Gefitinib (Iressa) is an EGFR tyrosine-kinase inhibitor that induces an early and dramatic response in 10% of patients with advanced NSCLC. However, CNS metastases have been increasingly recognized as the first site of disease progression among responders. In this study we describe the occurrence of brain metastases (BM) and leptomeningeal metastases (LM) in NSCLC patients who responded systemically to gefitinib. We conducted a retrospective review of long-term outcomes of all patients with NSCLC who started treatment with gefitinib between 1998 and 2002. A total of 139 patients with NSCLC started treatment with gefitinib within the period. Objective systemic response was observed in 21 patients (15%; 95% CI, 9%–21%). Among responders, median age was 64 (38–87); median KPS was 80 (60–90); 4 were men and 17 were women. All patients had adenocarcinoma; bronchioloalveolar features were present in nine of them. Eight patients (38%; 95% CI, 18%–61%) developed new CNS metastases (6 BM and 2 LM). In seven patients, CNS was the sole site of progression. Five patients with new BM were treated with whole-brain radiotherapy (RT); three of them re-started gefitinib following RT but progressed in the CNS shortly after. At last follow-up, only one (13%) of the patients with CNS relapse was alive, compared to seven (54%) in the group without CNS relapse. CNS is a frequent site of relapse in NSCLC patients who are otherwise well controlled with gefitinib. This pattern of relapse has a poor prognosis. Intrinsic resistance of metastatic clones, incomplete CNS penetrance of the drug, and longer survival are possible explanations for this high incidence and warrant further investigation.


David M. Peereboom, Cathy Brewer, Glen H.J. Stevens, John H. Suh, Steven Toms, Michael A. Vogelbaum, and Gene H. Barnett; Brain Tumor Institute, The Cleveland Clinic Foundation, Cleveland, Ohio, USA

Approximately 40% to 50% of GBM tumors overexpress EGFR, which activates tyrosine kinase to promote GBM cell infiltration in vitro. Radioresistance of some GBMs has been correlated with EGFR overexpression. Tyrosine kinase inhibitors (TKIs) cross the blood-brain barrier and inhibit downstream activation of the EGFR pathway. Erlotinib is a specific and potent inhibitor of EGFR tyrosine kinase activity against refractory GBM. TKIs are well-tolerated oral agents that lend themselves to maintenance therapy. Furthermore, the occurrence of erlotinib-induced rash of > grade 2 has been correlated with improved survival in NSCLC. A phase 2 trial of erlotinib added to concurrent RT-temozolomide (RT-tmz) was therefore initiated with pharmacodynamic dosing or erlotinib. Our study objectives are as follows; (1) to determine the progression-free survival (PFS) and overall survival (OS) in patients (patients) with newly diagnosed GBM treated with erlotinib and concurrent RT-tmz, (2) to evaluate the feasibility of dose escalation to a pharmacodynamic end point in these patients (3) to evaluate the toxicity of this regimen. Eligible patients have histologically proven newly diagnosed GBM, age at least 18, KPS at least 60, no cytochrome P-450 enzyme-inducing anticonvulsants, and no prior temozolomide (tmz), TKI, or cranial RT. Patients receive RT 60 Gy in 30 fractions with concurrent tmz (75 mg/m2/day × 7 days/week). Four weeks after RT-tmz patients receive tmz 150 mg/m2/day × 5 days, every 28 days for 6 cycles. Patients receive erlotinib daily from D1 of RT until disease progression. Erlotinib dosing starts at 50 mg/day with escalation every 2 weeks by 50 mg/day until the occurrence of grade 2 rash or 150 mg/day. All tumors have FISH assay for EGFR amplification. To date 5 patients have enrolled, with a median follow up of 11 weeks (range, 9–22). The median age is 47 (18–51) and median KPS is 90 (80–100). One patient had gross total resection. Four patients are EGFR negative. One patient reached a maximum erlotinib dose of 100 mg/day, and four reached 150 mg/day. Rash has been G2 (2 patients) and G1 (2), and one had no rash. Two patients have stable disease, two have progressed (both at 11 weeks), and one is too early to evaluate. No patient has had G4 toxicity. G3 toxicity has included neutropenia (one patient) and lymphopenia (two). Additional events unrelated to therapy have included wound infection, spontaneous pneumothorax, and rectus sheath hematoma. The combination of erlotinib with RT-tmz is feasible and overall has been well tolerated. Dose escalation to a pharmacodynamic end point is feasible in GBM. Accrual continues.


Richard A. Peterson; University of Washington, Seattle, Washington, USA

The patient is a 53-year-old man with a history of deja vu dating back to October 2001. MRI showed mild FLAIR signal in the right thalamus. He had a generalized seizure in November 2001, but had good control until August 2003. A repeat MRI showed a mass lesion in the right thalamus. A subtotal resection revealed glioblastoma. He then had external beam radiotherapy to 5940 cGy. Starting 1/9/04 he took temozolomide on days 1 and 2, receiving 2750 mg each day for a total of 5500 mg. At 200 mg/m2, his total 5-day dose would have been 1940 mg; thus he received 2.8 times the normal dose. These two days were complicated by nausea, vomiting, and diarrhea. While trying to obtain enough pills to fill the prescription, a pharmacist questioned the dosing, prompting a call from the patient’s wife. The patient received no further chemotherapy and was promptly admitted. On admit his WBC was 5.19 THOU/μl, ANC 4.57 THOU/μl, and platelets 134 THOU/μl. Empiric treatment with GCSF 480 mcg QD and Epogen 10,000 u thrice weekly was started. Empiric antibiotics, antifungals, and antivirals were started. Shortly after admission, he noticed tooth pain, and an abscess was found. This was treated with drainage and root canal. A mild transaminitis was noted, and to prevent any liver or hematotoxicity, Tegretol was changed to Topomanx. His labs peaked on day 9: GOT 82 U/L (10–44), GPT 223 U/L (12–59), ALk Phos 178 U/L 939–139). Despite GCSF support, his WBC dropped on day 8 to 2.19, and became neutropenic on day 9 at 0.79. This stabilized at 0.49 over the next 48 h, and since he was afebrile was discharged with QOD labs. During the course of admission, cardiac monitoring was normal, as were pulmonary function tests. On day 16 he developed fever and confusion, so admission was indicated. WBC was 0.21, ANC 0.00, platelets 17 K. He was started on IV antibiotics and Serratia bacteremia was identified. He was neutropenic by WBC for 14 days total, with a nadir of 0.04 on day 18. His ANC was 0.00 for 7 days, and below 1.5 for 16 days total. His lymphocyte count on original admission was 0.26 THOU/μl and took 2 months to recover to low-normal range. His platelets were below 50 for a total of 22 days, requiring 10 total transfusions. His hematocrit nadir was 24 and required 2 total transfusions. A bone marrow biopsy on day 25 was markedly hypocellular, with some signs of regeneration. He was maintained on Epogen and GCSF for 23 days total. He has since had a one-week course of Etoposide, with an ANC below 1.5 for 21 days. Temozolomide is generally well tolerated, but this case demonstrates the serious morbidity associated with excessive dosing. Complications were primarily a result of pancytopenia, and have limited further treatment of his tumor.


S. Phuphanich, S.A. Grossman, G. Lesser, T. Mikkelsen, K. Carson, S. Desideri, and J. Fisher; New Approaches to Brain Tumor Therapy (NABTT) CNS Consortium, Baltimore, Maryland, USA

Atrasentan is a highly potent and selective endothelin (ET) receptor antagonist that may inhibit cell proliferation by blocking the ET receptor that regulates angiogenesis of glioma growth. This ongoing study evaluates the safety, toxicity, PK, and therapeutic activity of escalating doses of atrasentan in the treatment of recurrent malignant glioma (MG). Sixteen patients with supratentorial progressive MG with previous radiation and < one regimen of chemotherapy were enrolled on this dose escalation trial. Six dose levels of atrasentan were tested: 10 mg/day (n = 2), 20 mg/day (n = 2), 30 mg/day (n = 3), 50 mg/day (n = 3), 70 mg/day (n = 3), and 90 mg/day (n = 3). The maximum tolerated dose (MTD) is defined as the maximum dose that causes no dose-limiting toxicity (DLT) in an expanded cohort of 10 patients. The cycle length was 28 days, and intra-patient dose escalation was not allowed. Patients were evaluated for response with MRI and neurological examination every 8 weeks. Blood samples for PK analysis were performed on day 29. Patients remained on study until tumor progression. Sixteen assessable patients (12 males, 4 females) have been enrolled. Their median age was 52.8 years; median Karnofsky performance status was 85%; 15 had glioblastoma multiforme (GBM) and 1 had an anaplastic astrocytoma (AA). Fourteen patients had received one prior chemotherapy regimen (1 BCNU, 1 PCV, 12 Temodar), and two had no previous chemotherapy. Eight patients were on enzyme-induced anti-epileptic drug, and eight were not. One patient had a DLT of hypoxia and peripheral edema at 90 mg/day after 18 days of dosing with atrasentan. Seven additional patients will be treated at 70 mg/day to confirm this dose as the MTD. The most common adverse events were rhinitis, headache, and peripheral edema. These events were reported as grade 1/2 and were probably related to atrasentan. There were no events of myelosuppression. To date, 14 patients have developed tumor progression. However, three of these GBM patients’ best responses was stable disease. One GBM patient remains with stable disease, and one GBM patient has a partial response. Five of the 16 patients are alive. Kaplan-Meier median survival is 6.0 (95% confidence interval, 5.4–9.5) months. Chronic oral dosing of atrasentan is safe in patients with recurrent malignant gliomas. The MTD of atrasentan in patients with recurrent high-grade gliomas (15 GBM, 1 AA) appears to be 70 mg/day. Pharmacokinetic data is currently being analyzed. Further study using atrasentan with radiation therapy in newly diagnosed malignant gliomas is planned to estimate the efficacy of this novel agent.


Jennifer A. Quinn, Annick Desjardins, David Reardon, James Vredenburgh, Jeremy Rich, Sridharan Gururangan, Allan Friedman, John Sampson, James Provenzale, Roger McLendon, James Herndon II, M. Eileen Dolan, Michael Badruddoja, Amy Walker, Sandra Tourt-Uhlig, and Henry S. Friedman; Duke University Medical Center, Durham, North Carolina; The University of Chicago, Chicago, Illinois; USA

Although the combination of Temodar and CPT-11 reveals a marked increase in activity compared with either agent used alone, the addition of O6-BG to this combination dramatically increased the growth delay of the O6-alkylguanine-DNA alkyltransferase (AGT)-positive malignant glioma D-456 MG. We are conducting a phase I trial of Temodar plus CPT-11 plus O6-BG to define the maximum tolerated dose (MTD) of CPT-11, along with toxicity and antitumor response associated with this drug combination in the treatment of adults with recurrent or progressive cerebral anaplastic glioma. Patients are administered intravenous O6-BG at a dose of 120 mg/m2 over 1 h, followed within 1 h by oral Temodar at a dose of 355 mg/m2. One hour after Temodar administration, CPT-11 is infused over 90 min. Immediately upon completion of CPT-11 infusion a 48-h O6-BG infusion at 30 mg/m2/day is administered. Repeat dosing of Temodar and CPT-11 occurs on day 1 of a 21-day cycle. CPT-11 dose is escalated in cohorts of 3–6 patients with two independent dosing schedules based on enzyme-inducing antiepileptic drug (EIAED) use. Radiographic response criteria are utilized to evaluate activity using T1-weighted, enhanced MRI images. Thirty-three patients have been treated to date, 31 with glioblastoma multiforme (GBM), 1 with anaplastic astrocytoma (AA), and 1 with anaplastic oligodendroglioma (AO). Eighteen patients have been accrued to the EIAED stratum at CPT-11 doses of 60, 90, 120, and 150 mg/m2. Fifteen patients have been accrued to the non-EIAED stratum at a CPT-11 dose of 40 mg/m2. In the non-EIAED stratum dose, de-escalation for Temodar was required from 355 to 267 mg/m2 and then to the present dose of 200 mg/m2. Dose-limiting toxicities observed thus far have been limited to hematologic events with six episodes of grade 4 neutropenia and three episodes of grade 4 thrombocytopenia. Sixteen patients have shown progressive disease after either the 1st or 2nd cycle. Four patients have remained stable for at least 4 to 8 cycles. One patient has shown a partial response for at least 4 cycles. Four patients are not assessable for response, and for three it is too early to determine response. The MTD of this drug combination has yet to be identified.


Jennifer A. Quinn, Annick Desjardins, David Reardon, James Vredenburgh, Jeremy Rich, Sridharan Gururangan, Allan Friedman, John Sampson, James Provenzale, Roger McLendon, James Herndon, II, M. Eileen Dolan, Michael Badruddoja, Amy Walker, Sandra Tourt-Uhlig, and Henry S. Friedman; Duke University Medical Center, Durham, North Carolina; The University of Chicago, Chicago, Illinois; USA

The major mechanism of resistance to alkylnitrosourea therapy involves the DNA repair protein O6-alkylguanine-DNA alkyltransferase (AGT), which removes chloroethylation or methylation damage from the O6-position of guanine. O6-BG is an AGT substrate and inhibits AGT by suicide inactivation. We completed a phase I trial of Temodar plus O6-BG which defined the maximum tolerated dose of Temodar to be 472 mg/m2 when administered as a single dose every 28 days. We are conducting a phase II trial of Temodar plus O6-BG to define the ability of O6-BG to restore Temodar sensitivity in patients with Temodar-resistant malignant glioma and to further define the toxicity of this drug combination. Patients are administered a 1-h O6-BG infusion at a dose of 120 mg/m2 followed immediately by a 48-h infusion at a dose of 30 mg/m2/day. Temodar is administered orally within 60 min of the end of the 1-h O6-BG infusion at a dose of 472 mg/m2. Repeat dosing of Temodar occurs on day 1 of a 28-day cycle. Patients are accrued into two independent strata based on histology (glioblastoma multiforme [GBM] vs. anaplastic astrocytoma [AA]). Radiographic response criteria are utilized to evaluate activity using T1-weighted, enhanced MRI images. Fifty-four patients have been treated to date: 34 with GBM and 20 with AA. Twenty-six patients have shown progressive disease after the first or second cycle. Nineteen patients have remained stable for at least 3 to 8 cycles. Three patients have shown a partial response for at least 2 to 14 cycles. Four patients are not assessable for response, and for two it is too early to determine response. Drug-related toxicities have been limited to hematologic events and include the following: 1 episode of grade 3 febrile neutropenia, 24 episodes of grade 4 neutropenia, 7 episodes of grade 4 thrombocytopenia, and 1 episode of grade 4 anemia. Further enrollment into this protocol will continue for a total of 32 assessable patients per stratum.


D. Reardon, G. Akabani, A. Friedman, H. Friedman, J. Herndon, R. McLendon, J. Quinn, J. Rich, J. Vredenburgh, K. Penne, J. Sampson, S. Gururangan, T. Shafman, T. Wong, J. Dowell, R. Dunn, M. Badruddoja, A. Desjardins, M. Affronti, D. Allen, S. Jackson, K. Zeigler, S. Silverman, S. Tourt-Uhlig, R. Coleman, M. Zalutsky, and D. Bigner; Duke University Medical Center, Durham, North Carolina, USA

We have previously shown that the administration of 131I-labeled anti-tenascin monoclonal antibody 81C6 (131I-81C6) into the surgically created resection cavity (SCRC) improves survival for patients with newly diagnosed and recurrent malignant glioma. Dosimetry analyses from these studies suggest that the delivery of 44 Gy to the SCRC by 131I-81C6 is associated with a low rate of toxicity and significant local tumor control. The primary objective of the present study is to evaluate the efficacy and toxicity of administering a dose of 131I-81C6 to achieve a 44-Gy boost to the SCRC perimeter. Eligibility criteria include the following: adults with newly diagnosed and previously untreated malignant glioma; gross total resection; no communication between the resection cavity and the CSF space; KPS > 60%; adequate bone marrow, kidney and hepatic function. 131I-81C6 is administered to achieve a 44-Gy boost to the SCRC based on a pretreatment dosimetry study performed with 5 mCi of 123I-81C6. Conventional external beam radiotherapy and systemic chemotherapy are prescribed to all patients following 131I-81C6 administration. We have treated 20 patients to date including 14 with GBM and 6 with AA/AO. The median age was 49.5 years (range, 24–70); 67% were male. The median dose of 131I-81C6 administered was 53 mCi (range, 25–150). All patients have achieved a 44-Gy (±10%) boost to the SCRC perimeter, and no patients have experienced grade 3 or greater toxicity attributable to 131I-81C6. With a median follow-up of 34 weeks, the median survival for all patients and those with GBM is 94 weeks. Additional accrual and follow-up are ongoing. These preliminary results suggest that further investigation of 131I-81C6 administered to achieve a 44-Gy boost to the SCRC is warranted.


D. Reardon, A. Friedman, J. Herndon, J. Quinn, J. Rich, J. Vredenburgh, M. Badruddoja, A. Desjardins, J. Sampson, S. Gururangan, J. Dowell, M. Affronti, D. Allen, S. Jackson, K. Zeigler, S. Silverman, S. Tourt-Uhlig, S. Silverman, A. Salvado, Z. Nikolova, D. Bigner, and H. Friedman; Duke University Medical Center, Durham, North Carolina; Novartis Pharmaceutical Corporation, East Hanover, New Jersey; USA

Imatinib mesylate (Gleevec) has multiple anti-tumor properties including inhibition of the PDGF receptor tyrosine kinase, anti-angiogenic activity and the ability to decrease tumor interstitial pressure. When combined with hydroxyurea, Gleevec was recently shown to induce radiographic response in 1/3 of patients with GBM resistant to both Temodar and nitrosoureas and achieved stable disease in another 1/3 of patients (Dresemann, 2003, Proc. Am. Soc. Clin. Oncol. Chicago). The present study is being conducted to further evaluate the activity of Gleevec plus hydroxyurea (G+H) in patients with recurrent malignant glioma. Eligibility criteria include recurrent malignant glioma, age > 18 years; KPS 60% or greater; less than grade 2 intratumoral hemorrhage; and adequate hepatic, renal, and bone marrow function. Hydroxyurea is administered at 500 mg BID while Gleevec is administered at 500 mg BID for patients on enzyme-inducing anticonvulsants (EIAC; phenytoin, carbamazepine, oxycarbamazepine, and phenobarbital) and at 400 mg QD for those not on EIAC. Each treatment cycle is 28 days, and patients are evaluated for response every other cycle. Forty-eight patients have been enrolled to date, including 32 with recurrent GBM and 16 with recurrent AA/AO. The median age is 50.2 years (range, 22.9–67.8) and 31 (65%) are male. Twenty-seven (57%) are on EIAC, while 20 (43%) are not on EIAC. Follow-up is limited to date with 40 patients (83%) remaining on therapy between cycles 1–3. Six patients (13%) discontinued because of progressive disease while 2 patients withdrew from study. Overall, the regimen has been well tolerated. Grade 3 or greater toxicity attributable to G+H includes: grade 3 pulmonary edema (n = 1), grade 3 fever and neutropenia (n = 1), and grade 4 neutropenia (n = 1). An update of the outcome and toxicity of G+H for patients with recurrent malignant glioma will be presented based on additional follow-up and enrollment


D. Reardon,1 H. Friedman,1 M. Brada,3 C. Conrad,2 J. Provenzale,1 E. Jackson,2 H. Serajuddin,4 and B. Chen,4; 1Duke University Medical Center, Durham, North Carolina, USA; 2The University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA; 3Royal Marsden NHS Trust, Surrey, UK; 4Novartis Pharmaceuticals Corporation, East Hanover, New Jersey, USA; 5Schering AG, Berlin, Germany

VEGF-mediated signaling through VEGF receptors is integral to GBM neovascularization and is therefore an attractive therapeutic target. PTK/ZK is a multi-VEGF receptor tyrosine kinase inhibitor that provides complete inhibition of intracellular signaling by all known VEGF receptors. The primary objective of this trial was to determine the safety profile and maximum tolerated dose (MTD) of oral once-daily PTK/ZK in combination with either temozolomide (TMZ) or lomustine (CCNU) in patients with GBM at first recurrence. Antitumor activity was also assessed. The starting dose of PTK/ZK was 500 mg/day, escalating to 1000, 1250, and 1500 mg/day in cohorts of three to six patients with dose expansion at 1500 mg/day in the TMZ treatment arm. Concurrently, patients also received either TMZ (200 mg/m2/day for 5 days every 28 days) or CCNU (130 mg/m2 every 6 weeks). Among 60 patients enrolled, median age was 54 years (range, 27–79 years), median KPS was 90 (range, 70–100), and the median number of prior regimens was 1 (range, 1–9). In 37 patients treated with PTK/ZK + TMZ, 1 dose-limiting toxicity (DLT) (grade 3 dizziness) was reported at 1500 mg/day and was resolved following a dose reduction to 1250 mg/day. Two of the 15 patients enrolled in the 1500 mg/day PTK/ZK expansion cohort experienced grade 3 or 4 transaminase elevations. MTD was not reached for those patients treated with PTK/ZK + TMZ. Among 23 patients treated with PTK/ZK + CCNU, DLTs included grade 3 febrile neutropenia in one patient each at the 500-mg and 1000-mg levels and grade 4 transaminase elevation in one patient treated at the 1000 mg level, which may be related to CCNU. The PTK/ZK dose level was reduced to 750 mg/day plus CCNU. PTK/ZK was absorbed with a mean tmax of 1.5 h after oral administrations of 500 mg/day. The mean terminal elimination half-life was 4.7 h. Coadministration of PTK/ZK with TMZ slightly increased tmax, but the terminal elimination half-life was not significantly different from that of PTK/ZK alone. Among 51 patients assessable for response, 4 had a partial response (3 with PTK/ZK + TMZ and 1 with PTK/ZK + CCNU) and 34 had stable disease (21 with PTK/ZK + TMZ and 13 with PTK/ZK + CCNU) as best response. Median time to progression was 16.1 weeks (95% CI, 9.1–19.3 weeks) for PTK/ZK + TMZ and 12.1 weeks (95% CI, 8.6–14.6 weeks) for PTK/ZK + CCNU. The median TTP for PTK/ZK + TMZ compares favorably with historical reports with TMZ alone (Br. J. Cancer, 2000, 83:588). Continuous daily dosing of oral PTK/ZK was well tolerated in patients with recurrent GBM. It demonstrated encouraging antitumor activity when combined with either TMZ or CCNU, warranting further study.


John H. Sampson,1 Allan H. Friedman,1 David A. Reardon,1 Henry S. Friedman,1 James M. Provenzale,1 Darell D. Bigner,1 Martin Brady,2 Raghu Raghavan,2 Christoph Pedain,3 Gary Archer,1 Denise Lally-Batts,1 Sharon McGehee,1 Sandra Tourt-Uhlig,1 Amy Grahn,4 Jeanne L. Dul,4 David Croteau,4 Jeffrey W. Sherman,4 and Raj K. Puri5; 1Duke University Medical Center, Durham, North Carolina, USA; 2Image-Guided Neurologics, Baltimore, Maryland, USA; 3BrainLAB AG, Munich, Germany; 4NeoPharm, Inc., Lake Forest, Illinois, USA; 5Center for Biologics Evaluation and Research, U.S. FDA, Bethesda, Maryland, USA

Convection-enhanced delivery (CED) utilizes positive pressure infusion to deliver therapeutic agents into brain tumors or into tissue adjacent to tumors or tumor resection sites. However, little is known about the extent of drug distribution achieved within the brain, or how target tissue selection affects distribution. IL13-PE38QQR is a recombinant cytotoxin being developed for therapy of malignant glioma. With a goal towards optimizing CED of IL13-PE38QQR, this study compared drug distribution assessed by SPECT imaging with that predicted by modeling based on pretreatment diffusion tensor imaging (DTI). Adult patients underwent pretreatment DTI, and patient-specific maps of tissue properties, including molecular diffusivity, hydraulic conductivity, and pore fraction, were constructed. Based on these maps, software simulations of the infusate flow and resulting drug distribution were computed. IL13-PE38QQR was administered by CED along with 123I-human serum albumin as imaging tracer during the first 48-h period of a 96-h infusion. The study involved both intratumoral and peritumoral administration. SPECT scans were performed at 6, 24, and 48 h after the start of infusion. Seven patients were treated, and all seven were assessable. The topography and extent of distribution were assessed by SPECT and were retrospectively compared to the simulated distribution. Results indicate that drug distribution is influenced by anatomic and pathologic properties of the target tissue, such as solid and infiltrating tumor components. Moreover, SPECT imaging data confirms that drug distribution was predicted by the software simulations. These imaging data support a role for predictive modeling in optimizing drug delivery. Further investigation of this DTI-based modeling to improve prospective, patient-specific target tissue selection, as well as resulting treatment efficacy, is warranted.


S.J. See,1 A. Ty,1 and M.C. Wong1,2; 1National Neuroscience Institute, Singapore General Hospital campus; 2Brain Tumour Research Laboratory, Division of Medical Sciences, National Cancer Centre; Singapore

Controversies still surround optimal first-line treatment for patients with newly diagnosed and progressive low-grade oligodendroglioma. Adjuvant radiation following biopsy or surgical resection of low-grade gliomas does not improve overall survival although the interval to progression is prolonged. The remarkable chemosensitivity reported in anaplastic oligodendrogliomas offers the prospect of using PCV chemotherapy to treat patients with symptomatic or progressive low-grade oligodendroglioma in an attempt to delay or avoid late radiation related neurotoxicity. Low-grade oligodendrogliomas are either non-contrast enhancing or have a heterogeneous appearance on MRI. The commonly used McDonald’s response criteria was written for tumors with contrast enhancement. Therapeutic trials in low-grade glioma have used differing response criteria. We report three patients with large, non-resectable, low-grade oligodendroglioma treated with PCV. The patients, aged 33, 35, and 51 years, had either progressive or symptomatic tumors. One of the two patients with progressive disease received radiation therapy 4 years earlier. All were chemo-naive. Histologic confirmation of low-grade oligodendroglioma was obtained within a 6-week period prior to initiation of chemotherapy. All had non-contrast enhancing tumor on MR imaging. Longest diameter of the tumors was 45, 80 and 99 mm prior to chemotherapy. During chemotherapy, one patient improved neurologically, the other two patients remained minimally symptomatic from occasional seizures. Serial MR images of these patients are shown. Dramatic radiographic improvements to PCV was appreciated by visual inspection of the MR images. These responses were seen within the first three courses of PCV. However, attempts to quantify improvement, by applying MacDonald’s response criteria to the areas of T2 signal abnormality, were fraught with difficulty. One patient progressed after 26 months. The other 2 patients remain well at 10 and 22 months. The MR images highlight the following: (1) low-grade oligodendrogliomas can respond to PCV chemotherapy, (2) presence of large tumor load (longest diameter > 60 mm) does not preclude response to PCV chemotherapy, (3) responses can be seen in the first few months of treatment and (4) improved radiographic response criteria for non-contrast enhancing tumors is needed. Magnetic resonance images of the three patients will be shown.


R. Soffietti, A. Costanza, A. Ducati, E. Laguzzi, M. Nobile, U. Ricardi, and R. Rudà; Neuro-Oncology, Neurosurgery and Radiotherapy, University and S. Giovanni Battista Hospital, Torino, Italy

There is an increasing tendency in clinical practice to use temozolomide (TMZ) as first-line treatment in recurrent malignant gliomas, reserving nitrosourea-based regimens as a salvage option, but few data are available in the literature regarding the results of this strategy. Goals of the study were to investigate in a prospective study the efficacy and toxicity of first-line TMZ followed by second-line PCV in patients with recurrent/progressive malignant gliomas after surgery and conventional radiotherapy. The inclusion criteria of the study were as follows: age, 18 years; biopsy-proven glioblastoma or anaplastic astrocytoma; tumor progression after standard conformal radiation therapy; Karnofsky performance status, 60. TMZ was administered at 200 mg/m2 for 5 days in cycles of 28 days up to tumor progression or unacceptable toxicity; PCV was administered in the standard schedule every 6 weeks. Tumor response was evaluated according to the conventional criteria of MacDonald et al. (1990). Fifty-four patients were assessable. A median of 5 cycles of TMZ (range, 1–18) were administered. Responses to first-line TMZ were as follows: CR 2/54 (4%), PR 9/54 (16%), SD 22/54 (41%), PD 21/54 (39%). Median TTP was 5 months, with a PFS at 6 and 12 months of 30% and 13%. Responding patients (CR + PR) had a TTP of 12 months with a PFS at 6 and 12 months of 65% and 45%. Grade 3–4 toxicity after TMZ included thrombocytopenia (17%), neutropenia (6%), and malaise (6%). Responses to second-line PCV (at tumor progression after TMZ) were as follows: CR 1/54 (2%), PR 3/54 (5%), SD 9/54 (17%), PD 41/54 (76%). Median TTP after PCV was 2.6 months, with a PFS at 6 and 12 months of 25% and 10%. Responding (CR + PR) patients had a TTP ranging from 9.2 to 38 months. Grade 3–4 toxicity after PCV included thrombocytopenia (15%), neutropenia (13%), neuropathy (11%), and skin rashes (6%). Overall median survival after first-line TMZ and second-line PCV was 14 months. TMZ is active as a first-line treatment for recurrent glioblastomas and anaplastic astrocytomas, with an overall response rate of 20% and mild toxicity. Some patients may benefit from second-line PCV, even if unresponsive to previous TMZ.


Eun-ik Son,1 Il-Man Kim,1 and Sang-Pyo Kim2; Departments of 1Neurosurgery and 2Pathology, Keimyung University, Dongsan Medical Center, Daegu, Korea

The authors report a case with a huge temporal schwannoma that originated from the region of the geniculate ganglion of the facial nerve. The technical aspect of our combined method for reconstructing the temporal bone defect and unique pathological findings are discussed with literature reviews. A 58-year-old woman, who had a right-sided progressive facial weakness of 17 years duration, presented with headache and convulsion. She underwent facial reanimation surgery about 13 years ago. The main clinical findings on admission are nearly complete facial paralysis of the peripheral type, progressive mixed hearing loss, diminution of lacrimal secretion, facial hyposthesia, and marked temporalis atrophy. CT scan demonstrated a large, mixed, dense mass in the middle fossa and an extensive bony destruction of the right petrous and tympanic portions. On MR images, this tumor was well encapsulated, 6-cm diameter, heterogenous but strongly enhanced. Radiological evidence of otitis media and mastoiditis was also noted. The tumor was initially approached intradurally for internal decompression and total tumor removal via a small right temporal craniotomy. The tumor had eroded the cortical bone including cochlea, superior semicircular canal, and tegmen tymani, and invaginated the dura mater of the basal and lateral temporal areas. A defect in the floor of middle fossa was repaired by a three-layer reconstruction technique using the fatpericranium-fat graft. The lower margin of the lateral temporal dura was anchored along the upper edge of temporal craniotomy after minimizing the middle fossa dead space with fat packing. Thereafter, tympanomastoid obliteration via retroauricular approach was performed. Eradication of all accessible air cells tracts and mucosa in the petrous pyramid, obliteration of the eustachian tube orifice, closure of the external auditory canal, and fat obliteration of the middle ear and mastoid clefts were carried out. This patient showed no subsequent CSF leak but had transient postoperative meningitis. Histopathological examination of the surgical specimen revealed a marked degenerated change of schwannoma, as an ancient type. A combined transmastoid and middle fossa surgical approach with bidirectional closure could be used to repair the defects in the tegmen and middle fossa dura to prevent postoperative CSF otorrhea.


G.H.J. Stevens, D. Peereboom, R.A. Prayson, J.R. Cook, and E.D. His; Brain Tumor Institute and Division of Pathology and Laboratory Medicine, Cleveland Clinic Foundation, Cleveland, Ohio, USA

Extranodal marginal zone B-cell lymphomas of mucosa-associated lymphoid type tissue (MALT lymphomas) are a well-described type of low-grade B-cell non-Hodgkin lymphoma. They typically arise at mucosal sites such as the gastrointestinal tract, lung, and conjunctiva but also arise in other epithelial containing sites such as skin, salivary gland, and thyroid gland. Unusual sites such as the genitourinary tract and meninges have been reported. We report a case of intraventricular MALT lymphoma. A 53-year-old male was noted to have an intraventricular lesion found in work-up for progressive headaches. Octreotide SPCET scan was negative. He underwent gross total resection on March 10, 2004. Pathology revealed MALT lymphoma. MRI cervical, thoracic, and lumbar spine were negative. Whole-body CT scanning revealed a renal mass, which on biopsy showed an angiolipoma. An Ommaya reservoir was placed and the patient is currently being treated with intrathecal long-acting cytarabine. MALT lymphomas involving the central nervous system are rare. We are aware of 14 cases reported in the English literature, and all but one have been associated with the dura. This is the first reported case arising within the ventricular system. Treatment has often involved a combination of radiation and chemotherapy. These tumors are thought to display indolent behavior and have a favorable prognosis. A review of the literature and pathology will be presented.


A. Taliansky, F. Bokstein, I. Lavon, Y. Shoshan, A. Lossos, E. Linetsky, and T. Siegal; Leslie and Michael Gaffin Center for Neuro-Oncology, Hadassah Hebrew University Hospital, Jerusalem, Israel

Anaplastic oligodendroglioma (AO) is a chemosensitive glial tumor. Chemotherapy with alkylating agents (PCV combination and TMZ) is of proven efficacy in recurrent AO with reported response rates ranging between 50% and 70%. Although the value of adjuvant chemotherapy in newly diagnosed AO still remains to be proven, it has become a common practice to give PCV combination prior to radiotherapy. TMZ has a lower toxicity profile as compared to PCV combination, but the response rate of newly diagnosed AO to this treatment has not been reported. The objectives of this study were to evaluate the response rate and progression-free survival (PFS) of newly diagnosed AO patients treated with TMZ following tissue diagnosis. Adult patients with newly diagnosed and histologically verified AO were eligible for treatment with TMZ as initial therapy. Treatment cycles were given every 28 days for 5 days at a starting dose of 200 mg/m2/d and were continued for 24 cycles unless toxicity or tumor progression required withdrawal of TMZ. MRI evaluations were repeated every 8 weeks. Seventeen patients (8 female, 9 male) with a median age of 49 years and a median KPS of 70% entered the study. In six patients (35%) AO evolved from a previously diagnosed grade II oligodendroglioma. All six patients were observed closely and remained chemotherapy and radiotherapy naïve until tumor progression to AO. Surgical procedures prior to TMZ treatment included a debulking surgery in eight patients and a biopsy in nine. Median number of TMZ cycles given until 4/2004 is 8 (range, 2–22). An objective response was documented in 12 patients (70%) and clinical improvement in 6 (35%). Median time to tumor progression has not been reached yet, with a median follow up of 13.7 months (range, 3–37 months). In six patients disease progression occurred after 3, 6, 15, 18, and 35 months. Four of these patients underwent re-operation, and all six patients received radiation therapy. PFS is 87.4% at 6 months and at 12 months. Eight tumors (47%) were evaluated for 1p/19q loss. Five showed deletion of 1p (63%), seven had loss of 19q (88%), and one tumor had no deletions. The latter is the only one of the eight patients who showed no response to treatment and progressed after 3 months. TMZ was well tolerated with only one grade 3/4 toxicity. Adverse effects included GI complaints, thrombocytopenia and neutropenia, mild lethargy, and allergic skin reaction. Our preliminary results show that newly diagnosed AO patients have a high rate of response to initial therapy with TMZ, similar to the response reported for PCV combination. Treatment is well tolerated, and most patients can complete therapy as scheduled. Further studies are needed to determine the optimal duration of treatment and whether radiotherapy should immediately follow chemotherapy.


Mizuhiko Terasaki, Takashi Tokutomi, Etsuyo Ogo, Naohisa Miyagi, Shintaro Fukushima, Yasuo Sugita, Toshi Abe, and Minoru Shigemori; Department of Neurosurgery and Radiology, Kurume University School of Medicine, Kurume; Department of Pathology and Biodefense, Saga Medical School, Saga; Japan

The objective of this study was to describe the clinical characteristics and outcomes of patients with insular diffuse astrocytoma variant (IDAV), defined as particular findings of MRI characterized by diffuse low-grade astrocyte infiltrate around the sylvian fissure at onset, high rate of malignant transformation, and specialized eloquent area involvements at progression. The clinical characteristics and outcomes of four patients treated with radio- or chemotherapy under the diagnosis of insular low-grade astrocytoma were analyzed retrospectively. All patients were diagnosed in the third or fourth decade of life. Three of the four patients presented with seizure onset at diagnosis. No patients responded to initial adjuvant therapy. Within 2.1 to 3.7 years (mean, 3.1 years) after pathological diagnosis, three of four patients developed speech disturbance or neuropsychological difficulties corresponding to newly enhanced lesions on MRI. After tumour excision, tissue analysis showed defined evidence of anaplastic astrocytoma. A rate to response for the second-line adjuvant therapy with a platinum-based regimen was 25%. We conclude that IDAV is frequently transformed to high-grade astrocytoma. Treatment strategy should be carefully considered to control the disease. Synthetic functional analysis is necessary to perform aggressive excision.


A. Ty,1 J. Rao,2 J.B.K. Khoo,3 S.J. See,1 and M.C. Wong1,4; 1National Neuroscience Institute, Singapore General Hospital campus; 2National Neuroscience Institute, Tan Tock Seng Hospital campus; 3Division of Diagnostic Radiology and 4Brain Tumour Research Laboratory, Division of Medical Sciences, National Cancer Centre; Singapore

The objective of this study was to determine the outcome of patients with oligodendroglial tumors treated with procarbazine, CCNU, and vincristine (PCV) chemotherapy at the Singapore General Hospital and National Cancer Centre. Between January 1998 and December 2003, patients with oligodendroglial tumors treated with PCV were evaluated for treatment response, adverse events, and durability of response. Prospective MR images were obtained after 2 to 3 cycles, and after 5 to 8 cycles of PCV. Regular assessment of neurological status was performed. Response was graded according to McDonald’s criteria. Patients with pathologically proven oligodendroglioma and anaplastic oligodendroglioma were studied (n = 15). Median age at diagnosis was 43 years. Pre-chemotherapy median Karnofsky performance score was 90%. Seizures were the most common presenting symptom, followed by headache and limb weakness. Overall objective response rate (CR + PR) after 6 to 8 cycles of treatment was 47% (7/15). The response rate for the oligodendroglioma group (n = 7) was 71% (5/7; 1 CR, 4 PR). Twenty-nine percent (2/7) had stable disease. The response rate for the anaplastic oligodendroglioma group (n = 8) was 25% (2/8; 1 CR, 1 PR) and 75% (6/8) had stable disease. Adverse events were mainly haematological. Twenty percent (3/15) of patients developed grade 3–4 leukopenia, and 13% (2/15) developed grade 3 thrombocytopenia. All 15 patients underwent dose reduction by the third or fourth cycle. Patients received a median of 7 cycles (range, 5–9 cycles) of PCV chemotherapy. To date, amongst the oligodendroglioma group (at median 19 months; range, 10–32 months from start of chemotherapy), 86% (6/7) remain well. Thirteen percent (1/7) progressed at 26 months (14 months after cessation of 8 cycles of PCV). Amongst the anaplastic group, 62% (5/8) remain well (median, 22 months; range, 16–53 months), whilst 38% (3/8) have progressed, with treatment failure at 22, 36, and 37 months from start of chemotherapy. Disease progression was seen 6, 20, and 19 months after completing 7, 8, and 7 cycles of chemotherapy respectively. Notably, of the four treatment failures, two had leptomeningeal metastases prior to chemotherapy. We conclude that in our Singaporean and Asian population, most oligodendroglial tumors respond with stable disease or better to PCV chemotherapy. Durable responses were observed in this group of patients. Dose-intensity-limiting, hematological adverse events are common. More efficacy and toxicity data should help optimize brain tumor chemotherapy regimens in Asian patients.


M. van den Bent, A.A. Brandes, A. van Oosterom, C. Dittrich, P. Fumoleau, B. Coudert, C. Twelves, C. de Balincourt, D. Lacombe, and E. Raymond; European Organization for Research and Treatment of Cancer (EORTC), Brussels, Belgium

Autocrine activation of PDGF alpha and beta receptors yields strong mitogenic effects and activates tumor angiogenesis in malignant gliomas. Pre-clinical data showed tumor growth inhibition of ras and v-sis transformed BALB/c, 3T3U87, and U343 human glioma xenografts in mice using imatinib mesylate. To assess the antitumor activity (measured by response and 6-month PFS) and the safety of imatinib mesylate in patients with histologically proven, CT-scan- or MRI-documented measurable recurrent glioblastoma, stable or decreasing doses of steroids, no more than one prior chemotherapy regimen, and no surgery or radiotherapy within 3 months prior to enrollment were entered. Imatinib mesylate was given until tumor progression at the daily dose of 600 mg (group A) and 800 mg (group B) in two cohorts of patients. A Fleming one-sample/one-stage testing procedure was used. A total of 51 patients (M/F, 26/25; median age, 54; range, 27–68; PS 0/1/2, 18/26/7) were entered. Thirty-three patients had been exposed to chemotherapy, 50 to radiotherapy, and 45 had prior resection. Nineteen patients were entered in group A (70 cycles) and 32 in group B (113 cycles). Grade 3–4 neutropenia was reported in five patients (1 in group A and 4 in group B) and was associated with fever in three patients. Dose reduction, 5 patients (4 in group A, 1 in group B); dose interruption, 21 patients (11 in group A, 10 in group B). Grade 3–4 nonhematological toxicity consisted of edema (1 patient), skin rash (2 patients), and reversible ALT elevation (4 patients). One patient presented an intratumoral hemorrhage associated with a documented tumor progression (considered not related to treatment). Three patients (2 in group A and 1 in group B) experienced confirmed partial responses that lasted 10, 10, and 12+ months. Responses slowly occurred after 3, 6 and 7 months of drug exposure. Prolonged, > 6-month tumor stabilizations were reported for 5 patients (1 in group A, 4 in group B). Imatinib mesylate as single agent displays promising antitumor activity and good safety profile in patients with recurrent glioblastoma. This study is currently extended to patients with recurrent malignant oligodendroglioma and anaplastic astrocytoma.


Matti Vapalahti,1,6 Arto Immonen,1 Kristiina Tyynela,2 Heleena Hurskainen,1 Anu Sandmair,1,3 Ritva Vanninen,4 Gillian Langford,7 Neil Murray,7 and Seppo Yla-Herttuala3,5,6; Departments of lNeurosurgery, 2Oncology, 3Molecular Medicine, A.I. Virtanen Institute, and 4Radiology, and 5Medicine, University of Kuopio, Kuopio, Finland; 6Gene Therapy Unit, Kuopio University Hospital, Kuopio, Finland; 7Ark Therapeutics Ltd., London, UK

High-grade glioma is a rapidly fatal brain tumour with no effective treatment. AdvHSV-tk, an adenoviral vector encoding the Herpes Simplex virus-thymidine kinase gene (EGO09, Cerepro), when used with subsequent ganciclovir may be a new effective therapy. Thirty-six patients (24 male, 12 female; age range, 35–75 years) presenting at a single centre with operable primary or recurrent high-grade glioma between 1997 and 2002 were randomised to receive either AdvHSV-tk gene therapy (3 × 1010 pfu) by local injections into the wound bed after the tumour was resected, followed by intravenous ganciclovir (5 mg/kg twice daily for 14 days) (n = 17) or tumour resection alone (n = 19). Patients in both groups with primary tumours received postoperative radiotherapy. AdvHSV-tk therapy increased mean survival by 81%, from 39.0 ± 19.7 (SD) in control patients to 70.6 ± 52.9 weeks in patients in the active group (log-rank regression P = 0.0095). Median survival increased from 37.7 to 62.4 weeks, an increase of 65%. The therapy was well tolerated as assessed by adverse events, clinical chemistry, haematology, and immunology. There was no evidence of any deterioration in quality of life or increased use of concomitant medications. AdvHSV-tk gene therapy with ganciclovir is a new, potentially effective therapy for operable high-grade glioma.


Michael A. Vogelbaum, David Peereboom, Glen H.J. Stevens, Gene H. Barnett, and Cathy Brewer; Brain Tumor Institute and Department of Neurosurgery, Cleveland Clinic Foundation, Cleveland, Ohio, USA

Erlotinib (Tarceva, OSI-774) is an orally available, small-molecule epidermal growth factor receptor (EGFR) tyrosine kinase inhibitor. We have evaluated, in a single center Phase II trial, the response rate of erlotinib in single agent therapy of recurrent glioblastoma multiforme (GBM). Patients with documented recurrent or progressive GBM who had received previous radiation therapy and cytotoxic chemotherapy were eligible for enrollment. No enzyme-inducing anti-epileptic agents were allowed. Patients were treated with 150 mg of erlotinib per day until tumor progression or study withdrawal. Tumor response was determined by MRI. Analysis for EGFR amplification was performed. Response data are available for 24 patients as of 5/2004; an additional 6 patients will have response rate data available prior to 11/2004. Of the 24 patients, 5 patients have shown partial responses (PRs), 1 showed a PR of his original tumor but also developed a separate unresponsive focus, and 5 have had disease stabilization for greater than 3 months (SD). Ten patients have had MRI-confirmed tumor progression (PD) within 3 months of starting erlotinib, and an additional 3 patients were taken off study because of neurological deterioration but without MRI evidence of tumor progression. Although a response has been seen in 10 of 24 patients (5 PR + 5 SD), the responses have not been durable; the median time to progression of responders has been 154 days after the start of therapy (N = 7). Seventeen patients have died since the start of the study. The pattern of subsequent treatment failure in responders has suggested a diffuse spread of tumor (e.g., gliomatosis cerebri or leptomeningeal spread). Although these results are preliminary in nature, we are encouraged by the response rate observed to date. The relationship of response to EGFR amplification and/or mutation will be discussed. The apparent lack of durability of response and the pattern of post-response failure may be due, in part, to limited drug penetration into the brain, an inadequate dose, and/or tumor heterogeneity.


F.D. Vrionis, M. Wotoczek-Obadia, M. Sullivan, and A. Hamm; Neuro-Oncology Program, H. Lee Moffitt Cancer Center & Research Institute, Department of Interdisciplinary Oncology, University of South Florida, College of Medicine, Tampa, Florida, USA

Surgery for metastatic tumors to the spine remains an important part of the treatment armamentarium. Maximum tumor resection with a minimum number of complications is one of the goals of surgery. Current surgical procedures include tumor resection and spinal stabilization for optimal results. The records of 96 patients who underwent surgery for a metastatic spine tumor at the H. Lee Moffitt Cancer Center & Research Institute were reviewed. Spinal instrumentation was used in the majority of patients. Ambulatory status was maintained in 91% and pain improved in 94% of patients. Complications included infection (5.2%), cerebrospinal fluid leak (2%), and delayed hardware failure (3.1%). The mortality rate was 4.1%; the main cause was due to tumor progression. Surgery is indicated in a select group of patients with metastatic tumors to the spine. A multidisciplinary approach is recommended for patient selection and complication avoidance. Surgical options, including approach, type of reconstruction, and extent of resection (including en bloc spondylectomy), need to be addressed for optimal outcomes.


F.D. Vrionis, M. Wotoczek-Obadia, M. Sullivan, and A. Hamm; Neuro-Oncology Program, H. Lee Moffitt Cancer Center & Research Institute, Department of Interdisciplinary Oncology, University of South Florida, College of Medicine, Tampa, Florida, USA

Kyphoplasty is a newly introduced procedure for the treatment of painful osteoporotic fractures of the spine. Limited experience suggests that this procedure could be also safely performed in patients with cancer. The medical records of 50 patients with cancer-associated fractures who underwent 56 consecutive kyphoplasties at the H. Lee Moffitt Cancer Center & Research Institute over the last 18 months were reviewed. The majority of patients (46%) suffered from multiple myeloma. Other malignancies included lung, breast, prostate, colon, esophageal, uterine, and sarcoma origins. One hundred twenty-eight spinal segments were treated, or approximately 2.56 per patient. Sixty-five percent of cases involved the thoracic spine, with the remainder involving the lumbar spine. There was one complication of a patient who experienced asystole at the end of the operation, requiring CPR. Complete or significant pain relief was obtained in 96% of patients. Most patients (69%) were discharged within 24 h. Kyphoplasty can be a safe and beneficial procedure in cancer patients who present with painful compression fractures of the spine. Kyphoplasty, due to its minimally invasive nature, should be considered as the procedure of choice for this patient population. Absence of radiographic and/or clinical spinal cord compression is a prerequisite for a kyphoplasty.


R. A. Weaver1, S. R. Burzynski1, M. Bestak2, R. I. Lewy3, T.J. Janicki4 , B.G. Szymkowski1, G.F. Jurida2 , M.I. Khan,5 and V. Dolgopolov5; 1Departments of Internal Medicine, 2Pediatric Oncology, 3Medical Oncology, 4Medical Documentation, and 5Radiology, deceased, Burzynski Clinic, Houston, Texas, USA

A Phase II study was conducted to evaluate the efficacy and safety of ANP in glioblastoma multiforme (GBM) which recurred or progressed post surgery, radiation therapy, and/or chemotherapy. Of 22 assessable patients who participated in the study , 6 were men and 16 were women, ranging in age from 27 to 63 with a median age of 47. Multicentric tumors were found in five patients. Previous treatments included surgery in 20 patients (10 patients had a tumor resection once, 8 patients had tumor resections performed twice, and the 2 remaining patients underwent biopsy only), radiation therapy in all patients, and chemotherapy in 10 patients. At least 8 weeks elapsed between initiation of ANP and previous radiation therapy and 6 weeks from chemotherapy with nitrosoureas. ANP was given in escalating doses by intravenous bolus injections. The median duration of ANP administration was 4.3 months, and the average dosage of Antineoplaston A10 was 6.37 g/kg/d and Antineoplaston AS2-1 was 0.24 g/kg/d. Responses were assessed by gadolinium-enhanced MRI repeated every 8 weeks and PET scan. The primary objective of the study, progression-free survival (PFS), was measured from the date of initiation of ANP to the date of progressive disease, death, or the last contact with patient and at six months was 50%. The median overall survival was 9.2 months, and maximum overall survival is over 8 years from initiation of ANP. A complete response (CR) was obtained by 4.5%, partial response (PR) by 4.5%, stable disease (SD) by 54.5% (CR + PR + SD = 63.5%), and progressive disease (PD) by 36.5% of patients. ANP was well tolerated, and there were no drug-related deaths. There was only one case of a grade 3 toxicity of anemia, and grade 2 toxicities included two cases of somnolence and hypernatremia and single occurrences of fatigue, fever, nausea, vomiting, headache and tinnitus. ANP contains three active ingredients: sodium salts of phenylacetylglutamine (PG), phenylace-tyisoglutamine (isoPG), and phenylacetic acid (PN). The proposed antineo-plastic activity of ANP in GBM consists of targeted therapy affecting the AKT2 and the TGFB1 pathways (PG), RAS, TP53, and p21 (PN), and apoptosis (PG and isoPG). In conclusion, ANP is well tolerated and provides encouraging survival and objective response data to warrant Phase III studies in GBM.


P.Y. Wen,1 W.K.A. Yung,1 K. Lamborn,1 B. Peng,2 P. Dahia,3 L. Abrey,1 J. Raiser,1 T. Cloughesy,1 K. Fink,1 M. Gilbert,1 S. Chang,1 L. Junck,1 D. Schiff,1 F. Lieberman,1 H. Fine,1 M. Mehta,1 H.I. Robins,1 L.M. DeAngelis,1 K. Hess,1 M. Groves,1 V.K. Puduvalli,1 V. Levin,1 C. Conrad,1 J. Kuhn,1 E. Maher,1 M. Hayes,2 S. Silberman,2 L. Letvak,2 R. Capdeville,2 R. Kaplan,4 A. Murgo,4 C. Stiles,3 and M.D. Prados1; 1North American Brain Tumor Consortium, The University of Texas, M.D Anderson Cancer Center, Houston, Texas; 2Novartis, East Hanover, New Jersey; 3Dana Farber Cancer Institute, Boston, Massachusetts; 4Cancer Therapy Evaluation Program, NCI, Bethesda, Maryland; USA

Imatinib mesylate is an inhibitor of platelet-derived growth factor (PDGF) receptors. PDGF and its receptors are frequently expressed together in malignant glioma (MG), raising the possibility that an autocrine/paracrine loop contributes to the pathogenesis of these tumors. In preclinical studies, imatinib inhibited the growth of glioblastoma (GBM) cell lines in vitro and in vivo (Kilic et al., 2000, Cancer Res. 60:5143), suggesting that it may have therapeutic potential in patients with MG. The North American Brain Tumor Consortium (NABTC) conducted a phase I/II study of imatinib in patients with recurrent MG (NABTC 99-08). Patients were assigned to one of two groups, according to whether they were taking enzyme-inducing anti-epileptic drugs (EIAED) or not (non-EIAED). In the phase I component, recurrent MG patients who received therapy for no more than three prior relapses were eligible. Patients in each group were independently escalated in standard cohorts of 3 from 400 mg/day to a maximum of 1200 mg/day. The phase II component was restricted recurrent MG patients who received therapy for no more than two prior relapses. Forty-seven eligible patients were enrolled in the phase I study (27 EIAED; 20 non-EIAED). There were 32 GBMs and 14 anaplastic gliomas (AGs). The MTD for non-EIAED patients was 800 mg/day. DLTs were neutropenia, rash, and elevated SGPT. EIAED patients received up to 1200 mg/day of imatinib without developing DLT. There were six SD and one PR. Plasma imatinib concentrations were determined by using HPLC/MS/MS assay. Imatinib was rapidly absorbed after oral administration with a mean tmax of 2 to 4 h. In EIAED patients, the mean Cmax and dose-normalized AUC of imatinib decreased by more than 50% (103.4 vs. 22.9 ng.h/ and mean t½ decreased from 14 to 7.8 h. Forty-eight eligible patients on non-EIAED were enrolled into the phase II component (29 GBM, 19 AG). Patients were treated initially with 800 mg/day of imatinib; 15 patients with AG received 600 mg/day after hemorrhages were observed. Of the 29 GBM patients, there were 0 CR, 1 PR, and 6 SD. The 6-month progression-free survival (PFS) was 3%. Among the 19 AG patients, there were 0 CR, 0 PR, and 2 SD. The 6-month PFS was 11%. Five patients in the phase II study (4 GBM, 1 AG) developed intratumoral hemorrhages in the setting of progressive disease. In the remaining patients, imatinib was generally well tolerated. Single agent imatinib appears to have only minimal activity in MG and may be associated with a slightly increased risk of intratumoral hemorrhage. The evaluation of the activity of combination regimen incorporating imatinib is under way in phase II trials.


P.Y. Wen,1,2 D.C. Gigas,1 L. MacDonald,1 T.T. Batchelor,3 D. Schiff,4 R. Xu,1 N. Ramakrishna,1,2 S. Weaver,5 J. Kracher,3 J. Bradshaw,1 B. Levy,1 E. Bailey,1 S. Kesari,1 E.A. Maher,1 J. Henson,3 and P.M. Black; 1Dana Farber Cancer Institute, Boston, Massachusetts; 2Brigham and Women’s Hospital, Boston, Massachusetts; 3Massachusetts General Hospital, Boston, Massachusetts; 4University of Virginia, Charlottesville, Virginia; 5Albany Medical Center, Albany, New York; USA

There is increasing evidence that the combination of chemotherapy with inhibitors of angiogenesis may have synergistic effects. We conducted a phase II study to determine if the combination of temozolomide (TMZ) with two inhibitors of angiogenesis, thalidomide (T) and celecoxib (C), administered following standard external beam radiation therapy (RT), would prolong survival in patients with newly diagnosed glioblastomas. Eligibility criteria included age >18; KPS = 60; no prior chemotherapy, investigational treatment, Gliadel wafers, or radiosurgery; normal hematologic parameters; normal renal and liver function; stable disease on MRI following RT; initiation of therapy within 5 weeks of completion of RT; and informed consent. TMZ was administered at a dose of 150 mg/m2 daily for 5 days for the first 4 weeks and then, if tolerated, at a dose of 200 mg/m2 daily for 5 days every 4 weeks. Thalidomide was started at 200 mg daily and increased by 100 mg every week to a maximum of 1200 mg/day, as tolerated. Celecoxib was started at a dose of 200 mg twice daily and increased over a week to 400 mg twice daily. Patients were seen at 2 weeks and 4 weeks and then at 4 weekly intervals. Neuro-imaging was performed every 8 weeks. Patients were treated until tumor progression, development of unacceptable toxicity, or for 1 year with TMZ, and indefinitely with T and C. Serum was collected for measurement of angiogenic peptides. To date 49 patients have been enrolled (19 female, 30 male). Median age is 54 (range, 30–78) and median KPS 90 (range, 70–90). Toxicities included neutropenia (1 grade 3, 1 grade 4, 1 grade 5), thrombocytopenia (1 grade 3, 1 grade 4), leukopenia (2 grade 3), 1 grade 4 infection, 3 DVT, and 2 PE. There was one treatment-related death from neutropenic sepsis. Fatigue was common but never more than grade 2. Fifteen percent of patients had PR, 2% CR, 7% MR, 51% stable disease, and 25% progressed at their first scan. Forty-nine percent of patients were alive at 1 year, and median survival was 11.8 months. Median progression-free survival was 4.5 months. To date 16 patients (32%) have survived at least 18 months, and 9 patients (18%) have survived for more than 2 years. In summary, this is a reasonably well tolerated regimen. Although median survival is not significantly prolonged compared to standard therapy, the proportion of long-term survivors appears to be increased. This study is supported by Celgene and Schering Plough Corporations.



Alberto Broniscer, Murali Chintagumpala, Daniel Bowers, Mehmet Kocak, Davonna Ledet, Thomas Merchant, Maryam Fouladi, Raja B. Khan, Matthew Krasin, Larry Kun, and Amar Gajjar; St. Jude Children’s Research Hospital, Memphis, Tennessee; Texas Children’s Hospital, Houston, Texas; Children’s Medical Center, Dallas, Texas; USA

We conducted a multi-institutional study in children with newly diagnosed HGG and unfavorable LGG consisting of an optional window therapy (2 cycles of CPT-11), followed by RT and 6 cycles of TMZ. Eligibility criteria included a histological diagnosis of HGG or unfavorable low-grade glioma (bi-thalamic or gliomatosis cerebri); age between 3 and 21 years; adequate bone marrow, hepatic, and renal functions; and treatment enrollment within 28 days of definitive surgery. Each cycle of CPT-11 [(20 mg/m2 daily × 5) × 2] was given every 3 weeks. The recommended local RT dose was 59.4 Gy. TMZ was started 4 weeks after the end of RT (5-day schedule of 200 mg/m2/day) with an interval between cycles of 21 to 28 days. Thirty-one eligible patients were treated in this study. Median age at diagnosis was 12.3 years (range, 3.4–21 years). Histological diagnoses included glioblastoma multiforme (48%) and anaplastic astrocytoma (32%). Only two patients had unfavorable LGG (both bi-thalamic tumors). Forty-five percent of the tumors were hemispheric, and the remaining originated in midline structures. Only six patients (19%) underwent a radical resection. Eighteen patients with HGG started window therapy. Only 14 patients completed 2 cycles of CPT-11, and four of them had no assessable disease at the start of window therapy. No objective responses to CPT-11 were observed among 10 assessable patients who completed two cycles (6 SD, 4 PD). Twenty-three patients received 112 cycles of TMZ. TMZ dose was reduced in 10 patients because of myelosuppression. Neutropenia and thrombocytopenia grade 3/4 were seen in 29% and 25% of all cycles, respectively. Only three patients have not experienced disease progression. The 1- and 2-year PFS values were 43% ± 9% and 10% ±5%, respectively. The 1- and 2-year OS values were 63% ±8% and 20% ±7%, respectively. In multivariate analysis, radical resection was the only risk factor associated with improved PFS (χ2 = 0.045). This regimen of CPT-11 showed no activity against pediatric HGG. The addition of TMZ did not impact on the outcome of this patient population.


S.R. Burzynski,1 R.A. Weaver,1 M. Bestak,2 R.I. Lewy,3 T.J. Janicki,4 G.F. Jurida,2 B.G. Szymkowski,1 M.I. Khan,5 and V. Dolgopolov5; Departments of 1Internal Medicine, 2Pediatric Oncology, 3Medical Oncology, 4Medical Documentation, and 5Radiology, deceased, Burzynski Clinic, Houston, Texas, USA

Patients with diffuse intrinsic brain stem gliomas (DBSG) have poor survival, which has not substantially changed despite various new treatments. ANP consists of three active ingredients including sodium salts of phenylacetylglutamine (PG), phenylacetylisoglutamine (isoPG), and phenylacetic acid (PN). Preclinical data support that the mechanism of antineoplastic activity in DSBG involves interruption of signal transmission in the RAS, (PN) AKT2, and TGFB1 (PG) pathways, activation of p53 and p21 tumor suppressor genes (PN) and apoptosis (PG and isoPG). The objective of this study is to determine the long-term survival of assessable patients who participated in Phase II studies with ANP and to identify which patient characteristics would predict their survival. A total of 60 patients received ANP; 46 were diagnosed with recurrent tumor after previous therapy and 14 with progressive DBSG, without prior treatment. A group of 31 patients did not meet admission criteria to the study because of a Karnofsky Performance Status (KPS) of 30 to 50; therefore, these patients were treated under Special Exceptions (SE). The overall survival for 29 patients who were admitted to the study at 2 years was 45% and at 5 years was 24%. The maximum survival is over 15.5 years (high-grade DBSG recurrent after radiation and chemotherapy). Corresponding data for SE patients are as follows: survival at 2 years, 32%; at 5 years, 16%; and maximum survival of 11 years (high-grade DBSG recurrent after radiation and chemotherapy). The progression-free survival (PFS) at one year was 41% for study patients and 29% for SE. Patient PFS is measured from the start date of study drugs to the date of progressive disease (PD), death, or the last contact with the patient. In study patients, a complete response (CR) occurred in 24%, partial response (PR) in 21%, stable disease (SD) in 21%, and PD in 34%. In the SE group, the corresponding results were as follows: CR 16%, PR 6%, SD 52%, and PD 26%. ANP was well tolerated. There were no grade 4 toxicities, and reversible grade 3 toxicities included single cases of allergic skin rash, hypertension, neutropenia, hypokalemia, and anemia. Patients in the ANP study group who were in better clinical condition and had higher KPS (60 or higher) than patients in the SE group had substantially longer overall survival and PFS and higher CR and PR rates. In conclusion, the use of ANP for DBSG compares favorably with standard radiation therapy, which offers less than a 10% survival at 2 years for newly diagnosed DBSG.


Jonathan Finlay, Vandana Batra, J. Russell Geyer, Richard Sposto, and Ian Pollack; The Childrens Center for Cancer and Blood Diseases, Childrens Hospital Los Angeles, California; Childrens Oncology Group, Arcadia, California; USA

Between 1985 and 1992, 250 children were enrolled on the CCG-945 trial for high-grade gliomas (HGGs) of the brain and spinal cord. This trial randomized children >3 years old (yo) to either the “standard” pCV or experimental “8-in-1” chemotherapy regimens, with local field irradiation (RT). Children <3 yo at diagnosis were non-randomly assigned to the 8-in-1 regimen, with delay or avoidance of RT. Eighty-seven children <6 yo (34.8% of all patients) were enrolled with institutional diagnoses of HGG, 51 being <3 yo. Forty-three percent of children <3yo and 38.8% aged 3–6 yo underwent >90% tumor resection. RT was delivered to 100% of the 3-to 6-yo children at diagnosis, while 21.6% <3yo received RT at some time point. Upon centralized review by five expert neuropathologists, 41.1% of the <3-yo children and 41.7% of the children aged 3 to 6 yo had diagnoses other than HGG. Fifty-five of 87 (63%) <6-yo children had MIB-1 proliferation index analyses performed centrally on the original tumor tissue; 67%, 53%, and 56% of children with MIB-1 indices, respectively, of <8%, 8%–18%, and >18% were <3 yo at diagnosis. Forty-seven of 87 children (54%) underwent centralized p53 mutation analyses, and 49 of 87 (56%) had centralized p53 expression analyses of the original tumor tissue performed. Fifty-three percent of children with p53 mutations were <3 yo, while 66% of children without p53 mutations were <3 yo. Fifty-seven percent of children with no p53 overexpression (0–1 copy number) were <3 yo, while 58% of children with p53 overexpression (2 or more copy number) were <3 yo. The MIB-1 indices and p53 expression and mutation profiles observed were not significantly different in children <3 yo and 3 to 6 yo. For children <3 yo, the 10-year event-free survival (EFS) and overall survival (OS) are 27% ± 6% and 37% ± 7%. For patients 3 to 6 yo, the 10-yr EFS and OS are 33% ± 8% and 42% ± 8%. For children with eligible, reviewed HGG pathology, those <3 yo had 10-year EFS and OS of 17% ± 8% and 26 ± 8%, while those 3 to 6 yo had 10-yr EFS and OS of 24% ± 9% and 24% ± 9%. There was no survival advantage for children <3 yo who received RT, either for the total population or for those with reviewed eligible HGG pathology. We conclude that there is a high discordance rate between institutional and consensus-reviewed diagnoses of HGG amongst children <6 yo at diagnosis. The long-term (10-year) survival for children <3 yo with reviewed diagnoses of HGG, treated with chemotherapy and avoidance of irradiation, is not different from the survival of children 3 to 6 yo treated with irradiation and chemotherapy at diagnosis. Differences in either treatment or tumor biology are not found, which might otherwise suggest that the younger group have an intrinsically more favorable outcome. A radical change in diagnostic and treatment strategies for young children with HGG is indicated by these findings.


Raja B. Khan, Jeffrey E. Schmidt, and Robert F. Tamburro; Divisions of Neurology and Neuro-Oncology and Department of Critical Care, St. Jude Children’s Research Hospital, Memphis, Tennessee, USA

Movement disorders as a complication of tumor or treatment are rare. We report five males and one female intensive care unit (ICU)–treated children with cancer who developed a generalized movement disorder. Median age was 2 years at the movement disorder onset (range, 1–6 years). Primary tumor was medulloblastoma in three, and one each had atypical teratorhabdoid tumor, hemophagocytic lympho-histiocytosis, and disseminated neuroblastoma; three had leptomeningeal disease at tumor diagnosis. Three children were post high-dose chemotherapy with stem cell rescue (−1 to +50 days). Median ICU stay at movement disorder onset was 21 days (range, 8–134). Five were receiving midazolam and narcotic (fentanyl or morphine) infusions for a median of 10 days (7–64), and one was a day after weaning of these drugs. Other medications included vancomycin in all six (7–28 days), amphotericin in five (6–35 days), meropenem in four, and acyclovir in two. Movements included myoclonus in all limbs, dystonic posturing of proximal and distal musculature, facial grimacing, and tongue thrusting. Movements were enhanced after sensory stimulation and resolved during sleep. EEG was obtained in five children during the movements and showed focal and diffuse slow waves without epileptiform activity. Movements gradually subsided in three children who died from progressive disease within 2 months of the movement disorder onset. In the three survivors, movements responded to baclofen and diphenhydramine in one each. Movements resolved gradually over a few weeks in two, and a few days in one. This is the first report of a generalized movement disorder in children with cancer likely resulting from the toxic effects of medications, especially midazolam and narcotics. However, young age, tumor burden, and other drugs may have contributed. It is important to differentiate this syndrome from seizures to avoid inappropriate treatment. Resolution or reduction of movements during sleep and augmentation with stimulation is a clue to the correct diagnosis. There are no long-term effects if the child survives the cancer.


Edward A. Neuwelt, D. Thomas Dickey, Kristin Gilmer Knight, and Leslie L. Muldoon; Oregon Health & Science University, Portland, Oregon, USA

Cisplatin (CDDP) is a widely used chemotherapeutic agent that is highly neurotoxic and ototoxic. Animal studies and clinical trials have shown that thiosulfates can protect against platinum-induced toxicities without reducing antitumor efficacy. To test ototoxicity, Long Evans rats were treated with CDDP 6 mg/kg delivered to the aorta 15 or 30 min after or 4 h before intravenous (IV) infusion of saline or N-acetylcysteine (NAC) 400 mg/kg. This was delivered via a retrograde right external carotid artery infusion, perfusing the vertebral arteries. Using the same method of infusion, CDDP was given 8 h before IV infusion of sodium thiosulfate (STS) 8 g/m2. Auditory brainstem response (ABR) thresholds were tested at 4–20 kHz 7 days after treatment and compared to baseline ABR values. The STS- and NAC-treated rats exhibited no significant change from baseline ABR values, while the saline-treated rats showed marked ototoxicity, especially at higher frequencies. All NAC and STS regimens were otoprotective; the 15-min NAC pre-treatment appeared most effective. As evidenced in weight loss 7 days post-treatment, the rats treated with NAC 15 min prior to CDDP exhibited less overall toxicity than the controls. Similar results of reduced weight loss were found when NAC was given 30 min before CDDP, but not 4 h after CDDP. To test for nephrotoxicity, CDDP 10 mg/kg was given intraperitoneally 15 min after i.v. saline or NAC 400 mg/kg. After 3 days, the NAC-treated rats all had a normal blood urea nitrogen (BUN) (mean ± 21.5 mg/dl) while four out of five saline treated rats had abnormally high BUN (69.8 mg/dl). Similarly, creatinine levels were significantly elevated in the saline-treated animals compared to the NAC group (mean ± 1.42 mg/dl vs. 0.75 mg/dl). This data shows that pre-treatment with NAC can prevent CDDP-induced ototoxicity, nephrotoxicity, and GI toxicity in rats. Future plans include testing the protection against CDDP-induced toxicities by using combinations of thiols and investigating their potential in preventing peripheral neurotoxicities. Clinical studies are under way to test thiol protection in both adult and pediatric patients, without decreasing platinum cytotoxicity of tumor.


R.J. Packer, A. Gajjar, G. Vezina, J. Langston, L. Rorke, P. Burger, P. Robertson, K. Murasko, D. Armstrong, D. Ris, D. LaFond, L. Boyer, and R. Sposto; Children’s National Medical Center, Washington, District of Columbia; Children’s Oncology Group, Arcadia, California; USA

Four hundred twenty-one children between 3 and 21 years of age with “average-risk” MB (nondisseminated and <1.5 cm2 postoperative residual) were prospectively treated with 2340 cGy of CSRT, 5480 cGy of local radiotherapy (RT), and either CCNU, cisplatin, and vincristine (Reg A) or cyclophosphamide, cisplatin, and vincristine (Reg B) chemotherapy during and after RT. Patients were enrolled on study between 12/1996 and 12/2000; 67 (17%) were 4 years of age or less and 205 (51%) were between 5 and 9 years of age. In analysis of 400 patients, at a median follow-up of 44 months, 3-year progression-free survival (PFS) was 85 ± 3% for those treated on Reg A and 83 ± 3% for those treated on Reg B. Ongoing neuroradiographic central review has disclosed “inadequate” imaging studies in nearly 10% and leptomeningeal disease dissemination in 3%. Twenty-one patients have been declared ineligible because of either evidence of dissemination at diagnosis or >1.5 cm2 of residual disease following surgery. Three-year PFS was 89 ± 2% in eligible patients, as compared to 64 ± 14% in ineligible (P = 0.04). Three-year PFS was 57 ± 18 in those with dissemination. Twenty-two percent had posterior fossa mutism. Hearing loss affected 25% of patients. Severe infection was more common after treatment in Regimen B than in Regimen A (18% vs. 29%, P < 0.005). There were two toxic deaths and four possible secondary tumors. Final review of neuroradiographic and neuropathologic eligibility are ongoing, but these results suggest that nearly 90% of appropriately staged “average-risk” patients will have disease control for at least 3 years with “reduced-dose” CSRT and post-CSRT chemotherapy. Staging remains of critical importance and high-risk patients are still inadvertently being placed on decreased-CSRT studies, with poorer outcome. Despite encouraging results, there is significant risk for treatment-related ototoxicity and neurotoxicity.

Articles from Neuro-Oncology are provided here courtesy of Society for Neuro-Oncology and Oxford University Press