|Home | About | Journals | Submit | Contact Us | Français|
Recent studies have demonstrated a high frequency of IDH mutations in adult “secondary” malignant gliomas arising from preexisting lower grade lesions, often in young adults, but not in “primary” gliomas. Because pediatric malignant gliomas share some molecular features with adult secondary gliomas, we questioned whether a subset of these tumors also exhibited IDH mutations.
We examined the frequency of IDH mutations, using real-time polymerase chain reaction and sequencing analysis, in a cohort of 43 pediatric primary malignant gliomas treated on the Children’s Oncology Group ACNS0423 study. The relationship between IDH mutations and other molecular and clinical factors, and outcome, was evaluated.
IDH1 mutations were observed in 7 of 43 (16.3%) tumors; no IDH2 mutations were observed. A striking age association was apparent in that mutations were noted in 7 of 20 tumors (35%) from children ≥14 years, but in 0 of 23 (0%) younger children (p=0.0024). No association was observed with clinical factors other than age. One-year event-free survival was 86±15% in the IDH-mutated group versus 64±8% in the non-mutated group (p=0.03, one-sided logrank test). One-year overall survival was 100% in patients with mutations versus 81±6.7% in those without mutations (p=0.035, one-sided logrank test).
IDH1 mutations are common in malignant gliomas in older children, suggesting that a subset of these lesions may be biologically similar to malignant gliomas arising in younger adults and may be associated with a more favorable prognosis.
High-grade gliomas are among the most therapeutically challenging central nervous system tumors of childhood . Although these neoplasms are histologically comparable to malignant gliomas in adults, they have several molecular features that distinguish them from their adult counterparts. For example, childhood primary malignant gliomas usually lack EGFR amplification or PTEN deletion [2-9], which are characteristic features of adult primary glioblastoma, but often exhibit TP53 mutations, particularly in tumors that arise in older children. Such molecular alterations resemble those observed in so-called secondary adult glioblastomas, which arise in the setting of a prior lower grade lesion [10-14], often in adults younger than 40 years of age, although a similar pattern of histological progression is uncommonly observed in childhood lesions.
Recent studies have demonstrated a high frequency of mutations of the IDH1 and IDH2 genes, which encode the isocitrate dehydrogenase (IDH) enzymes, in adult secondary malignant gliomas [14-24]. These alterations inhibit the normal function of the IDH enzyme in converting isocitrate to α-ketoglutarate , and instead drive the conversion of α-ketoglutarate to R(-)-s-hydroxyglutarate , a metabolite that may contribute to tumor development. IDH mutations likely represent an early step in tumorigenesis because such alterations are also observed commonly in grade II (diffuse) astrocytomas, oligodendrogliomas, and oligoastrocytomas, in some instances preceding the acquisition of other characteristic molecular features, such as TP53 mutations in astrocytomas and 1p/19q deletions in oligodendroglial neoplasms [15-17]. In contrast, IDH mutations are uncommon in primary adult malignant gliomas [14-17], supporting the existence of mechanistically distinct pathways of tumorigenesis.
Although several large surveys of IDH mutations that have included childhood brain tumors have reported that such alterations are uncommon [15, 19, 22, 24, 27], many of these studies have predominantly incorporated pilocytic astrocytomas and medulloblastomas, which are distinct histologically from gliomas that arise in adults. Given the molecular similarities that have been noted between primary pediatric malignant gliomas that arise in older children and secondary malignant gliomas that occur in adults, we questioned whether pediatric tumors exhibited a high frequency of IDH mutations, and if so, whether there were age-related differences in mutation frequency. Our results indicate the IDH mutations are common in malignant gliomas that occur in older children, suggesting that a subset of such lesions may be comparable on a molecular basis to lesions that arise in young adults. In contrast, such alterations are rare in tumors arising in younger children, supporting the existence of age-related pathways of tumorigenesis in childhood.
Tumor samples were obtained from children enrolled on the Children’s Oncology Group ACNS0423 study, which involved post-surgical administration of alkylator-based chemotherapy and irradiation. Institutional IRB approval was required for protocol enrollment. All patients were previously untreated and between 3 and 21 years of age at the time of diagnosis. Maximal surgical resection was encouraged, but was limited in some cases based on the location of the tumor. Adjuvant therapy consisted of daily temozolomide during irradiation followed by post-irradiation temozolomide and lomustine.
Eligibility required an institutional histopathologic diagnosis of anaplastic astrocytoma (AA), glioblastoma (GBM), or gliosarcoma and validation of the diagnosis by central pathology review to exclude patients with discordant histologies, such as low-grade glioma, which has been a concern on previous studies [28, 29]. Tissue accrual for the current study was coordinated by the Pediatric Branch of the Cooperative Human Tissue Network (CHTN). Specimens were de-identified by the CHTN to mask clinical and outcome data from investigators.
Formalin-fixed paraffin-embedded (FFPE) tissue specimens were used for all analyses. Slides were reviewed by a neuropathologist (RLH) to confirm that tumor tissue of sufficient quantity was available for the planned studies. Tumor targets were manually microdissected from 4-μm unstained histologic sections under the guidance of a corresponding hematoxylin- and eosin-stained slide using an Olympus SZ61 stereo microscope (Olympus, Hamburg, Germany). DNA was isolated from each target with the DNeasy Blood and Tissue kit on the automated QIAcube (Qiagen, Valencia, CA) instrument according to the manufacturer’s instructions. The quantity of isolated DNA was assessed using a NanoDrop 1000 spectrophotometer (Thermo Scientific, Wilmington, DE).
Detection of IDH1 and IDH2 mutations was performed using real-time polymerase chain reaction (PCR) and post-PCR fluorescence melting curve analysis (FMCA) on the LightCycler (Roche Applied Science, Indianapolis, IN). A pair of primers flanking each mutation site and two fluorescent probes were designed using LightCycler Design Software 2.6 (Roche, Indianapolis, IN; Table 1). All primers and probes were obtained from TIB Molbiol (Berlin, Germany). Amplification was performed in a glass capillary tube using 5–50 ng of DNA in a 20-μl reaction volume. In detail, 2 μl of 10× LightCycler Master Mix from the LightCycler FastStart DNA Master HybProbe Kit (Roche, Indianapolis, IN, USA) containing PCR buffer, deoxynucleotide triphosphates, 10 mM MgCl2, and Taq polymerase, 1.6 μl of 25 mM MgCl2, 40 pmol of each forward primer, 10 pmol of each reverse primer, and 5 pmol of each hybridization probe was used. The reaction mixture was subjected to 40 cycles of rapid PCR consisting of denaturation at 95°C for 5 s, annealing at 54°C for 20 s, and extension at 72°C for 12 s. Post-amplification FMCA was performed by gradual heating of samples at a rate of 0.1°C/s from 40°C to 95°C. Fluorescence melting peaks were built by plotting the negative derivative of fluorescent signal corresponding to the temperature (−dF/dT). All positive cases were directly sequenced.
Sanger (dideoxy) sequencing analysis was performed as previously described . In detail, IDH1 and IDH2 primers (Table 1) were used for PCR amplification and sequencing analysis. Each PCR reaction was performed using 5–50 ng of DNA, 0.2 μmol of each primer, and AmpliTaq Gold PCR Master Mix (Applied Biosystems, Inc., Foster City, CA) in a total volume of 50 μl. The reaction mixture was subjected to an initial denaturation of 95°C for 10 min, followed by 40 cycles of amplification consisting of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and extension 72°C for 60 s. The PCR products were sequenced in both sense and antisense directions using the BigDye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions and were analyzed on ABI 3130 (Applied Biosystems, Foster City, CA). The sequence electropherograms were analyzed using Mutation Surveyor software (SoftGenetics, LLC., State College, PA). Each case was classified as either positive or negative for the IDH mutation based on the sequencing results.
FFPE specimens were used for all analyses. Slides were reviewed by a neuropathologist (RLH) to confirm that tumor tissue of sufficient quantity was available for the planned studies. Tumor-containing sections were baked at 60°C for 30 min, deparaffinized in xylene, and rehydrated in graded concentrations of ethanol. Endogenous peroxidase activity was quenched by incubation in 0.3% hydrogen peroxide solution. Antigen retrieval  was performed by heating the slides in 10 mmol citrate buffer (pH 6.0) for 20 min. Nonspecific antibody binding was blocked by incubation in Protein Blocking Reagent (Thermo Corp, Pittsburgh, PA) for 20 min. Sections were then incubated with primary antibodies against Ki-67 (Immunotech, Westbrook, ME; USA 1:100), p53 (DO-7, Dako Corporation, Carpinteria, CA, 1:300), and pAkt (Se473) [Cell Signaling Technology, Inc., Danvers, MA; 1:100], in Common Antibody Diluent (BioGenex, San Ramon, CA, USA) at room temperature for 2 h. The p53 antibody recognizes a denaturation-stable determinant of wild-type and mutant p53 . Negative control sections were treated with diluent and mouse IgG (5 μg/ml, Dako Corporation, Carpinteria, CA) alone. Slides were then rinsed twice with PBS and antibody binding was localized using a Universal Labeled Streptavidin-Biotin 2 System (LSAB 2–HRP, Dako). Slides were incubated for 30 min in Biotinylated Link reagent at room temperature, followed by a 10-min PBS wash. Slides were then incubated in Streptavidin–HRP solution for 30 min at room temperature. Antibody binding was visualized using 3,3’-diaminobenzidine . The slides were counterstained with Mayer’s hematoxylin, dehydrated through graded concentrations of ethanol, cleared in xylene, and mounted and examined using a light microscope. Positive controls were included with each batch of sections to confirm the consistency of the analysis.
Because previous studies involving pediatric malignant gliomas have often included a high percentage of cases that were reclassified as low-grade gliomas on blinded central neuropathology review [28, 29], the current study incorporated a panel of neuropathology reviewers (PCB, MKR, DJB) to review the tumor specimens based on contemporary classification criteria [34, 35], restrict the eligible cohort to those cases confirmed to be high-grade gliomas, and establish a central review diagnosis to be used for outcome analyses. The IDH and other correlative analyses were performed independently of this review, so the review diagnosis of each specimen was not known during the molecular analyses.
For outcome analysis, tumors were classified according to the presence or absence of IDH mutations. The endpoints for analysis were overall survival (OS) and event-free survival (EFS). Overall survival was defined as the time from study entry to death from any cause. Event-free survival was defined as the time from study enrollment to the first occurrence of disease progression, relapse, second malignancy, or death from any cause. Nonparametric EFS and OS curves were computed using the product-limit estimates, with standard error via the Greenwood formula . Comparisons of outcome were based on one-sided logrank test. Comparisons of the distribution of IDH mutations between different patient subgroups defined by age, gender, histology, p53 expression, extent of resection, and MIB1 labeling were based on two-sided Fisher exact test .
Of 106 eligible patients enrolled on ACNS0423, 43 patients had specimens that were available for IDH mutation analysis in the current study. Seven of these tumors (16.3%) had mutations of IDH1; none had mutations of IDH2. As a group, the 43 patients who had specimens had a more favorable outcome than the cohort without specimens. None of these patients had a known prior lower grade lesion. Figure 1 shows an illustrative example of a tumor with an IDH1 mutation as contrasted against another case lacking this feature. Four tumors were positive for R132H mutations, two for R132S, and one R132C.
The distribution of IDH mutations as a function of clinical and other molecular factors is shown in Table 2. When the cohort was subdivided as a function of age, it was determined that IDH1 mutations were seen in 7 of 20 (35%) children older than 14 years versus 0 of 23 (0%) children younger than 14 years (p=0.0024). No association between IDH mutations and pAkt immunoreactivity or MIB1 labeling was apparent. Among the seven tumors with IDH mutations, p53 overexpression was seen in 1 (14.3%) versus 20 of the 36 tumors lacking mutations (55.6%; p=0.09). Average MIB labeling in the IDH1 mutated subset was 34.6±11.4 versus 32.0±3.18 in those without IDH mutations (p=0.83). The frequency of particularly high MIB1 indices (>36), which has been previously associated with an adverse prognosis, did not differ significantly between the two groups (3 of 7 versus 10 of 35, p=0.87).
We also examined whether IDH mutations might be more common in grade IV tumors compared to grade III lesions. Contrary to this assumption, IDH mutations were seen in 2 of 24 grade IV lesions (8%) versus 5 of 19 AAs (26%; p=0.21).
To assess the prognostic significance of IDH mutations in pediatric malignant gliomas, overall survival and event-free survival were compared between the subgroup of tumors showing IDH mutations and those lacking this feature. One-year event-free survival was 83±15% in patients with IDH mutations versus 64±8% in those without mutations (Fig. 2; p=0.03, logrank test). One-year overall survival was 100% in patients with mutations versus 81±6.7% in those without mutations (Fig. 3; p=0.035, logrank test). These results show that there is a significant difference in early outcome as a function of IDH mutation status. The median follow-up for non-failures was 1.5 years, and more extended follow-up data will be needed to determine whether IDH mutation status is associated with long-term outcome. Because extent of resection is a known prognostic factor in patients with high-grade glioma, stratified logrank tests were performed based on this factor. After stratifying the data based upon extent of resection (less than 90% versus greater than or equal to 90% resection), the early difference in outcome was still evident (stratified logrank test, p=0.04 for event-free survival and p=0.05 for overall survival).
Recent studies have demonstrated frequent mutations of the IDH1 and IDH2 genes in adult secondary malignant gliomas that arise from preexisting lower grade lesions [14-24]. Alteration in a single IDH1 allele significantly inhibits the normal function of IDH in promoting the oxidative decarboxylation of isocitrate to α-ketoglutarate, with resultant conversion of NADP+ to NADPH . Because α-ketoglutarate enhances the degradation of hypoxia-inducible factor subunit HIF-1α, inhibition of α-ketoglutarate formation increases the levels of HIF-1α, which favors tumor growth under hypoxic conditions. IDH1 mutations also drive the conversion of α-ketoglutarate to R(-)-s-hydroxyglutarate, a metabolite that may contribute to tumor development . The potentially contributory role of this alteration to tumor development is highlighted by the fact that IDH mutations are observed not only in secondary glioblastomas, but also in diffuse gliomas of lower histological grades, in some instances preceding the acquisition of other molecular alterations [15-19]. However, IDH mutations are uncommon in primary adult malignant gliomas, suggesting that these lesions arise by a mechanistically distinct pathway of tumorigenesis.
Although several large surveys of IDH mutations that have included pediatric brain tumors have reported that such alterations are uncommon [14, 15, 19, 22, 24, 27], many of these studies have predominantly incorporated pilocytic astrocytomas and medulloblastomas, which are distinct histologically from gliomas that arise in adults, and a comparatively small number of diffuse low-grade gliomas and malignant gliomas of childhood have been reported. The series of Balss et al.  included only 14 pediatric GBMs, one of which had an IDH1 mutation. Likewise, De Carli et al.  reported IDH mutations in only 4 of 73 children with non-pilocytic gliomas, with an average age of 16 years in the affected subset, but the histological composition of the analyzed tumors was not specified. Yan et al.  noted mutations in 0 of 15 primary pediatric glioblastomas (median age 5). Similarly, in a study of 1,010 diffuse gliomas that included a small subgroup of children, Hartmann et al.  noted that IDH1 mutations were rare in the pediatric subset, occurring in only 4 of 32 (12.5%) tumors, and IDH2 mutations were absent, leading the authors to conclude that IDH mutations were rare in childhood tumors, although the frequency of mutations as a function of age and histology was not provided. More recently, Antonelli et al.  noted no IDH1 mutations in a series of 27 pediatric high-grade gliomas, although the age range extended only to 17 years, and the mean age of their overall cohort was 9.4 years, whereas 20 of our 46 patients were 14 to 18 years of age, suggesting that the composition of our analyzed patient groups may have differed somewhat. In this regard, the incidence of TP53 mutations observed in their patients (18.5%) is also somewhat lower than the frequencies we have previously observed in older children (~40%) , suggesting that there may well be several age-related patterns of tumorigenesis within the broad group of childhood malignant gliomas.
Given the molecular similarities that have been noted between primary pediatric malignant gliomas that arise in older children and secondary malignant gliomas that occur in adults, in terms of their high frequency of TP53 mutations  and infrequency of EGFR amplification and PTEN deletions , we questioned whether a more focused analysis of childhood high-grade gliomas, including samples from both older and younger children, might identify an association between IDH mutations and age in these tumors. Our results indicate that IDH mutations are relatively common in malignant gliomas that occur in older children, although are less frequent than the reported incidence in secondary adult malignant gliomas, which exceeds 80% in some studies [14-18]. This observation supports the possibility that at least a subgroup of pediatric lesions occurring in older children may be biologically comparable to malignant gliomas that arise in young adults , although most lesions in older children appear to arise by a distinct pathway. Our finding that IDH-mutated tumors seemed to be associated with a more favorable prognosis than non-mutated tumors is consistent with observations in adults [14, 15]. In contrast, IDH mutations are rare in tumors arising in younger children, supporting the existence of age-related pathways of tumorigenesis in childhood.
The potential for multiple pathways of tumor development in childhood fits with our previous observations that TP53 mutations are common in malignant gliomas that occur in older children, but are rare in lesions that arise in younger children . The current results indicate that primary malignant gliomas in young children lack not only EGFR amplification, and PTEN and TP53 mutations, but also IDH mutations, implying that these lesions incorporate distinct, but so far, uncharacterized pathways of tumorigenesis. However, for tumors arising in older children, the finding of additional similarities to lesions in young adults has important implications for therapeutic stratification. Although strategies specifically directed toward IDH-mutated tumors have, to date, not been implemented in adults with malignant gliomas, this alteration or its downstream consequences constitute a logical target, and a similar approach could be considered in appropriately selected patients in the pediatric context.
This work was supported in part by NIH grants NS37704 (IFP), CA47904 (WALF), and CA98543 to the Children’s Oncology Group. The authors wish to acknowledge Judith Burnham for her technical assistance.
Ian F. Pollack, Department of Neurosurgery, Children’s Hospital of Pittsburgh, University of Pittsburgh School of Medicine, 4401 Penn Avenue, Pittsburgh, PA 15224, USA, Email: email@example.com.
Ronald L. Hamilton, Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA.
Robert W. Sobol, Department of Pharmacology and Chemical Biology, Hillman Cancer Center, University of Pittsburgh Cancer Institute and University of Pittsburgh School of Medicine, Pittsburgh, PA, USA.
Marina N. Nikiforova, Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA.
Maureen A. Lyons-Weiler, The Clinical Genomics Facility, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA.
William A. LaFramboise, Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA. The Clinical Genomics Facility, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA.
Peter C. Burger, Department of Pathology, Johns Hopkins University, Baltimore, MD, USA.
Daniel J. Brat, Department of Pathology, Emory University, Atlanta, GA, USA.
Marc K. Rosenblum, Department of Pathology, Memorial Sloan-Kettering Cancer Center, New York, NY, USA.
Emiko J. Holmes, Statistical and Data Center, The Children’s Oncology Group, Arcadia, CA, USA.
Tianni Zhou, Statistical and Data Center, The Children’s Oncology Group, Arcadia, CA, USA.
Regina I. Jakacki, Department of Pediatrics, Children’s Hospital of Pittsburgh, University of Pittsburgh School of Medicine, Pittsburgh, PA, USA.