|Home | About | Journals | Submit | Contact Us | Français|
Positron emission tomography (PET) imaging has become a useful tool for assessing early biologic response to cancer therapy and may be particularly useful in the development of new cancer therapeutics. RAF265, a novel B-Raf/vascular endothelial growth factor receptor-2 inhibitor, was evaluated in the preclinical setting for its ability to inhibit the uptake of PET tracers in the A375M(B-RafV600E) human melanoma cell line. RAF265 inhibited 2-deoxy-2-[18F]fluoro-d-glucose (FDG) accumulation in cell culture at 28 hours in a dose-dependent manner. RAF265 also inhibited FDG accumulation in tumor xenografts after 1 day of drug treatment. This decrease persisted for the remaining 2 weeks of treatment. DNA microarray analysis of treated tumor xenografts revealed significantly decreased expression of genes regulating glucose and thymidine metabolism and revealed changes in apoptotic genes, suggesting that the imaging tracers FDG, 3-deoxy-3-[18F]fluorothymidine, and annexin V could serve as potential imaging biomarkers for RAF265 therapy monitoring. We concluded that RAF265 is highly efficacious in this xenograft model of human melanoma and decreases glucose metabolism as measured by DNA microarray analysis, cell culture assays, and small animal FDG PET scans as early as 1 day after treatment. Our results support the use of FDG PET in clinical trials with RAF265 to assess early tumor response. DNA microarray analysis and small animal PET studies may be used as complementary technologies in drug development. DNA microarray analysis allows for analysis of drug effects on multiple pathways linked to cancer and can suggest corresponding imaging tracers for further analysis as biomarkers of tumor response.
RAF265 is a novel, orally dosed, small-molecule B-Raf kinase and vascular endothelial growth factor receptor-2 (VEGFR-2) inhibitor with potent antitumor activity in mutant B-Raf tumor models and is currently undergoing phase 1 clinical trials in melanoma [1,2]. Inhibiting mutant B-Raf as well as VEGFR-2 provides a dual mechanism of action: antiproliferative activity by inhibiting the Ras/Raf/mitogen-activated protein kinase (MAPK) pathway and indirect antitumor activity by inhibiting angiogenesis through VEGFR-2.
2′-Deoxy-2′-[18F]fluoro-d-glucose positron emission tomography (FDG PET) is a widely used clinical imaging test for many cancers and for a wide range of indications . Small animal PET scanners for rodents  has allowed for the assessment of tumor xenograft mouse models with FDG for preclinical oncology research and drug development [5–9]. Several authors have used small animal FDG PET to assess various therapies in mouse tumor xenograft models with FDG and the proliferation tracer 3′-deoxy-3′-[18F]fluorothymidine (FLT) [10–17]. We have recently shown that small animal PET studies are reproducible with moderately low variability, such that serial studies on mouse tumor xenografts are reliable in assessing therapy response [18,19].
DNA microarray analysis is a powerful technique to evaluate the expression of thousands of genes in a single experiment. Recent studies in clinical oncology have used DNA microarray analysis for identifying cancer subtypes, predicting prognosis, predicting therapy response, and understanding cancer biology [20,21]. Recently, several groups have begun to investigate the combination of FDG PET and DNA microarray analysis by correlating imaging findings with gene expression changes [22–27]. The two technologies are complementary and may provide unique insights into tumor biology. DNA microarray analysis of gene expression allows for analysis of multiple genes and multiple pathways but is limited by the need for invasive tissues sampling and may be restricted to a single time point. FDG PET is a non-invasive technology that allows for evaluation at multiple time points in the same subject without the need for invasive pathologic examination; however, it is limited to analysis of a single pathway, namely glucose metabolism albeit a very useful one for most cancers.
In this study, our first objective was to use DNA microarray analysis to suggest pathways affected by RAF265, which have corresponding imaging agents that could potentially serve as imaging biomarkers. Our second objective was to assess whether small animal FDG PET could be used to assess the efficacy of RAF265 in the A375M (B-RafV600E) mouse xenograft tumormodel. We show that RAF265 inhibited the glucose metabolism pathway and was confirmed by inhibition of FDG accumulation both in cell culture and in tumor xenografts.
RAF265 (Novartis, Emeryville, CA) is a novel, orally bioavailable, small-molecule inhibitor of Raf kinase/VEGFR-2 with a molecular weight of 518 g/mol. For cell culture experiments, the drug was dissolved in dimethyl sulfoxide. For in vivo mouse xenograft experiments, the drug was dissolved in polyethylene glycol-400 (PEG-400) to a concentration of 25 mg/ml.
A375M human melanoma cells, which express B-RafV600E, were grown in minimum essential medium with Earle salts and l-glutamine, supplemented with 10% fetal bovine serum, 1 mM sodium pyruvate, and 1x nonessential amino acids (Mediatech, Manassas, VA). MV4;11 human acute myelogenous leukemia cells, which express wild-type B-Raf, were grown in Iscove modified Dulbecco medium with l-glutamine and 25 mM HEPES, supplemented with 10% fetal bovine serum and 5 ng/ml granulocyte macrophage colony-stimulating factor. Cells were grown at 37°C with 5% CO2 in a humidified incubator. For xenograft implantation, cells were resuspended in Hank's balanced salt solution (Mediatech).
Fifty thousand A375M or MV4;11 cells were plated overnight onto 24-well plates. After changing the medium, 1.0 µM RAF265, 0.1 µM RAF265, or no drug (solvent control) was added to the wells in triplicate and incubated for 4 to 5 hours and 24 to 28 hours. Approximately 1 µCi of FDG (PETNET or Stanford Radiochemistry Facility) was added and incubated for 2 hours. Cells were rinsed with cold phosphate-buffered saline (Invitrogen, Carlsbad, CA) and then lysed with 1 N NaOH. Half of the sample was counted for radioactivity using a Cobra II gamma counter (Packard, Meriden, CT). The remaining sample was used to determine protein concentration with a Bradford protein assay (Bio-Rad, Hercules, CA). FDG accumulation was calculated as: FDG accumulation (%/mg) = FDG activity within the cells (cpm) ÷ total FDG added to the sample (cpm) ÷ total protein (mg) x 100%. FLT assays for A375M cells were performed in a similar manner using a 1-hour incubation period for FLT accumulation. FLT was synthesized by the Stanford Radiochemistry Facility with a specific activity greater than 45 TBq/mmol . Proliferation assays on cell lines were performed by incubating cells with RAF265 diluted in complete medium over a concentration range of 0.1 to 20 µM in 96-well plates. After 72 hours of incubation, cells were rinsed, and then the CellTiter-Glo assay was carried out as described by the manufacturer (Promega, Madison, WI).
Animal protocols were approved by the Stanford Administrative Panel on Laboratory Animal Care and the Novartis Institutional Animal Care and Use Committee. Forty-five 8-week-old nu/nu mice (Charles River Laboratory) were injected subcutaneously with 3 million A375M cells in the right flank. After approximately 1 week when xenografts reached a size of approximately 200 mm3, five mice were killed immediately as baseline controls (day 0). Tumors were harvested and frozen. The remaining mice were separated into groups of five. Half of the groups were orally dosed with 100 mg/kg of RAF265 (approximate volume of 100 µl) every 2 days for 14 days. The other half were orally dosed with vehicle PEG-400. Mice were killed, and tumors were harvested at 8 hours, day 1, day 8, and day 14.
RNA was extracted from the tumor xenografts using an RNeasy kit (Qiagen). Complementary DNA and RNA were synthesized. Complementary RNA was labeled and hybridized to a human genome U133 Plus 2.0 microarray (Affymetrix, Santa Clara, CA). Data were normalized using the MAS 5.0 software (Affymetrix).
The Gene Set Enrichment Analysis (GSEA) method [28,29] was used to identify coordinate changes in biologic pathways modulated by RAF265. The effects of RAF265 versus vehicle at each of the four time points were mapped to 4014 publicly available pathways (KEGG [Kanehisa Laboratories, Kyoto, Japan; n = 361 pathways], Panther [SRI International, Palo Alto, CA; n = 1698 pathways], and Metacore pathways [GeneGo, St Joseph, MI; n = 1955]) using an in-house implementation of GSEA as previously described . A subset of those results including pathways with corresponding molecular tracers available is presented. Thirty-four major reference pathways for apoptosis (apoptosis/annexin V) or containing glucose transporters and/or hexokinases (Entrez Gene IDs 6513, 6514, 6515, 6517, 6518, 11182, 155184, 29988, 56606, 81031, 66035, 154091, 114134, 144195, 3098, 3099, 3101, 2645), thymidine kinase (Gene ID 7083), and epidermal growth factor receptor (EGFR; Gene ID 1956) were identified (Table W1). Pathways considered significantly modulated by RAF265 had a q value (reflecting the false discovery rate) ≤ 1 x 10-5 .
A separate group of forty 8-week-old female nu/nu mice was injected subcutaneously with 3 million A375M cells in the right upper flank. When tumors reached an approximate volume of 100 mm3, eight mice were selected for cohort 1 (n = 4 control; n = 4 drug-treated group). When tumors reached an approximate volume of 200 mm3, 12 mice were selected for cohort 2 (n = 6 control; n = 6 drug-treated group). Mice were weighed, and tumors were measured with digital calipers every 2 to 3 days. Tumors volumes were calculated using orthogonal measurements as length x width x width ÷ 2.
The drug dosing and imaging schedules were as follows. Mice were administered 100 mg/kg RAF265 or vehicle (100% PEG-400) by oral gavage every 3 days starting on day 0 and continuing to day 15. For cohort 1 (starting tumor volume = 100 mm3), a day 0 baseline small animal FDG PET scan was performed immediately before the first drug dose, followed by imaging on days 1, 4, 7, 11, and 13. For cohort 2 (starting tumor volume = 200 mm3), mice were imaged on days 0, 1, 4, 7, 9, 10 or 11, 14, and 16. On day 9, imaging was performed before dosing. Owing to the limited availability of tracer, only four mice were scanned on days 9, 14, and 16 for the control group of cohort 2. Data for day 10 or 11 were combined and analyzed together using a middle time point of 10.5 days. One mouse in the drug-treated group of cohort 2 died during scanning on Day 11, thereafter only three mice were scanned on days 14 and 16 for the drug-treated group of cohort 2. The mouse that died had no overt signs of toxicity, such as illness, distress, or significant weight loss. In our experience, a death rate of approximately 1% during scanning is expected, which may be related to anesthesia, tumor burden, repeated tail injections, or a combination of these factors. The single mouse death was most likely related to scanning factors rather than drug toxicity. No body weight loss or other toxicity was observed in the remaining mice.
Imaging was performed as previously described and summarized below [18,19]. After a 4- to 6-hour fast, approximately 200 µCi of FDG was injected through the tail vein. One hour after injection, a 7-minute static prone scan was obtained in a MicroPET R4 (Siemens/CTI, Munich, Germany) without partial volume correction . A sample three-dimensional rendering of a tumor-bearing mouse is included as Figure W1.
Ellipsoidal regions of interest (ROIs) were drawn around the edge of the tumor activity using AMIDE software . The mean, the mean of the upper 20% of voxels, and the maximum activities were recorded. Percent injected dose per gram (%ID/g) and standardized uptake value (SUV) were calculated as follows: %ID/g = ROI activity ÷ injected dose. SUV = ROI activity ÷ injected dose x body weight. Two separate 5-mm background ROIs were drawn from the muscle adjacent to the tumor and in the opposite flank. Tumor-to-background ratios were calculated.
Immunohistochemistry staining was performed on a separate set of tumors treated with a similar drug regimen. Hematoxylin and eosin (H & E) staining was performed to assess areas of viable tissue and necrosis. Ki-67 staining was performed to assess areas of proliferation.
For cell culture and small animal FDG PET studies, t tests were performed to assess for significant differences between two groups of data. For a comparison among three groups of data, analysis of variance was first performed. If the analysis of variance was significant, pairwise comparisons were then performed. P < .05 was chosen to indicate significance. Data were reported with SEM error bars or error values.
A pathway-based analysis of gene expression in A375M xenografts from mice treated with RAF265 showed that RAF265 had profound inhibitory effects on the cell cycle at all time points examined (data not shown). Consistent with this, glucose metabolism, thymidine metabolism, and apoptosis pathways were modulated by RAF265 (Table 1 and Figure 1). A summary diagram of effects of RAF265 on the glucose metabolism pathway is provided in Figure W2. EGFR pathways were not transcriptionally modulated by RAF265 (Table W1). This suggested that FDG (glucose metabolism), FLT (thymidine metabolism), and annexin V (apoptosis) tracers may be useful in RAF265 imaging studies.
A375M cells were incubated with two different concentrations of RAF265 for 5 and 28 hours then assayed for FDG accumulation. At 28 hours, RAF265 treatment resulted in a significant dose-dependent inhibition of FDG accumulation compared with controls (untreated cells; Figure 2A). At 1.0 µM, RAF265 had a more significant decline in FDG accumulation (75.5%, P = .0003) compared with the lower concentration of 0.1 µM (38.2%, P = .02) compared with control. At the earlier 5-hour time point, RAF265 did not show significant inhibition. These results are consistent with potent antiproliferative activity of RAF265 in these cells (IC50 = 0.28 µM).
To determine whether the effect of RAF265 was selective for cells expressing mutant B-Raf, FDG accumulation was also evaluated in the human acute monocytic leukemia cell line MV4;11, which expresses wild-type B-Raf. At 4 and 24 hours, FDG accumulation in drug-treated MV4;11 cells was not significantly different from control at both 1.0 and 0.1 µM (Figure 2B). These results were consistent with the lack of antiproliferative activity of RAF265 on these cells (IC50 = 6 µM).
The DNA microarray data indicating inhibition of thymidine kinase containing pathways by RAF265 suggested that RAF265 should also inhibit the uptake of FLT. Therefore, A375M cells were incubated with two different concentrations of RAF265 for 4 and 24 hours then assayed for FLT accumulation. At the earlier 4-hour time point, RAF265 did not significantly inhibit the uptake of FLT; however, by 24 hours, there was a significant decrease in FLT accumulation at 1.0 µM (P < .0001) but not at 0.1 µM (P = .24; Figure 2C). Because FLT inhibition was only significant at the higher dose of RAF265 and only at 24 hours compared with FDG, it was concluded that FDG may be a slightly more sensitive tracer as an imaging biomarker to test in an in vivo tumor xenograft model. No further testing was performed with FLT.
Nude mice with A375M tumor xenografts were dosed orally with 100 mg/kg RAF265 every 3 days at two different starting tumor volumes of 100 mm3 for cohort 1 and 200 mm3 for cohort 2 (Figure 3A). Tumor volumes for the control mice increased two- to three-fold during the study, whereas the tumor volumes decreased 51.0% and 35.1% for drug-treated cohorts 1 and 2, respectively.
FDG accumulation was assessed during 2 weeks in the same mice. Representative small animal FDG PET images for a RAF265-treated mouse (Figure 4A) and a control mouse (Figure 4B) revealed inhibition of FDG accumulation by RAF265. For the RAF265-treated mice in cohort 1 (Figure 3B), the mean %ID/g of FDG accumulation decreased 28.2% on day 1 compared with the baseline day 0 and showed a statistically significant difference compared with the cohort 1 control group (P = .0001). The FDG accumulation continued to decrease for the drug-treated group, reaching a maximum decrease of 51.0% at day 13. The drug-treated group remained significantly decreased from the control group at all time points after day 0.
For cohort 2, the mean %ID/g of FDG accumulation decreased in the drug-treated group on day 1 by 30.9% compared with the baseline day 0 (Figure 3B). FDG accumulation was significantly lower in the drug-treated group than the control group on day 1 (P = .04) and day 4 (P < .0001). After day 7, FDG accumulation was lower in the control group compared with the RAF265-treated group. Visual inspection of the images revealed central photopenia, suggestive of tumor necrosis, in tumors greater than approximately 300 mm3. As the tumors increased in size, the central photopenia also increased (Figure 4C). Visual inspection of excised tumors suggested central necrosis in the larger tumors. This observation may partly explain the decrease in the mean %ID/g over time for the control tumors. To confirm, tissue sections of a representative tumor from a mouse under a similar dosing schedule also showed tumor necrosis with H & E staining and corresponding areas of decreased proliferation with Ki-67 staining (Figure 5).
Analysis of the max %ID/g, upper 20% %ID/g, SUV, and tumor-to-background ratios produced similar results and statistical significance to the mean %ID/g analysis (data not shown). Analysis of data normalized to the baseline day 0 was also similar.
To assess the early response to RAF265, day 1 data were further analyzed. Compared with baseline, FDG accumulation decreased 28.2% and 30.9% for cohorts 1 and 2, whereas tumor volumes decreased only 14.9% and 5.2%, respectively (Figure 3C). In cohort 1, both FDG accumulation (P = .0001) and tumor volume (P = .01) were significantly different compared with controls; however, in cohort 2, only FDG decreased significantly compared with control (P = .04), whereas the decrease in tumor volume was not significant (P = .25). This analysis supports the hypothesis that FDG accumulation is an earlier, more sensitive measure of antitumor activity.
In this study, our first objective was to use DNA microarray analysis to select appropriate tracers for PET imaging. We assessed whether changes in expression of genes involved in glucose metabolism, thymidine metabolism, apoptosis, or EGFR signaling could provide insight into which PET tracers might be effective in melanoma by performing a pathway-based analysis on DNA microarray data from RAF265-treated A375M xenografts. RAF265 significantly modulated glucose metabolism, pyrimidine metabolism, and apoptosis pathways, suggesting that FDG, FLT, and annexin V, but not EGFR-based imaging modalities, may have utility. The modulation of glucose metabolism was consistent with the subsequent results, demonstrating that FDG accumulation was inhibited in cell culture and in the mouse tumor xenograft model.
It will be interesting to examine cell accumulation and small animal PET imaging for all four tracers in both cell lines and animal studies. We plan to perform a larger study where several tracers are investigated in several drug-resistant and drug-sensitive cell lines and xenografts to establish a more comprehensive relationship between tracer efficacy and changes in gene expression. It may be possible to correlate tracer efficacy to gene expression changes using a much smaller cohort of genes representing imaging-relevant biologic pathways. In this case, the resource-intensive DNA microarray approach could be substituted with interrogation of a panel of key genes by quantitative polymerase chain reaction. Because RAF265 inhibits VEGF-induced angiogenesis  as well as mutant B-Raf, angiogenesis pathways and new emerging tracers for imaging angiogenesis will also be investigated to assess the relative strengths of the dual mechanism of action.
Our second objective was to determine whether FDG PET could be used to assess the efficacy of RAF265. We first demonstrated that RAF265 inhibited FDG accumulation in A375M human melanoma cell culture in a dose-dependent manner (Figure 2A). We subsequently demonstrated that RAF265 inhibited FDG accumulation in an A375M mouse tumor xenograft model as early as 1 day of drug treatment (Figure 3, B and C) using small animal FDG PET. Significant effects of RAF265 of FDG accumulation were observed before tumor volume changes and persisted during the entire 2 weeks of the experiment. Furthermore, significant inhibition of FDG accumulation was consistent with down-regulation of the genes involved in the uptake and trapping of FDG, as assessed by gene expression analysis.
RAF265 is currently in phase 1 clinical trials for melanoma, and we anticipate that clinical FDG PET will play a central role in assessing early response to treatment based on these preclinical results. We are unaware of any published studies that have used small animal FDG PET followed by validation with clinical FDG PET. This study may serve as a model for how PET and molecular imaging can improve the drug development process by validating imaging modalities before initiation of phase 1 trials. Follow-up analysis will be performed after the conclusion of phase 1 trials to determine the predictive value of these preclinical studies.
Although few studies that used FDG PET for monitoring therapeutic response in clinical trials exist in the literature, we believe that PET can and should be used to assess response to therapy based on its proven utility in clinical practice . FDG PET may serve as an exemplary marker in clinical trials of novel melanoma drugs based on the ability of PET to follow therapy response in other tumors  and based on the excellent ability of FDG PET to detect melanoma . As an example, a small phase 2 trial of a novel polyamine synthesis inhibitor showed that FDG PET was an early predictor of a poor response in metastatic melanoma . In addition, a recent phase 1 dose-escalation trial of a mutant BRAF inhibitor showed markedly decreased tumor FDG uptake at day 15 of treatment of metastatic melanoma . FDG PET could be invaluable in the clinical development of novel targeted agents such as RAF265. The current paradigm in dose determination for these agents is based more on the optimal biologic dose rather than the traditional maximal tolerated dose . A noninvasive method such as FDG PET could be an excellent tool to determine the optimal biologic dose and to assess tumor response for RAF265 in melanoma patients.
FDG PET has an advantage over anatomic imaging in that changes in FDG accumulation can often be seen earlier than change in anatomic size. In our study, we observed that changes in FDG accumulation were larger and observed earlier compared with tumor volume changes. These changes occurred on the time scale of days in the mouse model, which may translate to a time scale of weeks for patients or possibly sooner. The ability of FDG PET to see earlier changes may significantly impact treatment decisions and survival. Our preclinical imaging results are in agreement with previous work from our group , others , and in clinical studies .
On the basis of these results, we propose a stepwise approach to select a suitable imaging tracer for tumor evaluation. Gene expression analysis is first performed on tumor cells treated with drug in vitro to interrogate molecular pathways that have corresponding imaging tracers. Pathways that are most affected will suggest which imaging tracers can be used in subsequent tracer accumulation studies in cell culture and small animal PET studies in mouse xenograft models.
In human clinical studies, the choice of an imaging biomarker for cancer is limited to a few modalities. Most oncology studies are performed with contrast enhanced computed tomography (CT) and/or FDG PET/CT, with a few other imaging modalities for more specialized applications, such as contrast-enhanced magnetic resonance imaging (MRI) for brain and musculoskeletal tumors, somatostatin receptor imaging for neuroendocrine tumors, iodine 123 (123I) for thyroid cancer, and [123I]metaiodobenzylguanidine for pheochromocytoma. Many new imaging modalities and agents are in development, such as whole-body diffusion-weighted MRI, [18F]fluoromisonidazole PET for tumor hypoxia, and VEGFR imaging for angiogenesis [42–44]. With the increasing availability of molecular imaging agents, the selection of the most optimal tracer may be more challenging in the future. Gene expression profiling has the potential to assist with the selection of an appropriate imaging agent. To be useful, gene expression analysis of changes in gene expression must be correlated with imaging agent features. One revealing example was provided by the study of Segal et al. , where CT imaging traits of liver tumors were correlated with gene expression profiles. Other works have also been performed to correlate gene expression profiles with FDG PET [22–27]. In the future, tailored or personalized imaging biomarkers through rational selection may become more prevalent, much like personalized therapies have been envisioned.
This article attempts to unite DNA microarray and small animal PET analysis within the context of drug development and therapy evaluation. The studies presented here provide a model for translational research using DNA microarray technology, cell culture assays, and in vivo small animal PET studies, which can direct the selection of an appropriate PET tracer for human clinical studies.
We would like to offer one specific conclusion regarding RAF265 and two general conclusions regarding DNA microarray analysis and small animal PET. 1) RAF265 is an efficacious novel B-Raf/VEGFR-2 inhibitor, which causes decreases in tumor glucose metabolism and FDG accumulation as early as 1 day after treatment in our preclinical model. The small animal FDG PET results support the decision to proceed with clinical trials and also support the use of human clinical FDG PET to assess early tumor response in ongoing phase 1 trials of RAF265 for melanoma. 2) DNA microarray analysis and small animal PET are complementary technologies that can each serve as a biomarker to assess glucose metabolism within the context of drug development for drug evaluation. 3) DNA microarray analysis can screen multiple pathways involved in cancer. Significant changes can be used to select corresponding molecular imaging agents for further cell culture and in vivo evaluation.
The authors thank Andreas Loening for assistance with image analysis, Shay Karen at the Stanford Small Animal Imaging Facility for assistance with small animal PET imaging, and Frederick Chin and David Dick at the Stanford Radiochemistry Facility for assistance with radiotracer synthesis.
1This work was supported in part by a research grant from Chiron Corp, now a subsidiary of Novartis (S.S.G.), and National Cancer Institute In vivo Cellular and Molecular Imaging Centers P50 CA114747 (S.S.G.).