Vorasidenib, a Dual Inhibitor of Mutant IDH1/2, in Recurrent or Progressive Glioma; Results of a First-in-Human Phase I Trial

This article is featured in Highlights of This Issue, p. 4457

Gliomas represent the most frequent malignant primary brain tumors and are characterized by diffuse infiltration of the brain by malignant cells (1, 2). World Health Organization (WHO) grade II and grade III diffuse gliomas are often referred to as lower grade gliomas (LGGs; ref. 3). Compared to glioblastomas (WHO grade IV diffuse gliomas), LGGs afflict younger patients, initially grow at a slower rate, and typically do not show contrast enhancement on T1-weighted brain MRI at initial disease diagnosis (4, 5). Treatment of LGGs includes maximally safe tumor resection, followed by radiation and chemotherapy as appropriate (5, 6). Unfortunately, this treatment is not curative and most patients suffer disease recurrence and progress to a higher tumor grade (7), often associated with aberrant vascularization (8) and the appearance of tumor contrast enhancement on T1-weighted brain MRI (5). Even patients with long-term disease control suffer from disease-related or treatment-related symptoms, including neurocognitive changes (5, 9, 10). New treatment approaches targeting disease-defining genetic events at the earliest stage of the disease may delay the need for DNA-damaging therapies and perhaps delay the transformation of LGGs into more aggressive tumors.

Mutations in the metabolic enzymes isocitrate dehydrogenase 1 and 2 (IDH1/2) occur in various human malignancies, including acute myeloid leukemia (AML), cholangiocarcinoma, and glioma. They occur in up to approximately 80% of patients with LGGs (3). Cancer-associated IDH1/2 mutations occur early in tumorigenesis, cluster in the active site of the enzymes, and cause the mutant enzymes to produce D-2-hydroxyglutarate (2-HG; refs. 11, 12). Accumulation of 2-HG leads to competitive inhibition of potentially more than 60 α-ketoglutarate–dependent enzymes, causing epigenetic dysregulation and impaired differentiation (13, 14). Given the central role of 2-HG in the molecular pathogenesis of mutant IDH1/2 (mIDH1/2) cancer (13, 14), pharmacologic blockade of mIDH enzymes is being pursued as a potential therapy. Inhibition of mIDH1/2 promoted differentiation in experimental models of mIDH1/2 glioma, leukemia, and cholangiocarcinoma (15–17). Patients with relapsed or refractory AML harboring mIDH1 or mIDH2 showed clinical responses to isoform-selective inhibitors of mIDH1 (ivosidenib) and mIDH2 (enasidenib), respectively (18, 19). Ivosidenib also showed antitumor activity in patients with mIDH1 gliomas (20).

Vorasidenib (AG-881), a first-in-class, dual inhibitor of mIDH1/2, was specifically developed for improved penetration across the blood–brain barrier, and showed brain penetrance and reduced tumor growth in an orthotopic model of mIDH glioma (21, 22). Dual inhibition of mIDH1 and mIDH2 may be superior to isoform-selective inhibition of mIDH1 or mIDH2 because isoform switching from mIDH1 to mIDH2, or vice versa, has been reported as a potential mechanism of acquired resistance in AML (23). Here, we report the results of a phase I study of vorasidenib in patients with advanced mIDH1/2 solid tumors, with a focus on glioma.

This phase I, single-arm, multicenter, open-label, dose-escalation study of vorasidenib enrolled patients with mIDH1/2 advanced solid tumors, including glioma (ClinicalTrials.gov, NCT02481154).

Vorasidenib was administered orally, once daily, in continuous 28-day cycles. Dose escalation was conducted separately for glioma and non-glioma solid tumors. Cohorts of 3 to 6 evaluable patients were to be enrolled, including at least 6 patients receiving the MTD or recommended phase II dose (RP2D). Additional cohorts of 1 to 6 patients could be enrolled at any dose level below the estimated MTD or RP2D for the replacement of patients not evaluable for the dose escalation, the evaluation of alternative dosing regimens, or for further analyses used for RP2D selection. At least 18 patients in the glioma cohort and 21 in the non-glioma cohort were expected to be treated.

The study was conducted according to International Conference on Harmonisation of Good Clinical Practice guidelines and the principles of the Declaration of Helsinki. The protocol was approved by the Institutional Review Board/International Ethics Committee at each study location. Written informed consent was provided by all patients before screening and enrollment. The complete study protocol is available in the Supplementary Materials and Methods.

All enrolled patients had a confirmed diagnosis of solid tumor, including glioma, with documented mIDH1 or mIDH2 that had recurred after—or had not responded to—initial standard therapy. IDH1/2 mutation status was assessed locally. Eligible patients were aged ≥18 years, had an Eastern Cooperative Oncology Group performance status score of 0–2, and evaluable disease assessed using Response Assessment in Neuro-Oncology (RANO) or RANO-LGG criteria for patients with glioma, or RECIST version 1.1 for patients with non-glioma solid tumors (24–26).

Other eligibility criteria included an expected survival of ≥3 months and adequate bone marrow, hepatic, and renal function. Patients were excluded if they had received systemic anticancer therapy or radiotherapy <21 days before their first day of study drug administration, prior treatment with bevacizumab at any time, or an investigational agent <14 days before their first day of study drug administration. Patients with glioma had a baseline brain MRI scan within 14 days before day 1 while not receiving glucocorticoids, or receiving the same daily dose of glucocorticoids, during the 5 days before the baseline MRI scan.

The primary objectives were to evaluate the safety and tolerability of vorasidenib treatment and to determine the MTD and/or RP2D. Safety evaluation included the incidence of dose-limiting toxicities (DLTs) during the first treatment cycle and adverse events (AEs), serious AEs, and AEs leading to discontinuation.

Secondary objectives included clinical activity as measured by best overall response and progression-free survival (PFS). For enhancing glioma and non-glioma solid tumors, objective response was defined as complete response (CR) or partial response (PR), as determined by the investigator on the basis of RANO criteria (24) or RECIST version 1.1 (26), respectively. For patients with nonenhancing glioma, objective response was defined as CR, PR, and minor response (mR) as determined by the investigator on the basis of RANO-LGG (25). Given the challenges associated with accurate representation of tumor response on MRI in LGG, the RANO working group considers a 25%–50% reduction in tumor size compared with baseline clinically meaningful, and several classifications now include mR as a measure of treatment effect (25, 27). Therefore, mR was included in the objective response rate for nonenhancing glioma. PFS was defined as the time from first dose to the date of progression or death, whichever occurred first. Blood samples were drawn pre- and postdose to determine circulating levels of vorasidenib.

Patients attended study center visits as outlined in the schedule of assessments (Supplementary Materials and Methods).

Tumor volume measurements were evaluated in the nonenhancing glioma cohort as described previously (20). Exploratory assessments also included confirmation of baseline mIDH1/2 status and identification of co-occurring mutations by next-generation sequencing using the ACE Extended Cancer Panel (Personalis) whenever archival formalin-fixed, paraffin-embedded samples were available.

The safety analysis set included all patients who received at least one dose of study treatment. The dose-determining set comprised all patients considered evaluable for DLT assessment and MTD estimation [i.e., patients either had a DLT during cycle 1 or completed ≥75% of their planned cycle 1 doses (21 of 28 days) and were considered by the clinical study team to have had sufficient safety data available to conclude that a DLT did not occur during cycle 1]. AEs occurring after cycle 1 may have been designated as DLTs by the study team.

An adaptive Bayesian logistic regression model (BLRM) with two parameters guided by the escalation with overdose control principle (28) was used to make dose recommendations and estimate the MTD/RP2D. Dose-escalation decisions were based on all relevant data available for patients in the dose-determining set from all dose levels evaluated in the study, including observed toxicities and estimates of probability of DLTs using BLRM, safety information, and pharmacokinetic/pharmacodynamic data.

Disposition, demographic and baseline characteristics, safety, and pharmacokinetic parameters were summarized using frequency distributions or descriptive statistics.

Objective response rates were calculated along with two-sided 95% confidence intervals (CI). All time-to-event outcomes were estimated using Kaplan–Meier methods. Point estimates and 95% CIs were calculated. Estimates of the median and other quantiles were generated.

Data from the non-glioma solid tumor and glioma cohorts were analyzed separately in all analyses.

The study was designed by the sponsor in collaboration with the lead investigators. Clinical data were generated by investigators and research staff at each participating site. Safety data were reviewed at regular intervals by investigators and by the sponsor, which also had a role in data collection, analysis, and interpretation. I.K. Mellinghoff and T.F. Cloughesy wrote the first draft of the article and had final responsibility for the decision to submit for publication. Further medical writing support was provided by the sponsor.

The data collected for the study will not be made available to others. Qualified researchers may request access to related clinical study documents. Please submit your data sharing requests to https://clinicaltrials.servier.com/data-request-portal/.

Patients were enrolled from June 18, 2015 through June 23, 2017, across 10 sites in the United States. At the analysis cut-off date (April 29, 2020), the study was ongoing. Overall, 93 patients with mIDH1/2 advanced solid tumors were treated, including 52 patients with glioma (Table 1).

The glioma cohort included 22 patients with nonenhancing glioma (absence of enhancement on MRI by investigator assessment) and 30 with enhancing glioma. The median age of the patients with glioma was 42.5 (range, 16–73) years. Nearly all patients with glioma had WHO grade II [25 (48.1%)] or WHO grade III [22 (42.3%)] tumors as of the most recent assessment before screening. Most tumors harbored a mutation in IDH1 (92.3%). Thirty-nine (75.0%) patients had received prior systemic therapy for the treatment of glioma and 30 (57.7%) had received prior radiation therapy. Eight (36.4%) patients with nonenhancing glioma remained on treatment, with 10 (45.5%) discontinuing treatment due to disease progression, 2 (9.1%) due to AEs, and 2 (9.1%) withdrawing from the study. One (3.3%) patient with enhancing glioma remained on treatment, with 24 (80.0%) discontinuing due to progressive disease and 5 (16.7%) withdrawing from the study. Patient disposition is reported in Supplementary Fig. S1.

The non-glioma cohort (n = 41) comprised patients with a variety of other solid tumors. Most patients had an mIDH1 tumor [27 (65.9%); Table 1] and most had received prior systemic therapies. Enrollment to this cohort was stopped by the sponsor in October 2016, in favor of continued development in glioma.

The initial starting dose was 25 mg once daily. Dose escalation up to 300 mg once daily in glioma and 400 mg once daily in non-glioma was initially completed (Supplementary Table S1). On the basis of DLTs of elevated serum transaminases in patients with glioma, an additional 10 mg once-daily level was opened and an additional 6 patients were enrolled in the 50 mg once-daily dose level. Five AEs of grade ≥2 elevated transaminases that occurred at doses of ≥100 mg in patients with glioma were designated as DLTs by the sponsor. Transaminase AEs were dose dependent (Supplementary Table S2), not associated with a bilirubin elevation, and resolved to grade ≤1 with dose modification or discontinuation. Two patients discontinued because of this AE. The MTD was not reached in the glioma cohort based on BLRM; dose selection could be guided by the BLRM but was not dependent upon the BLRM. On the basis of dose-dependent DLTs, the sponsor and investigators recommended no further escalation beyond 300 mg, and that doses <100 mg be further explored in glioma. No DLTs were observed and the MTD had not been reached at doses of up to 400 mg once daily in patients with non-glioma tumors before termination of enrollment in this cohort based on the sponsor's decision to focus on the development of vorasidenib in glioma.

The most common (>10%) AEs are reported in Table 2. Ten (19.2%) patients with glioma and 19 (46.3%) with non-glioma tumors experienced a grade ≥3 AE. The most common grade ≥3 AEs among patients with glioma were seizure [4 (7.7%)] and increased alanine aminotransferase [3 (5.8%)] and aspartate aminotransferase [2 (3.8%)]. In the glioma cohort, 2 (3.8%) patients discontinued because of AEs and 7 (13.5%) required a dose reduction due to AEs. Treatment-related AEs were reported in 38 (73.1%) patients with glioma and 26 (63.4%) with non-glioma tumors. There were no treatment-related deaths.

In patients with nonenhancing glioma (n = 22), the objective response rate (CR + PR + mR) by investigator was 18%, including 1 PR (patient #22) and 3 mR (patients #19, #20, and #21). All 4 responses were sustained, ranging from 7.4 to 27.7 months in duration. Sixteen (72.7%) patients had stable disease as their best response, many with reductions in the sum of products of the diameters <25%, which did not qualify for mR (Fig. 1A; Table 3).

No patients with enhancing glioma had a confirmed radiographic response and 17 of 30 (56.7%) had stable disease as their best response. One patient with contrast-enhancing anaplastic oligodendroglioma (patient #25) had >50% reduction in the sum of products of the diameters that was not confirmed, and the patient was therefore categorized as stable disease (Fig. 1A; Table 3).

The median (range) treatment duration was 26.8 (1.0–50.9) months for nonenhancing glioma and 3.3 (0.2–53.6) months for enhancing glioma. Fifteen (68.2%) patients with nonenhancing disease and 4 (13.3%) with enhancing disease remained on treatment for >1 year (Fig. 1B).

With 75% of events reported, the median PFS in the overall glioma population was 7.5 months (95% CI, 3.7–12.9; Fig. 1C). In patients with nonenhancing glioma, the median PFS was 36.8 months (95% CI, 11.2–40.8), with 59% of events reported and 6 of 9 censored patients remaining on treatment (range of PFS for these 6 patients, 33.2–49.6 months). In patients with enhancing glioma, the median PFS was 3.6 months (95% CI, 1.8–6.5). Efficacy results for the non-glioma cohort are provided in the Supplementary Results.

Pharmacokinetic analyses were performed for the glioma and non-glioma cohorts separately as of March 11, 2019 (Supplementary Table S3). A dose-proportional increase in plasma exposure of vorasidenib was observed in patients with glioma at doses of 10–300 mg, and less than dose proportional in patients with non-glioma tumors at doses of 25–400 mg. Vorasidenib had a long half-life (46.9–87.3 hours in glioma; 45.5–176 hours in non-glioma).

Targeted sequencing was performed on archival tumor samples from 18 patients with enhancing glioma and 11 with nonenhancing glioma (Supplementary Fig. S2). There was no association identified between any single gene mutation and tumor response in this small sample set.

Evaluation of posttreatment tumor volumes by MRI was centrally performed for 21 of 22 patients with nonenhancing glioma. Additional post hoc analysis of pre- and posttreatment volume measurements was performed for 3 patients with available historical MRIs (patients #15, #19, and #22). Visual inspection of the images, as well as sequential tumor volume measurements, showed tumor shrinkage following the initiation of vorasidenib (Fig. 2).

Standard therapy for patients with LGGs includes maximally safe surgical tumor resection, with additional radiation and chemotherapy for high-risk tumors (5, 6). This treatment is not curative and most patients with LGG suffer considerable morbidity and premature death (5, 9). There remains an urgent need to develop novel treatment paradigms. Our study describes the first-in-human evaluation of vorasidenib, a dual mIDH1/2 inhibitor specifically developed for increased blood–brain barrier penetrance, in glioma. Vorasidenib was associated with a favorable safety profile at doses <100 mg once daily in this previously treated glioma population, with many patients remaining on treatment after several years of continuous treatment. On the basis of safety and pharmacokinetic data from this study, doses of 50 mg once daily and 10 mg once daily were tested in a subsequent perioperative phase I study in patients with nonenhancing glioma (ClinicalTrials.gov, NCT03343197). Preliminary data from that study confirmed sufficient central nervous system concentrations of vorasidenib 50 mg once daily and >90% reduction in intratumoral 2-HG concentrations compared with untreated controls, indicating near complete inhibition of the enzyme (29). On the basis of the findings from these phase I studies, a vorasidenib dose of 50 mg once daily was selected for further study in mIDH glioma.

Vorasidenib showed preliminary activity in patients with nonenhancing glioma, with an objective response rate (CR + PR + mR) of 18% (one PR, three mR) and a median PFS of 36.8 months. Although comparisons with historical data are difficult to make due to differences in patient populations and the heterogeneity of prior treatments in this recurrent patient population, the median PFS for patients with nonenhancing disease in our study compares favorably with outcomes reported for cytotoxic therapies (30, 31). Despite lacking historical MRI scans for all patients, sustained tumor shrinkage was observed in multiple patients with nonenhancing glioma with vorasidenib treatment. In contrast, there was no indication of antitumor activity of vorasidenib in patients with enhancing tumors (no objective responses; PFS, 3.6 months), reminiscent of our earlier findings with the mIDH1 inhibitor ivosidenib (20). The lack of single-agent antitumor efficacy of vorasidenib in patients with enhancing gliomas may be due to the presence of additional genetic alterations in these tumors that can bypass the need for the mIDH enzyme for tumor maintenance. Although this explanation seems plausible, given the general association between contrast enhancement and genetic tumor evolution in LGG (32), we are unable to address this question in our current study because our protocol did not mandate a tumor biopsy and genomic sequencing immediately preceding study enrollment.

A watch-and-wait approach following surgery remains a treatment option for patients with low-risk LGG. Given the acute and long-term toxicities associated with—and additional genetic alterations at disease recurrence resulting from—alkylating chemotherapy and radiation treatment for glioma (9, 10, 33), there is an opportunity to introduce a targeted therapy against a potential driver IDH mutation during the active observation period, in the hopes of delaying more toxic therapies and preserving quality of life for a younger patient population. The favorable safety profile and single-agent activity of vorasidenib in recurrent, progressive nonenhancing glioma in our current study support the further exploration of vorasidenib in the earliest stages of mIDH LGG, compared with a watch-and-wait approach. To that end, vorasidenib (50 mg once daily) is being tested versus placebo in the ongoing, randomized, phase III INDIGO study (ClinicalTrials.gov, NCT04164901). The INDIGO study is enrolling patients with recurrent grade II nonenhancing mIDH glioma treated with surgery only, and will seek to offer additional insight into the antitumor activity of vorasidenib at an early stage of disease.

I.K. Mellinghoff reports personal fees from Agio Pharmaceuticals during the conduct of the study, as well as personal fees from Black Diamond Therapeutics, Puma Biotechnology, Voyager, Amgen, Novartis, and Eli Lilly outside the submitted work. M. Penas-Prado is a member of the Independent Data Monitoring Committee for an ongoing Phase III trial of vorasidenib; this was approved as part of her official job duties and she does not receive any financial compensation. K.B. Peters reports grants and other support from Agios during the conduct of the study, as well as grants from Biomimetix, AbbVie, and Novocure outside the submitted work. H.A. Burris III reports other support from Roche/Genentech, Bristol Myers Squibb, Incyte, AstraZeneca, MedImmune, Macrogenics, Novartis, Boehringer Ingelheim, Lilly, Seattle Genetics, Merck, Agios, Jounce Therapeutics, Moderna Therapeutics, CytomX Therapeutics, GlaxoSmithKline, Verastem, Tesaro, BioMed Valley Discoveries, TG Therapeutics, Vertex, eFFECTOR Therapeutics, Janssen, Gilead Sciences, BioAtla, CicloMed, Harpoon Therapeutics, Arch, Arvinas, Revolution Medicines, Array BioPharma, Bayer, BIND Therapeutics, Kymab, miRNA Therapeutics, Pfizer, Takeda/Millennium, Foundation Medicine, EMD Serono, GRAIL, Daiichi Sankyo, and Vincerx Pharma outside the submitted work. E.A. Maher reports other support from Agios during the conduct of the study. F. Janku reports grants from Agios, Asana, Astellas, Bicara, BioMed Valley Discoveries, Bristol Myers Squibb, Fujifilm, Genentech, Novartis, Plexxikon, PIQUR, Proximagen, Symphogen, and Synthorx; grants and other support from Deciphera, IDEAYA, and SOTIO; other support from Guardant Health, IFM, Illumina, PureTech Health, and Synlogic; personal fees and other support from Cardiff Oncology; and personal fees from Immunomet outside the submitted work. G.M. Cote reports other support from Agios, Eisai, Macrogenics, Boston Biomedical, Plexxicon, Merck KGaA/EMD Serono, Springworks Therapeutics, Bavarian-Nordic, and Bayer; personal fees and other support from Epizyme and PharmaMar; and personal fees from C4T outside the submitted work. M.I. de la Fuente reports personal fees from Agios during the conduct of the study, as well as personal fees from Forma Therapeutics and Inovio outside the submitted work. J.L. Clarke reports grants and personal fees from Agios during the conduct of the study, as well as grants from Novartis and Merck outside the submitted work. B.M. Ellingson reports personal fees from Agios and MedQIA during the conduct of the study. B.M. Ellingson also reports personal fees from Medicenna, Imaging Endpoints, Kazia, VBL, Oncoceutics, Boston Biomedical, ImmunoGenesis, and Ellipses Pharma; grants and personal fees from Neosoma; and grants from Siemens and Janssen outside the submitted work. R.J. Young reports personal fees from Puma, NordicNeuroLab, and ICON plc, as well as grants and personal fees from Agios outside the submitted work. H. Liu reports other support from employment at Agios Pharmaceuticals Inc. and Servier Pharmaceuticals, LLC during the conduct of the study. M. Lu reports other support from employment at Agios Pharmaceuticals Inc. and Servier Pharmaceuticals, LLC during the conduct of the study, as well as other support from Agios Pharmaceuticals Inc. and Servier Pharmaceuticals, LLC outside the submitted work. I. Hassan reports employment with and stock ownership at Agios Pharmaceuticals Inc. and employment with Servier Pharmaceuticals, LLC. L. Steelman reports other support from employment at Agios Pharmaceuticals Inc., Servier Pharmaceuticals, LLC, and Infinity Pharmaceuticals outside the submitted work. S.S. Pandya reports other support from employment at Agios Pharmaceuticals Inc. and Servier Pharmaceuticals, LLC during the conduct of the study; S.S. Pandya also reports ownership of Agios stock. T.F. Cloughesy reports personal fees and non-financial support from Agios during the conduct of the study. T.F. Cloughesy also reports personal fees from Tocagen, Karyopharm, Odonate, Bayer, Amgen, MedScape, DelMar aka Kintara, Pascal Bioscience, GW Pharma, VBL, Oryx, Roche, Merck, Novartis, DNATrix, Boehringer Ingelheim, Kiyatec, Global Coalition for Adaptive Research, Sapience, Inovivo, Vigeo Therapeutics, Tyme, Brainstorm, Immvira, Gan & Lee, and Break Through Cancer; personal fees and other support from Katmai; and other support from Chimerix outside the submitted work. In addition, T.F. Cloughesy has a patent for 62/819,322 pending, licensed, and with royalties paid from Katmai, and stock options from Notable Labs. P.Y. Wen reports personal fees from Agios, AstraZeneca/MedImmune, Beigene, Celgene, Eli Lilly, Genentech/Roche, Kazia, MediciNova, Merck, Novartis, Nuvation Bio, Oncoceutics, Vascular Biogenics, and VBI Vaccines during the conduct of the study, as well as personal fees from Agios, AstraZeneca, Bayer, Boston Pharmaceuticals, CNS Pharmaceuticals, Elevate Bio Immunomic Therapeutics, Imvax, Karyopharm, Merck, Novartis, Nuvation Bio, Vascular Biogenics, VBI Vaccines, Voyager, QED, Celularity, and Sapience outside the submitted work. No disclosures were reported by the other authors.

I.K. Mellinghoff: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. M. Penas-Prado: Investigation, writing–review and editing. K.B. Peters: Investigation, writing–review and editing. H.A. Burris III: Investigation, writing–review and editing. E.A. Maher: Investigation, writing–review and editing. F. Janku: Investigation, writing–review and editing. G.M. Cote: Investigation, writing–review and editing. M.I. de la Fuente: Investigation, writing–review and editing. J.L. Clarke: Investigation, writing–review and editing. B.M. Ellingson: Methodology, writing–review and editing. S. Chun: Methodology, writing–review and editing. R.J. Young: Methodology, writing–review and editing. H. Liu: Methodology, writing–review and editing. S. Choe: Methodology, writing–review and editing. M. Lu: Methodology, writing–review and editing. K. Le: Methodology, writing–review and editing. I. Hassan: Supervision, writing–review and editing. L. Steelman: Resources, writing–review and editing. S.S. Pandya: Conceptualization, resources, supervision, writing–review and editing. T.F. Cloughesy: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. P.Y. Wen: Conceptualization, investigation, writing–review and editing.

We thank the participating patients and their families, and the nurses, research coordinators, and study management team. The clinical trial was supported by Agios Pharmaceuticals, Inc., and Servier Pharmaceuticals, LLC. Translational research studies were supported by the NIH (1 R35 NS105109 01 and P30CA008748; to I.K. Mellinghoff) and the National Brain Tumor Society Defeat GBM Initiative (I.K. Mellinghoff and T.F. Cloughesy). Marissa Arnofsky and Steven Schoenfeld of Servier Pharmaceuticals, LLC (formerly of Agios Pharmaceuticals, Inc.), provided operational support for this study. MedQIA (Los Angeles, CA) assisted with image analysis. Assistance in article preparation was provided by Vanessa Ducas, Excel Scientific Solutions, Fairfield, CT, funded by Agios Pharmaceuticals, Inc., and Servier Pharmaceuticals, LLC.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.