Phase 0 Clinical Trial of Everolimus in Patients with Vestibular Schwannoma or Meningioma

Meningiomas and vestibular schwannomas are among the most common intracranial neoplasms. According to the most recent Central Brain Tumor Registry of the United States (CBTRUS) report (1), meningioma represents the most common histology for this population, accounting for 37.8% of all tumors, with an annual incidence of more than 31,000 and an average annual age-adjusted incidence rate (AAAIR) of 8.58 per 100,000. For vestibular schwannoma, the histology-based annual incidence is over 6,000, with an AAAIR of 1.90 per 100,000. Given that a large proportion of patients with meningioma and schwannoma are diagnosed based on imaging and primary therapy is often nonsurgical (e.g., radiotherapy), the true incidence and prevalence of meningioma and vestibular schwannoma are substantially higher (2, 3).

Depending on the size and location, meningiomas cause neurologic symptoms including seizures and focal neurologic deficits. Vestibular schwannoma typically cause hearing loss and may also lead to facial nerve dysfunction, dysphagia, and brain stem compression. Primary treatment of meningiomas and vestibular schwannoma typically involves surgery and/or radiotherapy, which may cause further neurologic deficits and other complications, as well as late effects.

Although vestibular schwannoma and meningioma most often occur sporadically, they are also associated with the inheritable tumor predisposition syndrome neurofibromatosis type 2 (NF2), an autosomal dominant genetic disorder with an incidence of approximately 1 in 40,000. NF2 is caused by inactivation of the NF2 gene located on chromosome 22q (4, 5), that codes for the NF2 gene product, Merlin. Patients with NF2 are predisposed to develop cranial nerve schwannomas, most commonly involving the vestibular branch of the eighth cranial nerve (i.e., vestibular schwannoma), as well as meningiomas, ependymomas, and peripheral schwannomas. The majority of patients with NF2 develop progressive hearing loss in adolescence or young adulthood due to bilateral vestibular schwannoma and additional progressive neurologic deficits from multiple meningiomas, schwannomas, and/or ependymomas, which are often inoperable. As a result, patients with NF2 suffer from significant morbidity and reduced life expectancy (6).

Tumor formation in NF2 patients is thought to involve biallelic loss of NF2 through mutations and loss of heterozygosity. Biallelic somatic mutations in NF2 are also found in sporadic schwannomas and meningiomas (7) and thought to represent a primary oncogenic driver in these tumors. Biallelic inactivation of NF2 is present in approximately 40% of sporadic meningiomas (8), and both sporadic and NF2-relatedvestibular schwannoma consistently lack expression of detectable Merlin (9).

Prior studies have revealed that inactivation of Merlin activates the mTOR complex-1 (mTORC1) signaling pathway (10–12). Accordingly, the mTOR pathway was also identified as a priority therapeutic target at an NF2 Clinical Trials Consensus Workshop (12). The AKT effector mTORC1 is a key mediator of RTK signaling and a major sensor of metabolic fitness. mTOR regulates essential signal transduction pathways, linking growth stimuli to cell-cycle progression, and integrates signals involving nutrient availability, energy status, and stress (13, 14). Phosphorylation of 4E-BP1 and S6K downstream of mTORC1 promotes translation of mRNAs encoding proteins that regulate cell growth, survival, and cell-cycle progression (15). In addition, mTORC1 suppresses autophagy and activates transcription factors, which promote lipid biosynthesis and mitochondrial function (13, 14). The mechanism by which Merlin inhibits mTORC1 signaling has not yet been fully elucidated. Current evidence, however, suggests that Merlin acts at or below the level of TSC2/1, because loss of Merlin does not activate AKT or ERK, and silencing of TSC2 abrogates the growth-inhibitory effect of Merlin (10). Interestingly, unless phosphorylated by AKT, TSC2 accumulates in the nucleus (16). Furthermore, genetic analysis in Drosophila suggests that a CRL4 ligase ubiquitylates and targets TSC for degradation (17). These observations raise the possibility that CRL4^DCAF1 is the ligase that targets TSC2 for degradation. Alternatively, because CRL4^DCAF1 regulates a relatively broad program of oncogenic gene expression (18), it may regulate one or more genes, which in turn affect mTORC1 signaling. Importantly, inhibition of mTORC1 by rapamycin exerted a selective cytostatic effect, but did not induce apoptosis in Merlin-deficient arachnoidal, meningioma, and schwannoma cells in vitro (10, 11), with similar observations made in schwannomas in vivo (19). In mouse models of meningioma, mTORC1 inhibitors suppressed tumor growth in both NF2-deficient and nondeficient models (20, 21). Preclinical evidence also suggests that everolimus may inhibit VEGF production and therefore reduce tumor angiogenesis (22). This mechanism of action may be relevant investibular schwannoma, because prospective clinical trials have demonstrated that inhibition of VEGF has therapeutic efficacy in this tumor (23, 24).

Taken together, these data suggested that treatment with mTORC1 inhibitors, such as everolimus, should be explored clinically in the treatment of vestibular schwannomas and meningiomas. Although bevacizumab has emerged as a useful treatment option for a subset of patients with NF2 with progressive vestibular schwannoma (23, 24), there remains an urgent need for novel, effective, and less toxic therapies for patients with vestibular schwannoma and meningioma.

Everolimus is FDA approved for the treatment of renal cell carcinoma (RCC; ref. 25) and for the treatment of subependymal giant cell astrocytoma (SEGA) associated with tuberous sclerosis in patients who are not candidates for curative surgical resection (26). A prerequisite for efficacy of any molecular targeted antitumor therapy is target inhibition in the tumor. This is particularly true for brain tumors, which often exist in a protected environment behind the blood–brain barrier (BBB), blood–cerebrospinal fluid barrier, or blood–nerve barrier (27, 28). There is a paucity of data regarding the delivery of systemic drugs to vestibular schwannoma or meningioma tissue. Vestibular schwannoma may be protected by the BBB, blood–nerve barrier, or cerebrospinal fluid brain barrier. Meningiomas typically derive their blood supply from extracranial arteries and therefore considered outside the BBB (29), but it is unknown whether systemic drugs achieve adequate tissue concentration within meningiomas. Everolimus has been proposed to cross the BBB; however, the drug concentration achieved in these target tissues is unknown (30).

Our study was conceived as a phase 0 (or pharmacodynamic and biomarker endpoint driven) study, to inform interpretation of results from advanced stage clinical trials of everolimus in brain tumors, identify potential resistance mechanisms, and devise novel strategies to enhance therapeutic efficacy.

Adult patients (age ≥18 years) requiring surgical resection of a meningioma or vestibular schwannoma were eligible for this study. Histologic confirmation was not required prior to study entry. Additional key eligibility criteria included Karnofsky performance score (KPS) ≥60%, absolute neutrophil count ≥1,000/mm³ (unsupported), platelet count ≥100,000/mm³ (unsupported), hemoglobin ≥8 g/dL (transfusion support allowed), creatinine ≤1.5 times upper limit of normal (ULN) or corrected glomerular filtration rate ≥70 mL/minute, total bilirubin ≤1.5 times ULN, alanine aminotransferase (ALT) ≤2.5 times ULN, serum albumin ≥2 g/dL, INR <1.3 (or <3 on anticoagulants), fasting serum cholesterol ≤300 mg/dL, and fasting triglycerides ≤2.5 times ULN. Key exclusion criteria included prior therapy with mTOR inhibitors, known hypersensitivity to everolimus or other rapalogs, concurrent therapy with cytochrome 3A4 inducers or inhibitors, and chronic hepatitis B or C infection (due to the risk of disease reactivation with rapalogs).

All baseline evaluations were required within 14 days before starting everolimus and included history and physical including neurological exam, hepatitis screening, KPS, complete blood count, PT/INR, serum sodium, potassium, chloride, bicarbonate, blood urea nitrogen, creatinine, albumin, fasting glucose, and serum lipid profile (triglycerides, total cholesterol, HDL and LDL). For patients with positive hepatitis screening history, blood testing for HBsAb, HBsAg, HBcAb, HBV-DNA, and HCV-DNA was required. In addition, a serum pregnancy test was required for all females of childbearing age.

The study was conducted under a protocol approved by the Institutional Review Boards (IRBs) of all institutions enrolling patients on the clinical trial (NYU Langone Health, Johns Hopkins, Piedmont Healthcare, Massachusetts General Hospital) and registered at clinicalTrials.gov (NCT01880749). Informed consent was obtained from the patients in accordance with institutional policies. General oversight of the trial was by the principal investigator (M.A. Karajannis) and the Data and Safety Monitoring Board (DSMB) of NYU Langone Health.

This study was a prospective, open-label, phase 0 study. Given that only a single surgery is generally performed to resect meningiomas and schwannomas without prior biopsy, obtaining matched pre- and posttreatment samples from the same tumor was not feasible. We therefore relied on matched archival control tissue for comparisons.

The primary and secondary study endpoints were molecular target inhibition and intratumoral drug concentration, respectively. According to phase I trial data in patients with cancer, terminal half-life of everolimus administered daily is 30 hours and steady-state levels are reached within 1 week. Trough levels we shown to be stable thereafter, averaging 13.2 ng/mL with 10 mg daily dosing (31). Subsequent clinical studies established the safety and efficacy of everolimus administered on a 10-mg daily oral dosing schedule in patients with RCC (25).

The primary endpoint for each patient was the complete inhibition of phospho-S6 in tumor tissue after at least 10 days of exposure to everolimus at a daily dose of 10 mg, as determined by IHC. This endpoint was chosen based on prior pharmacodynamic data from a published trial, showing complete loss of phospho-S6 expression by IHC in solid tumor tissue of patients treated with everolimus (32).

The secondary endpoint was to assess the delivery of everolimus to tumor tissue along with pre- and postoperative blood concentrations. Exploratory endpoints included assessment of downstream pharmacodynamic biomarkers.

Only patients enrolled in the study who completed at least 10 days of everolimus at the full dose of 10 mg daily immediately before surgery and who had adequate tumor tissue available after clinical pathology were considered evaluable for analysis. Ten days was chosen as minimum treatment duration because steady-state is reached at that time point (33), and S6-kinase inhibition reaches a plateau with daily dosing, based on modeling of preclinical and clinical pharmacokinetic and pharmacodynamic data (34). On the basis of preclinical data, it was considered likely that everolimus will reach measurable concentrations in tumor tissue and it was expected that all of the samples will have measurable drug. Everolimus levels in tumor tissue and blood (immediately prior and after surgery) were assessed for each patient. Exploratory analysis was performed to assess this pharmacokinetic data in conjunction with the primary endpoint of inhibition of phospho-S6.

Everolimus was supplied by Novartis, Inc. and prescribed at a dose of 10 mg daily at bedtime by mouth for 10 days, immediately preceding the planned tumor resection. If necessary, patients were allowed to take everolimus for a maximum of 7 additional days to allow for unexpected delays in the surgical procedure. Patients were supplied with a study drug diary and asked to document the time of each dose of everolimus taken.

Any patient who received at least one dose of everolimus was considered evaluable for safety and for inhibition of phospho-S6; however, to evaluate the full potential of the drug to inhibit phospho-S6 in tumor tissue, the primary analysis was based on evaluable patients who complete at least 10 days of everolimus at the full dose of 10 mg daily immediately prior to surgery and who had adequate tumor tissue available after clinical pathology was completed.

Before the start of surgery and after completion of surgery, whole venous blood samples were collected in tubes containing K2 EDTA and sent to the participating hospital's clinical laboratory. Time of blood collections was recorded. Everolimus blood level testing was performed using a standard commercial laboratory assay (Quantitative Liquid Chromatography-Tandem Mass Spectrometry) approved for clinical use to monitor everolimus whole blood levels.

After obtaining tissue required for diagnostic pathology, additional tumor samples were collected for study purposes and snap frozen in liquid nitrogen. Time of sample collection was recorded. Care was taken to exclude tissue samples that have come in direct contact with heat (cauterization). Frozen samples were stored in liquid nitrogen until shipment and analysis. Fifteen unstained slides with freshly cut standard 4-μm tissue sections from formalin-fixed, paraffin-embedded (FFPE) tissue was also requested from each tumor sample for IHC analysis.

Snap-frozen samples were thawed and homogenized with 80% methanol after weighing. Control samples were spiked in with standard at 14 different amounts from 0 to 1,280 ng/mL. Three technical replicates were set for each point, and the standard curves were created. Each sample was analyzed in triplicate by LC/MS-MS using a 4.6- × 100-mm Acclaim120 C18 column (Thermo Scientific) and gradient of 0.1% [volume for volume (v/v)] formic acid in 2 mmol/L ammonium acetate to 100% methanol in 0.1% formic acid and 2 mmol/L ammonium acetate over 10 minutes at 600 μL/minute flow rate. The solvents were delivered by a Vanquish UHPLC coupled directly to a Q Exactive HF-X (Thermo Scientific). Data were acquired by parallel reaction monitoring (PRM) with a 2 Thomson window around 975.6 and a normalized collision energy of 25. Everolimus amount was calculated based on a linear standard curve using ion intensities of ammonium ion adducts of intact everolimus (m/z 975.6152) and fragment ions of m/z 908.5499 and 890.5396. Values were then normalized to pg/mg everolimus/mg tissue based on the weight of each tissue sample.

IHC was performed on FFPE, 4-μm tissue sections using unconjugated rabbit anti-human phospho-S6 ribosomal protein (pS6), clone 91B2 (Cell Signaling Technology, catalog No. 4857, AB_2181035), unconjugated rabbit anti-human phospho-AKT-1 (pAKT), clone S473/D9E (Cell Signaling Technology, catalog No. 4060, AB_2315049) and unconjugated rabbit anti-human phospho-MAPK (pERK) clone D13.14.4E (Cell Signaling Technology, catalog No. 4370, AB_2315112). IHC was performed on a Ventana Medical Systems Discovery XT instrument using Ventana's reagents and detection kits unless otherwise noted. Antibody was detected with Ventana biotinylated goat anti-rabbit followed by application of streptavidin–horseradish peroxidase conjugate. The complex was visualized with 3,3 diaminobenzidene and enhanced with copper sulfate. Appropriate positive and negative controls with known expression levels of antigen were included with the study sections. IHC results were scored using a histoscore, as previously described (32).

Tumor samples collected for study purposes were snap-frozen in liquid nitrogen until analysis. To prepare samples for immunoblotting, the tumor samples were lysed in RIPA buffer (Thermo Fisher Scientific) with protease and phosphatase inhibitors (Cell Signaling Technology) and quantified with Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). PVDF Membranes (Bio-Rad) were incubated in blocking buffer (5% skim milk in TBS with 0.1% Tween, Cell Signaling Technology) for 1 hour at room temperature and then with primary antibodies diluted in blocking buffer for overnight at 4°C. After three washes, the membranes were incubated with goat anti-rabbit HRP-conjugated antibody (Thermo Fisher Scientific) at room temperature for 1 hour and subjected to chemiluminescence using ECL (Thermo Fisher Scientific). Densitometry analysis was performed using Image Lab Software (Bio-Rad; ref. 25).

With 21 evaluable patients in each of the two tumor types under study, an optimum two-stage clinical trial design for each tumor type will test the null hypothesis that the proportion of patients with complete phospho-S6 inhibition in tumor (i.e., histoscore of 0 by IHC) posttreatment is ≤0.65, versus the alternative that this proportion is ≥0.90. If everolimus is actually ineffective in inhibiting phospho-S6, there is a 0.031 probability of concluding that it is effective (target α = 0.05); if the drug is actually effective, there is a 0.17 probability of concluding that it is not effective (power = 83%). For each tumor type, the trial will terminate after 10 patients are tested in the first stage if seven or more patients have complete inhibition of phospho-S6. If the trial continues to a second stage, a total of up to 21 patients will be tested; if 17 or fewer patients have complete inhibition of phospho-S6, the drug will not be considered biologically effective in that tumor type (calculations from PASS, NCSS, 2008, J. Hintze, Kaysville, UT).

Given significantly slower than expected accrual, we performed an early interim analysis and tested the available specimens as of October 2016 (i.e., eight meningiomas and one vestibular schwannoma), for phospho-S6 expression (primary endpoint). We found that five of nine samples had an incomplete p-S6 inhibition (i.e., histoscore >0), which meant that the requirement for opening stage 2 was not going to be met for meningioma and that timely completion of stage 1 accrual for vestibular schwannoma was unrealistic. Following DSMB recommendations, the protocol was amended to combine meningiomas and vestibular schwannoma into a single analytical stratum, and eventually closed after a total of 10 patients (eight with meningiomas, two with vestibular schwannoma) were accrued.

Exploratory analysis was performed to assess the relationship of everolimus tissue concentrations to blood levels and in conjunction with the primary endpoint of inhibition of p-S6, as well as downstream signaling effectors. Levels of p-S6, p-ERK, and p-AKT as determined by IHC (histoscore) were compared with 10 controls with matching distribution of tumor types (meningioma or vestibular schwannoma) and NF2 status (sporadic or NF2-related). Correlations of blood and tumor drug levels were estimated using a Spearman correlation. Distributions of biomarkers between study patients and controls were compared using a Mann–Whitney–Wilcoxon test.

Thirteen eligible patients were enrolled in the study between December 16, 2013 and October 11, 2016. There were nine females (69%) and four males (31%), with a median age of 39 years at enrollment (range, 21–69 years). Three patients did not complete the planned protocol therapy prior to surgery and were, therefore, considered nonevaluable; two patients discontinued drug early due to toxicities that were anticipated (mucositis and nausea), and one patient required urgent surgical intervention for clinically progressive disease. Of the remaining 10 evaluable patients, seven patients had a clinical diagnosis of NF2, five with meningiomas and two with vestibular schwannoma requiring surgery. The remaining three patients had sporadic meningiomas. All meningiomas were WHO grade I by histology, and only one tumor (M-10) was recurrent. Detailed clinical and demographic information for all evaluable study patients is provided in Table 1. All patients provided written informed consent for study participation prior to enrollment.

All 10 evaluable study patients completed a 10-day course of everolimus immediately prior to scheduled surgery. The three nonevaluable patients received at least one dose of everolimus. All observed toxicity at least possibly related to everolimus was expected and minor (CTCAE 4.0 grades 1 and 2), including mucositis (30%), nausea (15%), fatigue (8%), and rash (8%).

The suggested therapeutic range for the treatment of SEGA is 5 to 15 ng/mL, based on a predose (trough) specimen. Everolimus blood levels in the 10 study patients ranged between 6.9 to 49.6 ng/mL (median, 17.4) preoperatively and 6.3 to 43.4 ng/mL (median, 9.4 ng/mL) postoperatively. Frozen tumor tissue of sufficient quantity and quality for mass spectrometry was available in nine of 10 study patients, with everolimus concentrations in tumor tissue ranging between 9.2 to 169.2 pg/mg (median, 24.2). Detailed results for all patients are shown in Table 2 and tissue drug levels are displayed in Fig. 1.

Preoperative blood levels were higher compared to postoperative (Wilcoxon signed-rank test P = 0.002), and both values were strongly correlated (Spearman correlation = 0.85; Fig. 2A). Preoperative blood levels and tumor tissue concentrations were strongly correlated (Spearman correlation = 0.77; Fig. 2B). Postoperative bood levels and tumor tissue concentrations were moderately correlated (Spearman correlation = 0.67; Fig. 2C).

Detailed IHC results for all patients and controls are shown in Tables 2 and 3, respectively. Levels of p-S6 as determined by histoscore were lower in subjects (median, 90; range, 0–220) compared with controls (median, 122, range, 60–265; Mann–Whitney–Wilcoxon test P = 0.025; Fig. 3A). There was no evidence of a difference between levels of p-ERK in subjects (median, 180; range, 40–220) and controls (median, 135; range, 100–240; Mann–Whitney–Wilcoxon test P = 0.27; Fig. 3B). There was no evidence of a difference in value of p-AKT between controls (median, 42, range, 5–100) and subjects (median, 12, range, 0–100; Mann–Whitney–Wilcoxon test P = 0.18; Fig. 3C).

We further examined mTORC1 signaling in patients with sufficient remaining frozen tumor samples by immunoblot analysis. Detailed immunoblotting results for everolimus-treated patients and nontreated controls are shown in Fig. 4A. In keeping with IHC results, the p-S6/S6 ratio was decreased in subjects compared with controls (P < 0.005; Fig. 4B), while the p-ERK/ERK ratio did not show a significant difference between the two groups (Fig. 4C).

Despite many efforts, there remains a paucity of effective treatment options for patients with unresectable or recurrent meningiomas and schwannomas, with surgery and radiotherapy remaining the main therapeutic modalities for local tumor control. Recent prospective clinical trials have identified bevacizumab to be effective in a subset of patients with NF2-related and even sporadic vestibular schwannomas (23, 24), however the treatment benefit can only be sustained with continued drug administration in the majority of patients. Although biallelic loss of NF2 is considered a key molecular driver in the majority of schwannomas and approximately 40% of meningiomas (35), development of molecular targeted therapies has been challenging, and predominantly focused on signaling pathways that are aberrantly activated in NF2-mutant tumors, such as the ErbB, PDGFR, and mTOR signaling pathways (36). On the basis of preclinical in vitro data available at the time, we set out to assess the ability of everolimus to penetrate tumor tissue in human patients and abrogate mTORC1 signaling in a phase 0 study in preparation for subsequent phase II studies. Because of the study design that relied on volunteer subjects to assume additional risks without a reasonable likelihood of direct benefit, we encountered slower than anticipated approval, leading to early study termination. Nevertheless, we were able to acquire a unique patient-derived dataset shedding light on the results of published phase II studies of mTORC1 inhibitors in NF2-related tumors and informing future molecular targeted therapy studies in this patient population.

The first key finding from our study is the detection of measurable drug levels of everolimus in human meningioma and schwannoma tissue at steady-state. Median tumor tissue drug concentrations determined by mass spectrometry corresponded to the therapeutic range previously established in blood, and were strongly correlated with the preoperative blood levels. To our knowledge, this represents the first published data on everolimus drug levels in human tumor tissue. Tissue levels of temsirolimus, another rapalog, and mTORC1 inhibitor, were assessed in two studies including patients with malignant glioma (37, 38), with no qualitative reduction in tissue p-S6 levels observed based on very limited sample analysis.

The measurement of total drug levels in tissue is an inherent limitation of our study and similar clinical trials. Although measuring unbound (free) drug in the extracellular fluid (ECF) using microdialysis remains the gold standard of assessing drug penetration though the BBB in vivo (39), it is generally not feasible in a clinical trial setting such as ours. Another limitation of our study, which is inherent to phase 0 studies where the target tumor is removed with a single surgery, is the reliance on matched control tumor tissue from different subjects that were untreated.

In our study, we observed only partial inhibition of phospho-S6 in the treated tumors compared to controls, indicating incomplete target inhibition, despite reaching trough blood levels in all patients meeting or exceeding the therapeutic range for successful treatment of SEGA. The failure to meet this primary endpoint of phospho-S6 inhibition in our study is in contrast with published data in patients with breast cancer treated with everolimus at the same dose and schedule (32). Pharmacologic inhibition of mTORC1 may result in MAPK pathway activation through a PI3K-dependent feedback loop (40), and while primary target inhibition in our study was incomplete, we did not observe statistically significant elevation of p-ERK in tumors treated with everolimus. Taken together, our observations may explain the limited antitumor effect of everolimus as a single agent seen in clinical studies for NF2 patients to date (41, 42). More recently, combination therapy of everolimus and octreotide has shown encouraging clinical activity in a prospective clinical trial (43). Novel pharmacologic strategies to enhance mTORC1 inhibition in schwannomas and meningiomas, for example using different drugs, drug combinations, and/or delivery systems, merit investigation in future studies.

M.A. Karajannis reports grants from Novartis, Inc. and grants from the NIH/NCI during the conduct of the study; personal fees from AstraZeneca, Bayer, CereXis, and QED Therapeutics; and personal fees from Recursion Pharma outside the submitted work. S.R. Plotkin reports consulting for AstraZeneca; financial support from the Department of Defense and the NIH to run clinical trials of vistusertib (dual mTORC inhibitor); and drug (vistusertib) provided for clinical trials by AstraZeneca. D.G. Placantonakis reports other support from Monteris, Synaptive, and Tocagen and other support from Robeaute outside the submitted work. J.O. Blakeley reports grants from the NIH during the conduct of the study; personal fees from AbbVie, Astellas; and nonfinancial support from Bristol Myers Squibb outside the submitted work. No disclosures were reported by the other authors.

M.A. Karajannis: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. A. Mauguen: Data curation, formal analysis, visualization, methodology, writing–review and editing. E. Maloku: Formal analysis. Q. Xu: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration. E.M. Dunbar: Resources. S.R. Plotkin: Resources. A. Yaffee: Data curation. S. Wang: Data curation. J.T. Roland: Resources. C. Sen: Resources. D.G. Placantonakis: Resources. J.G. Golfinos: Resources. J.C. Allen: Resources. N.A. Vitanza: Resources. L.A. Chiriboga: Resources, Methodology. R.J. Schneider: Conceptualization, resources, funding acquisition, investigation, methodology. J. Deng: Resources, formal analysis, investigation, visualization, methodology. T.A. Neubert: Resources, formal analysis, supervision, investigation, visualization, methodology. J.D. Goldberg: Conceptualization, formal analysis, methodology. D. Zagzag: Resources, supervision, investigation, methodology. F.G. Giancotti: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, visualization, methodology, writing–review and editing. J.O. Blakeley: Conceptualization, resources, funding acquisition, methodology.

This study was supported by the NIH/NCI grant R01CA164295 to M.A. Karajannis and Novartis, Inc. grant CRAD001CUS205. This research was funded in part through the NIH/NCI Cancer Center Support Grant P30CA008748 to Memorial Sloan Kettering Cancer Center and NIH/NCI R01CA191222 (to F.G. Giancotti). The NYU Langone Experimental Pathology Immunohistochemistry Core Laboratory was supported in part by the Laura and Isaac Perlmutter Cancer Center Support Grant; NIH/NCI P30CA016087, and the NIH S10 grants NIH/ORIP S10OD010584 and S10OD018338. The NYU Mass Spectrometry Core for Neuroscience was supported by NIH grant S10OD023659 (to T.A. Neubert). We are grateful to the patients participating in this study and the clinical teams of the participating institutions for excellent study-related patient care. Results from this study were presented in part at the 2018 Joint Global Neurofibromatosis Conference, Paris, France, November 2018 and the Society of Neuro-Oncology 24th Annual Meeting, Scottsdale, Arizona, November 2019.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.