Search

2007 Background: Improvements in risk stratification of meningioma are needed to guide post-operative management and application of adjuvant therapy. Although profiling of DNA methylation, copy number variants (CNVs), RNA sequencing, and exome sequencing have better elucidated meningioma biology, these approaches have not revealed clinically tractable biomarkers for radiotherapy responses. In this study, we develop and validate a targeted gene expression biomarker to predict meningioma outcomes and benefit from radiotherapy. Methods: Targeted gene expression profiling was performed on a development set of 173 meningiomas (median follow-up 8.1 years) and a validation set of 331 consecutive meningiomas (median follow-up 6.1 years) treated at independent institutions (70% WHO grade 1, 24% WHO grade 2, 6% WHO grade 3). All patients underwent surgery (n = 504) with or without postoperative radiotherapy (n = 73 with radiation). Regularized Cox regression within the development set resulted in a continuous gene expression risk score for local freedom from recurrence (LFFR). The model (34 genes and 7 housekeeping genes) and thresholds for low, intermediate, and high-risk scores were locked and applied to the validation set. Results: The gene expression risk score outperformed WHO grade (validation 5-year LFFR delta-AUC 0.15, 95% CI 0.076-0.229, p = 0.001) and DNA methylation grouping (delta-AUC 0.075, 95% CI 0.006-0.130, p = 0.01) for LFFR, disease-specific survival, and OS, achieving a negative predictive value for recurrence at 5-years of 93.2%. The biomarker reclassified 35.8% of WHO grade 1 tumors as intermediate or high risk (5-year LFFR/OS 62%/79%), and 18.3% of WHO grade 2-3 tumors as low risk (5-year LFFR/OS 78%/100%). The biomarker was independently prognostic after accounting for WHO grade, extent of resection, primary versus recurrent presentation, CNV status, DNA methylation group, and Ki67 labeling index, and was predictive for LFFR after postoperative radiotherapy, with a hazard ratio of 0.41 for intermediate to high risk propensity-matched meningiomas (95% CI 0.2-0.9, p = 0.0002) versus 0.79 for low risk meningiomas (95% CI 0.1-8.8, p = 0.5182). Conclusions: Targeted gene expression profiling of 504 meningiomas resulted in a biomarker which improved discrimination of meningioma local recurrence, disease-specific survival, and overall survival, and also predicted benefit from radiotherapy.

2045 Background: Recurrence of GBM after initial treatment with surgery, radiation, and chemotherapy is nearly universal. Salvage therapies have limited efficacy with median overall survival (OS) of approximately 9 months and 6-month-progression-free survival (PFS-6) of 10-25% for both targeted and traditional therapies. Given GBM’s molecular heterogeneity, targeting a single molecular abnormality in isolation has consistently failed as a strategy, and precision combination approaches are needed. Methods: The primary objective was to demonstrate the feasibility of implementing a personalized drug regimen for patients (pts) with surgically resectable recurrent GBM within 35 days of surgery. Secondary objectives included safety and efficacy. Eligible pts signed consent before surgery, and tumor tissue was analyzed using the CLIA-approved “UCSF500” next-generation sequencing panel with paired tumor/germline sequencing. A specialized genomic tumor board made individualized treatment recommendations incorporating sequencing results of the recurrent tumor and clinical history for each pt, using up to 4 FDA-approved drugs in combination (all drugs provided by study). Correlative studies will be reported separately. Results: 19 pts signed consent and 16 pts had surgery on trial, 1 with pathology showing treatment effect only. The remaining 15 pts were all genetically profiled and successfully started their individualized treatment within 35 days of surgery, meeting the primary feasibility endpoint. Conclusions: Implementation of an individualized treatment regimen was feasible in a timely fashion in surgically resectable recurrent GBM pts, with encouraging preliminary efficacy results. Further investigation is warranted, both to validate efficacy and to streamline this approach in larger pt populations. Clinical trial information: NCT03681028. [Table: see text]

e13575 Background: Hyperprogressive disease (HPD) has been described in solid tumor patients treated with immune checkpoint inhibitors (ICI). HPD is defined as a ≥2-fold increase in tumor growth rate (TGR) following initiation of ICI. HPD has not been explored in patients with high grade gliomas (HGG) on ICI or standard cytotoxic regimens. In advanced cancer patients receiving ICI, MDM2/4 amplification or EGFR alterations, both found in HGG, correlated with HPD. We compared the rate of HPD in recurrent HGG patients receiving ICIs to those treated with non-immunotherapy agents. Methods: Patients with HGG on ICIs for 1st or 2nd recurrence were compared to a control group receiving other therapies at 1st recurrence. Patients with prior or concurrent bevacizumab or anti-VEGFR were excluded due to pseudoresponse and decreased enhancement with these drugs. HPD was calculated by comparing TGR immediately before and after treatment. Results: 49 patients met inclusion criteria (27 ICI, 25 control). 25/27 patients treated with ICIs and 20/22 patients in the control group had complete imaging and were eligible for analysis. In the ICI group, 60% were men (15/25) and 88% (22/25) had a diagnosis of GBM. 68% were treated at first progression (17/25). Controls were 80% male (16/20) and all had a diagnosis of GBM. 30% (6/20) were 65 years or older at diagnosis in the control group compared to 28% (7/25) in the ICI group. In total, 7/25 patients met criteria for HPD in the ICI group (28%) compared to 4/20 patients in the control group (20%). 10/25 patients (5/7 with HPD) in the ICI group and 8/20 patients (2/4 with HPD) in the control group had next generation sequencing of their tumors. EGFR alterations and MDM2/4 amplifications were not associated with HPD whereas PTEN mutations were more common in the HPD group (71% HPD vs. 33.3% no HPD). Conclusions: HPD is observed in patients with HGG treated with ICI at comparable rates to those with other cancers, but was also observed in 20% of patients receiving other therapies. While the numbers are small, PTEN mutations may be associated with HPD in patients with HGG. In contrast to other solid tumors, EGFR alterations and MDM2/4 amplifications were not associated with HPD in HGG.