(SS#I) Stryker CNS Tumor Award (2023 Award Winner): Next Generation Sequencing-derived Genomic Subclasses Predict the Impact of Extent of Resection in IDH Wild-type Glioblastoma

Introduction: Extent of resection (EOR) is a major driver of overall survival (OS) in IDH wild-type (IDHwt) glioblastoma. However, the value of EOR in genomically heterogeneous GBM subclasses is unknown. We leveraged a large cohort (N=520) of surgically resected IDHwt GBMs to establish volumetrically-defined residual tumor that predict EOR benefits in 5 genetically distinct subsets of GBM using next generation sequencing (NGS).

Methods: 520 surgically resected IDHwt GBM (2012-2021) with validated NGS (Glioseq) and FISH data were analyzed. Monte Carlo consensus clustering validated 5 biologically unique genetic subclasses identified by TERT promoter mutation (TERTm), MGMT methylation (MGMTm), EGFR pathway activation (EGFR) and p53 mutation (p53m) as modulators of EOR benefit. Clusters included: 1)TERTm/MGMTmeth/EGFR; 2) TERTm/MGMTm/p53 mutated, 3: TERTm/MGMTm/CDKN2A; 4: TERTwt; 5) TERTm/MGMT unmethylated. Impact of EOR/clinical variables on OS were analyzed across subtypes using Cox proportional hazard modeling. Volumetric residual tumor thresholds associated with optimized survival were identified (maximally-selected log-rank statistics).

Results: OS dependency on EOR varied significantly by cluster. TERTm/MGMTmeth/EGFR was associated with the most permissive contrast enhancing volume (CEV) in which survival benefits were seen (11.7cc CEV; 24.3 vs. 12.9mo, p=0.011). Conversely, the most stringent EOR standards were observed in TERTm/MGMTm/CDKN2A (0.1cc CEV, 28.1 vs 13.9mo, p=0.002). Univariate analysis demonstrated beneficial effect of EOR in all groups. Controlling for age, sex, performance, temozolomide, radiation, and CEV threshold, substantial inter-group variability was noted. Remarkably, after multivariate analysis, EOR emerged as the primary determinant of OS in Groups 2, 3, and 5, whereas radiation emerged as the primary determinant in Groups 1 and 3.

Conclusion : Our data provide a nuanced genomic context for interpreting GBM EOR effects in the current genomic era. Moreover, we provide volumetrically-defined CEV as independent predictors of surgical efficacy in defined subsets. These data may help drive surgical decision making as accurate preoperative genotyping methods are established in GBM.