PATH-08. A COMPREHENSIVE GENOMIC STUDY OF 390 H3F3A-MUTANT PEDIATRIC-TYPE DIFFUSE HIGH-GRADE GLIOMAS WHO CNS GRADE 4

Abstract

Abstract OBJECTIVES Histone gene mutant malignant gliomas - H3K27-altered diffuse midline glioma (DMG) and diffuse hemispheric glioma (DHG) H3G34-mutant - occur in all age groups and can have significant variation in clinical outcomes. Here, we report comprehensive genomic profiling from one of the largest collections of H3F3A-mutant gliomas analyzed to date, identifying subsets defined by recurrently co-mutated genes. METHODS We identified 390 H3F3A-mutant diffuse gliomas WHO grade 4 (201 females and 189 males) that were profiled in the comprehensive genomic profiling program at Foundation Medicine between 2013-2020. Information from pathology reports, histopathology reviews, and clinical data was assessed. RESULTS Our cohort comprised 304 (77.9%) H3K27M-mutant DMG WHO grade 4 (156 females and 148 males) and 86 H3G34-mutant DHG (45 females and 41 males) with a median age of 20 years (1-74 years). H3K27M-mutant DMG distributed equally between pediatric and adult patients, with 52% of the patients older than 20 years, 30% older than 30 years, and 18% older than 40 years at the time of first diagnosis. Clonal FGFR1 hotspot mutations were exclusively detected in K27M-mutant DMG (n = 64/304, 21%; p=0.0001), with a significant association with a higher age at first diagnosis (median 32.5 years), and with a wide tumor distribution across the diencephalon. Additional genes which were significantly more frequently altered in K27M-mutant compared to G34-mutant diffuse gliomas included NF1 (31% vs. 8.1%; p=0.0001) and PIK3CA/PIK3R1 (27.9% vs. 15.1%; p=0.016). Conversely, targetable alterations of the cell-cycle pathway (CDK4/6 amplifications and CDKN2A/B deletions) were enriched in H3G34-mutant DHG (26%) compared to H3K27M-mutant DMG (7%). Potentially targetable PDGFR alterations were present in 32.5% of H3G34-mutant DHG and in 18% of H3K27M-mutant DMG. CONCLUSIONS These data expand our understanding of the tumor-specific molecular features of pediatric-type high-grade gliomas, identifying cohort sub-structure by recurrent co-mutations, which can inform diagnosis and clinical trial design.