PATH-07. GENOMIC AND IMMUNE ANALYSIS OF PRIMARY REPLICATION-REPAIR DEFICIENT (RRD) GLIOMAS REVEALS THREE SUBGROUPS WITH DISTINCT DRIVERS AND RESPONSE TO IMMUNOTHERAPY: AN IRRDC REPORT.

Abstract

Abstract BACKGROUND Replication-repair deficiency (RRD) caused by germline/somatic defects in mismatch repair (MMRD) and/or polymerase-proofreading genes (PPD) drives 5-10% of high-grade gliomas in children, adolescents, and young adults (CAYA). Despite harbouring high mutation-burden (TMB), the basis of their heterogenous responses to immune-checkpoint inhibitors (ICI) is unknown. METHODS We performed multi-omic and immune analyses on human RRD-glioma specimens and developed murine models to investigate their differential responses to ICI. RESULTS RRD-gliomas in 202 CAYA-patients uniformly harbored high genomic microsatellite-instability and hypermutation (median TMB: 302-mutations/megabase), with specific MMRD mutational signatures contributing to frequent mutations in TP53 (88%), ATRX (83%), RAS/MAPK (77%) and IDH1/2 (17%). Strikingly, common pediatric mutations (K27M, G34R/V and BRAF;p.V600E) unrelated to MMRD signatures were absent, and a unique methylation profile distinct from non-RRD gliomas was observed. Molecularly, RRD-gliomas segregated into three genomic subgroups: RRD1 (MMRD+PPD; 56%), RRD2 (MMRD-only; 27%), and RRD3 (MMRD+IDH1/2; 17%). RRD1 included glioblastomas that developed early in germline CMMRD patients across the cerebral hemispheres and posterior-fossa, harbored extreme TMB (median: 420-mutations/megabase), balanced copy-number profiles, and immunogenic PPD signatures. RRD3 predominated in Lynch syndrome, presented in older children, were located in the forebrain, harbored lower TMB (median: 31-mutations/megabase) and demonstrated specific chromosomal gains/losses. Enrichment for immune-resistant JAK-STAT signalling in the RRD2/RRD3 transcriptome, and presence of high CD8+ T-cells in RRD1-glioma microenvironment, correlated with significantly improved 24-month post-ICI survival of 75% in RRD1 versus 43% and 33% for RRD2/3, respectively (p < 0.0001). Genetically engineered mouse models recapitulated the immuno-genomic phenotype and ICI-responses of the human RRD-subgroups, allowing successful preclinical testing of combinatorial checkpoint-inhibition strategies. Significant temporal and spatial heterogeneity linked to genomic instability and post-therapy immune-escape were observed. CONCLUSION Our data emphasizes an immediate need to distinguish RRD-glioma subgroups in clinical practice and translate our preclinical discoveries to subgroup-specific ICI-based combinatorial clinical trials.