PATH-23. GENOMIC LANDSCAPE OF IDH-MUTANT PRIMARY GLIOBLASTOMAS SHOWS DISTINCT CLINICAL AND MOLECULAR FEATURES AND THAT CDKN2A SHOULD BE SUPPLEMENTED WITH MGMTp AND G-CIMP FOR PRECISE PROGNOSTICATION

Abstract

Abstract There have only been rare studies of IDH-mutant primary glioblastomas (IDH-mutant astrocytoma IV); there were one or two studies on secondary glioblastomas. In a cohort of 70 cases, we conducted clinical analysis, methylation profiling, RNA sequencing, targeted sequencing, and TERTp seqeuncing on available FFPE tissues. Median follow-up was 58.2 months (n= 60). IDH-mutant primary glioblastomas had longer median OS (30.4 months) and median PFS (25.9 months) than IDH-mutant secondary glioblastomas as in the literature or established databases. MGMTp methylated cases had better OS (p= 0.001) and it was an independent prognosticator. We previously showed G-CIMP to be an independent prognostic marker for IDH-mutant glioblastomas (NOA 2019). Although CDKN2A deletion was an independent prognostic marker for poorer OS (p= 0.036) and PFS (p= 0.005), MGMTp methylation had a trend of superseding CDKN2A deletion (p= 0.055) for prognostication and G-CIMP subgroups could similarly partially supersede CDKN2A deletion (p= 0.582). Hence, CDKN2A deletion should be supplemented with these two biomarkers for finer prognostication. Targeted sequencing (n= 55) showed that there were more ATRX (35/55, 64%), TP53 (31/55, 56%), KMT2D (18/55, 33%), POLE (11/55, 20%) and MSH6 (7/55, 13%) mutations, but fewer TERTp (3/69, 4%) and PTEN (1/55, 2%) mutations than IDH-wildtype glioblastomas as from literature and databases. CNVs revealed by methylomes (n= 53) and mutations (n= 55) showed that there were more PDGFRA (amplification: 9/53, 17%, mutation: 10/55, 18%) alterations, but fewer MET (amplification: 3/53, 6%, mutation: 4/55, 7%) alterations and hypermutated (6/55, 11%) cases than IDH-mutant secondary glioblastomas from literature. GISTIC analysis revealed amplifications of CCND2, CDK4, MYC, and PDGFRA, deletions of CDKN2A, RB1, and chromosome 10q to be significant CNVs (q< 0.05). There were few EGFR amplifications (2/53, 4%), which was different from regular glioblastomas. RNA sequencing (n= 42) showed few fusions (4/42, 10%), which was different from IDH-mutant secondary glioblastomas.