Association between tumor mutational burden (TMB) and mutational profile and its effect on overall survival: A post hoc analysis of patients with TMB-high and TMB-low metastatic cancer treated with immune checkpoint inhibitors (ICI).

Abstract

2632 Background: High-TMB was recently approved as an agnostic biomarker for pembrolizumab in advanced cancers. Still, several patients with high-TMB do not respond to ICI, while some patients with low-TMB benefit from immunotherapy. Methods: We collected genomic (MSK-IMPACTassay) and survival data from 1,661 patients and assessed OS (Kaplan-Meier method) according to the mutational status. To better qualify patients with TMB ≥ 10 mut/Mb (TMB-H) (N = 488, 29%) and TMB < 10 mut/Mb (TMB-L) (N = 1,173, 71%) for ICI treatment, we analyzed single gene alterations impacting OS (P < 0.05). For all genes exhibiting a correlation with OS, we conducted a Cox multivariate analysis stratified by median TMB, sex, median age, microsatellite instability (MSI) status, and histology. Results: Survival to ICI increased with higher TMB. Median OS was 42 and 15 months for TMB-H and TMB-L (P < 0.05), respectively. For TMB-H tumors, 5 genes ( STK11, KEAP1, CIC, E2F3, TP53) exhibited reduced OS on ICI, and 22 genes ( NTRK3, TERT, NOTCH3, RNF43, TET1, PTPRD, NCOA3, TENT5C, ZFHX3, RIT1, CCNE1, PPM1D, GATA2, ALK, DNMT1, PTPRT, MET, EPHA7, BCL6, SMO, CDK6, MED12) were associated with better OS, P < 0.05. Cox multivariate analysis confirmed a correlation between mutations in STK11 and E2F3 with worse OS, while mutations in NTRK, PTPRD, RNF43, TENT5C, TET1, and ZFHX3 were associated with better OS (P < 0.05). Histology did not play a relevant role in ICI response except for melanoma (better OS; P < 0.05). MSI status did not significantly affect OS. For TMB-L tumors, 20 genes were related with reduced OS ( TP53, H3C2, DAXX, SMARC4, STK11, SOX17, RB1, PIK3CA, CTNNB1, KMT2D, HLA-A, FBXW7, CDH1, RBM10, KEAP1, IGF1R, H3C11, EGFR, RUNX1, B2M), while 8 genes were associated with better OS ( VHL, SETD2, PBRM1, BRAF, KDM5C, MAP2K1, CSF1R, RET), P < 0.05. Cox multivariate analysis confirmed 15 genes associated with superior ( KDM5C, PBRM1, and VHL) and inferior ( CTNNB1, DAXX, FBXW7, H3C2, H3C1, IGF1R, KMT2D, PIK3CA , RB1, SMARC4, SOX17, TP53) OS (P < 0.05). In the TMB-L context, melanoma histology and MSI status (except among VHL-mutated tumors) were independently associated with better OS. Conclusions: This pan-cancer analysis demonstrates that genomic alterations in single cancer genes can help define outcomes of TMB-H and TMB-L patients treated with ICI. [Table: see text]