Landscape of KRASG12C, associated genomic alterations, and interrelation with immuno-oncology (IO) biomarkers.

3127 Background: Sotorasib has shown promising activity in cancer patients (pts) specifically harboring the KRASG12C mutation. Response rates vary significantly by tumor type, suggesting KRASG12C pathogenesis may be cancer-type-dependent. Methods: We retrospectively analyzed de-identified records of 79,004 pts with various cancer types that underwent Tempus xT and xF next generation sequencing assays. Fisher’s exact test was used to analyze the association between cancer subtypes and KRAS variants. Logistic regression was used to study co-mutations between KRASG12C and other oncogenes, as well as the association between KRAS variants and IO biomarkers. False discovery rate-adjusted P-value (FDR P) was used for multiple testing. Results: In total, 13,578 (17.4%) tumors harbored KRAS mutations, of which 1,632 were G12C; 750 KRAS wild-type (WT) tumors gained KRAS mutation on follow-up testing, with 79 harboring G12C. The most frequent KRAS variants across all cancers were G12D, G12V, G12C, and G13D (see Table). The distribution of KRAS variants significantly varied by cancer type, with G12C and G12D being the most prevalent in non-small cell lung (NSCLC) and colorectal (CRC) cancers, respectively. G12C was most prevalent in NSCLC (9%), appendiceal cancer (3.9%), CRC (3.2%), tumors of unknown origin (1.6%), and pancreatic cancer (1.3%). Compared to non-G12C, G12C was more frequently identified in females (56% vs. 51%, FDR P = 0.0005), smokers/prior smokers (85% vs. 56%, FDR P 60 years of age (73% vs. 63%, FDR P = 0.0006). While no G12C tumors exhibited BRAFV600E co-mutations, BRAFnon-V600E co-mutations were seen in 3.1% of pts. Significant differences were observed in genomic alterations co-occurring with G12C compared to non-G12C (e.g., STK11 (20.6% vs 6%), KEAP1 (15.4% vs 4.6%), SMAD4 (7.2% vs 19%), and PDGFRA (5.8% vs 3%); FDR P < 0.0001). However, G12C and oncogene co-mutations were not significantly different between NSCLC and CRC, except for KEAP1 (FDR_ P = 0.04). Compared to non-G12C and WT, G12C tumors were associated with TMB-High and PD-L1 over expression but were less likely to have MSI-H status (FDR P < 0.0001; Table). Conclusions: Our data suggest that KRAS variants significantly differ by cancer type. Tumors harboring KRASG12C were significantly associated with high TMB and PD-L1 overexpression. KRASG12C mutation appeared to be associated with smoking status. These results may guide future therapeutic strategies.[Table: see text]