Abstract B004: Multi-omics profiling to identify mechanisms of resistance to KRAS inhibition: A comparative study on colorectal and lung cancer

Abstract

Abstract Background: Lung adenocarcinoma (LUAD) has a higher objective response rate to single-agent KRASG12C inhibitors (e.g., AMG 510) than colorectal cancer (CRC) patients (32.2% vs 7.1%). It is known that molecular heterogeneities exist between KRASMut LUAD and CRC tissues. We hypothesize that co-mutations and oncogenic pathways may confer resistance to KRAS inhibitors in CRC cells. Methods: Whole-exome and transcriptome profiles of KRASmut LUAD (n=168) and microsatellite stable (MSS) CRC (n=183) patients were downloaded from TCGA. Mutation profiles of KRASmut patients were also obtained from AACR Project-GENIE. Gene essentiality scores derived from the whole genome CRISPR screenings were downloaded from the DepMap to evaluate the relative oncogene addiction to KRAS in lung (n=133) and CRC (n=59) cell lines. Differential gene expressions studies were performed using EBayes, followed by gene set enrichment analysis (GSEA). CRC cell lines SW837, SW1463, SW403, and HCT15 were treated with AMG 510, a KRASG12C inhibitor, and Pictilisib, a PIK3CA inhibitor, and cell viability was measured up to 96 hours after treatment. Results: The KRASmut LUAD cohort had co-mutation clusters remarkable for (i) no co-mutation (59%, 99/168), (ii) KEAP1mut and/or STK11mut (20.2%, 34/168), and (iii) a mixed group with a frequent mutation in other genes (mean=2) but without a clear recurrent pattern (20.8%, 35/168). In CRC, KRASmut tumors formed clusters of (i) no co-mutation (59.6%, 109/183), (ii) PIK3CAmut (26.2%, 48/183), and (iii) a similar cluster with frequent co-mutation (mean=2) without a recurrent pattern (14.2%, 26/183). Similar co-mutation patterns were also observed in AACR Project-GENIE LUAD and CRC cohorts. KRAS oncogene addiction was seen in both KRASMut lung (median scores, -1.1 vs -0.4, p<0.001) and CRC (median scores, -1.3 vs -0.5, p<0.001) cell lines than KRASwt cells. However, KRASmut/PIK3CAmut co-mutated CRC cell lines (n=10) were significantly less dependent on the KRAS gene than KRASmut/PIK3CAwt cells (n=22) (median scores, -1.0 vs -1.4, p=0.03). Two KRASG12C/PIK3CAwt CRC cell lines that are highly dependent on KRAS (SW837 and SW1463 cells) also showed higher sensitivities (IC50<5uM) to AMG 510 and Pictilisib relative to KRASnon-G12C/PIK3CAmut CRC cell lines SW403 and HCT15 (IC50>10uM). GSEA revealed that the Hedgehog, spermatogenesis, and pancreas-beta cells genesets were significantly active (Normalized enrichment scores > 1, FDR<0.05) in KRAS KO-resistant CRC cells, but the same genesets were not enriched in KRAS KO-resistant lung cancer cells. Conclusion: Somatic co-mutations and transcriptomics signatures in KRASmut LUAD and CRC may predict sensitivities to KRAS inhibitors. KRASmut CRC cells with PIK3CAmut or active hedgehog pathway are resistant to KRAS gene KO, suggesting multiple drug resistance mechanisms to KRAS inhibitors exist in KRASmut CRC cells. Future studies should emphasize identifying new drug targets, as optimal drug combinations of KRAS inhibitors may not be same for all KRASmut CRC patients. Citation Format: Saikat Chowdhury, Vinay K. Pattalachinti, Valsala Haridas, Scott Kopetz, John Paul Shen. Multi-omics profiling to identify mechanisms of resistance to KRAS inhibition: A comparative study on colorectal and lung cancer [abstract]. In: Proceedings of the AACR Special Conference: Targeting RAS; 2023 Mar 5-8; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Res 2023;21(5_Suppl):Abstract nr B004.