Abstract C015: Selective and pairwise epistatic effects of somatic mutations in <i>KRAS</i> wild-type pancreatic cancer

Abstract About 8 to 10% of all pancreatic adenocarcinomas (PAAD) do not possess canonical mutations in KRAS. A sizable proportion of these KRAS wild-type (WT) pancreatic cancers harbor mutations in alternative MAPK-pathway driver genes. These alternative MAPK-pathway genes represent a set of therapeutic targets that might be exploited to treat individuals in this PAAD subset. Previous studies have prioritized the mutational- target landscape of KRAS WT PAAD by ranking genes determined by significance thresholds to be drivers by the prevalence of their somatic mutations in tumors. These prevalences usefully inform the size of the patient population that would be candidates for targeted therapeutics. However, prevalence is markedly influenced by underlying mutation rate. It is not a metric of cancer effect; it does not directly quantify the degree to which a driver mutation increases cancer cell survival and proliferation, nor do observations of co-occurrence or mutual exclusivity quantify epistasis. Therefore, to better guide the development of targeted therapeutics, we assembled a rich dataset of 6,391 PAAD tumor sequences from TCGA, UTSW, ICGC, QCMG, Yale-Gilead, and the AACR Project GENIE.We analyzed the cancer effect sizes of somatic variants in PAADs. Furthermore, we determined the selective epistatic interactions between KRAS and other PAAD driver genes. Nine of the 20 highest-effect variants were nonsynonymous substitutions in KRAS, with the prevalence of these mutations ranging from 18 to 2390 (median 93). This high frequency of highly oncogenic KRAS mutants supports the longstanding conception that KRAS mutations dominate the oncogenic landscape of PAAD. However, of the 6,391 cases analyzed, 627 (9.8%) were KRAS WT. To further investigate the oncogenic landscape of these samples, we employed a pairwise epistasis model of selection. We discovered that not only are alternative MAPK drivers (BRAF and NRAS) present in KRAS WT PAAD, but that they show antagonistic epistasis with KRAS mutation. Moreover, we found that non-MAPK drivers CTNNB1 and IDH1 also exhibit substantial antagonistic epistasis with KRAS. Also, in a KRAS-substituted background, selection for mutations in tumor suppressor genes TP53, CDKN2A, and SMAD4 was increased. Furthermore, mutations in BRAF, NRAS, CTNNB1, TP53, CDKN2A, and SMAD4, genes commonly considered to be drivers of PAAD, in addition to a previously unrecognized driver IDH1, suppressed selection for KRAS mutations. Not only MAPK drivers, but also other oncogenic drivers of PAAD at both high and low prevalence exhibit antagonistic somatic selective epistasis. Treatments targeting these variants that are antagonistically epistatic with KRAS may rescue the full strength of selection on KRAS variants, warranting continued monitoring for KRAS variants in resurgent cancer. Moreover, as resistance to KRAS inhibitors is likely caused by gained alterations in alternative MAPK drivers, exploration of treatments targeting these mutations may additionally provide an avenue to overcome therapeutic resistance in PAADs. Citation Format: Rishi M Shah, Can M Ersahin, Jeffrey D Mandell, Vincent L Cannataro, Jeffrey P Townsend. Selective and pairwise epistatic effects of somatic mutations in KRAS wild-type pancreatic cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Pancreatic Cancer Research; 2024 Sep 15-18; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2024;84(17 Suppl_2):Abstract nr C015.