Abstract Pancreatic ductal adenocarcinoma (PDAC) patients face poor prognoses attributable to delayed disease detection, resulting in distant metastases and thereby reducing treatment efficacy. The challenge of detecting PDAC lesions during early stages emanates from an inauspicious trifecta of clinical factors: early-stage malignancies are often asymptomatic, existing biomarkers are insufficiently sensitive, and the physical location of the pancreas stymies reliable screening. A promising strategy towards improving outcomes is the resection of high-risk, pre-malignant lesions deemed likely to progress into carcinomas. Intraductal papillary mucinous neoplasms (IPMNs) are both large enough to be detected during abdominal imaging and can originate PDAC lesions. However, a patient may have many IPMN cysts, many of which will never progress. As pancreatic resection carries with it a high risk of morbidity, distinguishing the the evolutionary trajectories of IPMNs that will and will not progress could inform treatment strategies. Additionally, these trajectories may serve as disruptable targets to manage pre- neoplastic lesions, preventing their progression to PDAC. The vast majority PDAC cases contain mutations in KRAS and GNAS with secondary driver mutations in TP53, SMAD4, CDKN2A, and RNF43. Recent work has compared the frequency of these mutations in PDAC with low-grade and high-grade IPMN lesions, finding that they more common in high than low- grade lesions. Interestingly, unlike KRAS or GNAS, KLF4 mutations are prevalent in low-grade IPMNs, rare in high-grade IPMNs, and almost never observed in full-fledged carcinoma, suggesting distinct evolutionary trajectories possibly mediated by epistatic interactions between KLF4 and KRAS. However, the prevalence of a mutation is not dispositive as to its importance in tumorgenesis. Mutations in TTN, for instance, are frequently observed in PDAC and other cancers, but are widely considered to play a modest role, if any, in these cancers. To understand the contribution of a mutation to the evolution of cancer, we must instead deconvolve selection from baseline mutation rate across genomic sites and among patients to distinguish truly oncogenic mutations from mutations that occur frequently merely due to high underlying mutational rates. Here, we quantify the evolutionary and epistatic interplay between KRAS, KLF4, and other frequent mutations in IPMNs and PDAC during malignant transition. First, we report the cancer effect sizes of single-nucleotide variants in IPMNs at varying dysplastic stages. Then, we infer the synergistic and antagonistic interactions between driver mutations, providing insight into the order in which driver mutations occur. We then corroborate this epistasis-inferred ordering with the ordering of mutations recovered from ancestral state reconstruction derived from time-calibrated phylogenies. Lastly, we probe the mutational processes most responsible for conferring highly oncogenic mutations in IPMNs and PDACs to identify potentially actionable or preventable pathways of disease progression. Metabolism Citation Format: Nic Fisk, Derek Song, Margaret Moore, Cole G Jensen, Vincent L. Cannataro, Mofeed Nagib, Jeffrey D Mandell, Luisa Escobar-Hoyos, John W Kunstman, Jeffrey P Townsend. Evolutionary and epistatic analyses reveal genic interactions with KRAS during malignant progression of pancreatic ductal adenocarcinoma [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 C030.