Pancreatic cancer is an aggressive malignancy associated with racial/ethnic disparities. Notably, Black patients have a higher incidence and mortality compared to their White counterparts. However, low representation of Black and Hispanic patients in cancer genomics studies hampers our understanding of genomic factors contributing to differential outcomes. Furthermore, comparisons between racial groups fail to capture differences in genomic ancestry. We studied the associations between genetic ancestry and race/ethnicity categories with mutational profiles, using a large, diverse real-world pancreatic cancer cohort. We analyzed 9,274 de-identified pancreatic cancer patients (8,483 with ductal adenocarcinoma, PDAC) who underwent tumor profiling with the Tempus xT 648-gene assay. We used 654 ancestry-informative markers to estimate continental ancestry for five regions: Africa (AFR), Americas (AMR), East Asia (EAS), Europe (EUR), and South Asia (SAS). We imputed Asian (ASN, n=431), Hispanic/Latino/Native American (HLN, n=593), Non-Hispanic Black (NHB, n=875) and Non-Hispanic White (NHW, n=7,309) categories from genetic ancestry using a previously published machine learning method. Associations between either race/ethnicity categories or ancestry proportions and somatic variants (copy number alterations or CNAs, small nonsynonymous, OncoKB L1/2 & R1, and predicted driver somatic mutations using boostDM) were assessed for 172 pancreatic cancer-related genes or gene sets using logistic regression models. Models with and without race/ethnicity or ancestry terms were compared with likelihood ratio (LR) tests, and p-values were adjusted to control the FDR at 5%. We tested for associations of race/ethnicity groups with TMB using a Kruskal-Wallis rank sum test. Tests of TMB and small variants used only patients with available matched normal tissue to avoid artifacts from misclassification of germline variants. Among pancreatic cancer patients, having either a CNA or small nonsynonymous mutation in a receptor tyrosine kinase (RTK) gene (ERBB2, EGFR, MET, FGFR1) was associated with genetic ancestry (LR adjusted p=0.027). Specifically, AMR ancestry was positively associated (OR=1.08 per doubling of ancestry proportion, p=0.004) and EUR negatively associated (OR=0.93, p=0.002). Associations in PDAC were similar but not significant. Race/ethnicity was associated with RTK changes in both groups, with ASN and HLN having more mutations than NHW, but the associations were not significant after correcting for multiple hypotheses. There was no association of race/ethnicity and TMB . Somatic changes in RTK genes were associated with AMR genetic ancestry; however, despite the relatively large cohort size, these results are modest, suggesting there is not a large somatic mutation component contributing to differences in pancreatic cancer outcomes by ancestry.
VIEW THE PUBLICATION
VIEW THE POSTER