Abstract 1166: Leveraging existing data to identify ancestry-associated features across multiple cancer types

Abstract

Abstract People with different ancestries inherit different risks and encounter different environmental exposures resulting in different somatic profiles. A lack of knowledge about ancestry-specific alterations is a major barrier to implementing precision medicine, leading to inequities in genetic testing, targeted treatment and clinical trial design for cancer patients from the underserved populations. However, a limited number of cancer cases from non-European populations have been sequenced in research setting, and paired normal samples are often not collected in routine clinical care. Here, we inferred African (AFR), South Asian (SAS), East Asian (EAS), American (AMR) and European (EUR) ancestry from comprehensive genomic profiling (CGP) of over 200,000 tumors and identified 18717, 8588, 6594 patients with significant proportions of AFR, AMR, EAS ancestry, respectively. Using logistic regressions with age of diagnosis, gender and tumor mutation burden (TMB) as covariates, we identified 165 ancestry-associated genes across 14 common cancer types. We used reported race/ethnicity data in the AACR project GENIE cohort for validation. We found that endometrial carcinoma patients with AFR ancestry were enriched with TP53 mutations (OR=2.3, FDR corrected p=5x10-31) but lacked PTEN (OR=0.3, FDR corrected p=2x10-41) and KRAS (OR=0.5, FDR corrected p=8x10-13) mutations. In colorectal cancer, KRAS (OR=1.5, FDR corrected p=4x10-44) and APC (OR=1.4, FDR corrected p=1x10-16) mutations were enriched in patients with AFR ancestry, whereas BRAF mutations were depleted in both AFR (OR=0.5, FDR corrected p=5x10-24) and EAS (OR=0.6, FDR corrected p=6x10-5) patients. CDK12 mutations were enriched in both AFR (OR=1.6, FDR corrected p=2x10-4) and EAS (OR=3.1, FDR corrected p=2x10-7) prostate cancer patients. For patients with AMR ancestry, increased mutation frequency was observed in SETD2 in pancreatic cancer (OR=3.4, FDR corrected p=1x10-4), and VHL in renal clear cell carcinoma (OR=1.9, FDR corrected p=2x10-4). In GBM, TERT alterations were associated with EUR ancestry (OR=1.3, FDR corrected p=0.003). Interestingly, TMB-high status was associated with EUR ancestry in melanoma (OR=5.9, p=6x10-36) and anti-correlated with AFR ancestry in endometrial carcinoma (OR=0.7, p=1x10-6). Our large-scale ancestry analysis characterized the somatic landscape of non-European cancer patients. Future directions include quantifying the germline heritability of identified ancestry-associated mutations using a local ancestry risk score (LRS) and investigating ancestral effects on cancer prognosis and outcome using the GENIE cohort. Citation Format: Jian Carrot-Zhang, Justin Newberg, Garrett Frampton, Rameen Beroukhim. Leveraging existing data to identify ancestry-associated features across multiple cancer types [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1166.