Abstract 2318: The association between somatic mutational profiles and germline polygenic risk scores in TCGA

Abstract

Abstract Background: Tumor somatic mutational profiles capture common and distinct mutagenic processes within and across tumor types. Mutational signatures may reflect the effects of exogenous exposures, so linking signatures of unknown origin to risk factors may provide insight into cancer etiology. Studies with comprehensive mutational profile data do not always have epidemiological exposure data. We used polygenic risk score (PRS) as proxies for exposures and examined the association between somatic mutational profiles and PRS in The Cancer Genome Atlas (TCGA). Methods: TCGA tumor somatic mutational profiles were identified using methods based on nonnegative matrix factorization. We collected total somatic mutational count data from 15 cancer types with sample size ranges from 334 (Kidney renal clear cell carcinoma, KIRC) to 930 (Breast invasive carcinoma, BRCA). The signature-specific mutational counts for 6 single base substitution (SBS) signatures (SBS1, SBS2, SBS3, SBS4, SBS5, and SBS13) were also retrieved. We calculated PRS for the 15 selected cancer types, cancer risk determinants (Cigarettes Per Day; Drink Per Week; Body Mass Index, BMI), hormone-related traits (Age at Menarche; Age at Menopause), and autoimmune diseases (Inflammatory bowel disease, IBD; Ulcerative Colitis; Crohn's disease; Rheumatoid arthritis) using germline genetic data from TCGA subjects. We tested the association between somatic mutational count and PRS using zero-inflated negative binomial and negative binomial models, adjusting for gender, age at diagnosis, and the top 10 genetic principal components. Results: We found 17 statistically significant associations between tumor somatic mutational profiles and PRS after Bonferroni correction p < 2.9 × 10−5). These include positive associations between the germline BMI PRS and somatic mutational count of SBS1 (clock-like signature) in prostate adenocarcinoma (PRAD) tumors (p = 9.8 × 10−8), the Crohn's disease PRS and SBS2 (APOBEC) in BRCA (p = 5.8 × 10−6), and the IBD PRS and SBS2 in BRCA (p = 2.1 × 10−6). Negative associations were found between the BRCA PRS and SBS2 in BRCA (p = 2.7 × 10−5), and the PRAD PRS and SBS1 in PRAD (p = 1.4 × 10−9). Consistent with prior studies that found an inverse association between the age at menarche PRS and risk of prostate cancer, likely reflecting the association between the PRS and hormone levels, we found a negative association between this PRS and SBS1 in PRAD (p = 9.7 × 10−8). Conclusions: Our analysis suggests that there are robust associations between tumor somatic mutational profiles and germline PRS. These may reflect mechanisms through hormone regulation and immunological processes that drive cancer etiology. However, we caution that the negative associations between cancer PRS and some somatic mutational counts may be due to collider bias. Future work includes increasing sample sizes and collecting epidemiological data on exposures. Citation Format: Yuxi Liu, Alexander Gusev, Yujing J. Heng, Ludmil B. Alexandrov, Peter Kraft. The association between somatic mutational profiles and germline polygenic risk scores in TCGA [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2318.