Established Cancer Predisposition Genes in Single and Multiple Cancer Diagnoses

Abstract

ImportanceMuch of the understanding of cancer risk associated with rare pathogenic variants (RPVs) is derived from family-based studies or clinically ascertained samples, which may be limited by ascertainment and selection bias.ObjectiveTo quantify associations between RPVs in previously implicated cancer predisposition genes and single and multiple cancer diagnoses in a large population-based study.Design, Setting, and ParticipantsIn this genetic association study, whole-exome sequencing data were used from the UK Biobank, a UK population–based cohort that enrolled participants aged 40 to 69 years between 2006 and 2010. Participants who were involved in the whole-exome sequencing release of 200 000 genomes in 2020 were included in this study. This analysis included White participants only, as findings in other racial and ethnic groups had small sample sizes. Participants were diagnosed before or after biobank enrollment until March 2024.ExposuresThe sequencing data of a set of 96 previously implicated cancer predisposition genes were analyzed and compared using 2 methods. To determine the statistical significance of an association, a robust optimal sequence kernel association test was used, while odds ratios (ORs) and 95% CIs were obtained through Firth logistic regression.Main Outcomes and MeasuresThe primary study outcome was the diagnosis of 1 of 11 cancers (bladder, breast, central nervous system, colorectal, lung, melanoma, ovary, pancreatic, prostate, renal, thyroid) defined by relevant diagnosis codes in inpatient hospital diagnosis, cancer registry, and/or death registry data.ResultsData from 183 627 participants (101 414 [55.2%] female) were analyzed, including 25 824 participants with at least 1 cancer diagnosis, of whom 23 704 (91.8%) had a single cancer diagnosis and 2130 (8.2%) had 2 or more cancer diagnoses. A total of 157 793 controls had no cancer diagnosis. The median (IQR) age was 62 (56-65) years in participants with at least 1 cancer diagnosis, compared to 57 (50-63) years in those without a cancer diagnosis. Genetic variation in 16 genes was significantly associated with at least 1 cancer of interest (ATM, BARD1, BRCA1, BRCA2, BRIP1, CDKN2A, CHEK2, HOXB13, MITF, MLH1, MSH2, MSH6, NF1, PALB2, RAD51C, and RAD51D). The presence of an RPV in 1 of these 16 genes was associated with increased odds of at least 1 cancer (OR, 1.87; 95% CI, 1.76-1.98) and multiple primary cancers (OR, 2.56; 95% CI, 2.18-2.99). Carrier frequency was 6.28% and 8.36%, respectively.Conclusions and RelevanceThis genetic association study demonstrates several established associations between cancer predisposition genes and cancer diagnoses in an unselected population-based study. These results also demonstrate that RPVs in cancer predisposition genes are associated with multiple primary cancer diagnoses, suggesting that multigene panel testing may be warranted in these individuals.