The pleiotropic landscape of rare variant associations with multiple cancers in large biobanks

Previous association studies between germline rare genetic variation and cancer risk have primarily examined a limited number of cancers in clinical samples, often with participants predominantly of European genetic ancestry. We conducted exome-wide rare variant association analyses across 70+ cancer types using data from more than 729,000 participants from the UK Biobank (UKB) and All of Us (AoU) Research Program cohorts. We used generalized linear mixed models to screen for cancer pleiotropic effects by conducting gene-based and single-variant tests of predicted loss of function (pLoF) and missense variants and six groups of cancer defined by biological and etiological similarities. We then assessed significant genes for associations with 27 individual cancer types that had data for at least 500 affected individuals. Of the 33 potential pleiotropic genes identified, 16 consistently showed significant associations across ≥3 individual cancer types. For example, the presence of at least one CHEK2 pLoF variant was associated with increased odds of diagnosis with 14 different cancers (odds ratio [OR] range: 1.34-3.21). Similarly, the presence of at least one RTEL1 missense variant was associated with lower odds of diagnosis with 9 cancers (OR range: 0.67-0.88). Our results expand our knowledge about these loci and point to a larger impact on overall cancer risk than previously appreciated.