Discordant familial risks between rare cancers

Abstract

Background: Recent large-scale DNA sequencing data indicate extensive genetic sharing between different types of cancer. Familial clustering of cancers at different sites was also confirmed by earlier population-wide epidemiological data. Cross-cancer familial risks tend to be lower in magnitude compared to cancer-specific risks and their reliable detection requires substantial sample sizes. Methods: In this study we utilize the 2021 update of the Swedish family and cancer data, the largest resource of its kind, to systematically study discordant familial clustering between 22 cancers, excluding the 4 most common cancers of the breast, prostate, lung and colorectum, as these have been covered in earlier publications. Familial risks were estimated as standardized incidence ratio (SIR), comparing the number of cases in a group with specified family history to background population. Separate analyses were conducted for groups with 1 and 2 probands diagnosed with the same cancer, that allowed quantification of familial risks and confirmation of associations in two independent populations. Results: Bidirectional 2-proband associations were observed for melanoma with squamous cell skin cancers and thyroid cancer with endocrine tumors; SIRs were 2.0 and 4.5, respectively. The second group included 7 cancer pairs that were significant in one 2-proband analysis, supported by 1-proband significance; endocrine and kidney cancers were the most common partners in these pairs. The final group of 14 cancer pairs were bidirectionally significant in 1-proband associations. Conclusions: The observed range of risk magnitudes reflected distinct etiological mechanisms spanning from shared environmental exposures to polygenic and possibly syndromic genetic influence, offering guidance to molecular studies.