Cancer Diagnosis, Polygenic Risk, and Longevity-Associated Variants

Abstract

Background: Polygenic risk scores (PRS) have been developed to predict individual cancer risk and their potential clinical utility is receiving a great deal of attention. However, the degree to which the predictive utility of individual cancer-specific PRS may be augmented or refined by the incorporation of other cancer PRS, non-cancer disease PRS, or the protective effects of health and longevity-associated variants, is largely unexplored. Methods: We constructed PRS for different cancers from public domain data as well as genetic scores for longevity (Polygenic Longevity Scores or PLS) for individuals in the UK Biobank. We then explored the relationships of these multiple PRS and PLS among those with and without various cancers. Results: We found statistically significant associations between some PLS and individual cancers, even after accounting for cancer-specific PRS. None of the PLS in their current form had an effect pronounced enough to motivate clinical cancer risk stratification based on its combined use with cancer PRS. A few variants at loci used in the PLS had known associations with Alzheimers disease and other diseases. Conclusion: Underlying heterogeneity behind cancer susceptibility in the population at large is not captured by PRS derived from analytical models that only consider marginal associations of individual variants with cancer diagnoses. Our results have implications for the derivation and calculation of PRS and their use in clinical and biomedical research settings. Impact: Extensions of analyses like ours could result in a more refined understanding of cancer biology and how to construct PRS for cancer.