Genetics, primary care records and lifestyle factors for short-term dynamic risk prediction of colorectal cancer: prospective study of asymptomatic and symptomatic UK Biobank participants

Abstract

ABSTRACTObjectivesTo quantify the contributions of polygenic scores, primary care records (presenting symptoms, medical history and common blood tests) and lifestyle factors, for short-term risk prediction of colorectal cancer (CRC) in both all and symptomatic individuals.DesignProspective cohort study.SettingUK Biobank with follow-up until 2018.ParticipantsAll participants with linked primary care records (n=160,526), and a subcohort of participants with a presentation of a symptom associated with CRC (n=50,728).Main outcome measuresOutcome was the first recorded CRC diagnosis within two years. Dynamic risk models with time-varying predictors were derived in a super-landmark framework. Contributions to model discrimination were quantified using novel inclusion-order-agnostic Shapley values of Harrel’s C-index using cross-validation.ResultsC-indices [95% CIs] were 0.74 [0.72-0.75] and 0.71 [0.67-0.77] for the models derived in all and symptomatic participants respectively. The Shapley contributions to model discrimination differed between the two groups of participants for different predictors: 31% (32% in the symptomatic participants) for core predictors (e.g., age, sex, smoking), 16% (12%) for polygenic scores, 27% (30%) for primary care blood tests, 14% (14%) for primary care medical history, 8% (0.5%) for additional lifestyle factors and 4% (12%) for symptoms.ConclusionsPolygenic scores contribute substantially to short-term risk prediction for CRC in both general and symptomatic populations; however, the contribution of information in primary care records (including presenting symptoms, medical history and common blood tests) is greater. There is, however, only a small contribution by the additional lifestyle risk factors which are not routinely collected in primary care.