Physical activity and colorectal cancer risk: a two-sample Mendelian randomisation study

Abstract

BackgroundEvidence from observational studies suggests a protective role for physical activity against colorectal cancer risk. However, whether the observed association is causal or not is not clear. We did a two sample Mendelian randomisation study to examine the causal relationship between physical activity and colorectal cancer risk. MethodsThis Mendelian randomisation study used common genetic variants associated with self-reported physical activity and accelerometer-based physical activity as instrumental variables. These variables were derived from the largest available genome-wide association study of physical activity, namely, UK Biobank. The effect of the instrumental variables for physical activity was analysed in a large colorectal cancer risk genome-wide association study using summary level data that included 31 197 cases and 61 770 controls. We applied inverse variance-weighted method as the main analysis method, with Mendelian randomisation-Egger, Mendelian randomisation–pleiotropy residual sum and outlier, Mendelian randomisation-robust, leave-one-out method, mode-based estimate, and the median-based method as sensitivity analyses. FindingsOur results show a significant protective effect between accelerometer-based physical activity and colorectal cancer risk (the outlier-adjusted odds ratios [OR] of inverse variance-weighted Mendelian randomisation was 0·92 per one standard deviation increase of accelerometer-base physical activity (95% CI 0·87–0·98, p=0·0137). The effect between self-reported physical activity and colorectal cancer risk was not statistically significant but was supportive of an inverse association (the outlier-adjusted OR of inverse variance-weighted Mendelian randomisation was 0·61 per 1 SD increase of moderate-to-vigorous physical activity [95% CI 0·36–1·06, p=0·0783]). InterpretationThe findings of this large Mendelian randomisation study show, for the first time to our knowledge, that objectively measured physical activity is causally implicated in reducing colorectal cancer risk. It suggests that physical activity is not merely a biomarker of good health. The limitations of the study are that it is based on only two genetic instruments and that it has limited power, despite the study size. Nonetheless, at a population level, these findings provide strong reinforcing evidence to support public health policy measures that encourage exercise, even in obese individuals. FundingThis work was supported by cancer research UK programme grant C348/A18927 (MD). ET is supported by a cancer research UK Career Development Fellowship (C31250/A22804). XZ has a PhD studentship from The Darwin Trust of Edinburgh.