Association of Physical Activity With Body Composition: A Mendelian Randomization Study

Abstract

Background. Over 70% of U.S. adults are classified as overweight or obese. To reduce risk of comorbidities associated with higher body fat, weight management interventions have principally emphasized behavioral modifications, including caloric restriction and increased physical activity (PA). Observational studies suggest that modest improvement in body composition can be achieved independently from dietary interventions with adequate PA. However, these studies are limited by recall bias, response bias, and confounding factors. A separate line of evidence suggests that PA may independently reduce high-risk visceral adipose tissue (VAT), without clinically meaningful weight change. In contrast to generalized anthropometric measures, VAT is exceptionally correlated with numerous comorbidities, and cardiometabolic diseases. Genome-wide association studies have previously identified genetic markers associated with PA behaviors and may provide an opportunity to critically evaluate hypothesized causal relationships with body composition. Objective. Test the incremental benefits between various PA exposures on body composition as measured by anthropometric indexes, body fat, and VAT outcomes in a Mendelian Randomization (MR) study. Methods. Genetic instruments were identified for both self-reported and accelerometer-measured PA at various levels of metabolic-equivalent tasks (METs), and non-MET thresholds including moderate-vigorous PA (MVPA). Outcomes included anthropometric and dual-energy X-ray absorptiometry measures of adiposity (e.g. body mass index, VAT), taken from the non-overlapping UK Biobank, GIANT and GEFOS Consortium to enable two-sample MR analyses. Multivariable MR (MVMR) included educational attainment as a covariate to address potential confounding. Sensitivity analyses were evaluated for weak instrument bias and pleiotropic effects. Results & Conclusions. Genetically predicted levels of PA were not independently associated with lower levels of total body fat percent or VAT. We did not observe a dose-relationship between increasing PA intensity and body composition. Genetically predicted high-intensity PA was protective against VAT in both MVMR analysis of accelerometer-measured vigorous PA (β = -0.12, 95% Confidence Interval (CI): -0.19, -0.04) and self-reported participation in strenuous sports or other exercises (β = -0.27, 95%CI: -0.52, -0.01). These results may support evidence of biological pathways through which higher-intensity PA decreases risk of downstream cardiometabolic diseases.