Abstract
OBJECTIVE: To examine specific physical activities' associations with biomarkers and compare metabolic equivalent of task (MET) score with biomarker-weighted score for type 2 diabetes (T2D) risk. METHODS: We used data from three large US prospective cohorts (1988/1991-2018/2019). Nine activities - walking, jogging, running, bicycling, lap swimming, tennis/squash, other aerobics, weightlifting, and heavy outdoor work - were assessed biennially. Biomarker analysis included 25,438 participants with data on C-peptide, C-reactive protein (CRP), intercellular adhesion molecule 1, leptin/adiponectin, and triglycerides/high-density lipoprotein cholesterol (TG/HDL). T2D risk analysis involved 183,892 participants followed for up to 30 years. Models were adjusted for confounders and total physical activity excluding the one of interest. RESULTS: Running demonstrated the strongest association with a healthier cardiometabolic profile, followed by jogging, aerobics and weightlifting. For every one hour/week of running, the percent difference in concentration was -4.4 % for C-peptide, -6.6 % for CRP, -11.5 % for leptin/adiponectin, and -5.6 % for TG/HDL (all P < 0.001). Adjusting for body mass index (BMI) slightly weakened the associations for these activities but they remained statistically significant. Walking and tennis/squash showed modest biomarker associations and became null after BMI adjustment. No association was observed for bicycling, lap swimming and heavy outdoor work, likely due to lacking intensity data. The biomarker-weighted score (90th vs. 10th of total physical activity, HR = 0.52; 95%CI, 0.50-0.55) had a stronger association with T2D risk than the MET score (HR = 0.61; 95%CI, 0.58-0.63). CONCLUSIONS: Types of physical activity had distinct associations with cardiometabolic biomarkers. A biomarker-weighted score performs better than MET score for T2D risk.