Dose–Response Relationships Between Exercise Modalities and Triglyceride Reduction in Adults: A Systematic Review and Model-Based Network Meta-Analysis

Abstract

Highlights What are the main findings? Among 76 randomized trials, AE, CE, and HIIT reduced TG; CE produced the largest average network estimate, whereas the effect of RT was inconclusive. TG reductions increased approximately linearly across observed exercise doses, with no supported plateau or interior optimal dose. What are the implications of the main findings? Exercise prescriptions should prioritize feasible, modality-specific dose ranges rather than a single universal optimal dose. Estimates at the upper dose boundaries, particularly for CE and HIIT, should be interpreted cautiously and confirmed in dose-ranging trials. Abstract Background/Objectives: Triglyceride-rich lipoproteins contribute to residual atherosclerotic cardiovascular risk, but the comparative and dose–response effects of exercise modalities on triglycerides (TGs) remain uncertain. We compared aerobic exercise (AE), resistance training (RT), combined exercise (CE), and high-intensity interval training (HIIT) and characterized modality-specific dose–response relationships. Methods: Five databases were searched from inception through January 2026. Randomized trials in adults were synthesized using contrast-based random-effects network meta-analysis (NMA) and model-based network meta-analysis (MBNMA), with dose harmonized as MET·min/week. Results: Seventy-six independent trials included 4483 participants in 208 randomized arms. The 122 active arms generated 122 exercise–control contrasts; one CE contrast with zero reported sampling variance was excluded from both network models, leaving 121 analyzed contrasts in 86 independent comparator blocks. Exercise reduced TG by 0.149 mmol/L, on average (95% CI 0.099–0.199), although the prediction interval crossed the null. CE had the largest average network estimate (MD −0.255 mmol/L), followed by HIIT (−0.225) and AE (−0.130); RT was inconclusive. The modality-specific linear model had the lowest AICc. Per 1000 MET·min/week, slopes were −0.128 mmol/L for AE, −0.307 for CE, −0.081 for RT, and −0.239 for HIIT; the RT slope crossed the null. No supported interior optimum was identified. Conclusions: AE, CE, and HIIT were associated with lower TGs, but certainty was low, and 50 trials were at high risk of bias. CE produced the largest average estimate within the available network, not definitive evidence of universal superiority. The reported lowest and upper observed doses are descriptive boundaries rather than efficacy thresholds or prescription targets.