Comparative effects of different exercise modalities and doses on lipid metabolism in patients with type 2 diabetes mellitus: a systematic review with Bayesian dose–response network meta-analysis

Objective This review aimed to determine whether the effects of structured exercise on triglycerides (TG) and low-density lipoprotein cholesterol (LDL-C) differed according to exercise modality and weekly training volume in adults with type 2 diabetes mellitus (T2DM). Methods Seven electronic databases were searched from inception to 18 December 2025. Randomized controlled trials were eligible when they enrolled adults with T2DM, evaluated a structured exercise intervention, and reported TG or LDL-C outcomes. Comparative effects across exercise categories were then estimated with a Bayesian random-effects network model. Weekly exercise volume was converted to MET-min/week and further examined using model-based dose–response network meta-analysis. Results Seventy RCTs comprising 5,317 adults with T2DM met the eligibility criteria. The TG analysis was based on 68 trials and 5,216 participants, whereas the LDL-C analysis included 61 trials and 4,473 participants. In the Bayesian network model, all active exercise modalities were associated with greater reductions in TG and LDL-C than the non-exercise control condition. Among the included modalities, high-intensity interval training (HIIT) produced the most favorable estimate for TG reduction (MD = -0.48 mmol/L, 95% CrI: -0.67, -0.30; SUCRA = 92.53%), while combined aerobic and resistance exercise (CE) showed the highest ranking for LDL-C improvement (MD = -0.46 mmol/L, 95% CrI: -0.63, -0.29; SUCRA = 85.45%). Dose-response modeling indicated that the lipid-lowering effects did not increase in a strictly linear manner across weekly exercise volume. The 95% CrIs remained below zero at approximately 1100–1500 MET-min/week for TG and 950–1500 MET-min/week for LDL-C. Conclusions The lipid response to exercise varied according to both training modality and weekly volume. HIIT generated the most favorable estimate for TG, whereas combined exercise showed the leading estimate for LDL-C. The identified dose intervals may provide quantitative reference points for exercise planning, but they should not be treated as fixed prescriptions because certainty of evidence, physical capacity, adherence, treatment background, and safety differ among patients. Systematic review registration https://www.crd.york.ac.uk/PROSPERO/view/CRD420261412956 , identifier CRD420261412956.