Effects of exercise intervention on glucose and lipid metabolism in patients with type 2 diabetes mellitus complicated by non-alcoholic fatty liver disease: a three-level meta-analysis.

OBJECTIVES: This systematic review and three-level meta-analysis evaluated the effects of exercise interventions on glycemic control, lipid metabolism, body composition/fat distribution, and liver function in patients with type 2 diabetes mellitus (T2DM) complicated by non-alcoholic fatty liver disease (NAFLD), and examined potential sources of heterogeneity relevant to individualized exercise prescription. METHODS: Randomized controlled trials were searched in Web of Science, PubMed, Cochrane Library, EBSCO, Embase, and CNKI from database inception to February 2026. The review followed PRISMA 2020 and used a three-level meta-analytic model to account for dependent effect sizes extracted from the same study. A priori subgroup analyses were conducted by exercise modality, and meta-regression was used to explore heterogeneity. RESULTS: Twenty-four studies involving 2,578 participants were included. Overall, exercise reduced BMI (Hedges' g = -0.36, p < 0.001), FPG (Hedges' g = -0.42, p < 0.001), HbA1c (Hedges' g = -0.52, p < 0.001), HOMA-IR (Hedges' g = -0.48, p < 0.001), LDL-C (Hedges' g = -0.36, p = 0.001), TG (Hedges' g = -0.45, p < 0.001), ALT (Hedges' g = -0.41, p < 0.001), AST (Hedges' g = -0.38, p < 0.001), and GGT (Hedges' g = -0.40, p < 0.001), and increased HDL-C (Hedges' g = 0.28, p = 0.009). Modality-specific analyses suggested that MICT may be more favorable for FPG, LDL-C, VAT, and HFC, whereas HIIT showed larger effects for HbA1c, HOMA-IR, TG, ALT, and GGT. Heterogeneity was substantial for TG, ALT, AST, HbA1c, and GGT in the primary analyses; however, sensitivity analyses reduced I 2 below 50% without materially altering the pooled effects. Evidence for resistance training was limited by the small number of studies and should be interpreted cautiously. Meta-regression suggested that medication status and exercise frequency may be potential contributors to between-study heterogeneity. CONCLUSION: Exercise, particularly aerobic exercise, appears to be a beneficial adjunctive strategy for improving metabolic and liver-related outcomes in patients with T2DM and NAFLD. Because clinical heterogeneity, medication use, dietary co-interventions, and study quality may influence the estimates, modality-specific conclusions should be applied cautiously and individualized according to patients' metabolic profile, comorbidities, and exercise tolerance. SYSTEMATIC REVIEW REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261327724, identifier CRD420261327724.