Comparative efficacy and safety of stem cell-based therapies for diabetes mellitus: a systematic review and Bayesian network meta-analysis of randomized controlled trials

Background Current glucose-lowering therapies improve glycemic control but do not directly reverse β-cell loss or dysfunction. Stem cell-based therapies may offer disease-modifying potential through immunomodulatory, paracrine, and regenerative mechanisms, but the relative efficacy and safety of different cellular strategies remain uncertain. Methods PubMed, Embase, Web of Science, the Cochrane Library, and ClinicalTrials.gov were searched from inception to June 10, 2026. Randomized controlled trials of stem cell-based therapies in type 1 or type 2 diabetes were included. Outcomes were glycated hemoglobin (HbA1c), fasting plasma glucose (FPG), fasting C-peptide, total daily insulin dose (TDD), and adverse events (AEs). Random-effects Bayesian network meta-analysis estimated mean differences (MDs) or odds ratios with 95% credible intervals (CrIs). Results Twenty-one reports representing 19 randomized controlled trials involving 691 participants were included. Compared with standard medical therapy (SMT), combined mesenchymal stem/stromal cell plus mononuclear cell therapy reduced HbA1c (MD, − 1.26%; 95% CrI, − 1.97 to − 0.54; moderate confidence) and FPG (MD, − 2.24 mmol/L; 95% CrI, − 3.93 to − 0.56; moderate confidence). Mesenchymal stem/stromal cell therapy also reduced HbA1c versus SMT (MD, − 0.95%; 95% CrI, − 1.70 to − 0.09; very low confidence), and mononuclear cell therapy reduced FPG versus SMT (MD, − 2.03 mmol/L; 95% CrI, − 3.75 to − 0.35; low confidence). Combined therapy showed the most favorable ranking probabilities for HbA1c, FPG, and TDD, but its effect on TDD was not statistically credible. No statistically credible differences were observed for fasting C-peptide or AEs, and meta-regression found no credible evidence of effect modification by diabetes type. Rankings should be interpreted cautiously because the evidence was sparse and confidence in many estimates was low or very low. Conclusions Combined mesenchymal stem/stromal cell plus mononuclear cell therapy may provide favorable glycemic effects, but evidence remains insufficient for firm conclusions or routine clinical use. Adequately powered, independently conducted multicenter randomized trials are needed. Systematic review registration: PROSPERO CRD420261420380