Comparative Efficacy and Safety of Intra-Articular Adipose-Derived, Bone Marrow-Derived, and Peripheral Blood-Derived Stem Cell Injections for Knee Osteoarthritis: A Systematic Review

Background: Intra-articular (IA) stem cell injection is an emerging treatment for knee osteoarthritis (KOA). Three principal cell sources—adipose-derived mesenchymal stem cells (ADMSCs), bone marrow-derived MSCs (BMMSCs), and peripheral blood-derived stem cells (PBSCs)—have been evaluated independently; however, a systematic review comprehensively comparing all three sources under unified eligibility criteria is absent from the literature. Methods: Systematic searches of MEDLINE, Embase, Cochrane CENTRAL, and Scopus were conducted from inception to December 2025, supplemented by manual reference screening and ClinicalTrials.gov. Eligible studies included randomized controlled trials (RCTs) and prospective comparative studies in adult KOA patients. Primary outcomes were pain (VAS/NRS) and function (WOMAC, KOOS) at ≥3 months. Risk of bias was assessed using RoB 2 and ROBINS-I; evidence certainty was rated using GRADE. Results: Thirty-one studies (n = 1247; ADMSC: 14 studies, n = 612; BMMSC: 12 studies, n = 487; PBSC: 5 studies, n = 148) met inclusion criteria. Pooled standardized mean differences (SMDs) for 6-month pain showed significant reduction versus comparators for ADMSCs (SMD −1.23; 95% CI −1.61 to −0.85; I2 = 62%) and BMMSCs (SMD −1.09; 95% CI −1.55 to −0.63; I2 = 70%). PBSCs demonstrated significant within-group improvement but were too few for formal pooling. Because no trial compared cell sources head-to-head, these estimates reflect within-source efficacy versus each study’s own comparator rather than comparative superiority between sources. Adverse events were mild and transient across all sources. GRADE certainty was moderate for ADMSCs, low for BMMSCs, and very low for PBSCs. Conclusions: IA injection of ADMSCs and BMMSCs provides pain reduction and functional improvement in KOA with point estimates reaching minimal clinically important difference thresholds, although the certainty of this evidence is only moderate (ADMSC) to low (BMMSC). PBSC evidence is insufficient for formal comparison. Adequately powered, three-arm head-to-head RCTs that share a common comparator and a core outcome set are needed to establish comparative efficacy. Because only indirect comparisons were possible, this review supports efficacy within each cell source but cannot establish the superiority of one source over another.