Efficacy based on dose-response and safety of bone marrow-derived mesenchymal stem cells for the treatment of osteoarthritis: a systematic review and meta-analysis Authors: Elnaz Lohrasbi, Soraya Babaie, Hadi Hamedfar, Samira Pourzeinali, Vahideh Toopchizadeh, Azizeh Farshbaf-Khalili Published: 2025 ABSTRACT AbstractIntroductionIn recent years, the use of mesenchymal stem cells (MSCs), especially those derived from bone marrow and embryonic sources, has been considered as an emerging therapeutic approach for repairing joint tissue and reducing osteoarthritis symptoms. This systematic review and meta-analysis evaluated bone marrow-derived mesenchymal stem cells (BM-MSCs) efficacy, safety, and optimal dosing for osteoarthritis treatment.MethodsThirteen randomized trials of the 1427 articles identified (11 meta-analyzed; n = 1150 patients) were identified via PubMed, Scopus, Embase, and other databases (search updated to May 2025). The primary outcomes were the optimal dose of BM-MSCs in improving pain intensity (VAS), WOMAC scores (pain, stiffness, function, total score), and their safety, analyzed using RevMan 5.4. Cochrane risk-of-bias assessment guided quality appraisal.ResultsBM-MSCs showed no significant pain reduction (VAS) or functional improvement (WOMAC) at 1–3 months’ post-injection. Significant improvements emerged at 6 months, peaking at 12 months: VAS decreased by WMD − 2.31 (95% CI − 3.83 to − 0.80; P = 0.003), and WOMAC total score reduced by WMD − 10.14 (95% CI − 18.26 to − 2.02; P = 0.01). Subgroup analyses indicated optimal doses of ≤ 25 × 10⁶ cells for VAS improvement and ≤ 25 × 10⁶ or 25–50 × 10⁶ cells for WOMAC total score at 12 months. Safety profiles were favorable, with no serious adverse events reported in most studies.ConclusionBased on the findings of this meta-analysis, BM-MSCs, especially in the long term, can be effective in reducing pain, stiffness, and improving the quality of life of patients with osteoarthritis. Also, the favorable safety of these cells makes them a promising option in joint reconstruction treatments. However, due to the heterogeneity of studies and the lack of standardized information on the optimal dose, it seems necessary to conduct prospective studies with careful design and long-term follow-up.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13287-025-04799-0.Keywords: Osteoarthritis, Mesenchymal stem cells, Bone marrow, Joint regeneration, Systematic review, Meta-analysis, Joint pain, Treatment safety Introduction Osteoarthritis is an inflammatory joint disease characterized by the destruction of cartilage and bone tissue. As a common and chronic disabling musculoskeletal disease worldwide, it imposes a heavy social and economic burden on patients and healthcare systems [1]. To date, nearly 400 million people worldwide have been living with osteoarthritis [2]. Knee osteoarthritis, as the most common site of osteoarthritis, is the most common diagnosis for arthropathy that causes discomfort and disability in older people [3]. Knee osteoarthritis becomes more common in adults with increasing age. Obesity, knee injuries, and a family history of knee osteoarthritis are risk factors for this disease [4]. The global prevalence of knee osteoarthritis is approximately 16% in people aged 15 years and older and 23% in people aged 40 years and older [5]. Epidemiological studies have reported a significant increase in the prevalence of knee osteoarthritis in the last years, highlighting the importance of early diagnosis and management approaches in suspected patients [6]. Knee osteoarthritis patients often suffer from reduced self-care ability, joint pain, and limited mobility [7]. Currently, osteoarthritis is treated using pharmacological and non-pharmacological techniques. Non-pharmacological approaches such as self-management, regular exercise, and weight control are the first-line treatment for osteoarthritis [8]. Pharmacological approaches, including nonsteroidal anti-inflammatory drugs (NSAIDs), paracetamol, topical analgesics, intra-articular corticosteroids, duloxetine, and some newer therapies such as nerve growth factor (NGF) antibodies, are recommended to relieve pain and reduce disability in knee osteoarthritis [9]. However, instead of regenerating damaged articular cartilage, the aforementioned therapies aim to reduce discomfort and increase joint mobility [10]. Although various treatment modalities have been proposed for the treatment of this disease, the use of bone marrow-derived mesenchymal stem cells (BM-MSCs) has been considered as a new and novel therapeutic approach [8, 11]. Mesenchymal stem cells (MSCs) have the potential for self-renewal, differentiation into cartilage, and immune modulation. Injection of MSCs has been demonstrated as an advanced therapeutic approach for the treatment of osteoarthritis by numerous preclinical and clinical studies [12]. The most common method of MSC administration for patients with knee osteoarthritis is intra-articular injection. Intra-articular injection of MSCs is associated with proliferation or differentiation into cartilage [13]. BM-MSCs constitute 0.001–0.01% of the bone marrow mononuclear cells. Due to their low abundance, culture and expansion of these cells under advanced in vitro conditions are required to obtain sufficient numbers for studies or clinical applications [14]. Some of the properties of BM-MSCs, such as ease of isolation from BM without causing immunological problems, ability to expand in vitro in a short period of time, bioavailability for implantation at the desired site with minimal loss of differentiation, and acceptable safety (no serious adverse reactions have been reported during autologous or allogeneic therapy), have made BM-MSCs an important tool in regenerative medicine [14]. Furthermore, BM-MSCs have demonstrated potent therapeutic potential for the treatment of a wide range of debilitating diseases, with more than 130 clinical trials currently registered/ongoing worldwide to elucidate the potential of BM-MSC-based cell therapy in the treatment of a variety of diseases [15]. The remarkable prevalence of osteoarthritis and its significant impact on patients’ quality of life underscore the importance of studies in the field of treatment methods for this disease. The use of new therapeutic methods such as regenerative medicine has been associated with promising results in recent studies. However, determining the optimal dose and safety of BM-MSCs for the treatment of osteoarthritis has not been done in previous studies, and this study was designed and implemented with this aim. This systematic review and meta-analysis was conducted to determine the appropriate dose, therapeutic effects, and safety of BM-MSCs on patients with osteoarthritis. In this study, scientific resources related to the use of BM-MSCs in the treatment of osteoarthritis was collected, reviewed, and meta-analyzed. By completing this systematic review and meta-analysis, comprehensive and complete information was obtained regarding the optimal and safe dose of BM-MSCs for the treatment of osteoarthritis, which can help improve the treatment of these patients. Methods Study design This systematic review and meta-analysis was designed and conducted according to the PRISMA 2020 (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. The study protocol was pre-written and registered in the PROSPERO database (registration ID: CRD420251104404). This study was approved by ethical committee of Tabriz University of Medical Sciences (Ethics number: IR.TBZMED.VCR.REC.1403.135). Search strategy Searches were exhaustively conducted on a systematic basis through Web of Science, Cochrane Library, PubMed, Scopus, ScienceDirect, ProQuest, ClinicalTrials.gov and Embase with no time limitation up to May 2025. The following keywords and medical subject headings (MeSH) were combined: (‘osteoarthritis’ OR ‘arthritis, degenerative’ OR ‘arthritis, noninflammatory’ OR ‘arthrosis’ OR ‘degenerative arthritis’ OR ‘degenerative joint disease’ OR ‘noninflammatory arthritis’ OR ‘osteo-arthritis’ OR ‘osteo-arthrosis’ OR ‘osteoarthritis’ OR ‘osteoarthrosis’ OR ‘primary osteoarthritis’ OR ‘rheumatoid arthrosis’) AND (‘mesenchymal stem cell’/de OR ‘mesenchymal progenitor cell’ OR ‘mesenchymal stem cell’ OR ‘mesenchymal stem cells’ OR ‘stem cell, mesenchymal’ OR ‘bone marrow cell’ OR ‘bone marrow cell’ OR ‘bone marrow cells’ OR ‘bone marrow derived cell’ OR ‘bone stem cell’ OR ‘cell, bone marrow’ OR ‘marrow stem cell’ OR ‘myeloid cell’ OR ‘myeloid cells’) AND (‘clinical trial’ OR ‘clinical drug trial’ OR ‘clinical trial’ OR ‘major clinical trial’ OR ‘trial, clinical’ OR ‘controlled clinical trial’ OR ‘clinical trial, controlled’ OR ‘controlled clinical comparison’ OR ‘controlled clinical drug trial’ OR ‘controlled clinical experiment’ OR ‘controlled clinical study’ OR ‘controlled clinical test’ OR ‘controlled clinical trial’ OR rct OR ‘randomized controlled trial’ OR ‘controlled trial, randomized’ OR ‘randomised controlled study’ OR ‘randomised controlled trial’ OR ‘randomized controlled study’ OR ‘randomized controlled trial’ OR ‘trial, randomized controlled’). A supplementary search of grey literature and reference lists from review articles was conducted using Google Scholar to identify additional relevant sources. No restriction filter was imposed on search strategy (Supplementary file 1). The PICO strategy used to establish the guiding question of this review is presented in Table 1. Table 1PICOS criteria for inclusion and exclusion of studiesParameter determined criteria for the present studyParticipantsPeople with osteoarthritis without restrictions in age and sexInterventionThe use of bone marrow mesenchymal stem cells without restrictions on dose or duration of interventionComparatorAny control group such as placebo, hyaluronic acid, corticosteroids, or no interventionOutcomesOptimal dose and safety as primary outcomes, as well as efficacy and cost-effectiveness as secondary outcomesStudy designRandomized controlled clinical trials PICOS criteria for inclusion and exclusion of studies Table 2The characteristics of the included studies investigating effects of bone marrow mesenchymal stem cells on the knee osteoarthritisStudy IDAuthors (year)CountryStudy designParticipants (e.g., number of participants, age, gender)Intervention details (e.g., MSC dosage, duration of supplementation)[30]Mautner et al. (2023)USAA phase 2/3, four-arm parallel, multicenter, single-blind, randomized, controlled clinical trial480 patients with a dx of knee OA (Kellgren– Lawrence II–IV) (4 × 120)Average age of patients in the overall study was 58.3 years. There were 214 (45.1%) males and 261 (54.9%) females who received injectionsArm 1: autologous bone marrow aspirate concentrate (n = 120), CSI (n = 40); arm 2: umbilical cord tissue-derived mesenchymal stromal cells (n = 120), CSI (n = 40); arm 3: stromal vascular fraction (n = 120), CSI (n = 40). (97-93-103-97 completed)[22]Garay-Mendoza et al. (2017)MexicoA prospective, open-label, phase I/II clinical trialA total of 61 patients were recruited: 30 in Group 1 (SC group) and 31 in control Group. Ten patients were lost during follow up, four and six, respectivelyOA grades II and III according to the Kellgren and LawrenceAutologous SC stimulation was started with the subcutaneous administration of 600 lg per day of granulocyte colony stimulating factor (G-CSF) (Biofilgran, Landsteiner Scientific, Mexico City, Mexico) for 3 consecutive days on an outpatient basis, before the procedure. SC were obtained by BM aspiration and administered in a single intra articular injection[23]Gupta et al. (2016)IndiaRandomized, double-blind, multicentric, placebo-controlled, phase II study60 patients with knee OAAt each dose level, 15 patients were randomized into two groups (Stempeucel® and placebo) in a 2:1 ratio using a computer generated randomization. Thus, 10 subjects receivedStempeucel® and 5 subjects received placebo at each dose level[24]Gupta et al. (2023)IndiaA phase 3 randomized, double-blind, multicenter, placebo-controlled study146 patients with grade 2 and 3 osteoarthritisPatients randomized to the BMMSC arm received a single intra-articular injection of 25 million BMMSCs suspended in 2 mL (20 mg) (1 mL of CryoStor CS5 1 1 mL of Plasma-Lyte A [Baxter]) of medium, followed by an injection of 2 mL of hyaluronic acid[25]Hernigou et al. (2020)FranceRandomized controlled trial140 adults aged 65 to 90 years. (mean age 75.4 ± 14.2 years) planned to undergo staged-bilateral total knee arthroplasty (TKA) for medial osteoarthritisA subchondral injection of MSCs (from iliac bone marrow concentrate) on one of the knees during the same anaestheticTreatment was performed by percutaneous injection of 20 mL BMC in the subchondral medial femorotibial compartment of each knee, i.e., 10 mL in the medial tibial plateau and 10 mL in the medial condyle.[21]Emadedin et al. (2018)IranPhase 1/2 single-center, triple-blind, randomized controlled trial (RCT) with a placebo control43 patients (Kellgren-Lawrencegrades 2, 3 and 4)Intra-articular implantation of MSCs (40 × 106cells)[19]Bastos et al. (2020)BrazilRandomized controlled, double-blind clinical trial47 patients (24 males & 23 females; 57.3 ± 10.7 years old) with knee OA in two groups (1:1)Intervention 1: one injection of 40 × 106 of autologous bone marrow-derived culture-expanded MSCs at 0 months (n = 16). Intervention 2: one injection of 40 × 106 of autologous bone marrow-derived culture-expanded MSCs + PRP at 0 months (n = 14)[28]Lamo‑Espinosa1 et al. (2018)SpainLong-term results of a prospective randomized controlled clinical trial25 patents (17 males and 8 males)Intervention 1:In the low-dose BM-MSCs group, a single intra-articular injection of 10 × 106 autologous cultured BM-MSC, followed by an intra-articular injection of 4 mL HA (n = 8)Intervention 2:In the high-dose BM-MSCs group, a single intra-articular injection of 100 × 106 autologous cultured BM-MSC, followed by an intra-articular injection of 4 mL HA (n = 8)[29]Lamo‑Espinosa1 et al. (2016)SpainPhase I/II randomized controlled clinical trial30 patients (19 males and 11 females), in two groups (1:1)Intervention 1:In the low-dose BM-MSCs group, a single intra-articular injection of 10 × 106 autologous cultured BM-MSC, followed by an intra-articular injection of 4 mL HAIntervention 2:In the high-dose BM-MSCs group, a single intra-articular injection of 100 × 106 autologous cultured BM-MSC, followed by an intra-articular injection of 4 mL HA[20]Bastos et al. (2018)BrazilRandomized clinical trial18 patients, 9 females and 9 males, 57.6 ± 9.6 years old, in two groups (1:1)Intra-articular injection of Autologous MSCs was done once at baseline 2–3 weeks after bone marrow aspiration[27]Lamo‑Espinosa1 et al. (2020)SpainPhase II randomized controlled clinical trial50 patients (33males and 17 females)Control group: 54.6 (33–70) years oldIntervention: 56 (40–62) years oldA single intra-articular injection of 100 × 106 autologous cultured BM-MSCs, followed by an intra-articular injection of 8 mL of Platelet Rich Plasma (PRGF)A weekly single intra-articular injection of PRGFR, in a final volume of 8 mL, was applied for 2 additional weeks (n = 26)[26]Ho et al. (2022)ChinaSingle-center, single-blind randomized controlled trial20 patients (12 males and 8 females), 58.00 ± 4.51years old with equal randomization in a 1:1 for two groupsA single injection of 6 mL of BM-MSCs at 1 × 106 cells per mL[31]Vega et al. (2015)SpainA multi-center Phase I-Ii Randomized Controlled Trial30 patients (13 men & 17 women); mean age ± SE (57 ± 9 years) in two groups (1:1)A single intra-articular injection of 40 × 106 cells of BM-MSCsComparator (if applicable)Dose of MSCsSource of MSCsOutcomes assessed (both primary and secondary outcomes)Results data (e.g., mean, standard deviation, confidence intervals)Adverse events & safety120 patients were candidate for Corticosteroid injection (97 completed)Corticosteroid: A 10 mL syringe containing 1 mL of depomedrol (40 mg/dL) and 6 mL of normal salineThe BMC production procedure results in 8 mL of product with 1 mL removed and 7 mL prepared for final injectionThe remaining 5 mL of the SVF were prepared for injection into the subjectsAfter thawing, 2–2.5 mL of solution remained and is aspirated from the cryobag, suspended in saline to produce 7 mL which is then prepared for injection into the subject (UCT)Of note, the BMAC and SVF were fresh autologous products while the UCT were cryopreserved, purified MSCs manufactured from donated allogeneic umbilical cord tissue in a cGMP (current Good Manufacturing Practice) facilityvisual analog scale (VAS) pain score and Knee injury and Osteoarthritis Outcome Score (KOOS) pain score at 12 months versus baselineIn addition to our primary outcome measures we also analyzed EQ-5D and PROMIS-29 between cohorts and CSIAt 1-year post injection, none of the three orthobiologic injections was superior to another, or to the CSI control none of the four groups showed a significant change in magnetic resonance imaging osteoarthritis score compared to baseline . For EQ-5D, the treatment by time interaction was not significant (P = 0.26), suggesting EQ-5D in the four treatment groups changed in similar ways (similar temporal patterns for the four treatment groups)Similarly, for PROMIS-29, we assessed all domains by a treatment by time interaction and there was no significance for the following subdomains: PROMIS-29 Anxiety (P = 0.78), Depression (P = 0.06), Fatigue (P = 0.56), Pain (P = 0.39), Physical Function (P = 0.048), Sleep (P = 0.91) and Social Rules and Activities (P = 0.82)There were no procedure-related serious adverse events (AEs) reported, which includes any allergic reactions or symptomatic infections seen in any treated patient. However, there were multiple related AEs reported that have been subdivided. The following related AEs demonstrated significance between cohorts: joint swelling (CSI 7.4% versus UCT 24.1%, P = 0.01), post-procedural contusion (SVF 38.6% versus BMAC 12.2% versus UCT/CSI 0%, P < 0.0001), post-procedural hematoma (BMAC 2.9% versus SVF 12.4%, P = 0.02)The control group received exclusively oral acetaminophen 500 mg every 8 h for 6 monthsBM-derived SC isolation was set to obtain a final volume of 10 mL for intra-articular administrationBone marrow aspirationVisual analogue scale and Western Ontario and McMaster Universities Osteoarthritis Index scores were performed at 1 week, 1 month and 6 months in both groupsThe BM-SC group showed significant improvement in knee pain and quality of life during the 6-month follow-upInitial scores for VAS and WOMAC were similar in both groups: VAS for Group 1 (x _ SD): 5.27 _ 2.196 and for Group 2 (x _ SD): 4.32 _ 2.35 (P = 0.10). WOMAC for Group 1 (x _ SD): 62.61 _ 18.55 compared to Group 2 (x _ SD) 69.93 _ 17.89 (P = 0.12). Further evaluations at 1 week, 1 month and 6 months showed a statistically significant improvement in VAS scale, and 1 month and 6 months in WOMAC for Group 1 (Table 2; Fig. 2)A patient from Group 1 presented swelling and pain in the knee the day after the SC-IA injection. Infection was ruled out and the patient recovered without further complications. Bone pain was referred by 12 patients during the stimulation with G-CSF. Some patients referred slight pain and stiffness during the first 48 h after the injectionPlacebo contained 15 mL PLASMA-LYTE A in similar cryobagsFour dose levels were studied in this trial: 25, 50, 75, and 150 million cells IMP injection was followed by injection of 2 mL hyaluronic acid (20 mg; Hyalgan, Fidia Farmaceutici S.p.A., Italy)Stempeucel® is a bone marrow-derived, ex vivo expanded, pooled, allogeneic human MSC populationSubjective evaluations—visual analog scale (VAS) for pain, intermittent and constant osteoarthritis pain (ICOAP), and Western Ontario and McMaster Universities Osteoarthritis (WOMAC-OA) index—were performed at baseline and at 1, 3, 6, and 12 months of follow-up. MRI of the knee was performed at baseline, and at 6 and 12 months follow-up for cartilage evaluationa trend towards improvement was seen in the 25-million-cell dose group in all subjective parameters (VAS, ICOAP, andWOMAC-OA scores), although this was not statistically significant when compared to placebo. The whole-organ magnetic resonance imaging score of the knee did not reveal any difference from baseline and the placebo groupAdverse events were predominant in the higher dose groups (50, 75, and 150 million cells). Knee pain and swelling were the most common adverse eventsPatients in the placebo arm received an intra-articular injection of 2 mL of placebo (1 mL of CryoStor CS5 1 1 mL of Plasma-Lyte A), followed by 2 mL of hyaluronic acidBMMSCs, which are bone marrow–derived, ex vivo– expanded, pooled, allogenic human MSCs. Cells from 3 healthy donors meeting the inclusion criteriaOf 25 million BMMSCs suspended in 2 mL (20 mg) (1 mL of CryoStor CS5 1 1 mL of Plasma-Lyte A [Baxter]) of medium, followed by an injection of 2 mL of hyaluronic acid. A total of 25 million expanded BMMSCs were cryopreserved in 1 mL of CryoStor CS5 (Sigma–Aldrich) in blinded 5-mL Crystal Zenith vials (West Pharmaceutical Services)The primary endpoint was the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) total score. The secondary endpoints were WOMAC subscores for pain, stiffness, and physical function; the visual analog scale score for pain; and magnetic resonance imaging findings using T2 mapping and cartilage volumeThe BMMSC group showed significant improvements in the WOMAC total score compared with the placebo group at 6 and 12 months (percentage change: 223.64% [95% CI 232.88 to 214.40] at 6 months and 245.60% [95% CI 255.97 to 235.23] at 12 months P\0.001; percentage change, 244.3%). BMMSCs significantly improved WOMAC pain, stiffness, and physical function subscores as well as visual analog scale scores at 6 and 12 months (P\ 0.001). T2 mapping showed that there was no worsening of deep cartilage in the medial femorotibial compartment of the knee in the BMMSC group at 12-month follow-up, whereas in the placebo group, there was significant and gradual worsening of cartilage (P \ 0.001). Cartilage volume did not change significantly in the BMMSC groupThere were 5 adverse events that were possibly/probably related to the study drug and consisted of injection site swelling and pain, which improved within a few days.No difference was shown in the incidence of adverse eventsbetwwen the study armsThey received TKA on the other kneeThe bone marrow graft of 20 cm3 volume (10 cc in the tibia and 10 cc in the femur) contained average 7800MSCs/mL (range 3120 to 11,560)Bone marrow was harvested on the contralateral anterior iliac crest of the TKA knee and preparedThe clinical results were graded according to the Knee Society knee score and the Knee Society function score [12]. During follow up, patients were evaluated with radiographs as well as functional outcomesAfter treatment with MSCs injection in bone marrow lesions of the subchondral bone, medial femorotibial compartmentBML volume experienced regression over 24 months (mean regression 1.5 cm3, range 0.8 to 3.2 cm3). At the most recent follow up (average of 15 years, range 10 to 20 years), a total of 25 (18%) of the 140 patients underwent total knee arthroplasty performed at a mean of ten years (range, 5 to 15 years) after the date of the cell therapy. The overall incidence of knee arthroplasty after cell therapy was 1.19% per person-year which was equivalent to the risk of a revision for a primary TKA in the contralateral knees of the same patient population (21 revisions, corresponding to 1.00% revision per person-year; p = 0.34). After adjusting for confounders, persistent BMLs larger than 3 cm3 after cell therapy was a strong independent risk factor for total knee arthroplasty (hazard ratio HR = 4.42 [95% CI 2.34 to 7.21]; p < 0.001), regardless of OA grade, with higher risks demonstrated for larger BMLs. Incidence rates of arthroplasty were also higher for young patients and for knees presenting severe malalignmentNot reported5 mL normal saline (placebo)MSCs (40 × 10 million cells)Bone marrow aspiration from their iliac crestsVisual analog scale(VAS), Western Ontario and McMaster Universities Arthritis Index (WOMAC) and its subscales, walking distance, painless walking distance, standing time and knee flexion compared with the placebo group at 3 and 6 months following the implantationsPatients who received MSCs experienced significantly greater improvements in WOMAC total score, WOMAC pain and physical function subscales and painless walking distance compared with patients who received placeboAccording to the WOMAC pain subscale, there was significantly greater pain relief in the MSC group compared with the placebo group. The WOMAC physical function subscale had a similar trend. This improvement was observed at the 3-month follow-up and improved further by the 6-month follow-upThe VAS score in both the MSC and placebo groups reduced over time, but this reduction was considerably higher in the MSCs group compared with the placebo groupThe PASS and MCII were not valid because data were insufficient in this studyThere were no major adverse events attributed to the MSC therapyControl group:One injection of 4 milligrams of corticosteroid at 0 months (n = 17)One injection of 40 × 106 cells at 0 monthAutologous bone marrow-derived from both posterior iliac crestsOutcomes assessed by the Knee Injury and Osteoarthritis Outcome Score KOOS and range of motion ROM at baseline, 1, 2, 3, 6, 9, and 12 months and intra-articular cytokines analysis at baseline, 6 and 12 monthsKOOS: all 3 groups showed a significant improvement from baseline to 1 month in most KOOS domains and global score (p < 0.05) and from baseline to last follow-up for most KOOS domains (p < 0.05).The MSCs and MSCs + PRP groups achieved higher percentages of expected improvement as compared to the corticosteroid group for most KOOS domains and global scoreROM: no significant differences between groups at the different follow-up endpointsCytokines analysis:All 3 groups showed a significant reduction in levels of human-IL10 from baseline to 12 months (p < 0.05)MSCs ( 0.26 ± 0.51, P = 0.023)MSCs + PRP (0.52 ± 0.99, P = 0.013)Corticosteroid (0.64 ± 1.32, P = 0.041)Not reportedReceived a single intra-articular injection of 60 mg Hyaluronic Acid (n = 9)Intervention 1:A single injection of 10 × 106 BM-MSCIntervention 2:A single injection of 100 × 106 BM-MSCAutologous bone marrow-derived from both iliac crestsVAS and WOMAC pain scoring systems assessed pain and function at 4 yearsVAS:A progressive improvement was shown during the follow-up in the BM-MSCs groups, median value (IQR) for Control, Low-dose, and High-dose groups changed from 5 (3, 7), 7 (5, 8) and 6 (4, 8) to 7 (6, 7), 2 (2, 5) and 3 (3, 4), respectively at the end of follow up (Low-dose vs. Control group, p = 0.01; High-dose vs. Control group, p = 0.004)WOMAC:A progressive improvement was shown during the follow-up in the BM-MSCs groups, median value (IQR) for Control, Low-dose, and High-dose groups changed from 27 (19, 32), 37 (30, 46) and 29 (22, 35.5) to 27 (17, 30), 17 (13, 25.5) and 16.5 (8, 23), respectively at the end of follow up (Low-dose vs. Control group, p = 0.04)No adverse effects were reported after BM-MSCs administration or during the follow-upReceived a single intra-articular injection of 60 mg Hyaluronic Acid (n = 10)Intervention 1:A single injection of 10 × 106 BM-MSCIntervention 2:A single injection of 100 × 106 BM-MSCAutologous bone marrow-derived from both iliac crestsVAS and WOMAC pain scoring systems assessed pain and function at baseline and 3, 6, and 12 monthsJoint space width was assessed by X-ray at baseline and at 6 and 12 monthsThe number and location of the lesions, cartilage thickness, signal intensity, and subchondral bone alteration and volume were assessed by MRI (WORMS protocol) at baseline and 6 and 12 monthsVAS:In the control group, pain status did not change significantlyIn intervention groups, the VAS score value was significantly reduced during all follow-ups. At 12 month, Low-dose BM-MSC group [2(1, 3), p < 0.01] and High-dose BM-MSC group [2(0, 4), p < 0.01]WOMAC:The high-dose BM-MSC group exhibited an improvement of 16.5 (12, 19) points at 12 months (p < 0.01)Improvement in control and low-dose patients could not be significantly sustained beyond 6 monthsX-ray:A borderline reduction was observed in the control group (p = 0.05 at 12 months)No changes were reported in high-dose BM-MSC groupsThe assessment in the low-dose BM-MSC group was not possibleMRI:Joint damage decreased only in the BM-MSC high-dose groupNo adverse effects were after treatment or during follow-upIntra-articular injection of Autologous MSCs with the injection of platelet-rich plasma was done once at baseline 2–3 weeks after bone marrow aspirationA single injection of 40 × 106 BM-MSCBone marrowMesenchymal stem cells derived from both posterior iliac crestsPain and knee joint function according to KOOS and clinical evaluation assessed at 0, 1, 2, 3, 6, 9, and 12 monthsthe global KOOS score improved 22.65 points (29%) for MSCsgroup(p = 0.025) and 26.4 points (30.7%) for MSCs + PRP group (p = 0.012), and there were no significant differences between the two groups, at 12 months10 adverse events(AEs) occurred in 3 patients in the MSCs group (33%) and in 2 of the MSCs + PRP group (22%). All the AEs reported were treatment-related and none required a specific procedure or hospitalization, and were resolved without any sequelae. The MSCs group reported 5 mild AEs (joint effusion, mild knee pain, low back pain, and moderate knee pain), 1 moderateAE (moderate knee pain, treated with concomitant dipyrone therapy) and 1 severe AE (intense knee pain). In the MSCs + PRP group 2 mild AEs were reported (mild knee pain) and 1 severe AE (intense knee pain, treated with concomitant dipyrone therapy)Weekly single intra-articular injection of PRGF of 8 mL, for 3 weeks (n = 24)100 × 106 autologous cultured BM-MSCsBM–MSCs from the pelvic bone (iliac crest)VAS and WOMAC pain scoring systems assessed pain and function at baseline and 3, 6, and 12 monthsJoint space width was assessed by X-ray at baseline and 12 monthsMRI (WORMS protocol) study was carried out at baseline and 12 monthsVAS:In the PRGF group, the mean (SD) at baseline and 12 months were 5 (1.8) and 4.5 (2.2), respectively (p > 0.05)In BM-MSC plus PRGFR, the mean (SD) baseline and 12 months were 5.3 (1.9) and 3.5 (2.5), respectively (p = 0.01)WOMAC:Overall WOMAC scores in the PRGF group, the mean (SD) at baseline and 12 months were 31.9 (16.2) and 22.3 (15.8), respectively (p = 0.002), and in BM-MSC plus PRGFR, the mean (SD) baseline and 12-month were 33.4 (18.7) and 23.0 (16.6), respectively (p = 0.053)Improvement in overall WOMAC scores was 37.3% and 23.4% in PRGF and BM-MSC plus PRGFR groups, respectivelyX-ray:The knee joint space during follow-up did not show any clinical effectMRI:MRI following the WORMS protocol did not show any significant change during follow-upNo adverse effects were reported after BM-MSC administration or during follow-upA single injection of 6 mL of HA6 mL of BM-MSCs at 1 × 106 cells per mLBone marrow was harvested from the posterior iliac crestOutcomes were assessed by VAS, the Knee Society Score (KSS), Knee Society Function Score (KSFS), WOMAC, and the 36-Item Short Form Survey (SF-36) questi onnaire at baseline, 3, 6, 9, and 12 months post-injection, and articular cartilage was evaluated by T2 mapping on serial MRI examinations.Between two groups:VAS scores reduced during 12 months at BM-MSCs while they increased in the HA group [− 2.91; 95% CI (− 4.98 to − 0.84); P = 0.007] and [− 2.93; 95% CI (− 4.85 to − 1.02); P = 0.003] at rest and with burden groups, respectively; but not significantly at during physical activity groupKSS [7.10; 95% CI (− 7.99 to 22.19); P = 0.35] and KFSS [4.81, − 3.88 to 13.500, P = 0.274] scores did not significantly increase over 12 monthsBM-MSCs showed a decrease in WOMAC total score [− 15.52; 95% CI (− 31.00 to − 0.04); P = 0.049]SF-36 score, showed an improvement in the physical functioning score [14.38; 95% CI (4.58 to 24.18); P = 0.005], the bodily pain score [25.26; 95% CI (7.69 to 42.82); P = 0.005], and the physical component score [9.03; 95% CI (2.81 to 15.250; P = 0.005] in the BM-MSCs groupT2 values tended to decrease more in the BM-MSCs group (17.5 ± 51.9 ms) than in the HA group (1.0 ± 5.5 ms) over the 12 month (P = 0.315)There were no observed complicationsA single injection of 60 mg of hyaluronic acid40 × 106 cells of BM–MSCsBM–MSCs was derived from the posterior iliac crestsOutcomes assessed by WOMAC, VAS, Lequesne also functional indices, SF-12, and quantitative MRI exploration at 8 days, and 3, 6, and 12 months after implantationVAS scores were reduced by 6 (P < 0.05) and 12 (P < 0.01) months after MSC transplantation, and after 12 months (P < 0.05) in the HA groupTotal WOMAC-general scores reduced by 6 and 12 months after MSC transplantation (P < 0.05), but the difference was not significant in the HA groupLequesne scores reduced by 6 (P < 0.01) and 12 (P < 0.05) months after MSC transplantation but the difference was not significant in the HA groupSF-12 scores did not show a significant differenceT2 demonstrated a significant decrease in poor cartilage areas in MSC-treated patientsNo serious adverse events occurred during treatment. Transient mild local pain and discomfort in the injected knee with inflammation and swelling during the first 1 to 7 days occurred frequently in both the control and cell-treated patients (50–60% of patients). This pain was controlled with ibuprofen The characteristics of the included studies investigating effects of bone marrow mesenchymal stem cells on the knee osteoarthritis 480 patients with a dx of knee OA (Kellgren– Lawrence II–IV) (4 × 120) Average age of patients in the overall study was 58.3 years. There were 214 (45.1%) males and 261 (54.9%) females who received injections A total of 61 patients were recruited: 30 in Group 1 (SC group) and 31 in control Group. Ten patients were lost during follow up, four and six, respectively OA grades II and III according to the Kellgren and Lawrence At each dose level, 15 patients were randomized into two groups (Stempeucel® and placebo) in a 2:1 ratio using a computer generated randomization. Thus, 10 subjects received Stempeucel® and 5 subjects received placebo at each dose level A subchondral injection of MSCs (from iliac bone marrow concentrate) on one of the knees during the same anaesthetic Treatment was performed by percutaneous injection of 20 mL BMC in the subchondral medial femorotibial compartment of each knee, i.e., 10 mL in the medial tibial plateau and 10 mL in the medial condyle. Intervention 1: In the low-dose BM-MSCs group, a single intra-articular injection of 10 × 106 autologous cultured BM-MSC, followed by an intra-articular injection of 4 mL HA (n = 8) Intervention 2: In the high-dose BM-MSCs group, a single intra-articular injection of 100 × 106 autologous cultured BM-MSC, followed by an intra-articular injection of 4 mL HA (n = 8) Intervention 1: In the low-dose BM-MSCs group, a single intra-articular injection of 10 × 106 autologous cultured BM-MSC, followed by an intra-articular injection of 4 mL HA Intervention 2: In the high-dose BM-MSCs group, a single intra-articular injection of 100 × 106 autologous cultured BM-MSC, followed by an intra-articular injection of 4 mL HA 50 patients (33males and 17 females) Control group: 54.6 (33–70) years old Intervention: 56 (40–62) years old A single intra-articular injection of 100 × 106 autologous cultured BM-MSCs, followed by an intra-articular injection of 8 mL of Platelet Rich Plasma (PRGF) A weekly single intra-articular injection of PRGFR, in a final volume of 8 mL, was applied for 2 additional weeks (n = 26) 20 patients (12 males and 8 females), 58.00 ± 4.51 years old with equal randomization in a 1:1 for two groups Corticosteroid: A 10 mL syringe containing 1 mL of depomedrol (40 mg/dL) and 6 mL of normal saline The BMC production procedure results in 8 mL of product with 1 mL removed and 7 mL prepared for final injection The remaining 5 mL of the SVF were prepared for injection into the subjects After thawing, 2–2.5 mL of solution remained and is aspirated from the cryobag, suspended in saline to produce 7 mL which is then prepared for injection into the subject (UCT) visual analog scale (VAS) pain score and Knee injury and Osteoarthritis Outcome Score (KOOS) pain score at 12 months versus baseline In addition to our primary outcome measures we also analyzed EQ-5D and PROMIS-29 between cohorts and CSI At 1-year post injection, none of the three orthobiologic injections was superior to another, or to the CSI control none of the four groups showed a significant change in magnetic resonance imaging osteoarthritis score compared to baseline . For EQ-5D, the treatment by time interaction was not significant (P = 0.26), suggesting EQ-5D in the four treatment groups changed in similar ways (similar temporal patterns for the four treatment groups) Similarly, for PROMIS-29, we assessed all domains by a treatment by time interaction and there was no significance for the following subdomains: PROMIS-29 Anxiety (P = 0.78), Depression (P = 0.06), Fatigue (P = 0.56), Pain (P = 0.39), Physical Function (P = 0.048), Sleep (P = 0.91) and Social Rules and Activities (P = 0.82) The BM-SC group showed significant improvement in knee pain and quality of life during the 6-month follow-up Initial scores for VAS and WOMAC were similar in both groups: VAS for Group 1 (x _ SD): 5.27 _ 2.196 and for Group 2 (x _ SD): 4.32 _ 2.35 (P = 0.10). WOMAC for Group 1 (x _ SD): 62.61 _ 18.55 compared to Group 2 (x _ SD) 69.93 _ 17.89 (P = 0.12). Further evaluations at 1 week, 1 month and 6 months showed a statistically significant improvement in VAS scale, and 1 month and 6 months in WOMAC for Group 1 (Table 2; Fig. 2) After treatment with MSCs injection in bone marrow lesions of the subchondral bone, medial femorotibial compartment BML volume experienced regression over 24 months (mean regression 1.5 cm3, range 0.8 to 3.2 cm3). At the most recent follow up (average of 15 years, range 10 to 20 years), a total of 25 (18%) of the 140 patients underwent total knee arthroplasty performed at a mean of ten years (range, 5 to 15 years) after the date of the cell therapy. The overall incidence of knee arthroplasty after cell therapy was 1.19% per person-year which was equivalent to the risk of a revision for a primary TKA in the contralateral knees of the same patient population (21 revisions, corresponding to 1.00% revision per person-year; p = 0.34). After adjusting for confounders, persistent BMLs larger than 3 cm3 after cell therapy was a strong independent risk factor for total knee arthroplasty (hazard ratio HR = 4.42 [95% CI 2.34 to 7.21]; p < 0.001), regardless of OA grade, with higher risks demonstrated for larger BMLs. Incidence rates of arthroplasty were also higher for young patients and for knees presenting severe malalignment Patients who received MSCs experienced significantly greater improvements in WOMAC total score, WOMAC pain and physical function subscales and painless walking distance compared with patients who received placebo According to the WOMAC pain subscale, there was significantly greater pain relief in the MSC group compared with the placebo group. The WOMAC physical function subscale had a similar trend. This improvement was observed at the 3-month follow-up and improved further by the 6-month follow-up The VAS score in both the MSC and placebo groups reduced over time, but this reduction was considerably higher in the MSCs group compared with the placebo group The PASS and MCII were not valid because data were insufficient in this study Control group: One injection of 4 milligrams of corticosteroid at 0 months (n = 17) KOOS: all 3 groups showed a significant improvement from baseline to 1 month in most KOOS domains and global score (p < 0.05) and from baseline to last follow-up for most KOOS domains (p < 0.05). The MSCs and MSCs + PRP groups achieved higher percentages of expected improvement as compared to the corticosteroid group for most KOOS domains and global score ROM: no significant differences between groups at the different follow-up endpoints Cytokines analysis: All 3 groups showed a significant reduction in levels of human-IL10 from baseline to 12 months (p < 0.05) MSCs ( 0.26 ± 0.51, P = 0.023) MSCs + PRP (0.52 ± 0.99, P = 0.013) Corticosteroid (0.64 ± 1.32, P = 0.041) Intervention 1: A single injection of 10 × 106 BM-MSC Intervention 2: A single injection of 100 × 106 BM-MSC VAS: A progressive improvement was shown during the follow-up in the BM-MSCs groups, median value (IQR) for Control, Low-dose, and High-dose groups changed from 5 (3, 7), 7 (5, 8) and 6 (4, 8) to 7 (6, 7), 2 (2, 5) and 3 (3, 4), respectively at the end of follow up (Low-dose vs. Control group, p = 0.01; High-dose vs. Control group, p = 0.004) WOMAC: A progressive improvement was shown during the follow-up in the BM-MSCs groups, median value (IQR) for Control, Low-dose, and High-dose groups changed from 27 (19, 32), 37 (30, 46) and 29 (22, 35.5) to 27 (17, 30), 17 (13, 25.5) and 16.5 (8, 23), respectively at the end of follow up (Low-dose vs. Control group, p = 0.04) Intervention 1: A single injection of 10 × 106 BM-MSC Intervention 2: A single injection of 100 × 106 BM-MSC VAS and WOMAC pain scoring systems assessed pain and function at baseline and 3, 6, and 12 months Joint space width was assessed by X-ray at baseline and at 6 and 12 months The number and location of the lesions, cartilage thickness, signal intensity, and subchondral bone alteration and volume were assessed by MRI (WORMS protocol) at baseline and 6 and 12 months VAS: In the control group, pain status did not change significantly In intervention groups, the VAS score value was significantly reduced during all follow-ups. At 12 month, Low-dose BM-MSC group [2(1, 3), p < 0.01] and High-dose BM-MSC group [2(0, 4), p < 0.01] WOMAC: The high-dose BM-MSC group exhibited an improvement of 16.5 (12, 19) points at 12 months (p < 0.01) Improvement in control and low-dose patients could not be significantly sustained beyond 6 months X-ray: A borderline reduction was observed in the control group (p = 0.05 at 12 months) No changes were reported in high-dose BM-MSC groups The assessment in the low-dose BM-MSC group was not possible MRI: Joint damage decreased only in the BM-MSC high-dose group Bone marrow Mesenchymal stem cells derived from both posterior iliac crests the global KOOS score improved 22.65 points (29%) for MSCsgroup (p = 0.025) and 26.4 points (30.7%) for MSCs + PRP group (p = 0.012), and there were no significant differences between the two groups, at 12 months 10 adverse events(AEs) occurred in 3 patients in the MSCs group (33%) and in 2 of the MSCs + PRP group (22%). All the AEs reported were treatment-related and none required a specific procedure or hospitalization, and were resolved without any sequelae. The MSCs group reported 5 mild AEs (joint effusion, mild knee pain, low back pain, and moderate knee pain), 1 moderate AE (moderate knee pain, treated with concomitant dipyrone therapy) and 1 severe AE (intense knee pain). In the MSCs + PRP group 2 mild AEs were reported (mild knee pain) and 1 severe AE (intense knee pain, treated with concomitant dipyrone therapy) VAS and WOMAC pain scoring systems assessed pain and function at baseline and 3, 6, and 12 months Joint space width was assessed by X-ray at baseline and 12 months MRI (WORMS protocol) study was carried out at baseline and 12 months VAS: In the PRGF group, the mean (SD) at baseline and 12 months were 5 (1.8) and 4.5 (2.2), respectively (p > 0.05) In BM-MSC plus PRGFR, the mean (SD) baseline and 12 months were 5.3 (1.9) and 3.5 (2.5), respectively (p = 0.01) WOMAC: Overall WOMAC scores in the PRGF group, the mean (SD) at baseline and 12 months were 31.9 (16.2) and 22.3 (15.8), respectively (p = 0.002), and in BM-MSC plus PRGFR, the mean (SD) baseline and 12-month were 33.4 (18.7) and 23.0 (16.6), respectively (p = 0.053) Improvement in overall WOMAC scores was 37.3% and 23.4% in PRGF and BM-MSC plus PRGFR groups, respectively X-ray: The knee joint space during follow-up did not show any clinical effect MRI: MRI following the WORMS protocol did not show any significant change during follow-up Between two groups: VAS scores reduced during 12 months at BM-MSCs while they increased in the HA group [− 2.91; 95% CI (− 4.98 to − 0.84); P = 0.007] and [− 2.93; 95% CI (− 4.85 to − 1.02); P = 0.003] at rest and with burden groups, respectively; but not significantly at during physical activity group KSS [7.10; 95% CI (− 7.99 to 22.19); P = 0.35] and KFSS [4.81, − 3.88 to 13.500, P = 0.274] scores did not significantly increase over 12 months BM-MSCs showed a decrease in WOMAC total score [− 15.52; 95% CI (− 31.00 to − 0.04); P = 0.049] SF-36 score, showed an improvement in the physical functioning score [14.38; 95% CI (4.58 to 24.18); P = 0.005], the bodily pain score [25.26; 95% CI (7.69 to 42.82); P = 0.005], and the physical component score [9.03; 95% CI (2.81 to 15.250; P = 0.005] in the BM-MSCs group T2 values tended to decrease more in the BM-MSCs group (17.5 ± 51.9 ms) than in the HA group (1.0 ± 5.5 ms) over the 12 month (P = 0.315) VAS scores were reduced by 6 (P < 0.05) and 12 (P < 0.01) months after MSC transplantation, and after 12 months (P < 0.05) in the HA group Total WOMAC-general scores reduced by 6 and 12 months after MSC transplantation (P < 0.05), but the difference was not significant in the HA group Lequesne scores reduced by 6 (P < 0.01) and 12 (P < 0.05) months after MSC transplantation but the difference was not significant in the HA group SF-12 scores did not show a significant difference T2 demonstrated a significant decrease in poor cartilage areas in MSC-treated patients Inclusion and exclusion criteria All randomized controlled trials (RCTs) published on investigating effects of bone marrow mesenchymal stem cells on human subjects diagnosed with knee osteoarthritis, involving intra-articular injection of BM-MSCs and reporting quantitative data on pain, function, stiffness or whole-organ magnetic resonance imaging score (WORMS), regardless of the study time was reviewed. Studies not being interventional randomized trials, not reporting intended outcomes, or involving composite injections were excluded. All observational (cross-sectional, case-control and cohort), animal studies, case reports, narrative reviews, letters to the editor, protocols, or conference abstracts without full data and systematic reviews were also excluded. In this study, the primary outcomes were efficacy based on dose-response and safety, and cost-effectiveness was as secondary outcome (Fig. 1). Flow diagram of study process Study selection Two reviewers (E.L, S.B) independently screened the titles and abstracts identified in the literature search after ethics committee approval. Both reviewers (E.L, S.B) downloaded the full texts of all potentially relevant studies for independent assessment and data extraction. Any disagreement with article selection was resolved through discussion with a third reviewer (A.F.Kh). Data extraction A predesigned extraction form was used by two authors (H.H, S.P) to independently screen the literature and extract the data. The following data were extracted: the first author’s name, journal, year of publication, study design, participants (e.g., number of participants, age, gender), intervention details (e.g., BM-MSC dosage, duration of supplementation), comparator, dose of BM-MSCs, source of BM-MSCs, outcomes assessed (both primary and secondary outcomes), adverse events & safety. Quality assessment Bias risk evaluation was conducted through dual independent assessment by two researchers following Cochrane Collaboration guidelines [16]. Domains analyzed included randomization procedures, allocation concealment, participant/personnel blinding, outcome assessor masking, attrition, and reporting biases. Discrepancies in judgments were resolved via third-party adjudication (V.T.T). Visual representation of bias profiles was generated using Review Manager 5.3 (RevMan; Cochrane Collaboration, UK). Statistical analysis Quantitative synthesis utilized RevMan 5.4.1 (Nordic Cochrane Centre, Denmark). Treatment effects were quantified as weighted mean differences (WMD) with 95% confidence intervals (CI) between intervention and control arms. Pre-post intervention changes were derived by subtracting follow-up means from baseline values. Standard deviations (SD) were calculated using: where SDb, SDf, and r represent baseline SD, follow-up SD, and baseline-follow-up correlation [17]. Median/interquartile range (IQR) data were converted to mean ± SD via validated methods [18]. Standard errors (SE) were transformed to SD using SD = SE×√n. For studies lacking numerical data, Universal Desktop Ruler 3.8 software extracted values from graphical representations. All conversions adhered to standardized methodologies to ensure analytical consistency. The chi-squared distribution and the I² statistic assessed the study heterogeneity. A random effects model was used when I² was greater than 50%; otherwise, a fixed effects model was adopted. If heterogeneity was observed, sensitivity analysis was performed through a sequential exclusion method, and a funnel plot was generated to assess publication bias. In cases of different scoring for Visual Analogue Scale (VAS) and Western Ontario and McMaster Universities (WOMAC), they converted to standard and identical scale. For different efficacy outcomes, sub-group analyses were done based on diverse applied BM-MSCs doses as: ≤25 × 106 cells, 25 to 50 × 106 cells, ≥50 × 106 cells, or non-reported dose. Two studies had no reported outcomes similar to other included studies, and so explained only as qualitative data (systematic review) [19, 20]. Results Study selection Figure 1 depicts the search and screening process for the literature based on the PRISMA statement. The initial search yielded a total of 1428 articles alongside four additional records identified through other sources. After eliminating duplicate entries (n = 612), 820 publications were subjected to eligibility assessment. Applying the inclusion and exclusion criteria led to the exclusion of 792 articles. Additionally, at final stage, 15 full-text articles excluded due to being not interventional randomized study, not intended outcomes and composite injections. Ultimately, an analysis was performed on 13 eligible publications comprising RCTs and prospective clinical trials were included in systematic review [19–31], and 11 publications entered in meta-analysis [21–31]. The included studies had publication dates ranging from 2015 to 2023 across various countries including the USA, Mexico, India, France, Iran, Brazil, China, and Spain. Study characteristics A total of 1150 participants were involved across the 13 studies. Among these participants, 824 were assigned to the intervention group (MSC-treated patients), while 466 were in the control group receiving usual care (In Hernigou et al., each patient served as their own control). From 18 to 480 was the sample sizes range. Participants (both genders) had the mean age between 55 and 75 years. According to Kellgren-Lawrence (KL) evaluations, the osteoarthritis grade usually ranges between II and IV. The intra-articular administration of MSCs were assessed in all included studies. Ex vivo-expanded cell preparations, stromal vascular fractions, or fresh bone marrow aspirate concentrates were used to deliver MSCs. There were significant differences in the dosages, ranging from 6 million to 150 million cells per injection, with most studies administering 25, 40, or 100 million MSCs in a single shot. Multiple trials have combined MSCs with hyaluronic acid (HA) or platelet-rich plasma (PRP). The comparators across studies comprised of placebo (saline or vehicle medium) [21, 23, 24], corticosteroid injections [19, 30], PRP [20], hyaluronic acid [26, 28, 29, 31], and conventional pharmacological therapy (e.g., acetaminophen) [22, 25, 27]. The contralateral knee receiving total knee arthroplasty in one study [25]. Primary and secondary outcomes varied across studies but generally included: pain and function (assessed using the Visual Analog Scale (VAS), Western Ontario and McMaster Universities Arthritis Index (WOMAC), Knee injury and Osteoarthritis Outcome Score (KOOS), Knee Society Scores (KSS), and Lequesne index, structural evaluation (assessed using MRI (T2 mapping, WORMS score), X-ray (joint space width), and quantitative cartilage assessments), and quality of life and global function (assessed using SF-12, SF-36, EQ-5D, PROMIS-29). Follow-up durations ranged from 6 to 24 months, with one study reporting long-term follow-up data up to 15 years [25]. Table 1 presents detailed information about the included studies, including their methodologies, participant characteristics, and key findings. This table serves as a valuable resource for understanding the context and outcomes of the research conducted on MSCs efficacy and safety in treatment of osteoarthritis. Risk of bias assessments Risk of bias assessment of the included studies showed that most studies had low to moderate risk of bias in most domains. The highest level of concern was related to “allocation concealment” and “blinding of outcome assessment”. Some studies also had ambiguity in the area of ​​“selective reporting”. In addition, more than 70% of the studies had low risk of bias in the area of ​​randomization. However, in the areas of allocation concealment and blinding, a higher percentage of studies had unclear or high risk of bias (Figs. 2 and 3). Risk of bias summary: review authors’ judgments about each risk of bias item for each included study Risk of bias graph: review authors’ judgements about each risk of bias item presented as percentages across all included studies Meta-analysis results In this section, the results of the meta-analysis of data extracted from RCTs included in the study are presented by outcome variables. VAS assessment In the analysis of VAS in the first month after the intervention, 4 studies (with 7 between-group comparisons) were included in the analysis. Heterogeneity analysis showed an I2 level of 52% and a p-value of 0.05. Therefore, a random effects model was applied. As a result of meta-analysis, MSC-treated patients and control groups showed no statistically significant differences (P = 0.44), which indicates there may not be a significant reduction in VAS after using MSCs in osteoarthritis patients [WMD: 0.20 (95% CI − 0.30, 0.70)]. Similarly, sub-group analysis based on dose classification illustrated no significant inter-group diversity in this regard (p > 0.05) (Fig. 4). In evaluating VAS score after 3 months intervention, six RCTs (with nine between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 83% and a p-value less than 0.00001, indicating statistical heterogeneity among five studies. Therefore, the random effects model was applied. As a result of meta-analysis, MSC-treated patients and control groups showed no statistically significant differences (P = 0.40), which indicates there may not be a significant reduction in VAS using MSCs in osteopathists patients after 3 months [WMD: − 0.25 (95% CI − 0.83, 0.33)]. Sub-group analysis based on dose classification indicated no significant between-group differences (p > 0.05) (Fig. 5) In evaluating VAS score after 6 months intervention, eight RCTs (with 11 between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 81% and a p-value less than 0.00001, indicating statistical heterogeneity among eight studies. Therefore, the random effects model was applied. As a result of meta-analysis, MSC-treated patients and control groups showed statistically significant differences (p = 0.01), which indicates there may be a significant reduction in VAS using MSCs in osteopathists patients after 6 months [WMD: − 0.65 (95% CI − 1.16, − 0.13)]. Sub-group analysis according to dose classification indicated a significant difference only in the applied dose less than 25 × 106 cells [WMD: − 1.33 (95% CI − 1.83, − 0.84); p < 0.0001] (Fig. 6a). Potential publication bias was assessed using a funnel plot. It appears that there is no substantial evidence of publication bias based on the distribution of studies on the plot. Although the number of studies is limited, it is difficult to detect asymmetry, particularly among certain subgroups (Fig. 6b). After 12-month intervention based on eight included RCTs (with 12 between-group comparisons), MSC-treated patients and control groups showed statistically significant differences (P = 0.003) using random effect (I2 = 92%), which indicates there may be a significant reduction in VAS using MSCs in osteopathists patients after 12 months [WMD: − 2.31 (95% CI − 3.83, − 0.80)]. Sub-group analysis according to dose classification indicated a significant difference only in the dose less than 25 × 106 cells [WMD: − 5.25 (95% CI − 9.19, − 1.32); p = 0.009] (Fig. 7a). The funnel plot indicated that there is no significant publication bias for VAS after 12-month intervention. The distribution of studies around the mean effect appears to be relatively uniform and symmetrical. Studies with higher precision (top) and lower precision (bottom) are well located on either side of the mean (Fig. 7b). Forest plot of comparison of VAS score at 1 month Forest plot of comparison of VAS score at 3 months a Forest plot of comparison of VAS score at 6 months. b Funnel plot of VAS score at 6 months a Forest plot of comparison of VAS score at 12 months. b Funnel plot of VAS score at 12 months WOMAC-overall score assessment In the analysis of WOMAC-overall score in the first month after the intervention, 3 studies (with six between-group comparisons) were included in the analysis. Heterogeneity analysis showed an I2 level of 0% and a p-value of 0.63. Therefore, a fixed effects model was applied. As a result of meta-analysis, MSC-treated patients and control groups showed no statistically significant differences (p = 0.37), which indicates there may not be a significant reduction in WOMAC-overall score after using MSCs in osteoarthritis patients after 1 month [WMD: 2.00 (95% CI − 1.30, 5.30)]. Sub-group analysis according to dose classification indicated no significant difference between groups (p > 0.05) (Fig. 8). In evaluating WOMAC-overall score after 3 months intervention, five RCTs (with nine between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 31% and a p-value of 0.17, indicating no statistical heterogeneity among five studies. Therefore, the fix effects model was applied. As a result of meta-analysis, MSC-treated patients and control groups showed statistically significant differences (p = 0.04), which indicates there may be a significant reduction in WOMAC-overall score using MSCs in osteopathists patients after 3 months [WMD: − 3.46 (95% CI − 6.68, − 0.23)]. Sub-group analysis according to dose classification indicated a significant difference only in the dose less than 25 × 106 cells [WMD: − 5.29 (95% CI − 8.98, − 1.60); p = 0.005] (Fig. 9). In evaluating WOMAC-overall score after 6 months intervention, seven RCTs (with 11 between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 96% and a p-value less than 0.00001, indicating statistical heterogeneity among seven studies. Therefore, the random effects model was applied. As a result of meta-analysis, MSC-treated patients and control groups showed no significant differences (p = 0.80), which indicates there may not be a significant reduction in WOMAC-overall score using MSCs in osteopathists patients after 6 months [WMD: − 1.11 (95% CI − 9.52, 7.30)]. Sub-group analysis according to dose classification indicated significant inter-group increases in the dose more than 50 × 106 cells [WMD: 1.87 (95% CI 0.42, 3.31); p = 0.01] and not reported dose [WMD: 9.67 (95% CI 1.50, 17.84); p = 0.02] (Fig. 10a). Potential publication bias was assessed using a funnel plot. As shown in Fig. 10b, there is a significant asymmetry in studies evaluating 6-month MSCs intervention in patients. Most of the studies with small sample sizes (bottom of the graph) are on the left side of the axis, which may indicate publication bias. This could mean that studies with negative or insignificant results are not published. After 12-month intervention, seven studies (with 12 between-group comparisons) were included. MSC-treated patients and control groups comparison using random effect (I2 = 85%) showed statistically significant differences (p = 0.01), which indicates there may be a significant reduction in WOMAC-overall score using MSCs in osteopathists patients after 12 months [WMD: − 10.14 (95% CI − 18.26, − 2.02)]. Sub-group analysis according to dose classification indicated significant inter-group reductions in the doses equal or less than 25 × 106 cells [WMD: − 17.51 (95% CI − 28.87, − 6.16); p = 0.003] and 25 < cells ≤ 50 × 106 [WMD: − 8.72 (95% CI − 16.78, − 0.65); p = 0.03] (Fig. 11a). The funnel graph is relatively symmetrical, but the lower part (studies with high standard errors) is more concentrated on one side. Although this asymmetry is slight, it could be an indication of a slight bias in the publication of small studies (Fig. 11b). Forest plot of comparison of WOMAC-overal score at 1 month Forest plot of comparison of WOMAC-overall score at 3 months a Forest plot of comparison of WOMAC-overall score at 6 months. b Funnel plot of WOMAC-overal score at 6 months a Forest plot of comparison of WOMAC-overall score at 12 months. b Funnel plot of WOMAC-overall score at 12 months WOMAC-pain score assessment In the analysis of WOMAC-pain score in the first month after the intervention, three studies (with six between-group comparisons) were included in the analysis. Heterogeneity analysis showed an I2 level of 88% and a p value less than 0.00001, indicating statistical heterogeneity among three studies. Therefore, the random effects model was used. As a result of meta-analysis, MSC-treated patients and control groups showed statistically significant increment [WMD: 19.40 (95% CI 4.28, 34.52); p = 0.01]. Sub-group analysis based on dose classification indicated significant inter-group increase just in the doses more than 50 × 106 cells [WMD: 53.81 (95% CI 37.44, 70.18); p < 0.00001] (Fig. 12). In evaluating WOMAC-pain score after 3 months intervention, five RCTs (with nine between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 0% and a p value of 0.61, indicating statistical heterogeneity among six studies. As a result of meta-analysis, MSC-treated patients and control groups showed no statistically significant differences (p = 0.58), which indicates there may not be a significant reduction in WOMAC-pain score using MSCs in osteopathists patients after 3 months [WMD: 0.27 (95% CI − 0.67, 1.20)]. Similarly, sub-group analysis based on dose classification illustrated no significant inter-group diversity in this regard (p > 0.05) (Fig. 13). In evaluating WOMAC-pain score after 6 months intervention, seven RCTs (with 11 between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 89% and a p value less than 0.00001, indicating statistical heterogeneity among eight studies. Therefore, the random effects model was applied. As a result of meta-analysis, MSC-treated patients and control groups showed no significant differences (p = 0.08), which indicates there may not be a significant reduction in WOMAC-pain score using MSCs in osteopathists patients after 6 months [WMD: − 3.66 (95% CI − 7.70, 7.30)]. Sub-group analysis based on dose classification indicated contradicting results, with significant decrease in the dose 25 < cells ≤ 50 × 106 [WMD: − 20.86 (95% CI − 30.40, − 11.31); p < 0.00001] vs. increase in the dose more than 50 × 106 cells [WMD: 1.23 (95% CI 0.06, 2.40); p = 0.04] compared to control group (Fig. 14a). As shown in Fig. 14b, several studies with large deviations from the center line are observed, creating asymmetry. This pattern could be an indication of publication bias or significant heterogeneity among studies. After 12-month intervention, six RCTs (with 10 between-group comparisons) with high heterogeneity (I2=87%) were included. MSC-treated patients and control groups showed statistically significant differences (p = 0.009) using random effects model, which indicates there may be a significant reduction in WOMAC-pain score using MSCs in osteopathists patients after 12 months [WMD: − 5.02 (95% CI − 8.77, − 1.27)]. Sub-group analysis based on dose classification indicated significant decrease merely in the dose 25 < cells ≤ 50 × 106 [WMD: − 9.57 (95% CI − 17.94, − 1.20); p = 0.02] (Fig. 15a). Funnel plot has a relatively symmetrical distribution, but the number of studies is relatively small. For this reason, it is not possible to definitively judge the absence of publication bias. In cases with small study sizes, the power of the funnel plot to detect bias is limited. (Fig. 15b). Forest plot of comparison of WOMAC-pain score at 1 month Forest plot of comparison of WOMAC-pain score at 3 months a Forest plot of comparison of WOMAC-pain score at 6 months. b Funnel plot of WOMAC pain score at 6 months a Forest plot of comparison of WOMAC-pain score at 12 months. b Funnel plot of WOMAC pain score at 12 months WOMAC-stiffness score assessment In the analysis of WOMAC-stiffness score in the first month after the intervention, three studies (with six between-group comparisons) were included in the analysis. Heterogeneity analysis showed an I2 level of 82% and a p value less than 0.00001, indicating statistical heterogeneity among three studies. Therefore, the random effects model was used. As a result of meta-analysis, MSC-treated patients and control groups showed no statistically significant differences [P = 0.43; WMD: − 5.90 (95% CI − 20.58, 8.79)]. Sub-group analysis according to dose classification indicated no significant difference between groups (p > 0.05) (Fig. 16). In evaluating WOMAC-stiffness score after 3 months intervention, five RCTs (with nine between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 0% and a p value of 0.44, indicating no statistical heterogeneity among five studies. As a result of meta-analysis using fix effects model, MSC-treated patients and control groups showed no statistically significant differences (P = 0.05), which indicates there may not be a significant reduction in WOMAC-stiffness score using MSCs in osteopathists patients after 3 months [WMD: − 0.60 (95% CI − 1.21, 0.01)]. Sub-group analysis based on dose classification indicated significant decrease merely in the dose equal or less than 25 × 106 [WMD: − 1.71 (95% CI − 2.93, − 0.49); p = 0.006] (Fig. 17). In evaluating WOMAC-stiffness score after 6 months intervention, seven RCTs (with eleven between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 76% and a p value less than 0.00001, indicating statistical heterogeneity among seven studies. Therefore, the random effects model was applied. As a result of meta-analysis, MSC-treated patients and control groups showed no significant differences (P = 0.10), which indicates there may not be a significant reduction in WOMAC-stiffness score using MSCs in osteopathists patients after 6 months [WMD: − 1.68 (95% CI − 3.71, 0.34)]. Sub-group analysis according to dose classification indicated no significant difference between groups (p > 0.05) (Fig. 18a). As shown in Fig. 18b, the Funnel graph is highly skewed. Most of the points are on one side of the central axis, and on the other side, there is a clear lack of studies. This could indicate serious publication bias or even a file drawer effect. After 12-month intervention, five RCTs (with nine between-group comparisons) were included. MSC-treated patients and control groups showed statistically significant differences (P = 0.004) using the random effects model (I2 = 92%), which indicates there may be a significant reduction in WOMAC-stiffness score using MSCs in osteopathists patients after 12 months [WMD: − 4.37 (95% CI − 7.31, − 1.42)]. Sub-group analysis based on dose classification showed significant decrease only in the dose ≤ 25 × 106 [WMD: − 10.07 (95% CI − 17.64, − 2.50); p = 0.009] (Fig. 19). Forest plot of comparison of WOMAC-stiffness score at 1 month Forest plot of comparison of WOMAC-stiffness score at 3 months a Forest plot of comparison of WOMAC-stiffness score at 6 months. b Funnel plot of WOMAC stiffness-score stiffness at 6 months Forest plot of comparison of WOMAC-stiffness score at 12 months WOMAC-function score assessment In the analysis of WOMAC-function score in the first month after the intervention, 3 studies (with six between-group comparisons) were included in the analysis. Heterogeneity analysis showed an I2 level of 0% and a p value of 0.67, indicating lack statistical heterogeneity among three studies. As a result of meta-analysis using fix effects moded, MSC-treated patients and control groups showed no statistically significant differences [p = 0.25; WMD: 2.15 (95% CI − 1.50, 5.81)]. Sub-group analysis according to dose classification indicated no significant difference between groups (p > 0.05) (Fig. 20). In evaluating WOMAC-function score after 3 months intervention, five RCTs (with nine between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 47% and a p value of 0.06, indicating lack of statistical heterogeneity among five studies. As a result of meta-analysis, MSC-treated patients and control groups showed no statistically significant differences (p = 0.61), which indicates there may not be a significant reduction in WOMAC-function score using MSCs in osteopathists patients after 3 months [WMD: − 0.74 (95% CI − 3.61, 2.12)]. Sub-group analysis based on dose classification indicated significant decrease merely in the dose 25 < cells ≤ 50 × 106 [WMD: − 32.20 (95% CI − 55.02, − 9.38); p = 0.006] (Fig. 21). In evaluating WOMAC-function score after 6 months intervention, seven RCTs (with eleven between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 84% and a P value less than 0.00001, indicating statistical heterogeneity among seven studies. Therefore, the random effects model was applied. As a result of meta-analysis, MSC-treated patients and control groups showed no significant differences (p = 0.68), which indicates there may not be a significant reduction in WOMAC-stiffness score using MSCs in osteopathists patients after 6 months [WMD: − 1.59 (95% CI − 9.17, 5.99)]. Sub-group analysis based on dose classification indicated significant decrease based on two included studies in the dose 25 < cells ≤ 50 × 106 [WMD: − 13.04 (95% CI − 19.78, − 6.30); p = 0.00001] vs. significant increase in one non dose-reported RCT [WMD: 13.21 (95% CI 4.94, 21.48); p = 0.002] (Fig. 22a). As shown in Fig. 22b, the Funnel graph is highly skewed. Most of the points are on one side of the central axis, and on the other side, there is a clear lack of studies. This could indicate serious publication bias or even a file drawer effect. After 12-month intervention, five RCTs (with nine between-group comparisons) were included. MSC-treated patients and control groups using random effects model (I2 = 90%) showed no statistically significant differences (p = 0.13), which indicates there may not be a significant reduction in WOMAC-stiffness score using MSCs in osteopathists patients after 12 months [WMD: − 8.51 (95% CI − 19.74, 2.57)]. Sub-group analysis according to dose classification indicated no significant difference between groups (p > 0.05) (Fig. 23). Forest plot of comparison of WOMAC-function score at 1 month Forest plot of comparison of WOMAC-function score at 3 months a Forest plot of comparison of WOMAC-function score at 6 months. b Funnel plot of WOMAC function score at 6 months Forest plot of comparison of WOMAC-function score at 12 months WORMS assessment In the analysis of WORMS in the 6 months after the intervention, three studies (with six between-group comparisons) were included in the analysis. Heterogeneity analysis showed an I2 level of 0% and a p value of 1.00, indicating lack statistical heterogeneity among three studies. As a result of meta-analysis, MSC-treated patients and control groups showed no statistically significant differences [p = 0.96; WMD: 0.23 (95% CI − 9.37, 9.83)]. Sub-group analysis according to dose classification indicated no significant difference between groups (p > 0.05) (Fig. 24). In evaluating WORMS after 12 months intervention, three RCTs (with six between-group comparisons) were included. Heterogeneity analysis returned an I2 level of 0% and a p value of 0.97, indicating lack of statistical heterogeneity among two studies. As a result of meta-analysis, MSC-treated patients and control groups showed no statistically significant differences (p = 0.52), which indicates there may not be a significant reduction in WORMS using MSCs in osteopathists patients after 12 months [WMD: − 3.12 (95% CI − 12.70, 6.45)]. Sub-group analysis according to dose classification indicated no significant difference between groups (p > 0.05) (Fig. 25). Forest plot of comparison of WORMS at 6 months Forest plot of comparison of WORMS at 12 months Discussion Despite the availability of multiple studies on the use of BM-MSCs for treating osteoarthritis, the optimal dosage and safety of these cells for this treatment have not yet been clarified; therefore, this study provides reliable evidence on the efficacy based on dose-response and safety of MSCs in the treatment of osteoarthritis, especially knee osteoarthritis, utilizing high-quality randomized clinical trials. This study, which was conducted by carefully reviewing 13 RCTs and analyzing data from 11 of them, was able to draw a more accurate picture of the improvement of symptoms over time after stem cell injection. As a chronic, debilitating musculoskeletal disease that mostly affects the knees, hips and hands, osteoarthritis is considered as one of the most common chronic diseases in the world [32]. In elderly people and those with sedentary lifestyles, this disease often decreases motor performance, causes chronic pain, disrupts daily activities, and reduces quality of life [33]. As a result of life expectancy increases and a growing elderly population over the past few decades, osteoarthritis rates have also increased significantly in the term of prevalence and incidence [34]. Nevertheless, health systems are burdened with the costs of treating and managing this disease, including medication, physiotherapy, joint replacement surgeries, and work absences [34]. It is estimated that millions of people suffer from pain and limitations caused by this disease [35]. In severe cases, total joint replacement is the only remaining treatment option, and it can be challenging as well [35]. Current treatments, including nonsteroidal anti-inflammatory drugs, corticosteroid injections, physiotherapy, and joint supplements, have only been able to provide temporary relief of symptoms, but none of them can stop or reverse the process of articular cartilage destruction [36]. Meanwhile, regenerative therapies, such as MSC therapy, are becoming increasingly popular. Besides inhibiting inflammation and relieving symptoms, these therapies can also stimulate repair processes and regenerate joint tissues, which can have longer-term effects [37]. In this regard, the present study was designed to evaluate the efficacy exploring the optimal dose and safety of bone marrowstem cells in the treatment of patients with osteoarthritis. The focus of this study was on examining clinical outcomes such as pain severity, joint stiffness, and motor function, as well as determining the optimal dose and analyzing safety data, which could help provide more precise evidence about the benefits or limitations of this method. The findings of this study, especially in the long term, will play an important role in clinical decision-making, the design of future trials, and the expansion of the application of cell therapy in degenerative joint diseases. What is clear from the review of the results is that the effectiveness of stem cells in improving osteoarthritis symptoms is mainly associated with a time delay. The most important outcomes of the intervention group were not significantly different from those of the control group after short-term assessments (such as 1 and 3 months after the intervention). In the long term, particularly after 6 and 12 months, some clinically important outcomes improved significantly. One of the most important pain measurement tools is the VAS scale. While no statistically significant difference was observed between the two groups during the first month and 3 months after the intervention, a significant reduction in pain was reported at the sixth month, which continued to be more pronounced at the twelfth month. The optimal and effective dose among studies according to sub-group analysis was minimal applied dose i.e. ≤25 × 10 cells. This indicates that MSCs, through regenerative and anti-inflammatory mechanisms, begin to exert their effects on pain reduction over time. The decrease in VAS score at the 12th month was clinically significant and indicated significant pain relief for the patients. The assessment of the WOMAC total score, a comprehensive measure of pain, joint stiffness, and motor function, also showed a similar pattern. A modest decrease in the overall WOMAC score was observed at the first and third months, but a more dramatic and statistically significant difference appeared at the 12 month. The optimal and effective dose among studies according to sub-group analysis was ≤ 25 × 106 cells or 25 < cells ≤ 50 × 106 cells. These findings suggest that MSC injection not only helps relieve pain, but also plays an effective role in improving patients’ daily functioning and reducing their physical limitations. An evaluation of the individual components of the WOMAC questionnaire yielded mixed results. There was a significant increment in WOMAC-Pain score at the first month compared to control which was observed in cases of using more than 50 × 106 cells, but this improvement did not continue at the third month, while significant reduction was seen at the 12 month assessment with optimal dose of 25 < cells ≤ 50 × 106 cells. Based on this pattern, it appears that the severity of disease, an individual’s immune response, or other intervening factors may influence pain response to treatment. The WOMAC-Stiffness scores also showed that the effect of stem cells in the short term was not very clear, but at 12 months, a significant improvement was observed with the effective and optimal dose of ≤ 25 × 106 cells. Since joint stiffness is one of the annoying aspects of osteoarthritis that reduces quality of life, especially in the early hours of the day, this finding is of great clinical importance. The WOMAC-Function score also showed a non-significant improvement at most time points, however, a significant decrease in the function score was reported at 12 months, which, although statistically at the threshold of significance, from a clinical perspective indicates an improvement in functional independence and a decrease in motor disability of patients. At 6 and 12 months, no significant differences were found in the WORMS, which measures knee function; this suggests that this scale may be less sensitive to detecting gradual changes due to cell therapy. In addition to pain-related measures, by analyzing instruments such as KOOS and SF-36, researchers have improved understanding of how interventions affect functional outcomes and quality of life among patients. In three studies, KOOS was used to assess functional outcomes. According to Mautner et al., a phase 2/3 clinical trial with four parallel arms compared changes in KOOS pain scores 12 months after injection, but no significant difference was found between stem cell and control groups (corticosteroid) [30]. Two other studies conducted by Bastos et al., the full version of the KOOS was applied to investigate several functional domains, such as quality of life [19, 20]. In one study, the total KOOS score in both MSC and MSC + PRP groups showed significant improvement at 12 months [20]. Also, in other study, all three treatment groups (MSC, MSC + PRP, and corticosteroids) experienced significant improvement in most KOOS subscales from baseline to 12 months, with the difference that the improvement in the stem cell groups was greater in terms of intensity and durability [19]. In the study by Ho et al., the standard SF-36 instrument was used to assess general quality of life at four time points (3, 6, 9, and 12 months after the intervention). The results showed that in the BM-MSC group, significant improvements were observed in several subscales, including physical function, bodily pain, and physical component, while in the control group (hyaluronic acid), no such improvements were reported. These findings indicate a positive effect of cell therapy on the quality of life of patients with knee osteoarthritis during 1 year of follow-up [26]. According to the study results, using MSCs, whether autologous or allogeneic, poses no serious complications from a safety perspective. A mild and transient local reaction such as inflammation or pain was reported, which resolved spontaneously after a short period of time. The safety of MSCs has been confirmed, and it is now possible to use them more widely in patients who are susceptible and refractory to other treatments. Likewise, in our meta-analysis, we noted that MSC injection did not have a standard dose. A total of 6 to 150 million cells were used in the reviewed studies, and in many cases, platelet-rich plasma (PRP) or hyaluronic acid was added. The optimal dose for improvement of pain and other symptoms in osteoarthritis was achieved by less than 50 × 106 cells (≤ 25 × 106 or 25 < cells ≤ 50 × 106). It seems that using high doses of stem cells may not result in better results and even week improvement compared to lower doses. A review and meta-analysis by Huang et al. [38], provided robust evidence on the efficacy and safety of adipose tissue derived-MSC therapy for osteoarthritis, particularly knee osteoarthritis. This study, by carefully reviewing 13 randomized clinical trials and analyzing data from 11 studies, provides a more detailed picture of the improvement in patients’ symptoms over time after stem cell injection. According to the findings of this study, the efficacy of MSCs is mainly associated with a time lag; in short-term assessments (1 to 3 months after injection), no significant difference was observed compared to the control group, but in longer time periods (6 and 12 months), significant clinical improvements in pain, joint stiffness, and motor function were reported. In terms of cell dose, this study identified doses below 25 million and between 25 and 50 million cells as optimal and effective doses for pain reduction (based on VAS) and functional improvement (based on WOMAC). Interestingly, doses above 50 million cells showed an initial improvement in pain in the first month, but this improvement did not persist, and ultimately the intermediate and lower doses had more sustained clinical efficacy. Also, joint stiffness and daily function symptoms improved significantly with doses of 25 million cells or less at 12 months post-injection. This pattern of results is consistent with the findings of the present study. Specifically, the meta-analysis results showed that high doses (>50 million cells), although they may have better or faster results in the short term, also resulted in more side effects and did not provide more sustained efficacy than intermediate or lower doses. On the other hand, optimal and balanced doses (less than 50 million cells) provide a good balance between efficacy and safety and significantly improve patients’ pain and function over a period of 12 months. It is important to consider the risk of side effects in addition to symptom improvement when choosing a therapeutic dose to achieve the best clinical outcome with the least risk. It should also be kept in mind that several factors such as disease severity, individual immune response, and the concomitant use of complementary therapies (such as platelet-rich plasma or hyaluronic acid) can influence the therapeutic response. The effectiveness of MSCs combined with other therapeutic modalities needs to be evaluated in future clinical trials focusing on dose-response and comparing different cell sources (bone marrow vs. umbilical cord). Moreover, the findings of this study demonstrated that treatment with MSCs for knee osteoarthritis could significantly improve patients’ pain, stiffness, and function over a 1-year period. Currently, conventional treatments mainly aim to relieve symptomatic relief and do not have regenerative properties. MSCs may play a crucial role in therapeutic strategies in the future due to their emerging and promising nature [39]. Despite the significant findings of this study, some limitations should be considered in interpreting the results. First, there was high heterogeneity between the included studies in some outcomes—especially at intermediate time points such as 6 months—which may be due to differences in the dose of stem cells injected, the source of cells (autologous or allogeneic), the concomitant use of complementary interventions such as PRP or hyaluronic acid, and variation in initial disease severity (based on the Kellgren-Lawrence classification). Second, some studies did not provide sufficient information on the method of randomization, blinding of patients and assessors, or allocation concealment, which could introduce bias and reduce confidence in the results to some extent. Third, in many studies, the sample size was relatively small and, in some cases, patients were followed up for only 6 or 12 months, whereas osteoarthritis is a chronic disease and it is important to examine the long-term effectiveness of cellular interventions. A key limitation is the insufficient data within individual dose subgroups. The preponderance of subgroups containing only one to three studies significantly undermines the reliability and generalizability of the dose-specific findings. In order to address these limitations, MSCs need to be assessed over a longer period of time to assess their long-term effects, especially on joint tissue structure (by using MRIs or arthroscopes). Future studies should have multicenter designs, larger sample sizes, and longer follow-up periods (at least 24 months). The analysis of factors predicting response to treatment (biomarker-based stratification) can also assist in optimizing treatment and optimizing outcomes, including comparing different cell doses, examining response to treatment in different subgroups (e.g., age, gender, disease severity), and comparing response to treatment across subgroups. The future of MSC treatments may benefit from studies that combine MSCs with other novel approaches, such as inflammation inhibitors, bioscaffolds, or gene therapy. Conclusion The present study, using a systematic review and meta-analysis of randomized clinical trials, provides acceptable evidence regarding the efficacy based on dose-response and safety of bone marrow-derived in the treatment of osteoarthritis. MSCs can significantly improve pain, reduce stiffness in joints, and improve motor function in patients who receive intra-articular injections, especially for longer periods (6 to 12 months), although a significant effect is not observed for short-term injections. The optimal dose for improvement of pain and other symptoms was achieved by less than 50 × 106 cells (≤ 25 × 106 or 25 < cells ≤ 50 × 106). These results are of great importance given the limitations of current therapeutic methods, which often only have a palliative role, and introduce MSCs as a promising therapeutic option with the potential to regenerate cartilage tissue. On the other hand, the results of this study also confirm the favorable safety of stem cells; no serious complications were reported, and the occurrence of mild and transient complications was limited and not worrisome. It is important to interpret the results of these studies with caution, however, since they differ in methods, doses, and patient characteristics. For these cells to be effective at optimal doses, structurally efficacious, and for long-term clinical use, prospective studies involving careful design, large sample sizes, and long-term follow-up are still necessary. It was concluded that this study is a critical step towards developing and applying cell-based regenerative therapies for osteoarthritis patients, paving the way for individualized, safe, and effective therapies. Search strategy Pubmed (‘osteoarthritis’/de OR ‘arthritis, degenerative’:ti, ab, kw OR ‘arthritis, noninflammatory’:ti, ab, kw OR ‘arthrosis’:ti, ab, kw OR ‘degenerative arthritis’:ti, ab, kw OR ‘degenerative joint disease’:ti, ab, kw OR ‘noninflammatory arthritis’:ti, ab, kw OR ‘osteo-arthritis’:ti, ab, kw OR ‘osteo-arthrosis’:ti, ab, kw OR ‘osteoarthritis’:ti, ab, kw OR ‘osteoarthrosis’:ti, ab, kw OR ‘primary osteoarthritis’:ti, ab, kw OR ‘rheumatoid arthrosis’:ti, ab, kw) AND (‘mesenchymal stem cell’/de OR ‘mesenchymal progenitor cell’:ti, ab, kw OR ‘mesenchymal stem cell’:ti, ab, kw OR ‘mesenchymal stem cells’:ti, ab, kw OR ‘stem cell, mesenchymal’:ti, ab, kw OR ‘bone marrow cell’/de OR ‘bone marrow cell’:ti, ab, kw OR ‘bone marrow cells’:ti, ab, kw OR ‘bone marrow derived cell’:ti, ab, kw OR ‘bone stem cell’:ti, ab, kw OR ‘cell, bone marrow’:ti, ab, kw OR ‘marrow stem cell’:ti, ab, kw OR ‘myeloid cell’:ti, ab, kw OR ‘myeloid cells’:ti, ab, kw) AND (‘clinical trial’/de OR ‘clinical drug trial’:ti, ab, kw OR ‘clinical trial’:ti, ab, kw OR ‘major clinical trial’:ti, ab, kw OR ‘trial, clinical’:ti, ab, kw OR ‘controlled clinical trial’/de OR ‘clinical trial, controlled’:ti, ab, kw OR ‘controlled clinical comparison’:ti, ab, kw OR ‘controlled clinical drug trial’:ti, ab, kw OR ‘controlled clinical experiment’:ti, ab, kw OR ‘controlled clinical study’:ti, ab, kw OR ‘controlled clinical test’:ti, ab, kw OR ‘controlled clinical trial’:ti, ab, kw OR rct OR ‘randomized controlled trial’/de OR ‘controlled trial, randomized’:ti, ab, kw OR ‘randomised controlled study’:ti, ab, kw OR ‘randomised controlled trial’:ti, ab, kw OR ‘randomized controlled study’:ti, ab, kw OR ‘randomized controlled trial’:ti, ab, kw OR ‘trial, randomized controlled’:ti, ab, kw) Embase Search: (((((“Mesenchymal Stem Cells“[Mesh]) OR “Umbilical Cord“[Mesh]) OR “Bone Marrow Cells“[Mesh]) OR “Myeloid Cells“[Mesh]) AND (“Osteoarthritis“[Mesh])) OR ((“mesenchymal stem cell“[Text Word] OR “mesenchymal progenitor cell“[Text Word] OR “mesenchymal stem cell“[Text Word] OR “mesenchymal stem cells“[Text Word] OR “stem cell, mesenchymal“[Text Word] OR “bone marrow cell“[Text Word] OR “bone marrow cell“[Text Word] OR “bone marrow cells“[Text Word] OR “bone marrow derived cell“[Text Word] OR “bone stem cell“[Text Word] OR “cell, bone marrow“[Text Word] OR “marrow stem cell“[Text Word] OR “myeloid cell“[Text Word] OR “myeloid cells“[Text Word]) AND (“osteoarthritis“[Text Word] OR “arthritis, degenerative“[Text Word] OR “arthritis, noninflammatory“[Text Word] OR “arthrosis“[Text Word] OR “degenerative arthritis“[Text Word] OR “degenerative joint disease“[Text Word] OR “noninflammatory arthritis“[Text Word] OR “osteo-arthritis“[Text Word] OR “osteo-arthrosis“[Text Word] OR “osteoarthritis“[Text Word] OR “osteoarthrosis“[Text Word] OR “primary osteoarthritis“[Text Word] OR “rheumatoid arthrosis“[Text Word])) Filters: Clinical Trial, Meta-Analysis, Randomized Controlled Trial, Systematic Review Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1. Acknowledgements The researchers hereby appreciate the assistance and financial support provided by the Research Department of Tabriz University of Medical Sciences. Author contributions Conceptualization: V.T., and A.F.K.; Methodology: V.T., E.L., and A.F.Kh.; Software: A.F.Kh.; Validation: V.T., E.L., and A.F.Kh.; Formal analysis: A.F.Kh.; Investigation: A.F.K., S.B., H.H., and S.P.; Data Curation: A.F.K., V.T., S.B., H.H., and S.P.; Writing-Original Draft: S.B., H.H., and S.P.; Writing-Review and Editing: V.T., E.L., and A.F.K.; Supervision: V.T., and A.F.K. All authors approved the final version of manuscript. Funding The Vice-Chancellor for Research and Technology, Tabriz University of Medical Sciences has funded the original research (Grant No: 74160). Data availability All data generated or analyzed during this study are included in this published article. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Vahideh Toopchizadeh, Email: toopchi.v@gmail.com. Azizeh Farshbaf-Khalili, Email: farshbafa@tbzmed.ac.ir. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement All data generated or analyzed during this study are included in this published article.