Mesenchymal stem cell-derived extracellular vesicles for osteochondral defect regeneration: a systematic review and meta-analysis of preclinical studies Authors: Yixin Wen, Guangran Meng, Shengtao Yang, Hongjun Mei Published: 2026 ABSTRACT BackgroundIn recent years, the number of animal studies investigating mesenchymal stem cell-derived extracellular vesicles (MSC-EVs) therapy for osteochondral defects has steadily increased. However, a comprehensive and systematic quantitative analysis is lacking. This meta-analysis aimed to evaluate the efficacy of MSC-EVs in animal models of osteochondral defect through a meta-analysis.MethodsPubMed, EMBASE, Web of Science, and Cochrane Library databases were searched up to January 31, 2025, to identify and select preclinical studies on MSC-EVs therapy for osteochondral defect models. No restrictions were applied regarding the specific characteristics of MSC-EVs or animal models. A random-effects model was used to perform pooled analyses of osteochondral repair scores, bone structural parameters, and histological scores. Subgroup analyses were conducted to explore potential sources of heterogeneity. Sensitivity analyses, funnel plots, and Egger’s test were employed to assess the robustness and stability of the results.ResultsA total of 19 studies met the inclusion criteria. MSC-EVs intervention improved International Cartilage Repair Society (ICRS) and Modified O’Driscoll Scale (MODS) scores at both 3 and 6 months, and significantly increased BV/TV, Tb.Th, as well as histological scores of cartilage and subchondral bone. Subgroup analyses indicated that none of the examined subgroups were significant sources of heterogeneity. Sensitivity analyses demonstrated the robustness of the results.ConclusionsMSC-EVs improved osteochondral repair scores, bone structural parameters, and histological outcomes in preclinical models, indicating their potential for promoting osteochondral regeneration. However, the evidence is primarily derived from small animal models with methodological variability and publication bias. Future studies should focus on large-animal models, standardized EVs preparation and dosing protocols, and long-term safety evaluations to facilitate clinical translation.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12967-025-07427-w. Introduction The self-healing capacity of articular cartilage is limited following cartilage defects, which often serves as a key contributing factor to degenerative joint diseases such as osteoarthritis [1, 2]. Globally, osteoarthritis (OA) affects approximately 250 million people [3], and over 60% of patients undergoing arthroscopic surgery present with articular cartilage injuries [4]. Anatomically, the avascular nature of cartilage tissue severely restricts its intrinsic regenerative potential [5]. When damage extends to the subchondral bone, which is rich in vascular networks, osteochondral defects are commonly formed, accelerating progression toward irreversible joint degeneration and impaired mobility [6, 7]. Current clinical interventions primarily include microfracture surgery and autologous chondrocyte implantation, which can improve joint function in the short term [8]. However, long-term follow-up studies reveal that the regenerated tissue often consists of mechanically inferior fibrocartilage and shows limited efficacy in repairing deep bone defects [9, 10]. Although tissue engineering approaches incorporating bioceramics, metal composites, and other scaffold materials have achieved promising therapeutic progress [11], single-component scaffolds face significant challenges in replicating the complex microenvironment of the osteochondral interface and in coordinating the bidirectional regeneration of cartilage and bone [12]. Mesenchymal stem cells (MSCs), owing to their multilineage differentiation potential and immunomodulatory properties, have emerged as a central therapeutic strategy in the field of osteochondral regeneration [13]. However, challenges such as limited post-transplantation cell survival [14], uncontrollable in vivo differentiation trajectories [15], and potential tumorigenic risks [16] have shifted research focus toward their paracrine mediators-extracellular vesicles (EVs). EVs, naturally occurring nanoparticles ranging from 10 nm to several micrometers in diameter, facilitate intercellular communication by delivering bioactive components such as proteins and microRNAs [17]. MSC-derived EVs retain the chondrogenic and anti-inflammatory properties of their parent cells [18], while offering advantages including low immunogenicity and absence of tumorigenic risks [19]. Compared to conventional cell-based therapies, EV-based cell-free strategies demonstrate unique potential in terms of storage stability and clinical translational feasibility [20]. A previous study have shown that MSC-EVs play a role in regulating chondrocyte function and maintaining cartilage matrix homeostasis, thereby mitigating the progression of OA [21]. Notably, in a comparative animal study conducted by Warmink et al. [22], MSC-EVs were found to outperform MSCs in slowing cartilage degradation, reducing osteophyte formation, and alleviating joint pain. Interestingly, the study also revealed potential adverse effects of MSC transplantation on OA models. In a systematic review published in 2020, To et al. [23] found that MSC-EVs reduced cartilage degradation in murine models of cartilage injury. However, due to the limited number of available studies, their review did not include a meta-analysis specifically evaluating the efficacy of MSC-EVs in osteochondral defect repair. Given the current lack of relevant clinical research, preclinical animal models remain a crucial source of evidence for the use of MSC-EVs in treating osteochondral injuries. Therefore, this meta-analysis aims to comprehensively assess the reparative effects of MSC-EVs on osteochondral defects in animal models, providing up-to-date evidence to support the clinical translation and application of MSC-EVs (Fig. 1). Fig. 1Potential mechanisms of mesenchymal stem cell-derived extracellular vesicles (MSC-EVs) from different sources in treating various cartilage defect animal models Potential mechanisms of mesenchymal stem cell-derived extracellular vesicles (MSC-EVs) from different sources in treating various cartilage defect animal models Materials and methods Systematic review The systematic review and meta-analysis were reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol statement [24]. The study was registered in the International Prospective Register of Systematic Reviews (PROSPERO 2025, https://www.crd.york.ac.uk/prospero/, CRD420251087284). Literature search strategy A systematic search was conducted in four electronic databases (PubMed, Embase, and Web of Science) for research articles published in English. Cochrane Library was additionally searched to capture any relevant systematic reviews and references, ensuring the comprehensiveness of the search. The search strategy combined MeSH/Emtree terms and free-text keywords using Boolean operators. Specifically, the following terms were included: (1) mesenchymal stem cells (“mesenchymal stromal cell” OR “mesenchymal stem cell” OR “MSC” OR “multipotent stromal cell”), (2) extracellular vesicles (“exovesicle” OR “exosome” OR “microvesicle” OR “apoptotic bodies” OR “EVs”), and (3) osteochondral defects (“osteochondral regeneration” OR “osteochondral lesion” OR “cartilage injury” OR “subchondral bone defect”). A manual search of the references in the included articles was also performed to avoid missing relevant studies. Two independent researchers conducted the search process, and discrepancies were resolved through discussion with a third researcher. We followed the PICOS framework: (1) Population: animal models of osteochondral defects (regardless of species or method of defect induction); (2) Intervention: local or systemic administration of MSC-EVs; (3) Comparison: blank control, vehicle control, or positive control; (4) Outcomes: ICRS scores, histological outcomes, and micro-CT parameters; (5) Study design: randomized or non-randomized in vivo studies. The detailed search strategy is provided in Supplementary Table 1. Eligibility criteria Inclusion criteria: (1) Preclinical animal models of osteochondral defects (no restrictions on species or methods of defect induction); (2) Local or systemic intervention using MSC-derived extracellular vesicles (MSC-EVs), regardless of dosage, administration route, or frequency. Studies using hydrogel scaffolds without EVs loading as carriers for defect filling were also included; (3) Included studies should have a control group (untreated osteochondral defect or PBS/placebo group; 4) Studies must report quantitative data for at least one predetermined outcome parameter (ICRS scores, histological indices, micro-CT bone measurements, or type II collagen deposition levels; 5) Randomized or non-randomized controlled in vivo trials. Exclusion criteria: (1) Studies using EVs derived from non-MSC sources (e.g., platelet-derived exosomes); (2) Non-original research (such as reviews, meta-analyses, or commentaries); (3) Studies without a controlled design; (4) Studies with incomplete or inaccessible outcome data; (5) Studies not published in English. Study selection All retrieved records were managed and de-duplicated using EndNote X20. Two researchers independently conducted the study selection in two consecutive stages: (1) Title and abstract screening: studies that clearly did not meet the criteria for animal models or interventions were excluded; (2) Full-text assessment: full-text articles were evaluated based on the predefined eligibility criteria, and studies that did not meet the inclusion criteria were further excluded. Any discrepancies in the selection process were resolved through discussion with a third researcher until consensus was reached. Data extraction Two authors independently extracted relevant data from the included studies using Excel spreadsheets. The extracted raw data were then cross-verified, and any discrepancies were resolved through discussion with a third researcher. Specifically, the following data were collected: (1) First author, study location, and year of publication; (2) Animal model characteristics: animal species, sex, sample size, age in weeks, and body weight; (3) MSC-EVs characteristics: source of EVs, diameter, engineering methods, route of administration, frequency, dosage, and duration of intervention. For studies that did not report raw numerical data, graphical data were extracted using Origin software (version 2021) and expressed as mean ± standard deviation (mean ± SD). For the data extracted from figures, we attempted to contact the corresponding authors via email to obtain further details. Outcome measures To evaluate the reparative effects of MSC-EVs on osteochondral tissue, primary outcome measures were stratified by 3-month and 6-month time points, including the International Cartilage Repair Society (ICRS) scoring system and the Modified O’Driscoll Scale (MODS). Secondary outcomes primarily included micro-CT assessments of bone volume fraction (BV/TV), trabecular number (Tb.N), and trabecular thickness (Tb.Th), as well as histological scores of cartilage and subchondral bone based on histological staining of osteochondral tissue. Additionally, the relative fluorescence intensity and mean integrated optical density (mean IOD/µm²) of type II collagen were evaluated. Quality assessment The methodological quality of the included studies was evaluated using the SYRCLE’s Risk of Bias (RoB) tool [25] for animal studies. The assessed domains included selection bias, performance bias, detection bias, attrition bias, reporting bias, and other potential sources of bias. Two researchers independently conducted the quality assessment, and any discrepancies were resolved through discussion with a third researcher until agreement was reached. The risk of bias for each item was categorized as low risk, high risk, or unclear risk (due to lack of description or insufficient methodological details). Statistical analysis A random-effects model was applied to calculate effect sizes, expressed as standardized mean differences (SMDs) with 95% confidence intervals. Heterogeneity among studies was assessed using the I² statistic, with values of < 25% indicating low heterogeneity, 25–50% indicating moderate heterogeneity, and > 50% indicating substantial heterogeneity. For pooled results with I² >50% and more than 10 included studies, subgroup analyses were performed based on animal and EVs characteristics to explore potential sources of heterogeneity. Sensitivity analyses were conducted to assess the robustness of the results (N ≥ 5), including ICRS score at 3 and 6 months, BV/TV at 3 and 6 months, Tb.Th at 3 months, histomorphology score of cartilage at 3 and 6 months, and relative staining intensity of II collagen protein at 3 months. Additionally, publication bias was evaluated using funnel plots and Egger’s test. A P-value < 0.05 was considered statistically significant. Meta-analysis and sensitivity analysis were performed using RevMan 5.3 (The Cochrane Collaboration) and Stata SE 15.0 software (StataCorp, College Station, TX, USA), respectively. Results Literature search and selection The study selection process is illustrated in Fig. 2. An initial total of 389 potentially relevant articles were identified through database searches. After removing duplicates using EndNote X20, 282 articles remained. Following title and abstract screening, 32 articles were selected for full-text review. Studies were excluded for the following reasons: in vitro studies only (N = 6), non-osteochondral defect models (N = 3), lack of primary outcome measures (N = 2), and incomplete data (N = 2). Ultimately, 19 studies [26–44]were included in the subsequent meta-analysis and systematic review. Fig. 2Flow diagram of the literature selection process for studies on MSC-EVs in the treatment of osteochondral defect animal models Flow diagram of the literature selection process for studies on MSC-EVs in the treatment of osteochondral defect animal models Study characteristics The characteristics of the animal models included in the studies are summarized in Table 1. The publication years of the included studies ranged from 2016 to 2024, with 84.2% (16 out of 19) published in or after 2021, highlighting the growing interest in the potential of MSC-EVs for treating osteochondral defects over the past five years. Except for three studies conducted in Singapore, all others were carried out in China. A total of 202 animals were included in the meta-analysis. Twelve studies used Sprague-Dawley (SD) rats, while six studies and one study used New Zealand White rabbits and micropigs, respectively. Regarding animal sex, eight studies utilized male animals, seven used female animals, and four did not report sex details. The age of the animals ranged from 6 weeks to 6 months. The body weight of the rats ranged from 200 g to 3.5 kg. Three studies respectively reported the average body weights of New Zealand White rabbits (2.5 kg), SD rats (250 g), and micropigs (16.23 kg). Table 1Main characteristics of animals included in the meta-analysisAuthorYearCountryOsteochondral defects modelSpecieGenderAgeWeightNumberChen 2019 [26]2019ChinaNew Zealand white rabbitsFemale6-week-oldNA24Chen 2024 [27]2024ChinaNew Zealand white rabbitsNANANA60Ding 2024 [28]2024ChinaSD ratsNANANA24Hu 2020 [29]2020ChinaSD ratsFemale8-week-oldNA28Jiang 2021 [30]2021ChinaNew Zealand white rabbitsNANANA50Li 2023 [31]2023ChinaSD ratsMale12-week-old400–450 g20Li 2024 [32]2024ChinaSD ratsMale12-week-oldNA40Liu 2024 [33]2024ChinaNew Zealand white rabbitsFemale4-6-month-old2.5 kg16Ma 2023 [34]2023ChinaNew Zealand white rabbitsFemaleNA3.0–3.5 kg15Shen 2022 [35]2022ChinaSD ratsMale8-week-oldNA24Sun 2022 [36]2022ChinaSD ratsNANANA54Tian 2024 [37]2024ChinaSD ratsMale8-week-oldNA20Xu 2024 [38]2024ChinaSD ratsMale8-week-oldNA28Yan 2021 [39]2021ChinaSD ratsFemaleNA200 g40Yan 2023 [40]2023ChinaSD ratsMale6-week-old250 ± 50 g30Yang 2022 [41]2022ChinaNew Zealand white rabbitsMale6-week-oldNA24Yang 2023 [42]2023SingaporeSD ratsMale12-week-old300–400 g48Zhang 2016 [43]2016SingaporeSD ratsFemale8-week-oldNA12Zhang 2022 [44]2022SingaporeMicropigsFemaleNA16.23 ± 1.41 kg12NA: Not applicable. SD: Sprague Dawley Table 1Main characteristics of animals included in the meta-analysisAuthorYearCountryOsteochondral defects modelSpecieGenderAgeWeightNumberChen 2019 [26]2019ChinaNew Zealand white rabbitsFemale6-week-oldNA24Chen 2024 [27]2024ChinaNew Zealand white rabbitsNANANA60Ding 2024 [28]2024ChinaSD ratsNANANA24Hu 2020 [29]2020ChinaSD ratsFemale8-week-oldNA28Jiang 2021 [30]2021ChinaNew Zealand white rabbitsNANANA50Li 2023 [31]2023ChinaSD ratsMale12-week-old400–450 g20Li 2024 [32]2024ChinaSD ratsMale12-week-oldNA40Liu 2024 [33]2024ChinaNew Zealand white rabbitsFemale4-6-month-old2.5 kg16Ma 2023 [34]2023ChinaNew Zealand white rabbitsFemaleNA3.0–3.5 kg15Shen 2022 [35]2022ChinaSD ratsMale8-week-oldNA24Sun 2022 [36]2022ChinaSD ratsNANANA54Tian 2024 [37]2024ChinaSD ratsMale8-week-oldNA20Xu 2024 [38]2024ChinaSD ratsMale8-week-oldNA28Yan 2021 [39]2021ChinaSD ratsFemaleNA200 g40Yan 2023 [40]2023ChinaSD ratsMale6-week-old250 ± 50 g30Yang 2022 [41]2022ChinaNew Zealand white rabbitsMale6-week-oldNA24Yang 2023 [42]2023SingaporeSD ratsMale12-week-old300–400 g48Zhang 2016 [43]2016SingaporeSD ratsFemale8-week-oldNA12Zhang 2022 [44]2022SingaporeMicropigsFemaleNA16.23 ± 1.41 kg12NA: Not applicable. SD: Sprague Dawley Main characteristics of animals included in the meta-analysis NA: Not applicable. SD: Sprague Dawley Regarding the characteristics of MSC-EVs, eight studies utilized EVs derived from bone marrow mesenchymal stem cells (BMSCs), three from umbilical cord mesenchymal stem cells (UC-MSCs), two from human Wharton’s jelly mesenchymal stem cells (hWJ-MSCs), three from adipose-derived mesenchymal stem cells (AMSCs), and two from human embryonic mesenchymal stem cells (hESCs). The reported EV particle sizes varied, with most studies indicating a range between 30 and 350 nm, except for one study that reported a size of 1245 nm. In the preparation of MSC-EVs, ultracentrifugation was employed in 15 studies, while two studies utilized tangential flow filtration (TFF). The remaining two studies used an isolation kit and size-exclusion chromatography, respectively. Regarding purification methods, except for one study that did not report purification details, nine studies used sterile membrane filtration, and eight studies employed PBS washing of the precipitate. In terms of intervention details, the majority of studies (N = 12) administered MSC-EVs via intra-articular injection. The remaining seven studies used hydrogel scaffolds to deliver EVs directly to the defect site, including cartilage extracellular matrix (ECM)/gelatin methacrylate (GelMA), sodium alginate/hyaluronic acid, and silk fibroin (SF) hydrogels. Intervention frequency varied across studies, including single administration, three times, once weekly, once every five days, and three times per week. Except for nine studies that administered a single injection and one study with three injections, the treatment duration in all other studies lasted at least one week. The details of EV characteristics and intervention protocols are summarized in Table 2. Table 2Characteristics of EVs and therapeutic methodsAuthorCharacteristics of EVsTherapeutic methodsSourceCell cultureIsolationPurificationDiameter (nm)TEMParticle concentrationMarkerRouteDoseFrequencyDurationChen 2019 [26]BMSCscultivate to 50–60%ultracentrifugationfiltered through a 0.22 μm filter40–110hollow spherical microvesiclesNATSG101, CD9, CD63cartilage extracellular matrix (ECM)/gelatin methacrylate (GelMA)/exosome scaffold200 µg/mLonceonceChen 2024 [27]hWJ-MSCscultivate to P3-P5Ribo™ Exosome Isolation Reagentthe pellet was washed with PBS60–350sphere-like morphologyNACD9, CD63, TSG101, TSP70knee joint cavity50 µg/mLthree times a week1 weeksDing 2024 [28]AMSCscultured under hypoxic conditionsultracentrifugationthe pellet was washed with PBS298.2spherical vesicle structureNATSG101, CD63, caspase-3knee joint cavity1 µg/mLonce every 5 days2 weeksHu 2020 [29]hUC-MSCsNAultracentrifugationthe pellet was washed with PBS50–150spherical microvesicle structure8.2 × 106 particles/mLCD81, CD63, CD9knee joint cavity10 × 108 particles/mLonceonceJiang 2021 [30]hWJ-MSCscultivate to P3-P5ultracentrifugationthe pellet was washed with PBS135shape of a discNACD9, TSG101, CD63knee joint cavity25 µg/mLonce a week5 weeksLi 2023 [31]hAMSCscultivate to P3-P4ultracentrifugationfiltered through a 0.22 μm sterile filter membrane140.3saucer-shaped vesicular morphologyNACD81, ALIX, TSG101exosome-reinforced hydrogel scaffoldNAonceonceLi 2024 [32]BMSCsNAultracentrifugationfiltered through a 0.22 μm filter63.58membranous vesicle structureNACD63, TSG101sodium alginate/ hyaluronic acid hydrogel scaffold125 µg/kgonceonceLiu 2024 [33]BMSCscultivate to P3-P8ultracentrifugationthe pellet was washed with PBS126.1spherical vesicle structureNAHSP60, TSG101, CD81bioenergetic-active exosomes scaffold100 µg/mLonceonceMa 2023 [34]BMSCsNAultracentrifugationNA141cup-shaped morphologyNACD81, TSG and Calnexinhyaluronic acid (HA)-based hydrogel1 × 1010/mLonceonceShen 2022 [35]BMSCscultivate to 80%ultracentrifugationfiltered through a 0.22 μm filter122round shapeNACD9, CD63, TSG101 and CalnexinSF hydrogel scaffold1 × 109 particles/mLonceonceSun 2022 [36]BMSCscultivate to P3 + TGFβ3 (50 ng/mL)tangential flow filtration (TFF)filtered through a 0.22 μm sterile filter membraneNAcup-like shape with double-layer membrane structuresNACD9, CD63, CD81, Alixexosome-reinforced hydrogel scaffold10 × 109 particles/mLonceonceTian 2024 [37]hUC-MSCsstaurosporine inducing apoptosisultracentrifugationthe pellet was washed with PBS1245round shapeNAhistone H3, histone H2B, complement C3b, and complement C1QCknee joint cavity1 µg/mLthree times a week1 weeksXu 2024 [38]AMSCscultivate to P3-P5ultracentrifugationfiltered through a 0.8 μm filter146.9 ± 13.3round shape7.5 × 109 particles/mLCD9, CD63, Tsg101, Hsp70, Alixknee joint cavity7.5 × 109 particles/mLonceonceYan 2021 [39]UC-MSCsNAultracentrifugationfiltered through a 0.22 μm filter120cup-shaped morphologyNACD63, CD81, TSG101knee joint cavity1 mg/mLonce a week4 weeksYan 2023 [40]hUC-MSCscultivate to P3-P5ultracentrifugationthe pellet was washed with PBS30–150bilayer biofilm structureNACD9, CD63, TSG101knee joint cavity25 µg/mLonce every 5 days3 daysYang 2022 [41]BMSCscultivate to P3-P4ultracentrifugationfiltered through a 0.22 μm sterile filter membrane131.2cub-like shape coated with bilayer1.2 × 1011 particles/mLCD81, TSG101, Flotillin 1knee joint cavity300 µL of 5 × 1010 particles/mLonce a week4 weeksYang 2023 [42]BMSCscultivate to P4size-exclusion chromatographythe pellet was washed with PBSNAspherical vesicle structureNACD9, CD63knee joint cavity100 µLonce a week6 weeksZhang 2016 [43]hESCscultivate hESC line H9ultracentrifugationNA100NANACD81, ALIX, TSG101knee joint cavity100 µg exosomes per 100 µL injectiononce a week6 weeksZhang 2022 [44]hESCscultivate to 80%tangential flow filtration (TFF)filtered through a 0.22 μm sterile filter membrane147.4bilipid membrane vesicles2.16 × 1011 particles/mLCD81, ALIX, TSG101knee joint cavity2.16 × 1011 particles/mLthree times2 weeksBMSC: Bone marrow mesenchymal stem cells; ECM: Cartilage extracellular matrix; GelMA: Gelatin methacrylate; hWJ-MSCs: Human Wharton’s jelly mesenchymal stem cell. hAMSCs: Human adipose-derived mesenchymal stem cells; hUC-MSCs: Human umbilical cord mesenchymal stem cells. hESCs: Human embryonic mesenchymal stem cells; NA: Not applicable; SF: Silk fibroin Table 2Characteristics of EVs and therapeutic methodsAuthorCharacteristics of EVsTherapeutic methodsSourceCell cultureIsolationPurificationDiameter (nm)TEMParticle concentrationMarkerRouteDoseFrequencyDurationChen 2019 [26]BMSCscultivate to 50–60%ultracentrifugationfiltered through a 0.22 μm filter40–110hollow spherical microvesiclesNATSG101, CD9, CD63cartilage extracellular matrix (ECM)/gelatin methacrylate (GelMA)/exosome scaffold200 µg/mLonceonceChen 2024 [27]hWJ-MSCscultivate to P3-P5Ribo™ Exosome Isolation Reagentthe pellet was washed with PBS60–350sphere-like morphologyNACD9, CD63, TSG101, TSP70knee joint cavity50 µg/mLthree times a week1 weeksDing 2024 [28]AMSCscultured under hypoxic conditionsultracentrifugationthe pellet was washed with PBS298.2spherical vesicle structureNATSG101, CD63, caspase-3knee joint cavity1 µg/mLonce every 5 days2 weeksHu 2020 [29]hUC-MSCsNAultracentrifugationthe pellet was washed with PBS50–150spherical microvesicle structure8.2 × 106 particles/mLCD81, CD63, CD9knee joint cavity10 × 108 particles/mLonceonceJiang 2021 [30]hWJ-MSCscultivate to P3-P5ultracentrifugationthe pellet was washed with PBS135shape of a discNACD9, TSG101, CD63knee joint cavity25 µg/mLonce a week5 weeksLi 2023 [31]hAMSCscultivate to P3-P4ultracentrifugationfiltered through a 0.22 μm sterile filter membrane140.3saucer-shaped vesicular morphologyNACD81, ALIX, TSG101exosome-reinforced hydrogel scaffoldNAonceonceLi 2024 [32]BMSCsNAultracentrifugationfiltered through a 0.22 μm filter63.58membranous vesicle structureNACD63, TSG101sodium alginate/ hyaluronic acid hydrogel scaffold125 µg/kgonceonceLiu 2024 [33]BMSCscultivate to P3-P8ultracentrifugationthe pellet was washed with PBS126.1spherical vesicle structureNAHSP60, TSG101, CD81bioenergetic-active exosomes scaffold100 µg/mLonceonceMa 2023 [34]BMSCsNAultracentrifugationNA141cup-shaped morphologyNACD81, TSG and Calnexinhyaluronic acid (HA)-based hydrogel1 × 1010/mLonceonceShen 2022 [35]BMSCscultivate to 80%ultracentrifugationfiltered through a 0.22 μm filter122round shapeNACD9, CD63, TSG101 and CalnexinSF hydrogel scaffold1 × 109 particles/mLonceonceSun 2022 [36]BMSCscultivate to P3 + TGFβ3 (50 ng/mL)tangential flow filtration (TFF)filtered through a 0.22 μm sterile filter membraneNAcup-like shape with double-layer membrane structuresNACD9, CD63, CD81, Alixexosome-reinforced hydrogel scaffold10 × 109 particles/mLonceonceTian 2024 [37]hUC-MSCsstaurosporine inducing apoptosisultracentrifugationthe pellet was washed with PBS1245round shapeNAhistone H3, histone H2B, complement C3b, and complement C1QCknee joint cavity1 µg/mLthree times a week1 weeksXu 2024 [38]AMSCscultivate to P3-P5ultracentrifugationfiltered through a 0.8 μm filter146.9 ± 13.3round shape7.5 × 109 particles/mLCD9, CD63, Tsg101, Hsp70, Alixknee joint cavity7.5 × 109 particles/mLonceonceYan 2021 [39]UC-MSCsNAultracentrifugationfiltered through a 0.22 μm filter120cup-shaped morphologyNACD63, CD81, TSG101knee joint cavity1 mg/mLonce a week4 weeksYan 2023 [40]hUC-MSCscultivate to P3-P5ultracentrifugationthe pellet was washed with PBS30–150bilayer biofilm structureNACD9, CD63, TSG101knee joint cavity25 µg/mLonce every 5 days3 daysYang 2022 [41]BMSCscultivate to P3-P4ultracentrifugationfiltered through a 0.22 μm sterile filter membrane131.2cub-like shape coated with bilayer1.2 × 1011 particles/mLCD81, TSG101, Flotillin 1knee joint cavity300 µL of 5 × 1010 particles/mLonce a week4 weeksYang 2023 [42]BMSCscultivate to P4size-exclusion chromatographythe pellet was washed with PBSNAspherical vesicle structureNACD9, CD63knee joint cavity100 µLonce a week6 weeksZhang 2016 [43]hESCscultivate hESC line H9ultracentrifugationNA100NANACD81, ALIX, TSG101knee joint cavity100 µg exosomes per 100 µL injectiononce a week6 weeksZhang 2022 [44]hESCscultivate to 80%tangential flow filtration (TFF)filtered through a 0.22 μm sterile filter membrane147.4bilipid membrane vesicles2.16 × 1011 particles/mLCD81, ALIX, TSG101knee joint cavity2.16 × 1011 particles/mLthree times2 weeksBMSC: Bone marrow mesenchymal stem cells; ECM: Cartilage extracellular matrix; GelMA: Gelatin methacrylate; hWJ-MSCs: Human Wharton’s jelly mesenchymal stem cell. hAMSCs: Human adipose-derived mesenchymal stem cells; hUC-MSCs: Human umbilical cord mesenchymal stem cells. hESCs: Human embryonic mesenchymal stem cells; NA: Not applicable; SF: Silk fibroin Characteristics of EVs and therapeutic methods BMSC: Bone marrow mesenchymal stem cells; ECM: Cartilage extracellular matrix; GelMA: Gelatin methacrylate; hWJ-MSCs: Human Wharton’s jelly mesenchymal stem cell. hAMSCs: Human adipose-derived mesenchymal stem cells; hUC-MSCs: Human umbilical cord mesenchymal stem cells. hESCs: Human embryonic mesenchymal stem cells; NA: Not applicable; SF: Silk fibroin Risk of bias assessment The risk of bias assessment for the 19 included studies is presented in Fig. 3A and B. All studies were rated as having a low risk of bias in terms of attrition bias, reporting bias, and other potential biases. However, none of the studies achieved a low risk of bias across all assessed domains. Specifically, all studies lacked detailed descriptions regarding random sequence generation, allocation concealment, random housing, blinding of caregivers and investigators, and blinding of outcome assessment, leading to an unclear risk of bias in these areas. Seven studies reported implementing blinding for outcome assessors. Additionally, four studies did not report key baseline information, such as animal sex, weight, and age, which was considered a high risk of bias in baseline characteristics. Overall, the majority of studies exhibited an unclear risk of bias, particularly in selection bias and blinding domains. Fig. 3Risk of bias assessment results for 19 studies based on SYRCLE’s ROB tool. (A) Risk of bias graph; (B) Risk of bias summary Risk of bias assessment results for 19 studies based on SYRCLE’s ROB tool. (A) Risk of bias graph; (B) Risk of bias summary Meta-analysis results MSC-EVs improved osteochondral repair scores at 3 and 6 months Seventeen studies reported the effects of MSC-EVs on osteochondral defect models at the 3-month time point. The pooled analysis showed that MSC-EVs intervention significantly increased ICRS scores (SMD = 3.13; 95% CI: 2.19 to 4.07; P < 0.0001) (Fig. 4A). Due to substantial heterogeneity (I² = 67%, P < 0.0001), subgroup analyses were conducted. However, animal species (New Zealand white rabbits, SD rats, and micropigs), animal ages (6-week-old, 8-week-old, and 12-week-old), EVs source (BMSCs, UC-MSCs, AMSCs, WJ-MSCs, and hESCs), EVs isolation methods (ultracentrifugation and TFF), EVs purification methods (washed with PBS and filtered through a filter), intervention routes (intra-articular injection and hydrogel scaffold), intervention frequencies (single dose, once weekly, three times a week, and once every 5 days), and intervention durations (single dose, one week, two weeks, and other time) were not identified as significant sources of heterogeneity (Supplementary Figs. 1–8). At the 6-month time point, MSC-EVs intervention also led to an improvement in ICRS scores (SMD = 4.06; 95% CI: 2.56 to 5.56; P < 0.0001) with notable heterogeneity (I² = 79%, P < 0.0001) (Fig. 4B). Subgroup analyses similarly indicated that none of the examined factors contributed significantly to the observed heterogeneity (Supplementary Figs. 9–16). Fig. 4Forest plots illustrating the effects of MSC-EVs on International Cartilage Repair Society (ICRS) scores in osteochondral defect models. (A) ICRS score at 3 months; (B) ICRS score at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Forest plots illustrating the effects of MSC-EVs on International Cartilage Repair Society (ICRS) scores in osteochondral defect models. (A) ICRS score at 3 months; (B) ICRS score at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Additionally, the Modified O’Driscoll Scale (MODS), another reliable scoring system for evaluating cartilage repair, was reported in two studies following MSC-EVs intervention. The pooled analysis indicated that MSC-EVs had no significant effect on MODS scores at the 3-month time point (SMD = 2.70; 95% CI: 1.52 to 3.87; P < 0.0001) (Fig. 5A). However, at the 6-month time point, MSC-EVs intervention resulted in a significant improvement in MODS scores (SMD = 4.27; 95% CI: 2.755 to 5.79; P < 0.0001) (Fig. 5B). Overall, these findings may suggest a positive effect of MSC-EVs on cartilage repair, as reflected in both ICRS and MODS scores. Nonetheless, due to the notable heterogeneity and the limited number of included studies, these results should be interpreted with caution. Fig. 5Forest plots demonstrating the effects of MSC-EVs on Modified O’Driscoll Scale (MODS) scores in osteochondral defect models. (A) MODS score at 3 months; (B) MODS score at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Forest plots demonstrating the effects of MSC-EVs on Modified O’Driscoll Scale (MODS) scores in osteochondral defect models. (A) MODS score at 3 months; (B) MODS score at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) MSC-EVs improved bone structural parameters and histological scores Changes in bone structural parameters observed via micro-CT serve as a key indicator for evaluating the efficacy of MSC-EVs. The pooled analysis using a random-effects model demonstrated that MSC-EVs significantly increased BV/TV at both 3 months (SMD = 2.60; 95% CI: 1.60 to 3.61; P < 0.0001) and 6 months (SMD = 1.67; 95% CI: 0.57 to 2.77; P = 0.003) (Fig. 6A and B). Regarding trabecular thickness (Tb.Th), MSC-EVs showed a promotive effect at 3 months (SMD = 1.94; 95% CI: 1.24 to 2.64; P < 0.0001) and 6 months (SMD = 1.16; 95% CI: 0.37 to 1.96; P = 0.004) (Fig. 7A and B). However, no increase in Tb.N was observed at the 3-month time point following MSC-EV intervention. (SMD = 1.20; 95% CI: -0.43 to 2.83; P = 0.15) (Fig. 7C). Fig. 6Forest plots showing the effects of MSC-EVs on BV/TV in osteochondral defect models. (A) BV/TV at 3 months; (B) BV/TV at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Forest plots showing the effects of MSC-EVs on BV/TV in osteochondral defect models. (A) BV/TV at 3 months; (B) BV/TV at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Fig. 7Forest plots showing the effects of MSC-EVs on bone structural parameters in osteochondral defect models. (A) Tb.Th at 3 months; (B) Tb.Th at 6 months; (C) Tb.N at 3 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Forest plots showing the effects of MSC-EVs on bone structural parameters in osteochondral defect models. (A) Tb.Th at 3 months; (B) Tb.Th at 6 months; (C) Tb.N at 3 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) In addition to improvements in bone structural parameters, the potential impact of MSC-EVs on tissue morphology was also assessed. Regarding osteochondral evaluation, MSC-EVs significantly improved histological scores at both 3 months (SMD = 3.76; 95% CI: 2.82 to 4.69; P < 0.0001) and 6 months (SMD = 4.34; 95% CI: 2.22 to 6.46; P < 0.0001) (Fig. 8A and B). For subchondral bone, MSC-EVs also resulted in increased histological scores at 3 months (SMD = 5.94; 95% CI: 3.60 to 8.29; P < 0.0001) and 6 months (SMD = 4.30; 95% CI: 2.49 to 6.11; P < 0.0001) (Fig. 8C and D). Type II collagen, a critical marker in osteochondral repair, was quantitatively assessed by immunohistochemical staining in seven studies. MSC-EVs significantly enhanced the relative staining intensity of type II collagen at 3 months (SMD = 6.79; 95% CI: 2.11 to 11.47; P = 0.004) but had no significant effect at the 6-month time point (Fig. 9A and B). Furthermore, no significant changes were observed in the mean integrated optical density (IOD) values of type II collagen at either time point (Fig. 9C and D). Overall, compared to control groups, MSC-EVs demonstrated a potential positive effect on both bone structural parameters and histological scores in animal models of osteochondral defects. Fig. 8Forest plots illustrating the effects of MSC-EVs on histomorphology scores in osteochondral defect models. (A) Histomorphology score of cartilage at 3 months; (B) Histomorphology score of cartilage at 6 months; (C) Histomorphology score of subchondral bone at 3 months; (D) Histomorphology score of subchondral bone at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Forest plots illustrating the effects of MSC-EVs on histomorphology scores in osteochondral defect models. (A) Histomorphology score of cartilage at 3 months; (B) Histomorphology score of cartilage at 6 months; (C) Histomorphology score of subchondral bone at 3 months; (D) Histomorphology score of subchondral bone at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Fig. 9Forest plots illustrating the effects of MSC-EVs on II collagen protein in osteochondral defect models. (A) Relative staining intensity at 3 months; (B) Relative staining intensity at 6 months; (C) Mean integrated optical density (mean IOD) at 3 months; (D) Mean IOD at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Forest plots illustrating the effects of MSC-EVs on II collagen protein in osteochondral defect models. (A) Relative staining intensity at 3 months; (B) Relative staining intensity at 6 months; (C) Mean integrated optical density (mean IOD) at 3 months; (D) Mean IOD at 6 months. Data are presented as standardized mean differences (SMD) with 95% confidence intervals (CI) Sensitivity analysis The robustness of the meta-analysis results was assessed through sensitivity analysis. Specifically, by sequentially excluding each individual study, the pooled results of the remaining studies remained consistent with the original findings. These results indicated that the pooled outcomes for ICRS score at 3 and 6 months, BV/TV at 3 and 6 months, Tb.Th at 3 months, histomorphology score of cartilage at 3 and 6 months, and relative staining intensity of type II collagen protein at 3 months demonstrated a certain degree of robustness (Fig. 10A and H). Fig. 10Sensitivity analyses demonstrating the robustness of different outcome measures. (A) ICRS score at 3 months; (B) ICRS score at 6 months; (C) BV/TV at 3 months; (D) BV/TV at 6 months; (E) Tb.Th at 3 months; (F) Histomorphology score of cartilage at 3 months; (G) Histomorphology score of cartilage at 6 months. (H) Relative staining intensity of II collagen protein at 3 months Sensitivity analyses demonstrating the robustness of different outcome measures. (A) ICRS score at 3 months; (B) ICRS score at 6 months; (C) BV/TV at 3 months; (D) BV/TV at 6 months; (E) Tb.Th at 3 months; (F) Histomorphology score of cartilage at 3 months; (G) Histomorphology score of cartilage at 6 months. (H) Relative staining intensity of II collagen protein at 3 months Publication bias Publication bias was further assessed using funnel plots and Egger’s test. The funnel plots for the 3-month and 6-month ICRS scores displayed asymmetry, and the results of Egger’s test confirmed the presence of publication bias (Fig. 11A and B). Additionally, BV/TV at 3 months, histomorphology score of cartilage at 3 and 6 months also showed evidence of publication bias (Table 3). Due to the limited number of studies reporting other outcome measures, funnel plots and Egger’s tests were not performed for those indicators. Fig. 11Funnel plots illustrating the presence of publication bias. (A) ICRS score at 3 months; (B) ICRS score at 6 months Funnel plots illustrating the presence of publication bias. (A) ICRS score at 3 months; (B) ICRS score at 6 months Table 3Egger’s test for publication biasOutcomeEgger’s test (p value)t valuePooling modelICRS scores at 3 months0.0007.68RandomICRS scores at 6 months0.0008.44RandomBV/TV at 3 months0.0016.07RandomHistomorphology score of cartilage at 3 months0.0233.24RandomHistomorphology score of cartilage at 6 months0.0016.35Random Table 3Egger’s test for publication biasOutcomeEgger’s test (p value)t valuePooling modelICRS scores at 3 months0.0007.68RandomICRS scores at 6 months0.0008.44RandomBV/TV at 3 months0.0016.07RandomHistomorphology score of cartilage at 3 months0.0233.24RandomHistomorphology score of cartilage at 6 months0.0016.35Random Egger’s test for publication bias Discussion Currently, the rapid development of regenerative cell therapies has ushered in a new era for the treatment of osteochondral injuries, though definitive conclusions regarding their efficacy and safety are still lacking [45]. Due to their low cost, low immunogenicity, and notable capacity to promote bone and cartilage tissue repair [46, 47], MSC-EVs are considered a promising alternative to conventional stem cell therapies. Since 2020, there has been an explosive growth in preclinical studies investigating the use of MSC-EVs for osteochondral defect models. However, no systematic summary has yet been conducted to comprehensively evaluate their potential reparative effects. To address this gap, the present meta-analysis included 19 preclinical animal studies, providing a thorough assessment for therapeutic efficacy of MSC-EVs. The pooled results demonstrated that, compared to control groups, MSC-EVs significantly enhanced osteochondral repair scores (ICRS and MODS), bone structural parameters (BV/TV, Tb.Th), and histological scores of cartilage and subchondral bone at both 3 and 6 months. This was accompanied by an upregulation of type II collagen expression at 3 months, indicating that MSC-EVs offer a novel therapeutic strategy for osteochondral defect repair. However, considering the presence of heterogeneity and publication bias, the interpretation of these results should be approached with caution. The ICRS and MODS scoring systems have become the gold standard for preliminary and rapid evaluation of osteochondral repair in various animal models under experimental conditions [48]. In our meta-analysis, we categorized evaluation time points at 3 and 6 months, systematically collecting and analyzing ICRS and MODS scores following MSC-EVs intervention. The results demonstrated that MSC-EVs significantly increased both ICRS and MODS scores at different time points, indicating favorable repair outcomes. Regarding imaging-based assessments, all included studies employed micro-CT scanning with quantitative analysis of regions of interest. BV/TV and Tb.Th are critical indicators reflecting changes in bone mass and microstructure during osteochondral defect repair, providing valuable reference points for clinical applications. Our findings showed that MSC-EVs significantly improved BV/TV and Tb.Th at both 3 and 6 months, suggesting notable enhancement of the osteochondral microarchitecture. In a study utilizing a miniature pig cartilage defect model, Hede et al. [49] further confirmed via micro-CT that MSC-EVs combined with bone marrow stimulation significantly promoted bone formation in the defect area compared to PBS controls and increased the tissue mineral density of the subchondral bone. Mechanically, MSC-EVs influence the function and behavior of recipient cells through three primary mechanisms: receptor-ligand interactions, membrane fusion, and endocytosis [50]. Through these pathways, EVs promote bone and cartilage repair and regeneration. An in vitro study confirmed that MSC-EVs enhance the viability and function of osteoarthritic chondrocytes, including the upregulation of genes and proteins associated with collagen and extracellular matrix (ECM) synthesis [51]. Additionally, MSC-EVs regulate the maturation of dendritic cells, suppressing pro-inflammatory cytokines such as IL-6 while promoting the expression of anti-inflammatory factors like TGF-β [52]. Furthermore, engineered MSC-EVs can be optimized to improve therapeutic efficacy and delivery protocols. Several researchers have utilized hydrogels or biomimetic scaffolds to encapsulate MSC-EVs, enabling localized sustained release and enhancing the regenerative effects on bone or cartilage tissue [29, 53]. It is well established that cartilage regeneration is an extremely complex process, involving the proliferation and migration of chondrocytes within both cartilage and subchondral bone [54]. For tissue regeneration in cartilage defect areas, in addition to direct injection, delivering MSC-EVs via bone tissue scaffolds is also a promising strategy. Ideally, scaffolds not only provide mechanical support but also structurally and functionally mimic cartilage tissue, thereby offering an optimal microenvironment for osteochondral regeneration [55]. In our included studies, six utilized hydrogel scaffolds to load MSC-EVs, achieving sustained release and prolonged bioactivity of the vesicles. Hydrogels possess excellent biocompatibility and porous structures, which better simulate the osteochondral microenvironment. For instance, Guan et al. [56] introduced aldehyde-functionalized chondroitin sulfate into gelatin methacrylate to construct a GMOCS hydrogel. After encapsulating BMSC-EVs, the composite hydrogel was shown to facilitate cartilage repair by promoting ECM synthesis. Additionally, EVs derived from synovium-derived mesenchymal stem cells integrated with thermosensitive hydrogels have been demonstrated to enhance chondrocyte anabolic activity and delay the progression of osteoarthritis [57]. Additionally, one of the included studies reported that BMSC-derived EVs preconditioned under hypoxic conditions effectively promoted bone regeneration [35]. Mechanistically, hypoxia-treated BMSC-EVs enhanced cartilage regeneration by upregulating the expression of miR-205-5p and activating the PTEN/AKT signaling pathway. Another study increased the efficacy of MSC-EVs through 3D culture, demonstrating a stronger capacity for cartilage repair [40]. However, due to the limited number of available studies, we were unable to perform further subgroup analyses based on the presence or absence of preconditioning treatments. Additionally, three studies demonstrated that MSC-EVs promote cartilage tissue repair by inducing M2 macrophage polarization [26, 30, 37]. Five studies further explored the specific molecular mechanisms by which MSC-EVs facilitate cartilage regeneration, involving multiple signaling pathways, including the TGF-β/Smad2/3 pathway [27], PI3K/AKT pathway [33], PTEN/AKT pathway [35], SOX11/FOXO pathway [36], and, MAPK pathway [37] (Fig. 1). Therefore, future research should focus on elucidating how targeted manipulation of specific molecules, such as mRNAs or miRNAs, through knockdown or overexpression strategies, influences these signaling pathways and contributes to cartilage defect repair. Clinical translation and challenges Notably, although MSC-EVs have demonstrated beneficial effects in cartilage defect regeneration, multiple challenges remain before their clinical translation. First, large-scale production continues to be a major bottleneck. During in vitro expansion, MSCs exhibit limited proliferative capacity, with reduced proliferation rates and morphological alterations observed as passages increase, thereby restricting large-scale EV preparation [19, 58]. To enhance yield, previous studies have employed genetic modification approaches, such as introducing the MYC gene, to obtain immortalized MSCs [59]. Other researchers have utilized three-dimensional culture systems, including perfusion bioreactors with 3D-printed scaffolds and microcarrier-based suspension cultures, which significantly increased EV production while preserving higher bioactivity [60, 61]. Second, EVs are limited by their short half-life in vivo and insufficient targeting ability. Hydrogels constructed from natural or synthetic polymers (e.g., collagen and gelatin) can encapsulate EVs to achieve sustained release [62, 63]. Moreover, surface modification with targeting peptides or nucleic acids, combined with external magnetic guidance (e.g., iron oxide nanoparticles), can enhance their accumulation at lesion sites [64]. In addition, the heterogeneity of EV preparations warrants attention, as differences in cell sources, culture conditions, and isolation methods may lead to variations in composition and function. Special caution is required regarding the impact of cellular senescence phenotypes on EV quality. During passaging or cryopreservation, MSCs may undergo senescence, and the EVs they secrete can carry components of the senescence-associated secretory phenotype (SASP), such as pro-inflammatory factors and growth factors [65]. These may induce senescence drift in recipient cells, trigger local inflammatory cascades, and increase therapeutic risks [66–68]. Therefore, evaluating senescence-related markers in MSC sources and analyzing the protein composition of EVs are essential for controlling product quality. The International Society for Extracellular Vesicles (ISEV) has also emphasized the importance of standardized procedures for ethical review, purity assessment, microbial contamination control, and potential toxicity evaluation [69]. Additionally, most existing studies are based on small animal models, which differ from humans in terms of size, biomechanics, and healing capacity, thereby necessitating further large-animal experiments to facilitate clinical translation [70]. A systematic evaluation and optimization of dosing regimens (including dosage, frequency, and routes of administration), as well as the effects of biomaterial scaffolds on the release kinetics and bioactivity of EVs, are required. At present, the lack of highly sensitive and specific in vivo tracing techniques may limit investigations into the biodistribution and pharmacokinetics of EVs [71]. In summary, advancing the clinical application of MSC-EVs will require breakthroughs across multiple aspects, including manufacturing processes, quality control, safety evaluation, and translational research. Strengths Firstly, through a rigorous literature search and screening process, we identified 19 preclinical animal studies and systematically evaluated the therapeutic effects of MSC-EVs on osteochondral defect models at different time points (3 and 6 months), focusing on three key aspects: osteochondral repair scores, bone structural parameters, and histological assessments. Secondly, we conducted subgroup analyses based on animal species, EVs sources, intervention routes, intervention frequencies, and treatment durations to explore potential sources of heterogeneity. Finally, we assessed the stability and reliability of the results using SYRCLE’s RoB tool, funnel plots, Egger’s test, and sensitivity analyses. Limitations Although this study is the first to systematically analyze the therapeutic effects of MSC-EVs on osteochondral injury models, several limitations need to be acknowledged. Firstly, the quality assessment of the included studies revealed a considerable number of items with unclear risk of bias, and publication bias was also detected, which may influence the pooled results. Future studies with higher methodological quality and robust evidence are needed to support these conclusions. Secondly, there was significant variability among the included animal models and the characteristics of MSC-EVs, such as their source, intervention frequency, delivery route, and duration. This heterogeneity necessitates cautious interpretation of the findings. Thirdly, the animal models primarily consisted of SD rats and New Zealand rabbits. It is important to recognize the substantial biological differences between these animals and humans, highlighting the need for further validation through large animal models and clinical trials to ensure the reliability of the results. Lastly, all included studies focused on evaluating the in vivo biocompatibility of MSC-EVs. Before clinical translation, thorough safety assessments of MSC-EVs are essential. In addition to the demonstrated efficacy and safety in animal models, the future clinical translation of MSC-EVs into advanced therapy medicinal products (ATMPs) will also require standardized protocols for EV preparation, characterization, and quality control, as well as careful evaluation of patient-specific clinical parameters. Conclusions The results of this meta-analysis showed that MSC-EVs intervention had a positive effect in improving osteochondral defect models compared to control groups, including enhancements in osteochondral repair scores, bone structural parameters, and histological assessments. However, future studies are still needed to verify the reliability of these findings using large-sample, multi-species animal models. Furthermore, prior to clinical translation, it is essential to emphasize the standardization of MSC-EVs preparation, intervention protocols, and safety evaluations. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1 Supplementary Material 2 Supplementary Material 2