The protocol for this study was following the guidance in the updated Pre)ferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) (Page et al., 2021). The study’s protocol was registered on PROSPERO(CRD420251080087)

Literature search strategy We searched PubMed, Cochrane, MEDLINE Complete, Web of Science databases, Embase, CNKI and Wanfang data for articles published without restriction on publication dates to 29 August 2025 With the combination of subject words and free words, the search terms included three categories: (1) “mesenchymal stromal cell”, “multipotent stromal cell”, “adipose-Derived Stem Cell”, and “bone mesenchymal stem cell”; and (2) “exosome”, “extracellular vesicle”, and “microparticle”; and (3) “keloid”, “hypertrophic scar”, and “pathologic scar”. The logical relationship was created with “OR” and “AND”; and the search formula was thereafter developed according to the characteristics of the different databases (supplementary data). The search was limited to animal trial studies. A preretrieval process improved the searches strategy. In addition, we performed a manual search of the references of the studies to obtain further potential studies. For the analysis we included studies reported in only English.

The inclusion criteria for studies were as follows:

Utilization of in vivo and in vitro experimental animal models.

Studies were excluded based on the following conditions:

Use of plant-derived EVs or EVs from sources other than mesenchymal stem cells.

All identified citations were imported into Zotero for systematic management of the search records, after the removal of duplicates. Two independent researchers screened the titles and abstracts of the studies to exclude those that did not meet the predefined inclusion criteria. As an additional precaution, the full texts of potentially relevant studies were assessed for final eligibility. Any disagreements between the two reviewers were resolved through discussion with a third team member. The study selection process was summarized in a flow diagram, in accordance with the PRISMA guidelines.

Relevant data were independently extracted by two reviewers from the included studies using a standardized, pilot-tested data extraction form in Excel (Microsoft, Seattle, United States). In cases where raw data were not provided, data were extracted from graphical representations using WebPlotDigitizer4.8. The following data were retrieved: 1) Study characteristics, including authors, study type, publication year, and country of origin; 2) Study population details, such as sample size, group, species, and scar size and depth; 3) Intervention characteristics, including criteria for MSC and EV characterization, MSC source, MSC-EV dose, and route of administration; 4) Study design information, such as comparator, sample size, MSC-EV isolation methods, and characterization protocols; 5) Measurement indicators, including scar size, cellular behaviors, collagen deposition, and protein levels of TGF-β1 and α-SMA; and 6) Risk of bias details.

STAIR was used to assess the quality of the included studies (Fisher et al., 2009). The items were as follows: (1) Sample size calculation. (2) Inclusion and exclusion criteria. (3) Randomisation. (4) Allocation concealment. (5) Reporting of animals excluded from the analysis. (6) Patients were blinded to the assessment of outcome. (7) Reporting potential conflicts of interest and study funding. Rate “Yes”, “No” or “Unclear”.

Data were analyzed using Stata 18 (Stata LP, Texas, United States). All results were treated as continuous variables and expressed as SMDs with 95% confidence intervals (CIs). A P value of <0.05 was considered indicative of a significant difference between the experimental and control groups, reflecting a substantial effect size (Higgins and Thompson, 2002). A fixed-effects model was utilized to quantitatively synthesize each outcome measure. In the presence of significant between-study heterogeneity, random-effects models were applied. The Cochrane Q test was employed to assess heterogeneity among studies, and the I² statistic was used to quantify the extent of this heterogeneity. A value greater than 50% was interpreted as substantial heterogeneity. When high heterogeneity was detected, the DerSimonian-Laird random-effects model (DL model) was used to estimate the average effect size and its corresponding confidence interval. A sensitivity analysis was conducted using Stata 18 to compare the new combined effect size with the original combined effect size, in order to determine if the results had changed substantially. If the effect size corresponding to an individual study fell outside the 95% CIs, the exclusion of that study was assessed for its impact on the total combined effect size. Additionally, if the combined effect size after excluding a study significantly differed from the combined effect size including all studies, the results were considered to be influenced by that study.

A total of 447 articles were initially screened. After removing duplicates and excluding Chinese-language publications, 200 papers were retained. Following a thorough screening of titles, abstracts, and full texts (Figure 1), 15 studies involving a total of 253 animals were ultimately included (Chen et al., 2023; Jiang et al., 2020; Liu et al., 2024; Li et al., 2021; Meng et al., 2024; Tian et al., 2024; Wang et al., 2025; Xu Z. et al., 2024; Xu C. et al., 2024; Ye et al., 2025; Yuan et al., 2021; Zhao et al., 2025; Zhen et al., 2025; Zhu et al., 2024; Zhu et al., 2020).

The baseline characteristics and methodological quality assessment results of these studies are summarized in Table 1 (Characteristics of the included studies) and Table 2 (Quality assessment). Table 1 outlines the characteristics of the 15 included studies, which were published between 2020 and 2025, with an aggregate sample size of 253. Of these, eight studies (Chen et al., 2023; Jiang et al., 2020; Liu et al., 2024; Li et al., 2021; Tian et al., 2024; Xu C. et al., 2024; Yuan et al., 2021; Zhu et al., 2024) utilized mice, two (Zhao et al., 2025; Zhen et al., 2025) employed rats and five (Meng et al., 2024; Wang et al., 2025; Xu Z. et al., 2024; Ye et al., 2025; Zhu et al., 2020) used rabbit as animal models. In terms of tissue sources, ten studies (Chen et al., 2023; Liu et al., 2024; Li et al., 2021; Meng et al., 2024; Tian et al., 2024; Wang et al., 2025; Xu Z. et al., 2024; Xu C. et al., 2024; Yuan et al., 2021; Zhu et al., 2020) used human adipose tissue, two (Zhao et al., 2025; Zhen et al., 2025) used human epidermal tissue, two (Jiang et al., 2020; Zhu et al., 2024) used human bone marrow tissue, and one (Ye et al., 2025) study utilized human umbilical cord tissue. All included studies were conducted in China. Exosome isolation and validation in these studies were performed using several methods, including ultracentrifugation for exosome extraction and identification, transmission electron microscopy (TEM) for morphological assessment of adipose-derived mesenchymal stem cells exosomes (ADSC-Exos), nanoparticle tracking analysis (NTA) for determining particle size distribution, and Western blotting to detect membrane surface markers such as Alix, TSG101, CD9, CD63, and CD81.All studies employed rodent models with full-thickness skin defects, ranging from 6 to 20 mm in diameter, located on the back or footpad. The control groups in the studies included various types of controls, such as phosphate-buffered saline and placebo treatments. Significant heterogeneity was observed across the intervention and control treatments for all outcome variables (P < 0.05).

The Systematic Review of Animal Intervention Studies (STAIR) tool was utilized to evaluate the methodological quality of the included studies. The majority of the studies were randomized controlled trials (RCTs); however, the specific randomization methods employed were not clearly specified. Several studies either failed to report the use of randomization or did not mention allocation concealment. A detailed summary of the study quality assessment is presented in Table 2.

Sensitivity analysis of scar reduction rate, collagen deposition, the expression of TGF-β1 and α-SMA, and fibroblast migration and proliferation was performed. No individual study’s effect size lay beyond the 95% confidence interval, nor did it overturn the pooled estimate, confirming that no single investigation exerted a disproportionate influence on the overall results; thus, the meta-analytic findings for every outcome measure remain robust.

Hypertrophic scars and keloids are fibroproliferative skin disorders characterized by excessive fibroblast activity. These conditions cause significant functional and cosmetic impairments, which contribute to a substantial psychological burden for affected individuals (Frech et al., 2023). Despite the availability of various therapeutic options, the treatment of hypertrophic scars and keloids remains challenging. No single modality is universally recommended, and none provide complete therapeutic efficacy (Tripathi et al., 2020). This review focuses on stem cell-derived exosome therapy, a novel cell-free treatment approach that has recently emerged as a promising strategy for intervention. Exosomes, due to their natural origin, exceptional biocompatibility, and multifunctional properties, are increasingly recognized as innovative platforms for drug delivery, immune modulation, and precision-based therapies (Zhou et al., 2023). Exosomes have the ability to regulate nearly all stages of wound healing and keloid suppression, including immune and inflammatory modulation, macrophage polarization, angiogenesis, cellular proliferation and migration, and ECM remodeling (Ma et al., 2022; Zhou et al., 2023; Ti et al., 2015; Shabbir et al., 2015). Achieving scar-free healing during the early stages of wound repair is the ideal strategy for minimizing scar formation. However, chronic wounds, particularly those with increasing prevalence in younger populations (<65 years), remain a significant public health concern, underscoring the ongoing relevance of this research (Sen, 2023; Carter et al., 2023). The transition from wound clot to granulation tissue in hypertrophic scars and keloids is regulated by a delicate balance between ECM protein deposition and degradation. Disruption of this process leads to abnormal scarring, ultimately resulting in the development of either keloids or hypertrophic scars (Tripathi et al., 2020). The transient transformation of fibroblasts into myofibroblasts is essential for acute wound healing; however, prolonged activation of myofibroblasts contributes to scar formation and fibrosis. (Younesi et al., 2024). Dysregulation of TGF-β/Smad signaling is a major contributor to scar formation and fibrosis, as it triggers excessive ECM deposition as well as the generation and activation of local myofibroblasts. This process is characterized by aberrant collagen synthesis and accumulation, an increased COL I/III ratio, and the formation of abnormally cross-linked collagen fiber bundles, ultimately driving the phenotypic transition of fibroblasts into myofibroblasts (Jiang et al., 2023; Zhang et al., 2020; Lichtman et al., 2016; Tang et al., 2018). During the wound-healing phase, activation of myofibroblasts promotes wound closure through ECM production and contraction. As myofibroblasts further mature, the incorporation of α-SMA, a specific actin isoform, not only enhances fibroblast contractility but also reinforces intracellular mechanotransduction feedback loops that sustain myofibroblast activation, leading to the development of excessively proliferative and rigid scars. Consequently, α-SMA is widely regarded as the most commonly used marker for identifying myofibroblasts in both cell cultures and fibrotic tissue sections Understanding Keloid Pathobiology (Skalli et al., 1986; Younesi et al., 2024; Hinz, 2016; Hinz et al., 2001; Talele et al., 2015). In summary, cutaneous wound healing involves the dynamic formation and remodeling of the collagen matrix by fibroblasts and myofibroblasts over time (Tan et al., 2019). Accordingly, the present study considers TGF-β and α-SMA as key regulatory targets through which MSC-EVs modulate scar formation.

Recent studies have increasingly focused on exosome-based therapies for wound healing. These analyses suggest that exosomes may attenuate the expression of genes involved in ECM remodeling, which is typically associated with scar formation. In the early phases of wound healing, there is an accelerated deposition and maturation of collagen. However, during the remodeling phase, the ECM and granulation tissue undergo dynamic restructuring, which involves both degradation and resynthesis processes, playing pivotal roles in the progression of Regulate the process of scar formation (Yuan T. et al., 2023). Our analysis highlights the primary role of exosome-mediated effects during the proliferative and remodeling stages of healing, while also acknowledging the influence of the initial inflammatory phase on fibrosis. Mast cells and macrophages are key players in this process. Mast cells contribute to fibrosis by releasing chemical mediators during degranulation, which activate multiple distinct signaling pathways (Nakajima et al., 2022; Yeo et al., 2024; Siddhuraj et al., 2022). M2 macrophages are particularly important in the pathogenesis of fibrotic disorders, emphasizing the need for further animal studies to explore the therapeutic potential of MSC-Exos interventions during the inflammatory phase (Ge et al., 2024; Dai et al., 2024).

Recent evidence increasingly supports the role of exosomes in alleviating organ fibrosis through various mechanisms. Deng demonstrated that in a myocardial infarction model, exosomes derived from ADSCs mitigated cardiac damage by activating the S1P/SK1/S1PR1 signaling pathway and promoting M2 macrophage polarization (Deng et al., 2019). Shi observed that in a bleomycin-induced idiopathic pulmonary fibrosis model, exosomes derived from human umbilical cord mesenchymal stem cells extracellular vesicles (hucMSC-EVs) alleviated fibrosis by delivering miR-21-5p and miR-23-3p. These microRNAs inhibited TGF-β signaling, leading to a reduction in myofibroblast differentiation and collagen deposition (Shi et al., 2021). In models of silicosis-induced pulmonary fibrosis, ADSC-Exos reduced collagen deposition and slowed fibrosis progression (Bandeira et al., 2018). In a liver fibrosis model, treatment with human amniotic mesenchymal stem cell-derived EVs resulted in a reduction of α-SMA expression and fibrotic areas by inhibiting the activation of Kupffer cells and hepatic stellate cells (Ohara et al., 2018). In a renal fibrosis model, human umbilical cord mesenchymal stem cell-derived exosomes (hucMSC-Exo) attenuated collagen deposition and interstitial fibrosis by inhibiting YAP activity through CK1δ/β-TRANP signaling (Ji et al., 2020). Despite these promising results, the therapeutic efficacy of exosomes is somewhat limited by the inherent inefficiency of native EVs and their inability to target multiple organs effectively. To maximize their therapeutic potential, modifications to improve delivery and targeting capabilities are often required (Zhang et al., 2024).

Given the therapeutic potential of exosomes in various diseases, there has been increasing interest in the development of engineered or hybrid exosomes. However, the clinical application of exosomes is limited by several challenges, including difficulties in large-scale production, instability under varying environmental conditions, and aggregation during storage (Mondal et al., 2023). While synthetic polymers used in nanocarrier design offer enhanced stability and controlled release properties, their clinical translation is hindered by concerns regarding safety, compatibility, and toxicity (Agrawal et al., 2014; Ahmad et al., 2022). This has spurred the exploration of hybrid exosomes, which aim to combine the benefits of both exosomes and synthetic systems, potentially enhancing their clinical efficacy. In addition, plant-derived exosome-mimetic nanoparticles exhibit inherent biocompatibility, anti-inflammatory, and antioxidant properties, along with renewable and sustainable characteristics, making them promising candidates for scalable production and clinical application (Feng et al., 2024). The convergence of tissue engineering and nanotechnology has accelerated the development of extracellular vesicle-based strategies, with engineered cargo loading, targeted delivery, and stimulus-responsive systems showing substantial promise for tissue regeneration and repair. Despite these advances, exosome-nanotechnology hybrid systems remain largely confined to preclinical studies (Liu et al., 2022). Extensive clinical trials are necessary to validate their translational potential and establish their efficacy in clinical settings.

The potential of EVs in endogenous tissue remodeling and therapeutic tissue repair has been increasingly recognized, leading to a growing number of early-phase clinical trials investigating EV-based therapies. Nevertheless, substantial challenges continue to hinder their clinical translation, including compliance with current Good Manufacturing Practice (cGMP) guidelines, high reproducibility in allogeneic settings, scalability, stability, storage, and large-scale production under stringent storage and clinical quality control requirements (Elsharkasy et al., 2020).To enable the successful clinical implementation of exosomes as a promising acellular therapeutic strategy across diverse medical fields, it is imperative that researchers and manufacturers strictly adhere to cGMP standards. This entails the establishment of standardized production processes, harmonized quality control systems, and well-defined clinical research protocols to ensure safety and efficacy (Elsharkasy et al., 2020; Lener et al., 2015). Furthermore, coordinated global efforts are required to develop industry-wide standards and specialized guidelines focused on cellular activity and drug mechanisms (Li et al., 2025). Such frameworks would provide consistent reference criteria for both preclinical and clinical studies and, to some extent, mitigate methodological heterogeneity across investigations.

In conclusion, exosomes play a pivotal role in cellular signaling due to their complex composition, which includes a diverse array of signaling cargos such as proteins, lipids, surface receptors, enzymes, cytokines, transcription factors, and nucleic acids (Hade et al., 2021). This extensive cargo enables exosomes to facilitate both targeted therapy and disease diagnosis. While the majority of studies included in our analysis focused on evaluating the effects of individual components of MSC-Exos on specific skin cells, there remains a gap in systematically assessing the bioactive components of exosomes and their broader impact on various skin cell types (Zhou et al., 2023). A more comprehensive evaluation of these bioactive components is essential to fully understand their therapeutic potential and mechanisms of action.

When evaluating the included studies using the SYRCLE tool, several studies failed to report the use of randomized controlled trial (RCT) methodologies, and nearly all experiments lacked detailed descriptions of specific random allocation methods. This resulted in a potential risk of bias across multiple study design domains. Furthermore, the limited number of studies eligible for meta-analysis was insufficient to adequately assess publication bias. Regarding study design, there was considerable variability in the animal models used, wound creation techniques, wound diameter or area measurements, and methods for assessing scar size, which may have impaired the precise evaluation of study outcomes. In addition, interspecies differences in anatomy and physiology between animal models and humans may substantially influence experimental outcomes. For example, wound healing in rodents occurs predominantly through wound contraction, whereas rabbits can partially recapitulate human metabolic characteristics (Gottrup et al., 2000; Abdullahi et al., 2014), and porcine skin morphology, physiological structure, and wound-healing processes more closely resemble those of human skin (Sullivan et al., 2001). In recent years, to enhance the translational relevance of basic research to clinical applications, increasingly sophisticated in vitro human skin–mimicking models have been developed, ranging from conventional two-dimensional (2D) skin cell cultures to three-dimensional (3D) tissue-engineered constructs. These models more closely approximate native human skin architecture and are able to capture key features of scar formation to a certain extent (Hofmann et al., 2023).