Study selection process
The PRISMA flow diagram illustrating the literature search process is presented in Fig. 1. Based on a systematic search across the specified databases, a total of 3,245 articles were initially retrieved. These included 539 articles from Embase, 1037 from PubMed, 847 from Scopus, 770 from Web of Science, and 16 from ProQuest. After removing 1388 duplicate records, two researchers independently screened the titles and abstracts of 1857 articles. Of these, 1,806 articles were excluded due to non-compliance with the inclusion and exclusion criteria. The full texts of 51 articles were assessed, and ultimately, 27 primary studies were selected for data extraction [27, 29, 42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66]. The remaining 24 articles were excluded for the following reasons: they did not focus on burn injuries, lacked direct or indirect data extraction possibilities, used alternative effect size metrics, did not utilize MSCs, or employed bilayered scaffolds.
PRISMA flow diagram detailing study screening and selection
Study characteristics
Animal models
Among the included studies, 14 studies utilized mice, 11 studies employed rats, one study used rabbits, and one study involved pigs (Fig. 2A). Of these, 16 studies focused on male animals, six studies on female animals, and one study included both sexes (Fig. 2B). The sex of the animals was not reported in four studies. Regarding the burn models, 16 studies induced third-degree burns, nine studies created second-degree burns, and the burn model was not specified in two studies (Fig. 2C). The methods employed to induce burns, along with other relevant information pertaining to the animal model, are concisely presented in Table 1.

Comperhensive overview of included study characteristics
Mesenchymal stem cells
Out of the 20 studies that utilized MSCs, 11 studies employed animal-derived MSCs, while nine studies used human-derived MSCs (Fig. 2D). Specifically, 11 studies used adipose tissue-derived MSCs, six studies utilized bone marrow-derived MSCs, and three studies employed umbilical cord-derived MSCs (Fig. 2E). A summary of the points mentioned, along with the methods for identifying and characterizing MSCs based on ISCT guidelines [67], is provided in Table 2.
Secretome
According to Table 3, among the seven studies that isolated MSCs, three studies focused on exosomes, one study investigated microvesicles, one study examined small extracellular vesicles, one study analyzed paracrine proteins, and one study utilized conditioned medium (Fig. 2F). Of these, four studies investigated secretomes derived from umbilical cord MSCs, two studies explored secretomes from adipose-derived MSCs, and one study used secretomes from iPSC-derived MSCs (Fig. 2G). For isolation, four studies employed ultrafiltration, and three studies used ultracentrifugation as the primary methods.
Scaffold types
In terms of scaffolds, six studies utilized biological-based scaffolds, while 21 studies employed hydrogel-based scaffolds (Fig. 2H). The hydrogel formulations included injectable hydrogels, foam hydrogels, amorphous hydrogels, film hydrogels, freeze-dried hydrogels, nanofiber hydrogels, and bioprinted hydrogels (Fig. 2I). The most commonly used biomaterials in these hydrogels were hyaluronic acid, collagen, chitosan, gelatin, and alginate (Fig. 2J). The information provided pertains to scaffolds, including Scaffold Compositions, Forms, and characterization methods, as well as the Structural and Biological Properties of scaffolds, all of which are comprehensively presented in Table 4.
Result of risk of bias assessment
The risk of bias assessment, conducted using the SYRCLE tool, yielded the following findings across the included studies. In the domain of selection bias, under the subcategory of sequence generation, 18 out of 27 studies (66.7%) were classified as having a low risk of bias. In the baseline characteristics subcategory, all but one study—i.e., 26 out of 27 studies (96.3%)—were deemed to have a low risk of bias. However, in the allocation concealment subcategory, all studies were rated as having an unclear risk of bias. In the domain of performance bias, the random housing subcategory indicated that 15 out of 27 studies (55.6%) were at low risk of bias, whereas in the blinding subcategory, all studies were assessed as having an unclear risk of bias. Regarding detection bias, the random outcome assessment subcategory showed that, with the exception of three studies, the remaining 24 out of 27 studies (88.9%) were classified as low risk. In contrast, the blinding subcategory within this domain revealed that all studies had an unclear risk of bias.In the domain of attrition bias, all but three studies—i.e., 24 out of 27 studies (88.9%)—were determined to have a low risk of bias. Finally, in the domain of reporting bias, all studies were consistently rated as having a low risk of bias. In total, of the 27 studies evaluated, 19 studies (70.4%) were classified as having an overall low risk of bias and were included in our review, while the remaining 8 studies (29.6%) were deemed to have an unclear risk of bias (Fig. 3). Also, to examine the impact of studies with unclear risk of bias on potential under- or overestimation of results, a subgroup analysis was conducted for the primary outcome. As shown in Table 5 and (Supplementary Data S2), these studies did not significantly influence the overall findings.

Quality assessment of individual studies using SYstematic Review Center for Laboratory animal Experimentation (SYRCLE) tool
Primary outcome
Our analysis demonstrates that the synergistic effect of MSCs and scaffolds significantly improves wound closure rates, measured as the primary outcome, across three time frames. To evaluate the progression of wound healing, these time points were categorized as short-term (1-week), mid-term (2 weeks), and long-term (3 weeks). These intervals align with the inflammatory, proliferative, and remodeling phases of the wound healing process, respectively, and were selected based on standard practices in preclinical burn wound research. Moreover, these time points were chosen due to their frequent reporting in the scientific literature, enabling standardization and comparison of results across studies. These results are derived from 23 studies for the 1-week time point, 25 studies for 2 weeks, and 12 studies for 3 weeks. The effect was most pronounced at 1 week (SMD = 3.97, 95% CI: 2.92 to 5.01), followed by 2 weeks (SMD = 3.47, 95% CI: 2.23 to 4.61), and 3 weeks (SMD = 3.03, 95% CI: 1.96 to 4.11) (Fig. 4A–C). These results indicate that the combination of MSCs and scaffolds is highly effective in promoting wound healing, with the strongest impact observed in the early stages. However, the high heterogeneity (I2 > 50%) across studies suggests variability in experimental conditions, which should be considered when interpreting these findings.

Forest plot demonstrating the therapeutic efficacy of MSC-scaffold combinations in promoting wound closure in a burn animal model. A Week 1, B Week 2, C Week 3
Subgroup analysis and meta-regression
Due to the presence of high heterogeneity and to investigate its underlying causes as well as identify factors influencing therapeutic efficacy, we conducted a subgroup analysis across three time frames: 1 week, 2 weeks, and 3 weeks. The results revealed noteworthy findings (all results are available in Table 5 and Supplementary Data S2). Our results indicate that, in the comparison between the use of MSCs and MSC-derived secretome, the administration of MSCs (SMD = 4.75, 95% CI: 3.15 to 6.36) demonstrated superior therapeutic efficacy in the short term (1-week) compared to secretome (SMD = 2.90, 95% CI: 1.77 to 4.03). However, in the medium term (2-week) and long term (3-week), specifically in 2-week the therapeutic efficacy of secretome (SMD = 3.94, 95% CI: 2.30 to 5.58) was greater than that of MSCs (SMD = 3.20, 95% CI: 1.71 to 4.68). I2 analysis in this subgroup suggests that one of the main sources of heterogeneity was the inclusion of MSCs and MSC-derived secretome combined with scaffolds. Separate analysis of these combinations reduced heterogeneity; however, due to the limited number of secretome studies, we included both scaffold-based MSC and secretome data in the pooled results. The investigation of the therapeutic efficacy of MSCs across various animal models indicates a significant reduction in I2 (heterogeneity). This suggests that the choice of animal model utilized in the studies may be one of the key factors contributing to the heterogeneity observed in our results. The evaluation of the therapeutic efficacy of MSCs in second- and third-degree burn models reveals significant findings. In the one-week time frame, the wound closure rate in second-degree burns (SMD = 3.95, 95% CI: 2.00 to 5.90) was notably sharp compared to third-degree burns (SMD = 6.36, 95% CI: 2.99 to 9.74). However, as might be expected, the therapeutic efficacy in the second and third weeks was better in second-degree burn models (SMD = 3.49, 95% CI: 1.41 to 5.56; SMD = 3.21, 95% CI: 0.99 to 5.43) compared to third-degree burns (SMD = 2.46, 95% CI: 0.31 to 4.60; SMD = 2.83, 95% CI: 1.64 to 4.01), respectively. Additionally, the reduction in I2 suggests that the type of burn model used may be one of the contributing factors to the heterogeneity observed in our results. In addition, the type of scaffolds used also significantly influences therapeutic efficacy. Our results demonstrate that, in the one-week time frame, MSCs combined with biological scaffolds exhibited superior therapeutic efficacy (SMD = 8.83, 95% CI: 0.76 to 16.90) compared to hydrogels (SMD = 4.02, 95% CI: 2.37 to 5.67). However, in the two- and three-week time frames, the therapeutic efficacy of hydrogels (SMD = 3.41, 95% CI: 1.89 to 4.93; SMD = 3.60, 95% CI: 1.31 to 5.89) was greater than that of biological scaffolds (SMD = 2.75, 95% CI: −0.54 to 6.04; SMD = 2.74, 95% CI: 1.56 to 3.91). The absence of a notable reduction in I2 within the scaffold type subgroup can be attributed to the substantial variation in scaffold nature (i.e., polymer-based vs. biological-based scaffolds) and in their constituent components (i.e., natural or synthetic biomaterials). This broad variability in both structural origin and material composition likely contributes to the persistent heterogeneity observed, which may, in turn, affect the robustness and certainty of the derived conclusions.The type of MSCs used significantly influences both heterogeneity and therapeutic efficacy. Our analysis highlights distinct performance patterns across different MSC sources—Umbilical Cord, Bone Marrow, and Adipose—over one-, two-, and three-week time frames. Umbilical Cord-derived MSCs demonstrated the highest therapeutic efficacy in the short term, with a SMD of 6.74 (95% CI: 4.88 to 8.60) at one week. This efficacy remained robust at two weeks (SMD = 6.30, 95% CI: 4.03 to 8.56) but declined slightly by three weeks (SMD = 3.29, 95% CI: 0.64 to 5.94). Similarly, Bone Marrow-derived MSCs exhibited strong therapeutic effects, with an SMD of 6.27 (95% CI: 4.74 to 7.81) at one week, 4.27 (95% CI: 1.73 to 6.81) at two weeks, and 4.42 (95% CI: 1.83 to 7.02) at three weeks. In contrast, Adipose-derived MSCs showed comparatively lower efficacy across all time frames: SMD = 4.16 (95% CI: 1.27 to 7.05) at one week, 1.56 (95% CI: 0.01 to 3.10) at two weeks, and 1.51 (95% CI: −0.28 to 3.30) at three weeks. Furthermore, the observed reduction in I2 suggests that the source of MSCs is a key factor contributing to the heterogeneity in our results. Regarding the source of MSCs, the results demonstrate that human-derived MSCs exhibit significantly higher therapeutic efficacy compared to animal-derived MSCs in the one- and two-week time frames. Specifically, human-derived MSCs showed an SMD of 6.66 (95% CI: 3.38—9.94) at one week and 3.86 (95% CI: 1.20 to 6.52) at two weeks. In contrast, animal-derived MSCs displayed lower efficacy, with an SMD of 3.60 (95% CI: 1.78 to 5.43) at one week and 2.52 (95% CI: 1.02 to 4.02) at two weeks. However, the results in the three-week time frame present a contrasting pattern. Here, animal-derived MSCs demonstrated higher therapeutic efficacy (SMD = 3.54, 95% CI: 0.91 to 6.17) compared to human-derived MSCs (SMD = 2.85, 95% CI: 1.62 to 4.07). Regarding MSC-derived secretomes, our results indicate that secretomes extracted from Adipose-derived MSCs exhibit higher therapeutic efficacy compared to those from Umbilical Cord-derived MSCs in the one- and two-week time frames. Specifically, Adipose-derived secretomes demonstrated an SMD of 3.67 (95% CI: 2.26 to 5.08) at one week and 4.83 (95% CI: − 2.26 to 12.16) at two weeks. In contrast, Umbilical Cord-derived secretomes showed lower efficacy, with an SMD of 2.78 (95% CI: 0.97 to 4.58) at one week and 4.10 (95% CI: 3.20 to 4.99) at two weeks. These findings suggest that Adipose-derived MSC secretomes may offer superior therapeutic benefits in the short to medium term. Furthermore, the results of our meta-regression analysis indicate that no dose–response relationship was observed regarding the number of MSCs administered and their therapeutic efficacy across the three specified time frames (Supplementary Data S2). Also, given the absence of a universally established criterion for classifying sample sizes in preclinical studies, we categorized sample sizes based on the ARRIVE guidelines and the distribution of sample sizes in the included studies [68]. Specifically, we classified sample sizes as small (n < 6, below the minimum recommended threshold), sufficient (n = 6–11), or large (n > 11). Our analysis revealed that, of the 27 included studies, 12 were classified as small, 13 as sufficient, and 2 as large. This distribution suggests that the overall sample sizes are sufficient to ensure the reliability of the meta-analytic results. Furthermore, subgroup analysis based on sample size across all three time frames demonstrated that the greatest therapeutic efficacy was observed in the adequate sample size group. Consequently, smaller sample sizes did not significantly influence the intervention outcomes and were not a determining factor in the strength of the evidence.
Secondary outcomes
Angiogenesis
Our analytical results demonstrate that the synergistic effect of MSCs and scaffolds significantly enhances the expression of CD31, a key marker of angiogenesis (SMD = 6.24, 95% CI: 3.90 to 8.58) (Fig. 5A). This finding underscores the potential of combining MSCs and scaffolds to promote vascularization, which is critical for effective tissue repair and regeneration. However, the I2 values exceeding 50% for both parameters indicate high heterogeneity among the studies included in the analysis.

Forest plot demonstrating the therapeutic efficacy of MSC-scaffold combinations on: A angiogenesis, B collagen deposition, C inflammatory cytokines, D growth factors
Collagen
The findings from our analysis highlight that the synergistic interaction between and scaffolds significantly enhances collagen deposition at burn wound sites (SMD = 4.97, 95% CI: 2.22 to 7.73) (Fig. 5B). This suggests that the combined application of MSCs and scaffolds plays a pivotal role in promoting tissue regeneration and improving wound healing outcomes. However, the high heterogeneity observed, as indicated by I2 values exceeding 50%, points to substantial variability across the studies included in this analysis.
Inflammatory cytokines
Our analysis reveals that the combined use of MSCs and scaffolds exerts a powerful anti-inflammatory effect, significantly lowering the levels of pro-inflammatory cytokines. Specifically, we observed notable reductions in TNF-α (SMD = − 4.06, 95% CI: − 1.72 to −6.4), IL-6 (SMD = − 6.24, 95% CI: − 2.23 to −10.26), and IL-1 (SMD = − 5.13, 95% CI: − 1.69 to −8.56) (Fig. 5C). These results suggest that the interaction between MSCs and scaffolds plays a critical role in dampening inflammatory pathways, which could have significant implications for therapeutic strategies in inflammatory diseases and tissue repair. However, the high heterogeneity reflected by I2 values exceeding 50% for all three cytokines indicates considerable variability across the studies analyzed.
Growth factors
The results of our analyses indicate that the synergistic effect of MSCs and scaffolds significantly enhances the expression of key growth factors, including TGF-β (SMD = 6.21, 95% CI: − 1.6 to 14.03) and VEGF (SMD = 7.30, 95% CI: 4.85 to 9.75) (Fig. 5D). The observed effect sizes suggest a substantial impact of this combination on promoting growth factor activity. However, the I2 values exceeding 50% for both parameters indicate high heterogeneity among the studies included in the analysis.
Publication bias
To assess publication bias, we employed three widely used methods: the trim and fill method, funnel plot analysis, and Egger’s and Begg’s tests. According to Table 6, for the primary outcome, the asymmetric distribution in the funnel plot and the addition of studies in the trim and fill method across all three time frames suggest the presence of significant publication bias. Furthermore, Egger’s and Begg’s tests yielded p-values < 0.05 at the 1-week and 2-week time points, further supporting the existence of notable publication bias. However, at the 3-week time point, the p-values from Egger’s and Begg’s tests exceeded 0.05, indicating no significant evidence of publication bias during this period. The evaluation of publication bias for secondary outcomes also revealed notable findings. For CD31 expression, inflammatory cytokines, and growth factors, the asymmetric distribution in the funnel plot, the addition of studies in the trim and fill method, and Egger’s test with p-values < 0.05 all indicate significant publication bias. However, Begg’s test yielded p-values of 0.11, 0.12, and 0.76, respectively. The asymmetric funnel plot, the addition of studies in the trim and fill method, and Begg’s test with a p-value < 0.05 suggest significant publication bias. In contrast, Egger’s test showed a p-value of 0.11 (all results are available in Supplementary Data S3).
Sensitivity analysis
For sensitivity analysis, we employed the one-out remove method to assess the influence of individual studies on the SMD as the effect size metric. According to Table 6, for both the primary outcome and secondary outcomes, the one-out remove method revealed that no outlier studies were identified, indicating that no single study had a significant impact on the overall SMD. This suggests that the results are not disproportionately influenced by any individual study (all results are available in Supplementary Data S4).