The researchers implanted a 0.6-by-0.8-centimeter piece of TiLOOP light titanium-coated polypropylene mesh in a pocket created beneath the posterior vaginal wall. Each animal then received one 0.3-milliliter injection at the implantation site: saline, 900,000 HUCMSCs or 3 × 10^10 HUCMSC-sEV particles. All animals had undergone bilateral ovariectomy and were operated on one week later, creating an early estrogen-depleted condition intended to approximate the postmenopausal environment relevant to pelvic-organ prolapse. Six animals per group were scheduled for terminal tissue collection at each time point, although no formal a priori sample-size or power calculation was performed. The interventions were administered at the same anatomical site and in the same injection volume against a common saline-treated mesh control, but their doses used fundamentally different units. Without a validated conversion between cells and vesicle particles, the experiment could not determine which treatment was more potent, equivalent or interchangeable.

Small extracellular vesicles are membrane-bound particles that can transport proteins, lipids, messenger RNA and microRNAs between cells. They offer a possible cell-free alternative to administering living stromal cells, which face practical problems involving viability, storage, transport, tissue retention and batch control. In this study, the vesicles were collected from serum-free cultures of commercially obtained HUCMSCs, then concentrated and purified by filtration, ultrafiltration and ultracentrifugation. Transmission electron microscopy and nanoparticle-tracking analysis examined the preparations, while Western blotting assessed vesicle-associated proteins including CD9, CD63, HSP70 and TSG101. Characterized samples had mean particle diameters of approximately 129 to 134 nanometres, and the suspensions used for administration were concentrated to 1 × 10^11 particles per millilitre. The researchers did not directly establish the vesicles’ specific endosomal origin or complete their biological reproducibility between batches, so they used the designation small extracellular vesicles in accordance with MISEV2023 recommendations.

At one week, the median inflammatory-cell infiltration score was 4 in the saline-control group, compared with 3 in each intervention group. The overall difference survived false-discovery-rate correction, but the HUCMSC-sEV group was significantly lower than control in the formal pairwise analysis, whereas the HUCMSC comparison did not meet the adjusted threshold. Both interventions also showed lower messenger-RNA expression for interleukin-10, interleukin-12 and tumor necrosis factor-alpha. TGF-beta expression was lower in the HUCMSC-sEV group; the corresponding HUCMSC finding was significant in the primary parametric analysis but not in the nonparametric sensitivity test, indicating limited robustness. Because the investigated panel included both proinflammatory and regulatory cytokine transcripts, the results describe a changed local cytokine-transcription pattern rather than a conventional anti-inflammatory profile. The researchers did not measure cytokine proteins, immune-cell subsets, macrophage polarization, or the timing and location of individual responses, so the findings cannot define a specific immunological phenotype or establish generalized suppression of inflammation.

Four weeks after implantation, fibrous encapsulation scores had medians of 3 in the saline group and 2 in both treatment groups, and both treatment-versus-control comparisons were statistically significant after the specified pairwise adjustment. The study assessed this feature histologically using fibroblast proliferation as a semiquantitative surrogate. Moderate fibrous tissue can help fix a mesh implant, whereas excessive encapsulation may increase stiffness or contribute to contraction, so a lower score cannot by itself be equated with better tissue integration. At 12 weeks, fibrous encapsulation remained lower in both intervention groups in the main analysis, with median scores of 3 in control and 2 in each treatment group. However, neither comparison remained significant in the exact permutation sensitivity analysis, and the week-12 finding therefore requires caution. No obvious mesh exposure or erosion was seen among surviving animals, but the 12-week observation period was too short to establish long-term safety, and biomechanical properties, pain, contraction and functional outcomes were not assessed. There was also no sham-operated or no-mesh control, meaning the results cannot show that any treatment normalized the tissue toward an uninjured state.

At the four-week assessment, MMP9 messenger-RNA expression was about 4.95 times the saline-control level in the HUCMSC-sEV group and 5.29 times in the HUCMSC group. MMP2 expression was approximately 12.73 and 13.83 times control, respectively. All four differences remained significant after correction and were supported by nonparametric sensitivity analyses. In the separate 12-week cohort, MMP9 was elevated in both treatment groups, at roughly 11.05 and 9.85 times control, while MMP2 was elevated at approximately 4.36 and 3.79 times control. These are differences among independent animals assessed at specified endpoints, not measurements of molecular trajectories within the same animals. Matrix metalloproteinases participate in extracellular-matrix degradation and reorganization, but the experiment measured only transcript abundance, not MMP protein concentrations or enzymatic activity. It also examined just one inhibitor gene, TIMP1, so the balance between matrix degradation and inhibition—and the actual rate of matrix turnover—remains unknown.

At 12 weeks, the median ratio of type I to type III collagen was 24.70 in the saline-control group, 18.61 in the HUCMSC-sEV group and 18.58 in the HUCMSC group. Both intervention groups had significantly lower ratios than control, and the findings persisted in an exact permutation sensitivity analysis. Type I collagen is associated with tensile strength, whereas type III collagen is associated with a more compliant reparative matrix, making the ratio a relative measure of collagen composition. However, neither type I-positive nor type III-positive area fractions differed significantly after correction. The researchers also did not measure collagen protein abundance, synthesis and degradation rates, fibre organization or tissue mechanics. The result therefore cannot be presented as direct evidence of reduced fibrosis, stronger repair or improved mesh integration. Collagen-gene measurements were not identical between the interventions: COL3 expression increased in the HUCMSC group at 12 weeks, but not in the HUCMSC-sEV group, even though their collagen ratios were similar. Earlier, at the four-week assessment, the intervention groups also had descriptively lower collagen ratios, but the overall comparison did not remain significant after false-discovery-rate correction.

Overall, the HUCMSC and HUCMSC-sEV groups showed partially overlapping response patterns. Both were associated with lower early expression of the three measured cytokine transcripts, less fibrous encapsulation at the four-week assessment, higher MMP2 and MMP9 expression in the four- and 12-week cohorts, and a lower type I-to-type III collagen ratio at 12 weeks. The vesicle group differed in showing significantly lower histological inflammatory-cell infiltration at one week, whereas only the cell group had higher COL3 transcript expression at 12 weeks. No adjusted exploratory comparison between the two interventions remained significant, and the experiments did not selectively block vesicle activity. The parallel findings are compatible with a contribution from paracrine vesicle signaling, but they do not establish that HUCMSC-sEVs caused the effects of living cells or explain how much of either response was cell-dependent. Differences in the cellular response at one endpoint and in COL3 expression at another show that the two treatments cannot be treated as biologically identical.

The experiment was also constrained by its size and design. Only three saline-control animals contributed evaluable four-week picrosirius-red data after four samples could not be prepared; the corresponding lower collagen ratios at that point were therefore descriptive after correction. Two of the 54 animals died before scheduled tissue collection, one shortly after anesthesia and another during postoperative follow-up with findings suggesting pelvic bleeding. Neither received a formal necropsy, leaving the causes uncertain and limiting safety interpretation. Commercial HUCMSCs lacked complete archived characterization, including quantitative flow cytometry and full differentiation testing, while donor-procurement, ethics and consent documentation was unavailable. For the vesicle preparation, specific endosomal biogenesis and batch-to-batch biological reproducibility were not established. Only one dose of each intervention was tested, and the distribution and retention of cells and vesicles at the implantation site were not tracked. The ovariectomized rat model also differs from clinical pelvic-floor reconstruction in anatomy, mesh dimensions and mechanical loading. Histology and gene expression cannot establish safety or benefit in human surgery, and findings from this single titanium-coated polypropylene mesh cannot automatically be extended to other materials or pelvic-floor implants.

Further experiments would need larger samples, prospectively defined independent vesicle batches with standardized release criteria, dose-ranging designs, protein and cytokine measurements, matrix-metalloproteinase activity assays and biomechanical testing. Longer follow-up would be needed to assess erosion, exposure, contraction and durable function, while direct tracking could establish how long cells and vesicles remain at the implantation site. The rat data support further investigation of HUCMSC-sEVs as a cell-free means of modifying mesh-associated host responses, especially because their storage and standardization may be simpler than those of living cells. But the central conclusion remains deliberately narrow: under the tested conditions, the two biological interventions were associated with shifts in selected inflammatory, fibrous and matrix-related measurements around vaginal mesh, without proving therapeutic efficacy, improved tissue function or equivalence between the treatments.

Subject of Research: Preclinical effects of stromal cells and small extracellular vesicles on the host response to vaginal polypropylene mesh in rats

Article Title: Human umbilical cord mesenchymal stromal cells and their small extracellular vesicles modulate host responses to titanium-coated polypropylene mesh in the vaginal wall of rats

Article References: Ai, F., Wang, C., Wang, G., Wen, X., Chen, Z., & Wang, S. (2026). Human umbilical cord mesenchymal stromal cells and their small extracellular vesicles modulate host responses to titanium-coated polypropylene mesh in the vaginal wall of rats. Frontiers in Bioengineering and Biotechnology, 14, Article 1985008. https://doi.org/10.3389/fbioe.2026.1985008

Image Credits: AI Generated

DOI: 10.3389/fbioe.2026.1985008

Keywords: vaginal mesh, small extracellular vesicles, mesenchymal stromal cells, polypropylene, tissue response, fibrous encapsulation, collagen remodeling, extracellular matrix, animal experiment, rat model, Human, umbilical