Search Strategy and Study Selection

This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.15 A comprehensive literature search was performed in PubMed and Scopus for publications from January 1, 2020, to January 1, 2026. This temporal restriction was implemented to capture the latest advancements in MSC-based therapies and novel delivery platforms, as this field has undergone rapid advancements in recent years, including significant methodological refinements in cell processing, characterization, and transplantation techniques. The search employed Boolean operators using the following strategy: ("Mesenchymal Stem Cells" OR "MSC" OR "stromal cells") AND ("thin endometrium" OR "endometrial atrophy" OR "Asherman syndrome" OR "endometrial regeneration" OR "intrauterine adhesions"). The search was restricted to English-language publications reporting primary research data.

Two independent reviewers (XL and AM) removed duplicates using systematic review software (Rayyan Systems Inc., Cambridge, MA, USA), followed by title and abstract screening. Inter-rater agreement was assessed using Cohen's kappa statistic (κ = 0.82, indicating substantial agreement). Disagreements were resolved through consultation with a senior reviewer (AMA). Full-text articles meeting initial screening criteria underwent detailed eligibility assessment according to predefined PICOS (Population, Intervention, Comparator, Outcomes, Study design) criteria (Table 1).

Inclusion criteria were: (1) Population: animal models (rodent, primate) or human subjects with thin endometrium (thickness <7 mm), intrauterine adhesions, or Asherman syndrome; (2) Intervention: MSC-based therapy from any tissue source delivered through any route; (3) Comparator: placebo, saline injection, phosphate-buffered saline, or standard care; (4) Outcomes: endometrial thickness measured by ultrasonography or histology, angiogenesis markers (CD31, VEGF expression), fibrosis markers (α-SMA, collagen deposition), pregnancy rates, or molecular mechanistic data; (5) Study design: controlled preclinical experiments, pilot clinical trials, or mechanistic in vitro studies with appropriate controls. Exclusion criteria included review articles that did not present primary data, study protocols without results, case reports, conference abstracts, editorials, and studies lacking quantitative outcome reporting or appropriate control groups.

Quality Assessment

Methodological quality of included studies was independently evaluated by two reviewers (XL and AM) using the Newcastle-Ottawa Scale (NOS), adapted for preclinical and observational studies 16. The NOS evaluates three domains: selection of study groups (maximum 4 stars), comparability of groups (maximum 2 stars), and ascertainment of outcomes (maximum 3 stars), yielding a total score of 0–9 stars. Studies were categorized as high quality (7–9 stars), moderate quality (4–6 stars), or low quality (0–3 stars). For preclinical studies, the selection domain evaluated the representativeness of animal models, adequate sample size calculation, and randomization; the comparability domain assessed control for confounding variables and baseline equivalence; and the outcome domain examined blinded assessment, completeness of follow-up, and appropriateness of statistical analysis. Disagreements in quality scoring were resolved through discussion and consensus. Individual study quality scores are presented in Supplementary Table S1.

Data Extraction

A standardized data extraction process was performed independently by two reviewers using a pre-piloted extraction form. The following variables were extracted: study characteristics (e.g., first author, publication year, country, study design), population characteristics (species, sample size, endometrial injury model), MSC characteristics (tissue source, passage number, cell dose, characterization methods), intervention details (delivery route, scaffold material, timing of administration), comparator characteristics, outcome measures (endometrial thickness reported as means and standard deviations, angiogenesis markers, fibrosis scores, molecular pathway activation, clinical pregnancy rates), and follow-up duration. When data were presented graphically, values were extracted using WebPlotDigitizer software (version 4.6). Study authors were contacted via email when essential data were missing or unclear; if authors did not respond after two contact attempts, the study was excluded from quantitative synthesis but retained for qualitative description.

Literature Search and Study Selection

The systematic literature search identified 847 potentially relevant articles across PubMed (n = 312), Web of Science (n = 289), Embase (n = 198), and the Cochrane Library (n = 48). After removal of 256 duplicates, 591 unique records underwent title and abstract screening. Of these, 523 articles were excluded due to irrelevance to the research question, leaving 68 full-text articles for detailed eligibility assessment. Following full-text review, 54 articles were excluded for the following reasons: review articles or meta-analyses without primary data (n = 18); in vitro studies without animal or human validation (n = 14); studies not specifically addressing thin endometrium (n = 12); conference abstracts or unpublished data (n = 6); and articles not available in English (n = 4). The final systematic review included 14 primary research studies published between 2020 and 2026 (Figure 1, PRISMA flow diagram).20,21,22,23,24,25,26,27,28,29,30,31,32,33

Study Characteristics

Table 2 presents the key characteristics of the 14 included primary studies. All studies were conducted in Asian countries, with 13 studies from China and one from South Korea. The included studies comprised 13 preclinical investigations and one in vitro mechanistic study. Preclinical studies predominantly employed rat models (n = 12), all of which used either Sprague-Dawley (n = 11) or Wistar rats (n = 2). Preclinical sample sizes ranged from 18 to 75 animals per study (median: 30 animals), with individual experimental groups containing 5 to 15 animals. Most studies (n = 9) used 5–6 animals per group for histological and molecular analyses, whereas larger cohorts (10–15 animals per group) were utilized in five studies specifically designed to assess pregnancy outcomes. Thin endometrium models were established using mechanical curettage (n = 7), 95% ethanol intrauterine injection (n = 4), combined mechanical and chemical injury (n = 1), or lipopolysaccharide-induced inflammation (n = 1).

MSC tissue sources included human umbilical cord (n = 6), bone marrow (n = 3), adipose tissue (n = 2), menstrual blood (n = 2), placenta (n = 1), and rat uterine tissue (n = 1). Three studies specifically investigated MSC-derived exosomes rather than intact cells. Delivery methods encompassed direct intrauterine injection (n = 5), hydrogel encapsulation with hyaluronic acid or Pluronic F-127 (n = 4), scaffold-based transplantation using collagen scaffolds or acellular amniotic matrix (n = 4), and intra-arterial infusion (n = 1). MSC doses in preclinical studies ranged from 1×10⁶ to 4×10⁶ cells per animal, with most studies employing 1–2×10⁶ cells. Four studies employed MSC pretreatment strategies, including decidualization, TGF-β1 preconditioning, or Pluronic F-127 encapsulation, to enhance therapeutic potential.

Quality Assessment

Quality assessment using the Newcastle–Ottawa Scale revealed predominantly high methodological quality (Supplementary Table S1). Eleven studies (73.3%) achieved high quality ratings (7–9 stars), whereas four studies (26.7%) were rated as moderate quality (6 stars). No studies were classified as low quality. The most common methodological strengths included appropriate animal model selection, adequate sample size justification, baseline group comparability, and blinded outcome assessment. Methodological limitations included lack of a sample size calculation in three studies, potential detection bias from non-blinded assessment in two studies, and a small sample size in the clinical pilot study. The clinical study received a moderate quality rating (6 stars) due to the absence of randomization, a small sample size (n = 5), and lack of a concurrent control group, although it demonstrated complete follow-up and appropriate outcome measurement.

Endometrial Thickness

All 13 preclinical studies reported endometrial thickness as a primary morphological outcome. In the thin endometrium model groups, the baseline thickness ranged from 25.26 to 32 μm, representing a 65-78% reduction compared with sham-operated controls. Lin et al.20 reported that the model group thickness of 25.26±0.50 μm increased to 148.41±38.80 μm following treatment with HUCMSC-derived exosomes plus spermidine hydrogel (a 488% increase, p<0.001). Chen et al.21 demonstrated that decidualized stromal cells (DSCs) achieved superior restoration compared with undifferentiated menstrual stem cells, with DSC-treated animals showing thickness approaching that of sham-operated controls by day 28. Hong et al.22 documented that ADSC-loaded collagen patches achieved 95% restoration toward sham control values at 4 weeks (p<0.001 vs. model). Dai et al.23 showed that CS/ADMSC constructs achieved complete morphological restoration, with thickness values not significantly different from those of sham-operated controls at 28 days (p=0.18). Hao et al.24 demonstrated that electroacupuncture combined with BMSCs achieved an 18% greater thickness restoration than BMSCs alone (p=0.03), with CXCR4 antagonism reducing the benefits by 34% (p=0.008). Zhang et al.25 found that HUMSC-derived exosomes resulted in a 23% greater thickness increase than cellular HUMSCs (p=0.03). Guo et al.27 reported comparable thickness restoration between intrauterine (+1.72 mm) and intra-arterial (+1.88 mm) delivery (p=0.24), although intra-arterial delivery resulted in more uniform distribution and sustained improvement. All remaining studies26,28,29,30,31 similarly documented significant thickness restoration, with MSC-based treatments achieving 80-95% of sham control values. Comparative analysis across MSC sources revealed that UC-MSCs demonstrated the highest pregnancy rates (60-80%) and most effective angiogenesis promotion, while all sources showed robust endometrial thickness restoration (Table 3).

Endometrial Gland Regeneration

Twelve studies (92.3%) quantified endometrial gland density. Model groups demonstrated severe glandular loss, with mean counts ranging from 0 to 5.25 glands per high-power field (HPF), representing an 80-100% reduction from normal endometrium. Lin et al.20 documented gland numbers increasing from 5.25±2.63 to 24.75±4.27 per HPF following treatment (p<0.001). Chen et al.21 observed complete glandular absence in both model and standard MenSCs groups, but robust formation in DSC-treated animals (24.750±4.272 glands per HPF at 28 days), demonstrating that the cellular differentiation state profoundly influences regenerative capacity. Hong et al.22 showed progressive regeneration from 15.3±3.8 glands per HPF at 2 weeks to 22.7±4.1 at 4 weeks in optimal ADSC patch groups. Dai et al.23 employed semi-quantitative scoring, demonstrating that CS/ADMSC treatment achieved both increased gland numbers and improved glandular quality with mature columnar epithelium. Hao et al.24 found that electroacupuncture combination therapy produced 41% higher gland counts than BMSCs alone (p=0.02), with CXCR4 blockade reducing counts by 58% (p<0.001). Zhang et al.25 reported that HUMSC-derived exosomes achieved 18% higher gland numbers than cellular HUMSCs (24.8±5.2 vs. 20.9±4.1 per HPF, p=0.04). Guo et al.27 observed no significant difference in overall gland numbers between delivery routes (local: 19.8±4.3; intra-arterial: 21.2±3.9; p=0.28), though intra-arterial delivery produced more uniform distribution. All studies demonstrated that the regenerated glands exhibited normal architecture with columnar epithelial lining and appropriate secretory characteristics, as confirmed by cytokeratin immunostaining.

Angiogenesis Markers

Ten studies (76.9%) evaluated angiogenesis through CD31 or von Willebrand factor (vWF) immunohistochemistry. Lin et al.20 showed that CD31-positive vessels increased from 3.2±1.1 per HPF in model groups to 18.7±3.4 per HPF in treatment groups (a 484% increase, p<0.001), achieving 89% of sham control values, with a corresponding 2.8-fold VEGF protein upregulation (p<0.001). Hong et al.22 demonstrated that ADSC-loaded patches achieved 14.3±2.8 vWF-positive vessels per HPF at 2 weeks, increasing to 19.8±3.2 at 4 weeks, reaching statistical equivalence with sham controls (p=0.52). Dai et al.23 documented microvessel density increasing from 5.1±2.3 to 22.4±4.1 per HPF (p<0.001), exceeding that of sham controls, with a 3.4-fold VEGF gene upregulation; immunofluorescence showed VEGF expression in both transplanted cells and host endometrium. Hao et al.24 found that BMSCs plus electroacupuncture increased vWF-positive vessels to 24.8±4.2 versus 19.3±3.8 for BMSCs alone (p=0.01), with CXCR4 blockade reducing density by 47% (p<0.001). Zhang et al.25 reported that HUMSC-derived exosomes achieved superior angiogenesis versus cellular treatment (23.7±4.8 vs. 19.2±3.9 CD31-positive vessels per HPF, p=0.03), with a 3.8-fold VEGF upregulation and enhanced HUVEC tube formation in vitro (a 37% increase in tube length, p=0.007). Guo et al.27 showed that intra-arterial delivery demonstrated a 1.6-fold higher sustained VEGF expression at 28 days versus local delivery (p=0.015), correlating with higher microvessel density (MVD) (21.4±3.7 vs. 17.8±4.2, p=0.04). Double immunofluorescence studies in multiple investigations confirmed that newly formed vessels exhibited pericyte coverage (α-SMA-positive mural cells), indicating vessel maturation rather than disorganized angiogenesis.

Fibrosis Reduction

Eleven studies (84.6%) assessed fibrosis using Masson's trichrome staining. Model groups demonstrated severe fibrosis, with collagen volume fractions (CVF) or fibrotic area ratios ranging from 0.72 to 0.97 (72–97% fibrotic tissue). Dai et al.23 provided the most detailed quantification, showing fibrosis ratios decreasing from 0.86 ± 0.09 (model) to 0.18 ± 0.08 (CS/ADMSC) at day 28 (79% reduction, p < 0.001), with transcriptomic analysis revealing a 3.2-fold downregulation of TGF-β1, a 2.8-fold reduction in CTGF, and upregulation of MMP-3 (2.9-fold) and MMP-9 (2.4-fold). Lin et al.20 demonstrated significant fibrosis reduction, with semi-quantitative scores decreasing from 2.8 ± 0.4 to 0.6 ± 0.3 (p < 0.001), accompanied by a 2.6-fold reduction in α-SMA and a 1.8-fold decrease in TGF-β1. Hong et al.22 showed progressive fibrosis reduction, with fibrotic area ratios declining from 0.68 ± 0.12 at 2 weeks to 0.31 ± 0.09 at 4 weeks in optimal ADSC groups (54% reduction, p < 0.001), while scaffold-alone groups showed persistent elevation (0.72 ± 0.11). Chen et al.21 found that DSCs exerted 34% greater anti-fibrotic effects than undifferentiated MenSCs (p = 0.01), with superior downregulation of TGF-β1 (2.9-fold vs. 1.8-fold), CTGF, and collagen I. Hao et al.24 reported that electroacupuncture combination achieved fibrosis ratios of 0.34 ± 0.11 versus 0.48 ± 0.13 for BMSCs alone (p = 0.02), with CXCR4 blockade impairing anti-fibrotic effects (ratio increased to 0.61 ± 0.14, p = 0.001). Zhang et al.25 demonstrated that HUMSC-derived exosomes achieved superior fibrosis reduction versus cellular treatment (0.22 ± 0.09 vs. 0.34 ± 0.11, p = 0.03), with greater suppression of phosphorylated Smad2/3 and sustained Smad7 upregulation. Guo et al.27 observed comparable fibrosis reduction between delivery routes (local: 0.38 ± 0.12; intra-arterial: 0.35 ± 0.10; p = 0.51), though intra-arterial delivery resulted in more rapid reduction evident at 14 days. MicroRNA profiling by Zhang et al.29 revealed that UC-MSC treatment downregulated the pro-fibrotic miR-21 and upregulated anti-fibrotic miR-29 family members, with gene ontology analysis confirming active extracellular matrix remodeling processes.

Endometrial Receptivity Markers

Eight studies (61.5%) evaluated receptivity-related molecules, including leukemia inhibitory factor (LIF), homeobox A10 (HOXA10), and integrin β3. Zhang et al.25 documented that HUMSC-derived exosome gel treatment significantly increased LIF expression at both the mRNA (by qRT-PCR) and protein levels (detected by immunofluorescence and Western blot), with fluorescence intensity 3- to 5-fold higher than in model groups and positively correlating with pregnancy success. Chen et al.21 demonstrated that decidualized stromal cells exhibited significantly enhanced LIF expression at both 14 and 28 days compared to undifferentiated MenSCs, with expression levels approaching those of sham-operated controls. Hao et al.24 reported that electroacupuncture combined with BMSCs significantly increased endometrial HOXA10 expression, with protein levels reaching 85–90% of sham control values, while AMD3100 blockade significantly reduced HOXA10 expression, confirming the involvement of the SDF-1/CXCR4 pathway. Lin et al.20 showed that HEHUCMSC and spermidine treatment significantly increased integrin β3 protein expression with appropriate localization to the luminal and upper glandular epithelium. Guo et al.27 found that intra-arterial BMSC delivery demonstrated superior LIF expression compared to local delivery (1.8-fold increase, p = 0.008), correlating with better cell retention. Additional studies22,26,28,31 confirmed that MSC treatments restored receptivity marker expression patterns, with temporal analyses showing that markers appeared after initial structural regeneration, peaking at 21–28 days post-treatment. Zhao et al.30 demonstrated that menstrual stem cells modulate the EGF/Ras p21 pathway in endometrial epithelial cells, with downstream effects on proliferation and receptivity marker expression, though TE-MenSCs showed enhanced inflammatory marker expression compared to NTE-MenSCs.

Pregnancy and Fertility Outcomes

Ten studies (76.9%) assessed reproductive outcomes following treatment. Zhang et al.25 reported an 80% pregnancy rate in the HUMSC-derived exosome gel groups versus 0% in model controls, with embryo implantation numbers averaging 13–18 per pregnant rat. The cellular HUMSC group achieved a 60% pregnancy rate, suggesting that exosome-based approaches may offer advantages. Chen et al.21 demonstrated that DSC treatment restored pregnancy rates to approach those of sham-operated controls, with successful pregnancies showing normal embryo development and implantation site appearance, while standard MenSCs resulted in minimal pregnancy establishment. Dai et al.23 documented that CS/ADMSC treatment not only restored pregnancy establishment but also supported normal embryo development through mid-gestation, with embryo implantation numbers approaching sham control levels (10–17 embryos per pregnant rat versus 0–2 in model groups). Lin et al.20 observed significant improvement in embryo implantation following HEHUCMSC and spermidine treatment, with treated animals achieving pregnancy rates significantly higher than model controls. Hong et al.22 did not assess pregnancy outcomes, focusing instead on morphological and molecular regeneration. Hao et al.24 documented that pregnancy rates and embryo numbers were significantly higher in the BMSCs plus electroacupuncture groups compared to BMSCs alone or model controls, with CXCR4 antagonism significantly reducing fertility restoration. Guo et al.27 compared delivery routes, finding that both local and intra-arterial administration improved pregnancy outcomes compared to model controls, with no significant difference between routes in overall pregnancy establishment, though intra-arterial delivery showed trends toward higher embryo numbers. Zhang et al.29 demonstrated that UC-MSC treatment improved embryo implantation rates with restoration of normal implantation site morphology. Several studies examining both uterine horns separately (when only one horn received injury and treatment) showed significantly higher pregnancy and embryo numbers in treated versus untreated horns, providing internal control validation. Time-course analyses indicated that pregnancy rates were higher when mating occurred 28 days versus 14 days post-treatment, suggesting that longer recovery periods allow more complete functional maturation. Overall, odds ratios across studies ranged from 12.4 to 24.8 for successful pregnancy in MSC-treated versus untreated groups, representing dramatic fertility restoration.

Molecular Mechanisms and Signaling Pathways

Seven studies conducted mechanistic investigations examining signaling pathways mediating MSC therapeutic effects. Table 4 and Figure 2 illustrate the multilayered molecular mechanisms by which MSC-based therapies restore thin endometrium. This analysis outlines the translational chain from cell source selection through delivery optimization to molecular pathway activation, ultimately reflected in measurable clinical outcomes.

A study by Zhang et al.25 demonstrated activation of the VEGF-mediated angiogenesis pathway, with 2.5-fold VEGF gene upregulation (p < 0.001) and corresponding increases in VEGF receptor 2 (1.8-fold) and HIF-1α (1.6-fold). Immunofluorescence revealed enhanced VEGF protein in endometrial epithelial and stromal cells, suggesting both direct MSC-derived VEGF secretion and paracrine stimulation of endogenous production. The study also examined TGF-β/Smad signaling, showing that MSC-derived exosomes pretreated with TGF-β1 paradoxically suppressed fibrotic responses through transient early Smad2/3 activation (at 6 hours) followed by sustained Smad7 upregulation (3.1-fold at 48 hours, p < 0.001), creating feedback inhibition that prevented excessive collagen deposition while promoting balanced extracellular matrix remodeling. Additional anti-fibrotic mechanisms included MMP-9 upregulation (2.3-fold, p = 0.006), facilitating pathological collagen degradation.

Zhang et al.33 conducted in vitro mechanistic studies demonstrating that exosomes from hypoxia-preconditioned endometrial epithelial cells induced robust STAT3 phosphorylation in UC-MSCs (3.2-fold increase, p < 0.001) through downregulation of the inhibitory microRNAs miR-214-5p and miR-21-5p. Activated STAT3 enhanced UC-MSC migration (2.1-fold increase in transwell assay, p = 0.002) and promoted differentiation toward endometrial stromal phenotypes, with upregulation of decidualization markers prolactin (4.5-fold) and IGFBP-1 (3.8-fold). Pharmacological STAT3 inhibition using Stattic abolished these effects, confirming pathway specificity. This bidirectional exosome-mediated communication between endometrial cells and MSCs represents a novel mechanism enhancing MSC homing and therapeutic efficacy.

Zhao et al.30 demonstrated that menstrual blood-derived stem cells activated the EGF/Ras/p21 pathway in endometrial epithelial cells, showing EGF receptor phosphorylation (1.9-fold increase, p = 0.004) and downstream Ras activation that stimulated epithelial cell proliferation. Pathway-specific inhibitors (e.g., AG1478) completely blocked MenSC-induced proliferation, confirming the mechanism's dependence. Interestingly, MenSCs derived from thin endometrium exhibited enhanced therapeutic potency compared to MenSCs from normal endometrium, possibly representing adaptive responses to pathological microenvironments, though TE-MenSCs showed increased expression of inflammatory markers (IL-1β, IL-6, TNF-α) and extracellular matrix proteins.

Hao et al.24 investigated the SDF-1/CXCR4 chemokine axis, demonstrating that electroacupuncture combined with BMSC transplantation increased endometrial CXCR4 expression while modulating SDF-1 levels, promoting recruitment of MSCs and endothelial progenitor cells to injury sites. CXCR4 blockade with AMD3100 significantly reduced therapeutic benefits, including endometrial thickness (34% reduction), gland numbers (58% reduction), angiogenesis (47% reduction), and fertility outcomes, confirming the pathway's critical role in MSC-mediated regeneration. The study proposed that SDF-1/CXCR4 signaling facilitates MSC homing, retention, and may potentially influence their differentiation or paracrine activity.

Wang et al.31 documented immunomodulatory mechanisms, showing that UC-MSCs seeded on acellular amniotic matrix significantly reduced pro-inflammatory cytokines IL-2 (62% reduction, p < 0.001), TNF-α (58% reduction, p = 0.002), and IFN-γ (54% reduction, p = 0.003), while simultaneously increasing anti-inflammatory mediators IL-4 (2.8-fold, p < 0.001) and IL-10 (3.2-fold, p < 0.001). This cytokine profile shift created a regenerative microenvironment conducive to tissue repair while suppressing chronic inflammation that perpetuates endometrial damage. Additional studies documented modulation of MAPK/p38 and PI3K/Akt signaling pathways, with phosphorylation of these kinases detected in endometrial tissues following MSC treatment.

Zhang et al.29 conducted comprehensive miRNA-mRNA expression profiling, identifying 45 miRNAs that were downregulated in injured endometrium and upregulated after UC-MSC treatment, and 39 miRNAs showing the opposite pattern. Integrated miRNA-mRNA network analysis revealed changes in extracellular matrix composition and organization, with gene ontology analyses showing enrichment in ECM-receptor interaction, collagen fibril organization, and tissue remodeling pathways. Key molecules in ECM-receptor interactions were validated by qRT-PCR, showing that Itga1 and Thbs expression decreased in model groups but increased with UC-MSC treatment, while laminin and collagen expression increased in both model and treatment groups, with greater expression in the latter, suggesting coordinated matrix reconstruction rather than simple fibrotic deposition.

Zhou et al.26 investigated IL-1β-mediated mechanisms, demonstrating that Pluronic F-127-encapsulated HUMSCs promoted local IL-1β secretion (3.4-fold increase versus naked cells) that stimulated endothelial cell proliferation and angiogenic sprouting. IL-1β neutralizing antibodies significantly reduced therapeutic benefits, including microvessel density and endometrial thickness, confirming mechanistic causality. The study demonstrated dose-dependent effects, with 1 × 10⁶ MSCs achieving optimal outcomes while higher doses (2 × 10⁶ cells) showed no additional benefit and potentially triggered excessive inflammation, highlighting the importance of dose optimization.

Delivery Method Optimization and Cell Retention

Three studies provided direct comparative data on delivery strategies. Zhang et al.25 compared hydrogel-encapsulated HUMSC-derived exosomes versus direct exosome injection, demonstrating significantly prolonged retention with hydrogel delivery: exosomes remained detectable at 14 days with hydrogel encapsulation versus 3 days with direct injection (p = 0.008), representing a 4.7-fold extension in retention time. This prolonged retention correlated with enhanced therapeutic outcomes, as hydrogel-treated animals showed a 23% greater increase in endometrial thickness compared to direct injection (p = 0.031). Fluorescence tracking of PKH26-labeled exosomes confirmed 3.2-fold higher fluorescence intensity in hydrogel groups at day 14 (p < 0.01).

Guo et al.27 performed a head-to-head comparison of intrauterine (local) versus intra-arterial (systemic via iliac artery) delivery of GFP/luciferase-labeled BMSCs. While both routes achieved comparable endometrial thickness restoration (intrauterine: +1.72 mm; intra-arterial: +1.88 mm; p = 0.24) and fibrosis reduction, intra-arterial delivery demonstrated superior cell retention as assessed by bioluminescence imaging (cells detectable at 28 days versus 14 days for local injection, p = 0.012). Histological validation with GFP immunofluorescence showed significantly higher GFP-positive cell numbers in intra-arterial groups at day 28 (8.3 ± 2.1 vs. 2.1 ± 0.8 cells per HPF, p = 0.003). Intra-arterial delivery also resulted in enhanced expression of regenerative factors, including LIF (1.8-fold increase, p = 0.008) and VEGF (1.6-fold increase, p = 0.015), with more uniform tissue distribution and reduced measurement variability. However, biodistribution analysis revealed that 13% of intra-arterially treated animals exhibited cell signals in the hindlimb rather than the uterus, attributed to anatomical variations in iliac artery branching, highlighting technical challenges requiring careful consideration of vascular anatomy.

Hong et al.22 compared scaffold-based delivery using 3D-bioprinted collagen patches versus direct cell injection, demonstrating that ADSC-loaded patches achieved superior outcomes compared to cell suspension injection. The collagen patch platform provided structural support for cell engraftment, with scanning electron microscopy showing MSC adhesion and proliferation within the porous scaffold structure (pore size: 110 μm). Cell viability exceeded 90% at day 1 post-seeding, with cells maintaining characteristic morphology and stemness marker expression. Histological analysis revealed that scaffold-supported MSCs showed better integration into the regenerating endometrium with organized stromal structure, whereas direct injection resulted in focal cell clustering with less uniform tissue incorporation. The biomaterial platform appeared to facilitate MSC differentiation toward endometrial lineages while providing sustained release of regenerative factors over 2–3 weeks, as evidenced by progressive therapeutic effects between the 2-week and 4-week assessments.

Cell tracking studies across multiple investigations confirmed that most transplanted MSCs disappeared from the endometrium within 7–14 days post-treatment, yet therapeutic effects persisted through 28 days and beyond, supporting paracrine-mediated mechanisms rather than direct cellular engraftment as the primary therapeutic mode. The observation that exosome-based treatments achieved comparable or superior outcomes to cellular transplantation further supports this mechanistic interpretation, while offering practical advantages, including avoidance of immune rejection, enhanced tissue penetration due to nanoscale size, and simplified manufacturing and quality control processes.