Therapeutic Effect of Mesenchymal Stem Cells on Sjögren's Syndrome: Systematic Review and a Preclinical Meta-Analysis

Evidence to support Mesenchymal stem cells (MSCs) treatment in Sjögren's syndrome (SS) has been veried. This study aims to evaluate the effectiveness of heterogeneous MSCs therapies, identify optimal experimental parameters and explore possible underlying mechanisms in animal models of SS. Methods Literature searches were performed in PubMed, Web of Science and EMBASE. Effect sizes of SS treatments with MSCs were extracted and analyzed by two authors independently. Results A total of 13 studies and 20 treatment arms met the inclusion criteria. When compared with the controls, MSCs treatment resulted in lower level of histological score (SMD= -2.208; 95%CI= -3.129, -1.286; P<0.001) accompanied by an improved trend of salivary ow rate (SFR) (SMD = 1.726; 95%CI= 1.340, 2.113; P <0.001) and Schirmer's test results (SMD= 3.379; 95% CI= 2.141, 4.618; P<0.001). In MSCs groups, levels of autoantibodies decreased to varying degrees. Treg cells were increased and Th17 cells were decreased in both lymph nodes and spleens. Additionally, IL-6 reduction and IL-10 elevation were found in local lesional tissues. Furthermore, TNF-α level dropped either in sera or glands. Notably, the cell injection frequency and routes may be two important factors affecting the effect of MSCs therapy. the knowledge, this is the rst meta-analysis to quantitatively evaluate MSCs therapeutic effects on SS. Our research emphasizes optimizing MSC treatment strategies to achieve better outcomes, thereby providing a valuable reference for clinical application.

(OA) [19]. Positive results in clinical trials have also been reported in terms of e cacy [20][21][22], as well as safety [23]. However, the application strategies of these studies are varied, involving various donor species and tissue sources, distinct injection methods, different treatment doses and frequencies [24][25][26]. Therefore, it has become critical to develop standardized stem cell application protocol, to nd the most effective treatment strategy and to explore treatment mechanisms.
In this study, we reviewed the preclinical studies of MSCs in the treatment of Sjögren's syndrome. The objective was to evaluate the quality and robustness of existing evidence, assess the therapeutic effects of MSCs, and identify optimal treatment conditions. This meta-analysis aims to supply insights into the therapeutic use of MSCs in Sjögren's syndrome, which can provide a framework for design of future applications of this therapy.

Methods
This meta-analysis review was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance [27].

Literature Search
To evaluate the therapeutic effects of MSCs in Sjögren's syndrome, relevant papers published up to May 1st, 2020 were screened comprehensively in PubMed, Web of Science and EMBASE databases using the following search terms: ("mesenchymal stem cells" OR "mesenchymal" OR "mesenchymal stromal cells" OR "MSCs") AND ("Sjögren's syndrome" OR "Sjogrens syndrome" OR "Sjogren syndrome" OR "Sicca syndrome" OR "SS"). Two reviewers (Y.L and Y.X.C) inspected all candidate articles independently and discrepancies were resolved by discussion (Y.F.P and Y.L).

Inclusion and Exclusion Criteria
Studies meeting all the following criteria were included: (1) animal models of Sjögren's syndrome; (2) the administration of MSCs; (3) the inclusion of a placebo arm; and (4) outcome measure-the primary outcomes: histological score, salivary ow rate (SFR) and Schirmer's test results; the secondary outcomes: Anti-nuclear antibodies (ANA), Anti-SSA/Ro52, Anti-α-Fodrin, Treg and Th17 expression, and the levels of IL-6, IL-10, TNF-α, TGF-β in serum or in the affected organs.
The exclusion criteria were as follows: (1) non-English articles, (2) studies with high risk of bias in the SYRCLE's risk of bias tool (below). (3) treatment methods using a combination of MSCs with other agents, (4) repeated data, (5) studies lacking quantitative data, and (6) case report, editorial, letter to editors, commentary, correspondence, conference abstract, expert opinion, or other meta-analysis.

Validated Terms
The following general study information was extracted: rst author, publication year, included treatment arms, modeling methods, number of animals, MSCs type, donor species, recipient species, transplant type, and treatment protocol. Secondly, to evaluate the e cacy of MSCs treatment, the outcome data was extracted from all available sources in each paper, including text and graph. When only graphical presentations were available, values for mean and SD or SEM were obtained using Engauge Digitizer version 12.1 software, under high magni cation by two independent investigators.

Quality assessment
The quality of the studies and risk of any bias were evaluated by two independent authors used the SYRCLE's risk of bias tool, which includes 10-item checklist: Q1: Whether the allocation sequence was adequately performed? Q2: Whether the baselines in groups were similar or they were adjusted for confounding factors? Q3: Whether the allocation was adequately concealed? Q4: Whether the animals were randomly housed during the experiments? Q5: Whether animal breeders and researchers were blinded against performance bias? Q6: Whether the animals were selected at random during results evaluation? Q7: Whether the result assessors were blinded? Q8: Whether incomplete data were adequately reported? Q9: Whether the research report was free of selective results report? Q10: Whether there were other issues that could induce high-risk bias? "Yes" and "NO" represent low bias risk and high bias risk respectively. If insu cient details were supplied to assess the risk of bias properly, the judgment was recorded as "unclear" [28]. The general quality was determined by taking all items together for presentation in a risk bias graph.

Statistical analysis
All statistical analyses were conducted using STATA14.1. Continuous outcomes were expressed as the standardized mean difference (SMD) with the 95% CI. Heterogeneity was analyzed among studies using the I 2 statistic. I 2 > 50% indicated signi cant heterogeneity and a random effect model was conducted. Otherwise, a xed effect model was performed. Subgroup, sensitivity, and meta-regression analyses were performed to explore and explain the heterogeneity and other potentially confounding factors among the results of different studies. Egger's linear regression method was used to detect publication bias. A P value < 0.05 was considered statistically signi cant.

Study characteristics
Literature searches retrieved 1573 articles from PubMed, EMBASE and Web of Science, in which 472 were duplicates, 1061 were excluded by title screened, 18 were excluded as conference abstract, review or in vitro study, 9 full-text did not meet the criteria, leaving 13 articles quali ed to be included in the meta-analysis review [29][30][31][32][33][34][35][36][37][38][39][40][41]. The literature selection process is summarized in In this meta-analysis, 13 studies with 20 treatment arms were included, of which 14 treatment arms used NOD mice for spontaneous Sjögren's syndrome models, 4 treatment arms used rabbits with active lymphocytes-induced models, 1 treatment arm used rats after X-ray irradiation, and 1 treatment arm used ovariectomized rats. The donor species of MSCs varied by studies, including human (11 treatment arms), mouse (7 treatment arms) and rabbit (2 treatment arms). Considering tissue source, 7 treatment arms were with bone marrow derived MSCs (BM-MSCs), 7 treatment arms with umbilical cord derived MSCs (UC-MSCs), 2 with adipose derived MSCs (AD-MSCs) and 4 were from other sources, such as 1 with induced pluripotent stem cells (iPSCs), 1 with human salivary gland stem cells (hSGSCs) and 2 with stem cells from human exfoliated deciduous teeth (SHEDs). The administration routes were also compared, with 55% of 20 treatment arms being xenogeneic and 45% were allogeneic. Transplant routes included 16 intravenous (IV), 3 intraperitoneal (IP) and 1 intra-glandular injection. In terms of injection number, 65% treatment arms used multiple treatments and 35% were with a single injection. The characteristics of the recruited studies are presented in Table 1.

Quality of included studies
The quality of the included studies was assessed with the SYRCLE scale. No studies were ranked as high risk of bias. 85% studies reported the baseline characteristics. Low risks of incomplete data bias and other sources bias were judged in the most studies. However, few studies attempted to report the strategies to handle performance bias and selection bias. A summary of the bias risks is displayed in Supplementary Fig. 1.

Primary outcome
Three primary indicators were used to evaluate effects in this meta-analysis review: (1) histological score, (2) salivary ow rate (SFR), and (3)  To explore the cause of heterogeneity, sensitivity analysis was performed. None of the single studies signi cantly in uenced the result (Supplemental Fig. 2A). Thus, we conducted further subgroup analysis according to the donor species, tissues of origin, transplant types, administration routes and injection frequency of MSCs in SS treatment (Table 2). Notably, all MSCs from various donor species showed consistent improvement of histological score, with slightly lower levels in the human-derived group (SMD= -2.176; 95% CI= -3.258, -1.093; P < 0.001) than the mouse-derived one (SMD= -1.925; 95% CI= -3.467, -0.382; P = 0.014). Rabbit-derived MSCs showed the most pronounced inhibitory effects (SMD= -6.124; 95% CI= -10.660, -1.589; P = 0.008), but the fact that this was a single study resulted in less reliability. Moreover, MSCs from different tissues of origin including BM-MSCs, UC-MSCs, AD-MSCs, iPSC-MSCs and SHEDs all demonstrated effectiveness to some degree. AD-MSCs and iPSC-MSCs provided more obvious therapeutic effects than the overall e cacy, but with just one study each. Xenogeneic MSCs had no signi cant difference with allogeneic ones in SS treatment. Interestingly, IV injection resulted in more profound and stable e cacy (SMD= -2.290; 95% CI= -3.216, -1.363; P < 0.001) as measured by the decrease of histological score, while IP injection (SMD= -1.769; 95% CI= -4.092, 0.553; P = 0.135) did not show statistically signi cant effect. When comparing the injection frequency, multiple injection showed signi cant advantages (SMD= -2.876; 95% CI= -3.799, -1.953; P < 0.001), which is indicated by the nonoverlapping con dence intervals of histological score reduction with the single injection (SMD= -0.723; 95% CI=-1.699, 0.253; P = 0.146). Single injection failed to show statistically signi cant e cacy for histological score improvement with 4 studies in the pool. The meta-regression analysis showed injection frequency was an independent and in uential factor contributing to histological score reduction (P = 0.02), explaining 63.36% of the heterogeneity and further con rming the subgroup analysis results. Variations between them were relatively small (I 2 = 39.5%, P = 0.07) (Fig. 2B). Stability of these results was further con rmed by the sensitivity analysis (Supplemental Fig. 2B) (Table 3). Meta-regression analysis did not nd valid covariates, being consistent with low heterogeneity.  (Fig. 2C). The sensitivity analysis found no single study interfered with the results (Supplemental Fig. 2C)

Effects of MSCs on Treg and Th17
Nine experimental arms were identi ed in this meta-analysis that examined the effect of MSCs on Treg and Th17 in Sjögren's syndrome models. Overall, MSCs increased Treg and reduced Th17 cells frequency. Intriguingly, the Foxp3 levels of MSCs group in spleen or lymph node were elevated consistently with little heterogeneity (SMD = 1.928, 95% CI = 1.067, 2.789, P < 0.001), while the differences in lesional tissue were not uni ed among the pooled studies (SMD= -2.408, 95% CI= -6.686, 1.869, P = 0.270). In contrast, the expression of IL-17 in MSCs group showed a decreasing trend, whether in lesions (SMD= -6.414, 95% CI= -18.639, 5.812, P = 0.304) or peripheral immune organs (SMD= -2.054, 95% CI= -2.985, -1.123, P < 0.001). Unlike to the effect on Foxp3 expression, the IL-17 levels in spleen or lymph node had speci cally consistent and robust reductions that were statistically signi cant (Fig. 4). These observations are in line with the immunoregulatory role of MSC on IL-17 in other autoimmune diseases [9,15,42].

Effects of MSCs on cytokines
In ammatory cytokines play a crucial role in the pathogenesis of autoimmune diseases [43,44]. We therefore also analyzed the

Publication bias
A linear regression plot generated for the primary outcomes with Egger's test suggested that there was somewhat of a publication bias in these studies (Histological score P < 0.001; Salivary ow rate P = 0.003; Schirmer's test results P = 0.13) (Supplemental Fig. 3).

Discussion
This meta-analysis of 13 studies and 20 treatment arms provided evidence of therapeutic bene t of MSCs on preclinical Sjögren's syndrome models. To our knowledge, this is the rst meta-analysis providing a comprehensive summary of the MSCs effect in SS, involving different donor species, tissues and treatment protocols. Furthermore, this work makes it possible to explain potential mechanisms of therapy by pooling large data samples. Our meta-analysis may provide important clues for the future clinical translation of MSCs in the treatment of patients with Sjögren's syndrome.
Overall, MSCs therapy shows de nite improvements in histological score, SFR and lacrimal gland secretion function. Speci c effect size varies depending on the MSCs donor species, tissues of origins, transplant types, administration routes or injection frequency. The three main outcome indicators that had similar results suggest that multiple injections are signi cantly better than single injection, and the IV route is superior to IP. Considering tissue origins, AD-MSCs have shown more favorable histology scores and lacrimal gland function, compared to the UC derived MSCs, while BM-MSCs provided weaker effects than these two.
Of note, in all studies that are included in the analysis, AD-MSCs are from rabbits, UC-MSCs came from human and BM-MSCs are from murine tissue. In order to distinguish whether tissue source or donor species play a dominate role in the observed differences, it would be prudent to explore the optimal combination of derived species and tissues. In addition, other data have drawn different conclusions in other diseases: UC-MSCs and G-MSCs (gingival tissue-derived MSCs) were more suitable sources for the RA treatment than other options [12,45,46]; IP injection of AD-MSC achieved more effective outcome on EAE than IV administration [47]. Therefore, the best MSCs treatment strategy may vary in different autoimmune diseases.
Notably, although derived from homologous organ tissues, local injection of SHEDs did not show signi cant effects compared with other tissue-derived MSCs or by other administration routes. Of the administration routes. This observation may imply that although tissue homology and local injection may enhance the targeting of MSCs, it could not achieve better regulation of systemic immunity. Most superior performances of targeted therapies are preferentially reported in tissue regeneration studies [48,49].
We summarized the comprehensive analysis of autoantibodies, immune cells and cytokines. Although only 5 of the 13 studies were included for this portion of the analysis, the results of autoantibodies showed a consistent decrease with active treatments. Indeed, targeting autoantibody has been associated with favorable outcome in the treatment of autoimmune diseases [50,51]. In terms of immune cells, Th17 reduction accompanied by Treg increase appeared in peripheral immune organs, but not uniformly in salivary or lacrimal glands. In contrast, in ammatory cytokines analysis presented signi cant difference in the local milieu (increased level of IL-10 and decreased trend of IL-6) after MSCs injection, while heterogeneity in serum was high. Our metaanalysis indicates that MSCs might control the local in ammatory response in lesions by regulating immune cell differentiation in peripheral immune organs but not by acting directly in the exocrine glands. Several reports have mentioned that MSCs will preferentially reach the lungs, liver, and spleen after infusion in vivo [52,53], whereas some researchers illustrated that viable MSCs could not pass through the lung [54]. There is no consistent report of MSCs targeting glandular lesions in our included literature [29,30], which further suggests that the mechanism involves indirect regulation of MSCs in exocrine glands. Overall, biodistribution of different MSCs in autoimmune disease models should be taken into account in the future, which may explain the similarities and differences in therapeutic mechanisms.
TNF-α plays an important in ammatory role in SS pathogenic process [55,56]. TNF-α neutralizing antibodies induce a remission of the SS clinical phenotype in animal models [57], but TNF-α inhibitors have not been shown to be effective in human studies [58,59]. It is likely that TNF-α exerts its differential effects based on the speci c type of TNF-α receptor engaged [44,60,61]. In addition, TNF-α also promotes the functional capacity of MSCs [62]. In our meta-analysis, MSCs treatment was associated with reduced TNF-α production in both glands and in serum. Considering that MSCs may regulate in ammatory and immunity more upstream, they would be a better choice for clinical application. Different from Graft-versus-Host Disease (GVHD) [63] and rheumatoid arthritis (RA) [64], our analysis did not suggest robust and consistent trend for changes in levels of TGFβ. As TGFβ abnormalities are clearly recognized as involved in the pathogenesis of SS, the results were more likely due to the heterogeneous effect of different treatment strategies [65][66][67].
Sjögren's syndrome is a relatively common autoimmune disease that mainly affects the exocrine glands. It is often secondary to another autoimmune diseases, such as RA or systemic lupus erythematosus (SLE); which results in signi cant heterogeneity. Although traditional and biological DMARDs are used empirically for Sjogren's symptoms, their e cacy is limited in modulating the systemic process [68]. Over the past few decades, cell therapy has emerged as a promising therapeutic strategy for treatment of autoimmune and other diseases. Optimization of the parameters in this therapeutic approach, such as the origin of MSCs, easiness of cell preparation, cost, functional stability, treatment protocol, as well as the disease clinical stage, merits further efforts to explore clinical applications in the future [69][70][71].
Limitations Quality assessment of studies was performed in this meta-analysis to investigate their credibility. Even though most of the studies were regarded as good quality with low or unclear risks of bias, few of them attempted to elaborate their strategies to prevent the potential performance bias or detection bias. Therefore, the results from this meta-analysis should be interpreted with caution. Moreover, the MSCs-based Sjögren's syndrome treatment would bene t from design modi cations in the future. In addition, the number of included trials was small, with several types of MSCs, different injection methods and unequal therapeutic doses involved. Although, we performed subgroup and sensitivity analyses, the above factors may weaken the reliability of the observations. Finally, there was a potential for the incomplete reports of identi ed research studies, which could have introduced publication bias.

Conclusions
Based on the published evidence, our meta-analysis is the rst to systematically evaluate the therapeutic effects of MSCs on Sjögren's syndrome in preclinical models. The analysis indicates that MSCs have signi cant potential to effectively improve the secretory function of the salivary and lacrimal glands and reduce in ammatory cell in ltration into exocrine glands. Overall, tissue homology and local injection of MSC did not present therapeutic advantages, multiple injections and use of the IV injection route displayed better performance. We also found that the therapeutic mechanism of MSCs might involve inhibiting the production of antibodies in serum and in ammatory cytokines in lesions, as well as balancing the Treg/Th17 ratio in peripheral immune organs. These preclinical results provide useful basis for future studies of MSCs treatment in patients with Sjögren's syndrome.

Declarations Acknowledgment
We thank all the authors of those original papers for their useful data.
Author contributions Y.F.P and Y.L had the initiative for drafting this article. Y.L and Y.X.C were study investigators and contributed to the process of data collection. The data layout, meta-analysis and writing were carried out by Y.L. The overall proof reading was achieved by both N.O and W.J. The authors read and approved the nal manuscript.
Funding Y.L and Y.F.P were in part supported from the National Natural Science Foundation of China (81771750).

Availability of data and materials
All supporting data are included in the article and additional le.
Ethics approval and consent to participate Not applicable Consent for publication