Efficacy and Safety of Umbilical Cord-Derived Mesenchymal Stromal Cell Therapy in Preclinical Models of Sepsis: A Systematic Review and Meta-analysis

Abstract Background In preclinical studies, mesenchymal stromal cells (MSCs), including umbilical cord-derived MSCs (UC-MSCs), demonstrate the ability to modulate numerous pathophysiological processes related to sepsis; however, a systematic synthesis of the literature is needed to assess the efficacy of UC-MSCs for treating sepsis. Objective To examine the effects of UC-MSCs on overall mortality (primary outcome) as well as on organ dysfunction, coagulopathy, endothelial permeability, pathogen clearance, and systemic inflammation (secondary outcomes) at prespecified time intervals in preclinical models of sepsis. Methods A systematic search was conducted on Embase, Ovid MEDLINE, and Web of Science up to June 20, 2023. Preclinical controlled studies using in vivo sepsis models with systemic UC-MSC administration were included. Meta-analyses were conducted and expressed as odds ratios (OR) and ratios of the weighted means with 95% CI for categorical and continuous data, respectively. Risk of bias was assessed with the SYRCLE tool. Results Twenty-six studies (34 experiments, n = 1258 animals) were included in this review. Overall mortality was significantly reduced with UC-MSC treatment as compared to controls (OR: 0.26, 95% CI: 0.18-0.36). At various prespecified time intervals, UC-MSCs reduced surrogate measures of organ dysfunction related to the kidney, liver, and lung; reduced coagulopathy and endothelial permeability; and enhanced pathogen clearance from multiple sites. UC-MSCs also modulated systemic inflammatory mediators. No studies were rated as low risk across all SYCLE domains. Conclusions These results demonstrate the efficacy of UC-MSC treatment in preclinical sepsis models and highlight their potential as a therapeutic intervention for septic shock.


Introduction
][4][5] Despite decades of research focused on immunotherapies and modulation of underlying biological drivers of disease, supportive care remains the mainstay of therapy for sepsis. 6,7As septic shock is associated with a high mortality rate of 20%-40%, identifying effective therapeutics remains an important area of research. 8,9esenchymal stromal cells (MSCs) are multipotent stem cells found in tissues such as bone marrow, adipose tissue, and umbilical cord. 10Recent preclinical sepsis studies suggest that MSCs represent a potential novel therapeutic for the treatment of sepsis.MSCs have been found to modulate the inflammatory response, 3,11,12 mitigate coagulopathy, and increase endothelial permeability largely through paracrine actions. 13,14MSCs have also demonstrated the ability to enhance pathogen clearance, reduce death, and improve surrogate measures of organ failure in preclinical studies. 15,168][19] Despite these in vitro differences, systematic reviews of animal models of sepsis that compared the in vivo efficacy of bone marrow-derived MSCs (BM-MSCs) 20 and all MSC types 21,22 to controls demonstrated reductions in death.Furthermore, subgroup analyses from the latter reviews suggested similar reductions in mortality for different MSC types (ie, adipose MSCs, BM-MSCs, umbilical cord-derived MSCs [UC-MSCs], IL-10-enhanced UC-MSCs, and menstrual fluid MSCs) as compared to controls. 21,22Although the reductions in death in animal models of sepsis appear similar for different MSC types, UC-MSCs have important advantages which include ready availability from donors, plentifulness, ease of procurement, and a noninvasive procedure to harvest. 23,24UC-MSCs also have the added advantage of being derived from very young tissue, with greater proliferative potential compared to cells from adult sources, making them a viable "off-theshelf" product and a potentially less expensive therapeutic for sepsis. 24Due to these advantages, our team is planning a phase II randomized controlled trial (RCT) comparing UC-MSCs to placebo for septic shock (Umbilical cord cellular immunotherapy for septic shock: a phase II RCT (UC-CISS II), NCT05969275).
In preparation for the UC-CISS II RCT, our team evaluated the effects of UC-MSCs, as compared to controls, in preclinical models of sepsis as no previous systematic review has reported exclusively on UC-MSCs.Furthermore, no systematic review has summarized the effects of UC-MSCs on surrogate measures of organ dysfunction, or other central aspects of sepsis pathobiology including coagulopathy, endothelial permeability, pathogen clearance, and systemic inflammation in preclinical models of sepsis.The evidence generated from this systematic review will provide hypotheses related to the biological effects of UC-MSCs, as well as the consideration of potential outcome measures for the UC-CISS II RCT and for other UC-MSC trials in sepsis going forward into the future.

Materials and Methods
The protocol for this review was posted on the University of Ottawa's Open Access Research Institutional Repository (http://hdl.handle.net/10393/43645).The approach for this systematic review and meta-analysis was adapted from a previously published systematic review by our group. 21

Eligibility Criteria
We included all preclinical controlled studies that used in vivo models of systemic sepsis, as mimicked through systemic infection (eg, cecal ligation and puncture [CLP], bacteria administration, and fecal slurry administration) or systemic endotoxin administration (eg, lipopolysaccharides [LPS] alone or with d-galactosamine [D-gal]), along with the systemic administration of UC-MSCs within 24 hours after sepsis induction.We excluded studies that used neonatal animal models of sepsis and models of direct acute lung injury.Studies that used engineered or modified MSCs (eg, altered the expression of a particular gene), multiple doses of MSCs, and co-treatments with other therapies (excluding antibiotics) or cell types were also excluded.

Intervention and Control Groups
The intervention group is defined as animals treated with UC-MSCs systemically following the induction of sepsis.The comparator or control group is defined as animals treated with vehicle or other controls (eg, phosphate-buffered saline, normal saline, fibroblasts, hydroxyethyl starch, Dulbecco's Modified Eagle Medium, and no treatment) following the induction of sepsis.

Literature Search
We conducted electronic literature searches on Embase, Ovid MEDLINE, and Web of Science using search strategies developed in conjunction with an information specialist to identify relevant studies published up to June 20, 2023.The full search strategy can be found in Supplementary Appendix.Additional searches were conducted by reviewing the bibliographies of eligible studies, reviews, abstracts, and relevant conference proceedings.

Study Screening and Selection
All studies were independently screened by 2 reviewers (CH and UT) on Covidence to determine if they met the inclusion requirements.Discrepancies were resolved through discussion with a third team member (LM).

Primary Outcome
The primary outcome was overall mortality measured at the latest time point reported.The overall mortality results were further analyzed according to subgroups determined a priori.These included animal species (eg, mice, rat, pig, and immunocompromised mice), sepsis model (eg, CLP, LPS, LPS with D-gal, live bacteria, and fecal slurry injection), resuscitation (eg, fluid alone, antibiotics, fluid and antibiotics, and no resuscitation), dose of MSC (eg, ≤500 000 cells, 500 000 to 1 × 10 6 cells, >1 × 10 6 cells), timing of MSC administration (eg, ≤1 hour, 1-6 hours, ≥6 hours [up to 24 hours] post sepsis induction), MSC preparation (eg, fresh or frozen), and the MSC administration route (eg, intravenous or intraperitoneal).Mortality outcomes were also measured for a priori determined time intervals of within 2 days, 2-4 days, and greater than 4 days, as per a previous review. 21

Data Extraction
Information on the animal model (eg, species, strain, age, weight, and sex), sepsis model, sample size, resuscitation, MSC characteristics (eg, origin, source, condition, and dose), control group, timing of UC-MSC delivery and administration route, along with the primary and secondary outcomes of interest, were extracted by 2 independent reviewers (CH and UT) using a predesigned standardized template.An experiment must have an intervention group and a corresponding control group for comparison.All data presented in graphical format were extracted using ImageJ (https://imagej.nih.gov/ij/).Discrepancies in the extracted data were resolved by a third team member (LM).Authors were contacted when further clarification of the data was required.

Risk of Bias Assessment
The SYRCLE risk of bias (ROB) tool for preclinical animal studies was used to evaluate the methodological quality of the included studies and assess elements for "high," "low," or "unclear" risk. 25Specific domains were identified a priori and evaluated independently by 2 reviewers (CH and CD) with discrepancies resolved through discussion with a third team member (LM).The 10 domains evaluated include (1) sequence generation, (2) baseline characteristics, (3) allocation concealment, (4) random housing, (5) blinding of personnel, (6) random outcome assessment, (7) blinding of outcome assessors, (8) completeness of outcome data, (9) selective outcome reporting, and (10) other sources of bias (eg, conflicts of interest, funding sources, a priori sample size calculations, etc.).

Statistical Analysis
Mortality results were pooled using a random effects model (Comprehensive Meta-Analysis Version 3.3070) to express results as odds ratios (OR) and 95% CI.Results from outcomes with continuous data (secondary outcomes) were pooled using a random effects model; these outcomes were expressed as the ratio of the weighted means (ROM) and 95% CI.A random effects model was used because it takes into account both within-study and between-study variation and the CI around the point estimate is often wider compared to if a fixed-effects model were used. 26An OR or ROM of <1 favored the UC-MSC group as compared to the control group.To reduce measurement bias and in accordance with our protocol, time of sepsis induction was the anchor used for the description of all outcome measures in accordance with the prespecified time intervals.Heterogeneity of the results was also assessed using the I 2 statistic and interpreted according to commonly adopted thresholds derived from Cochrane systematic review methodologies: 0%-40% may not be important, 30%-60% moderate heterogeneity, 50%-90% substantial heterogeneity, and 75%-100% considerable heterogeneity. 27The presence of publication bias was assessed through visual inspection of a funnel plot.Subsequent trim and fill analysis was used to estimate the number of missing studies and provide an estimated UC-MSC effect size "adjusted" for the publication bias. 28

Search Results and Study Characteristics
Our systematic search strategy yielded 7079 potential studies.After removing duplicates and applying our eligibility criteria, 26 studies comprised of 34 experiments (n = 1258 animals) were included in this review  (Fig. 1). Complte reporting of the included studies can be found in Table 1.A summary of the study characteristics (including the animal type, sepsis model, resuscitation, MSC timing, MSC preparation, MSC dose, and MSC administration route) can be found in Supplementary Table S1.

Cardiac
Cardiac dysfunction was evaluated through measures of ejection fraction (at 7-24 hours 29 post sepsis induction) and arterial lactate (at 0-6 37 and 7-24 hours 37 post sepsis induction; Table 3; Supplementary Fig. S6).No statistically significant change in ejection fraction or arterial lactate was observed with UC-MSC treatment as compared to controls.b Some studies used MSCs from multiple sources however we only report on the UC-MSCs in this study.c UC-MSC preparation was described as fresh or enough detail was provided to assume freshly cultured.d UC-MSC preparation was unclear and not enough detail was presented to assume fresh or frozen.e This study had 2 control groups whose results were combined for analysis.f This experiment had multiple UC-MSC groups whose results were combined for analysis.g In this experiment, antibiotics were given 24 hours after sepsis induction.h In this experiment, antibiotics were given 6 hours after sepsis induction.i Some studies were extracted as multiple experiments (as defined by having independent control and intervention groups for comparison) and were denoted with a letter (i.e.

Effects of UC-MSCs on Coagulopathy
Coagulopathy markers included measures of platelet concentrations (at 0-6 37 and 7-24 hours 37 post sepsis induction), fibrinogen (at 7-24 hours 33,37 post sepsis induction), PT (at 7-24 hours 33 post sepsis induction), and APTT (at 7-24 hours 33 post sepsis induction; Table 3; Supplementary Fig. S7).Treatment with UC-MSCs as compared to controls did not demonstrate a statistically significant increase in platelets or fibrinogen; however, both PT and APTT levels were significantly reduced at 7-24 hours post sepsis induction.
None of the 26 studies included in our systematic review reported harm individually.There were also no significant signals for harm found in the pooled analyses for our primary and secondary outcomes.

ROB Assessment
Of the 26 studies included in this review, none were rated as low risk for all 10 domains assessed (Table 4; Supplementary Table S2).Three domains, including sequence generation (randomization), allocation concealment, and random housing, were identified to have unclear ROB in all studies as insufficient detail was provided in the methods to classify  the level of risk.Four studies demonstrated a high ROB in one domain 38,45,50,52 while 3 studies demonstrated high risk in 2 domains 44,53,54 ; these 7 studies were all assessed to have a high ROB for selective outcome reporting due to the mortality and/or relevant secondary outcome being only presented in the results not prespecified in the methods.One study was found to have high ROB for incomplete outcome data as the sample values (n) were not consistent between methods and results for some of the secondary outcomes. 54One included study also reported a conflict of interest with one of the authors having served as an unpaid section editor of the journal in which the study was published. 44

Assessment for Publication Bias
Visual inspection of the funnel plot analysis of the overall mortality data suggests that publication bias may exist due to a lack of small to medium studies reporting negative or null results (Fig. 3).In our analysis, UC-MSC treatment compared to controls demonstrated a reduction in overall mortality (OR: 0.26, 95% CI: 0.18-0.36);after the trim and fill analysis, the adjusted effect size (OR: 0.38, 95% CI: 0.26-0.56)was more modest but still significantly reduced.

Discussion
Our comprehensive systematic review and meta-analysis summarizes the effects of UC-MSCs on overall mortality, surrogate measures of organ dysfunction, and key markers of sepsis pathobiology (coagulopathy, endothelial permeability, pathogen clearance, and systemic inflammation) in preclinical models of sepsis.Collectively, our review does suggest that UC-MSCs provide a benefit in preclinical sepsis without any significant risk for harm for any of the outcome measures.Our systematic review found that mortality was reduced in preclinical sepsis models with UC-MSC treatment, similar to what has been shown for all MSC types as reported by Lalu et al 21 (OR: 0.27, 95% CI: 0.18-0.40)and Sun et al 22 (OR: 0.29, 95% CI: 0.22-0.38).Prespecified subgroup analyses to examine the heterogeneity of the treatment effect found that UC-MSCs, as compared to controls, were associated with a reduction in overall mortality across different sepsis models (eg, CLP and LPS) and animal species (eg, mice and rats).In the one large animal (pig) model, mortality was lower in the MSC group as compared to the control group (2/6 [33%] vs 6/6 [100%]), but it did not reach statistical significance. 37Other potential sources of heterogeneity include the timing of MSC administration as well as the administration of fluid and antibiotic resuscitation.6][57] We observed that UC-MSC treatment reduced overall mortality even when MSC treatment was delayed to 6 hours post sepsis induction and when animals were resuscitated with fluids alone or in combination with antibiotics, thereby providing further evidence for the clinical applicability of UC-MSC therapy for sepsis.Additionally, overall mortality was significantly reduced for both freshly cultured and frozen UC-MSC preparations, supporting the use of frozen MSCs as an offthe-shelf product that would be necessary for acute treatment of time-sensitive syndromes such as sepsis.The early stages of sepsis are characterized by the excessive production of pro-inflammatory markers, which contributes to the later hallmarks of organ dysfunction and damage. 5Our systematic review demonstrated that UC-MSCs improved surrogate measures of renal, liver, and pulmonary dysfunction; reduced coagulopathy and endothelial permeability; and enhanced pathogen clearance in at least one of our prespecified time intervals.Our review also suggests that UC-MSC therapy may modulate several sepsis inflammatory mediators with some of these changes sustained over time. 58he preclinical efficacy signals of our secondary outcome measures should be considered hypothesis generating as there was substantial statistical heterogeneity (I 2 > 50) 59,60 for some of the outcomes, which may be due to small sample sizes, as well as varying animal species, sepsis models, MSC doses, and time of MSC administration.
Our preclinical systematic review has several strengths.These include a systematic and transparent search for all available studies; a protocol that is accessible in a public repository and describes the design, methods, outcomes, and approach to the analyses; and an examination of the effect of UC-MSCs on our primary outcome according to subgroups to explore the heterogeneity of the treatment effect.Despite these strengths, our study also has limitations.While we attempted to include relevant sepsis-related biological mediators as secondary outcomes, due to immense sepsis complexity, there may be other important mediators that we did not examine.Our funnel plot analysis suggested the presence of publication bias.However, after adjustment using the trim and fill analysis method for hypothetical missing studies not published, overall mortality remained significantly reduced.The time of UC-MSC administration and measurement of the outcomes did vary in these studies, although all studies administered MSCs within the first 6 hours of sepsis induction.To reduce measurement bias and in accordance with our a priori protocol, we used sepsis induction as the anchor for the description of secondary outcomes at our prespecified time intervals.Due to the number of comparative tests performed, results from our subgroup analyses and secondary outcomes should be interpreted with caution and be considered hypothesis generating.None of the included studies were graded as low risk across all SYRCLE ROB domains.This was often due to a lack of clear or incomplete methodological reporting, which also limits the strength of the interpretation of our study findings.Finally, the utility of preclinical models of sepsis to predict clinical efficacy is often questioned because they do not entirely replicate the pathobiology of clinical sepsis, the human population with sepsis (eg, the use of small, young animals without comorbidities) and the clinical treatment setting (eg, time of MSC administration and the use of co-interventions such as fluids and antibiotics). 61The included studies in our review have similar limitations (ie, the inclusion of young animals with no morbidities) with the exception of the administration of co-interventions.In an a priori subgroup analysis, the effect of UC-MSCs on mortality was similar when fluids and antibiotics were administered compared to when neither of these co-interventions was administered.Furthermore, while preclinical animal sepsis studies do not fully reproduce the complexity and heterogeneity of clinical sepsis, they do provide hypotheses for potential UC-MSC effects that can then be tested in the setting of clinical trials. 62

Conclusion
The results of our systematic review suggested that UC-MSC therapy, as compared to controls, reduced mortality in preclinical models of sepsis.UC-MSCs also improved surrogate measures of organ dysfunction (in the kidney, liver, and lung), coagulopathy, endothelial permeability, pathogen clearance, and systemic inflammation in at least one of the prespecified time intervals.The evidence generated from this systematic review has assisted the planning of outcome measures for our UC-CISS II RCT and we hope for other UC-MSC trials in sepsis going forward into the future.Our review also provides hypotheses related to the potential biological effects of UC-MSCs in sepsis (eg, effect on coagulopathy, endothelial permeability, pathogen clearance, and inflammation) that investigators may want to consider for measurement in future UC-MSC clinical trials in sepsis.

Figure 1 .
Figure 1.Preferred reporting items for systematic reviews and meta-analysis flow diagram for literature search and study inclusion.

Figure 2 .
Figure 2. Forest plot summarizing the effects of UC-MSC treatment on overall mortality in preclinical models of sepsis.Point estimates represent the OR for each individual study with the size of the point depicting the relative contribution to the pooled effect.The corresponding horizontal line represents the 95% CI.Meta-analysis was performed on the pooled results with random effects modeling and the overall effect is depicted with the diamond.

Figure 3 .
Figure 3. Funnel plot to detect publication bias.Trim and fill analysis was conducted on overall mortality results (the primary outcome).Open circles represent original study data and filled circles represent supplemented data through the trim and fill analysis.The open diamond represents the observed effect size and 95% CI.The filled diamond represents the adjusted effect size and 95% CI.

Table 1 .
Summary of the included preclinical studies.

Table 2 .
Overall mortality estimates of UC-MSC treatment according to prespecified subgroups.

Table 3 .
Secondary outcome estimates of UC-MSC treatment in preclinical models of sepsis.