Mesenchymal stromal cells as prophylaxis for graft-versus-host disease in haplo-identical hematopoietic stem cell transplantation recipients with severe aplastic anemia?—a systematic review and meta-analysis

Background Mesenchymal stromal cells (MSCs) are an emerging prophylaxis option for graft-versus-host disease (GVHD) in haplo-identical hematopoietic stem cell transplantation (haplo-HSCT) recipients with severe aplastic anemia (SAA), but studies have reported inconsistent results. This systematic review and meta-analysis evaluates the efficacy of MSCs as prophylaxis for GVHD in SAA patients with haplo-HSCT. Methods Studies were retrieved from PubMed, EMBASE, Cochrane, Web of Science, and http://clinicaltrials.gov from establishment to February 2020. Twenty-nine single-arm studies (n = 1456) were included, in which eight (n = 241) studies combined with MSCs and eleven (n = 1215) reports without MSCs in haplo-HSCT for SAA patients. The primary outcomes were the incidences of GVHD. Other outcomes included 2-year overall survival (OS) and the incidence of cytomegalovirus (CMV) infection. Odds ratios (ORs) were calculated to compare the results pooled through random or fixed effects models. Results Between MSCs and no MSCs groups, no significant differences were found in the pooled incidences of acute GVHD (56.0%, 95% CI 48.6–63.5% vs. 47.2%, 95% CI 29.0–65.4%; OR 1.43, 95% CI 0.91–2.25; p = 0.123), grade II–IV acute GVHD (29.8%, 95% CI 24.1–35.5% vs. 30.6%, 95% CI 26.6–34.6%; OR 0.97, 95% CI 0.70–1.32; p = 0.889), and chronic GVHD (25.4%, 95% CI 19.8–31.0% vs. 30.0%, 95% CI 23.3–36.6%; OR 0.79, 95% CI 0.56–1.11; p = 0.187). Furtherly, there was no obvious difference in 2-year OS (OR 0.98, 95% CI 0.60–1.61; p = 1.000) and incidence of CMV infection (OR 0.61, 95% CI 0.40–1.92; p = 0.018). Conclusions Our meta-analysis indicates that the prophylactic use of MSC co-transplantation is not an effective option for SAA patients undergoing haplo-HSCT. Hence, the general co-transplantation of MSCs for SAA haplo-HSCT recipients may lack evidence-based practice.


Introduction
Severe aplastic anemia (SAA) is a life-threatening bone marrow failure syndrome characterized by pancytopenia and hypoplastic bone marrow. Hematopoietic stem cell transplantation (HSCT) has been considered as a first-line therapy for young adults [1]. However, only 20-30% acquired SAA patients realistically hope to find a human leukocyte antigen (HLA)-matched sibling donor. With the improvement of conditioning regimens, like "Beijing protocol," haplo-identical HSCT (haplo-HSCT) has recently been widely used to treat SAA patients as an alternative strategy [2]. However, the main challenges facing current haplo-HSCT usage included the risk of graft-versus-host disease (GVHD) and a higher graft failure (GF) rate [3][4][5]. Therefore, improving the haplo-HSCT outcomes in SAA patients is of great concern.
Mesenchymal stromal cells (MSCs) are multipotent stromal cells characterized by modulating immune and inflammation response, supporting hematopoiesis, and repairing tissues, which are widely used in haplo-HSCT [6,7]. MSCs can be isolated from many tissues, including bone marrow (BM), cord blood and umbilical cord (UC) tissues, periosteum, adipose tissue, and fetal liver [8,9]. According to several previous clinical studies, the application of MSCs in haplo-HSCT can decrease the incidence and severity of acute or chronic GVHD, promote facilitation of HSC engraftment, and improve OS [10][11][12][13][14][15]. However, others found that MSCs may make little or no difference in reducing the risk of GVHD and death [16][17][18]. Thus, these conflicting results need to be addressed urgently [19].
To the best of our knowledge, there have been some excellent clinical studies about proposing haplo-HSCT as the first-line therapy for SAA patients [20][21][22]. Therefore, it is of great importance to clarify key factors related to the outcomes of SAA with haplo-HSCT. For example, some meta-analyses compared different donor sources in haplo-HSCT, evaluating whether peripheral blood (PB) or BM as graft source produces a more satisfactory outcome in SAA patients [23], while others sought the optimal conditioning regimen for haplo-HSCT in patients with SAA [24][25][26]. In addition, several meta-analyses have approved the efficacy of MSCs in haplo-HSCT recipients with hematological conditions, mostly in hematological malignancies [27,28]. Nevertheless, no meta-analysis has been done to evaluate the efficacy of MSCs combined with haplo-HSCT in SAA patients so far. Therefore, we performed the first systematic review and meta-analysis to investigate the efficacy of MSC co-transplantation following HSCT in patients with SAA.

Literature search
This meta-analysis was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines issued in 2009 [29]. We performed a systematic literature search in PubMed, EMBASE, OVID, Web of Science, and Cochrane Central Register of Controlled Trials (CENT RAL) from inception to January 2020 with the search terms "haplo-identical hematopoietic stem cell transplantation," "mesenchymal stem cells," and "severe aplastic anemia." In addition, we searched clinical trials in http://clinicaltrials.gov with "severe aplastic anemia [condition/disease] AND (haplo-identical hematopoietic stem cell transplantation [other terms])."

Study selection
Articles in the literature were identified and data were extracted by two investigators independently. Disagreements were resolved through discussion. The reference lists of relevant studies were also hand-searched. These searches were limited to the "first generation" reference lists. We removed duplicates and reviewed titles or abstracts. Studies that met the following criteria were included: (1) phase 2 or 3 clinical trials or retrospective studies evaluating the efficacy of MSC co-transplantation following HSCT in patients with SAA, (2) cases with > 5 patients, (3) studies with consistent criteria of observation items, and (4) studies reported a quantitative outcome of interest. Exclusion criteria were the following: (1) review papers or expert opinions, (2) individual case reports, (3) studies did not report a quantitative outcome of interest, (and 4) studies were reported in a language other than English. Meta-analyses do not involve human subjects and do not require Institutional Review Board review.

Data extraction
Data extraction from the eligible studies was carried out independently by 2 authors. We used a standardized extraction form to extract information about the first author, year of publication, study design, number of patients, median age, median intervals from diagnosis to treatment, the prophylaxis of GVHD, the conditioning regimen, acute GVHD (aGVHD), grade II-IV aGVHD, chronic GVHD (cGVHD), engraftment rate, all-cause mortality rate, and cytomegalovirus (CMV) infection rate. Because there are differences in study design, we extracted information in accord with the following criteria: (1) if data were used in two or more studies, data from the latter study were extracted on the basis of subtracting data published in the formal study; (2) we also summarized hematologic reconstitution time without performing a meta-analysis.

Quality assessment
Two authors worked independently on quality assessment. If disagreements occurred, an adjudicator was consulted. Single-arm studies were assessed using the Newcastle-Ottawa Scale modified for cohort studies without controls, as previously used by Lopez-Olivo [30]. Potential scores ranged from 0 to 6, with higher scores indicating higher quality. The following components were assessed: selection, which includes the representativeness of the exposed cohort; ascertainment of exposure; demonstration that the outcome of interest was not present at the start of the study; and outcome, which consists of an assessment of outcome, followed up long enough for outcomes to occur, and adequacy of the followup of cohorts. Because it was a meta-analysis of single proportions, publication bias was not advisable in this study.

Statistical analysis
Data manipulation and statistical analyses were performed using Stata statistical software version 15.0 (StataCorp, College Station, TX, USA) and R software (version 3.6.3). We conducted separate analyses for single-arm retrospective researches studying the treatment outcomes of haplo-HSCT with MSCs or without MSCs. Heterogeneity between studies was evaluated via I-squared statistic and p value. When heterogeneity was significant (p < 0.05 or I-squared > 50%), a randomeffects model was adopted to pool the results. Then, x 2 tests were applied to find if there are statistical differences of pooled estimates between groups, the effect measure was the adjusted odds ratio (OR) with 95% confidence intervals (CIs), and p < 0.05 was defined as statistically significant. The results of the meta-analysis were graphically displayed by forest plots and heterogeneity was further explored by subgroup analysis and sensitivity analysis. Notably, the event rates can be zero or one from some studies yet they still need to be included in the analysis to represent the whole population. In such cases, the resulting distribution of proportions tends to be 0 inflated. We made good use of the Freeman-Tukey double arcsine transformation to perform normalization and variance, which can be achieved by using the metaprop module to perform fixed and random effects metaanalysis of proportions.

Quality assessment
The results of risk of bias for each study are shown in Tables 1 and 2. All studies had a score ≥ 5, which indicates a relatively high research quality. The subjects included were representative, and ascertainment of exposure was confirmed by secure record. Outcome assessment was based on medical records, and the follow-up period was sufficient for outcomes to occur.

Discussion
This up-to-date meta-analysis comprehensively examined the published literature to evaluate the efficacy of co-transplantation of MSCs and haplo-HSCT in patients with SAA. To the best of our knowledge, this is the first meta-analysis to compare the clinical outcomes of MSCs with no MSCs in haplo-HSCT in patients with SAA. The results of our study are partially consistent with a previous meta-analysis examining the effects of MSCs posttransplantation of haplo-HSCT in hematological malignancies [15]. For example, we both concluded that MSCs make no difference in the incidences of aGVHD and CMV infection. They found a role of MSCs in reducing the incidences of cGVHD, while we did not.
Our study demonstrated no significant difference with regard to the pooled incidences of GVHD between MSCs and no MSCs groups. It is well known that GVHD remains the common and life-threatening complication limiting the widespread use of haplo-HSCT, as it associates with a high mortality and morbidity [57]. Since there were no controlled studies, we compared the incidence of aGVHD, grade II-IV aGVHD, and cGVHD in MSCs group and no MSCs group. Although the incidence of aGVHD was higher than no MSCs group, the incidences of grade II-IV aGVHD and cGVHD were lower than the pooled results in no MSCs group. However, no significant differences were found in pooled results between these two groups, which was different from the previous studies supporting a role of MSCs in reducing the incidence of GVHD [15,27]. Despite these previous studies showing that MSCs are effective in GVHD prophylaxis or treatment, most of them were conducted in vitro or in hematological malignancies. Moreover, the conclusions were drawn in HSCT area without highlighting on haplo-HSCT [58][59][60]. Hence, MSCs may make little or no difference to the risk of GVHD compared to no MSCs in haplo-HSCT for SAA patients.
Among the MSCs group, four of eight studies achieved 100% hematopoietic reconstitution and full donor chimerism after the application of MSCs in haplo-HSCT, which is higher than no MSCs group (4/17). Although MSCs were higher than no MSCs group with regard to the pooled results of 2-year OS and engraftment rates in our report, no statistically significant differences were found. Furthermore, it is reported that infections are the other major causes of death after haplo-HSCT in addition to GVHD [61]. We calculated the rates of death due to infection. The pooled result was 9.5% (95% CI, 5.8 to 13.1%) in the included studies in the MSCs group, which was much lower than those reported by the Center for International Blood and Marrow Transplant Research (CIBMTR), for all haplo-HSCT transplants conducted between 2009 and 2010 (infection 13-18%) [62]. CMV infections are opportunistic infections caused by low immune function. A reduction in CMV infection after allo-HSCT can It is reported that MSCs produce growth factors to aid tissue regeneration and accelerate the hematologic reconstitution [63]. The median post-HSCT times to neutrophil greater than 0.5 × 10 9 /L and platelet greater than 20 × 10 9 /L were summarized and listed in Table 1 and 2. The shortest and longest median time to achieve neutrophil engraftment and platelet engraftment were 11-14 days and 13-21 days respectively in the MSCs group and 10-19 days and 13-28 days respectively in the no MSCs group. Remarkably, all studies in the MSCs group reported results descriptively and stated that they observed either rapid engraftment [37] or a similar speed of engraftment after adding MSCs [31,32], which may demonstrate a role for MSCs in the enhancement of engraftment in SAA patients who underwent haplo-HSCT. According to the published papers, the treatment efficacy of MSCs varies among clinical trials, and MSC source might influence this [64]. The studies included in this meta-analysis used only BM-MSCs or UC-MSCs. Therefore, we conducted a subgroup meta-analysis for GVHD prevention according to MSC source. Consequently, the incidence of GVHD shows no significant difference with regard to the use of UC-MSCs versus BM-MSCs. Besides, we conducted "influence analysis" in Stata to explore the source of heterogeneity in no MSCs group. We could reasonably infer that studies (Jennifer et al. [39] and Gao et al. [40]) were one of the most important sources resulting in the heterogeneity of aGVHD and cGVHD respectively. Both of them were conducted in earlier years with different conditioning regimes and prophylaxis measures. There are some limitations in our meta-analysis. First, there may be a risk of confounding biases because various baseline characteristics or cointerventions including age, gender, donor type, conditioning regimen, and MSC originals may affect the treatment outcomes in SAA patients after haplo-HSCT; they were not fully controlled in this study. In addition, patients in no MSCs group usually had high heterogeneity. Although we tried to decrease the bias through statistical methods, sometimes errors were unavoidable. Second, because SAA is a rare disorder, few prospective control trials between the MSCs and no MSCs group are available so far, and all the included studies had small sample sizes. Besides, all were single-arm studies and case series that lacked a control group and likely suffered from a high risk of selection bias. Last but not least, we could not assess publication bias using funnel plots because we only had single-arm studies and case series.
In conclusion, our meta-analysis indicates that the prophylactic use of MSC co-transplantation does not reduce the incidence of GVHD and improve 2year OS in patients with SAA undergoing haplo-HSCT. Hence, the general co-transplantation of MSCs in routine clinical practice for SAA haplo-HSCT recipients is not recommended. However, since there is no direct evidence from comparative study to support this conclusion, more prospective, randomized controlled trials (RCTs) are needed to confirm whether MSCs convey a definite benefit for haplo-HSCT for SAA patients.