Safety of Cell Therapy with Mesenchymal Stromal Cells (SafeCell): A Systematic Review and Meta-Analysis of Clinical Trials

Background Mesenchymal stromal cells (MSCs, “adult stem cells”) have been widely used experimentally in a variety of clinical contexts. There is interest in using these cells in critical illness, however, the safety profile of these cells is not well known. We thus conducted a systematic review of clinical trials that examined the use MSCs to evaluate their safety. Methods and Findings MEDLINE, EMBASE, and the Cochrane Central Register of Controlled Trials (to June 2011), were searched. Prospective clinical trials that used intravascular delivery of MSCs (intravenously or intra-arterially) in adult populations or mixed adult and pediatric populations were identified. Studies using differentiated MSCs or additional cell types were excluded. The primary outcome adverse events were grouped according to immediate events (acute infusional toxicity, fever), organ system complications, infection, and longer term adverse events (death, malignancy). 2347 citations were reviewed and 36 studies met inclusion criteria. A total of 1012 participants with clinical conditions of ischemic stroke, Crohn's disease, cardiomyopathy, myocardial infarction, graft versus host disease, and healthy volunteers were included. Eight studies were randomized control trials (RCTs) and enrolled 321 participants. Meta-analysis of the RCTs did not detect an association between acute infusional toxicity, organ system complications, infection, death or malignancy. There was a significant association between MSCs and transient fever. Conclusions Based on the current clinical trials, MSC therapy appears safe. However, further larger scale controlled clinical trials with rigorous reporting of adverse events are required to further define the safety profile of MSCs.


Introduction
Mesenchymal stromal cells (mesenchymal stem cells; MSCs) are a heterogeneous group of cells that can be isolated from many adult tissues (e.g. bone marrow, adipose tissue). First described in 1974 [1] they have recently received attention in a number of different clinical fields for their potential therapeutic effects.
Although often described as 'adult stem cells', MSC's have limited cellular differentiation ability. Instead, pre-clinical evidence suggests that MSCs exert their beneficial effects largely through immunomodulatory and paracrine mechanisms. MSCs home to sites of inflammation and secrete bioactive molecules, and thus may be especially effective in proinflammatory diseases. [2] There is a growing body of literature demonstrating the efficacy of MSC therapy in a variety of pre-clinical models, including acute lung injury, [3,4] septic shock, [5] and acute myocardial infarction. [6] Several small clinical trials have investigated the efficacy and safety of MSCs in diseases including chronic heart failure, acute myocardial infarction, hematological malignancies and graft versus host disease.
There is interest in applying MSCs to pulmonary diseases (e.g. chronic obstructive pulmonary disease) and critical illness (e.g. acute respiratory distress syndrome); however, safety concerns represent a significant barrier to the successful translation of MSCs into an acceptable clinical therapeutic. These include neoplastic potential due to MSC's proliferative capacity, susceptibility to infection given their immunomodulatory effects, embolism of the cells, zoonoses associated with cell culture reagents, and acute or chronic immunogenicity of the cells themselves. [7] Therefore, we conducted a systematic review of the literature to evaluate the safety of MSC-based therapy in all clinical trials.

Eligibility Criteria
We included randomized and non-randomized controlled trials as well as uncontrolled clinical trials (Phase I/II trials with more than two participants) that examined the safety of intravascularly delivered MSCs in adult (at least 18 years old) or mixed adult and pediatric participants. All clinical settings were included. We excluded studies that exclusively used non-intravascular routes of administration, ex vivo differentiated MSCs, or co-administered MSCs with other experimental cells or treatments.

Search Strategy
We conducted electronic searches without language restriction of Ovid MEDLINE (1950 to June 2011), EMBASE (1980 to Week 21, 2011), Cochrane Central Register of Controlled Trials (2 nd Quarter 2011), and the Cochrane Database of Systematic Reviews (2 nd Quarter 2011). Given the non-standard terminology associ-  ated with MSCs a number of terms were used (Appendix S1, search strategy). ClinicalTrials.gov was searched for ongoing or recently completed trials. Abstracts and proceedings from clinical conferences were identified and searched using Web of Science (September 2010). Bibliographies of retrieved articles and relevant reviews were searched.

Assessment of Risk of Bias
RCTs that met inclusion criteria were assessed for risk of bias according to the Cochrane Collaboration methods. [8] Study Selection, Data Collection and Analysis All study selection and data extraction was performed independently by two reviewers (M.M.L., C.P.) using standardized forms. Discrepancies were resolved by discussion with a third author (L.L.M.).

Main Outcome Measure: Adverse Events
Adverse events were grouped according to the immediacy of the event (acute infusional toxicity, fever), the occurrence of organ system complications [neurological, pulmonary, cardiovascular (arrhythmias and other cardiac events), gastrointestinal and renal, and hematologic systems], infection, and the occurrence of longer term adverse effects (death, malignancy).
Completeness of adverse events reporting was assessed using the CONSORT approach to harm reporting. [9] Specifically, we examined whether expected adverse events were listed and defined in the methods section, whether events were described as serious versus non serious (e.g. as per Good Clinical Practice Guidelines), and if frequency and duration of follow up of adverse events was provided.

Statistical Analysis
Meta-analyses of adverse events was performed using Comprehensive Meta-analysis (Version 2, Biostat). Data was analyzed by Peto's method with correction of zero-count cells. Pooled events were described using odds ratios (OR) and 95% confidence intervals (95% CI). An odds ratio less than 1 favoured MSC treatment. Heterogeneity between trials was evaluated using the I 2 test [10] as well as the x 2 test. Sensitivity analyses were planned according to the patient population, MSC type (autologous versus allogeneic; fresh versus cryopreserved), and culture media (fetal bovine serum versus human).
Adverse events for non-randomized controlled trials with control groups that did not receive any dose of MSCs were pooled and reported according to numbers and proportions.

Search Results
Our search identified 2347 unique titles and 36 studies met inclusion criteria (see Figure 1). Seven unpublished studies were found in a search of clinicaltrials.gov (Appendix S2). Nineteen studies were found as abstracts only (Appendix S3).
Sample sizes ranged from 3 to 200 participants (28634, mean 6 standard deviation). The follow-up period was reported in all studies and the duration ranged from 0.5-60 months. Two studies (one RCT [13] and one non-RCT [21]) reported funding from a for-profit manufacturer of MSCs (Osiris Therapeutics, Inc.).

Sensitivity Analyses
The small number of RCTs in each meta-analysis precluded the conduct of planned sensitivity analyses.

Completeness of Reporting of Adverse Events
Twenty-eight of the 36 studies listed a priori at least one expected adverse event to be monitored, while the remainder did not. [17,18,26,27,29,31,33,35] Five studies explicitly reported and separated serious from non-serious adverse events; [13,14,20,24,46] two of these referenced a standardized approach to detailing adverse events developed by other organizations. [20,46] One study provided a priori a description of follow-up frequency and duration for all listed adverse events. [25] Eighteen studies provided this description for select adverse events, [11][12][13][14][15][16]18,19,[22][23][24]28,30,32,37,42,44,45] Seventeen studies provided no details for follow-up duration and frequency of reporting of adverse events. [17,20,21,26,27,29,31,[33][34][35][36][38][39][40][41]43,46] Discussion This is the first systematic review and meta-analysis to comprehensively summarize the safety of systemic MSC administration. Our analysis was unable to detect associations between MSC treatment and the development of acute infusional toxicity, organ system complications, infection, death, or malignancy. There was, however, a significant association between MSC administration and transient fever. Our systematic review of non-RCTs supported these results. Six of seven RCTs and all non-RCTs described equal or fewer deaths with MSC treatment compared to control treatment. The completeness of adverse event reporting in the included studies was variable. However, aside from fever, the published current clinical trials suggest that the administration of MSCs is safe.
Although malignant transformation is a theoretical risk, our pooled analysis found no association between MSCs and tumour formation. Concerns related to tumourgenicity of MSCs were raised by preclinical studies demonstrating increased tumour burden in vivo. [47] Although recent position papers have suggested low probability of malignant transformation and tumour formation with MSCs, [7] our review is the first systematic analysis of the issue. Malignancy occurred only in studies involving participants with ongoing or previous malignancies; no de novo malignancies were observed.
We found no evidence of increased susceptibility to infection with MSC administration. Although MSC immunomodulatory effects may be beneficial in pro-inflammatory diseases, these same effects may leave a patient susceptible to infection. [48] In our review, infections were common in already immunosuppressed      [13,14] There was a significant association between MSC administration and the development of fever. Fever was transient and not associated with long term sequelae. The mechanisms for fever are not clear but could be related to acute inflammatory reactions by a subset of patients to particular preparations of MSCs, not unlike similar reactions occasionally observed with red blood cell administration. [50] Our review also addresses several issues and theoretical concerns with the cell product used in studies. First, concerns for immunogenicity may be unfounded as 13 studies used unmatched allogeneic MSCs with no reports of acute infusional toxicity. This supports the idea that MSCs are 'immune-privileged', a characteristic that may be explained by their low expression of MHC proteins and T-cell co-stimulatory molecules. [51] Second, the use of fetal bovine serum for culturing MSCs has been criticized for potentially introducing zoonotic contamination to the cell product (e.g. prion disease), and also potentially increasing the immunogenicity of the cells. [52,53] Although the majority of included studies used fetal bovine serum, only one study specifically monitored for potential adverse events associated with its use. Concerns over fetal bovine serum will likely decrease in the future as expansion of MSCs in human blood products becomes more commonplace. The use of dimethylsulfoxide as a cryopreservative has been another potential concern with MSC therapy as this chemical is known to have toxic side effects and can cause hypersensitivity reactions. [54,55] In our review only one study documented the occurrence of acute infusional toxicity and attributed it to dimethylsulfoxide. [30] A final concern is the viability of cells administered, as the administration of necrotic cells or cellular by-products may increase immunogenicity. Less than half of the studies included assessed and reported on viability Abbreviations: ¥ = both follow-up duration and frequency defined a priori for adverse event, + = adverse event not defined a priori, * = clinical endpoint defined a priori, ++ = only follow-up duration listed for all events a priori, follow-up frequency not listed, AE = adverse event, DMSO = dimethylsulfoxide, FBS = fetal bovine serum, GVHD = graft-versus-host disease. doi:10.1371/journal.pone.0047559.t010 of MSCs prior to infusion. Thus, greater vigilance may be needed in future studies for reporting cellular viability and monitoring for potential dimethylsulfoxide related adverse events. No significant relationship between MSC administration and acute infusional toxicity was observed. The only RCT which described acute adverse reactions during infusion (acute and transient pulmonary edema in three participants) delivered MSCs to participants with chronic ischemic heart failure. [11] MSCs initially distribute to the lungs after intravascular administration; [56] thus, in susceptible patients this could cause a transient increase in pulmonary pressures and lead to pulmonary edema.
The reporting of adverse events was highly variable among the included studies. This may be related to editorial constraints of journals. Since use of MSCs may be associated with neoplastic growth long term, it is difficult to understand why approximately 50% of studies did not report follow up duration for adverse events. For highly experimental interventions with unestablished safety profiles, we contend that it is important to summarize the adverse reporting plan in the methods section of manuscripts and report short term and longer term events.
Our systematic review has several limitations. First, despite our comprehensive search strategy, there are a number of completed but unpublished industry sponsored studies and studies published in abstract form only that may alter the safety profile of MSCs. Second, we pooled adverse events across heterogeneous disease states. Given the limited number of clinical MSC studies, and the small sample sizes of each, it was important to pool data across trials to determine if any potential signals of harm existed. Previously, we have advocated this approach when individual trials are not adequately powered to detect potential harm. [57] However, we acknowledge that the occurrence, type, and severity of adverse events may vary significantly between different populations and according to different MSC characteristics (e.g. dose, type). The limited number of included RCTs precluded the conduct of these sensitivity analyses. Third, the majority of RCTs included in our analysis would be considered a high risk of bias. Although double blinding an MSC trial may be considered ethically unacceptable, it is difficult to justify the lack of concealment of the allocation of patients in many studies.

Conclusions
Our study provides a systematic examination for adverse events related to the use of MSCs. We did not identify any significant safety signals other than transient fever. Results from our systematic review should provide some assurance to investigators and health regulators that, with the present evidence, this innovative therapy appears safe.