Adult Bone Marrow–Derived Cells for Cardiac Repair:  A Systematic Review and Meta-analysis FREE

Ischemic heart disease (IHD) is a major cause of mortality and morbidity worldwide and accounts for approximately 20% of all deaths in the United States.^1- 3 Despite significant advances in medical therapy and interventional strategy, the prognosis of millions of patients with acute myocardial infarction (MI) and ischemic cardiomyopathy remains dismal.^4- 5 Although the underlying mechanism remains controversial, numerous studies in animals have documented that transplantation of bone marrow (BM)–derived cells (BMCs) following acute MI and in ischemic cardiomyopathy is associated with a reduction in infarct scar size and improvements in left ventricular (LV) function and perfusion.⁶ In humans, transplantation of BMCs and BM-derived circulating progenitor cells (CPCs) in patients with acute MI as well as chronic IHD has yielded similar encouraging results.^7- 8

However, these studies in humans are heterogeneous in their methods and have yielded disparate results. These studies have each enrolled a small number of patients and have fallen short of providing conclusive results. Thus, the extent to which BMC transplantation can improve outcomes in patients with IHD remains unclear. To our knowledge, there are no comprehensive syntheses of these data. Therefore, we performed a systematic review of the literature and meta-analysis to critically evaluate and summarize the potential therapeutic benefits of BMC transplantation for cardiac repair in patients with IHD.

The review question was to what extent does BMC transplantation affect cardiovascular outcomes in patients with IHD? We report this protocol-driven systematic review according to the Meta-analysis of Observational Studies in Epidemiology (MOOSE)⁹ and Quality of Reporting of Meta-analysis (QUOROM)¹⁰ statements.

Two reviewers (A.A.-L. and I.M.T.) judged eligibility of studies in duplicate and independently. Eligible studies were randomized controlled trials (RCTs) and cohort studies examining the effects of BMC transplantation on cardiovascular outcomes in patients with IHD. Because cytokines may exert cardiovascular effects, we excluded studies of cardiac repair solely via the mobilization of endogenous BMCs with systemic administration of cytokines.

We searched MEDLINE (January 1980 to July 2006), the Cochrane Central Register of Controlled Trials (CENTRAL) (July 2006), EMBASE (January 1980 to July 2006), CINAHL (Cumulative Index to Nursing and Allied Health) (January 1982 to July 2006), the US Food and Drug Administration Web site (http://www.fda.gov), and BIOSIS Previews (January 1980 to July 2006) using the following database-appropriate terms: coronary artery disease, myocardial infarction, stem cells, progenitor cells, bone marrow, circulating progenitor cells, myocardial regeneration, and cardiac repair. We sought additional studies by reviewing the reference lists of eligible studies and relevant review articles. The complete search strategy is available on request from the authors.

Two reviewers (A.A.-L. and I.M.T.) working in duplicate and independently used a standardized form to abstract the data from each study. The corresponding author (B.D.) solved disagreements that could not be solved by consensus. When necessary, LV end-diastolic volume was estimated from LV end-diastolic volume index, and infarct volume/mass was converted to infarct size expressed as a percentage of LV by calculating total LV myocardial volume from LV mass index. Data from echocardiography and cardiac magnetic resonance imaging were considered equivalent. When both echocardiographic and cardiac magnetic resonance imaging functional data were available, cardiac magnetic resonance imaging data were preferentially used.

We used the criteria by Jüni et al¹¹ to ascertain the methodological quality of included randomized trials¹¹ and a modified Newcastle-Ottawa scale¹² to assess the quality of cohort studies.

The main outcomes of our review were change from baseline in mean LV ejection fraction, infarct scar size, LV end-systolic volume, and LV end-diastolic volume. We conducted random-effects meta-analyses to pool these outcomes across included studies, estimating weighted mean differences between BMC-treated patients and control patients and their associated 95% confidence intervals (CIs). We estimated the proportion of between-study inconsistency due to true differences between studies (rather than differences due to random error or chance) using the I² statistic,¹³ with values of 25%, 50%, and 75% considered low, moderate, and high, respectively. Funnel plots graphically explored publication bias. We used RevMan version 4.2.7 (Cochrane Collaboration, 2004) for these analyses.

We conducted planned subgroup analyses and tested for treatment-subgroup interactions. Planned subgroups comprised the types of study design (RCTs vs cohort studies); the clinical scenario in which BMCs were used (acute MI vs chronic IHD); timing of BMC transplantation after MI and/or percutaneous coronary intervention (<5 days vs within 5-30 days); the number of cells injected (above vs below the median of 80 × 10⁶ BMCs used in the eligible studies); and the population of BMCs used (BM mononuclear cells vs nonmononuclear cells, including mesenchymal stem cells and BM-derived circulating progenitor cells). Because most of the included studies used the intracoronary route for BMC transplantation, the impact of the route of transplantation on outcomes could not be assessed.

Of 213 articles retrieved during the initial search (Figure 1), 81 were not reports of original investigations (review articles and editorials), 95 were conducted in animals, 6 used mobilization rather than transplantation of BMCs, 6 lacked control groups, and 7 were performed in vitro. Eighteen studies (12 RCTs and 6 cohort studies) with a total of 999 patients were eligible for review. The interreviewer agreement on study eligibility was 100%.

Table 1 summarizes the characteristics of all studies included in our meta-analysis. Notably, the sample size in each study was relatively small (range, 20-204 patients; median, 36 patients), and the follow-up duration was relatively short (range, 3-18 months; median, 4 months). There was considerable heterogeneity in the timing of cell transplantation after MI or percutaneous coronary intervention (range, 1 day to 81 months; median, 9.8 days) and in the number of BMCs used (range, 2 × 10⁶ to 60 × 10⁹ cells [median, 80 × 10⁶ BMCs]).

Table 2 describes the methodological quality of the RCTs, and Table 3 describes the quality of the cohort studies. All cohort studies and at least 6 RCTs failed to blind participants and caregivers, and at least 2 RCTs and 3 cohort studies failed to blind outcome assessors. The follow-up was complete in all eligible studies. The interreviewer agreement on these quality domains was greater than 90%.

Compared with control, BMC transplantation improved LV ejection fraction by 3.66% (95% CI, 1.93% to 5.40%; [I² = 71%; P<.001]; Figure 2), reduced infarct scar size by 5.49% (95% CI, −9.10% to −1.88% [I² = 66%; P = .003]; Figure 3); reduced LV end-systolic volume by 4.80 mL (95% CI, −8.20 to −1.41 mL; [I² = 0%; P = .006]; Figure 4); and reduced LV end-diastolic volume by 1.92 mL (95% CI, −6.31 to 2.47 [I² = 0%; P = .39]; Figure 5). We drew funnel plots to seek evidence of publication bias: where inconsistency was high, the funnel plots were not interpretable; where inconsistency was low, the funnel plots were inconclusive (available at: http://www.louisville.edu/medschool/medicine/cardiology/Archinternmed_2007_supplemental_data.pdf).

We did not find any treatment-subgroup interaction through any of our planned subgroup analyses (Table 4).

The injection of BMCs was found to be safe without significantly greater risk of major local or systemic complications. Except for Bartunek et al,¹⁵ who reported a higher incidence of in-stent restenosis in the BM mononuclear cell–treated group (9 of 19 patients vs 4 of 16 patients in the control group), the rate of restenosis was comparable among BMC-treated and control patients. The incidence of other complications, such as recurrent angina, MI, and sustained or nonsustained supraventricular or ventricular arrhythmias, was not significantly different between BMC-treated patients and controls. A supplemental table of reported incidence of complications in BMC-treated patients and controls is available at: http://www.Louisville.edu/medschool/medicine/cardiology/Archinternmed_2007_supplemental_data.pdf.

This systematic review and meta-analysis, the first, to our knowledge, to comprehensively summarize the available evidence of BMC transplantation in patients with IHD, indicates that BMC transplantation in patients with IHD is apparently safe and leads to modest benefits beyond those achieved with revascularization and conventional pharmacotherapy. Our results indicate that BMC transplantation may improve LV ejection fraction, infarct scar size, and LV end-systolic volume. However, the mechanisms explaining these benefits remain unclear.

Although the plasticity of adult stem cells remains debatable, extensive data from animal models indicate that BMCs are capable of differentiating into cells of cardiac and vascular lineages.^32- 38 Bone marrow–derived mesenchymal stem cells, mononuclear cells, and circulating endothelial progenitor cells have all been shown to differentiate into cardiomyocytes both in vitro and in vivo.⁷ Nevertheless, tracking cellular differentiation after transplantation in humans remains particularly difficult. Another potential mechanism is that transplanted BMCs may secrete a variety of growth factors and cytokines,³⁹ thereby enhancing myocyte survival following ischemic injury and facilitating the migration of resident cardiac stem cells⁴⁰ to the site of injury and their reparative activity. The finding of infarct scar size reduction with BMC therapy may signify new myocyte formation, superior preservation of existing myocytes, or both following BMC transplantation.

Beyond these mechanistic considerations, some technical issues remain unclear, such as the optimal number of BMCs, the optimal timing and route of transplantation, and the most effective type of BMC. Since only a small fraction of BMCs are retained in the myocardium following injection,⁴¹ we analyzed the pooled data based on the number of cells transplanted. There were no significant differences in outcomes between the groups that received less or more than the median number of cells. Although somewhat surprising, these findings perhaps underscore the importance of selective injection of the most efficacious cell subpopulation.

Furthermore, the impact of cell number may be affected by the timing⁴² and route⁴¹ of transplantation, both of which may influence cell retention. The retention of injected endothelial progenitor cells was much lower in sham-operated nude rats compared with nude rats 24 hours after acute MI.⁴² Furthermore, the benefits of BMC injection in the first few days after acute MI may be jeopardized by the local inflammation that renders the myocardium a hostile environment for the injected cells. In the Reinfusion of Enriched Progenitor Cells And Infarct Remodeling in Acute Myocardial Infarction (REPAIR-AMI) trial, the authors stratified data according to the time of BMC injection after acute MI.²⁸ While there was no correlation between the timing of the procedure and LV contractile recovery in the placebo group, a significant correlation was observed in the BMC-treated group. Transplantation of BMCs was more beneficial when performed 5 days or later after acute MI.²⁸ In our meta-analysis, injection of BMCs in the 5- to 30-day window resulted in a more than 3-fold greater reduction in infarct size and greater improvement in LV end-systolic volume compared with injection in the first 5 days after acute MI and/or percutaneous coronary intervention. Because the overall change in LV end-diastolic volume was not different between BMC-treated and control groups, a change in LV end-systolic volume may represent an improvement in global LV function. However, none of these interactions reached statistical significance, and the importance of these findings remains uncertain at this time. This lack of subgroup-treatment interaction may have resulted from a small number of studies with a small number of patients. Future meta-analyses with larger patient numbers or large randomized trials may identify potential interactions between treatment effects and the timing of BMC injection.

It is important to note that the majority of studies included in our review used unfractionated BM mononuclear cells^{18,20,23- 26,28- 30} and that BMC transplantation was reportedly safe in these studies. Although intracoronary injection of CD133⁺ BM mononuclear cells was associated with an increased incidence of in-stent restenosis,¹⁵ no other major adverse effects were noted in studies using different BMC populations. This safety profile of BMC transplantation as reported in these studies with follow-up durations of up to 18 months supports conducting further investigation of therapeutic efficacy. The possibility that reporting bias may be affecting the otherwise favorable safety picture emerging from our review, however, demands caution.

The duration of follow-up in the studies included in this meta-analysis was relatively short. Although the Transplantation of Progenitor Cells and Regeneration Enhancement in Acute Myocardial Infarction (TOPCARE-AMI) trial showed persistent benefits after 12 months of BMC and circulating progenitor cell therapy,⁴³ a longer follow-up of 18 months failed to demonstrate statistically significant improvements with cell therapy in the Bone Marrow Transfer to Enhance ST-Elevation Infarct Regeneration (BOOST) study.²⁴ Whether the benefits of BMC therapy are ephemeral remains to be assessed in larger trials with longer follow-up duration (eg, 5 years). Moreover, a single dose of BMCs may not be sufficient for myocardial repair, and repeated infusions may result in sustained benefits over a longer time frame, but this remains speculative. Genetic modifications of BMCs prior to transplantation may also potentially improve their regenerative capability.⁴⁴ These avenues may be explored in future trials. Overall, our findings support the recent consensus statement on the use of autologous adult stem cells for cardiac repair by the task force of the European Society of Cardiology that called for a pragmatic approach for demonstrating the efficacy of stem cell therapy in myocardial repair in humans.⁴⁵

Limitations in study quality (namely, lack of blinding), unexplained between-study inconsistency, sparse evidence, and indirectness of the outcomes (ie, exclusive reliance on surrogate outcomes) weaken the inferences. The methods for evaluating LV function, the type of BMC used, and the interval between acute MI and/or percutaneous coronary intervention and BMC transplantation varied among the included studies (Table 1), all of which are potential sources of heterogeneity. However, the consistency of the beneficial effect of BMCs in most of the prespecified primary end points and subgroups suggests that the association is valid. The fact that the beneficial effect of BMCs persisted across different study designs, BMC lines, timings and routes of transplantation, and clinical scenarios suggest that the association can cautiously be generalized to different patient populations.

We believe that combining data from RCTs and cohort studies was justified because for both designs patients were followed prospectively, accurate methods were used to assess the primary end points, and few patients if any were lost to follow-up. Importantly, the results were consistent even when the analysis was restricted to RCTs or cohort studies alone (Table 4 and Figures 2-5), strengthening the fact that the results of the meta-analysis are cautiously generalizable.

In conclusion, the results of our systematic review and meta-analysis suggest that BMC transplantation in patients with acute MI as well as chronic IHD is reportedly safe and is associated with modest improvements in LV ejection fraction, infarct scar size, and LV end-systolic volume, beyond those achieved with state-of-the-art therapy; however, there was no significant effect on LV end-diastolic volume. Although the benefits are modest, our results support the organization, funding, and conduct of larger randomized trials of BMC therapy designed to critically evaluate the long-term impact of BMC therapy on patient-important outcomes in patients with IHD.

Correspondence: Buddhadeb Dawn, MD, Division of Cardiology, 550 S Jackson St, Ambulatory Care Building, Third Floor, Louisville, KY 40292 (buddha@louisville.edu).

Accepted for Publication: January 24, 2007.

Author Contributions: Drs Abdel-Latif and Dawn had full access to all of the data in this study and take responsibility for data integrity and the accuracy of data analysis. Study concept and design: Abdel-Latif, Dawn, Bolli, Tleyjeh, and Hornung. Acquisition of data: Abdel-Latif, Tleyjeh, Perin, Zuba-Surma, Bolli, and Dawn. Analysis and interpretation of data: Abdel-Latif, Tleyjeh, Montori, Hornung, Perin, Bolli, Dawn, and Al-Mallah. Drafting of the manuscript: Abdel-Latif, Dawn, Zuba-Surma, Tleyjeh, Montori, Bolli, and Al-Mallah. Critical revision of the manuscript for important intellectual content: Dawn, Bolli, Montori, Abdel-Latif, Tleyjeh, Hornung, and Perin. Statistical analysis: Abdel-Latif, Montori, Tleyjeh, Hornung, and Dawn. Obtained funding: Bolli and Dawn. Administrative, technical, or material support: Bolli and Dawn. Study supervision: Dawn, Bolli, Hornung, and Perin.

Financial Disclosure: None reported.

Funding/Support: This meta-analysis and publication was supported in part by grants R01 HL-72410, HL-55757, HL-68088, HL-70897, HL-76794, and HL-78825 from the National Institutes of Health.

Additional Information: A supplementary table (reported incidence of complications in BMC-treated patients and controls) and figure (funnel plot [according to outcomes] for studies included in the meta-analysis) are available at: http://www.louisville.edu/medschool/medicine/cardiology/Archinternmed_2007_supplemental_data.pdf.