The present meta-analysis included 17 RCTs in which BMSC treatment was compared with a control group. The results demonstrated that, at 6 months, BMSC treatment leads to a 2.51% improvement in LVEF, a 4.98-ml reduction in LVESV, and a 3.46-ml reduction in LVEDV. Similar results were obtained in recent meta-analyses [30, 31]. Taken together, BMSC treatment improved LVEF outcome at 6 months.
We stress some important issues concerning rigorous study design in randomized controlled trials. All of the RCTs in the present analysis conducted random allocation to treatment and control groups. However, some researchers made an effort to maintain blindness to patients as well as healthcare providers in regard to group allocation, whereas others did not. Clinical trials of stem cell therapy should be strictly designed, as it is very difficult to maintain blinding throughout the entire research period, even when research is designed with double-blind controlled trials . We considered two procedures that are performed to maintain the blindness of treatment allocation to both patients and healthcare providers. First, bone marrow is harvested in all patients, including the control group, under full anesthesia just before surgery to ensure that the characteristics of the BMSC suspension do not differ significantly between the two groups [2–7]. Second, patients are returned to cardiac catheterization after bone marrow aspiration to ensure identical injection procedures in all patients. It is assumed that using autologous erythrocytes or patients’ own serum or plasma in the placebo preparation ensures double blindness. In the present results, no significant treatment effect was found in studies in which the control group underwent bone marrow aspiration, as indicated by the LVEF change of −1.29% (95% CI, −4.15 to 1.58), whereas studies that did not perform bone marrow aspiration in the control group showed significant improvement in LVEF by 3.81% (95% CI, 2.44 to 5.17). The intervention effect might therefore be overestimated because of the study design. Several trials used placebo preparations without cells in which the content of the syringes could be easily distinguished between the active treatment and the placebo. It is important that BMSCs were harvested in all patients, including the control group, under full anesthesia just before surgery, to guarantee that the characteristics of the BMSC suspension did not differ significantly between groups.
Nine of 17 trials implemented identical cardiac catheterization injection procedures after bone marrow aspiration in both the BMSC and control groups to maintain study blindness [2–7, 19, 21]. The surgeon was unaware whether cells or only saline was being injected. Trials that conducted cardiac catheterization as a sham procedure in the control group did not show significant changes in LVEF at 6 months (0.92%; 95% CI, -0.61 to 2.44); however, those without catheterization of control subjects showed significant LVEF changes (4.45%; 95% CI, 2.48 to 6.43).
One of the major differences of this study from other preexisting studies is that only conditions known to be effective in bias-minimized RCT studies were selected to evaluate how the rigorousness of method affected treatment effects. All included studies included PCI before the infusion of BMSC treatment or placebo. We included only trials in which the cell dose was higher than 108. Previous trials determined that the use of 108 or more injected cells shows improved outcomes in BMSC-treated patients [15, 30]. The mean change in LVEF was statistically significant in favor of administering BMSCs in studies using higher doses of BMSCs. These results suggest that significant effects on LVEF may be achieved only when the infusing doses are higher than 108 BMSCs. The route of BMSC delivery was intracoronary artery infusion by using a balloon in all included studies, as the present meta-analysis included only AMI patients. All included trials reported up to 6 months of short-term follow-up data to demonstrate the effects of rigorous study design.
We excluded whether the BMSCs were administered within 24 hours after primary PCI, because it was previously shown that the timing of cell transfer does not have a BMSC treatment effect in AMI patients [31, 32]. A larger randomized study suggested that BMSCs should ideally be administrated more than 4 days after STEMI to obtain the best benefit from this therapy . In a subgroup analysis, the REPAIR-AMI Trial found that the most favorable effects on LV function were observed with BMSC delivery on days 5 to 7 after MI .
We chose the studies to be included based on study design, route of delivery, cell type, cell dose, timing of injection, and follow-up duration to reduce heterogeneity and to estimate the effect of rigorous study design in randomized controlled trials. Heterogeneity was reduced from between 85% and 98% to 75% in the present study. When only rigorously designed studies were included in the analysis, heterogeneity was further reduced to 60%. Meta-regression analyses, exploring the source of heterogeneity, indicated that the treatment effects seen were associated with rigorous study designs. These findings were similar to those of the subgroup analyses. Design rigorousness seemed to explain the heterogeneity in the studies; however, a considerable degree of heterogeneity was still observed among the included trials. This might be due to differences in patient severity at baseline, the timing of infusions, or the duration of follow-up between studies.