Studies investigating bone marrow stem cell therapy (BMSCT) in patients with chronic ischemic cardiomyopathy have yielded mixed results. A meta-analysis of randomized controlled trials of BMSCT in patients with chronic ischemic cardiomyopathy was undertaken to assess its efficacy and the best route of administration. The MEDLINE, Embase, Cumulative Index to Nursing & Allied Health Literature, and Cochrane Library databases were searched through April 2012 for randomized controlled trials that investigated the impact of BMSCT and its route of administration on left ventricular (LV) function in patients with chronic ischemic cardiomyopathy and systolic dysfunction. Of the 226 reports identified, 10 randomized controlled trials including 519 patients (average LV ejection fraction [LVEF] at baseline 32 ± 7%) were included in the analysis. On the basis of a random-effects model, BMSCT improved the LVEF at 6 months by 4.48% (95% confidence interval [CI] 2.43% to 6.53%, p = 0.0001). A greater improvement in the LVEF was seen with intramyocardial injection compared with intracoronary infusion (5.13% [95% CI 3.17% to 7.10%], p <0.00001, vs 2.32% [95% CI −2.06% to 6.70%], p = 0.30). Overall, there were reductions in LV end-systolic volume of −20.64 ml (95% CI −33.21 to −8.07, p = 0.001) and LV end-diastolic volume of −16.71 ml (95% CI −31.36 to −2.06, p = 0.03). In conclusion, stem cell therapy may improve LVEF and favorably remodel the heart in patients with chronic ischemic cardiomyopathy. On the basis of current limited data, intramyocardial injection may be superior to intracoronary infusion in patients with LV systolic dysfunction.
Although a recent meta-analysis of bone marrow stem cell therapy (BMSCT) in patients with acute myocardial infarctions showed modest improvement in left ventricular (LV) function, the largest randomized controlled trial (RCT) to date evaluating BMSCT in patients with chronic ischemic cardiomyopathy with moderate to severe LV systolic dysfunction failed to reach its primary end point. Additionally, several smaller prospective studies that focused on patients with chronic infarction and heart failure yielded mixed results with stem cell therapy. Furthermore, earlier meta-analyses of stem cell therapy included patients with acute myocardial infarction or ischemic heart disease (acute and chronic myocardial infarctions) with preserved ejection fraction, who have a different pathophysiologic milieu and do not reflect the use of BMSCT in patients with chronic ischemic cardiomyopathy with moderate to severe LV systolic function. Consequently, the role of stem cell therapy in patients with chronic ischemic cardiomyopathy and LV systolic dysfunction remains unclear. Accordingly, we undertook a meta-analysis of RCTs that investigated the impact of BMSCT on the LV ejection fraction (LVEF) and LV remodeling.
Methods
A search of published studies was performed using the following databases: MEDLINE (1948 to April 2012), MEDLINE In-Process, the Cumulative Index to Nursing & Allied Health Literature, the Cochrane Database of Systemic Reviews, the American College of Physicians Journal Club Embase, and the Cochrane Central Register of Controlled Trials. The search terms used were “stem cell transplantation,” “stem cell therapy,” “coronary artery disease,” “myocardial infarction,” “acute coronary syndrome,” “angina,” “heart failure,” “chronic ischemic cardiomyopathy,” “bone marrow stem cells,” “cardiac repair,” and “cardiac regeneration.” We also hand-searched the reference lists of relevant reviews of stem cell therapy in humans. The inclusion criteria used to select studies for the meta-analysis were (1) investigation of the impact of autologous BMSCT in patients with chronic ischemic cardiomyopathy, (2) availability of echocardiographic measures (e.g., the LVEF), (3) baseline LVEFs of the patients included in the RCTs <45%, (4) inclusion of intracoronary (IC) and intramyocardial (IM) routes of administration, and (5) equal application of cointervention, if any, to the control and stem cell therapy groups.
The data extracted from selected RCTs were study characteristics, patients’ baseline characteristics, intervention characteristics in both study and control groups, echocardiographic data, and functional data. When outcome data were reported for multiple end points, we used 6-month data to ensure uniformity. Two investigators (JK and GAU) independently participated in study selection and data extraction. Disagreements were resolved by consensus after discussion with a third investigator (JPS). The criteria proposed by Juni et al were used to assess the methodologic quality of included RCTs. The components of quality assessment were description of adequate randomization sequence, concealment of treatment allocation, adequacy of follow-up, and blinding of investigators who assessed outcomes.
The primary outcome of this study was change in the LVEF from baseline in BMSCT-treated patients compared with controls. Changes in LV end-diastolic and end-systolic volumes were also assessed. A random-effects model was used to perform the meta-analysis. The number of subjects at the end of the protocol was considered the study sample size. We used this approach to be maximally conservative in our estimates of effect sizes. For each study, the reported mean change in the LVEF, with its SD, between the BMSCT and control groups was used in the statistical analysis. In studies in which mean changes in the LVEF were not reported, the unbiased estimator of variance method was used to calculate the SD. In studies in which medians were reported, means and SDs were estimated. Summary statistics are presented as weighted mean differences with 95% confidence intervals (CIs) between BMSCT and control groups. In 1 study, the randomization process was changed in the midst of the trial. Statistical analysis performed with and without this study demonstrated no impact on the magnitude and direction of the weighted mean change in the LVEF, so we included it in the final analysis. Heterogeneity was examined using the I 2 statistic. An I 2 value >75% was considered high, and further investigation to identify the sources of heterogeneity using sensitivity analysis was undertaken. In addition, subgroup analysis was performed on preplanned subgroups: type of cointervention, cell count, unselected versus selected bone marrow stem cells, route of administration, method of LVEF assessment, and the baseline LVEF. Funnel plots, along with Begg’s and Egger’s tests, were used to assess publication bias. Statistical analysis was performed using RevMan version 5 (Cochrane Collaboration, Copenhagen, Denmark) and Stata version 12.1 (StataCorp LP, College Station, Texas).
Results
The initial search yielded 2,367 studies, of which 226 reports were identified, including RCTs, cohort studies, and reviews. Of these, 10 RCTs including 519 patients met our inclusion criteria ( Figure 1 ). In these studies, 283 patients were in BMSCT groups and 236 in control groups. In 1 study with 3 arms (control group, IM arm, and IC arm), the control group was counted twice to perform the statistical analysis but was counted only once to generate the study sample size. Across all included studies, patients in BMSCT groups (mean age 61.6 ± 2.17 years, mean New York Heart Association [NYHA] class 2.25 ± 0.7) were comparable with those in control groups (mean age 60.6 ± 3.11 years, mean NYHA class 2.41 ± 0.62). Women accounted for 22% (n = 56) of control groups and 18% (n = 54) of BMSCT groups. The mean baseline LVEF in control groups was 32.03 ± 7.79%, whereas in BMSCT groups, it was 32.79 ± 6.53%. In all included studies, patients uniformly had histories of myocardial infarction, with echocardiographic, magnetic resonance imaging, or angiographic evidence of infarction. The time interval from infarction to stem cell therapy was highly variable not only across the studies but within the trials as well, varying from ≥3 months to 9 years. The inclusion criteria at the trial level common to all studies were chronic ischemic heart disease with evidence of previous infarction and LV systolic dysfunction on optimal medical therapy. Patients with chronic debilitating conditions, malignancies, atrial fibrillation, ventricular arrhythmias, and atrial fibrillation were excluded from all included RCTs.
Subjects included in the individual studies varied on several characteristics. Notably, in the included RCTs of Perin et al and Pokushalov et al, subjects were patients with “no-option” chronic ischemic heart disease, whereas in other studies, patients were electively scheduled for coronary artery bypass grafting (CABG). The baseline characteristics of patients enrolled in the included studies are listed in Tables 1 and 2 .
| Study | Age (yrs) | Women | NYHA Class | Baseline LVEF (%) | LVEF Assessment Method | ||||
|---|---|---|---|---|---|---|---|---|---|
| Control | BMSCT | Control | BMSCT | Control | BMSCT | Control | BMSCT | ||
| FOCUS-HF (2011) | 56.3 ± 8.6 | 60.5 ± 6.4 | 5 (50%) | 4 (20%) | 2.6 ± 0.8 | 2.3 ± 1.2 | 40 ± 3.2 | 37.5 ± 8.2 | Echocardiography |
| Zhao et al (2008) | 59.1 ± 15.7 | 60.3 ± 10.4 | 3 (16.7%) | 3 (16.7%) | 3.3 ± 0.5 | 3.3 ± 0.48 | 36.73 ± 9.15 | 35.8 ± 7.28 | Echocardiography |
| Stamm et al (2007) | 56.3 ± 8.6 | 60.5 ± 6.4 | 5 (50%) | 4 (20%) | 2.6 ± 0.8 | 2.3 ± 1.2 | 39 ± 8.7 | 37.9 ± 10.3 | Echocardiography, MRI |
| Hu et al (2011) | 58.27 ± 8.86 | 56.6 ± 9.721 | NR | NR | 2 ± 0.74 | 3 ± 0.74 | 24.95 ± 8.43 | 22.78 ± 7.7 | MRI |
| Hendrikx et al (2006) | 66.8 ± 9.2 | 63.2 ± 8.5 | 3 (30%) | 0 | — | — | 39.5 ± 5.5 | 42.9 ± 10.1 | MRI, SPECT |
| Patel et al (2005) | 63.6 ± 4.9 | 64.8 ± 3.9 | 2 (10%) | 2 (10%) | 3.4 ± 0.26 | 3.5 ± 0.27 | 30.7 ± 2.5 | 29.4 ± 3.6 | Echocardiography |
| Ang et al (2008) (IC) | 61.3 ± 8.3 | 62.1 ± 8.7 | 2 (10%) | 2 (9.5%) | 20.9 ± 8.9 | 28.5 ± 6.5 | DSE | ||
| Pokushalov et al (2010) | 62 ± 5 | 61 ± 9 | 7 (12.7%) | 8 (14.8%) | 3.5 ± 0.1 | 3.3 ± 0.2 | 26.8 ± 3.8 | 27.8 ± 3.4 | Echocardiography |
| Assmus et al (2006) | 61 ± 9 | 60 | 0 | 4 (11%) | 1.91 ± 0.73 | 2.2 ± 0.67 | 43 ± 13 | 41 ± 11 | LV angiography, MRI |
| Ang et al (2008) | 61.3 ± 8.3 | 64.7 | 2 (10%) | 6 (28.6%) | NYHA III–IV: 3 (15%) | NYHA III–IV: 4 (19%) | 20.9 ± 8.9 | 25.4 ± 8.1 | DSE |
| FOCUS-CCTRN (2012) | 62.32 ± 8.25 | 63.95 ± 10.9 | 2 (6.45%) | 8 (13.11%) | 2.41 ± 0.62 | 2.27 ± 0.63 | 30.19 ± 7.76 | 32.43 ± 9.23 | Echocardiography |
| Study | Randomization/Blinding | Total/BMSCT/Control (n) | Clinical Scenario | Follow-Up Duration (mos) | No. of Cells Transplanted | Control-Intervention | Cointervention |
|---|---|---|---|---|---|---|---|
| Perin et al (2011) | +/+ | 30/20/10 | No-option chronic HF | 3, 6 | 2 × 10 6 | Mock injection | 0 |
| Zhao et al (2008) | +/+ | 36/18/18 | Elective CABG | 6 | 6.59 × 10 8 to 5.12 × 10 8 | Saline injection | Off-pump CABG (61%), remainder on pump |
| Stamm et al (2007) | +/− | 40/22/21 | Elective CABG | 6, 18 | 5.8 × 10 6 | No sham injection | On-pump CABG |
| Hu et al (2011) | +/+ | 60/31/29 | Elective CABG | 6 | 13.17 ± 10.66 × 10 7 | Placebo injection | CABG |
| Hendrikx et al (2006) | +/+ | 20/10/10 | Elective CABG | 4 | 60.25 ± 31.35 × 10 6 | Saline injection | Off-pump CABG |
| Patel et al (2005) | +/− | 20/10/10 | Elective CABG | 6 | 0 | Off-pump CABG | |
| Ang et al (2008) (IC) | +/− | 63/20/21 | Elective CABG | 6 | 84 × 10 6 to 115 × 10 6 | 0 | On-pump CABG |
| Pokushalov et al (2010) | +/− | 109/55/54 | No-option chronic HF | 12 | 41 ± 16 × 10 6 | No sham injection | Medical therapy |
| Assmus et al (2006) | +/+ | 75/28/23 | Elective CABG | 3 | 205 ± 110 × 10 6 | 0 | Medical therapy |
| Ang et al (2008) (IM) | +/+ | 63/20/21 | Elective CABG | 6 | 84 × 10 6 to 115 × 10 6 | 0 | CABG |
| FOCUS-CCTRN (2012) | +/+ | 92/61/31 | No-option chronic HF | 6 | 100 × 10 6 | Bone marrow stem cell injection | Medical therapy |
The methodologic quality of the RCTs was assessed on the basis of the use of adequate methods to generate randomized sequences, allocation concealment, blinding of outcome assessors, percentage lost to follow-up, reporting of withdrawals, and use of intention-to-treat analysis ( Table 3 ). All included studies were adequately randomized, except 1 in which the randomization process was changed in the midst of the trial, and in all trials, outcome assessors were blinded. Placebo or sham injection was administered in 5 trials. On summarizing the data from the 10 RCTs (n = 519), compared with controls, BMSCT by either IM injection or IC infusion improved the LVEF by 4.48% (95% CI 2.43% to 6.53%, p <0.0001, I 2 = 96%) at 6 months. In a meta-analysis of data from 4 trials including 245 patients, BMSCT decreased LV end-systolic volume by −20.64 ml (95% CI −33.21 to −8.07, p <0.001, I 2 = 95%) and LV end-diastolic volume by −16.71 ml (95% CI −31.36 to −2.06, p = 0.03, I 2 = 92%) ( Figure 2 ).
| Study | Randomization Methods | Allocation Concealment | Blinding of Outcome Assessors | Loss of Subjects (n) | Intention-to-Treat Analysis |
|---|---|---|---|---|---|
| FOCUS-HF (2011) | Adequate | Adequate | + | 0 | + |
| Zhao et al (2008) | Adequate | Unclear | + | 0 | + |
| Stamm et al (2007) | Adequate | Adequate | + | 4 | + |
| Hu et al (2011) | Unclear | Adequate | + | 0 | + |
| Hendrikx et al (2006) | Adequate | Unclear | + | 3 | + |
| Patel et al (2005) | Adequate | Adequate | + | 0 | + |
| Ang et al (2008) | Unclear | Unclear | + | 2 | + |
| Pokushalov et al (2010) | Inadequate | Inadequate | + | 10 | + |
| Assmus et al (2006) | Unclear | Unclear | + | 4 | + |
| FOCUS-CCTRN (2012) | Adequate | Adequate | + | 2 | + |
Across all included studies, patients in BMSCT groups had a mean NYHA class of 2.25 ± 0.70, which was comparable with that of patients in control groups (2.41 ± 0.62). To make them more clinically meaningful, ordinal NYHA class data from 5 trials including 246 patients were collapsed into a binary variable with 2 levels: patients who improved in NYHA class by ≥1 class and those with no improvement or worsening of NYHA class. The odds of improvement in NYHA class by ≥1 class with BMSCT were 3.87 (95% CI 2.35 to 6.39, p = 0.00001). In addition, Minnesota Living With Heart Failure Questionnaire (MLHFQ) score decreased by −24.11 (95% CI −47.02 to 1.19, p = 0.04), and 6-minute walking distance (6MWD) increased by 103.9 m (95% CI −31.24 to 239.23, p = 0.13). Mean Canadian Cardiovascular Society (CCS) angina class increased by −0.73 (95% CI −1.55 to 0.09, p = 0.08).
The safety data reported in the included trials are listed in Table 4 . Except for inducible arrhythmia, reported by Hendrikx et al, and recurrent ventricular tachycardia, observed with IM injection in another study, BMSCT was not associated with an increased arrhythmia burden. One patient developed hematoma at the bone marrow harvest site. Overall, BMSCT was not associated with increased risk for death, myocardial infarction, or procedure-related complications.
| Study | Death | HF Hospitalization | Arrhythmia | MI | Stroke | Atrial Fibrillation | Respiratory Failure | Procedure Related | Renal Complications |
|---|---|---|---|---|---|---|---|---|---|
| FOCUS-CCTRN (2012) | C: 0; B: 1 death due to pump failure | C: 5; B: 3 | B: 1 recurrent ventricular tachycardia (procedure related) | B: 1 | — | — | — | Dissection of abdominal aorta | |
| FOCUS-HF (2011) | C: 0; B: 0 | C: 1; B: 2 | No ventricular arrhythmia | — | — | — | — | B: 1 received ICD therapy | |
| Zhao et al (2008) | C: 0; B: 1 | — | C: 0; B: 2 with VF | B: 7 (38.9); C: 5 (27.8) | B: 3 (16.7); C: 5 (27.8) | Renal failure: 1 in B; 1 in C | |||
| Stamm et al (2007) | C: 0; B: 0 | — | C: 0 I; B: 0 | C: 0; B: 0 | 1 in safety arm | 0 | — | Rethoracotomy | ARF: 1 in efficacy arm |
| Hu et al (2011) | C: 1; B: 0 | B: 3; C: 3 | |||||||
| Hendrikx et al (2006) | C: 1 (cardiac death); B: 1 (mediastinitis) | B: 6 with VT | |||||||
| Patel et al (2005) | C: 0; B: 0 | C: 0; B: 0 | C: 0; B: 0 | 0 | 0 | — | Hematoma at stem cell harvest site | ||
| Ang et al (2008) (IC) | C: 0; B: 1 | C: 0; B: 0 | C: 0; B: 0 | C: 1 | |||||
| Pokushalov et al (2010) | C: 21/54 (38.9%); B: 6/55 (10.9%) | C: 0, B: 0 | 0 | ||||||
| Assmus et al (2006) (3-mo follow-up) | C: 1; B: 0 | C: 1 | C: 1 | C: 0; B: 0 | C: 1 | ||||
| Ang et al (2008) (IM) | C: 1; B: 0 | C: 0; B: 0 | C: 0; B: 0 | C: 1 | Renal failure: 1 in C |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
