Objective: To investigate long-term safety and efficacy after intracoronary injection of autologous mononuclear bone marrow cells (mBMCs) in acute myocardial infarction (AMI).
Design: Randomised, controlled trial.
Setting: Two university hospitals in Oslo, Norway.
Patients: Patients from the Autologous Stem cell Transplantation in Acute Myocardial Infarction (ASTAMI) study were re-assessed 3 years after inclusion.
Interventions: 100 patients with anterior wall ST-elevation myocardial infarction treated with acute percutaneous coronary intervention (PCI) were randomised to receive intracoronary injection of mBMCs (n = 50) or not (n = 50).
Main outcome measures: Change in left ventricular (LV) ejection fraction (primary). Change in exercise capacity (peak VO2) and quality of life (secondary). Infarct size (additional aim), and safety.
Results: The rates of adverse clinical events in the groups were low and equal. There were no significant differences between groups in change of global LV systolic function by echocardiography or magnetic resonance imaging (MRI) during the follow-up. On exercise testing, the mBMC-treated patients had larger improvement in exercise time from 2–3 weeks to 3 years (1.5 minutes vs 0.6 minutes, p = 0.05), but the change in peak oxygen consumption did not differ (3.0 ml/kg/min vs 3.1 ml/kg/min, p = 0.75).
Conclusion: The results indicate that intracoronary mBMC treatment in AMI is safe in the long term. A small improvement in exercise time in the mBMC group was found, but no other effects of treatment could be identified 3 years after cell therapy.
Statistics from Altmetric.com
Despite advances in reperfusion strategies and medical treatment, myocardial infarction and subsequent heart failure remain major causes of morbidity and mortality in industrialised countries. The use of cell therapy to promote myocardial repair has gained profound scientific and public interest. The first study on intracoronary mononuclear bone marrow cells (mBMC) therapy shortly after acute myocardial infarction (AMI) in humans was reported by Strauer et al in 2002.1 Later, several randomised, controlled trials using the same method of cell delivery have been published, as recently reviewed.2 The primary endpoint has generally been the change in left ventricular (LV) ejection fraction (EF) after 3–6 months. Feasibility of this treatment has been confirmed, and it seems safe in the short term. However, the efficacy results vary between studies. Possible explanations for these differences are small study samples, different imaging techniques and differences in timing of treatment, cell dose, placebo treatment or cell processing protocols.3 4 5 6 7 Meta-analyses have found a modest beneficial effect on EF after 6 months.2 8 9 Strauer’s group reported a sustained beneficial effect of mBMC therapy on LV function, exercise capacity and mortality in the BALANCE study after 5 years, but the study was not randomised.10 Long-term results from randomised-controlled trials have only been published from 60 patients in the BOOST trial, where the relative improvement in EF observed in the mBMC group at 6 months was no longer significant at 18 months.11 Concerns have been raised about accelerated atherosclerosis, intramyocardial calcifications and risk for arrhythmias.12 13 14 Long-term data on safety and efficacy of this treatment are needed. Thus, we present the 3-year follow-up results from the Autologous Stem cell Transplantation in Acute Myocardial Infarction (ASTAMI) study.
The study design has been described previously.15 Briefly, patients presenting with their first ST-elevation AMI and successful acute percutaneous coronary intervention (PCI) with stent on the left anterior descending artery (LAD), were randomised to intracoronary injection of autologous mBMCs 4–8 days after the acute event (n = 50), or to a control group where neither bone marrow aspiration nor sham coronary procedure were performed (n = 50) (fig 1). Three patients allocated to the mBMC group did not receive intracoronary cell injections (due to stent-thrombosis preceding the cell infusion in two patients, cell viability <90% in one patient). The primary endpoint in ASTAMI was the change in LVEF from baseline to 6 months measured by single photon emission computed tomography (SPECT). Echocardiography and magnetic resonance imaging (MRI) were used for serial assessment of LV function. Secondary endpoints were changes in exercise capacity and quality-of-life (QoL). Clinical events, infarct size and multiple inflammatory and other biochemical markers were also evaluated. Echocardiography and clinical status were controlled after 12 months. No significant difference between groups in change in EF, clinical events or QoL was found. mBMC-treated patients had a larger increase in exercise time, but change in peak oxygen consumption (VO2) did not differ between groups. The mBMC therapy was associated with a transient increase in markers for inflammation. Coronary angiography at 6 months did not indicate luminal loss in the mBMC group compared to controls. These results have previously been presented elsewhere.15 16 17 18 We now performed a 3-year follow-up of patients from the ASTAMI study to evaluate long-term safety and effects on LV function, exercise capacity, biochemical parameters and quality of life. To minimise patient hazard, SPECT and routine coronary angiography were not performed at 3 years.
Cell preparation and injection
A 50-ml volume of bone marrow was harvested from the iliac crest 4–7 days after AMI. The mBMCs were isolated by Isopaque-Ficoll density gradient centrifugation and resuspended in 0.9% NaCl with 20% autologous heparin-plasma. Median cell viability was 95%, and median number of injected viable cells was 68×106 (25th–75th percentile: 54×106–130×106). The cell products were processed in the accredited cell production facility at Institute for Immunology (IMMI), the National Hospital, Oslo, Norway, under good manufacturing practice (GMP) conditions. A 10-ml volume of cell suspension was injected by percutaneous intervention in the LAD by stop flow technique.1
Transthoracic echocardiography was performed at baseline, 3, 6, 12 months and 3 years using a Vivid 7 cardiac ultrasound scanner with M3S transducer (GE Vingmed, Horten, Norway). LV end-diastolic and end-systolic volumes (EDV, ESV) and EF were calculated by tracing of endocardial contours from apical views and use of a modified Simpson biplane method.19 Wall motion was assessed by use of a 16-segment LV model, excluding the apical cap.20 All segments were given a wall motion score (1 = normal or hyperkinetic, 2 = hypokinetic, 3 = akinetic, 4 = dyskinetic and 5 = aneurismatic). The sum of scores was divided by the number of segments visualised to obtain the wall motion score index (WMSI). All analyses were performed offline by experienced observers blinded to treatment allocation, on dedicated software (GE Echopac). Intra-observer and inter-observer variability were assessed in a random subset of 25 patients. For LVEF, the mean intra-observer variability was 2.5% (SD 4.1%) with intraclass correlation coefficient 0.92 (95% confidence interval 0.83 to 0.96). Mean inter-observer variability was 1.1% (4.9%) with intraclass correlation coefficient 0.92 (0.81 to 0.96).
Patients without contraindications underwent MRI with a 1.5-T scanner (Magnetom Vision Plus or Magnetom Sonata, Siemens Medical Systems, Erlangen, Germany) 2–3 weeks, 6 months and 3 years after randomisation. Cine images in two long-axis (LAX) views, and after 3 years multiple short-axis (SAX) views to completely cover the left ventricle, were acquired (time resolution <50 ms). Corresponding late-enhancement images were obtained in the same session 10–20 minutes after the intravenous injection of 0.2 mmol/kg gadopentetate dimeglumine (Magnevist; Schering AG, Berlin, Germany), with a slice thickness of 8 mm, an increment of 10 mm and in-plane spatial resolution of 1.17×1.17 mm to 1.50 mm. A breath-hold segmented magnetisation-prepared turbo gradient-echo sequence was used, with an inversion time chosen to null the signal of the normal myocardium, typically 210–260 ms. Epicardial and endocardial borders and the infarcted area were manually delineated (PACS; Sectra Imtec AB, Linköping, Sweden). When in doubt, areas were considered infarcted if the pixel intensity was ⩾2 SD above the mean pixel intensity of adjacent normal myocardium. Infarct size and LV mass were determined from a stack of short-axis images covering the entire LV. EDV, ESV and EF were calculated by the biplane area-length method. All analyses were performed by experienced observers blinded to treatment allocation.
Maximal symptom-limited exercise tests were performed 2–3 weeks, 6 months and 3 years after AMI with an electrically braked bicycle ergometer (Jaeger ER900; Viasys Healthcare GmbH, Hochberg, Germany). At 6 months and 3 years, β-blockers, calcium-channel blockers and nitrates were postponed until after the test. Other medication was taken as usual. A 10-minute exercise time was pursued, starting at 25 W or 50 W with a 10 W or 25 W increase every second minute. For each patient the same protocol was used for every test. Oxygen consumption (VO2), CO2 production (VCO2), and ventilation (VE) were measured on a breath-to-breath basis (MVmax 229; Viasys Healthcare GmbH, Germany). Heart rate and ECG were continuously recorded. Blood pressure was measured at the end of each step. For parameters sampled continuously or breath-to-breath, average values from 20-second time intervals were used in the analyses. All parameters were calculated as previously described.16
For assessment of health-related QoL, patients responded to the Norwegian version 1.2 of the SF-36 health survey 2–3 weeks, 6 months, and 3 years after AMI. Scores were weighted and aggregated using normative data from the general Norwegian population to obtain physical component and mental component summary scores.
Continuous data are presented as mean (SD) or median (25th, 75th percentile) as appropriate. Categorical data are presented as frequency (percentage). The Student t tests or Mann-Whitney tests were used to compare groups at baseline. The χ2 or Fisher’s exact tests were used as appropriate. For analysis of continuous data measured at ⩾3 points of time, we used mixed model linear regression.21 Time and the interaction between treatment allocation and time were covariates. Intragroup changes from baseline were evaluated by paired sample t tests. Standard multiple linear regression analysis was used to identify predictors of long-term changes in EF. EpiData entry software version 3.1 and SPSS version 15.0 were utilised. The 3-year follow-up was not predefined in the original ASTAMI protocol. For interpretation of the statistical analysis, the parameters and endpoints defined in the original protocol were carried forward. All analyses were performed according to the intention-to-treat principle, tests were two-sided, and p values <0.05 were considered statistically significant.
Of the 100 patients originally enrolled in the ASTAMI study, two died in the follow-up period. Both deaths occurred during the second year of follow-up, one in the mBMC group due to subdural haematoma, and one in the control group due to acute stent-thrombosis. One patient refused participation in the 3-year follow-up. 97 patients underwent examination as scheduled 37 (3) months after randomisation. Clinical characteristics and medical therapy (table 1) did not differ between the groups, except that a higher proportion of patients in the mBMC group were taking diuretics after 3 years.
The rate of adverse clinical events was low and did not differ between the groups (table 2). All arrhythmic events, heart failure episodes and strokes occurred during the first few months of the study.
By echocardiography (fig 2, table 3), the groups did not differ at baseline. There were no significant effects of mBMC therapy on ESV, EDV, EF or WMSI, an indicator of regional wall motion abnormality, during the study period. EF and WMSI improved significantly in both groups from baseline to 3 months, where after there were only minor changes in systolic function.
Concordantly, by MRI (fig 2, table 3), there was no difference between the groups for EF, EDV, ESV, LV mass or infarct size at 2–3 weeks. No statistically significant effect of treatment was observed for these parameters during 3 years. However, in both groups infarct size was significantly reduced from baseline to 3 years (p<0.001). At 3 years, both long-axis and short-axis cine-sequences were acquired. A post hoc analysis of the correlation between the SAX and LAX methods for calculation of LV volumes was performed. Pearson’s r was for EDV 0.96 (p<0.001), ESV 0.98 (p<0.001) and for EF 0.88 (p<0.001). The relation between the groups in terms of volumes and EF was similar with both methods. No intramyocardial calcifications or tumours were noted.
Subgroup analyses on gender and baseline EF and infarct size, dichotomised at median, were performed, with no significant difference between treatment and control group in change in EF.
We also performed univariate and multivariate linear regression analyses to identify predictors for change in EF (by echocardiography) from baseline to 3 years. Age, gender, hypertension, hypercholesterolaemia, diabetes, smoking, time to PCI, peak creatine kinase isoenzyme (CK-MB), treatment and number of mBMCs (in subgroup analysis only) were explored to fit a multivariate model. EF at baseline, peak CK-MB, time to PCI, smoking and diabetes were independent predictors for change in EF in the multivariate analysis (table 4). In a subgroup analysis of the mBMC treated patients, EF, CK-MB and time to PCI were the only significant independent predictors. The number of mBMCs was significantly correlated with change in EF in the univariate analysis (r = 0.305, p = 0.04), but did not contribute in the multivariate model (standardised β = 0.08, p = 0.49).
On exercise testing (table 5), the groups performed similarly at 2–3 weeks. There was a greater increase in exercise time in the mBMC group than in the control group from 2–3 weeks to 3 years (1.5 minutes vs 0.6 minutes, p = 0.05). Change in peak VO2, VE/VCO2-slope, oxygen uptake efficiency slope (OUES) or other parameters, however, did not differ significantly between the groups. All patients with suspected angina pectoris had coronary angiographies before, or soon after, their 3-year follow-up. None of them had lesions eligible for further revascularisation. Analyses on clinical chemistry did not reveal values of concern regarding safety. Mean results at 3 years were similar in both groups, and within the normal range. Median pro B type natriuretic peptide (pro-BNP) was 24 (11–46) pmol/l in the mBMC group, and 23 (12–34) pmol/l in the control group. QoL did not differ significantly between the groups.
ASTAMI is hitherto the largest randomised, controlled study reporting long-term results after intracoronary mBMC injection in AMI. We found low rates of adverse events, with no differences between groups. This is reassuring in terms of safety, although the study was not powered to detect differences in clinical endpoints. At baseline, median EF was 41.9% by SPECT and median peak CK-MB 324 μg/l. Median peak CK-MB in patients with ST-elevation myocardial infarction (STEMI) screened, but not included, was 225 μg/l. Thus, the study comprises a cohort of patients with significant myocardial infarctions. Patient characteristics were similar in the mBMC treated group and the controls. Medical therapy was also similar between the groups, except that there was a higher proportion of patients on diuretics in the mBMC group. The reason for this difference was that more patients in the mBMC group had combination drugs containing thiazides prescribed as antihypertensives, a trend already present at baseline.
Comprehensive follow-up with echocardiography and MRI did not identify significant effects of intracoronary mBMC therapy on LV volumes, EF or infarct size.
The mBMC group displayed a larger increase in exercise time from baseline to 6 months and 3 years than the control group. Other studies have reported improved fractional flow reserve22 and improved diastolic LV function after mBMC therapy,23 factors that mainly improve cardiac performance during stress. Exercise capacity correlates poorly with LV function at rest.24 Thus, improved exercise capacity could conceivably be a result of cell therapy despite no proved effect on resting EF. Any placebo effect in the mBMC group should also be considered in an open-label trial. However, peak heart rate and peak respiratory exchange ratio (RER) did not differ between groups, indicating that the effort applied was equal between the groups. The predefined endpoint for exercise capacity in the ASTAMI study protocol was change in peak VO2. This parameter was not different between groups (p = 0.75), a finding which was supported by the similar results for VE/VCO2 slope and OUES. Although improvement in exercise time was better in the mBMC group at 6 months and 3 years, it was not consistently supported by the other parameters on exercise capacity and should be considered hypothesis generating only.
Previous meta-analyses on this topic2 8 9 claim an overall modest, but significant increase in EF in mBMC-treated patients. However, the results in different trials are heterogeneous. Methods and results in the earlier randomised-controlled mBMC trials have been discussed previously.3 Only the BOOST25 and REPAIR-AMI26 trials were positive after 3–6 months on the primary endpoint EF. The positive effect was no longer significant after 18 months in the BOOST-trial.11 To our knowledge, no complete long-term results on LV function have been published from REPAIR-AMI. Recently, the REGENT trial27 could not demonstrate significant benefit on EF from unselected mBMCs or selected CD34+ CXCR4+ cells compared to controls. The HEBE study (n = 200) on STEMI patients also turned out neutral (Piek JJ, AHA Scientific Sessions, 2008). The FINNCELL study (n = 80) showed significant (p = 0.05) improvement in EF evaluated by LV angiography and echocardiography after mBMC therapy in patients undergoing late PCI after successful thrombolysis.28 As PCI on the infarct-related artery was performed simultaneously, the effect of cell therapy cannot be directly compared with the other trials. Thus, the clinical data currently available suggest that the beneficial effect of intracoronary mBMC injection in AMI is, at best, modest, and may be transient.
The number of cells delivered and cell processing procedures used in this type of treatment have been discussed.4 5 6 Although a dose-response association could theoretically be expected in cell therapy, two meta-analyses8 9 failed to demonstrate a significant relation between cell dose and change in LVEF. Of note, in the univariate analysis, the number of mBMCs was significantly correlated to the change in LVEF in ASTAMI (r 0.305, p = 0.04). However, when baseline EF, peak CK-MB and time to PCI were entered in the multivariate analysis, the number of mBMCs did not contribute significantly. In general, our findings in the regression analysis are concordant with those in the study by Ndrepepa et al.29
The cell processing protocols applied in clinical studies have been compared in two different publications.4 7 Seeger et al retrospectively compared cell processing protocols used in ASTAMI and REPAIR-AMI, finding higher cell recovery and cell migration capacity with the REPAIR-AMI procedure.4 They suggested that density gradient centrifugation with Lymphoprep (Nycomed) and overnight storage of cells in NaCl with heparin and plasma in ASTAMI explained the differences in efficacy between the protocols. However, when van Beem et al prospectively compared cell processing protocols used in HEBE, ASTAMI and REPAIR-AMI, a similar cell recovery, cell viability and cell function was found with Lymphoprep as used in HEBE and ASTAMI, and Ficoll-Paque (Cambrex) as used in the REPAIR-AMI.7 Cell washing in NaCl and plasma (ASTAMI, HEBE) did not negatively affect cell numbers or cell function when compared to phosphate-buffered saline (REPAIR-AMI). As the HEBE study turned out neutral despite documented high cell numbers and cell quality, the relation between in vitro cell characteristics and clinical efficacy remains unsolved.
As ASTAMI was designed with 80% power to detect a 5% difference between groups on change in EF, smaller effects can not be excluded. It should also be noted that the first MRI was performed 2–3 weeks after AMI. Detection of early effects of the intervention was therefore precluded. On the other hand, the impact from tissue oedema and time to reperfusion at this time point was reduced, giving a more accurate baseline for evaluation of long-term effects of the intervention.
MRI volumes at baseline and 6 months were calculated by the area-length method from long-axis images. Calculation of volumes from short-axis images covering the entire LV is considered a more precise method, but requires a longer time for image acquisition. As SPECT and echocardiography served as the primary imaging modalities for evaluation of EF at baseline and 6 months, the area-length approach was chosen in a trade-off between MRI time consumption and accuracy. As the two methods provide volumes and EF with a high level of correlation,30 an association also confirmed in our post hoc analysis, we do not believe there is any lack of precision in the MRI data that would influence our results and conclusion.
In conclusion, intracoronary mBMC injection after AMI did not improve global LV function or clinical outcome during the 3 years of observation. A moderately larger increase in exercise time is observed in mBMC-treated patients. The treatment appears safe, with no adverse effects observed after 3 years. In our opinion, these results suggest that future trials on cell therapy in ischaemic heart disease should select patients at higher risk, and preferably await cell products with a documented greater therapeutic potential.
We would like to thank Professor Torbjørn Moum, Department of Behavioural Sciences, University of Oslo (UiO) for calculation of QoL scores, Associate Professor Magne Thoresen, Department of Biostatistics, UiO, for statistical advice, and Rita Skårdal, Cecilia Guevara and Maria Fornes for assistance in blood sampling and exercise testing. JOB, KL and SS have received research fellowships from the Norwegian Council on Cardiovascular diseases.
Competing interests None declared.
Ethics approval The study complies with the Declaration of Helsinki, and the protocol was approved by the regional committee for research ethics. All patients gave written, informed consent. The authors had full access to the data and take responsibility for its integrity.
Provenance and Peer review Not commissioned; externally peer reviewed.