OUP user menu

★ editor's choice

Improved clinical outcome after intracoronary administration of bone-marrow-derived progenitor cells in acute myocardial infarction: final 1-year results of the REPAIR-AMI trial

Volker Schächinger, Sandra Erbs, Albrecht Elsässer, Werner Haberbosch, Rainer Hambrecht, Hans Hölschermann, Jiangtao Yu, Roberto Corti, Detlef G. Mathey, Christian W. Hamm, Tim Süselbeck, Nikos Werner, Jürgen Haase, Jörg Neuzner, Alfried Germing, Bernd Mark, Birgit Assmus, Torsten Tonn, Stefanie Dimmeler, Andreas M. Zeiher
DOI: http://dx.doi.org/10.1093/eurheartj/ehl388 2775-2783 First published online: 10 November 2006


Aims To investigate the clinical outcome after intracoronary administration of autologous progenitor cells in patients with acute myocardial infarction (AMI).

Methods and results Using a double-blind, placebo-controlled multicentre trial design, we randomized 204 patients with successfully reperfused AMI to receive intracoronary infusion of bone-marrow-derived progenitor cells (BMCs) or placebo medium into the infarct artery 3–7 days after successful infarct reperfusion therapy. At 12 months, the pre-specified cumulative endpoint of death, myocardial infarction, or necessity for revascularization was significantly reduced in the BMC group compared with placebo (P=0.009). Likewise, the combined endpoint death, recurrence of myocardial infarction, and rehospitalization for heart failure was significantly (P=0.006) reduced in patients receiving intracoronary BMC administration. Intracoronary administration of BMC remained a significant predictor of a favourable clinical outcome by Cox regression analysis, adjusting for classical predictors of poor outcome after AMI.

Conclusion Intracoronary administration of BMCs is associated with a significant reduction of the occurrence of major adverse cardiovascular events after AMI. Large-scale studies are warranted to confirm the effects of BMC administration on mortality and morbidity in patients with AMIs.

  • Myocardial infarction
  • Prognosis
  • Cells
  • Catheterization


Experimental studies suggested that intravascular or intramyocardial administration of bone-marrow-derived progenitor cells (BMCs) or blood-derived progenitor cells may contribute to functional regeneration of infarcted myocardium and enhance neovascularization of ischaemic myocardium.14 Initial clinical studies demonstrated that intracoronary infusion of progenitor cells is feasible and may beneficially affect left ventricular contractile recovery510 or infarct size11 in patients with acute myocardial infarction (AMI). Recently, using a double-blind, placebo-controlled, randomized multicentre trial design, the REPAIR-AMI trial12 confirmed that intracoronary infusion of enriched BMCs is associated with improved recovery of global left ventricular contractile function within 4 months after a state-of-the-art treated AMI. However, it is unknown whether the enhanced contractile recovery of left ventricular function may also translate into an improved clinical outcome. Therefore, we analysed the occurrence of clinical events within 1 year after intracoronary infusion of BMCs in the REPAIR-AMI trial.


Study population and protocol

The study protocol has been described in detail previously.12,13 In brief, patients aged 18–80 were eligible for inclusion into the study, if they had an acute ST-elevation myocardial infarction successfully reperfused with stent implantation with a residual significant left ventricular regional wall motion abnormality (ejection fraction ≤45% by visual estimate). The Ethics Review Board of each individual participating centre approved the protocol and the study was conducted in accordance with the Declaration of Helsinki. The study is registered with clinicaltrials.gov, number NCT00279175.

In this double-blind, placebo-controlled randomized trial, performed in 17 centres, at a median of 4 days after AMI reperfusion therapy, bone-marrow aspiration was performed in 204 patients and the aspirate was sent to a central cell-processing laboratory (Institute for Transfusion Medicine, Frankfurt, Germany), where patients were randomized to placebo medium or BMC receiving groups. Cell processing has been described in detail elsewhere.12,13 BMC or placebo was infused using a stop-flow technique via an over-the-wire balloon, positioned in the infarct-related coronary artery within the segment of the previously implanted stent.

The primary endpoint, change of left ventricular ejection fraction by left ventricular angiography assessed at 4 months, as well as the 12 months clinical outcome of 168 of the 204 patients has been previously reported.12 For analysis of the primary endpoint, the study analysing centre has been unblinded after all 4 months data had been collected and finally analysed. However, patients, study centres, investigators, and those entering the data into the database still remained blinded until 12 months follow-up had been completed and clinical events had been finally categorized.

Clinical events

The definitions of individual clinical events are listed in detail in the supplementary data. In brief, death of any cause and type of death (cardiac, cardiovascular, or non-cardiovascular) were assessed. Repeated myocardial infarction was defined as creatinine kinase elevation above two times the upper normal limit with a significant MB fraction (>6%) or new Q-waves in the ECG. Revascularization procedures [percutaneous coronary intervention (PCI) or CABG] were defined as target vessel or non-target vessel revascularizations. Indications for revascularizations were at the discretion of the investigators. Stent thrombosis was defined as evidence of acute recurrent myocardial ischaemia and documented index infarct artery occlusion at the site of stent implantation. Rehospitalization due to heart failure was defined as hospitalization with typical clinical findings of heart failure requiring the addition of medication for the treatment of heart failure. Ventricular arrhythmias were defined as any documentation of ventricular fibrillation or ventricular tachycardia by ECG, Holter ECG, or pacemaker/ICD report. By protocol, patients were scheduled to undergo a 24 h Holter ECG at 4 months and 12 months after study inclusion. Syncope was defined as a transient, self-limiting loss of consciousness. Stroke was defined as a focal neurological defect that persists for ≥24 h confirmed by a neurologist or an imaging procedure. Cancer was defined as any new or recurrent neoplastic disease confirmed by histology.


Results for the primary endpoint, defined as the absolute improvement in global left ventricular ejection fraction from baseline to 4 months, measured by quantitative left ventricular angiography have been published previously.12 Pre-specified clinical endpoints included major adverse events (defined as death, repeated myocardial infarction, or any revascularization procedure) and rehospitalization for heart failure. Other clinical events or combined endpoints were assessed as a post hoc analysis. For analysis per patient, including Cox regression analysis and Kaplan–Meier analysis, only the first event of each patient was included into the analysis. The last 12 months follow-up was obtained on 21 August 2006.


Continuous variables are presented as mean ± SD (if not stated otherwise). Categorical variables were compared with the χ2 test or Fisher's exact test, as appropriate. Time-dependent event rates were estimated by Kaplan–Meier survival curves for the randomization status and P-values were determined by use of log rank statistics. Plotting log-minus-log function for each randomization group with respect to the combined clinical endpoint death, recurrent myocardial infarction, and revascularization procedures indicated proportional hazard. Therefore, Cox regression analysis was used to assess the hazard ratios of the randomization status (unadjusted) and, furthermore, after adjustment of additional single or multiple other variables potentially related to the clinical endpoint to be assessed. As such variables, we selected predictors commonly known to be associated with a poor clinical outcome after an AMI, namely age, diabetes mellitus, baseline ejection fraction, baseline endsystolic volume, and the use of aldosterone antagonist at hospital discharge as well as variables demonstrating an interaction with the treatment effect of BMCs on the primary endpoint (improvement of left ventricular function), namely days to intracoronary infusion and, once again, baseline ejection fraction.12 Statistical significance was assumed if P<0.05. All reported P-values are two-sided. Statistical analyses were performed using SPSS (Version 14.0, SPSS Inc.).


Patient population and baseline characteristics

A total of 204 patients have been randomized (103 to placebo and 101 to BMC). There were no significant differences in baseline characteristics (Table 1) (a detailed list of baseline characteristics and the study flow chart are available in reference12). Likewise, study medication did not significantly differ between placebo and BMC at hospital discharge and at 1-year follow-up, with the exception of aldosterone antagonists, which were significantly less frequently used in the BMC group (Table 1).

View this table:
Table 1

Baseline characteristics and concomitant therapy

Placebo (n=103)BMC (n=101)P-value
Risk factors
 Age, years57±11 (55)55±11 (56)
 Male gender84828382
 Smoking (active or former)70687473
Infarct treatment
 PCI additional non-infarct vessel stenosis12121010
 Time symptom onset to first reperfusion therapy, h7.0±6.5 (4.5)7.5±8.0 (4.5)
 TIMI flow pre-PCI0.75±1.1 (0)0.85±1.2 (0)
 Drug eluting stent13131616
 GP IIb/IIIa inhibitor during acute PCI81798281
 Maximal creatinine kinase (times UNL)19.3±13.7 (16.2)19.8±17.2 (14.6)
 Congestive heart failure16161111
 Intravenous catecholamine treatment87.855
Discharge medicationn=102n=100
 Oral anticoagulation32.944.00.72
 AT II blocker98.8550.29
 Calcium antagonist11.033.00.37
 Aldosterone antagonist1616550.013
 Non-statin lipid lowering drug22.011.01.0
Medication at 12 monthsn=94n=99
 Oral anticoagulation66.433.00.32
 AT II blocker151611110.32
 Calcium antagonist77.488.10.87
 Aldosterone antagonist121344.00.028
 Non-statin lipid lowering drug141516160.81
  • Values are expressed as mean ± SD. Median values are in parenthesis.

  • UNL, upper normal limit.

Four months follow-up data could be obtained in all 204 patients; at 12 months follow-up, there were three patients lost to follow-up in the placebo group, whereas data could be completely acquired in the BMC group (Figure 1).

Clinical events at 1 year

A total of eight deaths occurred during 1-year follow-up—six in the placebo group and two in the BMC group (Table 2). There was a significant difference in recurring myocardial infarctions between the two groups (P=0.029): none of the patients in the BMC group experienced a myocardial re-infarction, whereas six patients in the placebo group suffered a total of 8 myocardial infarctions during follow-up. Of those, six were located to the target vessel supplying the index infarct area, whereas two were located to a non-target vessel.

View this table:
Table 2

Clinical events during 1-year follow-up

Number of patients with eventsPlaceboa (n=103)BMC (n=101)P-value
 Cardiac death43.922.0
  Myocardial rupture11.011.0
  Myocardial infarction11.00
  Sudden death11.011.0
  Heart failure11.00
 Cardiovascular death (stroke)11.00
 Non-cardiovascular death (cancer)11.00
Myocardial infarction65.800.029b
Rehospitalization for heart failure32.900.25b
 Target vessel revascularization262516160.097c
  Stent thrombosis32.911.00.62b
 Non-target vessel revascularization161676.90.052c
Documented ventricular arrhythmia or syncope54.955.01.0b
 Ventricular arrhythmia43.955.00.75b
Combined events
Combined death or myocardial infarction109.722.00.019c
Combined death, infarction, or any revascularization424124240.009c
Combined death, infarction, or infarct vessel revascularization313018180.040c
Combined death, infarction, or rehospitalization for heart failure121222.00.006c
  • aIn 3 patients, only the 4 months follow-up was available.

  • bFishers exact test.

  • cChi-square test.

Likewise, additional revascularization procedures were significantly less frequently needed in the BMC group vs. the placebo group (placebo: 38 revascularizations in 37 patients vs. BMC: 25 revascularizations in 22 patients, P=0.026 comparing patients with revascularization procedures, and P=0.038 comparing number of revascularization procedures per patient). There was a trend—although statistically not significant—towards a reduction in both target vessel as well as non-target vessel revascularization (Table 2). The reasons for a target vessel revascularization were restenosis in 21 placebo and 15 BMC revascularizations, stent thrombosis in three vs. one revascularization, respectively, and a de novo intervention in two revascularizations in both groups. Reasons for non-target vessel revascularizations were restenosis in two placebo and one BMC revascularizations and a de novo intervention in 15 vs. eight revascularizations, respectively. One patient in the placebo group underwent coronary artery bypass grafting, whereas all other revascularizations were PCIs.

There were no statistically significant differences between the two groups with respect to the incidence of stent thrombosis (all after bare metal stent implantation), ventricular arrhythmia or syncope, stroke or cancer during follow-up (Table 2). There was one subacute stent thrombosis (within 30 days) in each group and two late stent thromboses (day 56 and 134, respectively) in the placebo group. The stent thrombosis in the BMC group (day 7) did not result in a recurrent myocardial infarction due to immediate successful recanalization.

The pre-specified combined endpoint death, recurrence of myocardial infarction, or revascularization procedures was significantly reduced in the BMC group compared with the placebo group (P=0.009). Likewise, the combined endpoint death, recurrence of myocardial infarction, and rehospitalization for heart failure was significantly reduced (P=0.006).

These findings are corroborated by time-dependent analyses such as Cox regression analysis (Figure 2) and Kaplan–Meier analyses (Figure 3). Importantly, time-dependent analyses clearly demonstrate that, in the placebo group, the combined endpoints death, recurrence of myocardial infarction, and rehospitalization for heart failure continuously accumulate during follow-up, whereas occurrence of these events is limited to the first 20 days in the BMC group (see Supplementary material online, Table S1 for individual events).

Figure 2

Hazard ratios of BMC infusion therapy with respect to individual and combined clinical events (Cox regression analysis).

Figure 3

Kaplan–Meier event-free survival analysis. (A) Death, recurrence of myocardial infarction, or revascularization procedures. (B) Death, recurrence of myocardial infarction, or rehospitalization for heart failure.

Predictors of clinical outcome

There was a trend, although statistically not significant, towards increased age to be associated with an adverse outcome as assessed by the pre-specified combined clinical endpoint death, recurrence of myocardial infarction, or revascularization procedures (Table 3). However, after adjusting for any of the selected variables, randomization to the BMC group remained a significant predictor of a reduced cardiovascular event rate. Most importantly, after adjustment of Cox regression analysis to all selected variables, randomization to the BMC group remained the only significant predictor of an improved clinical outcome as assessed by the combined endpoint death, recurrence of myocardial infarction, or revascularization procedures (P=0.022).

View this table:
Table 3

Unadjusted and adjusted hazard ratios of BMC infusion therapy with respect to death, recurrence of myocardial infarction, or revascularization procedure (Cox regression analyses)

Hazard ratio95% CIP-value
 Randomization to BMC0.520.32–0.860.011
Adjustment for age
 Randomization to BMC0.540.32–0.890.015
Adjustment for diabetes mellitus
 Randomization to BMC0.520.31–0.860.010
 Diabetes mellitus0.880.46–1.680.69
Adjustment for days to infusiona
 Randomization to BMC0.530.32–0.880.014
 Days to infusion1.180.95–1.470.16
Adjustment for baseline ejection fractionb
 Randomization to BMC0.530.32–0.880.015
 Baseline ejection fraction0.980.96–1.010.18
Adjustment for baseline endsystolic volumeb
 Randomization to BMC0.520.31–0.870.012
 Baseline endsystolic volume1.00.99–1.010.96
Adjustment for aldosterone antagonist at hospital dischargec
 Randomization to BMC0.550.32–0.930.026
 Aldosterone antagonist at discharge1.490.76–2.930.24
Adjustment for all variables listed aboved
 Randomization to BMC0.530.30–0.910.022
 Diabetes mellitus0.790.38 -1.640.53
 Baseline ejection fraction0.980.95–1.020.35
 Baseline endsystolic volume0.9950.98–1.010.41
 Aldosterone antagonist at discharge1.310.62–2.770.47
 Days to infusion1.140.90–1.440.27
  • a≤3, 4, 5 or ≥6 days, available in 202 patients with intracoronary infusion attempted.

  • bAvailable in 199 patients with baseline left ventricular angiogram.

  • cAvailable in 202 patients alive at hospital discharge.

  • dAnalysis includes 61 events in 197 patients with all variables available.


The principal finding of our study is that intracoronary administration of BMCs is associated with a significantly improved clinical outcome in patients with AMI. Clearly, the study was not powered to detect any significant differences in individual major adverse cardiac events. However, it is indeed reassuring that for every individual endpoint such as death, recurrence of myocardial infarction, and rehospitalization for heart failure, there was a trend, although statistically not significant, in favour of the BMC group. Moreover, both adjustment for other single or multiple potential predictors of outcome revealed that intracoronary administration of BMCs is significantly predicting an improved clinical outcome as assessed by the combined clinical endpoint death, recurrence of myocardial infarction, or revascularization procedures. Thus, these data indeed suggest that the enhanced contractile recovery of left ventricular function documented at 4 months12 may translate into an improved clinical outcome after 1 year.

Obviously, the present clinical study cannot disclose the cellular mechanisms responsible for the observed reduction in cardiovascular events. Experimentally, injection of BMCs was shown to stimulate neovascularization of ischaemic myocardium,3 to prevent cardiomyocyte apoptosis,4 and to alter scar formation by reducing the development of myocardial fibrosis.2,14 Interestingly, measuring coronary blood flow reserve in response to adenosine infusion using an intracoronary Doppler wire in a subset of our patients (n=54) demonstrated a significant improvement in the patients receiving BMCs compared with placebo infusion.15 Thus, in line with our previous observations in the TOPCARE-AMI pilot trial,16 intracoronary administration of BMCs appears to be associated with improved perfusion capacity of the infarct-related artery 4 months after an AMI. It is well established that coronary blood flow regulation is of prognostic importance after stent implantation17,18 as well as after myocardial infarction.19 Mechanistically, the improved perfusion capacity may increase epicardial artery shear stress and stimulate the endothelium to release nitric oxide (NO), which exerts antiatherosclerotic functions20,21 and may counteract the process of restenosis development and atherosclerotic disease progression. As such, the effects of intracoronary BMC administration on coronary blood flow reserve may well explain the reduction in revascularization procedures observed in the present study. Likewise, the increase in vascular conductance capacity may also reduce ongoing cardiomyocyte apoptosis by providing enhanced supply of oxygen and nutrients specifically to the border zone of the infarcted left ventricular segments.

Finally, the enhanced recovery of left ventricular contractile function paralleled by an abrogation of left ventricular endsystolic volume expansion observed at 4 months after intracoronary BMC administration12 may limit activation of the neurohumoral system in the post-infarction period and, by altering the levels of circulating cytokines, contribute to a reduction in the recurrence of myocardial infarction and death over time. It has been recently established that recurrent infarction causes the most deaths following myocardial infarction with left ventricular systolic dysfunction.22 Indeed, careful examination of the time course of the occurrence of death, re-infarction, and hospitalization for heart failure in the present study (Figure 3B and Supplementary material online, Table S1) indicates that these events continue to accumulate over the 1-year follow-up period in the placebo group, but not in the BMC group.

In contrast to the REPAIR-AMI trial, another controlled trial (ASTAMI), without a blinded placebo group, did not find a significant improvement in left ventricular function 6 months after intracoronary administration of BMC in patients with acute anterior myocardial infarction.23 The reason of the failure of the ASTAMI trial to show any benefit on left ventricular contractile recovery as well as on cardiovascular events is not clear, since no pre-clinical functional testing of the cells used for the ASTAMI trial was reported. However, it might well be that subtle differences in cell processing and storage may have accounted for the differing results. Indeed, it is noteworthy that despite using identical volumes of bone-marrow aspiration (50 mL) in both trials, patients in the REPAIR-AMI trial received an approximately three-fold higher number of mononuclear BMC and a 3.5-fold higher number of BMC bearing the haematopoietic stem cell marker CD34.12,23 Thus, there are very obvious differences in cell processing and storage between the two trials. Given the pivotal role of preserved functionality of isolated BMCs for mediating infarct size reduction in patients with AMI,7 it is at least conceivable that different functional BMC characteristics may have contributed to the different results of these trials.

In conclusion, intracoronary infusion of BMCs is associated with improved clinical outcome in patients with successfully revascularized AMI. Although clinical outcome (death, myocardial infarction, or revascularization) was a pre-specified combined endpoint in the present trial, the study was not powered to detect statistically significant differences in individual major adverse cardiac events. Therefore, large-scale, prospective, clinical endpoint trials are necessary to confirm the effects of intracoronary BMC administration on mortality and morbidity in patients with AMIs.

Supplementary material

Supplementary material is available at European Heart Journal online.


We are greatly indebted to Heike Braun for excellent trial organization and expert assistance. We also thank Hans Martin (Department of Haematology) for expert advice, Tina Rasper and Nicola Krzossok for technical assistance in cell processing, Wilhelm Sauermann (Datamap) for very helpful statistical advice, Tayfun Aybek for database assistance, and Florian Seeger and Jörg Honold for logistic support. In addition, we would like to thank the following non-profit research organizations, that have supported the research leading to the initiation of the study: Alfried Krupp Stiftung, German Research Foundation (D.F.G.) and European Vascular Genomics Network (E.V.G.N.). Supported by an unrestricted research grant from Guidant. Guidant provided balloon catheters, and Eli Lilly provided the abciximab.

Conflict of interest: V.S. reports having received consulting fees from Guidant. S.D. reports having received consulting fees from Guidant and Genzyme. A.M.Z. reports having received consulting fees from Guidant. S.D. and A.M.Z. report that they are cofounders of t2cure, a for-profit company focused on regenerative therapies for cardiovascular disease. They serve as scientific advisers and are shareholders.


Trial investigators and committee members

Germany: Herzzentrum Leipzig (S.E./R.H.) (35 patients included); J.W. Goethe Universität Frankfurt (V.S./A.M. Z.) (28); Kerckhoff Klinik Bad Nauheim (A.E./M. Stanisch/C.W.H.) (22); Zentralklinikum Suhl (W.H.) (18); Universitätsklinikum Giessen (H.H./H. Tillmanns) (15); Zentralklinkum Bad Berka (J.Y./B. Lauer) (14); Hamburg University Cardiovascular Center (D.G.M./T. Tübler) (13); Universitätsklinikum Mannheim (T.S./M. Brückmann/K. Haase) (11); Universitätsklinikum Homburg/Saar (G. Nickenig/N.W./M. Böhm) (9); Kardiologisches Centrum Rotes Kreuz, Frankfurt (J.H.) (8); Klinikum Kassel (C. Hansen/J.N.) (5); BG Klinik, Universität Bochum (A.G./A. Mügge) (4); Herzzentrum Ludwigshafen (B.M./J. Senges) (4); Herzzentrum NRW, Bad Oeynhausen (C. Hoffmann/M. Farr/D. Horstkotte) (3); Klinikum Lippe (A. Cuneo/U. Tebbe) (1); Universitätsklinik Mainz (S. Genth-Zotz/T. Münzel) (1).

Switzerland: Universitätsspital Zürich (R.C./T. Lüscher) (13).

Steering Committee: A.M.Z. (Principle Investigator), S.D., V.S., W.H., Karl K. Haase, D.G.M., and R.H..

Study Coordinating Center: Heike Braun, V.S., Frankfurt, Germany.

Central Cell Processing Center: T.T., E. Seifried, Institute for Transfusion Medicine and Immunohematology, Red Cross Blood Donor Service Baden–Württemberg–Hessen, Frankfurt, Germany.

Angio Core Lab: B.A., A.M.Z., Frankfurt, Germany.

Safety Committee: Tassilo Bonzel (Fulda, Germany), Wolfgang Kasper (Wiesbaden, Germany).


  • Present address. Universitätsklinikum Bonn, Med. Klinik II—kardiologie/Pneumologie, Sigmund-Freud-Straße 25, 53127 Bonn, Germany.


View Abstract