This editorial refers to ‘Intracoronary infusion of mononuclear cells from bone marrow or peripheral blood compared with standard therapy in patients after acute myocardial infarction treated by primary percutaneous coronary intervention: results of the randomized controlled HEBE trial’†, by A. Hirsch et al., on page 1736, and ‘Intracoronary autologous mononucleated bone marrow cell infusion for acute myocardial infarction: results of the randomized multicenter BONAMI trial’‡, by J. Roncalli et al., on page 1748
During the last two decades, the substantial decline in short-term mortality in patients with acute myocardial infarction (AMI) has markedly increased the prevalence of patients who are admitted to cardiology wards with left ventricular (LV) dysfunction. Despite state-of-the-art pharmacological and device-based treatment, the injured myocardium often appears unable to reverse massive cumulative cardiac cell death and loss of contractile function. The net result is progressive ventricular dilatation and development of heart failure, which remains a major cause of morbidity and mortality.1
The inadequate endogenous regenerative capacity of the heart following extensive ischaemic injury has stimulated research efforts using stem or progenitor cells as a unique and exciting opportunity for true cardiac repair and perhaps regeneration. The field of clinical cardiac cell therapy has witnessed a remarkable and rapid progress, since the very first reports of adult stem cell transfer in small animal models almost a decade ago. The observations that intramyocardial injection of adult bone marrow cells (BMCs) can improve LV contractile function through regeneration2 or neovascularization3 have triggered great enthusiasm about a possible treatment enabling cell-based biological repair.
The first generation of randomized controlled trials in patients with AMI explored the safety and efficacy of intracoronary injection of autologous mononuclear BMCs at variable time intervals following reperfusion. They almost uniformly focused on global left ventricular ejection fraction (LVEF) as a surrogate marker for efficacy, and have shown mixed and sometimes discrepant results on LV function recovery. Importantly, all of the trials using BMCs thus far have reinforced a message of reassuring safety. In one of the larger double-blind, randomized controlled studies to date, intracoronary infusion of mononuclear BMCs, as compared with placebo infusion, was associated with a 2.5% incremental increase in LVEF at 4 months.4 The effect was later reported to persist at extended follow-up. In contrast, in the smaller ASTAMI study in a comparable patient population with first AMI, no beneficial effect of BMC transfer could be documented using three non-invasive imaging modalities.5 Prior to these reports, the BOOST trial6 and the Leuven AMI trial7 had shown transient beneficial effects on global LV function recovery or on infarct size reduction, respectively, suggestive of accelerated healing processes following BMC transfer. Finally, the REGENT trial randomized 200 patients with anterior AMI to receive an intracoronary infusion of unselected mononucleated BMCs, selected CD34+/CXCR4+ BMCs, or standard therapy.8 The trial failed to show significant differences in LVEF improvement from baseline to 6 months follow-up between both cell groups and the control, but was limited by imbalances in baseline LVEF and incomplete follow-up, with paired MRI analysis available in 117/200 patients. Different outcomes and discrepancies in these first-generation trials are in part attributable to our limited understanding of critical features of progenitor cell preparations and of patients with AMI that are predictive of a favourable response to cell transfer (Figure 1). In an effort to help clear the smoke, a task force of the European Society of Cardiology emphasized in an earlier consensus statement the need for randomized controlled clinical trials, focusing on these specific cell-related and patient-related features.9
In this issue of the European Heart Journal, two important European multicentre, randomized controlled trials, the HEBE trial in The Netherlands10 and the BONAMI trial in France,11 respond to this call and present a refined analysis of a potential role for mononuclear BMCs in the aftermath of AMI. Both trials set out to include patients with poorer prognosis, i.e. larger infarcts with marked residual LV dysfunction, who have greater need for improved post-infarction treatment and represent a more favourable risk–benefit ratio. This strategy obviously makes sense and should facilitate confirmation—if any—of incremental BMC-mediated benefit as reported in subgroup or post-hoc analyses of previous trials.4,7,8
In the HEBE trial,10 initiated by the Interuniversity Cardiology Institute of The Netherlands, 200 patients with AMI and successful percutaneous coronary intervention (PCI) within 12 h after onset of symptoms were randomly assigned to receive intracoronary infusion of mononucleated BMCs, mononucleated cells isolated from peripheral blood (PBMCs), or standard therapy. The final results of this open trial with blinded evaluation of endpoints showed that intracoronary infusion of mononucleated BMCs or PBMCs did not improve global or regional LV systolic function at 4 months, as assessed by MRI. The primary endpoint of the trial, change in regional myocardial function, was defined as the percentage of dysfunctional segments with improved wall thickening at 4 months follow-up. Power calculations were based on the secondary endpoint, change in LVEF, assuming they would match the power for the primary endpoint. The three-arm study had a 90% chance of detecting a change of 6% in LVEF between active treatment and control; thus a smaller incremental improvement, as documented in earlier positive trials,4 could have been missed. However, the absence of trends for improved systolic wall thickening or LVEF in favour of BMCs and the lack of interactions between subgroups and intracoronary cell transfer argues against this possibility.
In the smaller BONAMI trial,11 patients with successfully reperfused AMI, residual LVEF of <45%, and decreased myocardial viability, defined by resting thallium 201-gated single photon emission CT (SPECT) imaging, were randomized to intracoronary BMC infusion or state-of-the-art therapy. The primary endpoint of the study was improvement of myocardial viability 3 months after AMI and the trial was powered to have a 90% chance of detecting an improvement of at least two non-viable myocardial segments between both groups. The BONAMI investigators report a trend in favour of the cell transfer group, with myocardial viability improving in 16/47 patients in the BMC group (34%) compared with 7/43 (16%) in the control group (P = 0.06). Interestingly, a multivariate analysis including major prognostic factors for LV function recovery after AMI suggested that smoking adversely affects the cell-mediated favourable response (P = 0.04), while patients with microvascular obstruction (MVO) seem to derive greater benefit from cell transfer (P = 0.07). Of note, no difference in global LV function recovery at 3 months follow-up was observed with any of the three imaging modalities used (radionuclide angiography, echocardiography, and MRI). Data on ventricular function at more extended follow-up are not available.
How do we integrate the results of both trials in a realistic perspective on cell therapy approaches for patients with ischaemic heart disease? As pointed out by many investigators in the field, the prevailing view after several first-generation trials was that adjunctive BMC therapy may offer greater—if any—benefit to patients with a high unmet clinical need, i.e. those with large infarcts, severely depressed LV function, and a high risk of developing clinical symptoms of heart failure. The HEBE and BONAMI investigators carefully restricted inclusion of first AMI patients to those with significant LV dysfunction (residual LVEF <40–45%), and thus studied a population with considerably greater injury than in all prior randomized trials with the exception of the REGENT study.8 Also, both studies were well matched with regards to baseline and procedural characteristics, and paired analysis of the primary endpoint was obtained in >90% of cases, for which the investigators are to be commended.
Although many would have predicted that enrolling patients with poorer prognosis would markedly facilitate our ability to demonstrate incremental benefit of cell therapy, the data at large seem to speak to the contrary, at least when strictly considering contractile recovery in the first months following the acute ischaemic injury. Is it possible that certain features of complex and heterogeneous cell preparations, such as mixed mononuclear BMCs, and of patients with ischaemic heart disease can predict a favourable response to cell transfer or not?
First, as far as the cell product is concerned, all six participating cell-processing laboratories in the HEBE trial were accredited stem cell laboratories, and a central core laboratory monitored the clonogenic potential and migratory capacity in a small sample of recovered BMCs. Because only the functionality of haematopoietic (CD34+) stem cells was tested in vitro, one cannot exclude that patient- or procedure-related variations in number and function of other important constituent bone marrow-derived cell types, including endothelial and vascular progenitor cells or mesenchymal stem cells, may have affected the results. Several authors have recently reported a negative impact of advanced age, risk factors for atherosclerosis, and red blood cell contamination on BMC functionality, with important repercussions on their in vivo capacity to promote blood flow recovery in a nude mouse model of hindlimb ischaemia.12,13 In the BONAMI trial, the investigators injected a 3-fold lower total cell number (98.3 ± 8.7 × 106 BMCs vs. 296 ± 164 × 106 BMCs in HEBE) but, in the absence of direct dose–response studies and extended in vitro phenotypic and functional characterization, the potential impact of different cell numbers remains uncertain in this trial. One of the interesting observations of the BONAMI trial, however, is the significant negative interaction of the observed BMC effect on myocardial viability with smoking status. Although cigarette smoke has been shown to impair BMC functionality, the clinical impact of this observation needs to be confirmed in future trials.
Secondly, with regards to patient-related characteristics and surrogate endpoints, both studies have made a laudable effort to expand our understanding of potential mechanisms for cell-mediated benefit after AMI. The HEBE investigators standardized a very comprehensive analysis of regional myocardial function and structure using state-of-the-art MRI with blinded core lab analysis, while the BONAMI investigators focused on segmental changes in myocardial viability over time, as evaluated by thallium-201-gated SPECT. In both studies, cell transfer was performed after acquisition of baseline MRI or SPECT images and at a time interval after coronary reperfusion that was shown to confer optimal benefit in previous studies.4 Because the HEBE trial integrated detailed quantitative MRI analysis in the largest patient population with extensive first AMI to date, the lack of a clear BMC-mediated effect on regional function recovery came as a surprise. Previous trials had reported that the most likely mechanism for enhanced cell-mediated functional recovery was greater regional systolic wall thickening or wall motion, resulting from paracrine, pro-angiogenic, and anti-apoptotic trophic effects in the infarct border zone.6,14 As the authors recognize, the short total ischaemic time (median 3.3 h) and short-term follow-up may possibly confound the data, and longer term follow-up is mandatory (and planned at 1 year). We know that maladaptive LV remodelling with progressive dilatation and systolic dysfunction is a slow but progressive process, taking much longer than 4 months to manifest its detrimental consequences. The discrepancy with BMC-mediated accelerated early contractile recovery in the BOOST study remains, but probably reflects the different severity of the AMI populations in both studies.6,10
Alternatively, the complex relationship between contractile recovery and presumed BMC-mediated myocardial cytoprotection in the infarcted myocardium is a cogent reminder of similar intricacies of functional, structural, and metabolic changes in the early post-infarction period.15 This is consistent with the observations in the BONAMI trial and raises the question of whether or not BMCs can enhance myocardial viability, quantitatively assessed as percentage thallium intake per myocardial segment, induce a ‘hibernation-like’ state in reperfused infarcted territory, and retard functional recovery for months. All these mechanistic questions remain at present unanswered but would require appropriately powered future studies, integrating rigorously controlled functional, structural, and metabolic imaging techniques. A closer look at the available MRI study results in post-AMI BMC studies highlights the complexity of such comprehensive mechanistic multicentre studies, because of site-related differences in image acquisition, post-processing, and image analysis protocols.
In the end, and most importantly, we need to remember that the ultimate success or failure of any innovative therapy including bone marrow cell transfer in patients with AMI, will depend on its ability to show clinical efficacy rather than on surrogate endpoints and the imputed mechanism of the effect. The consistently reassuring safety profile of intracoronary BMC transfer with accumulating long-term clinical follow-up should pave the way for large-scale clinical outcome studies, that are adequately powered to detect a significant reduction—if any—in the combined endpoint of death, recurrent myocardial infarction, and hospital admission for heart failure. This view has been articulated in the earlier consensus report of the task force of the European Society of Cardiology.9 The proof of the pudding remains in the eating.
Conflict of interest: none declared.