Although most cardiac cell therapy trials have focused on patients with acute myocardial infarction, attempts at “regenerating” chronically failing hearts in patients having suffered extensive myocardial infarctions have also been performed. In practice, these studies have entailed the use of skeletal myoblasts and bone marrow-derived cells.

Skeletal Myoblast Trials

After almost a decade of experimental studies, clinical trials of myoblast transplantation started in June, 2000, when we performed the first human transplantation of autologous myoblasts in a patient with severe ischemic heart failure [1] This case initiated a 10-patient series in which an average of 871 million cells, of which 87% were myoblasts, were injected in postinfarction scars [2]. Three other adjunct-to-coronary artery bypass grafting (CABG) were then performed [3–5]. Whereas the patient profile and technique of open-chest multiple injections were very similar to those used in our study, the number of transplanted myoblasts was highly variable (221 × 106 in the study of Gavira et al. [3] from 4 ± 105 to 5 ± 107 in the study of Siminiak et al. [4], 1, 3, 10, 30 × 107 and 3 × 108 in the dose-escalating study of Dib et al. [5]). It is important that the protocol of these three studies also differed from ours in that it systemically entailed a concomitant revascularization of the myoblast-injected areas.

Put together, these studies primarily demonstrated the feasibility of the procedure (i.e., the possibility of growing several hundreds million cells from a 10-g muscular biopsy under good manufacturing practice (GMP) conditions and within a 2- to 3-week time frame) as well as the safety of multiple needle punctures in the postinfarction scar and along its borders. Likewise, none of the myoblast-injected patients has developed a cardiac tumor (our longest survivor was operated on in December 2000). Indeed, the only safety concern has been an increased risk of postoperative sustained ventricular tachycardia [2, 4], and this susceptibility to arrhythmias following myoblast transplantation has been later confirmed in rat experiments [6]. Currently, the most commonly accepted mechanism of these arrhythmias is the electrical insulation of myoblast clusters from the surrounding cardiomyocytes [7] leading to a slowing of conduction and subsequent reentries [8]. This hypothesis is primarily supported by data derived from co-culture experiments showing that myoblast transfection with connexin 43 decreases arrhythmogenicity [8]. However, the origin of these posttransplantation arrhythmias may not be univocal because of the possible involvement of the intrinsically arrhythmogenic myocardial substrate characteristic of heart failure and the role of needle-induced tissue disruption and inflammation that may further contribute to conduction blocks [9]. The location of myoblast injections could also modulate the proarrhythmic potential of the procedure as those lining the border zone of the scar seem less arrhythmogenic than those performed in its core [10].

Although these initial studies were neither designed nor powered to provide efficacy data, the functional effects of myoblast injections were nevertheless assessed up to 4 years [5] and even later (58 months in our trial) [11]. Outcomes were found to range from stabilization of left ventrical (LV) ejection fraction and volumes [11] to improvements in regional and global LV function from baseline values [3, 4] and, occasionally, in metabolic viability of transplanted areas, as assessed by positron emission tomography and magnetic resonance imaging (MRI) [3, 5]. It is clear, however, that the small size of these series, their open-label type of design and the lack of controls made these data inconclusive.

To overcome these hurdles, we have implemented a randomized, double blind, placebo-controlled trial (MAGIC, an acronym for myoblast autologous grafting in ischemic cardiomyopathy), which involved 21 centers in Europe and included patients with severe left ventricular dysfunction, a postinfarction nonviable scar, and an indication for CABG. Muscular biopsies were cultured in two core laboratories and 3 weeks later, either 400 or 800 million cells or a placebo solution were injected in approximately 30 sites in the core and the margins of the infarct area during the bypass operation. Of note, an implantable cardioverter defibrillator (ICD) was implanted in every patient before hospital discharge and an independent blinded committee then adjudicated ventricular arrhythmias detected by the ICD read-outs. Out of 120 randomized patients, 97 were effectively treated. At the 6-month study point, the proportion of patients who had experienced arrhythmias was not significantly different between the myoblast-treated and the placebo-injected groups despite a trend toward a greater incidence of these events early after operation in the myoblast-treated groups. In terms of efficacy, the trial failed to meet its primary end point as neither regional nor global LV function, as assessed blindly by echocardiography in a core laboratory, were significantly improved by myoblast injections, regardless of the dose, compared with controls; however, the highest dose of cells resulted in a significant reversal of remodeling, evidenced by a decrease in LV end-diastolic and endystolic volumes (a prespecified secondary end point) compared with the placebo group [12]. Overall, these mixed outcomes reflect a commonly observed discrepancy between some signals supporting the proof of concept and the lack of translation of these effects in a clinically meaningful improvement of LV function.

In parallel to these surgical trials, three phase-I catheter-based studies have been reported. One has entailed administration of myoblasts through the coronary sinus with a dedicated catheter, which allows direct cell injections into the scar area under endovascular ultrasound guidance [13] and the trial has confirmed both the feasibility and safety of this approach, although this route of cell transfer may be technically challenging, particularly in patients who have previously undergone lead implantation for cardiac resynchronization therapy. In the other two percutaneous trials, myoblasts have been injected through an endoventricular catheter under electromechanical guidance. One of the studies (10 patients) reported a 1-year improvement in systolic velocity of the cell-injected segments and an increase in global ejection fraction during low-dose dobutamine infusion [14]. The second study also demonstrated an improved function in six treated patients who were compared with six case-matched controls [15]. However, the discrepancy in outcomes between the surgical phase I trials and the MAGIC trial highlight how data collected in such small-sized, uncontrolled, and open-label studies can be misleading. To address these issues, a randomized study has then been performed, which has included 23 patients with LV ejection fraction below 40% and old (>10 years) infarction allocated to endoventricular myoblast injections or optimal medical management alone [16]. The results reported orally at the 2007 Scientific Sessions of the American Heart Association look encouraging, and a more detailed publication of these data is now awaited.

Bone Marrow Cells

A more limited number of studies have assessed the effects of bone marrow cell transplantation in patients with heart failure (reviewed in [17]). In the surgical setting, cells have been injected epicardially into the target areas, except for one study in which they were also infused directly into the coronary artery through the bypass graft [18]. Because small-sized uncontrolled and open-label trials are of questionable relevance, two studies deserve further analysis. Mocini et al. [19] have studied 36 patients with a recent (<6 months) myocardial infarction and reasonably well-preserved LV function. These patients were nonrandomly allocated to injections of autologous bone marrow mononuclear cells (MNC; mean number: 292 × 106) in the border zone of the infarct area during CABG or to a control group. There were no safety concerns except for an initially higher troponin I peak after cell therapy. Three months after the procedure, both regional and global LV function had significantly improved compared with baseline values. However, outcome measures did not differ between the two groups and, unexpectedly, ejection fraction did not improve following bypass in control patients, which might have driven the results in favor of the treated group. In the study of Hendrikx et al. [20], 20 patients with an ejection fraction in the range of 40% were randomized to receive in-scar injections of autologous bone marrow mononuclear cells (mean number: 60 × 106) or saline in addition to CABG. Four months later, cell injections resulted in a significantly greater regional function, as assessed by magnetic resonance imaging, but failed to improve ejection fraction or perfusion defects beyond values seen in control patients. Altogether, these results confirm previous experimental data showing that transplantation of unfractionated bone marrow in chronically infarcted myocardium does not provide a functional benefit [21]. Furthermore, although these trials did not report safety problems, the recent experimental finding that in-scar implanted mesenchymal stem cells (MSC), some of which are present in the unfractionated bone marrow cell injectate, may cause intramyocardial calcifications [22] raises a cautionary note that requires further investigation.

In the more specific perspective of exploiting the angiogenic properties of some bone marrow cell populations, other investigators have investigated the effects of CD133 progenitors epicardially injected during CABG [23]. Although this trial was not strictly randomized, it has been rigorously conducted and it shows the capacity of the CD133 population to improve LV function and perfusion at 6 months postoperatively, particularly in patients with the poorest preoperative LV function. As one would not expect these cells to transdifferentiate into new cardiomyocytes, the prevailing assumption is that their paracrine effects may have led to increased angiogenesis, rescue of reversibly injured native cardiomyocytes and, more hypothetically, recruitment of putative endogenous cardiac stem cells which, altogether, could account for the improved LV function.

Catheter-based studies of bone marrow cells in patients with heart failure are also limited. They have been pioneered by an open-label nonrandomized trial [24] in which 11 patients received endoventricular injections of MNC and were reported to have an improved exercise capacity and a reduced perfusion defect at the 1-year follow-up. In two other studies, MNC have been infused directly into the coronary arteries. The IACT study [25] has claimed myocardial “regeneration” on the basis of improved outcomes, but these results are highly questionable because of the multiplicity of potential flaws in the trial design (such as the small sample size, the lack of true randomization, and the huge heterogeneity in baseline LV function). Using a more elaborate cross-over type of design in a 75-patient study, Assmus et al. [26] have similarly reported, at a follow-up of 3 months, the benefits of intracoronary infusions of either circulating progenitor cells isolated from venous blood or of bone marrow-derived MNC on both global and regional LV function (the latter being more effective). Although still limited, these data open interesting perspectives for the catheter-based treatment of chronic heart failure by bone marrow cells although they currently remain weakened by the lack of robust preclinical models and of subsequent mechanistic insights.

Limitations and Remaining Hurdles

In a clinically oriented perspective, they can be stratified into three main categories.

Origin of Cells

So far, a strong argument favoring the use of skeletal myoblasts or bone marrow-derived cells has been their autologous origin. However, with accumulated clinical experience, the limitations of patient-specific products have become increasingly apparent. They include: (1) the naturally occurring individual variability between patients, which makes it difficult to end up with a reproducible cell therapy product, particularly in patients with ischemic cardiomyopathy whose progenitor cells may be functionally impaired [27]; (2) the cost of customized quality controls, and (3) the logistical complexity related to back-and-forth shipments of the cellular products when their processing is centralized in a core laboratory. These hurdles can be overcome by cell banks able to supply a readily available “off-the-shelf”, controlled, reproducible, and thoroughly characterized product. In turn, these allogeneic cells have the disadvantage of immunogenicity, except, maybe, in the case of mesenchymal stem cells, which are credited for an immune privilege [28]. Whether this issue can be addressed by appropriate donor-recipient immunomatching still remains uncertain, but it is clear that autologous and allogeneic have to be investigated with regard to risk-benefit and cost-effectiveness ratios.

Transfer and Engraftment of Cells

A consistent finding of cell therapy studies is the very low rate of sustained cell engraftment, which is usually in the range of 1% of the initial number of donor cells a few weeks after transplantation [9]. Assuming that the cardiomyocyte deficit resulting from an infarction large enough to cause heart failure is on the range of one billion cells [29], one cannot reasonably expect a meaningful clinical benefit from such a tiny number of persisting donor cells, particularly in the case of skeletal myoblasts for which a clear dose–effect relationship has been documented [30].

This low rate of engraftment is initially caused by a mechanical leakage of cells. A recent study has thus shown, in a porcine model of cardiopulmonary bypass, that only 10% of intramyocardially injected microspheres approximating the size of MSC are retained within the sites of injection after 30 min, regardless of whether the heart is arrested or beating [31]. Experimentally, it seems that both surgical epicardial and percutaneous transendocardial injections result in equivalent engraftment rates [32], whereas studies performed at the acute stage of myocardial infarction have shown that only 2–5% of intracoronarily infused MNC are retained in the myocardium after a few hours [33]. It is critical to address this issue of cell transfer, and different strategies are currently being investigated, which include computer-driven injection devices, replacement of the injection concept by cell sheets in the case of surgical cell therapy (see below) and techniques, which enhance myocardial homing if cells are delivered intravascularly.

The second event that decreases engraftment is the high percentage of death of initially retained cells. This loss occurs over the first weeks after transplantation and results from the interplay of three main factors: inflammation, ischemia caused by the poor vascularization of the injected areas, and apoptosis subsequent to detachment of anchorage-dependent cells from their extracellular matrix (anoikis). The recognition of these contributing factors is now leading to move from isolated cell delivery to more composite grafts that incorporate a scaffold and eventually growth factors. In brief, whereas inflammation can be blunted by a pulse of corticosteroids, the ischemic component of cell death can be counteracted by a variety of strategies including direct revascularization, whenever feasible, co-transplantation of angiogenic bone marrow-derived cells or cell engineering with genes encoding angiogenic growth factors [34]. Cell survival can be enhanced by an equally wide array of techniques like graft incorporation into biocompatible matrices, physical or pharmacological preconditioning of cells, or gene-based boosting of survival pathways although, conceptually, the most appealing approach for preventing anoikis could be to respect cell-to-matrix and cell-to-cell connections, which have been successfully achieved with cell sheets prepared by culturing cells on temperature-sensitive or fibrin-coated films that are then stacked and overlaid onto the infarct area [35].

Clearly, the development of these survival-enhancing strategies should be paralleled by that of techniques of cell tracking allowing a noninvasive and reliable assessment of engraftment rates. A great deal of interest is currently paid to cell loading with iron superparamagnetic iron particles for detection by MRI, but this technique has its own caveats [36] and improvements in this area are eagerly awaited.

Functionality of Cells

A major conceptual question is to figure out what is the expected mechanism of action of the transplanted cells and what their ultimate objective should be. If the premise is that cells are going to act paracrinally by releasing factors that can stimulate angiogenesis, favorably affect the composition of the extracellular matrix, promote cell survival pathways, and even possibly recruit putative cardiac stem cells, then they do not necessarily need to be phenotypically matched to host cardiomyocytes as long as they supply the appropriate mediators. In this view, adult cells like myoblasts or bone marrow cells appear reasonable candidates, particularly the CD34 population, which is known for its angiogenic capabilities [37]. Conversely, if the assigned goal is to restore contractility of akinetic myocardial areas characteristic of heart failure by new functional cells, it has become clear that neither skeletal myoblasts [38] nor bone marrow cells [39] are the good players. The reason is that none of these cell types expresses gap junction proteins allowing formation of a syncytium, which is the prerequisite for the graft to beat in synchrony with the recipient heart and consequently to contribute to augment its pump function. In this case, it is not unexpected that cells that best achieve this remuscularization are those which recapitulate the developmental cardiomyogenic program. In this setting, the limited availability and poor scalability potential of fetal cardiomyocytes [40], along with the uncertainty about the persistence of cardiac stem cells in adulthood, highlight the potential interest of human embryonic stem cells (ESC), provided they have been appropriately specified in vitro toward a cardiac lineage. Under these conditions, there is compelling growing evidence that following engraftment in postinfarction scars, these ESC-derived progenitors complete their differentiation in cardiomyocytes [41].

Because there is no animal model that can fully duplicate the complex situation of patients with coronary artery disease, we believe that it is legitimate to continue clinical trials provided that they are randomized, adequately powered, placebo-controlled and blinded. As it is likely premature to launch large-scale mortality trials, surrogate end points and imaging modalities should be selected so as to confirm the proof of principle and help in providing mechanistic insights. In this perspective, the assessment of cell-related morphological changes such as infarct size, LV remodeling, or regional wall thickness are probably as informative as the commonly used measurements of global LV function. However, it is likely that the outcomes of these trials will be optimized if their protocols integrate lessons learned from the first wave of studies and thus successfully address issues related to transfer, survival, and functional integration.