Granulocyte Colony-Stimulating Factor iN Cardiac Repair: Time for a Reappraisal?

This editorial refers to ‘Granulocyte colony-stimulating factor attenuates left ventricular remodelling after acute anterior STEMI: results of the single-blind, randomized, placebo-controlled multicentre STem cEll Mobilization in Acute Myocardial Infarction (STEM-AMI) Trial’ by F. Achilli et al., published in this issue on pages 1112–1122. Despite early reperfusion of infarct-related coronary arteries followed by best medical therapy, salvage of viable myocardium is often incomplete and will result in the development of heart failure, compromising clinical outcome. The combined clinical endpoint of death, recurrent myocardial infarction (MI), and rehospitalization for worsening heart failure in the largest carefully monitored trial population was reached in about one in every four patients.1 It is very likely that clinical outcome in the real-world experience of daily practice in a continuously ageing population will be even more dismal. These data reinforce an urgent need for better therapies to prevent the vicious cycle of progressive deterioration of left ventricular (LV) function after myocardial infarction. In the last decade, major advances in stem-cell biology have created great promise for what has been generally termed ‘regenerative medicine’. The existence of both cardiac and non-cardiac progenitor cells that have been shown to impact on cardiac repair in preclinical models has spurred a wide-ranging interest in potential clinical applications of these promising basic concepts. The enthusiasm for these early clinical trials has been fuelled by experimental data, which suggested that autologous bone marrow-derived or circulating progenitor cells or cardiac resident cells may be beneficial for post-infarction LV function recovery. Many of the preclinical studies had suggested beneficial effects on cardiac repair irrespective of whether progenitor cells were being mobilized from the bone marrow niche using potent cytokines, including granulocyte colony-stimulating factor (G-CSF), or whether purified progenitor cell populations were administered in vivo using a variety of delivery routes. In parallel, carefully performed clinical studies have unambiguously shown that in the early post-myocardial infarction period several progenitor cell populations, characterized by the expression of specific surface markers such as CD34, c-kit, and CxCR4, are transiently increased in the circulation, suggesting an endogenous cross-talk between the bone marrow niche and a remote site with severe ischaemic damage.2–4 Although these observational studies are scientifically exciting, they do not allow to identify whether such cellular response is merely reactive or reflects an incipient, albeit insufficient, reparative phenomenon. Against this background, many researchers have set out to investigate whether cytokine-mediated amplification of such endogenous and spontaneously occurring bone marrow response to remote ischaemic injury could be a useful strategy to enhance post-myocardial cardiac repair. Unfortunately, thus far, clinical trials with the most rigorous methodological design that have incorporated the use of G-CSF have failed to live up to these high expectations.5,6 G-CSF is a haematopoietic cytokine produced by monocytes, fibroblasts, and endothelial cells with an important role in myeloid development and neutrophil maturation, hence its frequent application in haematological practice for granulocyte collection from donors for blood banking and allograft therapy.7 When G-CSF was administered in animal models of myocardial infarction, improved haemodynamic function and infarct remodelling were thought to be related to enhanced mobilization of bone marrow-derived stem cells, which subsequently home to the damaged myocardial tissue and induce neovascularization.8,9 Such homing response was mediated by strong up-regulation of the chemokine stromal cell-derived factor 1 (SDF-1) in ischaemic myocardium, while in the early post-infarction period the G-CSF receptor was also up-regulated in the injured myocardium, suggesting a sensitization of the ischaemic heart to this cytokine.10 Although most recent studies have documented a reassuring safety profile of G-CSF in patients with ischaemic heart disease,11,12 reasons for the lack of efficacy are likely multifold. The selection of patients and severity of the ischaemic insult, dose and timing of growth factor administration, lack of understanding of local G-CSF receptor expression, choice of clinical or surrogate endpoints, and investigative imaging modalities are generally considered to impact on the results of mobilization strategies.11,13 In the present issue of the European Journal of Heart Failure, Achilli et al.14 report a randomized, single-blind, placebo-controlled study that was designed to evaluate the efficacy of G-CSF administration on LV function (primary endpoint) and volumes (co-primary endpoint) in patients with anterior myocardial infarction and depressed LV ejection fraction (STEM-AMI trial). The authors randomized 60 consecutive patients with anterior STEMI undergoing primary angioplasty within 12 h of symptom onset. They carefully excluded patients with aborted infarctions (<2 h of symptom-to-reperfusion time) or patients with residual ejection fraction >45%, all of whom will likely have a favourable post-revascularization outcome. Using magnetic resonance imaging (MRI) for the primary and co-primary endpoints, they detected no change in LV ejection fraction from baseline (Day 5) to 6 months’ follow-up between the groups, but a strikingly different pattern of LV remodelling with significant expansion of end-diastolic and -systolic volumes in placebo-treated patients, vs. virtually no change in G-CSF-treated patients. Importantly, a rigorous multivariate analysis of covariance whereby treatment effects were estimated by computing the differences between adjusted means of G-CSF and placebo-treated patients indicated that the change in LV end-diastolic volume from baseline to 6 months was significantly lower in G-CSF-treated patients. Of interest, a similar relation was observed for the changes in infarct remodelling, potentially suggesting a common or similar pathogenic mechanism. Finally, serial determinations of NT-proBNP, a validated biomarker of LV stress in heart failure, showed a significant decline from baseline to 6 months’ follow-up in G-CSF-treated patients only. The elegantly designed and carefully conducted study by Achilli et al.14 differs in several important aspects from the previously reported prospective, double-blind, randomized clinical study by Engelmann et al. in patients undergoing delayed revascularization after subacute ST-segment myocardial infarction.6 In the latter study, patients were included who presented late after MI involving any perfusion territory (mean time to PCI 32–45 vs. 4–6 h in the STEM-AMI trial, which exclusively randomized patients with anterior MI), were started on SC cytokine injections after a greater time delay (31 h after re-establishing TIMI2-3 flow vs. 9 h in STEM-AMI), and received a follow-up MRI scan to evaluate LVEF and the degree of systolic wall thickening 3 months later (vs. 6 months in STEM-AMI). The trial also significantly differs from the REVIVAL-2 study, which to date still represents one of the largest prospective, double-blind, randomized, and placebo-controlled G-CSF trials including 114 patients with successfully reperfused acute myocardial infarction within 12 h of symptom onset. Treatment with a similar G-CSF dose regimen was started 5 days after revascularization and was unable to further influence the prespecified primary endpoint, a reduction in infarct size as evaluated by technetium-99 (99mTc) sestamibi scintigraphy at 4–6 months’ follow-up.5 Of note, serial MRI analysis in 96 patients did not show changes in LV volumes over time in either placebo or G-CSF-treated patients, consistent with a relatively low-risk population as confirmed by the modest reduction in ejection fraction at baseline (49–51%). Taken together, a critical look at these studies suggests that perhaps the best window of opportunity for G-CSF mobilization strategies is the early post-revascularization phase in patients with large (anterior) myocardial infarctions, who are at increased risk of developing maladaptive LV remodelling and heart failure. In addition, Achilli et al.14 report that the observed effects on LV remodelling were independent of baseline NT-proBNP, Troponin-I, number of mobilized CD34+ cells, and time-to-reperfusion or Killip score. Unfortunately, the small size of the study does not allow the identification of underlying mechanisms that may have contributed to the observed effects, a limitation acknowledged by the authors. It is very plausible that the early timing of G-CSF administration is critically important, as suggested by the powerful cardioprotective effect of the cytokine in animal models.15 Whether these direct anti-apoptotic effects predominate over any progenitor cell mobilizing effects remains unknown at the present time and needs to be addressed in future, specifically designed larger clinical studies. The suggestion by the authors that the observed benefit may be derived from improved infarct remodelling is at present premature and needs to be substantiated in follow-up studies with a more refined analysis of infarct remodelling. In summary, the authors of STEM-AMI are to be complemented for integrating multiple complementary imaging modalities in this carefully designed exploratory study, which will undoubtedly revive interest in the potential therapeutic role of mobilization strategies. While they acknowledge the limitations of small-scale exploratory studies most notably with respect to underlying mechanisms, the reported encouraging effects on LV remodelling in combination with the confirmed safety profile will facilitate larger clinical trials in targeted AMI patients at high risk for developing maladaptive LV remodelling and heart failure. These prospective, randomized multicenter studies in AMI patients at increased risk for developing heart failure are required to evaluate whether or not the observed changes in remodelling parameters following early G-CSF treatment translate into better clinical outcome and survival.