Phase I cardiovascular cell therapy clinical trials: Are we running with scissors?

The results of unblinded phase I and early randomized placebo-controlled cardiovascular stem cell clinical trials are encouraging, but considerable uncertainty exists regarding mechanism(s) of benefit, ideal cell(s), and method(s) of delivery for specific patient populations.1, 2, 3 Questions with regard to dosing, timing of delivery, and long-term safety also remain. These issues have led prominent cardiologists to voice concerns that our enthusiasm is misplaced, and we have “jumped the gun” in our haste to provide novel therapeutic options when mechanistic questions and safety have not been definitively addressed. Some have even suggested that cardiovascular stem cell therapy is at “a crossroads” and a moratorium should be placed on new clinical trials.4 We disagree, but with a few caveats. Well-designed clinical trials have the potential to complement and even stimulate basic science and preclinical research.5 In our excitement to bring new cell therapy–based treatments to very ill patients, we need to ensure that clinical studies are conducted such that we will be able to measure the effect of treatment and detect safety issues.

In this issue of the Journal, Kovacic et al6 report the results of an open-label, safety, and feasibility trial in 20 patients with refractory ischemia. Patients received subcutaneous granulocyte-colony stimulating factor (G-CSF) at 10 μg/kg per day for 5 days in conjunction with “controlled induction of myocardial ischemia” using an exercise treadmill test on days 4 and 6 during G-CSF administration. Three months after randomization, patients received a second course of G-CSF at 10 μg/kg per day for 5 days with the stress test to induce myocardial ischemia. After the second course, patients were randomized to receive intracoronary CD133+ cells (n = 10) versus intracoronary infusion of unselected cells (n = 6). During the 6-month follow-up, there were no deaths but 4 patients (20%) had a myocardial infarction. In addition, cardiac biomarkers (troponin I) were elevated on an additional 17 occasions in 8 patients. The authors noted the patients who completed the study (n = 16) had an improvement in angina frequency, exercise time, and the Duke treadmill score. However, there was no significant difference in ischemic burden as measured by dobutamine stress echocardiogram or persantine stress perfusion imaging. Cell infusion did not appear to have an effect on outcome. This trial raises a number of issues that deserve comment.

There are several novel aspects to this phase I trial: exercise to “simulate ischemia,” the repeat administration of G-CSF therapy, and attempted labeling of cells. Ultimately, however, the study is crippled by multiple errors in design and execution. First, nearly every patient received a different experimental regimen: if consistency cannot be achieved in a trial of 20 patients, study in a larger patient population will only magnify the error. Second, G-CSF therapy was administered at high dose (10 μg/kg): previous trials in patients with refractory ischemia using this dose were stopped early because of safety concerns.7, 8, 9 Third, there was no placebo group: patients with refractory angina are known to exhibit significant improvement during clinical trials, making this a crucial component of trial design. Fourth, although exercise treadmill testing during G-CSF administration was used to “simulate myocardial ischemia” (which theoretically might serve as a homing signal to attract mobilized progenitor cells), the lack of a placebo group means that the trial did not actually attempt to assess the efficacy of the principle hypothesis. Fifth, after the second round of G-CSF administration, patients were randomized to receive intracoronary CD133+ stem cells versus unselected cells: this was the only treatment subjected to study, and it appeared to be ineffective. Finally, attempts at cell labeling did not allow for interpretation: it was impossible to analyze what percentage of cells were delivered to the myocardium and how long the cells persisted. The trial's complexity and lack of focus make it difficult to answer the question, “What did we learn from this trial and how can we apply the knowledge to future studies?”.

The trial also raises an issue that may seem simplistic but is completely relevant: what is “safe”? In other words, how do we monitor phase I trials with novel agents and how do we decide when a trial should be stopped? In this trial, there was an active data safety monitoring board, in fact more active than usual in that each patient's eligibility was reviewed by the data safety monitoring board before enrollment. Despite this, 4 patients had elevated troponin I consistent with myocardial infarction after the first round of G-CSF and 8 patients had 17 other episodes of elevated troponins that would, at minimum, qualify as myocardial injury. We have seen this problem before. In 2005, 2 groups reported acute coronary syndromes in conjunction with injection of subcutaneous G-CSF (10 μg/kg per day for 5 days) and granulocyte-macrophage colony-stimulating factor (GM-CSF) (10 μg/kg per day for 2 weeks). In the GM-CSF–treated patients, 2 of 7 patients with stable angina developed an acute occlusion of a coronary artery within 12 days of treatment8; and in an National Institutes of Health–sponsored trial, 2 of 16 patients with refractory angina suffered an acute myocardial infarction, one of which was fatal.9 Based on these 2 trials and the known proinflammatory and procoagulant effects of GM-CSF and G-CSF, we felt it was prudent to restrict G-CSF and GM-CSF to patients with no other therapeutic options and, if necessary, to proceed cautiously with lower doses. Based on the early clinical trials, it appeared unlikely that G-CSF or GM-CSF alone would provide sufficient clinical benefit especially in light of the potential safety concerns.9 A phase I randomized, placebo-controlled trial using lower doses to facilitate stem cell mobilization followed by apheresis and local delivery of CD34+ cells appeared safe with clinical improvement in the treated patients.10 The phase I trial results led to a 150-patient, phase II, randomized, placebo-controlled trial in patients with refractory ischemia who have completed enrollment.

Kovacic et al used the high dose of G-CSF and specifically added exercise testing to provoke myocardial ischemia: this combination should have raised a red flag in the concept phase of the trial in light of the previous trials. Although the frequency of troponin elevations in an unselected cohort of refractory ischemia patients is unknown, only 1 of 17 elevations was not temporally related to G-CSF and the associated leukocytosis. Although there was no mortality in the brief 6-month follow-up period, elevation of troponin predicts adverse clinical outcomes in almost every clinical situation. The authors propose that patients with refractory ischemia are a high-risk population expected to have significant morbidity and mortality over this time frame: critical readers should ask, “Is that actually true?”.

To properly design a clinical trial, one needs to understand the patient population. So what do we actually know about “no option” patients with refractory angina? We recently reported long-term follow-up on more than 1,100 patients with refractory angina, which demonstrated a 3.3% mortality rate at 1 year and 25.8% at 9 years after diagnosis.11 Nearly one third of deaths were from noncardiovascular causes. In a meta-analysis of 13 randomized, placebo-controlled, refractory angina trials, the 1-year mortality was 1% to 2%, and the rate of myocardial infarction was 2% in placebo patients.12 This is consistent with long-term data from the International EECP (Enhanced External CounterPulsation) registry, which includes predominately patients with class III and IV angina who are not candidates for revascularization.13 A 6-month myocardial infarction event rate of 20% (excluding the “troponin elevations”) would be high by any standard.

The major issue for patients with refractory angina is not survival but quality of life. Trials should be designed with this in mind. A second major issue in the design of refractory angina trials is the well-known placebo effect that complicates the assessment of treatment effects. In the 13 placebo-controlled, refractory angina trials that measured exercise time, the 4- to 6-month improvement in placebo patients was nearly 1 minute. Trials need to be designed and powered with this expectation for the placebo group. Although the authors of the current study introduce a number of hypotheses worthy of testing, the complexity of the trial and the small numbers make it impossible to make firm conclusions.

There is a need for an international working group to develop standards for cardiovascular stem cell trials. It is imperative that we design future cell therapy trials to provide clear insights into safety and efficacy and allow the field to advance: “running with scissors” is no longer an option.