This randomized, double-blinded, controlled trial of autologous stem cell therapy for critical limb ischemia was a pilot trial and intended to present a safety profile for the technique and to identify trial design issues for a pivotal trial. Although not powered to demonstrate efficacy, the outcomes provide valuable trends about future trial design, not only for this technique, but for other trials in CLI.
There was a trend toward improved amputation rates in Rutherford 5 patients treated with BMAC, although this did not achieve statistical significance. (p = 0.1337) More important was the difference in amputation rates between those patients with rest pain at screening (Rutherford 4) and those with tissue loss (Rutherford 5). Patients with tissue loss demonstrated a much higher amputation event rate (46.7%) than did those with rest pain alone (5.6%) and this difference was statistically significant (p = 0.0029). Only a single patient with rest pain underwent an amputation by six months; patients with tissue loss at screening accounted for 93.3% of all amputations. This association was further confirmed by logistic regression modeling which demonstrated the correlation of tissue loss with amputation with an Odds Ratio of 18.48 (p = 0.013).
With regard to change in Rutherford class as an instrument to measure effectiveness of CLI therapy, we found the opposite. A higher percentage of Rutherford 4 patients demonstrated an improvement in Rutherford classification than did Rutherford 5 patients. This makes biologic sense, because an ischemic ulcer requires more blood flow to heal than intact skin does to survive. Therefore salvaging a limb with tissue loss requires a far greater degree of perfusion than does achieving resolution of rest pain . While change in Rutherford class may be a used to evaluate therapy in patients with rest pain, it may not be as appropriate for advanced CLI patients.
ABI was a difficult endpoint to evaluate because of multiple missing data points and the difficultly of incorporating amputation into this measure. Hemodynamic endpoints may not be appropriate for NOCLI patients at high risk for amputation.
Since the results in our pilot trial differed depending on the Rutherford classification of patients at screening, we surveyed the literature for evidence of the impact of tissue loss on the amputation rate. To determine the amputation rate in no option CLI patients overall, we identified a group of NOCLI patients gathered from the control groups of eight randomized trials and one observational study in the literature. This surrogate natural history population demonstrated a six month amputation rate of 23%. However, only the CIRCULASE trial, Klomp's spinal cord stimulation trial and the current BMAC pilot trial present amputation data according to Rutherford class. Meta-analysis of these trials demonstrated that patients with tissue loss have a higher amputation rate than those with rest pain (Hazard Ratio 8.650, p = 0.0513). This is further supported by a separate analysis of patients in a spinal cord trial in which those with tissue loss demonstrated a higher amputation risk (HR 2.38, p = 0.018) .
Because most published clinical trials group all CLI patients together regardless of disease severity, it is difficult to quantify the impact of tissue loss on amputation rate. However, there is evidence to support that tissue loss in CLI correlates with poor outcomes. In addition to an analysis of NOCLI studies, risk stratification models underline the correlation between tissue loss and amputation in patients undergoing surgical or endovascular revascularization. The first of two validated risk stratification models, the PREVENT III tool was derived from a multivariate analysis of nearly a thousand CLI patients, and it identified tissue loss as second only to dialysis as a factor negatively impacting AFS. In a population of patients undergoing surgical bypass, tissue loss was associated with a hazard ratio of 2.2 for increased risk of death or amputation at one year . The Finnvasc model identified tissue loss as a predictor of negative outcome during the immediate postoperative period . Worse scores on either of these tools correlate with decreased AFS at one year in patients undergoing revascularization .
Tissue loss correlates with a variety of poor outcomes in CLI. Khan et al. demonstrated that in patients undergoing surgical revascularization, tissue loss is associated with a higher rate of amputation even if the revascularized segment remained patent - the so-called anatomic success but functional failure . Nguyen et al. found that in patients undergoing surgical bypass those with tissue loss required increased resource utilization such as length of stay and experienced a higher rate of graft related events (GRE) such as thrombosis, need for revision or amputation . In a prospective observational study of 1560 CLI patients, of whom 36.5% underwent revascularization, Bertele et al. found that tissue loss was associated with increased risk of amputation at six months (7.8% vs. 13.9%) . Taylor et al. reviewed 2240 limb revascularizations for PAD according to preoperative indication . The population included patients with claudication (999), ischemic rest pain (464), and tissue loss (777). There were significant differences in multiple endpoints, including limb salvage and survival, among the groups. The 1 year limb salvage rate for patients with rest pain was 88.6% while for those with tissue loss it was 75.1% (p < 0.001). The survival rate at one year was 79.1% for rest pain and 65.8% for tissue loss (p < 0.001).
Implications for trial design
This difference in outcomes between patients with tissue loss and those with rest pain has implications for study design in CLI. Patients with tissue loss demonstrate a higher rate of amputation, which is a component of amputation free survival (AFS). If amputation rate or AFS are used as the primary endpoint in a clinical trial, the lower number of amputations in Rutherford 4 patients may dilute the event rate in the overall patient population making it difficult to demonstrate a treatment effect. This is supported by sensitivity analysis which demonstrates that increasing the proportion of Rutherford 4 patients in the study population requires increasingly larger sample sizes to maintain appropriate statistical power. Patients with tissue loss are appropriately evaluated using amputation or AFS as an endpoint.
On the other hand, if the outcome measure is change in Rutherford class, which has been suggested by some as component of a combined endpoint, a sample weighted towards Rutherford 4 may better demonstrate changes by this measure.
Recently some authors have begun to question whether grouping all CLI patients together obscures the evaluation of therapy [23–25] and our findings confirm that patients with tissue loss should be analyzed separately from those with rest pain.