On Wednesday the 28th Allos announced the results of their randomized phase 2 study comparing Folotyn (pralatrexate; Allos) to Tarceva (erlotinib; Genentech/Roche; OSI) in 201 patients with stage IIIB/V NSCLC who had been previously treated with a platinum containing regimen. The primary endpoint of the trial was overall survival. In the overall patient population a reduction in the risk of death of 16% was achieved (HR = 0.84). In the “primary efficacy population” a reduction in the risk of death of 13% was achieved (HR = 0.87). There was no statistical significance reported and the company claims that was not part of the analysis plan for the trial.
Why would the company exclude a statistical analysis of the study? Well a lot of other people are wondering as well and presuming the worst. The Allos share price dropped 9.8% yesterday when this data and quarterly earnings were reported. In the company’s defense, phase 2 trials are used to demonstrate product efficacy and provide data to help in the design of future phase 3 studies. Typically, phase 2 studies have fewer patients so there is always a risk that the trial will not have enough statistical power to detect a difference versus the control. The Folotyn trial however was quite large with 201 evaluable patients and it probably had sufficient power to detect a difference.
So the company intentionally omitted doing a statistical analysis. I think Allos is paying the price on two fronts for this decision: the share price fell anyway because of the lack of a convincing result and they aren’t really gaining sufficient insight for their phase 3 design. They don’t know if the drug is really more effective than erlotinib. Is a 16% reduction in risk enough to be significant? Even if these results bear out in phase 3 will there be sufficient evidence to differentiate Folotyn and drive usage in a space that is becoming more crowded?
The company did report a more dramatic risk reduction of 35% and 37% in non-squamous cell histology patients and light smokers respectively. Although the reduction in risk appears large, without knowing the number of patients in these cohorts or the statistical significance, we can’t draw any real conclusions. That being said, it is clear that Folotyn has efficacy in NSCLC. We know that erlotinib has efficacy, therefore these results demonstrate that Folotyn is probably at least as effective as it—and possibly more active. What was not being said is even more important—squamous cell patients and heavy smokers didn’t benefit. This reduction in the size of the potential market for Folotyn was probably another reason the share price fell.
Finally, designing a study which includes the evaluation of pre-defined patient subsets but does not include a statistical plan can pose a real danger to the phase 3 study design and ultimately which patients have access to the drug. Imagine if squamous cell patients and heavy smokers were excluded from the phase 3 trial and the drugs’ subsequent labeling (or compendia listing) and we later learn that it actually works for these patients. How many patients would have lost Folotyn as a treatment option in the meantime? I am sure that the Allos team recognizes this danger and won’t exclude these patient cohorts in the phase 3 study(s).
So the question remains whether or not Folotyn will likely be approved or obtain compendia listing for NSCLC. Since the drug probably has efficacy in NSCLC its success in phase 3 will depend on what treatment is selected for the control arm. If the control is erlotinib I think there is some risk of failure unless a post-hoc statistical analysis of this phase 2 trial demonstrates a p-value of less than 0.01 in favor of Folotyn. Allos announced that this analysis is being conducted and will be presented at an upcoming medical meeting. I am going to wait and see before buying any stock.