Pharmacyclics Announces Results from Phase 3 Smart Trial of Xcytrin(R) for Lung Cancer Brain Metastases
For the record, I have no financial interests in Pharmacyclics or its competitors. I’ll assume you will read the linked press release, so I can keep my comments brief. I have a question. Why was there a 50% improvement in the primary endpoint measure without statistical significance? One person on this morning’s conference call asked about variability in response but didn’t really understand what he was asking, and the respondent apparently didn’t either. No one on the call asked how the p-value was calculated, or about the confidence interval around the hazard ratio, or even how the hazard ratio was derived. Any of these would have been reasonable questions; take your pick. When a drug purportedly fails to deliver on its promises in Phase 3 investors, patients, regulators and investigators all want to know why. It’s not an excuse that there is a limited time (in this case 48 hours) to promulgate the top-line data. If a company discloses data and infers truths about its drug based on those data, it had better make the data clear. To do less is, at best, unfair to all interested parties. In this case, interested parties were not told why a 50% increase in the median time to progression of neurological symptoms (the primary endpoint) was associated with a hazard ratio of just 0.78 (p=.12). In my experience, this apparent discrepancy is unusual in cancer trials. Usually, the median TTP corresponds closely with the hazard ratio, reflecting proportional hazards of therapy. So, something strange is going on with the data or the analyses or both; too bad the public has no idea what.
