In brief, the data showed that out of the 240 patients in the trial, those who had no apoE4 had statistical improvement compared to those who did. The skeptics made their case, to which I reply:
- The prospective analysis failed. The success was with retrospective analysis. My response to that is that in phase II studies where there are sometimes too many variables, often variables that are not known or identified when the study originated, it is not unusual to undertake retrospective analysis to identify positive trends. And in this case, it was more than a trend – it was statistically significant. Keep in mind that this was a phase II study – not a pivotal phase III study, where the analysis would have to be prospectively defined. I hope Elan/Wyeth made appropriate changes to their Phase III study and statistical analysis plan [SAP] to reflect this.
- There was no mention of dose response. Fair enough - but that does not mean there was no dose response. Absence of affirmation does not mean affirmation of absence. While a dose response is scientifically appealing, take a look at it practically. If the drug worked at a dose, and there was no effect whatever with smaller doses, would you refuse to take the effective dose if you had AD?
- The sample size was small. Of course it was – it was a phase II study. On the flip side – maybe it is impressive that there was statistical significance even with such a small sample size?
- If you data-mine enough, you will find something statistically significant. Not entirely true – having data mined through many studies, I can state with confidence that there are some truly negative studies. More importantly, if the data mining finds something significant that is scientifically reasonable, should it not be given some importance and explored in a prospective, controlled, larger study? Many drugs, including Viagra, came out of such exercises.
- It only worked on patients with ‘milder’ disease. Even if true, does that imply those with mild disease should not be treated? Should they be allowed to progress to severe disease?
- It worked in only a small subset of patients (approximately 40%). I would not call 40% small, and 40% of tens of millions (just in the US) is still a large number.
The details of the study will be presented on Tuesday, July 29, during an afternoon session at the ICAD meeting in Chicago. Let the data speak for itself.