There's no telling how the FDA will react to data when two methods of statistical analysis have yielded conflicting results and the trial concerns a drug candidate for a condition having no treatment precedent. Just ask Transcept, which is working to address issues from a 2009 complete response letter regarding Intermezzo, its treatment for the unique condition of middle-of-the-night awakening.
Transcept has just released results of a highway driving study intended to test residual effects of the drug on next-day driving ability, according to investor Larry Smith, a contributor to Seeking Alpha. The drugmaker used a commonly accepted measure of driving performance--standard deviation of lateral position--as the primary end point, and it used a statistical analysis approach that it calls symmetry analysis.
The driving test results show that Intermezzo produces no adverse effect on a subject's capacity to operate a motor vehicle four hours after a middle-of-the-night dose. Smith is not so sure that closes the case, however. The FDA had OKed the statistical methodology as a measure of potential next-morning driving impairment. But, Smith asks, will the FDA ultimately accept this endpoint or the statistical analysis in answer to the complete response letter?
Transcept also used a secondary statistical-analysis technique that compares group mean effects. Researchers found small but statistically significant differences between Intermezzo and placebo taken four hours prior to driving. By this statistical measure, Intermezzo narrowly missed the primary endpoint.
It's hard to guess what the FDA will do, Smith says, given the absence of clinical precedents. Intermezzo achieved the primary endpoint using its FDA-approved statistical analysis plan. But the regulator could still decide the endpoint was incorrect, particularly given the second statistical analysis' finding that showed modest driving impairment.
- here's the article