Sarepta's Elevidys had a single primary endpoint and a bunch of secondary endpoints. It failed the primary on p-value but trended in the desired direction. The primary was an old standard for measuring DMD (Duchenne), but was diagnostically very crude. (The recent Anavex PR on Rett pointed out how CGI-I lacked granularity because it scored patient performance on a seven-point scale; the test for the primary endpoint that Elevidys failed scored on a three-point scale.) The secondary endpoint measures, OTOH, were mainly timed tests (e.g., how fast a subject could walk or run ten meters), so they were very granular. On those, Elevidys did great compared to the placebo, with generally excellent p-values (at least for the younger of the two cohorts in the trial). So the secondary endpoints very clearly demonstrated efficacy.
It's easy to see how a three-point scale would not do a good job of showing a Duchenne boy's deterioration over the course of the one-year trial (although it may have over two of three years), while tests of abilities timed in tenths of a second would.
Additionally, there is a clear biomarker for Duchenne improvement: creation of functional dystrophin, a critical muscle protein that Duchenne boys lack because of a genetic defect. Sarepta showed success on that biomarker.
I don't see Sarepta's trial results as analogous to the results of EXCELLENCE, where none of the primary or secondary efficacy endpoints had p-values better than 0.05. The biomarkers for Rett also aren't clear like the creation of functional dystrophin.
OTOH, there are ways that I see Anavex's Alzheimer's P2b/3 as analogous. It's arguable that ADCS-ADL, which failed statistical significance in the P2b/3*, is a measurement that, like Sarepta's primary endpoint, does not accurately reflect deterioration in a timespan as short as 48 weeks, though it may do so over a two or three year period. Even more important, the P2b/3 showed statistical significance for its other primary, and its secondary, endpoints. Also, the brain shrinkage and amyloid-tau-related measurements were biomarkers that demonstrated modification of the disease trajectory.
Sarepta is applying for unconditional approval for Elevidys on the basis of that "failed" trial. I can see Anavex applying for approval for the Alzheimer's indication on the basis of the "failed" P2b/3, similar to Sarepta applying for approval based on its "failed" trial.
* Whether it failed is contested on this board by the WGT crowd, but I think it's very nearly unquestionable that it did.
