FDA Guidelines are not regulations and therefore allow for discretion. But they are also not nothing.
So, unlike with a regulation, a failed co-primary endpoint can still result in an approval so long as the Agency is able to use its scientific judgment to support the decision. In the case of an early AD trial, like the 2b/3, given the state of the science this approval possibility appears open in the case of a failed co-primary functional endpoint; on the other hand, I don't see how FDA discretion could be supported to ignore a failed cognitive co-primary endpoint. Even discretion has its limits, and can't tip over into "arbitrary and capricious" territory.
As I've consistently said, the use of the older FDA decisional guidance -- and the fact that other early AD applicants used it -- will complicate the path to an NDA approval for blarcamesine, but it will not necessarily block it.