Eli Lilly’s phase III clinical trial of semagacestat, a gamma-secretase inhibitor, was a notable disaster. The drug did absolutely nothing at all for Alzheimer’s patients – in fact, if anything, it made them slightly worse. The amyloid hypothesis was not illuminated in any useful way by this work – in fact, if anything, the situation was more confused afterwards than it had been before, and that’s saying something. For its part, Lilly spent a great deal of time and money that (in retrospect) could have been more usefully employed doing almost anything else.
Here’s a paper that combs through the wreckage looking for what we can learn from all this. It’s not an easy task:
Semagacestat was one out of the many candidate drugs that advanced as a γ-secretase inhibitor blocking Aβ generation in cell cultures, animals, and finally human. Only relatively late in the studies did it become clear that the potential side effects caused by blocking Notch-signaling were severely limiting the clinical use of semagacestat. In fact, from a scientific point of view, it remains puzzling why the company decided to move forward toward phase III tests with a dose (maximal 140 mg/day) that in none of the phase II tests had shown any significant impact on Aβ levels in the cerebrospinal fluid (CSF) of humans. Looking back, it seems clear that such a phase III trial was unlikely to test the amyloid hypothesis, as elaborated below.
The author, Bart de Strooper of Leuven and University College (London) argues that the bad effects of the drug can likely be assigned to its effects on Notch signaling, and that the eventual once-a-day dosing schedule may well have exacerbated the problem. Even the peripheral side effects (skin trouble, nausea) could have been enough to decrease the cognitive test scores of an Alzheimer’s population. As for the drug’s effects in the brain, it had a short enough half-life that combining this with q.d. dosing meant that there were periods where no drug at all would be expected to remain in the CSF. This may have actually stimulated beta-amyloid production during those periods – in fact, de Strooper argues that Lilly may have chosen one of the worst possible dosing protocols for the drug.
And he goes on to point out just how little we know about gamma-secretase, both in the disease state and under normal conditions. He characterizes the entire Lilly trial as “premature”, given the state of the field, and believes that it has in fact set back research in this area, industrial and academic. The article is a brief for the defense, from someone who clearly believes that gamma-secretase as a target has been ill served. It’s hard to argue with him on that point, but at the same time, one reason it’s been so poorly characterized is that coming up with good therapeutic approaches has been very difficult. Finding a gamma-secretase inhibitor with good properties has really been a long haul. In the end, the main lesson from the trial seems to be that we can’t draw many lessons from the trial, which reminds me of what Hegel had to say about learning from history.
I’d be willing to bet that some of the people who are upset about semagacestat’s failure and what it’s done to the field were actually glad to see it go into the clinic at the time. Finally, we’d get some sort of data on this mechanism, even if this wasn’t the best way to look at it. Had to be better than nothing, right? Well, this experience was more proof that there are worse things than nothing. The gamma-secretase field would have been better off if had continued to travel hopefully, rather than arriving at the end of that trial.