The FDA is letting the industry know that it’s interested in streamlining the accelerated approval process for new cancer drugs, but that they’re not interested in making it too easy, either. They’re open to the idea of a “one-trial” route to such approvals, but it had better be a good trial.
The agency is singling out single-armed trials for special disfavor, especially non-randomized ones. You see these run in oncology to get a first quick read on safety and efficacy in some defined patient population, and that’s all fine – but using data from these trials to support accelerated approval doesn’t look like it’s going to cut it any more. This article at The Lancet will illustrate why: it’s about the PI3K inhibitors for leukemia and other malignancies. Several of these compounds were given accelerated approvals, but they’ve proven to be more toxic than the first trials indicated. The drugs still work; it’s just that their risk/reward isn’t as favorable as hoped.
Smaller single-arm trials aren’t good at picking up rare-but-serious adverse events, for obvious statistical reasons, but there are bigger problems. Tumor progression and even survival endpoints in such trials are not very interpretable under those conditions, either, and relying on comparison to other historical trials is asking for trouble, too. But a “real” randomized controlled trial brings things into better focus, not least because you at least have a direct comparison to a control arm (usually standard-of-care). And you might well be able to use longer-term follow-up in the same trial to fulfill an approval requirement later on.
Now, such trials take longer to set up and are more expensive to run, so from that perspective this is slowing things down. But trying to speed things up by generating incomplete or lower-quality data is not the answer:
Given the limitations of single-arm trials, a randomized controlled trial is the preferred approach to support an application for accelerated approval. Sponsors can, as appropriate, elect to conduct a single randomized controlled trial to support an accelerated approval and to verify clinical benefit (i.e., follow a “one-trial” approach) or, they can conduct separate trials – one to support the accelerated approval and another, a confirmatory trial, to verify clinical benefit.
This also addresses the agency’s requirement that confirmatory trials be enrolling before accelerated approval is even granted. This draft guidance is in the public-comment phase now, but I don’t see any reason why it won’t turn into the pathway for accelerated approval from here on out. It seems to strike a good balance between running things more quickly while at the same time not cutting corners. People who see the agency as a roadblock in the way of zippy progress aren’t always wrong about that, but some of the “progress” you can show by flooring the accelerator is an illusion, too. . .
