I wrote last year about KRAS-targeting drugs as cancer therapies, and that pretty much means G12C-KRAS targeting ones. That’s because years of effort had yielded no solid candidates for inhibiting “plain” KRAS in tumors, but a great deal of hard work has paid off in compounds that target one of the mutant forms of it that shows up, where a glycine at position 12 has turned into a cysteine.
Amgen (NASDAQ:AMGN) got a conditional approval in 2021 for the first one of these (sotorasib, or Lumakras). And as all drug development fans will appreciate, that means that the company is required to run a confirmatory trial.
Well, that trial (CodeBreaK 200, an open-label one with 345 patients) has been run, Amgen has been presenting the results from it this summer, the new data are up for FDA review, and now things are really hitting the huge high-speed industrial fan. If you’re a subscriber to Stat, this article from Tuesday went into excellent detail – otherwise, read on.
Here are the FDA briefing documents for the meeting held on Thursday, October 5. If you read Amgen’s press releases from the last two or three months and then read this stuff, you will feel as if you have stepped through a portal into another dimension.
Unfortunately, that new world may be the one that the rest have agreed to be the real one. Which means that the Amgen Press Release World starts to look pretty tenuous. Here’s why I say that:
1. The FDA notes first that the median progression-free survival (PFS) in this new trial was five weeks. But the standard imaging interval is six weeks, which makes you wonder how robust that five-week estimate is, given the mismatch in the data collection.
And in fact, it’s not so robust at all – even Amgen says in its submission that the median PFS might be as little as five days. If you’re not sure which of those it is, then you’re not sure if your drug works.
2. The control arm of the study (doclataxel treatment) had 23 patients who were randomized but were never actually dosed. There were only 2 such patients in the Lumakras arm, and that’s enough of a discrepancy to make you wonder about systemic bias in the results.
3. The FDA reviewers say that in their opinion, the clinicians evaluated the progression of cancer in these patients in ways that favored Lumakras. Remember, this was an open-label trial: the clinicians knew which patients were getting which drug, and open-label trials are open invitations to both conscious and unconscious bias in such calls.
Now, this trial did have blinded independent central review (BICR) independent of these physicians, but the FDA notes that there were notably more early calls of PFS for Lumakras by the investigators as compared to the BICR and more late calls of PFS for the control patients. Ideally, such disagreements should have a more random distribution.
4. The agency also says that in the crossover portion of the trial that there was early crossover of patients from the control arm to the Lumakras arm, with insufficient evidence of disease progression.
19 of the 46 patients who crossed over did so before a BICR call, and the agency believes that it is likely that these patients were relatively healthier than usual, distorting the eventual results.
This is a similar problem to the number of early dropouts in the control arm mentioned above, and having both of these problems pointing in the same drug-favoring direction is not a good look.
It turns out that there were discrepancies between the trial clinician evaluations and the BICR team all the way along, to the point that re-reads of the BICR data actually changed the progression-free survival figure from “not significant” to “statistically significant” at the trial’s interim read.
The document says that Amgen and the FDA discussed these numbers back in May, and that Amgen asked if these data would be enough by themselves. The FDA replied that no, this was insufficient evidence and requested that the trial continue.
But recurring trouble of this sort, they now say, “depletes confidence in the overall trial conduct and data integrity”. They go on to summarize by saying, “Multiple features of CodeBreaK 200 do not appear consistent with an adequate and well-controlled trial”.
Progression-free survival aside, this trial actually showed worse overall survival for the treatment group than the controls, but the FDA is more forgiving and believes that the most likely outcome was that there was no actual difference. But is that what you want in a confirmatory trial?
It’s all very bad news for Amgen, and the question now is not whether Lumakras will go on to full approval, but whether it will (eventually) remain on the market at all.
The FDA is just asking its committee, on September 6, to consider whether this trial can be reliably interpreted. The answer is probably “no”, and the recommendation will probably be to go run a real trial. With real numbers. Ones where you can say what they really mean.
Original Post
Editor’s Note: The summary bullets for this article were chosen by Seeking Alpha editors.
Read the full article here