Well, everyone was just dealing with the news that Gilead’s antiviral remdesivir had had two trials suspended in China, when Stat‘s Adam Feuerstein and Matthew Herper broke news last night about a trial here in the US.
125 patients had been recruited at the University of Chicago to receive daily doses of remdesivir, and the only reason we have news of what’s going on is because of a series of mistakes. A Chicago infectious disease specialist overseeing the trial there mentioned during a video chat with colleagues that most of these patients had been discharged, with only two fatalities. The video was recorded and then leaked to Stat. And let’s be honest about this part: this was a severe breach of the trial protocol, the sort of thing that under other circumstances could lead to the whole thing being invalidated and not accepted by the FDA as evidence. That’s not going to happen here; everyone is just going to roll their eyes, kick their desks, and find a way to deal. But this is not the way to handle trial data – it complicates everything at a time when we don’t need to be doing that. The Chicago trial supervisor should have known better than to discuss results like this to a crowd of people over video, and whoever leaked it should have known better than to do it. Let’s not have any more of this.
Should Stat have published? That’s a classic journalism question: this was certainly news, and journalists report news. It seems certain that if they had declined that the video would have been sent to any number of other news outlets; the cat was already exiting the bag. Someone was going to run with this story, and overall I’m glad that this got sent to Feuerstein and Herper rather than to Sean Hannity or Dr. Oz. I don’t like this leak at all, but the parts I like the least are the steps that happened before the news got published.
But now that it’s out there, let’s talk about what’s in the leak. Gilead stock jumped like a spawning salmon in after-market trading on this, and one of the reasons was that that 113 of the 125 patients were classed as having “severe disease”. People ran with the idea that these must have been people on ventilators who were walking out of the hospital, but that is not the case. As AndyBiotech pointed out on Twitter, all you had to do was read the trial’s exclusion criteria: patients were not even admitted into the trial if they were on mechanical ventilation. Some will have moved on to ventilation during the trial, but we don’t know how many (the trial protocol has these in a separate group).
Note also that this trial is open-label; both doctors and patients know who is getting what, and note the really key point: there is no control arm. This is one of the trials mentioned in this post on small-molecule therapies as being the most likely to read out first, but it’s always been clear that the tradeoff for that speed is rigor. The observational paper that was published on remdesivir in the NEJM had no controls either, of course, and that made it hard to interpret. Scratch that, it made it impossible to interpret. It will likely be the same with this trial – the comparison is between a five-day course of remdesivir and a ten-day course, and the primary endpoint is the odds ratio for improvement between the two groups.
So we will have the choice to like remdesivir or to love it; there will be no direct standard of comparison for how these patients will have done without it. Everyone will be trying to synthesize such a comparator from other clinical trials and reports, but that’s a dangerous business. I hope that we can learn something from the subgroup analyses, but there’s a limit.
Bottom line: I’m sounding like a defective parrot here because I say this so often, but we have to wait for controlled trials in order to say anything definite. Such trials are underway, with actual comparisons to standard of care, but they take longer. Fast trials are generally not very interpretable, interpretable trials are generally not fast. I will be glad to see these numbers when they appear, but don’t believe anyone who runs with a “Cure for Covid!” headline, because it’s extremely unlikely that remdesivir (a single agent with a broad mechanisms that’s not optimized for this virus) is any such thing. Remember, there are as yet no single-small-molecule antiviral cures for anything, coronavirus or not. My hope for the drug is that it is effective enough to get people out of the hospitals more quickly and to keep more of them off ventilators than if they were not taking it. For that hope to be realized, we need that comparison to the people who are not taking it. This trial doesn’t have it.