So you can see that these two came out rather differently, with the Zhejiang study showing no detectable difference on treatment and the Wuhan one showing what looks like a real effect, especially on radiological progression of pneumonia (which I have to say is a very strong endpoint to measure). Here, then, is a good exercise in interpreting clinical trial statistics: we are now one-and-one after
two small hydroxychloroquine trials: which one (if either) reflects the real-world situation? Update, for medicinal chemists et al.: people have asked about the once-a-day dosing in the Zhejiang study versus splitting to twice-a-day in the Wuhan one. The human pharmacokinetics of hydroxychloroquine are well worked out (as they should be for a drug of that vintage). A 200mg oral dose hits its Cmax in the 3-hour range, but boy, does it tail off slowly after that: plasma half-life is 123 days (!) with a large volume of distribution (extensive uptake in tissue). About 10% of the dose is excreted as parent, with metabolites still showing up in urine after three months. The metabolites peak in the blood about the same time as the parent compound after a dose, so it’s not that the compound doesn’t get metabolized – that long half-life is due to distribution. doses, so you’d figure 200 and 400 oral would likely compare in the same way. I believe that toxicity and QT prolongation are CMax driven, and that’s the likely reason for dosing b.i.d. Since we don’t have a good idea of the mechanism for any antiviral effects, it’s hard to say if those are more CMax or more AUC driven, though. You could argue that overall we’re seeing either no benefit or some benefit here, which is good. As for adverse events, neither trial reported anything serious, But both of them excluded patients with any sort of cardiac arrhythmias, a wise precaution since one of the most acute worries with high doses of hydroxychloroquine is QT-interval prolongation, and you don’t want to do that to anyone with any underlying problems. So as long as such patients are excluded, for now hydroxychloroquine is in the “might do nothing, might do some good” category, which under the current conditions seems sufficient for treating patients, pending further data. You will notice that we are not exactly in the “total cure” category that the Marseilles group has been putting itself in, but frankly, these results from China are more like what I expect from the clinic (at best!) when using a repurposed drug against such a pathogen. One thing to note about the clinical data in this situation: fatality rates are notoriously hard to estimate in an epidemic, and often only become clear after things have completely settled down. But you would probably not go far wrong estimating this one for the total infected population as around 1% – maybe a bit better, maybe a bit worse. It’s higher when you look at the patients who are admitted into hospitals, naturally, and it most certainly gets higher as you stratify by age groups. So if you want to see how many people you’re keeping from dying with any given therapy, you need a larger sample than anything we’ve seen so far. Getting people out of the hospital more quickly, or keeping them out of the ICU or off ventilators, though, are very worthy goals in themselves, and if HCQ treatment can help with those it’ll be most welcome. We have little or no data on these yet. There are some things that need to be noted about this latest work, though. As Leonid Schneider has commented on PubPeer, the original trial as registered in China looks quite different from what we see here. Update: see this comment as well. The design was for 100 control patients, another 100 patients to receive one dose of hydroxychloroquine, and 100 more to receive a higher dose. What we have, though, are only two groups of 31 patients each, which suggests that there were problems with the inclusion criteria for the trial and/or with patient recruitment. The trial design also called for endpoints of negative results for viral RNA, and for “T cell recovery time”, to be collected by sputum and throat swabs and by blood samples, respectively, and none of this shows up in the preprint at all.
