I’ve been meaning to write about the fluvoxamine story, so today’s the day. Its history is as a CNS drug, a serotonin reuptake inhibitor that also has activity on sigma-1 receptors. It’s generally prescribed for depression, obsessive-compulsive disorder, anxiety, panic disorder, etc. and it’s been in use since the mid-1980s (and since the early 1990s in the US). Its biggest disadvantage is that it strongly inhibits several CYP metabolic pathways, which means that you have to step carefully in patients under treatment with other medications, lest their real dosages in the blood suddenly alter when starting on fluvoxamine.
In 2019, it was reported (PMC open version here) that its sigma receptor agonist activity might be useful in systemic inflammation (and in related conditions as well). That’s part of a hypothesis that the endoplasmic reticulum in cells was an underappreciated target in toxic inflammation. Fluvoxamine lowered cytokine levels and improved survival in mouse models of sepsis, and it was suggested as a possible new therapeutic agent in that area. Sepsis most certainly needs some of those – there have been multiple clinical trial failures over the decades, as one good idea after another has wiped out due to the complications of systemic inflammation.
With the recognition that coronavirus infections were leading to mortality and morbidity due to an overactive inflammatory response, a whole list of antiinflammatory therapies have been tried to dial that down. Many haven’t worked, while dexamethasone famously has. Fitting in with that “complications of systemic inflammation” take, there has really been no other way to sort this out other than in the clinic. Some of the failures have been with therapies that you might have thought would work (antibodies against different immune system components, for example), while dexamethasone itself didn’t seem to be useful in SARS or MERS, two previous outbreaks of coronavirus-mediated human disease. There are just too many pathways operating at the same time (and operating on each other) for us to think our way to the best solution: clinical data rule.
So news of a completed fluvoxamine trial in coronavirus patients certainly falls into that category. An observational study of patients taking antidepressants had picked up a possible effect, and a smaller trial did as well. This one examined about 1500 patients in 11 treatment centers in Brazil (Minas Gerais), divided almost exactly 50/50 into treatment and control groups (blinded), and the trial was actually stopped for superiority (always a welcome event). These were patients who were at risk for progression to severe disease, and the endpoint was whether they ended up hospitalized or in some other urgent-care setting when treated with fluvoxamine 100mg b.i.d for ten days. This is part of the TOGETHER trial, an adaptive-design effort that’s been evaluating a number of potential therapies (and whose treatment arms for hydroxychloroquine and for lopinavir/ritonavir had already been stopped for futility), so the recommendation was to stop assigning patients to the fluvoxamine treatment group while the trial continues in other areas.
79 patients had a primary outcome event in the treatment group, as opposed to 119 in the placebo group. 87% of those events were outright hospitalizations. Secondary endpoints such as viral clearance at day 7 and overall hospitalizations did not reach significance, but you have to wonder if that’s because the primary endpoint hit and the Bayesian design redirected the patients at that point (perhaps a frequentist design would have run longer and hit some of the secondary endpoints as well, but at the cost of the flexibility of the adaptive framework). No particular subgroups stood out as having different effects due to age, sex, comorbidities, etc.
This is good news, since the drug is cheap and widely available, and I hope that this news is immediately affecting clinical practice. The authors note that the sigma receptor anti-inflammation hypothesis is probably the explanation for these effects, but they can’t rule out antiplatelet effects and others. They also point out that this is a very rare example of a repurposed drug actually showing useful effects during the pandemic. Actually – my own editorial comment here – it’s a relatively rare example of a useful completed clinical trial during the pandemic, if we’re going to be honest about it. I think that the authors have similar feelings:
Since the start of the COVID-19 pandemic, there have been more than 2800 randomised controlled trials registered on ClinicalTrials.gov. However, fewer than 300 have been reported and most clinical trials have been small and underpowered, with sample sizes less than 100. In many cases, these trials have been unsuccessful at recruiting as the local epidemics occur in waves and sustainable infrastructure to maintain staff or local interest for recruitment is lacking. The trials that provide the clearest medical understanding tend to be the larger platform trials, such as SOLIDARITY,17 RECOVERY,18 PRINCIPLE,11 and REMAP-CAP.19 As a result, we actively collaborate with other investigators running trials with overlapping interventions so that they can be aware of our study decisions and establish whether they should influence their respective trials.
It’s been said before, but it needs to be said several more times: the clinical response to the pandemic, taken has a whole, has not been good, and we’re going to have to take steps not to repeat these mistakes if (or when) another such infectious agent appears. That’s another topic entirely, but going into what I refer to as Headless Poultry Mode shouldn’t be an option. This trial and the ones listed above have been the real beacons in this area, but the US has not exactly been performing up to where it should in this area.
