It’s been a while since I wrote about what we generally call “Long Covid” – the continuing medical problems reported by some patients who went through infections with one of the SARS-CoV2 variants. The main reason has been that I have not been able to make very much sense out of the area. And it isn’t just me: this issue is baffling a lot of people, just as previous reports of other long-term viral effects (Lyme, Epstein-Barr, and several others) have. For decades now these topics have been a guaranteed way to start a fight in any mixed crowd that has some interest in medicine. Do some viral infections have long-term effects? Definitely – there are well-documented cases, although we don’t always understand how they work.
To that question, do such things happen because of a (semi)permanent derangement of the immune response, something that could keep rolling along after the original viral infection has long been cleared? We certainly know about things like this; there’s no doubt that this is a real mechanism. Or does the virus never really get cleared? We know about diseases and conditions that work this way, too, and it could be happening with the coronavirus. Many Lyme disease sufferers who report long-running symptoms after multiple courses of treatment lean towards this explanation. And there is another explanation that’s always waiting out there, too: that some of the cases are more in the realm of somatic symptom disorders, and that you may never be able to find a conventional biochemical explanation at all. This suggestion will almost invariably raise the temperature in one of those mixed-crowd discussion I was imagining earlier. Some people are ready to apply this explanation almost tout court, but there are plenty of others who regard it as an insulting cop-out on the part of medical science.
In case you’re wondering, I certainly don’t rule out that last explanation, at least on a case-by-case basis. But I have trouble applying it to the whole mass of people who report these sorts of symptoms. Given how little we know about the human immune system (and perhaps how little we know about its relation to the central nervous system), I find it irresponsible to dismiss the whole list of long-virus cases as “all in the head”. Not to mention finding it arrogant. Humility should never be far from our minds in this business, because humiliation itself is never far from us when we get too sure of ourselves – pride goeth before destruction and all that
So this new report definitely gives us all something to think about! It’s just come out from under embargo, and it details a what seems to be a well-controlled trial (fully blinded, randomized, placebo-controlled) in 1162 patients (30-85 years old, median age 45, 56% female). These patients were all overweight and all had confirmed coronavirus infections, which were treated with standard-of-care. In addition, patient groups were also treated with either fluvoxamine, metformin, or our old friend ivermectin: there was a metformin plus ivermectin group, a metformin plus fluvoxamine group, a metformin plus placebo group, an ivermectin plus placebo group, a fluvoxamine plus placebo group, and finally a control group that got (you guessed it) placebo plus placebo. These multifactorial designs are not fun to set up and run across widely separated medical centers, but they can tell you a lot if you can get them to land. And it looks like this one did.
8.3% of the patients total reported a “long Covid” diagnosis by day 300 of the study, if you’re looking for another rough estimate of how common this is. It should be noted that there was no apparent connection between the severity of disease and the likelihood of a long Covid diagnosis. And here’s how those different patient groups break down with those diagnoses: first off, there was no effect with either ivermectin alone or fluvoxamine alone. compared to placebo. That leaves metformin, and in that case the metformin placebo group had a 10.3% cumulative incidence rate of long Covid diagnosis, while the actual treatment group showed only 6.3% incidence. Metformin’s effects held up in subgroup analyses, and what’s more, this result is consistent with an earlier report (the TOGETHER trial) of metformin treatment lowering long Covid incidence. These two together raise the take-it-seriously level considerably.
One’s thoughts naturally turn to how metformin could be working, and the authors have this to offer:
The exact pathophysiology of long COVID is unknown but is likely to be multifactorial, including the inflammatory cascade during acute infection and persistent viral replication. Mechanistic in-silico modelling predicts that translation of SARS-CoV-2 viral proteins is a particularly sensitive target for inhibition of viral replication, and previous studies have shown that metformin is capable of suppressing protein translation via mammalian target of rapamycin (mTOR) inhibition.
That’s as good a place as any to start, but it’s almost too straightforward, if that makes any sense. It has to be noted that metformin has pleitropic effects that have shown up with a number of diseases. That simultaneously gives you some belief that the drug really could be working here too, while making the actual mechanism hard to pick out. But the work to do that seems very much warranted. This also makes a person wonder if metformin could help with other long-virus syndromes – you could learn some interesting things by seeing where it works and where it doesn’t, for sure.
An overarching point, though, is made in a valuable commentary in the same issue of the Lancet by Jeremy Samuel Faust: this effect against a placebo control really shows us that there is something there. Long Covid is looking less like a cloudy mass of speculation and shoulder-shrugging, and more like a problem that medical science can start to unravel, using the tools we already have:
When a disease is too poorly defined, it follows that it is almost impossible to modify either the incidence of that disease or the distribution of its outcomes—that is, unless the treatment effect is so great, and the true target population so common in the assembled denominator, that any corresponding signal dilutions are offset. The present study suggests that, even with definitions as amorphous and heterogenous as those currently in use for diagnosing long COVID, there was to be found within this study population an ample cohort of individuals with syndromes similar enough that disease incidence could be modified, and metformin appeared to achieve that. Furthermore, the finding that long COVID is modifiable, although here showing prevention, offers hope. . .
That is does! A story to watch.
