How Does Dexamethasone Help COVID-19 Patients?
In mid-June, a press release brought some hope for treating patients with COVID-19: “Dexamethasone is the first drug to be shown to improve survival in COVID-19.” These findings came from the RECOVERY trial, a large randomized clinical trial in the UK designed to evaluate four different treatments for hospitalized COVID-19 patients. The results made a big splash, both in the optimistic headlines that followed, and the scientific community, where some were miffed that a press release was published before any data or journal paper. The full results have since been published as a preprint (which is to say, before full review by journals and scientific peers), so let’s dig in to better understand what the trial found.
According to the preprint, the dexamethasone (dex) arm of the trial enrolled 2,104 eligible COVID-19 patients across 176 National Health System hospitals in the UK between March and June. These 2,104 patients were randomly selected from among the ~11K participating in the larger RECOVERY trial, and they were generally well-matched with the 4,321 patients in the control arm in terms of disease severity and demographics. At the start of treatment, 16% of patients across both arms were on ventilators (or equivalent) and 60% were receiving oxygen only, leaving 24% with neither treatment. By the nature of the trial, all patients were hospitalized and COVID-19-positive, and thus we should not try to generalize the results to cases of COVID-19 that don’t require hospitalization.
Because the trial was randomized without stratification (that is, patients were placed randomly without regard for matching age or other demographics across treatment groups), there were some minor differences between those assigned to the dex arm versus the control. Most notably, the dex patients were 1.1 years older than control patients on average. The researchers were able to adjust for age when running the analysis, though, and they did not find that this difference had any impact on the results.
One of the most important parts of planning a clinical trial is picking the right outcomes to measure and monitor. In some recent COVID-19 trials, such as for remdesivir, the primary endpoint was “time to recovery.” In the RECOVERY trial (perhaps ironically given the name), the primary endpoint was 28-day mortality-- that is, how many patients had died after 28 days of treatment. The difference is subtle, but it does affect the way results are counted and assessed. For example, if a patient was still in the hospital after 28 days, this would be considered a successful treatment in the RECOVERY trial, but could be a failed treatment in the remdesivir trial. We might bicker about what the ideal endpoint is here, but, hopefully, these are at least directionally equivalent.
So, what happened with those 2,104 patients receiving dexamethasone treatment? Looking across all of them, there was a small but statistically significant reduction in deaths: 24.6% of control patients died within 28 days, but only 21.6% of dex-treated patients died within 28 days.
“Errr,” you say, “that doesn’t seem to be much of a difference.” And it’s true; that difference doesn’t change the mental math of, “Will patient X die?” However, if we analyze the numbers according to the severity of symptoms, a clearer picture emerges. Among the 16% of patients on ventilators, 40.7% of control patients died within 28 days, but only 29% of dex-treated patients died within 28 days. To put those numbers another way: for COVID-19 patients requiring ventilators, treatment with dexamethasone increased survival by 35%. That’s big, and has the potential to save many lives, especially among the sickest of patients.
Patients requiring oxygen but not on ventilators likewise saw a decrease in mortality when treated with dex, though the difference looked very similar to the overall count: 25.0% of control patients on oxygen died within 28 days, but only 21.5% of dex-treated patients on oxygen died within 28 days.
Now, if you’re doing the math, you should conclude that the patients receiving neither ventilator support nor oxygen actually did worse with dex treatment, and you wouldn’t be wrong: 13.2% of control patients without ventilators or oxygen died within 28 days, but 17.0% of dex-treated patients without ventilators or oxygen died within 28 days. Given the number of patients in this group (hundreds), this difference is not statistically significant, but it does strongly imply that giving dexamethasone to everyone or to mild COVID-19 patients is a bad idea, and could potentially cause more harm than good.
|Figure 1 from the paper shows that fewer patients die when treated with dex across all patients, patients on ventilators, and patients with oxygen support. However, patients not receiving any oxygen do slightly better without dex treatment (not statistically significant).|
What is dexamethasone, and what does it do?
These results might at first seem paradoxical: how could dexamethasone help the most sick patients, but hurt those with more mild COVID-19 symptoms? The answer lies in the nature of dexamethasone, and the reason it was chosen as a possible treatment to begin with. Dex is a type of steroid called a corticosteroid, and it is a cheap and common pharmaceutical used for many illnesses requiring steroid treatment, including rheumatoid arthritis and croup.
The primary role of dex is to bind to glucocorticoid receptors, which are steroid receptors that exist on the surface of most cells across the body. Once a glucocorticoid receptor is bound, it triggers signaling pathways within the cell that turn off transcription of inflammatory genes and turn on transcription of anti-inflammatory genes. The net effect of dexamethasone, therefore, is as an anti-inflammatory hammer. Many different cells, especially within the immune system, turn down their alarms and stop anti-viral or anti-bacterial behavior. This is why dex is an effective treatment for illnesses like rheumatoid arthritis, where the primary cause is immune system overreaction, and this is also why dex is often used in laboratory settings to study anti-inflammatory transcription in macrophages and other cells.
Dexamethasone is a common and relatively cheap pharmaceutical.
What does that mean for COVID-19? Well, you’ll note that I’ve been talking about the disease condition (COVID) and not the actual virus (SARS-CoV-2), and that’s very intentional. Dex does not do anything (that we know of) to prevent the spread of the virus, or to help the body overcome the infection itself. On the contrary, we would expect Dex to quiet the immune system, and thus to make the body less able to fight off SARS-CoV-2. This may in fact be what the results in the more mild hospitalized patients show us-- that dex treatment too early or in mild cases may have a negative affect. (Though we should be careful here: the RECOVERY trial does not provide sufficient data to prove that hypothesis, and we would have to look to further studies to draw that conclusion.)
The beneficial effect of dexamethasone on more severe cases, researchers reason, is likely due to the fact that the acute respiratory failure many patients suffer from is not a direct result of the virus, but a result of immune overreaction in the lungs following viral infection. In their attempt to quell the infection in the lungs, cells of the innate and adaptive immune system create a hyper-inflammatory environment in lung tissues, and this in turn causes respiratory failure. Thus, patients on ventilators often show a loss of circulating immune cells (since many are localized to the lungs and other affected tissues) as well as high levels of inflammatory immune signaling molecules. In other words, for these patients, the threat of the virus has been superceded by the threat of an immune storm, and dex, in turning down immune and other cells, increases the chance of survival.
So, we can say based on the RECOVERY trial that dexamethasone significantly improves the chance of survival for some of the most severe COVID-19 cases, but we should be careful to caveat that this does not make it an all-purpose treatment for COVID-19, and it certainly is not a vaccine or early treatment.