Skip to Main Content

An arthritis drug that was seen as a promising treatment for some Covid-19 patients delivered disappointing results in clinical studies, its makers, the drug firms Regeneron and Sanofi, said Monday.

The result could have an impact not only for their treatment, Kevzara, but also for a similar drug from Roche, Actemra, that is being used off-label in many hospitals. It also may lower the odds that other repurposed medicines used against autoimmune diseases like rheumatoid arthritis will benefit Covid-19 patients.


“When you try everything under the kitchen sink, most of the time it’s not going to deliver the results that you want, no matter what the small 20-patient or 30-patient studies say,” said George D. Yancopoulos, Regeneron’s co-founder and chief scientific officer.

Kevzara was not expected to directly block the coronavirus, SARS-CoV-2, that is causing a global pandemic. But it was hoped that the drug would help ease the immune system’s overreaction to the virus — a “cytokine storm” that causes inflammation and fluid buildup in the lungs of many of the sickest patients — potentially helping to keep patients off of ventilators or saving their lives. Early data from a 21-patient study in China using Actemra had appeared promising.

But there was no benefit in the group of patients most like those in that Chinese study: those termed “severe,” meaning they needed oxygen, but not with air pressure that is faster than normal breathing and certainly not those on ventilators. A 276-patient study in this group was stopped because there was no chance it would succeed.


There is still a glimmer of a positive result in sicker patients, whom the companies call “critical.” These patients need what is called high-flow oxygen, or are on ventilators. In that group, an extra 1 in 10 patients who received high-dose Kevzara was discharged from the hospital compared to those who received placebo. A larger study is continuing in the hopes of proving this benefit.

“It’s not, unfortunately, black and white that’s going to cure everybody,” said Yancopoulos.  But he said that such a difference, if confirmed, would still be important. For that 1 person in 10, “it’s all the difference in the world.”

The Kevzara trial had been designed to be run in two stages: a Phase 2 study in which researchers would try to understand which measures of the disease were important, and a larger Phase 3 study in which they would try to confirm those results with the rigor necessary to achieve regulatory approval. In both stages, patients were randomly assigned to receive one of three options: a placebo; a 200-milligram dose of Kevzara; or a 400-milligram dose of Kevzara. Going forward, the patients in Phase 3 will receive either a placebo or the 400-mg. dose of Kevzara.

When it came time to analyze the Phase 2 results, there had been no apparent benefit for patients taking the drug when the severe and critical groups were combined. But in the severe group, it appeared that patients who received Kevzara were actually doing worse. This led a panel of outside experts who were looking at the data for the larger Phase 3 study to look at the severe group. They found that patients were not, in fact, being harmed by Kevzara. But there was also no hope the study would show a benefit for the severe patients.

Part of the reason: The results in the Kevzara arms were similar to those seen in the 21-patient study that had been published in China, which had lacked a control group of patients who had received a placebo or standard treatment. But the patients in the new study’s placebo arm did as well.

Overall, all of these patients, though hospitalized, did better than researchers had expected. In the severe group, 80% of patients were discharged from the hospital, 10% died, and 10% are still hospitalized.

In the critical group, the results were more positive, but still subtle. In the placebo group, 55% of 77 patients died or were on a ventilator; this compared to 46% of the 94 who received 200 mg. of Kevzara, and 32% of the 88 who received 400 mg. of Kevzara. Similarly, 41% of the placebo patients were discharged from the hospital, compared to 39% in the 200-mg. group and 53% in the 400-mg. group. Those differences would be positive if shown in a larger study, but in the current study they could be the result of the play of chance.

Regeneron and Sanofi said that they would share the results with the data monitoring committee of a study of Kevzara they are running outside the U.S. Roche is conducting its own study of Actemra in Covid-19 patients.

Yancopoulos said that the experience with Kevzara holds a lesson for other drugs being tested against Covid-19: Data from large, randomized trials where a defined group of patients are assigned to get an experimental drug or the best treatment available are necessary to know what works and what doesn’t.

“It shows how hard it is,” he said. “It shows that a lot of times when people think they’re seeing things in desperate situations, when they’re just trying whatever they can, that a lot of times that leads to misleading hypotheses and misleading results. Which is why it’s so important, so crucial, to get this sort of data to really guide physicians and patients in terms of which drugs are really working, in what setting, and for what patients.”

Yancopoulos expressed hope regarding Regeneron’s other effort at developing a Covid-19 drug: creating synthetic antibodies against the virus, an approach that had positive results against the Ebola virus. Those studies, he said, are expected to begin in June.

  • The science behind Kezvara as an interleukin inhibitor given as one dose never suggested that it would prevent cytokine storm prior to its onset. Once patients have cytokine storm (the so called critical patients) Kezvara will ameliorate the degree of damage from the deadly storm. At 400 mg dose, the data so far is pretty compelling that it is effective in reducing death in patients in ICUs on ventilators experiencing cytokine storm. I did not find the results disappointing, having never expected it be ever be used on the less ill patients. 60,000 patients in the US have died of Covid-19, the majority on respirators in the midst of a cytokine storm. Kezvara could have reduced that number by at least 12,000 people. Great results so far.

  • Vast data analysis is essential for the direction of what might work for a new disease. There is ample material to work with for technology to amass vast databases from the hundreds of thousands of Covid cases. To be serious about speeding up a cure / vaccine, data must be shared and analyised : race, gender, social habits, medical history, family medical history, prior vaccinations and how recent, prior medications, Covid tests and results, drugs used for Covid treatment, disease progress and timelines, treatment results. Everything is available in patients charts – hundreds of thousands of patients charts. This information must be used – it is of utmost value.

  • Two observations:

    1) The high dose vs. placebo Kevzara results in critical patients is a fair bit better than a ‘subtle’ result. 58% were completely off oxygen vs. 41% in the placebo. While most of the mortality/ventilation differential between the arms was in patients remaining on ventilation in the control group, experience elsewhere has shown patients who have been on ventilators for weeks with COVID-19 rarely pull through. I.e. Odds are the mortality rate rises materially higher in the control group over the coming days/weeks vs. the high dose critical patient group.

    2) Clinicians need to remember that this trial DOES NOT necessarily give a good indicator of how their patients will recover in the absence of any treatment ex supportive care. The control group in both the severe and critical groups received a placebo. We have no way of knowing whether mortality in patients not receiving any care other than ventilators / high flow oxygen will be higher without the potential psychological benefit of the placebo effect. In fact the article notes that clinicians were surprised by how well the severe patients did in both groups.

    Was a P-value provided for mortality?

Comments are closed.