Skip to Main Content

When the patient came back 10 days later, coughing repeatedly and complaining of headache, Davey Smith feared the worst.

Smith had prescribed the patient Pfizer’s new antiviral pill, Paxlovid, on the previous visit, after a Covid-19 test came back positive. A resurgence of symptoms probably meant one thing, especially after Smith tested the patient and got another positive.


“I was pretty sure it was resistance,” said Smith, an infectious disease physician at the University of California, San Diego. “I’m a virologist, I combat resistance all the time.”

In this case, he was wrong. When Smith and his colleagues analyzed the virus, they found the patient was actually an early example of Paxlovid rebound, where the virus returns without evolving around the drug.

But that doesn’t mean scientists can drop the issue. Smith was on watch for a reason. Resistance is the hobgoblin of antiviral medicine, even with antivirals as effective as Paxlovid. After doctors deployed nearly every new virus-killing infusion or pill in history, strains popped up — either immediately or eventually — with machinery warped in just the right way to evade the threat.


Exactly how much of a problem resistance will be for Paxlovid is complicated. In some patients, the coronavirus will inevitably find ways to evade the pill, as it did prior Covid-19 drugs.

“If there is anything we know about viruses and antiviral drugs is that eventually we will see some sort of resistance,” Andrew Pavia, chief of pediatric infectious diseases at University of Utah Health, said in an email.

What’s less clear, Pavia and other experts say, is whether any resistant variants will spread widely. The coronavirus may have particular difficulty getting around Paxlovid compared to other drugs because patients take it for only five days and because it targets a protein the virus can’t easily change. Any mutation or modification the virus makes may impair its ability to replicate or survive.

“At some point, there will be Paxlovid-resistant virus,” said Adam Lauring, who studies RNA virus evolution at the University of Michigan. “Whether that clinically becomes a problem or not, it’s hard to say.”

As the drug becomes more widely available, though, some researchers fear we may be using it in ways that raise the risk such a strain emerges. Paxlovid was authorized in December but only about 33,000 prescriptions were filled in the U.S. per week until this month, when the rate skyrocketed to 160,000. On Thursday, the White House announced further measures to educate physicians and make it even easier to get treatment.

Those efforts can help vulnerable patients who are cut off from the health care system or had a reluctant doctor to get access to a potentially life-saving drug. But some experts fear the White House’s efforts may also encourage doctors to prescribe Paxlovid to relatively low-risk patients.

In other words, the drug may currently be under-prescribed to high-risk people and over-prescribed to low-risk people, said Walid Gellad, director of the Center for Pharmaceutical Policy and Prescribing at the University of Pittsburgh. And all those low-risk prescriptions can create new opportunities for resistance.

“People shouldn’t be taking Paxlovid unless they bloody need it,” said Katrina Lythgoe, evolutionary virologist at Oxford.

Pfizer has been thinking about resistance throughout the drug’s development, said Annaliesa Anderson, the pharma’s chief scientific officer of bacterial vaccines.

Researchers started by trying to create resistance in the lab. At a biosafety level-3 facility they passed the virus through different cell dishes in the presence of low levels of nirmatrelvir, the molecule within Paxlovid that inhibits the virus.

It’s like an obstacle course tournament. In each dish, the drug is present at low enough levels to block some but not all the virions. After a set interval, researchers take the surviving virions from one dish and put them in a fresh one with the same setup. Initially, virions don’t necessarily have to be resistant to survive — some will just be lucky — but over time the virions with random mutations that allow them to sidestep Paxlovid will be more likely to make it to the replication stage.

How long it takes to find resistant strains, how many different strains arise and how “fit” those viruses are, can indicate how easily the virus will evade an antiviral in the real world. Similarly, Pfizer researchers analyzed a massive online database of coronavirus sequences to track the virus’s evolution and see if there’s been significant change in its protease, the protein Paxlovid targets.

Both efforts suggested that, while resistance was possible, it would be difficult. Since the start of the pandemic, Anderson said, the protease had changed 10 times as slowly as the spike protein that vaccines and monoclonal antibodies target.

They also analyzed virus from patients in their clinical trial.

“At this point, we haven’t seen anything that concerns us in those clinical studies,” she said. “So we’re not looking at something where one could expect potentially a very rapid rate of resistance.”

If the drug holds up, it may be due to that unmoving target: The protease, a slicing-and-dicing protein essential for viral replication. Numerous viruses have difficulty shifting their protease to evade medicinal threats. HIV patients, for example, can develop resistance to protease inhibitors, but it takes longer than with other drugs. And it comes at a cost.

Some researchers believe the same will be true with the coronavirus and Paxlovid.

“If resistance does emerge within an individual, my guess [is] it will be detrimental to how well the virus replicates,” Lythgoe added in an email. “If so, it may be unlikely to spread.”

Others, though, are less sanguine. “I think it’s too early for us to be that optimistic,” said Jonathan Li, director of the Harvard/Brigham Virology Specialty Laboratory.

He’s not alone. Jun Wang, a medicinal chemist at Rutgers, analyzed Omicron-variant sequences in March and found the same thing Pfizer did: Paxlovid still neutralized the virus. But he also spotlighted 25 different mutations in the protease.

None conferred resistance or were even at the site of the protein that the drug binds to. But changes on any point in any protein can affect the shape of the binding and how it interacts with other molecules.

“Everybody understands this: Resistance is not a question of if, it’s a question of when,” Wang said.

Much will depend on how the drug is used and in whom. Although experts emphasized the drug should still be used for anyone high-risk, many said they were concerned about three groups.

First, there are patients who don’t take the complete five-day, 30-pill course, because they decide they feel better, even if the virus hasn’t been entirely eliminated. Additionally, they may stockpile the leftover pills and pull out a few when someone else in the household gets sick.

As in Pfizer’s lab experiments, it can expose the virus to subtherapeutic doses of the drug and promote resistant variants. It’s the same reason doctors emphasize taking antibiotics exactly as prescribed.

We should be “using the drug widely, but carefully,” Pavia, the Utah physician, said in an interview. “We don’t want people to be taking one day or two days of drug, we don’t want people stockpiling it, because that could cause problems in the future.”

They are also concerned about patients experiencing Paxlovid rebound who decide to take a second course of the drug when the virus comes back, because they’re giving the virus more opportunities to find escape strategies. Although the FDA and CDC explicitly advise rebounders not to take a second course, Li suspects many are.

At one point Pfizer CEO Albert Bourla outright encouraged patients to do so, noting that doctors often prescribe more antibiotics if the initial dose didn’t clear the infection. (In an email, A Pfizer spokesperson said they advise patients to speak with a health care professional.)

“The more days, the more hours the virus gets exposed to the drug,” said Li, “the more likely it is to develop resistance.”

Lythgoe, however, argued you might want to give some high-risk rebounding patients Paxlovid, because there’s a non-zero chance they face severe disease. And, as the infection persists, and the patients’ immune systems struggle to fend off the rebound on their own, you run the risk of creating new immune-evasive variants.

“It’s layers and layers and layers of complication,” she said. “It’s a really difficult problem and it’s really unclear to me what the best course of action is.”

Sara Cherry, an evolutionary virologist at University of Pennsylvania, also argued researchers weren’t monitoring enough rebound cases for possible resistance.

“We haven’t been systematic,” she said. “So we don’t really know.”

The most complicated group, however, is the immunocompromised, such as transplant patients and patients on certain cancer drugs. These patients stand to benefit most from Paxlovid’s profound efficacy, because they often get inadequate protection from vaccines.

But these patients are also most likely to give rise to resistance. Throughout the pandemic, researchers have shown immune-resistant variants rising in immunocompromised patients, where there’s enough of an immune response to train the virus but not enough to clear it entirely.

Some researchers fear a similar scenario with Paxlovid. In most people, the immune system teams with Paxlovid to suppress the virus. In the immunocompromised, however, Paxlovid may be on its own.

That might mean immunocompromised people need a longer course of Paxlovid or combination therapies that hit the virus at multiple points and leave it little room to maneuver — strategies that might also benefit everyone. But so far Pfizer hasn’t launched any such trials and academics told Bloomberg that the company has been unwilling to provide the drug for such trials.

Myron Cohen, an infectious disease physician at University of North Carolina-Chapel Hill, said he wanted to start a study combining Eli Lilly’s antibody bebtelovimab with Paxlovid in immunocompromised patients, although he has yet to ask Pfizer. Ravindra Gupta, who studies antiviral resistance at Cambridge University, said he’d like to see studies combining Paxlovid with fluvoxamine, an antidepressant that has shown limited efficacy against Covid-19.

David Boulware, an infectious disease physician at the University of Minnesota, said he also received “a strong no” when he tried combination trials with the other antiviral pill now available, Merck’s molnupiravir.

Reiterating that they had seen few signs of resistance, a Pfizer spokesperson said researchers were “proactively evaluating” approaches should resistance emerge, such as combinations. And a Merck spokesperson said the company initially prioritized supply commitments after molnupiravir was authorized but were “currently working through opportunities for future investigator-initiated studies.”

But Gupta argued there was little time to spare.

By the time we have the information, it may be too late to stop the spread of resistance,” he said. “That’s the worst-case scenario.”

Correction: An earlier version of this story misspelled Sara Cherry’s name and misstated the number of pills in a course of Paxlovid.

Create a display name to comment

This name will appear with your comment

There was an error saving your display name. Please check and try again.