By Daniel A. Sweeney, MD, Cameron R. Wolfe, MBBS, MPH & Andre C. Kalil, MD, MPH 

In the early days of the pandemic there was an understandable rush to define and optimally treat COVID-19. Anecdotal evidence and the opinions of eminent scientists and non-scientists overwhelmed social and mainstream media platforms only to eventually be overruled by the results of careful, scientific analyses and well-designed clinical trials. Now, more than 2 years into the pandemic, we must return to the standard of careful and thorough analyses for all interventions and treatments.

The debate about "COVID-19 rebound" after nirmatrelvir/ritonavir (Paxlovid) treatment is one of these timely areas warranting further investigation. Continuing down the current path of uncertainty has consequences for how and by whom this antiviral should be used. However, by applying lessons learned from the early days of the pandemic -- including acknowledging the importance of randomized controlled trials (RCTs) -- we can avoid repeating the same mistakes. To do this, it is necessary to start by defining the question, identifying current knowledge gaps, and only then can one propose scientific solutions to bring a rapid resolution to the COVID-19 rebound controversy.

The Need for a Sound Definition Before Studying COVID-19 Rebound

Paxlovid consists of two drugs: nirmatrelvir, which inhibits a SARS-CoV-2 protease inhibiting viral replication, and ritonavir, which slows the inactivation and breakdown of nirmatrelvir. Per a CDC health advisory released in May, COVID-19 rebound is defined as a return of symptoms or a "new positive viral test after having tested negative" occurring "2 to 8 days after initial recovery." We just saw this over the weekend in the case of President Biden.

This definition of rebound is challenging and prone to inflating the incidence of rebound. It is possible some individuals identified as having "Paxlovid rebound" may have been experiencing a waxing and waning of COVID-19 symptoms while some unknown number of other reported rebound cases could be due to the known limitations of COVID-19 testing.

First, infectious disease specialists have long witnessed that some respiratory illnesses can have a bimodal course. This phenomenon has been reported in patients with COVID-19, albeit best described in hospitalized patients with more severe disease. Additionally, secondary bacterial infections (e.g., sinusitis, bronchitis, pneumonia) have been known to occur after COVID-19, which could also be misidentified as COVID-19 rebound.

Second, COVID-19 antigen testing -- a convenient diagnostic tool -- does have limitations. It is plausible, for example, that some individuals, after getting COVID-19 and completing a course of nirmatrelvir/ritonavir, initially get a false negative test (especially if they have a low viral load); then, after testing again out of precaution a few days later, could get a positive result due to viable or dead viruses -- an antigen test cannot distinguish between them. Finally, some patients who have clinically recovered will continue with positive PCR COVID-19 testing for several weeks, which could explain some of the anecdotal reports of COVID-19 rebound.

Third, recall bias is another challenge to defining COVID-19 rebound and its incidence. It is a well-known phenomenon that patients who receive a particular therapy are more likely to report side effects or symptoms during their clinical course compared to individuals who are not treated. Thus, the problem of identifying cases of COVID-19 rebound solely based on observational or anecdotal reports, and not as part of a controlled trial, is that the number of cases may be falsely increased.

Few Studies Specifically Describe Paxlovid Rebound

The data supporting the existence of COVID-19 rebound after nirmatrelvir/ritonavir treatment is limited. Most studies were conducted prior to the emergence of Omicron and employ a retrospective design prone to various biases. Nonetheless, the data suggest the rebound phenomenon may occur in only a small number of high-risk patients and the findings to date don't offer sufficient evidence that it is definitely a result of nirmatrelvir/ritonavir.

A secondary analysis of the EPIC-HF trial (the basis for the antiviral's authorization) revealed that approximately 1% to 2% of all unvaccinated individuals (treated and untreated) tested positive after having a negative test. And a recently published study showed that COVID-19 rebound occurred in only four of 483 (0.8%) vaccinated, high-risk patients. The incidence of rebound among vaccinated and unvaccinated patients infected with the Omicron variant is unknown.

There are three notable, albeit small, case reports recently published in the medical literature regarding COVID-19 rebound after nirmatrelvir/ritonavir treatment. One non-peer-reviewed study describes 10 patients who experienced a relapse of symptoms and SARS-CoV-2 viral load following treatment with nirmatrelvir/ritonavir. The authors identified two instances of transmission during relapse. The second report, also not yet peer-reviewed, describes a single patient who experienced the return of symptoms and had culturable virus 5 days after completing a nirmatrelvir/ritonavir course. Because nirmatrelvir/ritonavir viral resistance was not noted nor did the patient demonstrate an absence of neutralizing antibodies, this raises the question of whether rebound could be a function of drug level in the individual's blood (while the dosage in the antiviral pill is consistent, not everyone absorbs the same amount or metabolizes it the same way). The third and most recent study has been peer-reviewed and describes a series of seven patients who experienced COVID-19 rebound, six of whom had recurrence of symptoms and one who was asymptomatic but had a repeat positive antigen test. While also limited in size, this study supports the findings of the previous two studies: three of the seven patients had culturable virus, suggesting that patients who experience COVID-19 rebound after nirmatrelvir/ritonavir may be infectious. However, there is no defined level of culturable virus that can identify patients who remain infectious/contagious. And fortunately, sequencing of SARS-CoV-2 virus from six of the seven patients did not identify any known antiviral resistant mutations.

Studies That Could Shed Light on Paxlovid Rebound

The EPIC-SR trial enrolled 1,553 individuals between August 2021 and December 2021 with confirmed SARS-CoV-2 who had a standard risk of progressing to severe disease. Pfizer has shared some of these results; however, these press releases have not mentioned COVID-19 rebound.

Another study, the PANORAMIC trial, is currently enrolling high-risk outpatients in the U.K., but like the EPIC-SR trial, it is unclear if rebound is among the study outcomes.

Together, these studies have the potential to both further define and report the incidence of rebound after nirmatrelvir/ritonavir. If these data were collected from the EPIC-SR trial then it should be released. And if the PANORAMIC trial is not already designed to identify cases of rebound, then the protocol should be amended accordingly.

Only an RCT Can Definitively Answer the Question of Paxlovid Rebound

Whether nirmatrelvir/ritonavir is associated with COVID-19 rebound and how often this occurs can only be determined by a prospective double-blind RCT. The basis for the standard of care of hospitalized COVID-19 patients was largely defined by the results of large RCTs that meticulously detailed adverse events. In a similar fashion, an RCT performed with individuals at low risk for progression to severe disease who have been vaccinated or previously infected with SARS-CoV-2 need to be randomized to placebo or nirmatrelvir/ritonavir and then followed closely for rebound by evaluating symptoms, signs, and serial collection of both molecular tests and viral cultures, in conjunction with monitoring of close contacts to determine whether this is transmissible. This will define more precisely the rebound prevalence, clinical consequences, and infectiousness rates. One could also test to see if there is any benefit to treating rebound with an additional course of nirmatrelvir/ritonavir. Despite reports of nirmatrelvir/ritonavir being prescribed to treat rebound, the CDC is not currently recommending this approach.

Without such a scientifically rigorous study, uncertainty will persist at a dire cost. Not knowing who best benefits from nirmatrelvir/ritonavir can result in its overuse, more drug-induced side effects, and the development of viral resistance. Additionally, some individuals who would benefit from nirmatrelvir/ritonavir may be hesitant to be treated because of the perceived risk of COVID-19 rebound. On the other hand, if COVID-19 rebound does occur due to nirmatrelvir/ritonavir, then it is important to know the incidence, clinical, and microbiologic course, and whether additional treatment with nirmatrelvir/ritonavir is warranted.

Once again, the pandemic reminds us to hurry but not rush to judgement, to conduct well-designed clinical trials, and to learn from the results to improve patient care.

Daniel A. Sweeney, MD, is an associate clinical professor of medicine at the University of California San Diego. Cameron R. Wolfe, MBBS, MPH, is an associate professor of medicine at Duke University in Durham, North Carolina. Andre C. Kalil, MD, MPH, is a professor of medicine at the University of Nebraska Medical Center.