by Gertrud U. Rey
The antiviral drug Paxlovid is highly effective at inhibiting SARS-CoV-2 replication and reducing symptoms associated with COVID-19. Nevertheless, there have recently been numerous reports of recurrence of positive tests and symptoms after completing treatment with Paxlovid, leading some to infer that the drug triggers the recurrence. Is this inference actually correct, or would the recurrences happen regardless of treatment? In other words, is “Paxlovid rebound” really just COVID rebound?
Most studies aiming to address this question have been retrospective analyses, which use existing data collected from events that have already happened. A major disadvantage of examining data retrospectively is that it is impossible to randomly assign participants to experimental or control groups, or to even apply the proper controls as is typically done in a prospective study. These drawbacks often lead to a biased selection of participants such that they do not always represent the population that is intended to be analyzed, which leads to inaccurate results and false conclusions.
In an attempt to remedy this shortcoming, a group of investigators led by Michael Mina carried out a prospective study in which they compared the outcomes between two groups of COVID-19 patients: a group of 127 subjects who chose to be treated with Paxlovid and a control group of 43 subjects who chose not to be treated. The aim of the study was to determine whether Paxlovid recipients experience a higher incidence of rebounds than non-treated individuals.
To qualify for the study, all participants had to test positive for SARS-CoV-2 using a rapid antigen test. The day of the first test was then documented as day 0 and the participants continued testing themselves and recording their symptoms on days 2, 5, 7, 9, 11, 13, 15, and 17 of the study period. Any positive antigen test after a negative test within the 17-day period was defined as a viral rebound, and any recurrence of symptoms after initial symptom clearance within the same period was defined as a symptom rebound. At the 17-day time point, among the Paxlovid group, 14% of subjects had experienced a viral rebound and 19% had experienced a symptom rebound. In contrast, only 9% of subjects in the (untreated) control group had a viral rebound and only 7% had a symptom rebound. There were no noteworthy differences in the number of rebounds between the two groups at the one-month time point. Although the incidence of rebound was slightly higher in the Paxlovid group, this difference between the two groups was not statistically significant; it was likely due to random chance and the small sample sizes of the groups. In other words, the slightly higher incidence of viral and symptom rebounds in the Paxlovid group has no clinical meaning, and one can interpret the rate of rebounds between the Paxlovid and control groups to be similar, meaning that Paxlovid probably does not cause viral and/or symptom rebounds.
The authors thoughtfully note that the study has several limitations. First, the overall sample size of 170 participants is small and there was a large difference between the sizes of the two groups (i.e., 127 subjects in the Paxlovid group and 43 subjects in the control group). Large and balanced sample sizes are critical for reducing the margin of error and for obtaining results that are both accurate and clinically useful. Second, the participants tested themselves, which could have introduced unknown errors such as whether the tests were carried out properly or at the correct time. Third, participants were asked to only test every other day to ensure compliance; however, daily testing would have provided additional data points and more comprehensive findings. Larger surveys done under more controlled and standardized conditions are needed to validate the results obtained in this study.
In contrast to popular opinion, rebounds can happen after most viral infections, so there is nothing unique about SARS-CoV-2 in this regard. Even if Paxlovid does cause viral and/or symptom recurrence in a small subset of people, a preponderance of the evidence indicates that early treatment with Paxlovid results in an overwhelming reduction in hospitalization and death for COVID-19 patients. Understanding the underlying mechanisms leading to rebounds can help guide practitioners to modify timing and length of treatment with Paxlovid or other antiviral drugs to reduce the incidence of rebound.
[The material in this blog post is also covered in Catch This Episode 41.]
Trudy, I think that your discussion covers the question “In other words, is “Paxlovid rebound” really just COVID rebound?” quite well.
Especially concerning the less-than-the-best Mina study that was at least a serious attempt, yet not sufficient for a proof beyond a reasonable doubt standard.
I am not fond of our current state of affairs within virology … at the nucleotide collection level, but much less so at the nomenclature level (On The Origin of The Containment Entity – 2).
This is a bit vague: when saying there was only a ‘slight’ difference in rebound between the two groups, it presumably is referring to the end of month position rather than the 17 day, which seems to show around twice as much recurrence with Paxlovid than without: not a ‘slight’ difference.
However: no indication is given of the relative severity of symptoms, and the apparent difference is only there because they are not comparing similar processes: this sort of test would be good for comparing two different antivirals, but not antiviral plus virus with virus alone.
If I have absorbed discussion on TWiV reasonably, there is actually no rebound in either group: the untreated group simply experiences a gradually diminishing set of viral symptoms as their immune system gears up to fight and learns how to beat the new invader; the treated group suffers less severe symptoms at the outset as the antiviral keeps viral numbers down while the immune system is finding and producing the right antibodies to control the infection with; after antiviral treatment ends there is more work for the immune system itself to take on, and so ‘rebound’ symptoms are experienced, and then tail off the same way they did in the untreated group. The antiviral just gives the immune system time to get going and reduces the chance of it being overwhelmed by the initial exponential virus production.
Seems to me the sequence of events known as ‘rebound’ are exactly as would have been predicted. The hospitalised/non-hospitalised figures are the only ones that count in comparing treatment and non-treatment.
Another point to think of also, is that, in the UK, it is hard to imagine how anyone could get Paxlovid within the time it takes to realise one is ill, get a test, and get through to their GP (here we are only allowed to use our one registered, very oversubscribed, doctor’s practice) to try and make an appointment, when, typically, all the slots for a month are filled within the first day booking starts!
For those who can afford it, there are some advantages to the US competitive health system, but at least here we can all get help eventually without bankrupting ourselves. :/