HIV and ID Observations

Filippo Balbi, Testa Anatomica, 1854.

Over at Clinical Infectious Diseases, researchers from the Mayo Clinic published a retrospective analysis of nirmatrelvir/r (Paxlovid) treatment, with a careful review of each patient’s chart.

The goal was to determine the clinical outcomes after the 5-day treatment course, with a focus on the frequency of rebounds — a topic of great clinical interest but with little real-world data with a decent-sized denominator.

The study included 483 patients with at least some risk factors for severe COVID-19 based on the institution’s Monoclonal Antibody Screening Score. Adjudication of eligibility was done centrally through the hospital’s COVID-19 outpatient treatment program. The mean age was 63, and 93% were vaccinated. Patients were given the option of telemedicine follow-up and encouraged to self-report symptoms or other problems; this information and their full electronic medical records were reviewed after the treatment.

Out of this group, 4 (0.8%) clearly experienced clinical rebound — these cases are described. None of the 4 received re-treatment or were hospitalized. Two other patients (neither of them rebounders) did require hospitalization, but the authors note these were unrelated to COVID-19. There were no deaths.

We can be reassured by the information on the low rate of severe outcomes, an outcome mirrored in low hospitalization rates nationally despite lots of ongoing cases. Rates of hospitalization for each COVID-19 case are now extremely low, especially among those vaccinated, and observational studies (peer-reviewed and published and unpublished) suggest that Paxlovid may reduce this risk even further. A large retrospective study from Kaiser Permanente Southern California involving 5,287 patients found similarly  low hospitalization rates.

(Note that these observational studies include patient populations treated with Paxlovid with established risk factors for COVID-19 adverse outcomes — that’s how it’s mostly prescribed under the Emergency Use Authorization (EUA). By contrast, those enrolled in the EPIC-SR study, which was recently stopped due to futility, were “standard risk”. Critically important will be information on baseline characteristics in EPIC-SR.)

For the Mayo study, the authors are to be credited for collecting and reporting the data so quickly, especially at a time when clinicians and patients increasingly observe the rebound phenomenon and wonder what to do about it. The centralized program used for distributing COVID-19 treatments is an additional strength of this report, as it allowed participants the opportunity to self-report progression of symptoms, and upfront ensured that all who were treated met the criteria for high risk based on the EUA.

The big limitation of this Mayo study, however, is that it’s retrospective. As a result, I strongly suspect the 0.8% rate of rebound substantially underestimates the true incidence. Pfizer reported rebounds in 2% of study participants in the prospective EPIC-HR trial, a rate more than twice as high. Based on anecdotal experience — there’s lots of Paxlovid treatment out there! — I would not be surprised if it’s at least 10 times this high, or in the 5-10% range.

What still remains unknown, frustratingly, is whether treatment with subsequent rebound reduces — or paradoxically increases — the total amount of time that people are contagious, or, if the total time is the same, does it just delay the time that a person can confidently say that they’re in the clear?

Why is this important? I’ve had people tell me that they don’t want to take the drug because they’ve heard it might lengthen the time they’re contagious. And I’ve had other patients who experienced rebound express regret on having taken the drug — even though we have no idea whether they would have tested positive for a prolonged period even without the treatment.

Let’s hope further data from prospective studies give us some answers — in particular, the placebo-controlled trials, either Pfizer’s two studies or the large PANORAMIC trial done in the United Kingdom.

In the meantime, I’m still advising people at higher risk for COVID-19 adverse outcomes — even if vaccinated — to be treated with Paxlovid, uncertainties about the above issues notwithstanding.

Unless you’ve been hiding under a rock, you’ve heard that some people treated for COVID-19 with nirmatrelvir/ritonavir (Paxlovid) experience a relapse in illness shortly after stopping treatment.

It’s both a recurrence of symptoms and a positive antigen test — sometimes after the test became negative. One case report published as a pre-print shows that a relapse can have a very low cycle threshold, meaning a high viral load:

Relapses vary in severity, from very mild and brief to worse than the initial illness. Check out the comments in my previous post for a sampling.

Some posit that the immune system doesn’t get a chance to mount a full response since the drug quickly reduces exposure to viral antigens. Others note that certain individuals experience prolonged viral replication, symptoms, and antigen test positivity, and perhaps it’s this group that’s destined to rebound after the 5 days of Paxlovid use. The drug knocks down replication for a while, but not long enough, so it rebounds.

Maybe it’s both — these aren’t mutually exclusive hypotheses. Regardless, as noted in my prior post, there’s still plenty we don’t know. But given the rapidly increasing use of Paxlovid nationally, I thought it worth this brief update about two important things we’ve learned since then, plus my opinion on a commonly asked question about contagiousness.

First, Pfizer offered additional details from their EPIC-HR study, citing that late viral rebounds occurred in roughly the same proportion of treated and untreated participants in their trial — around 2%:

UPDATE: Pfizer gave me the figures regarding the proportion of people who had Covid rebound in the Paxlovid clinical trial: About 2% of participants who got the drug experienced rebound, compared with ~1.5% of those who received the placebo rebounding within the same time frame.

— Benjamin Ryan (@benryanwriter) April 28, 2022

Whether the rate of relapse is higher than this 2% in clinical practice is unknown. Certainly, it seems that way — it even happened to a famous virologist! — but without knowing the denominator of total treated patients, we can only speculate.

And remember, clinicians have noted biphasic illness patterns since the start of the pandemic, and continue to do so. We’ve attributed that second period of worsening mostly to a robust (and sometimes hyperactive) immune response, but potentially there could be a component of concomitant heightened viral replication.

Second, in communications with clinicians, the FDA and Pfizer have made it clear that the people who relapse are in fact eligible for re-treatment under the Emergency Use Authorization (EUA). In other words, the within-5-day symptom clock starts over with the relapse. This would be justified clinically for our highest-risk patients (severely immunocompromised, medically fragile, or with severe recurrent symptoms), and favored over other outpatient treatments (all of which have logistical or efficacy issues) until we know more. So far, viral resistance has not been identified.

So where does that leave us with regard to infection control issues? In my opinion, we should assume that people who experience a symptomatic relapse with a positive antigen test could transmit the virus to others. Until data prove otherwise — meaning studies show these late relapsers have defective virions that are not replication-competent — this should be the recommended approach for work, school, and socializing. Symptoms plus positive antigen test equals potentially contagious.

In the meantime, I completely agree with this editorialist that we should be doing more to monitor how Paxlovid is doing since the EUA launched.  I’m hoping large healthcare systems can quickly assemble observational studies evaluating how often this relapse phenomenon occurs out there in what is often called “the real world” — which means, in this context, “not in a clinical trial”. Prospective studies collecting virologic and immunologic data would also provide great value, especially given the theoretical concern that treatment blunts a favorable immune response.

So let’s keep learning.

You know, baby steps.