Does Paxlovid Cause COVID-19 Rebound? What Two Placebo-Controlled Trials Showed

This page summarizes an analysis by Food and Drug Administration (FDA) scientists, published in CDC's MMWR, of trials run in 2021–2022.

Nirmatrelvir/ritonavir (Paxlovid) is an oral COVID-19 treatment. Nirmatrelvir blocks the SARS-CoV-2 main protease (Mpro, also called 3CLpro), and ritonavir is a pharmacokinetic enhancer. It was first made available under FDA Emergency Use Authorization on December 22, 2021, and approved on May 25, 2023, for mild-to-moderate COVID-19 in adults at high risk of severe disease, including hospitalization or death.

Some patients have had a rebound, SARS-CoV-2 detected again or COVID-19 symptoms returning, after taking it, and others after no antiviral at all. Most reports came from single cases or nonrandomized studies, so whether the drug itself causes rebound was unclear. Concern about it has reportedly led to less use of the drug.

The trials

FDA scientists reanalyzed data the drug's sponsor, Pfizer, had submitted from two phase 2/3, randomized, double-blind, placebo-controlled trials. In both, adult outpatients with mild-to-moderate COVID-19 took nirmatrelvir 300 mg with ritonavir 100 mg, or placebo, twice daily for 5 days.

Health care providers took nasopharyngeal swabs on days 1, 3, 5 (end of treatment), 10 and 14. Rebound meant viral RNA rising from day 5 to day 10 or day 14. A level of at least 5 log10 copies/mL was treated as possibly signaling infectious virus.

Findings

Rebound rates were similar with drug and placebo.

The first comparison was significant, but it did not account for how much the virus fell during treatment; once that was accounted for, the difference narrowed and was no longer significant. Across EPIC-HR and both periods of EPIC-SR, rebound among people whose virus had responded by day 5 ranged from 6.4% to 8.4% with the drug and 5.9% to 6.5% with placebo. In EPIC-SR, neither approach found a significant difference in either period, and a stricter definition requiring at least 5 log10 copies/mL also gave similar rates.

Rebound also happened during treatment. In EPIC-HR, viral RNA often rose between days 3 and 5, while people were still taking the drug, at rates numerically higher than after treatment.

Rebound was not linked to severe outcomes. Through day 28:

• in EPIC-HR, 1 of 77 drug recipients (1.3%) and 3 of 53 placebo recipients (5.7%) with rebound were hospitalized for COVID-19, with no deaths, in line with the trial's overall hospitalization rates;
• in EPIC-SR, three people with rebound were hospitalized, one on the drug and two on placebo, all in the 2021 period;
• viral RNA levels showed no consistent timing relationship between rebound and hospitalization.

Other findings.

• Rebound was not linked to immunosuppression, but EPIC-HR had only 13 immunosuppressed participants: six on the drug (none rebounded) and seven on placebo (one rebounded).
• Of 59 drug recipients with rebound whose virus was sequenced, two (3%) had a treatment-emergent substitution in Mpro associated with nirmatrelvir resistance on day 10. Neither was immunosuppressed or hospitalized for COVID-19.
• At every visit after baseline, a similar or higher share of drug recipients had viral RNA below the limit of quantification, and a similar or lower share had 5 log10 copies/mL or more. The drug did not delay viral clearance overall.

What it means

Across roughly 3,000 trial participants, rebound was not consistently linked to the drug. The authors note it might be somewhat more common with treatment, since several comparisons were modestly, though not significantly, higher, but rebound also occurred with placebo at generally similar rates.

Because rebound was seen during treatment too, it cannot simply be blamed on the drug wearing off. At least some rebound likely reflects natural swings in virus production, shedding tied to the patient's own body, or variation in how swabs are taken, which would also explain rebound on placebo. Genomic databases should keep watching for nirmatrelvir-resistant variants.

The findings support FDA's conclusion that the drug is safe and effective for eligible patients at high risk of severe COVID-19.

Limitations

• Rebound rates depend heavily on definitions and sampling schedules. This analysis counted rebound at a single post-treatment visit, but events between visits or after day 14 could be missed.
• Only 13 EPIC-HR participants were immunosuppressed, a group in whom rebound might matter more.
• High-risk people infected with Omicron or later sublineages were not studied, because the drug was already available under Emergency Use Authorization by then.
• The data allowed a look at viral levels, hospitalization and death, but not a detailed study of milder symptoms.

Sources

• Harrington PR, Cong J, Troy SB, et al. "Evaluation of SARS-CoV-2 RNA Rebound After Nirmatrelvir/Ritonavir Treatment in Randomized, Double-Blind, Placebo-Controlled Trials — United States and International Sites, 2021–2022." MMWR 72(51). https://www.cdc.gov/mmwr/volumes/72/wr/mm7251a2.htm
• FDA Integrated Review of New Drug Application 217188 (Paxlovid): https://www.accessdata.fda.gov/drugsatfda_docs/nda/2023/217188Orig1s000IntegratedR.pdf