SARS-CoV-2 virologic rebound with nirmatrelvir-ritonavir therapy

Abstract

Objective: To compare the frequency of replication-competent virologic rebound with and without nirmatrelvir-ritonavir treatment for acute COVID-19. Secondary aims were to estimate the validity of symptoms to detect rebound and the incidence of emergent nirmatrelvir-resistance mutations after rebound.

Design: Observational cohort study.

Setting: Multicenter healthcare system in Boston, Massachusetts.

Participants: We enrolled ambulatory adults with a positive COVID-19 test and/or a prescription for nirmatrelvir-ritonavir.

Exposures: Receipt of 5 days of nirmatrelvir-ritonavir treatment versus no COVID-19 therapy.

Main outcome and measures: The primary outcome was COVID-19 virologic rebound, defined as either (1) a positive SARS-CoV-2 viral culture following a prior negative culture or (2) two consecutive viral loads ≥4.0 log₁₀ copies/milliliter after a prior reduction in viral load to <4.0 log₁₀ copies/milliliter.

Results: Compared with untreated individuals (n=55), those taking nirmatrelvir-ritonavir (n=72) were older, received more COVID-19 vaccinations, and were more commonly immunosuppressed. Fifteen individuals (20.8%) taking nirmatrelvir-ritonavir experienced virologic rebound versus one (1.8%) of the untreated (absolute difference 19.0% [95%CI 9.0-29.0%], P=0.001). In multivariable models, only N-R was associated with VR (AOR 10.02, 95%CI 1.13-88.74). VR occurred more commonly among those with earlier nirmatrelvir-ritonavir initiation (29.0%, 16.7% and 0% when initiated days 0, 1, and ≥2 after diagnosis, respectively, P=0.089). Among participants on N-R, those experiencing rebound had prolonged shedding of replication-competent virus compared to those that did not rebound (median: 14 vs 3 days). Only 8/16 with virologic rebound reported worsening symptoms (50%, 95%CI 25%-75%); 2 were completely asymptomatic. We detected no post-rebound nirmatrelvir-resistance mutations in the NSP5 protease gene.

Conclusions and relevance: Virologic rebound occurred in approximately one in five people taking nirmatrelvir-ritonavir and often occurred without worsening symptoms. Because it is associated with replication-competent viral shedding, close monitoring and potential isolation of those who rebound should be considered.

Figure 1.

Virologic decay curves with semiquantitative viral cultures and quantitative viral load among individuals with acute COVID-19 taking no therapy or nirmatrelvir-ritonavir (N-R). Black lines indicate individuals without rebound, whereas blue lines indicate individuals with virologic rebound. Panels A (viral load) and B (viral culture) depict decay curves for those not receiving therapy. Panels C (viral load) and D (viral culture) depict individuals who received N-R. Panels E and F compare our primary outcome with all available time points (E) or restricted to days 5, 10 and 14 only (Panel F) as defined in prior studies [1]. Using only three timepoints to detect rebound resulted in missing 81% of the observed virologic rebound events of replication-competent virus.

Figure 2.

Comparative frequency of virologic rebound by nirmatrelvir-ritonavir use, stratified by demographics and clinical characteristics (A), and by number of days between the first positive SARS-CoV-2 test and initiation of nirmatrelvir-ritonavir therapy (B). For the sub-group comparisons, the bottom P-values represent Fisher’s exact tests comparing rebound rates between those taking versus those not taking nirmatrelvir-ritonavir. The upper P-values represent Fisher’s exact tests comparing rebound rates among those taking nirmatrelvir-ritonavir across the sub-groups, for example comparing those taking nirmatrelvir-ritonavir with immunosuppression present versus those taking nirmatrelvir-ritonavir with immunosuppression absent.

Figure 3.

Kaplan Meier survival curves demonstrating time from initial positive SARS-CoV-2 test until initial negative viral culture (A-C) and final negative culture (D-F). In Panel A, we demonstrate that there is a faster time to first negative culture in those receiving nirmatrelvir-ritonavir (N-R) versus no therapy (No Rx). In Panels B and C, we find similar patterns in time to initial negative culture, when dividing the N-R group into those who rebounded (B) and those who did not (C). However, as shown in Panel D, there is no difference in time to final negative culture between N-R and No Rx groups. This appears to be due to the prolonged time to final negative culture among N-R users who rebound (Panel E), because the time to final negative culture remains shorter in N-R users who did not rebound compared to the No Rx group (Panel F).