Study Design and Procedures
The protocol, protocol amendments, Informed Consent Documents, Investigator Brochure, and other relevant documents (e.g., advertisements) were submitted to an Institutional Review Board / Independent Ethics Committee (IRB/IEC) by the investigator and reviewed and approved by the IRB/IEC before the study was initiated.
This study was conducted in accordance with the protocol and consensus ethical principles derived from international guidelines including the Declaration of Helsinki Council and the Council for International Organizations of Medical Sciences International Ethical Guidelines, applicable International Council for Harmonisation Good Clinical Practice Guidelines, applicable International Organization for Standardization 14155 guidelines, medical device guidelines, and other applicable laws and regulations, including privacy laws.
EPIC-HR is a Phase 2/3, randomized, double-blind, placebo-controlled study conducted in 2,224 symptomatic adult participants with COVID-19 who were non-hospitalized and at increased risk for progressing to severe illness. Eligible participants with confirmed diagnosis of SARS-CoV-2 infection and symptom onset within 5 days were randomly assigned in a 1:1 ratio to receive either nirmatrelvir 300 mg co-administered with ritonavir 100 mg or matched placebo orally every 12 hours for 5 days (10 doses total). Randomization was stratified by geographic region and by receipt or expected receipt (based on investigator opinion) of COVID-19 monoclonal antibodies at the time of randomization. Efficacy and safety assessments were conducted through Day 34 visit; a long-term follow-up period continued through Week 24. Further details of the study design have been previously described1. Participants utilized an electronic diary to capture daily COVID-19 symptoms through Day 28.
Nasopharyngeal (NP) swabs were collected by health care professionals from participants at baseline, Day 1 (D1), Day 3 (D3), Day 5 (D5), Day 10 (D10), and Day 14 (D14). Nasal swabs were self-collected by participants at D3, D10, and D14 if an in-person study visit did not occur (Supplemental Table 1 for collection site details). For analysis, sample collections (e.g., D1, D3, D5, D10, and D14) visits were mapped as baseline (Day -2 to Day 1; within 1 hour post first dose) and post-baseline (>1 hour post first dose). NP/nasal swab samples were shipped on dry ice to the University of Washington Medicine Clinical Virology Laboratory for analysis. Upon receipt, samples were evaluated for quality control/swab deviations and logged into the sample management system. Swabs were placed in Universal Transport Medium and then aliquoted for viral load, viral sequencing, and subsequent infectivity/phenotypic analysis.
Viral Load (VL; RNA) Quantitation
SARS-CoV-2 VL was measured as log10 copies/mL using a validated Abbott Real Time Quantitative SARS-CoV-2 assay at the University of Washington Medicine Clinical Virology Laboratory5. In brief, the VL assay is a reverse transcriptase (RT)-PCR assay intended for the quantitative detection of nucleic acid from SARS-CoV-2 in NP or nasal swabs by detecting the RNA-dependent RNA polymerase (RdRp) and nucleocapsid (N) genes using the Abbott m2000 System. The assay has a dynamic range with an upper limit of quantitation (ULoQ) of 8 log10 copies/mL and the lower limit of quantitation (LLoQ) of 2.0 log10 copies/mL. Data reported as less than 2.0 log10 copies/mL was imputed as 1.70 log10 copies/mL, and data reported as “not detected” was imputed as 0 log10 copies/mL.
SARS-CoV-2 genomes from clinical samples were sequenced using the SWIFT biosciences next generation amplicon based sequencing methodology, and sequencing was performed at the University of Washington Medicine Clinical Virology Laboratory6. Virus from D1, D3, D5, D10, and D14 participant samples that meet the sequencing limit of detection (LoD) (viral concentrations ≥500 copies/mL [≥2.70 log10 copies/mL]) were sequenced and genotyped by mapping against the reference sequence (NCBI: NC_045512.2). In brief, to ensure quality control of sequence data, the genome acceptability criteria were as defined as positive controls reported with 1 million raw reads, >85% identity to the reference sequence (NCBI: NC_045512.2), 750X average genome coverage, 1000X mean Mpro gene coverage, >90% of 3CL at 100X or higher, and <10% Ns in the consensus genome.
Viral Load Rebound (VLR)
The study population for VLR analysis included participants with matched D5 and D10 and/or D14 samples (N=990 for nirmatrelvir/ritonavir and N=980 for placebo). In brief, rebound was classified based on a half log increase in viral load at follow-up (D10 and/or D14) relative to the end of treatment (D5) viral load levels. The lower threshold for rebound was set at 2.7 log10 copies/mL (the LLoQ for the viral sequencing assay). For the present study, participants with half-log increases at follow-up were then split into two categories. For the first category, if only one follow-up sample was available and showed a half-log increase, it was classified as present. If the rebound was present at D10 and levels were still elevated by a half log or more by D14, it was classified as persistent. In brief, the statistical definition for present and/or persistent was: If D14 VL was not available, D10 VL change from D5 (CFD5) ≥0.5 log10 copies/mL and D10 VL ≥2.7 log10 copies/mL, or If D14 VL is available, D14 VL CFD5 ≥0.5 log10 copies/mL and D14 VL ≥2.7 log10 copies/mL. If both follow-up samples were available, but the rebound was only evident at D10, it was classified as transient. The classification for “transient” was defined as follows: If D10 VL was >0.5 log10 copies/mL relative to D5 and D10 VL was >2.7 log10 copies/mL AND if D14 VL was <0.5 log10 copies/mL relative to D5 or D14 VL was <2.7 log10 copies/mL.
Viral Genotyping and Treatment Emergent Mutation Analysis
Virus sequenced from participant samples were genotyped per the University of Washington standard operating procedures. The LLoQ for the sequencing assay was 2.7 log10 copies/mL. Mutations resulting in an amino acid (AA) substitution within a sample were called if the amino acid frequency (AAFREQ) was ≥1% and the amino acid substitution (AASUB) was different from the reference sequence (NCBI: NC_045512.2). Resequencing was conducted if lineage differences were observed across participant samples. VLR variant calls were cross-examined for strand bias. Mutations with significant strand read bias were detected within the present/persistent VLR population and excluded from the TEM analysis. Any distinct mutation was called TEM if the mutation was absent (AAFREQ <5% or not called) for the participant in their baseline sample, but the same mutation was present (AAFREQ ≥5%) for the same participant in any post-baseline sample (D3, D5, D10, or D14). Mutations resulting in frameshift (‘fs’) were excluded from the TEM analysis. Full treatment emergent mutation analysis results are described in a separate treatise.
Lineage classification. For each participant, the SARS-CoV-2 lineage was called based on classification at D1, or if D1 was not available or had insufficient viral concentrations (<2.70 log10 copies/mL), then at D3, if not, then at D5, if not, then at D10 and if not, then at D14.
Mpro Genomic Regions of Interest
Mutations were grouped into SARS-CoV-2 genomic regions of interest including: 1) Mpro residues within 5 Angstroms of nirmatrelvir-Mpro binding and Mpro cleavage regions external to Mpro gene (Supplemental Tables S2 and S3).
Plasma Assay for Nirmatrelvir
Pharmacokinetic samples were to be collected at D1 between 30 to 90 minutes post-dose, and at D5 preferably as pre-dose or any time post-dose. Additionally, one sample was to be collected at any time post-dose during in-person visit at D2, D3, or D4. Nirmatrelvir concentrations were determined by a specific and sensitive bioanalytical method using liquid chromatography with tandem mass spectroscopy (LC-MS/MS). The calibration curve range for the plasma method was 10.0 to 10,000 ng/mL for nirmatrelvir. The assay validations and study sample analyses were conducted at York Bioanalytical Solutions, York, UK in compliance with the current US FDA and European Medicines Agency Guidance requirements.
Plasma samples were collected for serological analysis. Roche Elecsys® Anti-SARS-CoV-2 (against N – qualitative) and Elecsys® Anti-SARS-CoV-2 “S” (both quantitative and qualitative) immunoassays were used for the detection of total host immunoglobulins to SARS-CoV-2 N or spike (S) proteins. The immunoassays are set to a sandwich format. Assays were validated following CRO SOPs (PPD) and run per manufacturer’s instructions. Serostatus at baseline was defined as positive in either N or S qualitative assay.