Plasma concentrations of nirmatrelvir and molnupiravir required for inhibition of SARS-CoV-2 replication differ between rhesus macaques and humans.

Early during the COVID-19 pandemic, non-human primate (NHP) infection models emerged as highly useful tools for preclinical screening of antiviral drugs. However, it is uncertain whether NHP models can be used to precisely inform optimal dosing in humans. We previously established and validated mathematical models, which were fit to SARS-CoV-2 viral loads from human clinical trials. These models identified that plasma drug concentrations required to inhibit viral replication by 50% in humans (in vivo EC50) differ substantially from in vitro EC50 estimates in cell culture systems. Here, we apply models to sequential viral load data from SARS-CoV-2 infected rhesus macaques (RMs) that were untreated or treated with nirmatrelvir/ritonavir, molnupiravir, or both drugs. We identify that equivalent plasma drug concentrations correspond to greater antiviral potency in the lungs compared with nasal passages for nirmatrelvir and molnupiravir. Average nirmatrelvir antiviral efficacy in RM (30% in nasal passages and 46% in the lungs) was estimated to be less than in humans (82%) due to shorter plasma drug half-life. Molnupiravir efficacy in RMs (95% in nasal and 99% in the lungs) is estimated to be similar to efficacy in humans against omicron variants. Our model estimates that 10-fold higher plasma nirmatelvir concentrations are needed in humans versus RMs to achieve 50% reduction in viral replication, whereas 20-fold lower plasma molnupiravir concentrations are needed. Our results suggest that dose optimization in humans based on modeling of NHP viral loads is limited by drug-specific differences in pharmacokinetic, pharmacodynamic, and virologic profiles, and that data from human phase 1 and 2 trials are better suited for this task.