Abstract
Background: An in silico screen was performed to identify FDA approved drugs that inhibit SARS-CoV-2 main protease (M(pro)), followed by in vitro viral replication assays, and in vivo pharmacokinetic studies in mice. These studies identified atovaquone as a promising candidate for inhibiting viral replication. Methods: A 2-center, randomized, double-blind, placebo-controlled trial was performed among patients hospitalized with COVID-19 infection. Enrolled patients were randomized 2:1 to atovaquone 1500 mg BID versus matched placebo. Patients received standard of care treatment including remdesivir, dexamethasone, or convalescent plasma as deemed necessary by the treating team. Saliva was collected at baseline and twice per day for up to 10 days for RNA extraction for SARS-CoV-2 viral load measurement by quantitative reverse-transcriptase PCR. The primary outcome was the between group difference in log-transformed viral load (copies/mL) using a generalized linear mixed-effect models of repeated measures from all samples. Results: Of the 61 patients enrolled; 41 received atovaquone and 19 received placebo. Overall, the population was predominately male (63%) and Hispanic (70%), with a mean age of 51 years, enrolled a mean of 5 days from symptom onset. The log(10) viral load was 5.25 copies/mL vs. 4.79 copies/mL at baseline in the atovaquone vs. placebo group. Change in viral load did not differ over time between the atovaquone plus standard of care arm versus the placebo plus standard of care arm. Pharmacokinetic (PK) studies of atovaquone plasma concentration demonstrated a wide variation in atovaquone levels, with an inverse correlation between BMI and atovaquone levels, (Rho -0.45, p = 0.02). In post hoc analysis, an inverse correlation was observed between atovaquone levels and viral load (Rho -0.54, p = 0.005). Conclusion: In this prospective, randomized, placebo-controlled trial, atovaquone did not demonstrate evidence of enhanced SARS-CoV-2 viral clearance compared with placebo. However, based on the observed inverse correlation between atovaquone levels and viral load, additional PK-guided studies may be warranted to examine the antiviral effect of atovaquone in COVID-19 patients.