Characterisation of SARS-CoV-2 genomic variation in response to molnupiravir treatment in the AGILE Phase IIa clinical trial

Abstract

Molnupiravir is an antiviral, currently approved by the UK Medicines and Healthcare products Regulatory Agency (MHRA) for treating at-risk COVID-19 patients, that induces lethal error catastrophe in SARS-CoV-2. How this drug-induced mechanism of action might impact the emergence of resistance mutations is unclear. To investigate this, we used samples from the AGILE Candidate Specific Trial (CST)-2 (clinical trial number NCT04746183). The primary outcomes of AGILE CST-2 were to measure the drug safety and antiviral efficacy of molnupiravir in humans (180 participants randomised 1:1 with placebo). Here, we describe the pre-specified exploratory virological endpoint of CST-2, which was to determine the possible genomic changes in SARS-CoV-2 induced by molnupiravir treatment. We use high-throughput amplicon sequencing and minor variant analysis to characterise viral genomics in each participant whose longitudinal samples (days 1, 3 and 5 post-randomisation) pass the viral genomic quality criteria (n = 59 for molnupiravir and n = 65 for placebo). Over the course of treatment, no specific mutations were associated with molnupiravir treatment. We find that molnupiravir significantly increased the transition:transversion mutation ratio in SARS-CoV-2, consistent with the model of lethal error catastrophe. This study highlights the utility of examining intra-host virus populations to strengthen the prediction, and surveillance, of potential treatment-emergent adaptations.

Fig. 1. Protocol overview and the detection of the molecular signatures of molnupiravir mechanism of action.

a (i) A simplified AGILE CST-2 Phase IIa trial protocol. Molnupiravir was administered to outpatients as four oral pills (200 mg each, 800 mg total) every 12 hours for five days. Participants were randomised placebo to drug 1:1, with nasopharyngeal swabs taken for viral load monitoring. (ii) Sequencing protocol. RNA extracted from nasopharyngeal swabs, taken at days 1, 3 and 5 post treatment initiation, was used for amplicon library preparation using the EasySeq™ RC-PCR SARS-CoV-2 WGS kit (Nimagen, Netherlands). Resulting sequence reads were mapped to the Wuhan-Hu-1 reference (NC_045512.2). b Molnupiravir mechanism of action via the RNA template leads to the accumulation of transition mutations in viral progeny. c Average Ts/Tv ratio values per RNA sample from all participants (placebo n = 65, green; molnupiravir n = 59, purple). SARS-CoV-2 RNA from molnupiravir (purple) participants show a statistically significant accumulation of transition mutations over time compared to placebo (green). d the same genomic data as in c separated into the individual base changes that contribute to the transition mutation counts. A two-sided Wilcoxon rank sum test was performed in c and d; ****P ≤ 0.0001, ***P ≤ 0.001, **P ≤ 0.01, ns = P > 0.05 (Bonferroni adjusted). The boxplots indicate the median, interquartile range, and the minimum and maximum values (excluding outliers). Exact p values are reported in the Source Data. RC-PCR reverse complement-polymerase chain reaction, WGS whole genome sequencing, GCPLab good clinical practice laboratory (University of Liverpool). Fig. a and b were created with Biorender.com.

Fig. 2. Predicted amino acid variations derived from SARS-CoV-2 RNA in the whole genome, NSP12, NSP14 and Spike sequences.

a Predicted amino acid variation derived from RNA sequence information across the whole genome in all Delta infected participants (n = 52). Each sample is assigned a predicted “Top” (green), “2nd” (blue) and “3rd” (dark purple) amino acid (AA) based on proportion of reads at every genome position. Minimum read depth = 200. Minor genomic variants (>0.1 and <0.5; grey dashed lines) increase in frequency over time, with viral RNA from molnupiravir treated participants showing more diversity. b NSP12 showed very little minor genomic variation over the five days. c NSP14 also showed minor genomic stability, but had sites of low-level minor variation at codon positions 18,634 and 18,643 (indicated as amino acids 199 and 202 with black arrows) that were present in all samples tested and may represent a persistent sub-population. d Spike had two sites with an amino acid mixed population at codon positions 21,617 and 21,845 (indicated as amino acids 19 and 95 with red arrows) in all Delta samples analysed. These are known VOC sites in all the Delta sub-lineages.

Fig. 3. Predicted amino acid variations derived from SARS-CoV-2 RNA in the whole genome of B.1.1.7/Alpha, B.1.177/EU1 and BA.1/Omicron lineages.

Predicted amino acid variation derived from RNA sequence information across the whole genome in all a, alpha (placebo n = 14, molnupiravir n = 11); b B.1.177/EU1 (placebo n = 10, molnupiravir n = 8); and c BA.1/Omicron (placebo n = 9, molnupiravir n = 11) infected participants. Each sample is assigned a predicted “Top” (green), “2nd” (blue) and “3rd” (dark purple) amino acid (AA) based on proportion of reads at every genome position. Minimum read depth = 200. Minor genomic variants (>0.1 and <0.5; grey dashed lines) increase in frequency over time, with viral RNA from molnupiravir treated participants showing more diversity.