This article is part of a larger series on the risks of molnupiravir. For a summary of the risks at the level of the individual, see here, and for an overview of the risks on a societal level, see here

Mutations to the SARS-CoV-2 genome, especially in the spike protein, help the virus elude vaccines, natural immunity, and monoclonal antibody treatments. Such mutations can also help improve viral fitness, including quicker viral replication, shorter incubation periods, and increased immunosuppressive capabilities. Together, these issues pose a serious challenge to pandemic control. One potential area of hope are antiviral drugs. Unfortunately, none of our current options are effective in preventing infection. In fact, one of the more widely-used drugs against Covid-19 —molnupiravir— may supercharge viral mutation and spawn new variants. When I first voiced this concern, it was sidelined as largely hypothetical. Now, two novel preprints suggest the risk is all too real, and the concern justified. 

Following the Genetic Breadcrumbs 

Molnupiravir works by inserting itself into the viral ribonucleic acid (RNA) and, once there, introducing a series of errors into the genetic code. Over time, these small mistakes build up, culminating in viral particles that cannot efficiently replicate due to the number of errors in their genome. The modus operandi, then, can be described as death by mutation. But the worry is that the mutation portion of the process might occur without the subsequent neutralization, leaving us with exotic and unpredictable variants. Theoretically, this could happen if someone breaks off their molnupiravir regimen prematurely, or in immunocompromised patients who, despite treatment, do not fully clear the infection.

The errors that molnupiravir introduces are not entirely random. Instead, it leaves behind a certain mutational signature — think of this as fingerprint. Nucleic acids like RNA and DNA are composed of basic building blocks called nucleotides. Nucleotides can themselves be broken down into smaller components, including parts called nitrogenous bases. There are four different nitrogenous bases in RNA: adenine (A), cytosine (C), guanine (G) and uracil (U). Molnupiravir is associated with mutations that change guanine to adenine and mutations that change cytosine to uracil (Figure 1, left). It is also associated, albeit to a lesser extent, with the reverse bait-and-switch, where adenine is changed to guanine and uracil to cytosine (Figure 1, right)

FIGURE 1. In the most common scenario, shown on the left-hand side, Molnupiravir (M) is incorporated opposite guanine (G) nucleotides. It can then pair to adenine (A) in subsequent replication, creating a G-to-A mutation. If the original guanine resulted from negative strand synthesis from a coding-strand cytosine (C), then the G-to-A change ultimately creates a C-to-U coding change (Scenario 1). In the second, less common, scenario shown at the right hand side molnupiravir is initially incorporated by pairing with adenine, which can result in an A-to-G mutation or, if the original A came from a coding uracil (U), a U-to-C mutation. SOURCE: “Identification of a molnupiravir-associated mutational signature in SARS-CoV-2 sequencing databases” SANDERSON ET AL. 2023

Following the genetic breadcrumbs, Sanderson et al. scoured global SARS-CoV-2 sequencing databases to check for viral lineages containing molnupiravir’s mutational signature. They analyzed a phylogenetic tree based on more than 13 million SARS-CoV-2 sequences submitted to the Global Initiative on Sharing Avian Influenza Data (GISAID) and the International Nucleotide Sequence Database Collaboration (INSDC). For each branch of the tree, they plotted out the number of each substitution class (A-to-U, A-to-G, etc.). Filtering branches by the ratio of their substitutions, they located some that had unusually high proportions of the G-to-A substitution so characteristic of molnupiravir. These same branches also happened to have high proportions of C-to-U substitutions, again associated with molnupiravir. 

Almost all of these samples originated in 2022, with barely any such branches cropping up before then. For reference, molnupiravir became widely available in November and December of 2021. The researchers also note that the sequences with high G-to-A ratios mostly come from countries that use molnupiravir: Australia (more than 380,000 prescriptions by the end of 2022), the United Kingdom (more than 30,000 prescriptions), and the United States (more than 240,000 in the early months of 2022). In contrast, France and Canada, which both have a strong sequencing infrastructure but have not authorized molnupiravir for use, had only a low number of sequences with high G-to-A ratios. Another piece of evidence comes in the form of age data; the sequences with unusually high G-to-A ratios were primarily derived from older patients, the same demographic that would be most likely to receive molnupiravir in the first place. 

Worryingly, Sanderson et al. recorded at least three separate clusters —one in Australia and two in the United Kingdom— that displayed onwards transmission of the molupiravir-associated lineages. In Australia, the cluster included up to 20 people infected over a span of one month, 

Molnupiravir and Immunocompromised Patients

Despite being convincing, the above evidence remains circumstantial; it is not quite a smoking gun. But in combination with the findings of a second study, it grows in strength. 

Fountain-Jones et al. closely monitored nine immunocompromised patients infected with SARS-CoV-2. Five of the patients received treatment with molnupiravir, the remaining four did not. As early as 10 days after treatment, the SARS-CoV-2 genomes of patients had accumulated on average 30 new mutations. In contrast, no such viral variation was observed in patients who did not receive molnupiravir.  

Many of the mutations that the researchers witnessed were located within the spike protein, which the virus uses to bind and enter host cells. It is also the target of our current vaccines. And although most mutations were transitory —present in one sample, but gone the next— some of them were fixed, showing up swab after swab. Mutations that remain steady over time are generally indicative of some kind of advantage to viral fitness. 

In one of the patients, Fountain-Jones et al. recorded 10 different fixed mutations which, taken together, accounted for a third of the mutations seen in the Omicron variant. 

Of note, all of the patients remained SARS-CoV-2-positive following treatment, presenting a clear opportunity for onward transmission of new variants in the hospital and beyond. 

Not Just Dangerous, But Ineffective Too

Initial efficacy data for molnupiravir suggested that it reduced the risk of hospitalization or death by 50% in patients with mild-to-moderate COVID-19. By the time the final clinical trial results were released, this number had plummeted down to 30%. Although far from optimal, there were few antiviral treatment options available at that point in the pandemic — even now, effective anti-Covid-19 drugs remain scarce. After a hesitant endorsement by the Antimicrobial Drugs Advisory Committee —by the narrow margin of 13 votes against 10— molnupiravir was authorized for emergency use by the United States Food and Drug Administration (FDA) in December of 2021. 

Since then, additional trials on molnupiravir efficacy have been conducted. One of the most recent, funded by the UK National Institute for Health and Care Research (NIHR), enrolled more than 25,000 participants. All participants were over the age of 50 or, if younger, had at least one comorbidity that placed them under increased risk of severe Covid-19. Unlike the Merck and Ridgeback trials, the majority of participants were vaccinated. 

Participants were randomly picked to receive either molnupiravir plus usual care, or usual care alone. They then submitted reports over a period of 28 days, updating the researchers on their status through a daily online diary. 

The data speak for themselves: among high-risk, vaccinated adults, Molnupiravir did not reduce the frequency of Covid-19-associated hospitalizations or death. Whether or not you received molnupiravir, your odds of ending up in the hospital were essentially the same. 


To be clear, highly-active combinatorial drugs can help us contain, and maybe even eliminate, SARS-CoV-2. This is not in question. What is in question is whether our current options are up to par. The aforementioned research suggests that at least one of them, molnupiravir, falls short. The safety concerns raised by these findings should be taken seriously both by the manufacturers and the United States Food and Drug Administration (FDA).