Persistent SARS-CoV-2 Infections in the Community Show Fast Within-Host Change but Little Late Transmission

Prof Katrina Lythgoe

Image credits: University of Oxford

Mahan Ghafari, DPhil

Image credits: University of Oxford

In a UK population-based surveillance study, investigators identified 576 persistent SARS-CoV-2 infections. Longer infection duration correlated with higher evolutionary rates. Late-stage onward transmission appeared rare, though a subset of infections acquired hallmark mutations seen in emerging variants.

Email interview with Prof Katrina Lythgoe, lead geneticist for the ONS-CIS study and associate professor in the Department of Biology at Oxford, and Mahan Ghafari, DPhil, from Pandemic Sciences Institute at the University of Oxford.

Contagion: Your study identified accelerated viral evolution in a subset of persistent infections. What do you see as the most significant risks these cases pose for the emergence of new variants?

Lythgoe and Ghafari: "In our population-based cohort, persistent infections (we define this as those with infections lasting longer than 30 days and high viral load activity) were not very rare — roughly 0.7–3.5% of all covid infections can last at least that long. Within this group, only a small slice (about 2%) showed the kind of accelerated evolution that produces highly mutated viruses. Most importantly, among our high-confidence prolonged persistent infections, we didn’t detect any household transmission from persistent infections; based on our data the chance of household spread looked low (on the order of <1 in 84 persistent infections)."

Lythgoe and Ghafari: "So what does that mean for variant emergence? It tells us the full chain of events needed for a “new variant” — rapid evolution and successful spread — seems possible but rare in the community. A lot of these persistent infections may be transmission dead-ends. The headline risk isn’t zero, but it’s lower than many feared from community-based persistence alone. That said, we should remain alert in settings where the virus has both time and opportunity to transmit (eg, immunocompromised hosts with prolonged shedding or those under strong treatment pressure)."

Contagion: You found no strong associations between host factors (such as vaccination status or lineage) and evolutionary rates, apart from infection duration. What does this suggest about which patients clinicians should prioritize for monitoring?

Lythgoe and Ghafari: "The short answer is infection duration was the only strong signal: the longer someone stays positive, the more likely that we find highly mutated viruses during infections. A key limitation of our study was that we didn’t have access to participants’ medical records (eg, immunocompromising conditions and antiviral treatments), so we couldn’t directly test clinical risk factors in this study. That said, other studies from immunocompromised patients—for example those with haematological malignancies, transplant recipients, or with poorly controlled HIV—suggest evidence for a higher likelihood of prolonged infection and faster within-host viral change. So, a practical takeaway from a clinical point of view would be: (a) Flag people who remain PCR-positive beyond ~3–4 weeks for closer follow-up and, where feasible, targeted sequencing. (b) In clinical settings, immunocompromised patients are sensible priorities based on the broader literature."

Contagion: ORF6 emerged as a highly conserved region under strong purifying selection. From a clinical or therapeutic perspective, how promising is this as a potential drug target?

Lythgoe and Ghafari: "We found ORF6 to be highly conserved and under strong evolutionary pressure to stay intact, which usually means the virus can’t easily change it without losing its fitness. ORF6 plays an important role in helping SARS-CoV-2 switch off the cell’s alarm system by blocking interferon responses, essentially by interfering with the cell’s nuclear import machinery. If you could block ORF6, you might restore those early immune defences and hurt the virus across different variants. The challenge is that ORF6 is a small accessory protein without a clear “active site” to target. That makes it harder to design drugs against than, say, the polymerase or main protease, which already have effective antivirals, or the spike protein, which is the target of antibodies. Developing therapies against ORF6 would likely require more innovative approaches, such as drugs that disrupt its interactions with host proteins, or host-directed therapies that tweak the nuclear import pathways."

Data were obtained from the Office for National Statistics COVID-19 Infection Survey, a large longitudinal, population-based study in the UK (April 2020–March 2023). Participants were randomly sampled from the general population and underwent regular reverse transcription polymerase chain reaction (RT-PCR) testing irrespective of symptoms. The analysis included infections with high viral load (cycle threshold [Ct] ≤30) and available genome sequences from seven major lineages (alpha, delta, BA.1, BA.2, BA.4, BA.5, XBB). Persistent infection was defined as sustained or rebounding high viral RNA titers for ≥26 days. Using raw sequence data, investigators identified de novo mutations and estimated within-host synonymous and nonsynonymous evolutionary rates across the SARS-CoV-2 genome.

Within-host SARS-CoV-2 evolution during persistent community infections was highly heterogeneous. A small subset of high-viral-load, persistent infections showed accelerated evolution or recurrent acquisition of variant-defining mutations, yet onward transmission beyond the first month appeared uncommon. These findings support prioritizing genomic surveillance of persistent infections and inform management strategies for affected patients.

Reference

Ghafari M, Kemp SA, Hall M, et al. SARS-CoV-2 genomic diversity and within-host evolution in individuals with persistent infection in the UK: an observational, longitudinal, population-based surveillance study. Lancet Microbe. Published online July 30, 2025. doi:10.1016/j.lanmic.2025.101154