SARS-CoV-2 evolution during prolonged infection in immunocompromised patients

Abstract

Prolonged infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in immunocompromised patients provides an opportunity for viral evolution, potentially leading to the generation of new pathogenic variants. To investigate the pathways of viral evolution, we carried out a study on five patients experiencing prolonged SARS-CoV-2 infection (quantitative polymerase chain reaction-positive for 79-203 days) who were immunocompromised due to treatment for lymphoma or solid organ transplantation. For each timepoint analyzed, we generated at least two independent viral genome sequences to assess the heterogeneity and control for sequencing error. Four of the five patients likely had prolonged infection; the fifth apparently experienced a reinfection. The rates of accumulation of substitutions in the viral genome per day were higher in hospitalized patients with prolonged infection than those estimated for the community background. The spike coding region accumulated a significantly greater number of unique mutations than other viral coding regions, and the mutation density was higher. Two patients were treated with monoclonal antibodies (bebtelovimab and sotrovimab); by the next sampled timepoint, each virus population showed substitutions associated with monoclonal antibody resistance as the dominant forms (spike K444N and spike E340D). All patients received remdesivir, but remdesivir-resistant substitutions were not detected. These data thus help elucidate the trends of emergence, evolution, and selection of mutational variants within long-term infected immunocompromised individuals.

Importance: SARS-CoV-2 is responsible for a global pandemic, driven in part by the emergence of new viral variants. Where do these new variants come from? One model is that long-term viral persistence in infected individuals allows for viral evolution in response to host pressures, resulting in viruses more likely to replicate efficiently in humans. In this study, we characterize replication in several hospitalized and long-term infected individuals, documenting efficient pathways of viral evolution.

Keywords: COVID-19; SARS-CoV-2; coronavirus; long-term infection; prolonged infection.

Fig 1

Timeline showing days since symptom onset along the x-axis and patient data on the y-axis. Treatments are colored as boxes below the x-axis, where the length of the box represents the window of treatment administration. Orange and purple line graphs represent absolute neutrophil count (ANC) and absolute lymphocyte count (ALC), respectively, across the left and right y-axis with units of thousands of cells per microliter. Black shapes represent replicates for a given timepoint, and the horizontal black line indicates the period for which sequence data are available. The black vertical line represents the day of onset either by symptom if applicable or by first positive test if initially asymptomatic.

Fig 2

Overview of SARS-CoV-2 evolution in long-term infected individuals. (A) Heatmap showing timepoints and replicates for each patient. Columns represent the genome sequence samples, and rows represent mutations, where light blue indicates a match to the ancestral Wuhan reference strain, and the darker shades indicate increasing prevalence. Only mutations that were found in more than one replicate with >0.03 occurrence and showed changes over the course of infection are included in the heatmap. (B) Phylogenetic tree showing patients in a representative tree covering the course of the pandemic. (C) Phylogenetic tree of patient 663, the case of suspected reinfection, shown in the context of subsampled sequences representing viral lineage XBB.1. (D) Root-to-tip plot for infection in patient 486 (red) and background (black) of the same viral strain, B.1.311. (E) Root-to-tip plot for infection in patient 637 (orange) and background (black) of the same viral strain, BA.1.1. (F) Root-to-tip plot for infection in patient 640 (light blue) and background (black) of the same viral strain, BA.1. (G) Root-to-tip plot for infection in patient 641 (dark blue) and background (black) of the same viral strain, XBB.1.

Fig 3

Summary of the top 10 most variable iSNVs within each patient. Different mutations are distinguished by color. Error bars show variability across replicates; lack of error bars indicates less than 0.001 deviation in proportion. (A) Patient 486’s top 10 most variable iSNVs. (B) Patient 637’s top 10 most variable iSNVs. (C) Patient 640’s top 10 most variable iSNVs. (D) Patient 641’s top 10 most variable iSNVs.

Fig 4

iSNV accumulation over time and by the coding region. (A) iSNV accumulation across patients, shown over the full duration of sampled infection. Linear regression lines were fitted for a cutoff of 0.01 (iSNV found between 1% and 99% prevalence). (B) iSNV accumulation for a cutoff of 0.25 (25%–75% prevalence). (C) Number of iSNVs by the coding region showing length as a nucleotide on the x-axis and iSNV counts on the y-axis. Shaded region indicates the 95% CI for linear regression of coding region length to iSNV counts. (D) Occurrence of iSNVs throughout the genome by position. The left axis indicates the number of patients with any iSNV in a given position, indicated by either circles or diamonds. Circles indicate that the iSNV was synonymous, and diamonds indicate that the iSNV was nonsynonymous. The right axis indicates the moving average centered on each position spanning a window of 500 bases. Units for the moving average are average iSNVs per patient per 500 bases. Coding regions are colored in alternating black or gray, with the regions of interest labeled and colored.

Fig 5

Rate of change of mutations and evasion of monoclonal antibody therapy. (A) Plot showing the change in proportion per month for iSNVs and P-values indicating a certainty that there was a change in iSNV proportion, colored by the coding region. (B) Plot showing spike E340D, which confers sotrovimab resistance, across all patients. (C) Plot showing spike K444N substitution, which confers resistance to bebtelovimab, across all patients. (D) Structure of SARS-CoV-2 spike showing nonsynonymous iSNVs as red, synonymous iSNVs as blue, and known drug-resistant mutations that rose to dominate in our cohort marked in black.