Genotype Analysis of Respiratory Syncytial Virus Before and After the COVID-19 Pandemic Using Whole-Genome Sequencing: A Prospective, Single-Center Study in Korea From 2019 to 2022

Respiratory syncytial virus (RSV) is a highly transmissible virus that is the leading cause of lower respiratory tract infections (LRTI) in infants and young children.1 Globally, approximately 94.6 infants per 1,000 person-years (uncertainty range [UR]; 70.8–131.6) experience RSV-associated LRTI. Furthermore, 15.9 infants per 1,000 person-years (UR; 12.6–21.2) require hospitalization, even though a majority of them were previously healthy without underlying illnesses.1, 2 Moreover, this virus is recognized as life-threatening in the elderly population.3

RSV is an enveloped, negative-sense, single-stranded RNA virus that contains 10 genes: nonstructural protein 1 (NS1), NS2, nucleoprotein (N), phosphoprotein (P), matrix (M), short hydrophobic protein (SH), glycoprotein (G), fusion protein (F), transcription processivity factor (M2), and large polymerase complex (L).4 Traditionally, RSV genotypes (strain or clade) have been determined based on the genetic characteristics of the hypervariable region 2 within the G gene.5 Recently, several RSV-specific antiviral drugs have been developed, which mostly target the F protein; however, they can also target the NS2 and L proteins, and the selective pressures imparted by these drugs can result in mutations in non-G genes.6, 7, 8 Therefore, genetic surveillance based on whole-genome sequencing (WGS) has become increasingly important.9

The coronavirus disease 2019 (COVID-19) pandemic led to an abrupt interruption of seasonal RSV outbreaks in Korea.10 The RSV epidemic in Korea during the 2019–2020 period terminated early in February 2020, and there were almost no RSV outbreaks for approximately 1.5 years (until October 2021), when domestic non-pharmaceutical interventions (NPIs) were eased.11, 12 This phenomenon provided a serendipitous opportunity to investigate the impact of the COVID-19 pandemic and the corresponding effects of NPIs on the molecular trends of RSV in Korea. Furthermore, we investigated the presence of resistance-associated substitutions against RSV-specific monoclonal antibodies (palivizumab, nirsevimab, and suptavumab).

During the study period, 276 children with RSV infections were hospitalized and 288 respiratory samples were collected. Among these, 133 samples (46.2%) were available for amplification: 71 (53.4%) from the pre-COVID-19 period and 62 (46.6%) from the COVID-19 period (Fig. 1A). Among the 133 samples included in the WGS, 6 cases are samples obtained from three patients with an interval of more than a week between collections. The male to female ratio was 1.1:1, and infants accounted for 47.5% of the study population. The detailed demographics are presented in Supplementary Table 5.

Fig. 1
Flowchart of study population selection and monthly trend of RSV hospitalizations. (A) Selection schematic of study population and samples. (B) Monthly RSV hospitalization data at SCH during the study period. The shaded portion represents clinical surveillance data on RSV hospitalizations from the KDCA.
PCR = polymerase chain reaction, RSV = respiratory syncytial virus, COVID-19 = coronavirus disease 2019, WGS = whole-genome sequencing, SCH = Severance Children’s Hospital, KDCA = Korea Disease Control and Prevention Agency.

• Click for larger image
• Download as PowerPoint slide

RSV epidemics and molecular epidemiology in Korea from 2019 to 2022

The data on monthly RSV hospitalizations at Severance Children's Hospital closely aligned with the RSV trends in the clinical surveillance operated by the KDCA (Fig. 1B). RSV-B was predominant during the 2019–2020 season; however, the RSV epidemic ended abruptly in February 2020. A resurgence in RSV cases began in November 2021; RSV-B was initially predominant; however, RSV-A became predominant after February 2022 (Fig. 1B).

RSV-A was dominated by the GA2.3.5 (ON1) strain, whereas RSV-B was dominated by the GB5.0.5a (BA) strain. No significant changes were observed before or after the pandemic (Fig. 2). The P-distance, which represents the degree of nucleotide differences by the total number of nucleotides between seasons (i.e., 2019–2020, 2020–2021, 2021–2022), was not substantial (ranging from 0.008 to 0.011). Additionally, within each season, the degree of simple variation between strains was not significant (ranging from 0.005 to 0.010) (Table 1). An examination of the degree of variation in the 10 RSV genes through WGS revealed that the G protein had the highest nucleotide substitution rate with 2.53 × 10⁻³ substitutions per site per year (s/s/y) (95% highest posterior density [HPD]: 2.05–3.05 × 10⁻³ s/s/y) and 4.31 × 10⁻³ (95% HPD: 3.35–5.36 × 10⁻³ s/s/y) for RSV-A and RSV-B, respectively. Regarding the F gene, the nucleotide substitution rate was low (ranging from 6.34–8.65 × 10⁻⁴ s/s/y in RSV-A and 8.41 × 10⁻⁴–1.30 × 10⁻³ s/s/y in RSV-B) (Fig. 3). RSVs with identical G regions were identified, and the number of changes outside the G region were determined (Supplementary Figs. 2 and 3). In all determined identical G regions, at least 1 nucleotide difference was noted in all panels; notably, more gaps in genome were more prevalent in post-pandemic samples. Tracking WGS nucleotide differences rather than G region alone may be beneficial in tracking short-term transmission patterns.

Fig. 2
Phylogenetic tree of RSV-A and RSV-B before and after the coronavirus disease 2019 pandemic based on whole-genome sequencing. (A,C) depict the phylogenetic tree of pre-pandemic RSV (from September 2019 to February 2020), whereas (B,D) represent the phylogenetic tree of post-pandemic RSV (from November 2021 to December 2022). The red circles represent RSV samples from the 2019–2020 season, the blue circles represent samples from the 2021–2022 season, and the circles with blue borders represent samples from the 2022–2023 season.
RSV = respiratory syncytial virus.

• Click for larger image
• Download as PowerPoint slide

Table 1
Estimation of average P-distances between and within RSV seasons in Korea

• Click for larger image
• Click for full table
• Download as Excel file

Fig. 3
Estimates of nucleotide substitution rates for RSV-A and RSV-B using whole-genome sequencing. The analysis was conducted using the search-thinned datasets and was carried out using the BEAST program.
RSV = respiratory syncytial virus.

• Click for larger image
• Download as PowerPoint slide

Site-specific substitution analysis of the RSV F gene

Table 2, 16, 17 shows the site-specific substitutions in the RSV F gene. I206M:Q209R and I206M:Q209R:S211N substitutions were observed in all RSV-B strains (n = 77). I206M:Q209R accounted for 100% (n = 43) of substitutions identified in the pre-pandemic period, whereas the I206M:Q209R:S211N combined substitution accounted for 63.9% (n = 22) of substitutions in the post-pandemic period (I206M:Q209R accounted for 36.1% (n = 12)). There were no known reported major mutations that confer resistance to nirsevimab in the F2 and F1 regions of either the RSV-A or RSV-B strains. Mutations conferring resistance to palivizumab, which is currently available in Korea, were not observed in any of the analyzed RSV strains. Notably, mutations known to confer resistance to suptavumab, such as L172Q and S173L, were observed in 100% of the analyzed RSV-B strains (n = 77).

Table 2
Substitution analysis of binding sites on F protein-targeting monoclonal antibodies and comparisons with previous studies

• Click for larger image
• Click for full table
• Download as Excel file

In this prospective surveillance study, we assessed changes in the molecular epidemiology of RSV before and after the COVID-19 pandemic in Korea using WGS for the first time. Despite the unprecedented disruption of the seasonal RSV epidemic, there were no significant genotypic changes in the circulating RSV strains in Korea between the pre- and post-pandemic periods, and their genotypic trends aligned with global patterns. No substitutions in the F gene, which is known to confer resistance to RSV-specific monoclonal antibodies nirsevimab and palivizumab, but not suptavumab, were identified.

Our findings revealed that all the analyzed RSV-A strains belonged to GA2.3.5 (ON1), and all the analyzed RSV-B strains were GB5.0.5a (BA). According to the INFORM-RSV study conducted in eight countries (the United Kingdom, Spain, the Netherlands, Finland, Japan, Brazil, South Africa, and Australia) during the pre-pandemic season (2017–2018), the BA9 genotype of RSV B and ON1 genotype of RSV-A were predominant in all countries except South Africa.5 Similar findings were reported during the pre-pandemic period (2017–2018) in Korea.18, 19 Yun et al.18 reported that in Korea, the ON1 genotype of RSV-A had replaced all other RSV-A genotypes since 2012, whereas the BA genotype of RSV-B (mainly BA9) had replaced all other RSV-B genotypes since 2007. The minimal changes in the RSV genotypes in the post-pandemic period could be explained by a genetic bottleneck. Recent studies comparing the prevalent RSV genotypes pre- and post-pandemic have reported similar results to ours, indicating that there were no significant genotypic changes. According to recent research by the Australian RSV study group, there was no seasonal RSV outbreak in the winter of 2020. However, in late 2020, there was an out-of-season RSV-A outbreak of the previously prevalent ON1 genotype, which showed limited genetic diversity based on WGS analysis.20 Therefore, it is possible that in Korea, the RSV strain that was prevalent prior to the pandemic might have been circulating at undetectable levels locally and re-emerged after NPIs were lifted.

However, the limitation of strain diversity by NPIs does not necessarily guarantee the resurgence of previously dominant strains. Lin et al.21 conducted a genetic analysis of an RSV-A outbreak in Taiwan, and targeted the G and F genes. They reported the prevalence of the RSV-A ON1 strain in the second half of 2020, which exhibited distinct features from the predominant ON1 strains during the 2013–2016 epidemics. Pierangeli et al.22 exclusively analyzed the G gene, revealing that the RSV-A ON1 strain was predominant in Rome from October to December 2021, and closely resembled strains from the pre-pandemic RSV season. However, the RSV-B BA10 subtype, which dominated the 2022–2023 season, was different from the predominant RSV-B strains during the 2018–2019 season in Rome.

Our findings underscore the high degree of conservation in the F gene, which enhances the likelihood of stable drug efficacy for RSV-specific drugs. In an international study on nirsevimab, which recently received Food and Drug Administration approval in 2023 for pregnant women and older individuals, the authors also reported that a prevalence of < 1% of substitutions would affect the RSV neutralization capacities of both nirsevimab and palivizumab.5, 14 However, suptavumab, a fully human monoclonal antibody targeting site V of the pre-F protein, failed to meet the primary endpoint in a recent clinical trial due to high levels of resistance against RSV-B strains with L172Q/S173L variants.23 Moreover, these RSV-specific monoclonal antibodies can potentially exert selective pressure after their widespread use in clinical practice, even though their prevalence is currently extremely low. In a study analyzing the F gene of RSV samples collected from 2009 to 2015 in Korea, Choi et al.13 reported the prevalence of the N201S substitution (16.7%), which reduced the neutralization capacity of nirsevimab. Therefore, continuous monitoring of RSV genotypes at the national level is imperative. Continued genetic surveillance is crucial because there’s a possibility that some rare strains or variants could emerge as predominant strains during the post-pandemic period.

The correlation between RSV genotype and disease severity has been suggested in the literature but with limitations. Li et al.24 analyzed the G gene in a study conducted during an RSV-A ON1 epidemic in Guangzhou, China during the 2018–2019 season, reporting a cluster with a novel set of five amino acid substitutions that exhibited milder symptoms such as reduced wheezing and shorter symptom duration. Additionally, Midulla et al.25 reported in a study conducted in Italy on hospitalized RSV-infected infants under one year old that RSV-A NA1 showed more severe chest retractions and a higher likelihood of hypoxia compared to RSV-A ON1 or RSV B BA. However, these studies require careful interpretation due to risk factors such as older siblings and smoke exposure not being fully controlled, and variations in the extra-G regions were not evaluated. Therefore, more well-designed clinical studies using the WGS method are necessary for a precise analysis.

This study also has limitations. First, this was a single-center study, which may not represent the overall Korean pediatric population. Second, as this study used residual samples, we were unable to gather clinical information on disease severity in the participants. However, regardless of the severity, the genotypes of all specimens analyzed in this study were similar to those of ON1 and BA. Moreover, because factors such as immunological debt due to the absence of natural boosting during the NPI period can also influence severity, there may be limitations in explaining virulence based solely on viral genotypes. To address this issue, future studies that include both clinical and immunological information are needed.26 Third, we were unable to analyze the genotypes of unamplified samples, which accounted for approximately 53.8% of collected specimens. However, positive rates were also observed in other studies and were higher than the historical RSV culture rate of 10–30%.18, 27 Finally, we were unable to conduct neutralization tests to directly assess drug resistance. To address this, we are developing a microneutralization assay for RSV to evaluate the effectiveness of RSV-specific drugs against domestic RSV variants.

Nevertheless, our study results can help to elucidate the molecular epidemiology of RSV using WGS. The long-read next-generation sequencing technology and interpretation pipeline established in this study are expected to play a significant role in the molecular surveillance of RSV in Korea. In conclusion, our WGS analysis revealed no significant genetic changes in RSV genomes related to nirsevimab or palivizumab resistance when comparing the pre- and post-pandemic periods in South Korea. Furthermore, the unpredictability of RSV seasonality after the pandemic highlights the importance of year-round surveillance, rather than relying solely on traditional seasonal monitoring.

Supplementary Method 1

RNA extraction and amplification

Supplementary Table 1

Primer sets used in this study

Supplementary Table 2

Quality control of SMRTbell run summary

Supplementary Table 3

GenBank accession number, country, genotype, year of RSV-A and RSV-B reference sequences

Supplementary Table 4

The parameter settings for phylogenetic analysis

Supplementary Table 5

Baseline characteristics of study subjects and samples

Supplementary Fig. 1

Schematic diagram of long-reads whole-genome sequencing of RSV.

Supplementary Fig. 2

Comparison of RSV-A genomes with identical G regions. Each panel represents RSVs with identical G gene sequences highlighted in gray. Orange vertical lines are for nucleotide substitutions, blue lines are for deletions, and red lines are for insertions.

Supplementary Fig. 3

Comparison of RSV-B genomes with identical G regions. Each panel represents RSVs with identical G gene sequences highlighted in gray. Orange vertical lines are for nucleotide substitutions, blue lines are for deletions, and red lines are for insertions.