Clinical, Genomic, and Immunological Characterization of RSV Surge in Sydney, Australia, 2022

Nasopharyngeal swabs discarded after routine diagnostic testing were collected from RSV-positive infants (aged <12 months) who presented with respiratory symptoms to Sydney Children’s Hospital Randwick, Australia between May and September 2022. RSV infection was diagnosed by using the Allplex Respiratory Panel 1 (Seegene, South Korea) at the Serology and Virology Division, New South Wales Health Pathology. Specimens (for which there was a sufficient remaining sample) were collected immediately after diagnostic reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and frozen at −80°C until genomic sequencing (Supplemental Information).

Clinical details for identified cases were ascertained via retrospective medical record review to extract data on demographics, comorbidities, level of respiratory support, level of care (outpatient, ward, intensive care), length of hospitalization, and death (Supplemental Information). Treatment with palivizumab did not exclude infants from the study. For analysis, severity was grouped as mild, moderate, or severe. Patients who (1) did not require treatment or received low-flow oxygen were determined to have mild disease, (2) required high-flow oxygen, noninvasive ventilation, or continuous positive airway pressure were determined to have moderate disease, and (3) required mechanical ventilation, or extracorporeal membrane oxygenation or more invasive management, were determined to have severe RSV disease.

Continuous variables were summarized by using median and interquartile range. Comparisons of continuous variables between groups were performed by using 2-tailed Mann-Whitney U tests. χ-square tests with Yates’ continuity correction were used for the analysis of categorical variables, and Fisher’s exact test was used when the expected value of a cell was <5.

Hybridization-capture sequencing of RSV-positive specimens was performed as described in the Supplemental Information. Taxonomic classification of viral reads in all samples was achieved by using CZ ID v7.1, a cloud-based open-source bioinformatics pipeline for metagenomic sequencing data. Viruses other than RSV that were detected at ≥100 reads based on nucleotide alignments were deemed positive and included in coinfection analysis. Deduplicated host-filtered reads were downloaded from CZ ID for in-house variant calling, consensus genome assembly, and phylogenetic analysis, as described in detail in the Supplemental Information.

Analysis of consensus variants focused on nonsynonymous mutations that were supported by a minimum of 20 reads at the consensus level (allele frequency ≥50%). We also investigated minor variants (MV), defined as those variants with an allele frequency of ≥3% and <50% that occurred at genomic positions with a minimum depth of coverage of 200 reads (ie, ≥6 reads supporting the MV out of ≥200 total). Fisher’s exact tests were performed to evaluate variants differentially occurring between mild and severe disease groups.

Discarded clinical sera of women 20 to 40 years of age were convenience sampled from the Serology and Virology Division New South Wales Health Pathology Laboratory serology archive before each RSV outbreak occurring between 2020 and 2022 in the Southeastern Sydney Local Health District. Twenty samples were collected during each time period, which included February 2020 to March 2020, October 2020, February 2021 to March 2021, and February 2022 to March 2022 (Fig 1). NAb titers to RSV-A and RSV-B were measured by plaque reduction neutralization test in duplicate, as we have previously described.²¹  Ninety-percent inhibitory concentrations (IC₉₀) were calculated by using 4-parameter dose-response curves (GraphPad Prism). Unpaired, 2-sided Student t tests were used for statistical comparison between groups.

This study was approved by the Sydney Children’s Hospitals Network Human Research Ethics Committee (2020/ETH00718).

Capture-hybridization sequencing was performed on 297 nasopharyngeal swabs collected from 264 RSV-infected infants <12 months of age in Sydney, Australia. There were 104 RSV-A (39.4%) and 160 RSV-B (60.6%) cases.

Clinical data were available for all 264 patients, although severity information was unavailable for 1 patient. Based on the level of respiratory support intervention, 71 (27%) infants were determined to have severe disease. Younger patients (median age 79 days), and those presenting in the month of July 2022 were significantly overrepresented in the severe disease group (P = .0001 and P = .0005, respectively). Patient sex, comorbidities (including cardiovascular, respiratory, and immunologic), prematurity, RSV subtype, and RSV RT-qPCR cycle threshold value were not significantly associated with disease severity (Table 1 and Supplemental Table 4).

Viral coinfections were detected by using capture-hybridization sequencing in 66 (25.1%) infants (Table 1). In order of frequency, coinfecting viruses included non-SARS coronaviruses, picornaviruses, bocavirus, adenovirus, parainfluenza virus, and influenza A virus (H3N2; Table 2). There was 1 case with 2 co-detections and 1 case with 3 co-detections. Coinfection with any specific virus was not significant (P = .61).

Sequencing was performed in 4 batches of 22, 70, 58, 80, and 67 samples, respectively (Supplemental Table 3). Among specimens from which a complete RSV genome was obtained (>95% coverage), genotyping of the G gene revealed that all RSV-A sequences were GA2.3.5 lineage (ON1-like) and all RSV-B sequences were GB5.05a lineage (BA-like). Phylogenetic analysis with publicly available sequences collected since 2017 revealed that 2022 Sydney RSV (Syd-RSV) sequences generated as part of this study were closely related to RSV viruses circulating globally (Fig 2). All Syd-RSV-A sequences were most closely related to historical US and UK viruses. Syd-RSV-B were clustered with Australian RSV sequences detected during an off-peak RSV outbreak from late 2020 to early 2021 and sequences from the United States (collected from mid- to late 2022), Spain, the United Kingdom, and the Philippines (collected in 2019).

For variant analysis, we excluded (1) any consensus-level variants (allele frequency ≥50%) occurring at genomic positions with a total read depth of <20 and (2) minor variants (allele frequency ≥3% and <50%) occurring at genomic positions with a total read depth of <200. The distribution and allele frequencies of nonsynonymous mutations across the fusion gene are shown in Fig 3. Antigenic sites Ø (targeted by nirsevimab) and II (targeted by palivizumab) are the target of currently licensed prophylactic mAbs. Antigenic site IV is the target of an mAb undergoing phase 3 clinical trials (clesrovimab), and site V is the target of the discontinued mAb suptavumab.

At the consensus level, for RSV-A, there were 1, 1, and 2 mutations within the antigenic sites II, IV, and V, respectively. For RSV-B, there were 3, 1, 3, and 2 consensus-level mutations detected, within antigenic sites Ø, II, IV, and V, respectively (Fig 3, Supplemental Table 5).

Of note, 1 patient (1%) who was receiving palivizumab prophylaxis had RSV-B with a single-nucleotide polymorphism resulting in K272Q, which has been reported to result in complete loss of palivizumab activity.²²  Within the nirsevimab epitope (site Ø), 1 RSV-B sample (1%) contained the minor variant I64M, which has not been reported elsewhere. The site Ø I206M:Q209R polymorphism detected in the majority of RSV-B samples within our cohort became frequent worldwide between 2015 and 2021, with I206M reported to reduce the neutralization potency of nirsevimab 5-fold.^23,24  Likewise, S211N (detected in 7% of RSV-B genomes) has been extensively characterized and, as a single amino acid change, has no known effect on neutralization by nirsevimab. Across the whole genome, no specific mutations were statistically associated with disease severity.

The possibility of waning population immunity to RSV since the beginning of the COVID-19 pandemic was investigated by using serum from women of childbearing age collected at baseline of the past 4 Syd-RSV epidemics (Fig 1). These were tested for NAbs to both RSV-A and RSV-B. The geometric mean of RSV NAb titers did not significantly differ between any of the collection time points for both subtypes (Fig 4). Across all time points, mean neutralizing titers were 318.9 (SD factor 2.66) for RSV-A and 105.1 (SD factor 3.08) for RSV-B (Supplemental Table 6).

The significant clinical burden of RSV infection occurs for both immunocompromised and healthy children.²⁵  In keeping with this notion, our study revealed that in RSV-infected infants <12 months of age, more than one-quarter of hospital visits by otherwise healthy children resulted in admission and receipt of high-flow oxygen support. Although not statistically associated with disease severity, we frequently (25.1%) detected viral coinfections with a hybridization-capture sequencing method utilizing a panel targeting 29 respiratory viruses (>41 000 probes). Although reported rates of viral coinfection with RSV vary widely, partly due to historical limitations in multiplexing capacity and clinical awareness,^26 –28  such a high frequency of viral coinfections in our cohort is likely reflective of the coinciding resurgence of other respiratory viruses after recent years of disrupted circulation during the COVID-19 pandemic.^6,29 

The Syd-RSV genotypes detected in our cohort largely reflect those circulating worldwide since 2017, similar to recently reported RSV outbreaks in the United States.^9,10  Evolutionary rates for RSV-A and RSV-B have recently been estimated at 1.48 × 10⁻³ and 1.92 × 10⁻³ nucleotide substitutions per site per year, respectively.³⁰  The branch lengths separating Syd-RSV sequences from historical international samples (collected between 2017 and 2022) are consistent with these published estimates of the evolutionary rates. The close phylogenetic relationship between some RSV-A and -B sequences with recent viruses sampled globally suggests several viral introductions from the northern hemisphere into Sydney in 2022. This follows the dramatically reduced RSV genetic diversity observed in Australia throughout 2020 and 2021 when public health measures related to the COVID-19 pandemic were implemented.³  Together, our findings reveal both the continued endemicity of RSV in Australia during this time period (through clustering of Syd-RSV-B with historical Australian sequences) and a rapid rebound in local genomic diversity after the resumption of international travel.

Repeated maternal viral exposure is considered an important factor in maintaining protective antibody levels in infants. In the absence of circulating RSV infection, NAb titers have been shown to wane rapidly.³¹  Indeed, a recent study revealed significant differences in peak–trough NAb titers in 2021 after an absence of RSV cases during the 2020 season.³²  However, although NAb titers vary significantly from peak–trough after an RSV outbreak, we found no difference in the baseline (preoutbreak) NAb titers of serum from women of childbearing age collected before each of the past 4 RSV epidemics occurring in Sydney. These contrasting findings are due to differences in the timing of specimen collection between the 2 studies.³²  Our data indicate that before the 2022 season, antibody-mediated immunity to RSV was, at the population level, no lower than in previous seasons.

Across all seasons, the mean RSV-A neutralizing titers were 3-fold higher than RSV-B neutralizing titers. This may be a consequence of previous immunity to RSV-A in the population stemming from previous epidemics. Whether this contributed to the predominant circulation of RSV-B seen in our study (in contrast to the RSV-A outbreaks observed in the northern hemisphere) requires further investigation. High RSV-A neutralizing titers have previously been associated with protection against all RSV infections, suggesting a level of protective antibody cross-reactivity.³³  The authors of ongoing studies will investigate subtype-specific NAbs in the population after the RSV-B-dominated 2022 season.

An absence of RSV exposure in 2020 resulted in delayed off-season outbreaks seen globally in 2021. Although both the detection of globally established RSV genomes and the high frequency of coinfecting viruses appear to support hypotheses of immune debt as a significant contributor to the size and burden of RSV outbreaks during the 2022 season,^9,10,34  our findings suggest that a more complex interplay of unknown factors, and not waning antibody-mediated immunity, led to the increased RSV case numbers and disease burden observed in 2022 outbreaks. It is also important to note that significant changes in infrastructure, social attitudes, and clinical practice since the COVID-19 pandemic have led to an increased frequency of testing for respiratory infections, making direct comparisons of pre- and post-pandemic RSV detections and rates of positivity difficult. Although the observed surge in cases may be partially attributed to the increased frequency diagnostic of testing, combined clinical and hospital surveillance reports of corresponding RSV outbreaks in the northern hemisphere suggest a genuine rise in RSV transmission and clinical severity in 2022.^4,9,10  In combination with testing data, the consideration of clinical presentations and hospitalization rates is needed in comprehensive epidemiologic studies to provide a holistic understanding of RSV dynamics post-COVID-19.

Because of challenges in developing an infant RSV vaccine, passive immunization through virus neutralization with mAbs is the primary therapeutic option to protect infants against infection in the first 6 months of life, when they are most vulnerable to RSV-induced lower respiratory tract disease. The recently European Medicines Agency- and FDA-approved nirsevimab has high neutralizing activity and enhanced stability and has revealed promising results in protecting infants against severe RSV disease.^35,36  However, prophylaxis with mAbs relies on a high level of conservation with viral epitopes. A primary concern of the enhanced global use of mAbs is the potential selection for, and spread of, mAb-resistant viral variants. We detected the palivizumab resistance mutation K272Q in the sole patient in the cohort receiving palivizumab for RSV. Breakthrough infections by resistant RSV variants are detected in ∼5% of patients treated with palivizumab.²²  Although palivizumab-resistant viruses reveal impaired fitness in vitro and only occur at low frequencies within the community,³⁷  molecular surveillance programs remain important for monitoring therapeutic resistance. For example, in the 2015 to 2017 RSV seasons, a phase 3 clinical trial of mAb suptavumab failed because of the rapid emergence of a high-frequency neutralization-resistant RSV-B variant.²⁰ 

The detection of a single minor I64M variant within antigenic site Ø in our study has not been described elsewhere to date. This amino acid mutation does not alter the polarity from the wild type; however, the characterization of this binding site substitution should be investigated to understand functional changes potentially affecting nirsevimab neutralization. Although conservation within the nirsevimab binding site is high, consensus-level resistance mutations at fusion protein positions 68 and 201 have been found at low frequencies (<1%) in other studies.^23,38  It is unknown whether low-frequency variants detected in RSV molecular surveillance programs will be selected over time with widespread mAb prophylaxis in infants. With the recent approval of the first RSV vaccine for older adult populations by the FDA and UK Medicines and Healthcare products Regulatory Agency,^39,40  ongoing RSV surveillance is also needed to assess selective pressure potentially applied by mAbs and vaccines at the population level and to understand implications of a changing RSV genome for therapeutic efficacy and potential clinical phenotypes.

Limitations of our study include the convenience sampling method evaluating clinical specimens from a single tertiary pediatric hospital, excluding infants with mild RSV illness not seeking medical attention. Although RSV became a nationally notifiable disease in Australia in 2022, this did not change clinical practice with respect to testing or treatment because all children presenting with respiratory symptoms at Sydney Children’s Hospital Randwick are tested for RSV as standard of care. Because of sample availability, we also measured RSV-neutralizing titers in women of childbearing age as a proxy for pediatric samples. Ideally, prospective clinical, virological, and immunologic data from a broad range of infections, including community cases from a larger cohort of infants would enable a more comprehensive analysis of variables related to RSV. However, our data are broadly reflective of the local and international reports of increased RSV detections and burden during the COVID-19 pandemic.^5,7 

Our findings suggest that a complex interplay of factors, beyond a simplistic reduction in population antibody-mediated immunity, contributed to the variability of seasonal RSV epidemics. Continued genomic and immunologic surveillance programs are required to understand these factors and devise optimal strategies for managing RSV outbreaks, particularly in the context of the evolving landscape of immunotherapeutics. Such studies can inform the future design and use of vaccines, particularly mRNA redesign postvaccine utilization.

We would like to thank all authors who generously shared RSV genome data on GenBank.

COVID-19

coronavirus disease 2019

ECMO

extracorporeal membrane oxygenation

FDA

US Food and Drug Administration

IC₉₀

90% inhibitory concentrations

mAb

monoclonal antibody

MV

minor variants

NAb

neutralizing antibody

RSV

respiratory syncytial virus

RT-qPCR expansion

reverse transcription-quantitative polymerase chain reaction

Syd-RSV

Sydney respiratory syncytial virus