Results from the second WHO external quality assessment for the molecular detection of respiratory syncytial virus, 2019–2020

Abstract Background External quality assessments (EQAs) for the molecular detection of human respiratory syncytial virus (RSV) are necessary to ensure the standardisation of reliable results. The Phase II, 2019–2020 World Health Organization (WHO) RSV EQA included 28 laboratories in 26 countries. The EQA panel evaluated performance in the molecular detection and subtyping of RSV‐A and RSV‐B. This manuscript describes the preparation, distribution, and analysis of the 2019–2020 WHO RSV EQA. Methods Panel isolates underwent whole genome sequencing and in silico primer matching. The final panel included nine contemporary, one historical virus and two negative controls. The EQA panel was manufactured and distributed by the UK National External Quality Assessment Service (UK NEQAS). National laboratories used WHO reference assays developed by the United States Centers for Disease Control and Prevention, an RSV subtyping assay developed by the Victorian Infectious Diseases Reference Laboratory (Australia), or other in‐house or commercial assays already in use at their laboratories. Results An in silico analysis of isolates showed a good match to assay primer/probes. The panel was distributed to 28 laboratories. Isolates were correctly identified in 98% of samples for detection and 99.6% for subtyping. Conclusions The WHO RSV EQA 2019–2020 showed that laboratories performed at high standards. Updating the composition of RSV molecular EQAs with contemporary strains to ensure representation of circulating strains, and ensuring primer matching with EQA panel viruses, is advantageous in assessing diagnostic competencies of laboratories. Ongoing EQAs are recommended because of continued evolution of mismatches between current circulating strains and existing primer sets.

Results: An in silico analysis of isolates showed a good match to assay primer/probes.
The panel was distributed to 28 laboratories. Isolates were correctly identified in 98% of samples for detection and 99.6% for subtyping. across the six WHO regions to establish a global surveillance infrastructure to monitor the incidence and prevalence of RSV, particularly in LMICs, and to provide standardised scientific data to guide the introduction of RSV vaccines or other immunoprophylaxis/treatments once they become available. 2 Participating laboratories were selected from those within the Global Influenza Surveillance and Response System (GISRS) 3 with a history of successful performance in the molecular detection of influenza viruses. One of the aims of the initial phase of the global surveillance was to evaluate and standardise WHO recommended and other existing RSV molecular tests to enable the accurate detection of RSV by laboratories in all six WHO regions.
To achieve this, in 2016, the first External Quality Assessment exercise for the molecular detection of RSV was successfully developed, launched, and completed by the 14 participating national laboratories. 4 Following the successful implementation of Phase I in 2018, this was expanded to 26 countries in Phase II, each with a national laboratory participating in the Global RSV surveillance Program (Table S1).
This study describes the second of the WHO External Quality Assessments (EQA) performed during 2019-2020 for the molecular detection of RSV.

| METHODS
The objective of the RSV EQA 2019-2020 was to assess laboratory proficiency in the molecular detection and subtyping of RSV into the two main subtypes, RSV-A and RSV-B. In the preparation of an EQA, it is important that the composition of panels is representative not only of historical but also of current circulating RSV strains, and that primers and probes of assays are sufficiently sensitive to the target sites to be able to detect currently circulating lineages.  downloaded from GenBank, aligned with MUSCLE, 9 curated and used for the matching analyses of the CDC and VIDRL primers and probes.
The criteria used to curate the sequences include removing sequences with any ambiguous nucleotides ("N" s), and/or sequences with deletions causing frameshifts. Sequences with incorrect RSV subtype annotation were identified and added to the database of correct subtype.  Consensus trees were visualised with Figtree (1.4.0). 14 P-distances were calculated with MEGA 7 15 for the G-gene ectodomain to characterise the most genetically distinct isolates for inclusion in the panel.

| Sharing and selection of viruses for EQA panel
An in silico analysis was conducted to ensure that the sequences of all RSV isolates shared matched the recommended primer/probe sets. Isolate sequences were aligned using MAFFT (7.305) 16 and then manually edited in AliView (1.28) 17 to retain the primer binding sequences and isolates tested against the recommended WHO CDC RSV detection and VIDRL subtyping assays. Sequences were named in a standardised manner using the approach recently proposed by Salimi et al. 18  from each specimen were randomly selected for quality checks for sterility, stability, and other predistribution testing requirements. They were also tested in-house by a commercial semiquantitative RT-PCR assay for the relevant targets to verify the calculated Ct value and sent to an external UK reference laboratory for confirmation of content. On successful completion of quality control checks, the specimens were labelled and packed for dispatch the first week of December 2019. The panels were stored and shipped at room temperature.

| Development of qualitative survey
The purpose of the survey was to assess the quality of the WHO RSV surveillance system to ensure the data collected by participating laboratories from all countries met set targets and standards. As such the survey was designed by UK NEQAS and the Reference Laboratory Working Group to collect information on general laboratory practices, amplification platforms, nucleic acid extraction methods, RSV molecular detection and subtyping assays used in the EQA. Survey reporting requirements were used to custom design the UK NEQAS website for reporting of information.

| Reconstitution of panel and reporting of results
Participants were required to reconstitute the EQA specimens in 1.2 ml of sterile molecular grade water. For nucleic acid extraction protocol and molecular testing, participants were instructed to follow their existing laboratory protocols/kits instructions. Laboratories were advised to store the lyophilised specimens preferably at 4 C. Once reconstituted, samples were to be extracted, and RNA was stored at À80 C for long-term use.
Countries were given the option to use the RSV singleplex or multiplex RT-qPCR assays provided by the CDC, USA, 7 or VIDRL Australia, 8 respectively, for commercial assays or in-house assays.
Results for the EQA from all assays were required to be reported through the UK NEQAS online platform. Distribution of the panel by UK NEQAS was made to 28 laboratories in 26 countries and commenced in December 2019 with the expected closing date and return of results to be by the end of February 2020. However, it was subsequently agreed that an extension of the closing date to September 2020 be approved given the staffing issues, border closures and shipping challenges of the COVID-19 pandemic.

| Scoring of results
A scoring system was established after consultation between UK NEQAS and the WHO Reference Laboratories (Table 1) (Table S2). Nine RSV strains, the most diverse based on G gene P-distances, were included in the final panel (Table 2) in addition to the historical control.

| Genotyping of EQA isolates
Using the standard approach to genotyping, the four non-historical RSV-A isolates selected for the RSV EQA 2019 were ON1-like strains, and the 5 RSV-B isolates were BA-like ones, with the previously

| Matching of detection and subtyping primers to globally circulating RSV strains
The CDC detection and VIDRL subtyping primers were matched to the curated database of 1010 RSV-A and 633 RSV-B WGS ( Table 3).
The sequences with the highest number of mismatches over the CDC detection primers/probe was found to be a strain from Jordan (HRSV/

| RSV detection and subtyping results
The  (Table  4). Of these, 21 obtained a score of good, and 4 obtained a score of acceptable.  Participant results showed CVs between 0.10 and 0.13 compared with the CV range of 0.04 to 0.08 for Reference Laboratories (Table S3). This suggests that the participating laboratories had lower sensitivities for the detection of RSV than the Reference Laboratories and UK NEQAS.

T A B L E 3 Primers and probe mismatches to current circulating RSV-A and RSV-B strains
Results from the qualitative survey consisted the following: extraction, amplification platforms and detection and typing assays used by National Laboratories.
The QIAamp Viral RNA kit was the most frequently used nucleic acid extraction method (Table S4). Eight amplification platforms were used by participants in the 2019 EQA: the ABI 7500 was the platform most frequently used followed by the BioRad, Rotor-Gene and Cepheid GeneXpert. A total of 10 detection methods were used, with the CDC pan RSV detection assay 7 and multiplex assay 19 being the most frequently used methods, followed by the Cepheid Gene Xpert and the VIDRL duplex assay. 8 The most common typing assay used was the CDC multiplex RSV Assay. 19

| DISCUSSION
The WHO EQA 2019-2020 for the detection of RSV by real-time RT- As the extraction methods, platforms, amplification kits and detection assays used were not unique to these sites, extraction methods (automated/manual) and laboratory techniques and analysis may have contributed to a decrease in sensitivity.
The CDC detection assay provided for this EQA 7 targets the M gene and the VIDRL assay the L gene. 8  Despite the identification of some mismatches, they were judged insufficient to affect the assays' ability to detect the RSV strains in the panel. However, the described mismatches might have slightly impacted assay sensitivity. Of note, a third assay (a CDC multiplex assay targeting the N gene 19 ) was provided to participating laboratories from February 2020; a formal analysis of primer/probe matching was not conducted as part of the design of this EQA, and is therefore a limitation of this study.
Genomic variability in RSV is especially relevant in laboratories that have been using in-house developed assays. A recent survey 20 found that the N gene is the most common RT-PCR target for RSV detection assays, and therefore, ongoing review of WGS of circulating RSV strains is required, as there is the potential for resultant genetic drift which may impact the effectiveness of such diagnostics, the accuracy of which is needed for the RSV detection, management and outbreak investigation. The importance of ongoing surveillance of the entire RSV genome is highlighted by a recent example where mutations in the N gene of circulating RSV-B strains affected the efficacy of a custom rRT-PCR assay for the detection of RSV. 21 The objective of the first WHO EQA 4 was to ensure that laboratories and countries enrolled into the RSV pilot project could acquire suitable surveillance samples and implement a CDC real-time RT-PCR assay for RSV detection. The objectives of the second EQA were to ensure that an expanded network of countries/laboratories using a greater diversity of laboratory methods could detect recently circulating strains, known to be genetically diverse from the selected reference strains. The performance of laboratories at RSV detection showed an improvement in the second EQA compared with the first.
There were fewer errors and closer concordance between observed and expected results. This may reflect greater familiarity with the protocols as well as more experience gained by laboratory workforce during the expansion of global molecular testing during the COVID-19 pandemic.
Overall, the pandemic has had a significant impact on the Phase II of the WHO RSV Surveillance Program. In many countries during the Covid-19 pandemic, RSV circulation largely disappeared or was greatly reduced, meaning few RSV positive cases were present with a consequent reduction in sampling, 22 while in others, the urgent need for surveillance of SARS CoV-2 has diverted resources away from RSV surveillance. Previous patterns of circulation of RSV may continue to be seriously disrupted for some years to come, 23 although co-infection with SARS-CoV-2 and RSV has been documented 24,25 and may increase with time. The fact that both SARS-CoV-2 and RSV present initially with similar clinical manifestations highlights the importance of accurately identifying RSV with implications for costeffective management and control of both diseases. 26

| CONCLUSIONS
The RSV EQA 2019-2020 for countries participating in the Phase II of the WHO Global RSV Surveillance Program showed that the laboratories performed at high standards. Some laboratories reported RT-PCR Ct values more than 2 SD above the mean with higher CVs, com-