Seroprevalence and Virologic Surveillance of Enterovirus 71 and Coxsackievirus A6, United Kingdom, 2006–2017

Enterovirus A71 (EV-A71) and coxsackievirus A6 (CVA6) cause hand, foot and mouth disease (HFMD) and are occasionally linked to severe neurologic complications and large outbreaks worldwide. We estimated EV-A71 and CVA6 seroprevalence using cross-sectional age-stratified samples collected in 2006, 2011, and 2017. Seroprevalences of EV-A71 and CVA6 increased from 32% and 54% at 6–11 months to >75% by 10 years of age. Antibody titers declined after 20 years, which could indicate infrequent re-exposure in older populations. Age profiles for acquiring infections and mean titers were comparable in the 3 testing years, despite the marked increase in incidence of CVA6-related HFMD from 2010. The uncoupling of changes in disease severity from the infection kinetics of CVA6 as we inferred from the seroprevalence data, rather than incidence of infection over the 11-year study period, provides further evidence for a change in its pathogenicity.

the effects of the virus on public health have been substantial; in Europe these infections are considered mild and often remain undiagnosed (5), although severe neurologic manifestations and small outbreaks have been reported more recently (6)(7)(8)(9)(10). EV-A71 is classifi ed into 7 genogroups (A-G) and several subgenogroups (B0-B5, C1-C5) based on the viral protein 1 gene; the appearance of novel EV-A71 genogroups has been associated with large HFMD outbreaks (5).
Coxsackievirus A6 (CVA6) has become another major cause of HFMD since 2008 (11,12). CVA6 infections have often been linked to a febrile atypical form of HFMD, affecting both pediatric and adult populations (13)(14)(15). The severity of the clinical manifestations associated with CVA6 infections and the recent increase of HFMD cases associated with EV-A71 and CVA6 in Europe (10) may have originated through the evolution of recombinant forms or changes in pathogenicity of emerging strains (16,17). Alternatively, their clinical prominence may have resulted from an increase in infections in a larger previously unexposed and susceptible populations. To investigate that theory, we determined the age-stratifi ed seroprevalence of EV-A71 and CVA6 in representative cross-sections of the UK population in 2006, 2011, and 2017; we used serotype-specifi c microneutralization assays and compared our fi ndings with the numbers of infections reported through public health surveillance.
The 2011 timepoint corresponded to the approximate timing of large EV-A71 outbreaks, especially in Vietnam and China (12,18) in addition to emergence of CVA6 infections associated with atypical clinical phenotypes (11,19). Whereas the 2006 timepoint was selected to precede these recorded events and the 2017 to measure population immunity post-CVA6 emergence period, the last timepoint also corresponded to recorded EV-A71 outbreaks in Spain and elsewhere in Europe in 2016 (4,7,8 Enterovirus A71 (EV-A71) and coxsackievirus A6 (CVA6) cause hand, foot and mouth disease (HFMD) and are occasionally linked to severe neurologic complications and large outbreaks worldwide. We estimated EV-A71 and CVA6 seroprevalence using cross-sectional agestratifi ed samples collected in 2006, 2011, and 2017. Seroprevalences of EV-A71 and CVA6 increased from 32% and 54% at 6-11 months to >75% by 10 years of age. Antibody titers declined after 20 years, which could indicate infrequent re-exposure in older populations. Age profi les for acquiring infections and mean titers were comparable in the 3 testing years, despite the marked increase in incidence of CVA6-related HFMD from 2010. The uncoupling of changes in disease severity from the infection kinetics of CVA6 as we inferred from the seroprevalence data, rather than incidence of infection over the 11-year study period, provides further evidence for a change in its pathogenicity.
timepoints reflected changed activity of both viruses and hence enabled us to measure their effects on population immunity.

Virus Strains
We obtained 2 CVA6 strains isolated in Finland in 2008 and 2016 from the National Institute for Health and Welfare (Helsinki, Finland). The CVA6/2008 isolate was obtained during a HFMD outbreak in Finland (20), and the CVA6/2016 isolate was a contemporary clinical strain. We used the EV-A71 genogroup B4 strain isolated in Singapore (5865/ SIN/000009). We propagated EV-A71 viruses in a rhabdomyosarcoma cell line obtained from the American Type Culture Collection. We propagated CVA6 viruses in TE32 or 130T cells obtained from the UK National Institute for Biologic Standards and Control. We determined the 50% tissue culture infective dose (TCID 50 ) of virus stocks by means of endpoint dilution using the Reed and Muench method: in a 96-well format, 8 replicates of a 10-fold serial dilution were incubated with cells in Dulbecco minimum essential medium (DMEM; Sigma-Aldrich, https://www.sigmaaldrich.com) containing 2% vol/vol fetal bovine serum (FBS; Sigma-Aldrich) and penicillin/streptomycin (10,000 U/mL; Sigma-Aldrich) at 37 °C in 5% CO 2 for 4-5 days.

Neutralization Assays
The microneutralization assay was performed as previously described (21) (Appendix). In brief, we inactivated serum samples for 3 min at 56°C, and then diluted 2-fold serially in 2% DMEM-FBS from 1:8 to 1:1,024. We mixed 50µL of diluted samples and 100 TCID 50 of virus stock diluted in 50 µL in 96-well microplates and incubated at 37°C for 1 hour. We added 100 µL of cell suspension containing average of 20,000 rhabdomyosarcoma cells in 10% DMEM-FBS for EV-A71 assays and average of 20,000 TE32 cells in 5% DMEM-FBS for CVA6 assays. We observed cytopathic effect in an inverted microscope after incubating at 37°C in 5% CO 2 for 4-5 days. We used pooled adult serum with known neutralizing antibody titer (nAb; 13/328, obtained from the UK National Institute for Biologic Standards and Control) as a positive control and inactivated horse serum (obtained from American Type Culture Collection) as negative control. We included a virus control and an uninfected cell control for each batch of tests. We tested each sample in duplicate and calculated results as their geometric mean titers (GMT).
To determine the optimal strain for the CVA6 neutralizing assay, we compared titers of 36 serum samples collected in 2006 against the 2 CVA6 clinical isolates. We selected 18 samples each for the 1-5-year (representing serologic responses acquired during 2001-2006) and >40-year (representing serologic responses acquired substantially before 2006) age groups. For the 1-5-year age group, 16/18 samples tested were seropositive for the CVA6/2008 and 17/18 samples tested were seropositive for CVA6/2016 isolates. All 18 samples tested from the >40-year age group were seropositive for both CVA6 isolates. GMT to both CVA6 isolates were comparable between the 1-5-year and >40-year age groups (Appendix, Figure 1). Samples collected from the >40year age group in 2006 had proportionately higher nAb against the CVA6/2008 isolate (p = 0.008 by paired Wilcoxon signed rank test). Because the differences in GMT between the CVA6 isolates were minor, we selected the more contemporary CVA6/2016 strain for the assay used in this study.

Virological Surveillance Data
We collected information on enterovirus-positive samples submitted for typing to the PHE Enteric Virus Unit (London, UK), during 2006-2017. Local diagnostic laboratories in England and Wales were asked to forward samples in which EV RNA has been detected for typing, for the purposes of national enhanced enterovirus surveillance. Data collected included a total number of enterovirus-positive samples submitted for typing and the number identified as EV-A71 or CVA6 per month, patient age group, and sample type.
We used these data to compare the prevalence of infections estimated from serologic data with EV-A71and CVA6-associated infections reported through this voluntary enhanced enterovirus surveillance.

Statistical Analysis
We compared rates of seropositivity in different groups using χ 2 or Fisher exact test, with Bonferroni adjustment for multiple comparison. We compared age-stratified GMTs between the serum collection time points using the Mann-Whitney U or Kruskal-Wallis test with Dunn's post hoc analysis. We calculated 95% CIs of the seroprevalence rates according to the Wilson method (http://vassarstats. net/prop1.html) and considered p<0.05 statistically significant. We computed all the statistical analyses in R (https://www.r-project.org). Most EV-A71 infections were reported in even years; ≈10% of all enterovirus-positive samples were identified as EV-A71 in 2006, 2008, and 2010, whereas this proportion has remained at ≈3% since 2012. The peak months for EV-A71 detections were July-August and for CVA6 detections were October-December. The highest monthly detections were 20 of EV-A71 in July 2013 and 74 of CVA6 in October 2017 (Figure 1, panel B).
The seropositive samples from infants (<6 months of age) in 2006 had a GMT 5-fold higher than the same age group in 2017, whereas those from children 6-11 months of age in 2006 had a geometric mean titer 3.6fold higher than the same age group in 2017. Similarly, the samples from children 1-5 years of age in 2011 had a GMT 5.5-fold higher than in 2017 (Appendix, Appendix Table 1). Significant increases in titers of seropositive samples were found among children <6 months (p = 0.014 by Kruskal-Wallis test) and 1-5 years of age (p = 0.0026) and also among patients aged 11-20 years of age (p = 0.0067) (Appendix, Appendix Figure).
The proportion of infants <6 months of age with titers >64 was significantly higher in 2006 (75%) than   Table 2).

Discussion
Seroepidemiology findings in this study showed that EV-A71 and CVA6 infections were highly prevalent among children and adults in the United Kingdom. From the minimum values in the 6-11-month age group after the decline of maternally conferred immunity (24,25), we determined that EV-A71 and CVA6 neutralizing antibody detection frequencies and titers increased steadily with age, which indicates ongoing exposure and infection throughout childhood. EV-A71 seropositivity rates observed in the United Kingdom were comparable to those observed among preschool children <6 years of age (63.4%) in Germany (26) and in children <5 years of age in the Netherlands (27). EV-A71 seroprevalence in adults (>75%) was comparable for the United Kingdom, the Netherlands, and Germany. The number of persons with high titers of EV-A71 neutralizing antibodies declined with age; this finding is consistent with previous seroepidemiological studies, including the report of high EV-A71 antibody titers in the 10-14-year age group in Germany (28), and comparable to the peak titers recorded in the 11-20year age group in our study. These findings indicate that EV-A71 primarily circulates in and infects children, and the subsequent decline in titers but not fre-quencies of seropositivity indicates that re-exposure in the older population is uncommon (28)(29)(30). The decline in titers may also reflect the differences between acute serologic responses post-infection in the younger population and homeostatic antibody levels in the older population that become established years after infection (30). Related to this decline, the >4-fold attrition in mean EV-A71 neutralizing antibody titers in the 21-40-year age group (Appendix Table 1) may also create the low mean titers of maternally derived antibodies observed in children <6 months of age. This finding may underpin the high incidence of EV-A71 diagnosis reported in the 0-3-month age group when infants are most susceptible to severe infection outcomes ( Figure 3). Of note, the largest share (39%) of enterovirus-positive samples were obtained from this age group, which might attest to infants' vulnerability and higher likelihood of sampling.
The global emergence of CVA6 since 2008 has been linked to an increase in pathogenicity of CVA6 around 2010 (31), becoming another major causative agent for HFMD in several countries worldwide (23). This change was reflected in the number of atypical HFMD caused by CVA6 in Scotland in 2014 (19) and also in the increasing numbers of reported CVA6 infections in our study (Figure 1). Our seroprevalence data show that CVA6 circulated widely before the emergence of atypical HFMD in 2008 (25); seroprevalence approached 90% in adults >40 years of age as recorded in 2006 (Figure 3). This observation discounts the idea that the increased incidence of CVA6-associated HFMD simply reflects a change in its infection incidence and the existence of a widely susceptible population. Comparing the 2 serotypes, CVA6 seroprevalence was higher than EV-A71 seroprevalence in younger children (1-10 years) in each study year (Figure 3; Appendix Tables 1, 2). However, this difference was not reflected in the peak age group for CVA6 infections (1-5 years) (Figure 2), which contrasts with the predominance of EV-A71 infections recorded in neonates and infants. CVA6 infections were predominantly detected in skin vesicle fluids (Table; Appendix), which would primarily be associated with HFMD manifestations (32)(33)(34).
Over the study period, the number of samples referred to PHE substantially increased (Figure 1), but rather than indicating more enterovirus-associated disease, this finding is more likely a reflection of improvements in detection through exclusive introduction of PCR in the clinical laboratories (35). Diagnostic practices in general, and for enteroviruses in particular, have changed over time in England and Wales as previously described (35). The use of PCR has increased rapidly, from 36% in 2000 to 45% in 2011, and probably approached 100% in 2015, replacing the slow and laborious virus culture entirely.
Changes in clinical practice or diagnostic procedures, such as the threshold for investigating and hospitalizing patients with suspected viral infections, or performing lumbar puncture (35), may have further influenced the number of samples submitted to PHE. Controlled cohort-based surveillance studies are required to better infer EV incidence.
A limitation of this study is that we based our inferences of incidences of EV-A71 and CVA6 infections on referral of clinical samples for typing at PHE. The much lower numbers of EV-A71-positive samples identified from older children and adults ( Figure 2) at a time when seroprevalence was increasing ( Figure 3) is indicative of subclinical infections or benign disease in these age groups. Differences in clinical practices could have also influenced the number of samples obtained and referred from older children and adults to PHE. For instance, CSF samples are more likely to be obtained for enterovirus testing from these patients who had any neurologic symptoms, compared with throat, fecal, or rectal swab specimens from which the viral loads would be higher and virus excretion prolonged (36,37). In addition, delayed lumbar puncture also reduces the likelihood of a positive pathogen detection. Atypical and varying clinical manifestations, especially in older adults, and the absence of CSF pleocytosis may also impede the timely diagnosis of enteroviral infections and consequently reduce the number of samples found to be positive and referred to PHE.
We used a convenience sample of residual serum samples from diagnostic laboratories throughout England. Although we attempted to include equal sample sizes for all ages, the serosurvey was not powered to provide precise seroprevalence estimates for certain age groups. The volume of available specimens, particularly for the younger age groups, was insufficient, thus limiting the number of samples tested and generalization of our results to the larger pediatric population. Convenience samples are also prone to chance variations in sampling between geographic regions. Lack of additional information on participants' risk factors for exposure was another limitation.
In summary, we provide an analysis of age-stratified seroprevalence of EV-A71 and CVA6 in the UK population. Prevalence of infection by both viruses inferred from age-related changes in seroprevalence varied little over the 11-year study period despite the emergence of CVA6-associated HFMD in 2010, implying changes in CVA6 pathogenicity rather than changes in population susceptibility to severe infection outcomes. This study will enable a more detailed understanding of population susceptibility, the emergence of enterovirus serotypes, and potential changes in serotype pathogenicity and transmissibility.