Age at cytomegalovirus, Epstein Barr virus and varicella zoster virus infection and risk of atopy: The Born in Bradford cohort, UK

Abstract Background The prevalence of allergic diseases has increased in recent decades, but the causes remain unclear. Changes in the epidemiology of childhood infections could have contributed, but the current evidence is inconclusive. This study aims to investigate whether age at cytomegalovirus (CMV), Epstein‐Barr virus (EBV) or varicella zoster virus (VZV) infection is associated with the development of atopy. Methods A total of 2559 children were enrolled in the Born in Bradford Allergy and Infection Study. Serum samples collected at 12 and 24 months were tested for CMV‐IgG, EBV‐IgG and VZV‐IgG for 1000 children to establish age at infection. Skin prick testing (SPT) was conducted at age 4 years. Results Serology and SPT results were available for 740 children. Of these, 135 (18%) were atopic. In girls, there was a strong association of CMV infection in the second year with increased odds of atopy (adjusted OR 4.38, 95% CI 1.87‐10.29) but this was not observed in boys. Age at EBV or VZV infection was not associated with risk of atopy in unadjusted analysis, but there was effect modification by sex; girls infected with VZV in the second year of life had increased odds of atopy (adjusted OR 2.85, 95% CI 1.29‐6.30). Conclusions Our results highlight potential sex‐specific effects of age at CMV infection and age at VZV infection on risk of atopy, which provide insight into the mechanisms involved in the development of atopy.

Cytomegalovirus (CMV), Epstein-Barr virus (EBV) and varicella zoster virus (VZV) are persistent herpesviruses commonly acquired in childhood. CMV and EBV are usually asymptomatic, and VZV causes chickenpox. There is good evidence that CMV and EBV affect the developing immune system. 8,9 CMV impacts on T-cell 10 and NK cell 11,12 differentiation, and CMV seropositivity is associated with inflammation, atherosclerosis and immunosenescence. 13,14 EBV is associated with immune disorders such as multiple sclerosis and lymphoma. 15,16 The features of CMV and EBV in particular indicate that changes in the epidemiology of these infections could have contributed to the increases in eczema, hay fever and allergic asthma.
In a Swedish study, no association was found between CMV or EBV seropositivity at age 4 and asthma, hay fever or eczema, although CMV-positive/EBV-negative children were more likely to have specific IgE to common allergens than children seronegative for both infections. 17,18 In another study, children who were EBV seropositive at 24 months were less likely to be IgE-sensitized at this age compared to EBV-seronegative children and this association was enhanced among children co-infected with CMV. 19 Another analysis by the same group showed that children infected with EBV before 2 years were less likely to be IgE-sensitized at age 2 and 5 years while those infected after 2 years were more likely to be IgE-sensitized at 5 years. 20 There was no association between CMV, EBV or VZV infection at age 12 months and eczema, asthma, hay fever, total serum IgE levels or allergic sensitization at age 7 in a German study. 21 In a UK casecontrol study, there was no association between EBV or VZV infection and eczema in 1-to 4-year-olds. 22 Other studies have examined the association between seropositivity to CMV, EBV, VZV and other infections and risk of atopy or eczema, hay fever, or asthma, but the combined evidence is inconclusive. [23][24][25] In some, the findings were limited by a lack of statistical power, but in most, infection status was measured at only one time point, often at age 4 or older, so it was not possible to pinpoint age at infection more precisely. As herpesvirus infections are common, it is unlikely that infection per se is associated with risk of disease, but alteration in risk could result from infection at a critical period in immune development.
Measuring infection status at more than one age in the first years of life is therefore important. In the Born in Bradford Allergy and Infection Study, children were tested for CMV-, EBV-and VZV-IgG at 12 and 24 months. We have previously reported earlier infection in Pakistani compared to white British children. 26 The aim of this analysis is to investigate whether earlier CMV and/or EBV infection protect against the development of atopy.

| ME THODS
The Allergy and Infection Study (ALL IN) is a sub-study of the Born in Bradford birth cohort (BiB), described elsewhere. 27 Children were eligible to participate if they were enrolled in BiB with a maternal baseline questionnaire available and were born on or after 1 March 2008. Parents were invited to participate in ALL IN just before the child's first birthday. If they agreed, a home or clinic visit was arranged and informed consent was obtained. A blood sample was taken, and a questionnaire was completed at 12 months and again at 24 months of age. Serum aliquots were stored at −80°C. The questionnaires included detailed information on breastfeeding and childcare, and the standard International Study of Asthma and Allergies in Childhood (ISAAC) questions on potential risk factors for atopy (pets, damp, mould, fuel for cooking and heating, type of windows in the child's bedroom, type of flooring, type of pillow and bedding for the child). 28 Of the 2559 children enrolled in ALL IN, serum samples were tested for CMV-IgG, EBV-IgG and VZV-IgG for 1000 children to establish age at infection; infected by 12 months; infected between 12 and 24 months; or uninfected at 24 months. 26 Samples with concentration values of ≥6.0 AU/mL were considered as positive for CMV-IgG, and VCA IgG concentrations of ≥20 U/mL were considered as positive for EBV-IgG. For VZV, samples containing 160 mIU/mL were considered positive, those between 140 and 160 were equivocal and those less than 140 mIU/mL were deemed negative.
When the children were 4 years old, the parents were re-contacted and invited for a visit including a questionnaire and skin prick test (SPT), as part of the MeDALL (Mechanisms of the Development of Allergy) study. 29,30 The SPT was performed according to a standard protocol using the following allergens: cat hair, dog hair, grass mix, house dust mites Dermatophagoides pteronyssinus and Dermatophagoides farinae plus a positive control (histamine 1%) and a negative saline control (Allergopharma kits supplied by Diagenics Ltd). The mean wheal diameter was recorded for each allergen (and the diameter of any wheal from the negative control subtracted).
A wheal diameter of ≥3 mm was considered a positive reaction.
Children with a positive reaction to at least one allergen were defined as atopic.
The community research staff were trained in SPT, with quality control procedures to assess intra-observer variability; all staff had to perform two monthly quality checks and achieve a coefficient of variation of less than 20 in their SPT series.

Key Message
We present skin prick test data from a large population-

| Statistical analysis
Data analysis was conducted using Stata version 14. 31 Frequency distributions for key variables described the children and their mothers. Cross-tabulations showed associations between these variables and atopic status. Age at CMV, EBV and VZV infection were the key exposures of interest. The outcome for this analysis was measured at age 4 years only and in all children at the same age. Our data therefore provide an estimate of prevalence, not risk, of atopy, so it is appropriate to present prevalence odds ratios from logistic regression analysis. If univariable analysis demonstrated an association with atopy, the relevant variable was included as a confounder in the multiple logistic regression analysis. The findings for age at CMV infection were compared between the univariable and multiple logistic regression analyses, and no evidence of multicollinearity was found, so the multiple regression analysis was accepted as the final model. Birth order and duration of breastfeeding were included in the CMV model a priori as these are strongly associated with age at CMV infection 26 and associated with atopy in other studies. [32][33][34] Effect modification by sex and ethnic group was investigated. 35 Previous studies have indicated interaction between EBV and CMV infection on atopy/IgE sensitization, 18,19 so we compared the proportion of children who were atopic for combinations of age at CMV/EBV/VZV infection.
The number and type of positive allergens per child were described and examined by ethnic group and sex.

| Unadjusted analysis
Among the children who were CMV infected by 12 months and those who remained uninfected at 24 months, 17% were atopic (31/181 and 82/485, respectively). However, of the 74 children who were CMV infected between 12 and 24 months, 30% (22) were atopic. There were no differences in the proportion of children atopic by age at EBV infection or by age at VZV infection (18% or 19% in each group).
Boys were twice as likely to be atopic than girls, and Pakistani children were twice as likely to be atopic than White British children.
Childcare attendance by age 24 months, reporting pets in the home and central heating at the 12-month visit, and smoking in pregnancy were associated with reduced risk of atopy. Low birthweight and damp reported at 12 months were associated with an increased risk of atopy (Table 1). There was not strong evidence of an association with pets and damp reported at the 24-month visit. There was no evidence of associations between atopy and mould in the home, type of windows in child's bedroom, gas for cooking, gas fire, wood or coal fire with chimney, type of flooring or bedding (data not shown).
Only nine respondents reported making changes due to asthma/ allergies (mainly changing to hard floors or new carpets, different bedding) at the 12-month questionnaire (three were atopic), and only 17 reported making changes at the 24-month questionnaire (five were atopic).
The associations of atopy with breastfeeding, childcare, pets and smoking in pregnancy may reflect differences in the prevalence of these factors between white British and Pakistani children; Pakistani women were more likely to breastfeed, less likely to use childcare and far less likely to smoke than white British women. Pakistani families were less likely to have pets than white British families.
The 740 children included here were generally similar to the total 1000 children with serological data, although there was a slightly higher proportion of Pakistani children.

| Adjusted analysis
There was some evidence of effect modification of the association between age at CMV infection and atopy by sex (LRT χ 2 = 4.80, P = 0.09), adjusting for ethnic group, birth order, duration of breastfeeding, regular childcare attendance by 24 months, low birthweight,  Table 2).
There was also evidence of effect modification by sex of the association between age at EBV infection and atopy (LRT χ 2 = 5.87, P = 0.05) and age at VZV infection and atopy (LRT χ 2 = 8.52, P = 0.01); the sex-specific estimates are presented in Table 2. Girls infected with VZV in the second year of life were at increased odds of atopy (aOR 2.85, 95% CI 1.29-6.30). There was no evidence of effect modification by ethnic group of the association between age at CMV, EBV or VZV infection and atopy.

| Which allergens were positive?
There were five allergens in the panel. Most of the 135 atopic children were positive for one (n = 56) or two (n = 52) allergens (Table 3).
Positive reactions to house dust mites (HDM) were most common,  TA B L E 2 Sex-specific estimates for the association between age at CMV, EBV and VZV infection and atopy Our results highlight potential sex-specific effects of age at CMV infection on risk of atopy. Overall, boys had a greater risk of atopy than girls, as reported in other studies, 38,39 but differences in risk of atopy by age at CMV infection were only observed for girls. So why is "delayed" CMV infection associated with atopy in girls but not in boys?

| D ISCUSS I ON
There are other examples of sex differences in response or susceptibility to infections early in life, 40,41 and neonatal vitamin A supplementation (NVAS) was associated with increased atopy in girls but not boys in long-term follow-up of children enrolled in a previous randomized controlled trial in Guinea-Bissau. 42 As in our study, females were less likely to be atopic than males overall, but those receiving NVAS were at increased risk of atopy (adjusted RR = 1.78, 95% CI 1.17-2.71) and wheeze during childhood (adjusted RR = 1.78, 95% CI 1.02-3.09) compared to those receiving placebo. Possible mechanisms to explain these associations are that retinoic acids from vitamin A can lead to an enhanced Th2 response and a reduced Th1 response. High oestrogen levels can also promote a Th2 response, and girls experience surges in oestrogen levels in early infancy as well as in puberty. A further study by the group demonstrates that NVAS is associated with differences in response to BCG vaccination by sex and increased pro-to anti-inflammatory cytokine ratios in females but decreased ratios in males, at 4-6 months of age. 43 In the Generation R cohort (population-based cohort from foetal life onwards, in Rotterdam, the Netherlands), no interaction of sex and CMV infection was observed in relation to immune phenotypes but the children were tested at age 6 years, 35  immune development, that is the first 2-3 years, influences risk of atopy at age 4 years, and later infection has less impact.
We did not observe any differences in risk of atopy by age at EBV infection, in contrast to the Swedish studies, 19  SPT is widely used and accepted as a valid tool to define atopy. 28 Several members of the team carried out the tests, and there may have been variations in technique between them, despite training and quality control checks. However, any measurement error and subsequent misclassification of atopic status would have been nondifferential, as the research team were unaware of the infection status of the children. Some studies use sIgE to define atopic status, instead of, or in addition to, SPT. There is generally good correlation between sIgE and SPT (>70% concordance), at least in high-income country settings. 50,51 Not all children with a positive SPT will develop clinical symptoms of eczema, hay fever or allergic asthma. Some non-atopic children may develop different types of asthma caused by non-atopic mechanisms.
Further work will investigate whether the higher odds of atopy among girls infected with CMV in the second year translates to an increased risk of eczema, hay fever, wheeze or asthma among these children.
In conclusion, our results highlight potential sex-specific effects of age at CMV infection and age at VZV infection on risk of atopy, which may provide insight into the mechanisms involved in the development of atopy. Goodier (LSHTM) for helpful comments on the manuscript.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest in relation to this work.

AUTH O R S' CO NTR I B UTI O N S
LP conceived and designed the study, analysed and interpreted the data and drafted and revised the manuscript. DW co-ordinated data collection, interpreted the data and revised the manuscript. PG designed the study, interpreted the data and revised the manuscript.
JW designed the study, interpreted the data and revised the manuscript. All authors have given final approval of the version to be published.