Association between nasal and nasopharyngeal bacterial colonization in early life and eczema phenotypes

Abstract Background An association has been reported between early life Staphylococcus aureus nasal carriage and higher risk of childhood eczema, but it is unclear whether this relationship is causal and associations with other bacterial species are unclear. Objective To examine the associations of early life nasal and nasopharyngeal bacterial carriage with eczema phenotypes, and the direction of any associations identified. Methods Among 996 subjects of a population‐based prospective cohort study, nasal swabs for Staphylococcus aureus, and nasopharyngeal swabs for Streptococcus pneumoniae, Moraxella catarrhalis and Haemophilus influenzae were collected and cultured from age 6 weeks to 6 years. Never, early, mid‐, late transient and persistent eczema phenotypes were identified from parental‐reported physician‐diagnosed eczema from age 6 months until 10 years. Multinomial regression models and cross‐lagged models were applied. Results Staphylococcus aureus nasal carriage at 6 months was associated with an increased risk of early transient and persistent eczema (OR (95% CI): 2.69 (1.34, 5.39) and 4.17 (1.12, 15.51)). The associations between Staphylococcus aureus nasal carriage and eczema were mostly cross‐sectional, and not longitudinal. No associations of Staphylococcus pneumoniae, Moraxella catarrhalis and Haemophilus influenza nasopharyngeal bacterial carriage with eczema and eczema phenotypes were observed (OR range (95% CI): 0.71 (0.35, 1.44) to 1.77 (0.84, 3.73)). Conclusions Early life Staphylococcus aureus nasal carriage, but not Staphylococcus pneumoniae, Moraxella catarrhalis and Haemophilus influenza nasopharyngeal carriage, was associated with early transient and persistent eczema. Staphylococcus aureus nasal carriage and eczema were mostly cross‐sectionally associated, and not longitudinally, making a causal relationship in either direction unlikely.


| INTRODUC TI ON
Childhood eczema is a common chronic skin disorder with variable age of onset and persistence. 1 We previously identified eczema phenotypes taking into account the variability of eczema onset and persistence within and between individuals over time. 2 The use of eczema phenotypes, instead of the simplified dichotomous outcome of eczema, might better reflect the natural course of eczema and help understand their specific underlying risk factors.
Both genetic and environmental factors seem to influence the development and persistency of eczema. 3 Additionally, bacterial carriage of the main commensals Staphylococcus aureus, Haemophilus influenzae, Moraxella catarrhalis and Streptococcus pneumoniae in the nasal cavity and nasopharynx was suggested to be associated with eczema. 4,5 The nasal and nasopharyngeal area may function as important reservoirs for bacteria to spread to different body sites. In addition, competitive and cooperative inter-bacterial, and host-bacterial interactions affect the microbial colonization dynamics and the priming of the host's immune responses, and thereby altering the susceptibility of developing atopic diseases. 6 A previous meta-analysis of mainly hospital-based cohorts showed that nasal carriage of S. aureus was associated with an increased risk of eczema in children and adults. 4 We previously showed in a population-based cohort that early life nasal carriage of S. aureus was associated with increased risk of eczema and eczema severity in children aged 1-2 years, but persistent effects at older ages were not clear. 7 Also Haemophilus influenzae, Moraxella and Streptococcus pneumoniae in the nasopharynx are suggested to be associated with increased risk of eczema. 5,8,9 However, studies only used vaccinations against Haemophilus influenzae and Streptococcus pneumoniae, not bacterial carriage, and were performed in hospital-based or adult populations. 5,8,9 Furthermore, it remains unclear whether bacterial nasal and/or nasopharyngeal carriage leads to increased risk of the development of eczema, is a consequence of eczema, or occurs simultaneously with eczema due to other mechanisms. 10 Therefore, we aimed to examine the associations of early life bacterial nasal and nasopharyngeal carriage with eczema phenotypes from birth until age 10 years among 996 subjects of a population-based prospective cohort study.
Next, we aimed to disentangle whether the direction of associations was from bacterial nasal and nasopharyngeal carriage leading to an increased risk of eczema or reversely.

| Design
This study was embedded in the Generation R Study, a populationbased prospective cohort study from early foetal life onwards in Rotterdam, the Netherlands. 11 Figure 1).

| Bacterial nasal and nasopharyngeal carriage
Swabs of the nose and nasopharynx area were taken by trained research nurses at the research centre at ages 6 weeks, 6 and 14 months, and 2, 3 and 6 years, as previously described. 13,14 For this, sterile transport swabs with liquid Amies medium were used. Nasal swabs were put in phenol red mannitol broth at 35°C for 5 days.
Material from tubes that turned yellow was plated on a blood agar plate with 5% sheep blood at 35°C for 1 day to isolate Staphylococcus

| Eczema phenotypes
Information on physician-diagnosed eczema was obtained from parental-reported questionnaires at the ages of 6 months, and 1,

| Covariates
Information on pet keeping and maternal psychiatric symptoms using the Global Severity Index (GSI) was obtained by questionnaires during pregnancy. 15 The mode of delivery was obtained from midwives and hospital records. Postnatal questionnaires provided information on daycare attendance and antibiotic use in the first year after birth.

| Statistical analysis
We compared characteristics of those included and not included in our study using Pearson's Chi-square and Mann-Whitney U tests.
First, we examined the associations of bacterial nasal and nasopharyngeal carriage with ever eczema and with five eczema phenotypes from birth until age 10 years using logistic and weighted multinomial regression models, respectively. Weights were based on class probabilities. Next, cross-lagged models were used to examine bidirectional associations of bacterial nasal carriage with eczema from birth until 10 years. Cross-lagged models allow associations between two repeatedly measured variables to be examined in both directions simultaneously while accounting for continuity between the repeated measures over time. A conceptual model of the studied cross-lagged associations is presented in Figure S1. For example, the effect estimates of the association of bacterial carriage at earlier age with eczema at later age will be adjusted for all earlier associations between and within bacterial carriages and eczema. With this method, we aimed to disentangle the predominant direction of the observed association between bacterial nasal carriage and eczema. We examined cross-lagged effects, cross-sectional effects and stability effects in the period from birth until age 3 years, and only cross-lagged and stability effect in the period from 4 until 10 years due to the uneven distribution of repeated measures of bacterial nasal carriage and eczema at those ages. As a sensitivity analyses for increased statistical power, we applied generalized estimating equation (GEE) models with an unstructured and autoregressive correlation matrix to examine the associations between bacterial nasal and nasopharyngeal carriage at age 6 weeks with repeated measures of eczema from 6 months until 10 years. All analyses were adjusted for potential confounders, which were first selected from literature including known potential underlying biological mechanisms. 2,[16][17][18] Next, confounders were selected if they were associated with both the exposure (bacterial nasal/nasopharyngeal carriage) and the outcome (eczema phenotypes), and were not within the causal pathway based on epidemiological concept. Additionally, they were included if they changed the effect estimates or the unadjusted analyses with ≥10% in adjusted analyses. Family history of atopic diseases, maternal age, parity and education, and child's gestational age, birth weight, sex and breastfeeding did not meet our defined statistical criteria of confounding, and therefore, were not included in the models. For better interpretation, we adjusted all analyses models for the same confounders. We assumed that data were missing at random.
Twenty data sets were created to handle missing data in covariates (≤12%) using multiple imputation by chained equations. Missing data in bacterial nasal and nasopharyngeal carriage and eczema were not imputed. The size and direction of the effect estimates were similar when we used complete-case analyses, and therefore, we only present the results based on imputed data. We did not adjust for multiple testing in the main analyses (nasal carriage with S. aureus and nasopharyngeal carriage with any bacteria), because the bacterial carriages were examined under the same hypothesis. For the sensitivity analysis of the separate nasopharyngeal bacteria, we corrected for multiple testing using alpha .05 divided by the effective independent number of tests calculated based on the correlation structure between the bacteria. 19 All measures of association are presented as odds ratios (OR) together with their corresponding 95% confidence intervals (95%CI). Imputation and regression analyses were performed using the packages 'mice' (version 3.6.0), 'stats' (version 3.6.1) and 'nnet' (version 7.3.12), cross-lagged analyses were performed in Mplus (version 8.2), and using package 'MplusAutomation' (version 0.7-3), and GEE analyses were performed using the package 'geepack' (version 1.2-1) in R version 3.6.1. [20][21][22][23][24][25] 3 | RE SULTS

| Subject characteristics
Characteristics of children and their mothers are shown in Table 1.
Compared with children included in the analysis, those not included had mothers who had more psychiatric symptoms during pregnancy (Table S1). The number of children eligible for inclusion during follow-up was 1190 children at ages 6 weeks to 4 years, 1166 children at age 5 years, and 1109 at age 10 years. Physician-diagnosed eczema ranged from 13.4% at age 6 months to 6.0% at age 10 years (Table S1).  Table S3). This association attenuated to non-significant after correcting for multiple testing.

| Early life bacterial nasal and nasopharyngeal carriage and eczema phenotypes
We observed no associations of nasal carriage of S. aureus or nasopharyngeal carriage with any bacteria at ages 6 weeks with overall eczema from birth until age 10 years in the sensitivity analyses with GEE models (data not shown).   Figure 2A; cross-sectional or cross-lagged associations were observed between nasopharyngeal carriage with any bacteria and eczema ( Figure 2B).

| Direction of associations between bacterial nasal and nasopharyngeal carriage and eczema
Results from cross-lagged models were similar in effect size and direction when examining nasopharyngeal carriage with H. influenzae, M. catarrhalis and S. pneumoniae separately (Table S5).

| DISCUSS ION
In this population-based prospective cohort study, we observed that only nasal carriage of S. aureus at age 6 months was associated with an increased risk of ever eczema, and specifically with an increased risk of early transient and persistent eczema phenotypes until age 10 years. The direction of effects between nasal carriage of S. aureus and eczema was largely cross-sectional, making causality either way unlikely. Nasopharyngeal bacterial carriage with H. influenzae, M. catarrhalis and/or S. pneumoniae from age 6 weeks until age 6 years was not associated with ever eczema or eczema phenotypes from birth until age 10 years.

| Comparison with previous studies
A previous meta-analysis, cohort and case-control studies showed that nasal carriage of S. aureus was associated with an up to fivefold increased risk of eczema in children and adults. 4,26 We observed in our current study that nasal carriage of S. aureus at age 6 months was associated with an increased risk of ever eczema at age 10 years, which is in line with results of our previous study in children until age 2 years. 7 We now additionally explored eczema phenotypes across childhood, taking the onset and persistence of eczema into account, Note: Values are odds ratios (OR) with 95% confidence interval from logistic and multinomial regression models on imputed data. One cohort study showed that H. influenzae vaccination at age 6 months was associated with an increased risk of eczema at age 18 months. 8   Values are odds ratios (95% confidence interval) derived from logistic regression models, using cross-lagged modelling. Models were adjusted for maternal psychiatric symptoms, pet keeping, mode of delivery, daycare attendance and antibiotic use. *p-value < .05, **p-value < .01. The corresponding 95% confidence intervals of the cross-lagged and cross-sectional effects, and the effect estimates of the stability effects are shown in Table S3 S. aureus 6 months

| Strengths and limitations
The strengths of this study include that it is embedded in a shared towel use and shared bedrooms). [39][40][41][42] Also, since we only used one set of confounders mostly measured at early age, residual confounding could be greater for associations between exposures and outcomes at later ages.

| CON CLUS ION
We observed that early life nasal carriage with S. aureus, but not na-

CO N FLI C T O F I NTE R E S T
The authors have no potential conflicts of interest to disclose.