Introduction

Allergic respiratory diseases have increased markedly in urbanized, westernized, and affluent populations [1,2]. Immunoglobulin E (IgE) sensitization refers to the production of allergen-specific IgE. IgE sensitization to an allergen is not the same as being allergic to the allergen. Individuals may produce IgE to allergens without developing symptoms if exposed to the allergen, i.e. individuals may or may not have allergic symptoms. It is unclear why some individuals demonstrate only sensitization while others have active allergic disease. IgE sensitization is associated with a lymphocyte T-helper type 2 (Th2) polarization, favouring Th2-derived cytokine production and production of allergen-specific immunoglobulin E (IgE) antibodies and a reduced lymphocyte T-helper type 1 (Th1) response [3]. Hence, it has been hypothesised that the reduced Th1 response could lead to a higher risk of infection [3].

Previous studies have suggested that persons with IgE sensitization and atopic disease may have a higher susceptibility to infection [3–6]. Since IgE sensitization has been shown to be inversely associated with smoking [7,8], and smoking may be associated with a higher risk of pneumonia and other respiratory diseases, smoking habits may be a (negative) confounder of the association between IgE sensitization and pneumonia. Furthermore, in epidemiological studies it is often difficult to discriminate between asthma and chronic obstructive pulmonary disease (COPD), the latter having a well-known high risk of hospitalization with pneumonia and exacerbation of disease. Hence, to avoid misclassification of COPD as asthma and confounding by smoking we performed additional analyses stratified by smoking status and expected any possible effects of IgE sensitization on infection and disease risk to be more evident in never smokers.

To our knowledge, only one relatively small previous study has investigated the prospective association between an objective marker of sensitization and risk of respiratory infections in adults [5]. Allergic respiratory disease is often underdiagnosed and the risk associated with objective biomarkers of atopic disease may add knowledge about the potential burden of atopic disease. IgE sensitization is sometimes asymptomatic, although it is associated with lower airway inflammation. We investigated the association between IgE sensitization, defined as specific IgE positivity against inhalant allergens, and the incidence of acute airway infection, infection and chronic lower airway disease in five Danish population-based cohorts.

Materials and methods

Study populations

We used the following five population based studies: Monica1, Inter99, Health2006, the 1936-cohort, and the Allergy98 study. They were recruited from the Danish Central Personal Register as random samples of the population in the southern part of Copenhagen and included questionnaires, physical examinations, and blood tests as previously described in detail [9,10]. An overview of the studies has previously been published [11].

The cohorts have been described in detail previously [9,10], but to recapture this information: A sample of 7,931 persons 18–69 years of age was invited to a general health examination in the Health2006 study (14). A total of 3,471 (43.8%) persons were examined from 2006–2008. In The Monica1 study, 4,807 persons with Danish citizenship at the ages 30, 40, 50 and 60 years old were invited (1982–84). The participation rate was 79%, yielding a total of 3,785 participants (12). In the Inter99 study (1999–2001), 12,934 persons from 30–60 years of age were invited. A total of 6,784 persons participated, and the participation rate was thus 52.5% (13). The study was a population-based randomized controlled trial (CT00289237, ClinicalTrials.gov), and it investigated the effects of lifestyle intervention on cardiovascular disease (13). The Copenhagen Allergy study that was initiated in 1990 included a group of persons randomly selected from the general population and a selected group of persons with allergic respiratory symptoms (that was recruited from a random sample of the general population by a screening questionnaire). We used data from the follow-up health examination in 1997–1998 (that is referred to as ‘Allergy98’) where 1,966 persons aged 15–77 years with Danish citizenship were invited. A total of 1,216 persons (61.9%) participated (15). And finally, the 1936-cohort was initiated in 1976 where a total of 1,200 randomly selected persons born in 1936 (40 years of age at the time of the study) were invited for a health examination focusing on cardiovascular disease risk factors. Between 1976 and 1977, 1,052 persons were examined (participation rate 87.7%). The 1936 cohort has previously been described (16;17).

From these five studies, 15,098 individuals had measurements of serum specific IgE, and of these, a total of 249 persons participated in more than one of the studies, and these persons were included only in the study in which they were first examined as described in detail previously [9,10]. We thus included a total of 14,849 persons (Inter99: 5,961; Allergy98: 1,172; 1936-cohort: 989; Health2006: 3,246; Monica1: 3,481) (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart of the included study populations.

https://doi.org/10.1371/journal.pone.0171525.g001

Registry-based diagnoses

The permanent and unique personal civil registration number of every person who lives in Denmark enables linkage of data from national registers on an individual level. Information on diseases according to the International Classification of Diseases (ICD) was obtained from the Danish National Patient Register [12]. The register holds information on all admissions to Danish hospitals since 1978. Outpatient visits such as emergency department visits have been recorded in the registries since 1995 and are included in the analyses. The Danish Registry of Causes of Death provided diagnoses suspected to be the cause of death [13]. Fatal and non-fatal events were combined. The included diagnoses are summarized in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Categorization of acute airway infection, infection, and chronic lower airway disease according to ICD-8 and ICD-10 codes.

https://doi.org/10.1371/journal.pone.0171525.t001

Participants with a registry-based diagnosis of the event before baseline were excluded from the prospective analyses (Fig 1). The Danish Civil Registration System provided information on death and emigration status [14]. We followed the participants until 31 December 2010.

Results

Table 2 shows the baseline characteristics of the participants according to study population and sensitization status expressed as mean (standard deviation, SD) or % (number). The IgE sensitization frequency ranged from 14.9–37.3% in the five study populations. IgE sensitization was associated with male sex, young age, higher consumption of alcoholic drinks, higher BMI, and vocational training.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Baseline characteristics according to study population and sensitization status.

https://doi.org/10.1371/journal.pone.0171525.t002

The results of the meta-analyses of the study-specific estimates from Cox regression analysis showed that IgE sensitization was associated with a higher risk of pneumonia, asthma and chronic lower airway disease (Fig 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Fixed effects meta-analysis of the association between IgE sensitization and disease.

Abbreviations: CI, confidence interval; HR, hazard ratio.

https://doi.org/10.1371/journal.pone.0171525.g002

The median follow-up was 11.3 years (Inter99: 11.0; Allergy98: 12.6; 1936-cohort: 33.8–33.9; Health2006: 3.5; Monica1: 27.3–27.4 years). The number of events was for pneumonia: N = 918, other acute airway disease: N = 216, infection: N = 956, asthma: N = 329, and other chronic lower airway disease: N = 713.

Fig 3 shows the associations of IgE sensitization with the included diseases for never smokers. In general, the analyses stratified by smoking status showed that never smokers compared to ever smokers had an increased risk of the included outcomes, except for risk of infection that was relatively unaffected by smoking habits (data not shown). In particular, IgE sensitized never smokers had a statistically significantly higher risk of pneumonia (HR = 1.73, 95% CI: 1.23, 2.44) and asthma (HR = 3.17, 95% CI: 2.10, 4.76) (Fig 3). However, there was also a higher risk of asthma among IgE sensitized former and light smokers, but no association between IgE sensitization and asthma among moderate and heavy smokers.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Fixed effects meta-analysis of the association between IgE sensitization and disease among never-smokers.

Abbreviations: CI, confidence interval; HR, hazard ratio.

https://doi.org/10.1371/journal.pone.0171525.g003

In additional analyses of the Health2006, Inter99 and the Allergy98 study, we excluded persons with self-reported asthma at baseline (including 8445 persons and a total of 214 events), and this yielded a HR of 1.12 (95% CI: 0.83, 1.51) of pneumonia for IgE sensitized vs. non-sensitized and restricting these analyses to only include never smokers (3241 persons and 57 events) yielded a HR of 1.20 (95% CI: 0.69, 2.08). In further analyses with exclusion of persons with a registry-based diagnosis of asthma before or during follow-up (including 13331 persons and 829 events), we found a HR of 1.12 (95% CI: 0.93, 1.33) of pneumonia for IgE sensitized vs. non-sensitized, and restricting these analyses to only include never smokers (4643 persons and 165 events) yielded a HR of 1.53 (95% CI: 1.08, 2.19). The estimates were largely similar when we only included primary diagnoses and inpatient contacts and excluded deaths (data not shown). Regarding mortality, we found that death caused by pneumonia for sensitized vs. non-sensitized had an adjusted HR = 0.91 (95% CI: 0.60, 1.40; p = 0.678). The odds ratio for pneumonia without asthma during follow-up was 2.78 (95% CI: 1.47, 5.23, p = 0.002) for IgE sensitized vs. non-sensitized.

Discussion

In a large population-based study using registry-based follow-up to obtain information on respiratory infections and diseases, we found that IgE sensitization was associated with a significantly higher risk of a registry-recorded event of pneumonia, asthma and other chronic lower airway diseases. The higher risk of asthma associated with IgE sensitization is well-known, while the observed increased risk of pneumonia is not generally recognized as a co-morbidity of IgE sensitization. We found no association between IgE sensitization and infections in general.

Our results are in line with a study by Jung et al who found that atopic disease other than asthma was associated with an increased risk of serious pneumococcal disease [6]. Cirillo et al found that young adult patients with allergic rhinitis had more numerous and prolonged respiratory infections than non-allergic persons, and Rantala et al reported that atopic adults had more respiratory infections than non-atopics [5] and the number of respiratory infections increased with increasing levels of specific IgE antibodies to common allergens [3]. Ciprandi et al found that allergic children with a positive skin prick test or a history of allergic disease had more and more severe respiratory infections than non-allergic children [4]. Regarding atopy and chronic airway disease, Fattahi et al found that atopy is a risk factor for respiratory symptoms in COPD patients [16]. The development of immune tolerance with allergen-specific immunotherapy has long been successfully used in the treatment of IgE mediated allergic diseases. Of note, Ciprandi et al found that 40 children treated with allergen-specific immunotherapy had less severe extra-allergic surrogate markers of infection in comparison with controls [17]. In another study, Ciprandi et al found that sublingual immunotherapy may be able to reduce the number and duration of upper respiratory infections in patients with allergic rhinitis [18].

The potential mechanisms that may underlie the positive associations between IgE sensitization and pneumonia are unclear. It may involve both the suggested reduced Th1 response among IgE sensitized that may be important to the protection against infections as well as the mucosal inflammation and decreased defence against infectious agents associated with allergic diseases in general, e.g. among patients with atopic dermatitis [3]. Other possible mechanisms for the observed association between IgE sensitization and infection include a deficient interferon response; impaired alveolar macrophage function; changes of airway epithelium which may increase the susceptibility to infection; and genetic effects since many genes that are associated with the susceptibility to e.g., atopic disease are functional immune response genes [19]. In addition, several common external factors are associated with both increased susceptibility to infection and with asthma exacerbations or asthma control, e.g., smoking [19]. Of note, however, another study by Rantala et al, showed that recent respiratory infections were strong determinants for adult-onset asthma [20]. Other previous studies have found more frequent and more severe infections such as pneumonia in persons with atopy and asthma, both among children and adults [19,21–23]. This suggests that IgE sensitization may not be causally related to the susceptibility to infections but rather that IgE sensitization and infection share common pathways, or that persons more prone to infection develop atopic disease more frequently while keeping their propensity to get infections. However, we excluded persons with a history of the disease of interest prior to the time of assessment of IgE sensitization thereby minimizing the risk of reverse causation, i.e. infection/airway disease having an influence on IgE sensitization status. Regarding chronic lower airway diseases, the observed significantly higher risk of registry-recorded events among IgE sensitized, it is possible that, in persons with lower chronic airway disease, IgE sensitization does confer an increased risk of hospital admission due to exacerbation of disease. Another explanation could be misclassification of asthma as chronic lower airway disease. In some patients, there may be overlap between disease phenotypes or it may be difficult to discriminate between different diagnoses.

The strengths of our study include the large population-based samples and the longitudinal design which is superior to cross-sectional ones to demonstrate temporal relationships. The main analyses in the current study are Cox regression analyses since our outcome variable of interest is the time to event where the hazard ratio can be roughly interpreted as the incidence rate ratio [24]. Cox regression analysis works for censored survival times and suits our study with long-term and variable length follow-up, where a considerable number of persons whose observed times are censored before the end of the study which is not well taken care of by e.g., logistic regression analysis where these persons must be omitted or assumed not to have had the event. Also, we used standardised registry-based diagnoses and a long-term follow-up with few persons lost. Another strength is the use of measurements of serum specific IgE positivity to inhalant allergens. Commercially available immunoassays were used to determine serum specific IgE. We have previously reported that these results are in agreement with skin prick test reactivity and allergic respiratory disease [25,26]. However, it should be borne in mind that IgE sensitization does not reflect allergic respiratory disease per se. The use of three different methods to determine serum specific IgE positivity could potentially bias the results. However, there were no signs of significant heterogeneity between the studies.

The limitations of the study include the relatively low participation in some of the cohorts and the fact that we did not include information on self-reported diagnosis of allergy and asthma in the main analyses. We only had information on self-reported asthma diagnosis in three of the cohorts. However, the results from additional analyses with exclusion of persons with self-reported asthma diagnosis at baseline or a registry-based diagnosis of asthma either before or during follow up were in line with the main results, although the results from the additional analyses were only statistically significant when we included never smokers only. Also, as we only included cases where persons died, were seen in hospital either as inpatient, outpatient or emergency room contacts, we likely missed a substantial part, i.e. because most cases are probably treated by their private practitioner. However, we included the most severe cases, i.e. those that were hospitalized. There were no reported cases of the diagnoses of interest before 1976 for the 1936-cohort, probably because reporting to the register was incomplete. Consequently, there may be some misclassification of some of the persons (classified as incident cases in the 1936-cohort) that should have been classified as having a history of the event and excluded from the Cox regression analyses, had the register been more complete before 1976. It is also a potential limitation that we used data from five different studies, as this may induce biases. However, the meta-analyses showed no sign of heterogeneity across studies. In addition, we did not exclude immunocompromised subjects which may confound the finding if the immunocompromised were unequally distributed in sensitized and non-sensitized. However, the very low prevalence in the general population makes this a minor risk. It may be discussed whether reporting of incidence and the use of Cox regression analysis are appropriate since it only considers the first incident of, e.g., pneumonia, while failing to notice a subsequent diagnosis of, e.g., asthma. Indeed, there is a risk of misclassification of the outcome since pneumonia may be a sign of undiagnosed asthma, and this may explain some of the association between IgE sensitization and pneumonia. Although we used both self-reported and registry-based diagnosis of asthma, underreporting of asthma may still be an issue, since it may be missed when treating the exacerbations only. The exclusion of persons with a known asthma diagnosis at baseline and stratification by smoking habits may not entirely solve this problem.

The high and increasing prevalence of allergic disease warrants studies that investigate possible health effects of IgE sensitization beyond allergic diseases. Previously, we have found that IgE sensitization was not associated with diseases such as autoimmune disease or cancer [9,10,27,28]. In the current study we found that IgE sensitization was associated with a significantly higher risk of asthma and other chronic lower airway diseases as expected but also with a significantly higher risk of registry-recorded events of pneumonia. However, the positive association between IgE sensitization and pneumonia may be explained by an undiagnosed asthma causing the pneumonia, and further studies are needed for confirmation.