|Home | About | Journals | Submit | Contact Us | Français|
Rationale: Aeroallergen sensitization and virus-induced wheezing are risk factors for asthma development during early childhood, but the temporal developmental sequence between them is incompletely understood.
Objective: To define the developmental relationship between aeroallergen sensitization and virus-induced wheezing.
Methods: A total of 285 children at high risk for allergic disease and asthma were followed prospectively from birth. The timing and etiology of viral respiratory wheezing illnesses were determined, and aeroallergen sensitization was assessed annually for the first 6 years of life. The relationships between these events were assessed using a longitudinal multistate Markov model.
Measurements and Main Results: Children who were sensitized to aeroallergens had greater risk of developing viral wheeze than nonsensitized children (hazard ratio [HR], 1.9; 95% confidence interval [CI], 1.2–3.1). Allergic sensitization led to an increased risk of wheezing illnesses caused by human rhinovirus (HRV) but not respiratory syncytial virus. The absolute risk of sensitized children developing viral wheeze was greatest at 1 year of age; however, the relative risk was consistently increased at every age assessed. In contrast, viral wheeze did not lead to increased risk of subsequent allergic sensitization (HR, 0.76; 95% CI, 0.50–1.1).
Conclusions: Prospective, repeated characterization of a birth cohort demonstrated that allergic sensitization precedes HRV wheezing and that the converse is not true. This sequential relationship and the plausible mechanisms by which allergic sensitization can lead to more severe HRV-induced lower respiratory illnesses support a causal role for allergic sensitization in this developmental pathway. Therefore, therapeutics aimed at preventing allergic sensitization may modify virus-induced wheezing and the development of asthma.
Allergic sensitization and virus-induced wheezing in early life are risk factors for childhood asthma inception, but the developmental sequence between them is unknown.
Allergic sensitization precedes rhinovirus wheeze, but the converse is not true. This timing and the plausible mechanisms by which allergic sensitization can lead to more severe rhinovirus illnesses support a causal role for allergic sensitization in this developmental pathway.
Aeroallergen sensitization and virus-induced wheezing episodes during infancy and early childhood are risk factors for subsequent asthma development, and children with both risk factors in early life are at particularly high risk of having asthma at school age (1, 2). Understanding the sequence of events that lead to sensitization and virus-induced wheezing during early life is important to identifying potential causal relationships in the development of atopic asthma. Early allergic sensitization has consistently been identified as a risk factor for asthma development, and plausible mechanisms by which allergic sensitization leads to greater severity of viral respiratory illnesses have also been proposed (3). If allergic sensitization causes subsequent virus-induced wheezing and asthma inception, then preventing or interrupting the development of allergic sensitization would be a potential strategy for asthma prevention.
In contrast to this concept, some animal models of early life infection suggest that virus infection could lead to allergic sensitization (4, 5). In clinical studies, Sigurs and colleagues reported that respiratory syncytial virus (RSV) bronchiolitis leading to hospitalization during infancy was a risk factor not only for asthma but also for allergic sensitization at 13 years of age (6). However, others have been unable to replicate these findings, and a systematic, longitudinal assessment of the development of allergic sensitization and virus-induced wheezing is needed to better understand these relationships during early childhood.
We hypothesized that allergic sensitization during early life predisposes children to more severe viral respiratory illnesses and wheezing, particularly with human rhinovirus (HRV), and that wheezing respiratory illnesses do not increase the risk for the subsequent development of allergic sensitization. To address this hypothesis, we prospectively and longitudinally assessed the timing and etiology of viral illnesses and the development of aeroallergen sensitization in children participating in the Childhood Origins of ASThma (COAST) birth cohort study. Some of the results of this study were previously reported as an abstract (7).
A total of 289 children at high risk for asthma and allergic disease based on parental histories of asthma and allergic sensitization were enrolled at birth and followed prospectively in the COAST study (8). The sample size was originally calculated to assess the contributions of viral respiratory infections and immune dysregulation to the inception of asthma. Of the original cohort, 285 children had sufficient data available for inclusion in this ancillary analysis. The human subjects committee at the University of Wisconsin School of Medicine and Public Health approved the study, and informed consent was obtained from the children's parents.
The etiology and timing of specific wheezing viral respiratory illnesses during the first 6 years of life were assessed using nasal lavage, culture, and PCR-based viral diagnostics as previously described (1). Viral pathogens identified included HRVs, enteroviruses, adenoviruses, influenza, parainfluenza, coronaviruses, RSV, metapneumovirus, and bocavirus. Using these methods, a viral etiology was identified in 90% of wheezing episodes during the first 3 years of life (1).
A wheezing respiratory illness was defined as meeting one or more of the following criteria: (1) physician-diagnosed wheezing at an office visit, (2) an illness for which the child was prescribed short- or long-acting β-agonists or controller medications, or (3) an illness given the following specific diagnoses: bronchiolitis, wheezing illness, reactive airway disease, asthma, or asthma exacerbation (9).
Peripheral blood was drawn on an annual basis, and aeroallergen sensitization was assessed in serum by fluoroenzyme immunoassay (Unicap 100; Pharmacia Diagnostics, Uppsala, Sweden) at 1, 2, 3, 4, 5, and 6 years of age. Allergen-specific IgE was measured to dog, cat, Dermatophagoides farinae, Dermatophagoides pterynissonus, and Alternaria alternata at every age. At 5 and 6 years of age, additional aeroallergens tested included cockroach, ragweed, birch, and timothy grass. Any result 0.35 or greater was considered positive.
Development of viral wheezing and aeroallergen sensitization was modeled using multistate Markov models in continuous time. At any given time, each child belonged to one of the following states: State 1, nonsensitized, nonwheezing; State 2, sensitized, nonwheezing; State 3, nonsensitized, wheezing; State 4, sensitized, wheezing. At birth, each child is in State 1. Instantaneous transitions are only permitted from States 1 to 2, States 1 to 3, States 2 to 4, and States 3 to 4. Direct transitions from States 1 to 4 are not allowed because the two events would not happen at exactly the same moment (if continuously monitored). These transitions are governed by a set of transition intensities for each pair of states. Each transition intensity, the instantaneous risk or hazard of moving from one state to another, is modeled using log-linear models that incorporate age (year of life as a categorical variable) and other covariates. Age and other covariate effects are assumed to be identical for the nonwheezing to wheezing transitions (e.g., States 1 to 3 and States 2 to 4) and for the nonsensitized to sensitized transitions (e.g., States 1 to 2 and States 3 o 4). Differences in the intercepts of these log-linear models represent the effect of sensitization on wheezing and the effect of wheezing on sensitization, respectively. Estimates from these log-linear models are exponentiated and interpreted as hazard ratios. The multistate Markov model naturally incorporates the available, incomplete data on state transitions. The exact times of viral wheezing respiratory illnesses are observed, although the sensitization status at the time of viral wheezing is not necessarily known. Current sensitization status is only observed at scheduled annual visits.
The above four-state model was used to look separately at (1) sensitization and wheezing by any etiology, (2) sensitization and wheezing associated with HRV, and (3) sensitization and wheezing associated with RSV. Models are presented unadjusted and adjusted for risk factors including dog exposure, smoke exposure, older siblings, maternal allergy, paternal allergy, maternal asthma, and paternal asthma. An expanded eight-state model was used to compare transitions among sensitized, HRV wheezing, and RSV wheezing states. At any given time, each child belongs to one of the following states: State 1, nonsensitized, nonwheezing; State 2, sensitized, nonwheezing; State 3, nonsensitized, HRV wheezing only; State 4, sensitized, HRV wheezing only; State 5, nonsensitized, RSV wheezing only; State 6, sensitized, RSV wheezing only; State 7, nonsensitized, HRV and RSV wheezing; and State 8, sensitized, HRV and RSV wheezing. Instantaneous transitions are only permitted from States 1 to 2, States 1 to 3, States 1 to 5, States 2 to 4, States 2 to 6, States 3 to 4, States 3 to 7, States 4 to 8, States 5 to 6, States 5 to 7, States 6 to 8, and States 7 to 8. A two-sided P value < 0.05 was regarded as statistically significant. Analyses were conducted in the R statistical computing environment (10) using the msm software package (11).
Children in the COAST study were classified in one of four states at each yearly visit: State 1, neither sensitized to aeroallergen(s) or viral wheeze; State 2, viral wheeze only; State 3, sensitized to aeroallergen(s) only; and State 4, sensitized to aeroallergen(s) and viral wheeze (Figure 1). The absolute rates during the first 6 years of life of viral wheezing, aeroallergen sensitization, and each of the four States in the model are depicted in Figures 2A through 2C. The relative probability of moving from one state to another during the subsequent year was determined at 1 to 6 years of age (see Table E1 in the online supplement). For example, when a child was in State 1 at 1 year of age, the probability of remaining in State 1 for the next year was 0.63, and the probabilities of transitioning to States 2, 3, or 4 were 0.26, 0.07, and 0.04, respectively (Table E1).
By comparing the probability of moving from State 3 (sensitized only) to State 4 (sensitized and viral wheeze) with the probability of moving from State 1 (neither) to State 2 (viral wheeze only), one can determine whether allergic sensitization increased the risk of subsequent viral wheeze (Figure 1). Children who were sensitized to aeroallergens had greater risk of developing viral wheeze than nonsensitized children (Table 1) (hazard ratio [HR], 1.9; 95% confidence interval [CI], 1.2–3.1). The absolute risk of children moving from State 3 (sensitized) to State 4 (sensitized and viral wheeze) was greatest at 1 year of age; however, the relative risk (States 3 to 4 and States 1 to 2) was consistently increased at every age assessed (Table E1).
To assess whether the observed relationship of allergic sensitization preceding viral wheeze was virus specific, we used the four-state longitudinal model, except we replaced any viral wheeze with the two most commonly detected viral pathogens during wheezing illnesses in early life, HRV (Figures E1A–E1C) and RSV (Figures E2A–2C).
Beginning at 1 year of age and continuing throughout the first 6 years of life, allergic sensitization led to an increased risk of wheezing illnesses caused by HRV(HR, 2.3; 95% CI, 1.3–4.0) (Table 1). In contrast, allergic sensitization did not lead to a statistically significant increase in risk of RSV wheezing illnesses (HR, 1.6; 95% CI, 0.87–2.9) (Table 1). To further determine whether this relationship was virus specific, we developed an eight-state model that assessed relative risk including whether a child began to have RSV wheezing only, HRV wheezing only, or both RSV and HRV wheezing. In this model, sensitized children were at increased risk of transitioning to HRV wheezing (HR, 2.8; 95% CI, 1.5–5.1) but not RSV wheezing (HR, 0.71; 95% CI, 0.25–2.0). The difference between these ratios was statistically significant (P = 0.02).
By comparing the probability of transitioning from State 2 (viral wheeze only) to State 4 (sensitized and viral wheeze) with the probability of transitioning from State 1 (neither) to State 3 (sensitized only), we were able to determine whether viral wheeze led to a greater risk of subsequent allergic sensitization (Figure 1). Children with viral wheeze were not at increased risk of developing subsequent allergic sensitization (HR, 0.76; 95% CI, 0.50–1.1) (Table 1). In fact, there was a nonsignificant trend for viral wheeze to reduce the risk of developing allergic sensitization. This finding was not affected by viral etiology of the wheezing episode (HRV: HR, 0.69; 95% CI, 0.41–1.2; RSV: HR, 0.83; 95% CI, 0.51–1.3) (Table 1).
In this study, we longitudinally assessed the developmental relationships between allergic sensitization and virus-induced wheezing from infancy to 6 years of age, reasoning that the time sequence of these events in early life would provide information about the direction of causality. We have demonstrated that sensitization to aeroallergens, beginning in the first year of life, consistently predisposes children to viral wheezing illnesses. The reverse was not true; we found no evidence that outpatient virus-induced wheezing illnesses increased the risk of sensitization to aeroallergens in subsequent years. Finally, there was evidence of virus specificity, in that allergic sensitization specifically increased the risk of wheezing with HRV but not RSV infections.
This relationship between allergic sensitization and HRV wheezing could be explained by a common underlying susceptibility to both conditions or by a causal relationship. In support of a common underlying predisposition, low IFN-γ responses during early life have been associated with greater risk of early childhood wheeze and allergy (12–14). Furthermore, impaired innate immune responses have been associated with allergy and asthma (15, 16). However, our data regarding the directionality of the observed relationship between allergic sensitization and HRV-wheezing illness demonstrate that the former is significantly more likely to precede the development of the latter.
A number of underlying mechanisms by which allergic sensitization increases the risk of HRV-induced wheezing have been proposed. Recent work has suggested that interactions between innate and allergic inflammatory mechanisms may lead to more severe viral illnesses in allergic individuals (17). In addition, ongoing allergic inflammation in the airways may directly lead to impairment of the epithelial cell barrier and antiviral response. For example, goblet cells, commonly found in the airways of individuals with allergic asthma, were recently reported to be particularly susceptible to HRV infection (18). Finally, allergic inflammation may directly inhibit host antiviral responses. Gill and colleagues found that increased expression and cross-linking of the high-affinity IgE receptors on plasmacytoid dendritic cells impaired production of IFN-α to influenza infection in vitro (19). This counterregulation between IgE and antiviral responses could lead to increased severity of viral illnesses in allergic individuals. In support of this potential mechanism, a recent clinical trial by Busse and colleagues demonstrated that omalizumab, a monoclonal anti-IgE antibody that down-regulates high-affinity IgE receptor expression on plasmacytoid dendritic cells and other cell types (20), was effective in preventing virus-induced asthma exacerbations in a highly allergic inner-city population (21).
Our findings that early life wheezing illnesses did not lead to subsequent aeroallergen sensitization differ from the Sigurs study, which showed that severe RSV bronchiolitis was associated with increased risk of sensitization at school age (6). This may reflect differences in the population because the COAST children were nearly all outpatients, whereas Sigurs and colleagues studied children who were hospitalized with bronchiolitis. However, the rates of sensitization in the control group in the Sigurs study were low compared with the general population, and other groups have been unable to replicate these findings (22, 23).
The strengths of our study include the meticulous characterization of viral respiratory illnesses during the preschool years, with virus detection rates of 90% during wheezing illnesses (1). In addition, annual assessment of allergen-specific IgE allowed comprehensive and repeated assessments of transitions in our four-state model. We specifically chose to focus on sensitization to aeroallergens, rather than foods, because this is a stronger risk factor than food sensitization for asthma inception (1), and, from a physiologic perspective, responses to respiratory allergens would be expected to affect the risk of wheeze more than responses to food protein. A limitation of our study is that allergic sensitization was not assessed until 1 year of age, whereas wheezing episodes were assessed continuously from birth. However, only about 10% of the cohort developed allergic sensitization by 1 year of age, suggesting that assessment of sensitization earlier during infancy would not have provided much additional information. Another potential limitation of our study is that COAST is a high-risk cohort; however, about one half of the cohort did not develop allergic sensitization by 6 years of age, providing a large nonallergic population for comparison.
In summary, using a longitudinal model in a well characterized birth cohort, we have clearly demonstrated that allergic sensitization increases the risk for HRV wheezing and that the converse is not true. When considered together with plausible mechanisms for allergic sensitization to inhibit antiviral responses, this sequential relationship supports a causal role for allergic sensitization in this developmental pathway. To definitively prove causality, one would need to prevent or modify allergic sensitization in early life and demonstrate lower risk of subsequent HRV wheezing illnesses. Allergen avoidance strategies have not led to reductions in sensitization (24); therefore, novel strategies aimed at the prevention of allergic sensitization are needed. This is an important goal given the close relationship between HRV wheezing illnesses in early life and the subsequent development of asthma (1, 25).
Supported by NIH grants R01 HL61879, P01 HL70831, M01 RR03186, and 1 UL1RR025011 from the Clinical and Translational Science Award (CTSA) program of the National Center for Research Resources (NCRR).
Author Contributions: Conception and design, D.J., M.E., R.G., J.G., R.L.; analysis and interpretation, D.J., M.E., R.G., C.T., T.P., W.L., J.G., R.L.; drafting the manuscript for important intellectual content, D.J., M.E., R.G., J.G., R.L.
This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org
Originally Published in Press as DOI: 10.1164/rccm.201104-0660OC on September 29, 2011