Background
Acute bronchiolitis is a common lower respiratory tract infection in infants, often caused by respiratory viruses, and accounts for up to 15–17% of all hospitalizations in infants under 2 years [
1]. Respiratory syncytial virus (RSV) is the most common cause of bronchiolitis, although other viruses, mainly rhinovirus (HRV) are also frequently identified in these infants. RSV and HRV account for 60–80% of bronchiolitis in infants [
2].
It is well known that severe bronchiolitis is often associated with subsequent respiratory morbidity and up to 30–40% of infants hospitalized for bronchiolitis will develop recurrent wheezing or asthma later in life [
3‐
6]. However, not all children hospitalized with bronchiolitis will develop asthma. The mechanisms underlying asthma following bronchiolitis hospitalization are complex and immune responses to respiratory viruses may underlie both, bronchiolitis severity and long-term sequela such as asthma [
7]. However, the underlying immune mechanisms of chronic respiratory illness development after acute bronchiolitis remain unclear. Several cytokines have been involved in Th2-asthma, among them, the alarmins, initiators of T2 inflammation: IL-33, IL-25, and thymic stromal lymphopoietin (TSLP). Alarmins promote the expression of IL-4, IL-5, and IL-13 and increase the levels of periostin [
8].
TSLP, an epithelial cell-derived cytokine, is synthesized in response to various stimuli such as RSV infections and is considered a master regulator of type 2 immune response in the respiratory tract, that links the innate and adaptive immune responses by activating innate lymphoid cells (ILC2), as well as inducing Th2-type T cell differentiation [
9,
10].
Periostin is a matricellular protein produced in response to inflammatory stimuli mediated by IL-4, IL-5, and IL-3 by many cells, including epithelial cells and fibroblasts. There is evidence that periostin modulates upper respiratory tract inflammation and remodeling, can induce the differentiation of fibroblasts into myofibroblasts and increase fibrosis, can influence epithelial remodeling, and can change the underlying matrix by modifying the deposition of collagen fibrils [
11].
There is growing evidence that TSLP and periostin are elicited in the upper airways of infants with RSV and HRV bronchiolitis, and increased TSLP levels are related to more severe disease and intensive care unit (ICU) admission [
12]. Lee et al. [
13] reported that viral antigen recognition triggers a signalling cascade that results in TSLP production and strong T2 response that seems to play a key role in the pathogenesis of asthma. Indeed, the Th1/Th2 imbalance has been proposed as a key event in the inflammatory process after severe bronchiolitis that could predispose to recurrent wheezing and asthma [
14]. It has been also proposed that several T2 cytokines such as TSLP or periostin could be used as prognosis biomarkers for the development of asthma [
15].
The main goal of this study was to evaluate whether there is some association between the nasal detection and the levels of TSLP and periostin, in infants admitted for bronchiolitis, and the subsequent development of recurrent wheezing and asthma at 4 years of age. Other secondary outcomes, also related with the association of nasal TSLP and periostin with the long-term respiratory morbidity, were evaluated, mainly need of chronic asthma treatment and respiratory admissions during the follow-up period.
Methods
Study design
This was an observational, longitudinal, post-bronchiolitis, hospital-based, follow-up study, which is a part of an ongoing prospective investigation of respiratory tract infections in children, approved by the Medical Ethics Committee. Written informed consent was obtained from all the parents/caregivers after a full explanation of the study protocol. All methods were carried out in accordance with relevant guidelines and regulations.
Clinical assessment
Children hospitalized for their first episode of acute bronchiolitis in the Severo Ochoa University Hospital (Spain), between October 2013 and July 2017, currently aged 4 years, included in a previous study to investigate whether infants exhibit enhanced nasal airway secretion of TSLP, IL-33, and periostin during natural respiratory viral bronchiolitis, were included [
12]. Parents were contacted by telephone, and were invited to a clinical interview based on a structured questionnaire to obtain information on wheezing episodes; bronchodilator and oral corticosteroid prescription; related hospital admissions; chronic asthma treatment; physician-diagnosed atopic dermatitis; allergic rhinitis; food allergy; pet contacts; daycare attendance; parental smoking habits; allergy; eczema and asthma in first order family members diagnosed by a medical doctor. The ISAAC questionnaire for asthma symptoms for 6–7-year-old children, previously validated and translated to Spanish was also employed [
16]. To minimize recall bias, data reported by parents were confirmed by reviewing electronic medical records from both, hospital and primary care.
Current asthma prevalence was estimated by the proportion of patients who responded positively to question number 2 of the ISAAC questionnaire (wheezing or whistling in the chest in the last 12 months), the one which has demonstrated the greatest correlation with current asthma prevalence in validation studies [
16]. The prescription of chronic asthma treatment and the need for respiratory admissions during the follow-up were considered and evaluated as indicators of asthma severity.
Recurrent wheezing was defined as the presence of wheezing diagnosed by a doctor in the first 4 years of life [
17].
The classic criteria, an initial episode of acute onset expiratory dyspnoea with previous signs of viral respiratory infection—whether this was associated with respiratory distress or pneumonia—were applied in diagnosing
bronchiolitis [
18].
Virus detection
Two nasopharyngeal samples (NPA) were obtained at admission for bronchiolitis by a standard, routine technique, consisting of gently washing the nasal cavity with 1 ml of phosphate buffered saline in each nostril and collection into a standard mucus extractor. The samples were refrigerated at 4 °C until being processed within 24 h of collection. One of the two samples was processed in the Respiratory Virus and Influenza Unit at the National Microbiology Centre (ISCIII, Madrid, Spain). Detection of respiratory virus was performed by 3 independent multiplex reverse transcription-polymerase chain reaction (RT-PCR) assays. The first assay detected Influenza A, B, and C viruses; the second was used to detect parainfluenza viruses (PIV) 1 to 4, HRV, and enteroviruses; and the third assay detected the presence of RSV types A and B, human metapneumovirus (HMPV), human bocavirus (HBoV), and human adenoviruses (ADV). These 3 assays were real-time multiplex RT-PCRs and used the SuperScript™ III Platinum® One-Step Quantitative RT-PCR System (Invitrogen). The other sample was used for immunological testing at the Immunology Department of IIS-Fundación Jiménez Díaz as described below.
A recent study by Lopez-Guisa et al. [
19] demonstrated a good correlation between bronchial and nasal epithelial expression of pro-remodelling factors. NPA is a non-invasive method, especially useful in infants and young children.
Detection of cytokines and proteins in nasal secretions
Nasopharyngeal aspirate processing
Previous to NPA filtrations with a 40-µm nylon filter, NPAs were centrifuged and cellular pellet and supernatant were obtained. Supernatants were directly frozen at – 80 °C.
Immunological analyses in nasopharyngeal aspirate
In NPA supernatant TSLP and periostin were analysed by ELISA Kit (R&D Systems, Abingdon, UK), according to the manufacturer’s instructions using provided standards and quality controls. The intra-assay and inter-assay coefficients of variation were: TSLP: 8.2% and 7.47%, respectively, and periostin: 2.19% and 9.99%, respectively. The lower detection limit of these assays was 32.5 pg/ml for TSLP and 62.5 pg/ml for periostin.
Statistical analysis
Values were expressed as percentages for discrete variables, or as mean and standard deviation or median and interquartile range for continuous variables. Comparisons used either X2 or Fisher exact test (2-tailed) for categorical variables and Student T-test, Mann–Whitney U test, Kruskal–Wallis test, and analysis of variance (ANOVA) for continuous variables. To control for potentially confounding variables (maternal/paternal/siblings’ asthma and atopy, atopic dermatitis, prematurity, viral identification during acute bronchiolitis, and cigarette smoke exposure) and to examine the independent association between nasal TSLP and periostin and the likelihood of developing asthma, a backward stepwise binomial logistic regression model was built. All the variables with p-value < 0.1 were introduced in the multi-variate analysis. Adjusted odds ratios (OR) with 95% confidence intervals were calculated. A probability of < 0.05 was considered statistically significant. All analyses were performed using the Statistical Package for the Social Sciences (SPSS), Version 23.0.
Discussion
Our results showed, for the first time, that infants admitted for bronchiolitis, who later developed asthma, with need of chronic treatment, by the age of 4, were significantly more likely to have nasal TSLP detection at admission. Indeed, regardless of their atopic status or the viral etiology, infants with nasal TSLP production were more likely to receive maintenance asthma treatment, including the combination of montelukast plus IGC, usually indicated for higher severity levels of asthma. Children with TSLP detection also tended to require more hospital admissions for recurrent wheezing. According to our data, and those obtained from experimental studies, TSLP seems to play a key role in asthma inception after respiratory viral infections. Han et al. [
20] showed, in experimentally RSV-infected mice, that the administration of anti-TSLP antibodies before neonatal RSV infection significantly attenuated airway response to inhaled methacholine and reduced eosinophil numbers in the bronchoalveolar fluid. They postulated that neonatal RSV infection initiates a cascade that involves increased TSLP release from infected epithelial cells, which induces the upregulation of OX40 ligand (OX40L) expressed on lung dendritic cells, which may be critical for promoting the initial differentiation and expanding existence of Th2 cells and regulatory T cells (Tregs) [
21]. This, in turn, initiates the polarization of RSV-specific T cells to a T2 phenotype so that re-exposure to RSV triggers the expansion of RSV-specific T2 memory cells and the enhanced development of airway hyperresponsiveness (AHR), accompanied by eosinophilic airway inflammation, mucus hyperproduction, and IL-13 release. More recently, Fan et al. [
22] provided the first direct evidence that RSV non-structural protein (NS)1 breaks immune tolerance and induces airway inflammation and AHR in infected mice. Animal studies have also demonstrated that TSLP is necessary and sufficient for the development of T2 cytokine-associated airway inflammation, mediated through distinct immune cell cascades in the context of innate and adaptive T2 inflammation [
23]. Salka et al. [
24] recently reported that human infant airway epithelial cells respond to a virus mimic (double-stranded RNA) with robust production of TSLP, and in vivo, they also found that infants with higher TSLP nasal levels at admission for respiratory infections (not necessarily bronchiolitis), had an increased probability of respiratory hospitalizations or emergency room visits 12 months after discharge, suggesting a role of TSLP secretion during severe bronchiolitis in infancy and in asthma inception later in life. However, Chen et al. [
25] did not find any association between several serum cytokine levels, including TSLP, in infants hospitalized for bronchiolitis and the frequency of recurrent wheezing episodes in a 2-year follow-up study. The authors explain their unexpected results to the fact that they tested serum samples rather than nasopharynx aspirates.
According to our results, the association between TSLP and asthma development seems to be independent of the atopic status. In line with our data, Vrsalovic et al. [
26], recently described higher serum concentration of TSLP in asthmatics than in healthy children, but without any difference among the three different asthma phenotypes: allergic asthma, virus-induced asthma, and non-allergic asthma. Lin et al. [
27] also reported higher levels of TSLP receptors in asthmatic patients than in healthy children, but similar concentrations between allergic and non-allergic asthmatic patients.
However, despite the strong association of TSLP with asthma, some studies have failed to find a consistent association between circulating TSLP and asthma development. The study conducted in the birth cohort from the Urban Environment and Childhood Asthma (URECA) found that the early presence of circulating TSLP was significantly associated with reduced incidence of recurrent wheeze in those children not sensitized to aeroallergen [
28]. These differences could be explained by several factors. Firstly, there is a great deal of variation in the methodology among the different studies. Thus, while Chen et al [
25] study included, as ours, only infants admitted with bronchiolitis, some authors [
24] recruited infants less than 24 months hospitalized for a PCR-confirmed viral respiratory infection, regardless of whether it was bronchiolitis or a recurrent wheezing episode. Other studies recruited asthmatic children with a wide range of age [
26,
27], whereas others followed up a cohort of newborns. In addition, most studies evaluate serum TSLP levels [
25‐
28] that could be less reliable than nasal TSLP to express local TSLP production [
24]. Also, genetic variation may have an impact beyond circulating TSLP protein expression, including on other genes or pathways as Biagnini Myers et al. [
29] demonstrated. Murrison et al. [
30] recently found that 90% of children with some defined TSLP risk genotypes and high nasal TSLP mRNA expression, had asthma compared with 40% of children without risk genotypes and with low nasal TSLP expression, finding no association between serum TSLP and asthma. These data suggest that childhood asthma may be modified by the combined effect of TLSP genotype and TSLP expression in the nasal epithelium and both factors should be taken into account when evaluating asthma risk.
In contrast to TSLP, the nasal detection of periostin at admission for bronchiolitis was associated, in our series, with a more favourable respiratory outcome. In fact, higher nasal periostin levels were significantly related to lower frequency of current asthma at age 4, ever diagnosis of asthma, maintenance asthma treatment prescription, and admissions for recurrent wheezing. Periostin is a distinct signature protein for the T2-high asthma phenotype in adults [
31,
32] although its role in children is controversial. Some studies have found higher levels of serum periostin in children with asthma [
19,
33‐
35] and some of them even reported a significant correlation between serum periostin levels and asthma severity [
34]. However, other studies found similar serum periostin levels in children with severe asthma compared to those with controlled asthma [
36,
37] or even lower periostin levels in children with severe uncontrolled asthma than in children with controlled asthma [
38].
Regarding recurrent wheezing in preschool children, there are also inconclusive data about the association between serum periostin level and airway inflammation. Yooma et al. [
39], in 2–5-year-old children, reported higher serum periostin levels in children with recurrent wheezing and in those who developed acute wheezing exacerbation in the subsequent year compared to healthy control children. Anderson et al. [
40], in the Childhood Origins of Asthma (COAST) cohort study, demonstrated that, in children with atopic risk, a high periostin level at age 2 years was associated with a greater risk of asthma at age 6. However, a recent study by Guvenir et al. [
41] evaluated the usefulness of serum periostin in wheezy preschool children for predicting the development of asthma in school ages, founding no difference in the levels of periostin between children with transient wheezing and children with asthma in both, preschool and school periods.
In relation to periostin detection in infants with acute bronchiolitis, our group, in a previous study, demonstrated for the first time that naturally occurring severe infections by the most common respiratory viruses, in hospitalized infants with bronchiolitis, induces nasal airway secretion of periostin when compared with healthy controls [
12]. Regarding the association between periostin detection in infants with bronchiolitis and asthma development, Nanishi et al. [
42] recently published a multicenter cohort study of infants with severe bronchiolitis, measured the serum periostin level at hospitalization, and grouped infants into three groups: low, intermediate, and high levels. They examined the association of periostin levels at entry with the development of asthma at 6 years of age. After adjusting for confounding factors, they found that, compared to the low periostin group, the asthma risk was significantly higher among infants in the intermediate group but non-significantly greater in the high-level group. After the stratified analysis, infants without IgE sensitization or parental asthma or eczema showed no significant periostin-outcome association, suggesting that high and moderate serum levels of periostin are associated with increased risk of asthma by age 6 years only among infants with severe bronchiolitis and allergic predisposition. The causal mediation analysis performed in that study demonstrated that there was no indirect (mediation) effect of periostin, suggesting that the effect of IgE sensitization on developing asthma was driven through pathways other than periostin. Our follow-up study, with a similar design to Nanishi’s et al [
42], although conducted at a single center, with smaller sample size and with nasal rather than serum periostin detection, found, on the contrary, an inverse association between periostin detection and the development of asthma by age 4 years, independently of atopic risk factors such as atopic dermatitis or family history of asthma or atopy. Previous experimental studies found that periostin decreases allergic airway inflammation in mice. Gordon et al. [
43] demonstrated that, compared with wild-type controls, periostin deficient mice developed increased AHR and serum IgE levels following allergen challenge. They speculated that periostin’s role in the airway is to act as a brake on allergen-induced IgE production and AHR. The mechanism of this effect could be explained by periostin’s regulation of the TGF-beta signaling pathway and the anti-inflammatory effects of TGF-beta-induced T regulatory cell differentiation. The study by Kondoh et al. [
44] suggests that periostin strengthens the extracellular matrix structure of the intact alveolar wall and acts protectively during acute lung injury in mice. In view of the contradictory data and the limited evidence available, it seems that the role of periostin in the development of asthma in children with a history of severe bronchiolitis is far from being clarified.
Our study has several potential limitations. Only 27 infants had detectable nasal TSLP levels. We followed-up our patients up to the age of 4 years and as asthma can develop later in life, some of them could be misclassified now. That is why we have extended the follow-up until the age of 10 years.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.