Periostin levels and eosinophilic inflammation in poorly-controlled asthma

Asthma is a chronic heterogeneous inflammatory disorder of the airways where approximately 50 % of adults have eosinophilic airway inflammation which is referred to as eosinophilic asthma. Sputum eosinophils are an effective monitoring tool in asthma when used in therapeutic decision making and there is good evidence that this approach can significantly reduce severe exacerbations [1, 2]. The technical difficulties associated with sputum collection and processing limit the use of this management strategy in clinical practice. Alternative and accurate biomarkers that can predict sputum eosinophils are needed to facilitate inflammation-based asthma management.

We have recently shown that blood eosinophil counts are a suitable surrogate for sputum eosinophil proportion in identifying patients with eosinophilic asthma, being simple and readily available and thus may provide a suitable asthma management tool [3]. Similarly, reports also show that serum periostin levels are associated with airway eosinophilia in asthma [4]. In bronchial brushings, periostin gene expression is upregulated in a subset of patients in which it was associated with the expression of Th2 cytokines. These patients also had higher blood and BAL eosinophil counts and were most responsive to inhaled corticosteroids (ICS) [5]. Periostin is expressed by lung epithelial cells and fibroblasts and in-vitro experiments show that periostin is secreted towards the basolateral side of the airway [6, 7]. However little is known about the airway levels of periostin and the relationships between airway periostin levels, asthma inflammatory subtypes and asthma control.

This study sought to examine the ability of serum periostin to predict the presence of eosinophilic asthma and to determine the relationship between airway and serum periostin levels, inflammatory subtype and asthma control in a group of adults with poorly-controlled asthma.

Participants consisted of 83 adults with a mean age of 61 years who were predominantly atopic (73 %) and poorly-controlled (mean ACQ6 1.83) on treatment (Table 1). A sputum cell count was available for 82/83 participants to assign inflammatory phenotype. Thirty-seven participants had eosinophilic asthma, defined as sputum eosinophils ≥3 % and 45 had non-eosinophilic asthma. Most participants with non-eosinophilic asthma (n = 28) had a paucigranulocytic phenotype (34 %); 17 had neutrophilic phenotype (21 %) and five participants had a mixed eosinophilic and neutrophilic phenotype (6 %).

In this well-characterised cohort with poorly-controlled asthma, periostin was detected in 78 % of sputum samples at a 1:2 dilution. While levels of periostin in induced sputum were lower than those measured in serum, both sputum and serum periostin levels were significantly associated with airway eosinophilia. Additionally, both sputum and serum periostin levels were significantly higher in patients with eosinophilic asthma compared with non-eosinophilic asthma. Our data are supported by similar correlations between serum periostin and sputum eosinophils in other cohorts [4, 14].

There may also be a biological explanation for the observed lower levels of periostin in sputum. It has been shown that periostin is strongly expressed at the RNA level in epithelial cells but protein expression could not be detected in these cells [6, 7]. However, periostin is immunolocalized to the basement membrane immediately below the epithelial cells as well as mesenchymal cells [7]. This discrepancy between RNA and protein expression has led to the postulation that periostin protein is rapidly secreted by airway epithelial cells into the subepithelial layer. In vitro tissue culture experiments using an air/liquid interface, representing the airway lumen and bronchial tissue respectively, support this hypothesis, as periostin was abundantly present in the basal medium, but could not be detected in the apical washes [7]. It is therefore feasible that periostin may not be present in high concentrations in the sputum supernatant collected from the airway lumen. Alternately, the low sputum periostin levels may represent plasma leakage or the presence of periostin in the sputum supernatant may indicate a sub-group of patients who have a specific alteration in the epithelial barrier in some way or suggest a different cellular source of periostin in the sputum samples.

The utility of identifying eosinophilic asthma is well established with improved management and reduced exacerbations as key outcomes [1, 15]. However, sputum induction is not widely available and is unlikely to be adopted into routine clinical practice. As such, there is an unmet need for surrogate biomarkers to assist clinical decision making in asthma. Despite a statistically significant correlation between both serum and sputum periostin and the sputum eosinophil proportions, the strength of the association was relatively modest such that neither sputum nor serum periostin levels could sufficiently predict the presence of an eosinophilic inflammatory subtype with AUC in the poor to fair range [16]. We then compared blood eosinophils and serum periostin levels as surrogates to identify patients with eosinophilic asthma as identified by sputum eosinophils ≥3 %. In agreement with Wagener et al [17], we found blood eosinophils were an accurate predictor of patients with eosinophilic asthma (AUC 0.829) and serum periostin did not adequately distinguish eosinophilic from non-eosinophilic asthma in adults. Sputum periostin similarly lacked the rigor to be an adequate biomarker to detect eosinophilic asthma in adults with poorly-controlled asthma being neither sensitive nor specific and has the added disadvantage of not being easily accessible in clinical practice. In the BOBCAT study, serum periostin measured using the ELISA platform was found to be a sensitive but not specific marker of airway eosinophilia and performed better than blood eosinophils at predicting the presence of airway eosinophilia as defined by eosinophils in both sputum and tissue [4]. Of course, all studies so far including the current study reported here may be limited by the numbers of participants included with varying patient characteristics, levels of treatment and severity of asthma, and larger studies may provide more insight into the ability of serum periostin to predict the presence of eosinophilic asthma.

There are limited studies supporting the utility of periostin in the management of asthma. There is evidence that periostin levels can indicate better treatment outcomes in studies involving the IL-13 monoclonal antibody lebrikizumab or anti-IgE antobody omalizumab [8, 13, 18]. In these studies blood eosinophils and fractional exhaled nitric oxide (FeNO) were also able to identify people with greater benefit and future clinical studies need to determine how each of these biomarkers will be used in clinical practice. While periostin levels may not be suitable as surrogate markers for detecting the presence of eosinophilic asthma, its biomarker utility may lay in its ability to predict treatment response to anti-IL-13 therapy [13, 19], although this has yet to be tested in prospective studies.

In poorly-controlled asthma treated with inhaled corticosteroids, periostin levels are low in airway secretions, but can be detected. Although sputum and serum periostin levels are significantly related to sputum eosinophil proportions, their ability to predict eosinophilic asthma in poorly-controlled asthma appears relatively modest.