Confounding Patient Factors Affecting the Proper Interpretation of the Periostin Level as a Biomarker in Asthma Development

Abstract

Introduction: The proper use of serum periostin (POSTN) as a biomarker for asthma is hindered by inconsistent performance in different clinical settings.

Objective: To explore patient's factors that may affect POSTN expression locally and systematically and its utility as a biomarker for asthma development.

Materials and methods: Here we used bioinformatics analysis of publicly available transcriptomics data to confirm that POSTN is an asthma specific gene involved in core signaling pathways enriched in the bronchial epithelium during asthma. We then explored a large number of datasets to identify possible confounders that may affect the POSTN gene expression and consequently, its interpretation as a reliable biomarker for asthma. Plasma and saliva levels of POSTN were determined in locally recruited asthmatic patients (mild, moderate and severe) compared to healthy controls to confirm the bioinformatics findings.

Results: Our bioinformatics results confirmed that POSTN was consistently upregulated in the bronchial epithelium in asthma, chronic obstructive pulmonary disease (COPD) and idiopathic pulmonary fibrosis (IPF) bronchial epithelium. In asthma, its mRNA expression was affected by gender, sample anatomical site and type, steroid therapy, and smoking. In our cohort, plasma POSTN was upregulated in severe and non-severe asthmatic patients. Saliva POSTN was significantly higher in non-severe asthmatic patients compared to healthy and severe asthmatic patients (specifically those who are not on Xolair (omalizumab)). Patients' BMI, inhaled steroid use and Xolair treatment affected POSTN plasma levels.

Conclusion: Up to our knowledge, this is the first study examining the level of POSTN in the saliva of asthmatic patients. Both plasma and saliva POSTN levels can aid in early diagnosis of asthma. Saliva POSTN level was more sensitive than plasma POSTN in differentiating between severe and non-severe asthmatics. Patients' characteristics like BMI, the use of inhaled steroids, or Xolair treatment should be carefully reviewed before any meaningful interpretation of POSTN level in clinical practice.

Keywords: biomarkers; periostin; transcriptomics.

Figure 1

Genes that are differentially expressed in asthma compared to healthy controls in the three datasets GSE64913 (12 genes), GSE4302 (11 genes) and GSE13396 (7 genes). POSTN and IGF2BP3 were consistently differentially expressed among the three transcriptomic datasets.

Figure 2

Degree of relatedness of a gene to another gene in the meaning of annotation profiles of the identified genes as a group to each other using eXploring Genomic Relations for enhanced interpretation (http://galahad.well.ox.ac.uk:3024/). POSTN showed a high degree of similarity with 14 out of the 49 identified DEG.

Figure 3

Top pathways shared by the DEGs in at least 2 of the 3 selected datasets. The graph was generated using metascape (http://metascape.org). The DEG are enriched in pathways related to response to wounding, bacteria, and regulation of cytokine secretion.

Figure 4

Raw mRNA expression of POSTN in the extracted 10 datasets to identify confounding variables that affect POSTN gene mRNA expression in healthy “H” and asthmatic “SAS”. (A, B, C) Gender (female “F” vs male “M”) in central “C” vs peripheral “P” airways epithelium “E” (GSE64913), (D) Nasal vs Bronchial (GSE4302, GSE64913, GSE13396 and GSE41861), (E) Different asthma phenotypes in nasal samples (GSE41862), (F) Flovent inhaler use and smoking (GSE4302), (G) Rhinovirus infection (GSE13396), (H) COPD and smoking (GSE30063) and (I) IPF (GSE21369). The Kruskal–Wallis with corrected Dunn’s post hoc nonparametric test was performed where p<0.05 was considered significant.

Figure 5

ELISA Quantification of Plasma and Saliva POSTN protein level using a locally recruited cohort of asthmatics patients compared to healthy controls. Plasma (A-B) and Saliva (C-D) POSTN protein levels in healthy controls versus nonsevere asthmatics (mild and moderate) and severe asthmatics (on Xolair and without Xolair). The Kruskal–Wallis with uncorrected Dunn’s post hoc nonparametric test was performed where p<0.05 was considered significant.

Figure 6

ELISA Quantification of Plasma and Saliva POSTN protein level using ELISA in a locally recruited cohort of asthmatics patients compared to healthy controls. (A-B) POSTN level in patients using Symbicort compared to those who are not (C-D) plasma and saliva POSTN level in different categories of participants according to their BMI. The Kruskal–Wallis with uncorrected Dunn’s post hoc nonparametric test was performed where p<0.05 was considered significant.