Integrated molecular networks reveal latent components distinguishing type 2-high and type 2-low asthma

Background Type 2 inflammation (T2) biomarkers are used to identify T2-low and T2-high endotypes of asthma in the clinic and understanding of these endotypes has helped to improve clinical management of patients. The T2-low asthma endotype remains relatively poorly characterised and these patients have few treatment options. Despite its clinical relevance, this endotype lacks specific biomarkers. Objective To identify transcriptional and microRNA signatures associated with asthma endotypes. Methods We studied 88 participants with asthma stratified into T2-high and T2-low asthma by blood eosinophils and/or fractional exhaled nitric oxide. We defined whole-blood microRNA and mRNA profiles for each participant and used Data Integration Analysis for Biomarker discovery using Latent Variable approaches for Omics integration to identify latent components (LCs). The results were validated in two independent cohorts of subjects with asthma, including patient samples taken before and after treatment with benralizumab. Results The T2-high endotype was the most prevalent (67%, n=55) within the study population. We identified two LCs associated with T2-high and T2-low asthma. LC1 was characterised by eosinophil-derived mRNAs and included multiple T2-low microRNAs. LC2 was associated with both endotypes and non-eosinophilic cell types. Pathway analysis revealed increased representation of taste and smooth muscle pathways in the T2-low endotype. The T2-low microRNA miR-574-3p was enriched in LC1. Furthermore, we found that the LC signatures were present in two independent cohorts and modulated by benralizumab. Conclusions T2-high and T2-low asthma are distinguished by unique transcriptomic networks and have minimal microRNA overlaps. T2-low asthma was enriched for pathways involved with smooth muscle contraction. The LCs signatures are also modulated by benralizumab.