Clinical features and later prognosis of replicable early-life wheeze clusters from two birth cohorts 12 years apart.

Abstract

Clustering techniques can define the heterogeneity of asthma and wheezing. Defining early-life wheezing clusters and associated asthma risk could potentially inform patient management strategies. Clustering models that yield replicable cluster groups will have greater validity and clinical utility. This study sought to identify early-life wheezing clusters that are translatable into clinical practice and assess their stability over time in two whole-population birth cohorts established a decade apart from the same geographical location. Nonparametric K-means cluster analysis was performed separately on two birth cohorts from the Isle of Wight, UK; the Isle of Wight Birth Cohort (IOWBC) and Food Allergy and Intolerance Research Cohort (FAIR), using clinically defining variables in wheezing subjects in the first 3-4 years. Associations of resulting clusters with potential early-life risk factors and 10-year asthma outcomes were further assessed. Five clusters were identified in both cohorts: (1) infantile-onset-transient-non-atopic-wheeze, (2) infantile-onset-persistent-non-atopic-wheeze, (3) infantile-onset-atopic-wheeze, (4) early-childhood-onset-non-atopic-wheeze, and (5) early-childhood-onset-atopic-wheeze. Two atopic wheezing clusters (3 and 5) were associated with greatest early-life wheeze frequency, highest wheeze persistence, and asthma prevalence at 10 years. Cluster 1 was commonest but had lowest early-life wheeze frequency and asthma prevalence at 10 years. Cluster 2, characterized by limited atopy but recurrent infantile respiratory infections and ongoing early-life wheezing, had high 10-year asthma prevalence only in IOWBC. Early-life wheeze comprises several disease clusters (two more severe and three mild-moderate) with differing relationships to later childhood asthma, which can be replicated over time supporting their potential validity and clinical utility.