Identification of distinct immunophenotypes in chronic pulmonary aspergillosis using cluster analysis.

Abstract

Whether chronic pulmonary aspergillosis (CPA) has different immunophenotypes remains unknown. To identify different CPA immunophenotypes using cluster analysis. We used a subject-centred multivariate clustering approach without prior assumptions to identify CPA phenotypes. We retrospectively included the data of treatment-naïve subjects with CPA and excluded subjects with asthma and allergic bronchopulmonary aspergillosis (ABPA). We performed a scalable two-step cluster analysis using the log-likelihood distance measures to identify CPA phenotypes based on the blood immunological profile (total IgE, eosinophil count and Aspergillus-specific IgE and IgG). We included 351 CPA subjects and found two clusters. Cluster 2 (n=118) had significantly higher serum total IgE, peripheral blood eosinophil count, and serum A. fumigatus-specific IgE and IgG than cluster 1 (n=233). Cluster 2 subjects had a lower FEV1:FVC ratio on spirometry and were more likely to have a fungal ball (88 [74.6%] vs. 145 (62.2%), p=.023) on the CT thorax than cluster 1. After treatment discontinuation, cluster 2 had a longer median (interquartile range) time to relapse than cluster 1 (11.5 [7.3-27.4] vs. 4 [1.1-8.9]months, p=.005). We identified two distinct CPA phenotypes, type-2 dominant and non-type-2, with different clinical and radiological findings and treatment outcomes. Future studies should confirm our findings and investigate different treatment strategies based on CPA phenotypes.