Open AccessPhenotyping asthma using an unsupervised prediction model based on blood granulocyte responsiveness
Author(s) -
Hilvering Bart,
Vijverberg Susanne,
Jansen Jeroen,
Houben Leo,
Schweizer Rene,
Lammers JanWillem,
Koenderman Leo
Publication year - 2015
Publication title -
clinical and translational allergy
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.979
H-Index - 37
ISSN - 2045-7022
DOI - 10.1186/2045-7022-5-s2-o2
Subject(s) - medicine , asthma , sputum , exhaled nitric oxide , eosinophil , immunology , eosinophilic , pathology , spirometry , tuberculosis
Method Clinical parameters, activation of blood granulocytes and sputum phenotypes were assessed in 115 adult asthma patients (NCT01611012). Blood granulocytes were stained with antibodies against the active FcgRII receptor (CD32, clones A17/A27) and a-chain of MAC-1 (CD11b) in the presence or absence of fMLF, and analysed by flow cytometry. NLPCA was used to reduce dimensions in a combined cellular and clinical dataset, followed by discriminant analysis to generate a prediction model. Phenotypes identified by the model were cross-validated by inflammatory sputum phenotypes based on sputum induction (reference test). Results Peripheral blood eosinophil count, FeNO (Fraction of Exhaled Nitric Oxide), ACQ (Asthma Control Questionnaire), medication use, nasal polyposis, aspirin sensitivity and neutrophil/eosinophil responsiveness upon stimulation, are important parameters to differentiate between eosinophilic and non-eosinophilic phenotypes. The combinations of these parameters lead to an accurate prediction model for eosinophilic asthma with 90.5% sensitivity and 91.5% specificity.