TY - JOUR T1 - Multivariate cluster analysis of immunopathological and clinical data from the U-BIOPRED severe asthma study: Preliminary results JF - European Respiratory Journal JO - Eur Respir J DO - 10.1183/13993003.congress-2015.PA871 VL - 46 IS - suppl 59 SP - PA871 AU - Susan Wilson AU - Borislav Dimitrov AU - Ratko Djukanovic AU - Jonathan Ward AU - Ana Sousa AU - Julie Corfield AU - Peter Sterk AU - K. Fan Chung AU - Brabro Dahlen AU - Pascal Chanez AU - Dominick Shaw AU - Norbert Krug AU - Thomas Sandstrom AU - Peter Howarth AU - UBIOPRED Consortium Y1 - 2015/09/01 UR - http://erj.ersjournals.com/content/46/suppl_59/PA871.abstract N2 - The U-BIOPRED Consortium aims to phenotype severe asthma. We have used the method of unbiased multivariate principal component analysis (PCA) by varimax rotation with Kaiser normalisation to describe the variance and possible relationships in our immunopathology and clinical data collected from a subgroup of adult subjects following bronchoscopy. Complete data for 33 variables (59 subjects) were analysed with at least 80% explained total variance as a criterion to identify the first, most important components (clusters of variables). The following 14 principal components, with *core variables, are shown in the figure. When added to the above 33 variables in a semi-biased approach, the cohort descriptor (healthy, mild/moderate, severe) clustered within the second component, blood neutrophils. Based on the 14 principal components scores, we are further going to explore the grouping of the subjects by agglomeration with hierarchical analysis and K-means clustering. This will allow an identification of several, better defined sub-cohorts (phenotypes) according to the newly-described clinico-immunopathological clusters. Funded by IMI. ER -