Fig. 1.

Multivariate OPLS model of histological response to topical corticosteroids. The outcome Y-variable (blue) was set as a histological response of < 15 peak eosinophil counts/HPF in the esophagus after treatment. Input X-variables with discriminatory power are shown as bars and consisted of clinical data, immune parameters and questionnaire scores, collected before (A) and after treatment (B). Immune parameters (green) are presented as percent (P) blood eosinophils expressing a particular molecule and its median fluorescence intensity (M), and absolute numbers (N) of leukocyte subpopulations in the blood. Scores from the EORTC QLQ-OES18 (purple) and Watson dysphagia scale (red) questionnaires and clinical data (yellow) are shown. Δ is the difference between values before and after treatment for each parameter. The X-variables indicated by bars in the same direction as the output Y-variable were positively associated, and the variables indicated by bars in the opposite direction were negatively associated with being a responder. The model had an explanatory power (R²Y) of 50% and a stability (Q²) of 27%