Figure 3

Classification of treatment response based on pleiotropic or disease-specific genes. (A) Glucocorticoid (GC) treatment of CD4⁺ T cells from patients with seasonal allergic rhinitis (SAR) had the largest effect on the expression of genes that participated in many disease modules. This effect was not observed following natazulimab treatment of CD4⁺ T cells from patients with multiple sclerosis (MS). The figure shows the correlation between the mean treatment ± standard error of the mean effect on mRNA expression measured by the squared student t-values between GC-treated and untreated cells and the number of disease modules a gene participated in. PCC, Pearson's correlation coefficient. (B) Pleiotropic or disease-specific genes accurately classified high and low responders to treatment in SAR and MS. The estimated probabilities (cross-validated) of a sample being a high responder (HR) based on the LASSO classifiers after drug treatment. The horizontal black line at 0.5 represent the classification border of HR and low responders (LR). The probability estimates of each group of patients are summarized into box-plots showing the median, inner quartile range, whiskers and outliers.