BackgroundBenralizumab is highly effective in many, but not all, patients with severe asthma. Baseline characteristics alone are insufficient to predict an individual's probability of long-term benralizumab response.ObjectivesTo (1) study whether parameters at 3 months –in addition to baseline characteristics– contribute to the prediction of benralizumab response at 1 year and to (2) develop an easy-to-use prediction tool to assess an individual's probability of long-term response.MethodsWe assessed the effect of benralizumab treatment in 192 patients from the Dutch severe asthma registry (RAPSODI). To investigate predictors of long-term benralizumab response (≥50% reduction in maintenance oral corticosteroid (OCS) dose or annual exacerbation frequency) we used logistic regression, including baseline characteristics and 3-month Asthma Control Questionnaire (ACQ-6) score and maintenance OCS dose.ResultsBenralizumab treatment significantly improved several clinical outcomes and 144 (75%) patients were classified as long-term responders. Response prediction improved significantly when 3-month outcomes were added to a predictive model with baseline characteristics only (AUROC 0.85versus0.72, p=0.001). Based on this model, a prediction tool using gender, prior biologic use, baseline blood eosinophils, FEV₁and at 3 months OCS dose and ACQ-6 was developed which classified patients into 3 categories with increasing probability of long-term response (95%CI): 25%(3–65), 67%(57–77) and 97%(91–99) respectively.ConclusionIn addition to baseline characteristics, treatment outcomes at 3 months contribute to the prediction of benralizumab response at 1 year in patients with severe eosinophilic asthma. Prediction tools as proposed in this study may help physicians optimize the use of costly biologics.