Prediction of the effectiveness of Esketamine nasal spray in patients with difficult-to-treat depression

Abstract

Background: Patients with difficult-to-treat depression face significant challenges. Despite the promising results of esketamine nasal spray in reducing depressive symptoms, the response and remission rates are low. Together with the high cost and time investment, early prediction of treatment outcomes is crucial. This study examines whether changes in Montgomery-Åsberg Depression Rating Scale (MADRS) scores after one week of esketamine treatment can predict response and remission at weeks 4 and 8 in depression according to MADRS scores, as well as in quality of life. Methods: This retrospective study included 395 patients with unipolar depression recruited through the Esketamine Nasal Spray Consortium of the Netherlands (ENC-NL). Depression severity was assessed using MADRS, and quality of life was measured with WHODAS and EQ-5D. Data were split into training (75%) and test sets (25%), with 10-fold cross-validation applied. A logistic regression model, implemented as a generalized linear model (GLM) with a binomial link function, was trained and optimized based on receiver operating characteristic (ROC) metrics. Model performance was assessed using ROC curves, area under the curve (AUC), negative predictive values (NPV), and specificity. Results: A 10% decrease in MADRS score after one week strongly predicted non-response and non-remission at weeks 4 and 8 across both depressive symptoms and quality-of-life measures. The model demonstrated high reliability for identifying non-responders and non-remitters but had limited accuracy in predicting positive outcomes such as response and remission. Conclusions: Early improvements in MADRS scores are robust predictors of treatment outcomes and quality-of-life changes. This study provides clinically relevant thresholds to guide timely and informed adjustments to treatment strategies, improving care for patients with difficult-to-treat depression.