Analysis of depressive symptom trajectory patterns with zuranolone: Time‐series clustering of Japanese Phase 2 and Phase 3 trial data

Abstract Aim Major depressive disorder (MDD) shows varied treatment responses. Previous Phase 2 and 3 trials demonstrated the efficacy of 2‐week oral zuranolone in Japanese MDD patients, but individual symptom trajectories were not analyzed. To address this gap, this study aimed to characterize patterns of depressive symptom trajectories among these patients. Methods We conducted a pooled analysis of Phase 2/3 randomized, double‐blind, placebo‐controlled trials in Japanese adults with MDD. Participants received zuranolone 30 mg or placebo once daily for 14 days, with a 6‐week follow‐up. Patients with complete 17‐item Hamilton Rating Scale for Depression (HAM‐D17) data at baseline and 10 subsequent time points through Day 57 were included. Change from baseline scores underwent Dynamic Time Warping–based k‐means clustering identified symptom trajectory patterns in the zuranolone group, with optimal cluster number determined by the elbow method. A Random Forest classifier then assessed scale items linked to symptom worsening trajectories. Results Four distinct trajectories were identified. Three clusters showed varying degrees of symptom reduction during treatment, stabilizing during follow‐up. The fourth cluster showed symptom reduction during the treatment period and a mean increase in HAM‐D17 scores after Day 15, primarily due to worsening insomnia. Each Patient Health Questionnaire‐9 (PHQ‐9) score showed a similar course to the corresponding HAM‐D17 scores. No notable differences were observed among clusters in baseline demographics, episode duration, or initial severity. Conclusion This study reveals diverse symptom trajectories in zuranolone‐treated patients and provides clinically relevant insights, highlighting the importance of early post‐dosing monitoring—particularly of sleep symptoms—to inform clinical management and future research.