Unsupervised neurobiology-driven stratification of clinical heterogeneity in depression

Introduction One of the main obstacles in providing effective treatments for major depressive disorder (MDD) is clinical heterogeneity, whose neurobiological correlates are not clearly defined. A biologically meaningful stratification depressed patients needed to promote tailored diagnostic procedures. Objectives Using structural data, we performed an unsupervised clustering define clinically clusters patients. Methods T1-weighted and diffusion tensor images were obtained from 102 MDD In 64 patients, symptoms, number stressful life events, severity exposure adverse childhood experiences evaluated using Beck Depression Inventory (BDI), Schedule Recent Experiences (SRE), Risky Family Questionnaire (RFQ), Childhood Trauma (CTQ). Clustering analyses with extracted tract-based fractional anisotropy (TBSS, FSL), cortical thickness, surface area, regional measures grey matter volumes (CAT12). Gaussian mixture model was implemented clustering, considering Support Vector Machine (SVM) as classifier. 10x2 repeated cross-validation grid search hyperparameters tuning clusters’ stability. The optimal determined by normalized stability, Akaike Bayesian information criterion. Analyses adjusted total intracranial volume, age, sex. relevance identified assessed through MANOVA, domains scales dependent variables labels fixed factors. Discriminant analysis subsequently assess discriminative power these variables. Results Cross-validated approach 2 highly stable (normalized stability=0.316, AIC=-80292.48, BIC=351329.16). MANOVA showed a significant between-clusters difference scores (p=0.038). distinguished two accuracy 78.1%, BDI behavioural CTQ minimisation/denial showing highest discriminant values (0.325 0.313). Conclusions Our results defined informed associated trauma specific profiles, which may assist targeting interventions treatments. Disclosure Interest None Declared