Description

Background and rationale Severe psychiatric disorders such as major mood disorders, psychotic disorders and autism spectrum disorders are clinically, genetically and biologically heterogeneous, which limits the efficacy of current treatments. Precision psychiatry requires deep and objective phenotyping of large samples to identify homogeneous subgroups that share common clinical and biological features and may benefit from targeted prevention and treatment strategies. Based on the literature and international consensus, social withdrawal and anhedonia are two particularly robust transdiagnostic dimensions with an underlying biological hypothesis, validated French measurement tools, existing translational models, and recognised relevance for patients.

French Minds is part of the French national research programme PEPR PROPSY (Precision Psychiatry), funded by the French government under France 2030. It extends pre-existing French psychiatric cohorts to build a multimodal database of more than 10,000 deeply phenotyped patients.

Study design National, multicentre, prospective, longitudinal cohort study (French "Jardé" law, category 1 interventional research not involving a health product). Patients are recruited in French Minds centres, i.e., the FondaMental Foundation expert centres and the First Episode Psychosis (PEP) / RHU PsyCare centres. Healthy volunteers are recruited through advertising and pre-screened by telephone, then assessed with the MINI interview at inclusion.

Study procedures

• Visit 0 (inclusion): eligibility check, informed consent, sociodemographic and medical history, self-questionnaires; routine biological check-up prescribed for patients.
• Visit 1 (within 8 weeks of V0, one day): fasting blood sample for research (genotyping, whole-genome sequencing, DNA methylation, RNA and microRNA sequencing, metabolomics and lipidomics, proteomics, inflammatory and immune markers, autoantibodies, immune cell subpopulations), metabolic and anthropometric measures, neuropsychological assessment (CANTAB, WAIS-IV subtests, CVLT, theory of mind tasks, verbal fluency, attention), motor strength, minor neurological signs, clinician-rated scales (BPRS, C-SSRS, treatment response and resistance, life events, neurodevelopmental history, autoimmune psychosis criteria), and self-questionnaires.
• Visit 2 (within 6 weeks of V1): 3T brain MRI (structural and functional sequences, harmonised across sites), approximately one hour.
• Visit 3 (12 to 14 months after V1, one day): same assessments as V1, plus course of illness since V1 (episodes, hospitalisations, care use, treatment history).
• Visit 4 (within 6 weeks of V3): 3T brain MRI identical to V2.
• Optional stool collection at home at V0 and V3 for gut microbiome (metagenomic) analysis, subject to specific consent.
• Between visits (over one year): digital phenotyping using a study-provided smartphone with the Kanopsy app (validated self-questionnaires on anhedonia, sleep, loneliness, pain, anxiety and depression, repeated over one year) and a connected watch for passive collection of activity, sleep and heart rate variability. A safety alert redirects participants to the French national suicide prevention line (3114) when scores indicate clinical concern.
• Environmental exposure is estimated from residential history and housing characteristics. Linkage with the French National Health Data System (SNDS) is planned to study long-term care trajectories, subject to specific regulatory authorisations (CESREES, CNIL).

Biological samples are stored in a dedicated biobank (Hôpitaux Universitaires Henri Mondor, AP-HP). Genomic analyses are performed at the CNRGH (CEA) and MRI acquisition harmonisation, centralisation and quality control are handled by the CATI national neuroimaging platform.

Statistical analysis Approximately 500 patients per diagnostic group (2,500 patients) and 500 healthy volunteers will be included, allowing estimation of subgroup proportions (expected 20-50% per disorder) with 5% precision (alpha 5%), accounting for 20% loss to follow-up or refusal. Analyses will combine clustering methods, multivariate regression, analysis of variance, growth and survival models for longitudinal data, and exploratory data-mining / machine-learning approaches. Missing data will be handled by multiple imputation and multiple testing will be controlled using the false discovery rate (FDR).

Expected benefits There is no direct individual benefit; the expected benefit is collective, through improved understanding of prognostic factors and biological mechanisms in severe psychiatric disorders, with the potential to support the development of new, more personalised therapeutic strategies.