Phenotyping of acute stress response in persons with impulsive and non-impulsive suicide attempts: a multimodal neuroimaging and peripheral biomarker study

• Applicant: Lejla Colic, Ph.D.
• Subject Area: Biological Psychiatry
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 569177901

Suicide attempts (SA) are a public health concern. Despite research efforts, reliable biological markers of SA risk are lacking. This may stem from the heterogeneity in the neurobiological mechanisms underlying SA. Evidence points toward the existence of distinct SA phenotypes - impulsive and non-impulsive - characterized by divergent stress-response profiles. Impulsive SAs are poorly planned, triggered by acute emotional distress, and marked by heightened sensitivity to psychosocial stressors. In contrast, non-impulsive SAs are premeditated, associated with persistent suicidal ideation, and rooted in chronic stress-system dysregulation. Yet, no study has comprehensively validated this distinction using multimodal biological markers and longitudinal follow-up. The PASTIS project aims to fill this critical gap by identifying and validating neurobiological stress-response signatures that differentiate impulsive and non-impulsive SA subtypes. 150 individuals will be recruited across two national cohorts (France, and Germany), with three groups: impulsive SA, non-impulsive SA, and affective controls experiencing a current major depressive episode without history of SA. Each participant will undergo clinical phenotyping, including SA, suicidal ideation, depression severity, impulsivity, emotional regulation, and stress exposure. Participants will complete multimodal neuroimaging, including resting-state and a paradigm for inducing psychosocial stress in the scanner. Salivary cortisol will be collected at five time points to map the HPA-axis response to stress. Heart rate variability will be continuously monitored using photoplethysmography to capture autonomic stress reactivity. Plasma will be sampled before and two hours after the stress task to quantify dynamic changes in inflammatory, vascular, and neuroplasticity markers. In parallel, neuron-enriched extracellular vesicles will be isolated from baseline plasma to evaluate molecular signals derived from brain tissue. Hair samples will be collected to assess chronic stress physiology. Follow-up assessments at 1 week, 1 month, and 6 months will track suicidal ideation, impulsivity, depressive symptoms, and life stressors, allowing to test the predictive validity of biological profiles for future outcomes. Multimodal data will be integrated using regularized machine learning approaches (Elastic Net, XGBoost) to identify features that best differentiate SA subtypes. Models will undergo cross-validation and cross-national replication (France vs Germany) to ensure generalizability across healthcare contexts. Explainable AI techniques (SHAP values) will be implemented to provide interpretable outputs for clinical translation. PASTIS is the first study to combine stress-challenge neuroimaging, biomarker profiling, longitudinal follow-up, and explainable machine learning to dissect heterogeneity in suicide risk. Results will include validated risk models, a harmonized open dataset, and shareable pipelines.

• DFG Programme: Research Grants
• International Connection: France
• Cooperation Partner: Dr. Aiste Lengvenyte