Disentangling the Effects of Suicide Attempts and Psychiatric Diagnosis Based on a Genotype-Informed Dynamic Model of the Serotonin Presynapse

Abstract

Background: Suicide attempts often co-occur with bipolar disorder (BD), major depressive disorder (MDD), and schizophrenia (SCH). Although impairments of the serotonin (5-HT) system have been associated with suicide attempts, it remains unclear whether these alterations reflect suicidal behavior or are confounded by underlying psychiatric diagnosis. This study used a genotype-informed dynamic model of the 5-HT presynapse to disentangle the effects of suicide attempts and psychiatric diagnosis. Methods: We applied a personalized dynamic model of the 5-HT presynapse to 392 psychiatric patients (with BD, MDD, or SCH), categorized by suicide attempt status, and 140 unaffected individuals. The model incorporated five variants across TPH2, SLC6A4, and MAOA genes simulating individual-specific concentration changes of five 5-HT-related molecular species. Model outputs were summarized by six statistical measures (mean, median, maximum, standard deviation, skewness, and kurtosis) and compared across groups. Results: No significant differences were found across groups defined by suicide attempt status and unaffected individuals. However, diagnosis significantly influenced 5-hydroxyindoleacetic acid (5-HIAA) mean, median, maximum, and standard deviation (all p < 0.05). BD patients had lower 5-HIAA levels than SCH patients (mean: p = 0.013; median: p = 0.013; maximum: p = 0.014; standard deviation: p = 0.014). MDD patients also showed lower 5-HIAA levels than SCH patients for the same measures, with differences approaching significance. No significant difference was observed between BD and MDD patients. A diagnosis-by-suicide attempt status interaction was observed for 5-HIAA skewness (p = 0.013). Conclusions: Model-derived 5-HT profiles were shaped primarily by diagnosis, while temporal dynamics of 5-HIAA, rather than its absolute levels, was associated with suicide attempt status. Thus, personalized dynamic modeling incorporating genetic variants may aid in detecting subtle molecular signatures across diagnoses and suicidal behavior.