Abstract
A growing body of evidences suggests that suicidal ideation (SI) and suicidal behaviors have biological bases. However, no biological marker is currently available to evaluate the suicide risk in individuals with SI or suicide attempt (SA). Moreover, the current risk assessment techniques poorly predict future suicidal events. The aim of this study was to examine the association of 39 new and already described peripheral cells and proteins (implicated in the immune system, oxidative stress and plasticity) with lifetime SA, past month SA, current SI, and future suicidal events (visit to the Emergency Department for SI or SA) in 266 treatment-seeking individuals with mood disorders. Equal parts of patients with and without past history of SA were recruited. All individuals at inclusion gave blood, were evaluated for SA recency, current SI, and were followed for two years afterwards. The 39 peripheral blood cellular and protein markers were entered separately for each outcome in Elastic Net models with 10-fold cross-validation, followed by single-analyte covariate-adjusted regression analyses for pre-selected analytes. Past month SA was associated with increased plasma levels of thrombospondin-2 and C-reactive protein, whereas current SI was associated with lower plasma serotonin levels. These associations were robust to adjustments for key covariates and corrections for multiple testing. The Cox proportional hazards regression showed that higher levels of thrombospondin-1 and of platelet-derived growth factor-AB predicted a future suicidal event. These two associations remained after adjustment for sex, age, and SA history, and outperformed the predictive value of past SA. Thrombospondins and platelet-derived growth factors have never been investigated in the context of suicide. Altogether, our results highlight the involvement in the suicidal process of platelet biological response and plasticity modifiers and also of inflammatory factors. They also suggest that SI and SA may have different biological correlates and that biomarkers associated with past SA or current SI do not automatically also predict future events.