Citation

Vai B, Calesella F, Lenti C, Fortaner-Uyà L, Caselani E, Fiore P, Breit S, Poletti S, Colombo C, Zanardi R, Benedetti F. Psychiatry Res. Neuroimaging 2023; 331: e111627.

Copyright

(Copyright © 2023, Elsevier Publishing)

DOI

10.1016/j.pscychresns.2023.111627

PMID

36924742

Abstract

Suicide attempts in Bipolar Disorder are characterized by high levels of lethality and impulsivity. Reduced rates of amygdala and cortico-limbic habituation can identify a fMRI phenotype of suicidality in the disorder related to internal over-arousing states. Hence, we investigated if reduced amygdala and whole-brain habituation may differentiate bipolar suicide attempters (SA, n = 17) from non-suicide attempters (nSA, n = 57), and healthy controls (HC, n = 32). Habituation was assessed during a fMRI task including facial expressions of anger and fear and a control condition. Associations with suicidality and current depressive symptomatology were assessed, including machine learning procedure to estimate the potentiality of habituation as biomarker for suicidality. SA showed lower habituation compared to HC and nSA in several cortico-limbic areas, including amygdalae, cingulate and parietal cortex, insula, hippocampus, para-hippocampus, cerebellar vermis, thalamus, and striatum, while nSA displayed intermediate rates between SA and HC. Lower habituation rates in the amygdalae were also associated with higher depressive and suicidal current symptomatology. Machine learning on whole-brain and amygdala habituation differentiated SA vs. nSA with 94% and 69% of accuracy, respectively. Reduced habituation in cortico-limbic system can identify a candidate biomarker for attempting suicide, helping in detecting at-risk bipolar patients, and in developing new therapeutic interventions.

Language: en

Keywords

Suicide; Depression; Machine learning; Bipolar disorder; Habituation