RESUMO
Introduction: Suicide is a current leading cause of death in adolescents and young adults. The neurobiological underpinnings of suicide risk in youth, however, remain unclear and a brain-based model is lacking. In adult samples, current models highlight deficient serotonin release as a potential suicide biomarker, and in particular, involvement of serotonergic dysfunction in relation to the putamen and suicidal behavior. Less is known about associations among striatal regions and relative suicidal risk across development. The current study examined putamen connectivity in depressed adolescents with (AT) and without history of a suicide attempt (NAT), specifically using resting-state functional magnetic resonance imaging (fMRI) to evaluate patterns in resting-state functional connectivity (RSFC). We hypothesized the AT group would exhibit lower striatal RSFC compared to the NAT group, and lower striatal RSFC would associate with greater suicidal ideation severity and/or lethality of attempt. Methods: We examined whole-brain RSFC of six putamen regions in 17 adolescents with depression and NAT (MAge [SD] = 16.4[0.3], 41% male) and 13 with AT (MAge [SD] = 16.2[0.3], 31% male). Results: Only the dorsal rostral striatum showed a statistically significant bilateral between-group difference in RSFC with the superior frontal gyrus and supplementary motor area, with higher RSFC in the group without a suicide attempt compared to those with attempt history (voxel-wise p<.001, cluster-wise p<.01). No significant associations were found between any putamen RSFC patterns and suicidal ideation severity or lethality of attempts among those who had attempted. Discussion: The results align with recent adult literature and have interesting theoretical and clinical implications. A possible interpretation of the results is a mismatch of the serotonin transport to putamen and to the supplementary motor area and the resulting reduced functional connectivity between the two areas in adolescents with attempt history. The obtained results can be used to enhance the diathesis-stress model and the Emotional paiN and social Disconnect (END) model of adolescent suicidality by adding the putamen. We also speculate that connectivity between putamen and the supplementary motor area may in the future be used as a valuable biomarker of treatment efficacy and possibly prediction of treatment outcome.
RESUMO
Machine learning (ML) techniques have gained popularity in the neuroimaging field due to their potential for classifying neuropsychiatric disorders. However, the diagnostic predictive power of the existing algorithms has been limited by small sample sizes, lack of representativeness, data leakage, and/or overfitting. Here, we overcome these limitations with the largest multi-site sample size to date (N = 5365) to provide a generalizable ML classification benchmark of major depressive disorder (MDD) using shallow linear and non-linear models. Leveraging brain measures from standardized ENIGMA analysis pipelines in FreeSurfer, we were able to classify MDD versus healthy controls (HC) with a balanced accuracy of around 62%. But after harmonizing the data, e.g., using ComBat, the balanced accuracy dropped to approximately 52%. Accuracy results close to random chance levels were also observed in stratified groups according to age of onset, antidepressant use, number of episodes and sex. Future studies incorporating higher dimensional brain imaging/phenotype features, and/or using more advanced machine and deep learning methods may yield more encouraging prospects.