Neuroimaging pattern interactions for suicide risk in depression captured by ensemble learning over...

Area of Science:

• Neuroscience
• Genetics
• Psychiatry

Background:

• Urgent need for reliable neuroimaging biomarkers for suicide in major depression disorder (MDD).
• Requirement for biomarkers interpretable through molecular tissue signatures.
• Development of an ensemble learning scheme over transcriptome-defined parcellations (TDP) to explore brain patterns.

Purpose of the Study:

• To develop and validate a novel neuroimaging approach for identifying suicide risk biomarkers in MDD.
• To explore homogeneously parcellated brain patterns and their interactions using transcriptome data.
• To quantify brain dysfunction related to suicide using resting-state functional abnormality (RSFA) scores.

Main Methods:

• Recruited 96 MDD patients without suicide attempt (SA), 86 with SA, and 102 healthy controls for resting-state fMRI.
• Generated spatially-continuous TDPs based on gene expression from the Allen Human Brain Atlas.
• Applied a three-layer ensemble learning scheme to integrate TDPs and quantify RSFA scores for suicide risk prediction.

Main Results:

• Ensemble learning over TDPs achieved higher suicide predictive accuracy (73.23 ± 1.07%) compared to regional analysis or null models.
• A specific parieto-occipital TDP (PO-TDP) pattern, quantified by RSFA score, was crucial for suicide risk prediction.
• Alterations in the PO-TDP were associated with transcriptional profiles of GRIN2A and GABRG2, with overrepresentation of glutamatergic and GABAergic synapses.

Conclusions:

• Glutamatergic and GABAergic dysfunction in the visual cortex, indicated by the PO-TDP pattern, is implicated in suicide risk.
• An inherent excitatory/inhibitory imbalance in these pathways may contribute to aberrant emotional processing and neurocognitive deficits.
• These findings suggest potential molecular targets and neuroimaging biomarkers for suicide prevention in MDD.