Hyperexcitability Across Brain Networks in Late-Life Depression

Background: Late-life depression (LLD) remains a major clinical challenge due to the lack of robust neurobiological markers, limiting the development of targeted and effective treatments. Methods: Here, we introduce a novel computational framework that applies the Ising model to multimodal neuroimaging data, inferring excitation–inhibition balance (EIB) by integrating structural and functional connectivity. In a large cohort from the Alzheimer’s Disease Neuroimaging Initiative (113 LLD subjects, 165 controls, all aged 65+), we identified group-level differences in EIB, independent of common neurodegenerative risk factors including apolipoprotein E and amyloid burden. Results: We identified increased whole-brain EIB in LLD ( p = 0.0014, β = 0.039), independent of common neurodegenerative risk factors. After multiple comparisons correction, network-level analysis revealed pronounced EIB dysregulation in the limbic ( p < 0.001), dorsal attention ( p < 0.001), and basal ganglia networks ( p = 0.007). At the regional level, significant alterations emerged in the hippocampus, thalamus, and anterior cingulate cortex (all p < 0.05)—areas central to emotional regulation, executive function, and cognitive processing. Notably, mild cognitive impairment displayed the opposite pattern with decreased whole-brain EIB ( p = 0.042, β = −0.023), providing a clear neurobiological distinction between depression and neurodegeneration. Conclusions: These findings establish EIB dysregulation as a core mechanism in LLD, offering a novel explanatory framework for why traditional antidepressants often fail in this population. The regional specificity of these alterations suggests new approaches for circuit-based interventions, particularly relevant to emerging rapid-acting antidepressants that modulate the EIB. This mechanistic framework offers a biomarker for diagnosis and treatment monitoring and informs novel, circuit-based therapeutic strategies.