Stage-Progressive Subcortical BOLD Signal Dynamics in Alzheimer's Disease: Hippocampal Hyperactivation, Thalamic Rigidification, and Pallidal Gate Dysregulation

Background: The thalamus has been identified as an early site of Alzheimer’s disease (AD)
pathology, but its functional trajectory relative to hippocampal degeneration and its specificity for
distinct cognitive domains remain incompletely characterised in large-scale longitudinal data.
Objective: To test the Thalamic Filter Model prediction that thalamic hardware (filter layer)
degrades before hippocampal hardware (memory buffer) across the AD spectrum, with thalamic
integrity specifically predicting executive function but not memory, language, or visuospatial
domains.
Methods: Longitudinal structural MRI and diffusion tensor imaging (DTI) data were drawn from
the Alzheimer’s Disease Neuroimaging Initiative (ADNI), encompassing 804 participants (222
cognitively normal [CN], 391 mild cognitive impairment [MCI], 191 dementia) with 2,412
longitudinal observations and 1,232 DTI observations. Thalamic volume, posterior thalamic
radiation fractional anisotropy (PTR-FA), and hippocampal volume were extracted and compared
across disease stages. Partial correlations examined the independent relationship between thalamic
volume and specific cognitive domains after controlling for hippocampal volume. APOE4 carrier
effects were examined within each disease stage.
Results: Thalamic volume showed monotonic, statistically significant decline across the
CN→MCI→Dementia spectrum (F(2,801)=13.11, η²=0.032, Cohen’s d=0.61). PTR-FA declined
concurrently (F(2,653)=8.90, d=0.73). Hippocampal volume showed a larger overall effect
(d=1.15) but with decline concentrated at the MCI→Dementia transition rather than CN→MCI,
consistent with filter-before-transmitter degradation. Thalamic volume independently predicted
executive function (partial r=0.094, p=0.010) after controlling for hippocampal volume, with zero
independent relationship with memory (partial r=0.011, p=0.78), language (partial r=0.019,
p=0.61), or visuospatial domains (partial r=0.024, p=0.51). APOE4 specifically amplified thalamic
atrophy at the MCI stage (d=−0.224, p=0.039) with null effects at CN and Dementia stages and
null hippocampal APOE4 effects at all stages.
Conclusion: Thalamic filter hardware degrades before hippocampal memory hardware across the
AD spectrum, specifically predicting executive function loss. APOE4 amplifies thalamic atrophy
during the transitional MCI vulnerability window. These findings identify a potential early
intervention window targeting thalamic integrity, with implications for patient stratification in
clinical trials.