Background: The presence of AD pathology can result in diverse behavioral phenotypes, including the typical amnestic variant characterized by memory deficits, and an atypical aphasic variant characterized by language deficits. Previous research has identified unique cortical atrophy patterns in each phenotype, though less focus has been drawn to subcortical involvement. The current study sought to dissociate these behavioral phenotypes by characterizing their thalamic volume and shape features using high-dimensional brain mapping procedures. Relationships between brain metrics and specific language and memory deficits were also investigated in aphasic AD and amnestic AD, respectively. Method: Thalamic integrity was examined in aphasic AD (n = 25), amnestic AD (n = 21), and healthy control participants (n = 44). Age and supratentorial volume (STV) were used as covariates in all analyses. MR scans were acquired using high-resolution T1-weighted MPRAGE volumes following the ADNI protocol. Thalamic shape features were estimated using Large Deformation Diffeomorphic Metric Mapping. General linear models compared differences in thalamic shape between groups. Pearson correlation coefficients characterized relationships between thalamic nuclei (pulvinar, anterior, and mediodorsal) and language and memory performance in aphasic AD and amnestic AD, respectively. Results: After controlling for age and STV, thalamic volume did not differ between groups [F (2,85) = 2.55, p = 08]. However, AD phenotypes exhibited bilateral inward shape deformation in dorsal and ventral regions extending in an anterior to posterior fashion [left: F(20, 154) = 2.61, p < .001; right: F(20,154)= 2.26, p < .01]. Amnestic AD demonstrated right ventrolateral localized volume loss relative to aphasic AD. Pearson models revealed lower confrontation naming was correlated with localized volume loss of bilateral pulvinar (left: r = .59, p < .01; right: r = .55, p < .01), and bilateral anterior (left: r = .50, p = .01; right: r = .49, p = .01) thalamic nuclei for aphasic AD; lower delayed recall was significantly correlated with localized volume loss in left anterior (r = .46, p = .04) thalamic nuclei in amnestic AD. Conclusions: In the absence of volumetric differences, shape measures captured distinct patterns of localized volume loss in aphasic AD and amnestic AD behavioral phenotypes relative to control participants. Comparisons of AD variants demonstrated inward deformation in amnestic AD, particularly in right ventrolateral regions. Thalamic changes appear to be implicated in AD pathology, with relationships to the expected cognitive impairments, although thalamic atrophy patterns are unable to fully dissociate behavioral phenotypes.



College and Department

Family, Home, and Social Sciences; Psychology



Date Submitted


Document Type





Alzheimer's disease, primary progressive aphasia, shape analysis