Frontiers in Aging Neuroscience · Nov 2, 2026
Baseline amygdala and hippocampal–amygdaloid transition area volumes in cognitively normal older adults with a subsequent increase in Clinical Dementia Rating Sum of Boxes: an exploratory ADNI study
By Juan Zhang, Runyang Li, Peixue Li et al.
The amygdala is affected early in Alzheimer's disease, but whether its volume already differs in cognitively normal older adults before the first clinically rated change is uncertain. We analysed baseline T1-weighted MRI from 289 cognitively normal participants in the Alzheimer's Disease Neuroimaging Initiative whose Clinical Dementia Rating Sum of Boxes (CDR-SB) was 0 at baseline: 91 had a CDR-SB above 0 at one or more later assessments (increase group) and 198 remained at 0 (stable group). This operational endpoint is not equivalent to a global CDR of 0.5 or to mild cognitive impairment. Amygdala nuclei and hippocampal subregions were segmented with FreeSurfer 7.3.2 and compared with linear models adjusted for age, sex, estimated total intracranial volume, APOE ε4 status and ADNI phase, which differed between groups. The increase group had a smaller right whole amygdala (d = −0.31, 95% CI −0.58 to −0.04), but this difference did not survive false discovery rate correction across the four whole-structure tests (q = 0.093), the left amygdala difference was smaller and imprecise (d = −0.24, 95% CI −0.51 to 0.03), and whole hippocampal volumes did not differ detectably. No amygdala nucleus survived correction. Among 44 hippocampal measures, only the left hippocampal–amygdaloid transition area (HATA) survived correction (d = −0.48, 95% CI −0.75 to −0.21; q = 0.021), and it remained smaller after adjustment for whole amygdala volume. No hemisphere-by-group interaction was significant. Whole-amygdala estimates were about 30% smaller after phase adjustment and smaller among ADNI-2/3 participants. The differences were evident in amyloid-negative but not in the few amyloid-positive participants. In nested cross-validation, adding the HATA volumes to the covariates raised the AUC from 0.69 to 0.72, but in a whole-pipeline bootstrap the gain was uncertain (median 0.02; 95% interval −0.04 to 0.10). A smaller left HATA, and less certainly a smaller right amygdala, was associated with a subsequent CDR-SB increase in this sample, but the effects were small, partly dependent on adjustment for ADNI phase and of no demonstrated value for individual prediction. Because volumes were measured once, the data cannot distinguish early atrophy from long-standing differences in volume.