Clinical phenotypes of Alzheimer's disease: investigating atrophy patterns and their pathological correlates

In Alzheimer's disease (AD), MRI atrophy patterns can distinguish between amnestic (typical) and non-amnestic (atypical) clinical phenotypes and are increasingly used for diagnosis and outcome measures in clinical trials. However, understanding how protein accumulation and other key features of neurodegeneration influence these imaging measurements, are lacking. The current study aimed to assess regional MRI patterns of cortical atrophy across clinical AD phenotypes, and their association with amyloid-beta (Aβ), phosphorylated tau (pTau), neuro-axonal degeneration and microvascular deterioration.

Post-mortem in-situ 3DT1 3 T-MRI data was obtained from 33 AD (17 typical, 16 atypical) and 16 control brain donors. Additionally, ante-mortem 3DT1 3 T-MRI scans of brain donors were collected if available. Regional volumes were obtained from MRI scans using an atlas based parcellation software. Eight cortical brain regions were selected from formalin-fixed right hemispheres of brain donors and then immunostained for Aβ, pTau, neurofilament light, and collagen IV. Group comparisons and volume-pathology associations were analyzed using linear mixed models corrected for age, sex, post-mortem delay, and intracranial volume.

Compared to controls, both typical and atypical AD showed volume loss in the temporo-occipital cortex, while typical AD showed additional volume loss in the parietal cortex. Posterior cingulate volume was lower in typical AD compared to atypical AD (- 6.9%, p = 0.043). In AD, a global positive association between MRI cortical volume and Aβ load (βs = 0.21, p = 0.010), and a global negative association with NfL load (βs = - 0.18, p = 0.018) were observed. Regionally, higher superior parietal gyrus volume was associated with higher Aβ load in typical AD (βs = 0.47, p = 0.004), lower middle frontal gyrus volume associated with higher NfL load in atypical AD (βs = - 0.50, p < 0.001), and lower hippocampal volume associated with higher COLIV load in typical AD (βs = - 1.69, p < 0.001). Comparing post-mortem with ante-mortem scans showed minimal volume differences at scan-intervals within 2 years, highlighting the translational aspect of this study.

For both clinical phenotypes, cortical volume is affected by Aβ and neuro-axonal damage, but in opposing directions. Differences in volume-pathology relationships between clinical phenotypes are region-specific. The findings of this study could improve the interpretation of MRI datasets in heterogenous AD cohorts, both in research and clinical settings.