Structural heterogeneity reveals distinct amyloidopathies in Alzheimer's disease

Within the amyloid/tau/neurodegeneration (ATN) framework, hippocampal (H) and inferior parietal lobule (IPL) atrophy are established markers of neurodegeneration (N+), whereas the biological significance of choroid plexus (ChP) enlargement remains uncertain. Although amyloid positivity (A+) underpins the current biological definition of Alzheimer's disease, the extent to which it is consistently associated with structural neurodegeneration remains unclear. We investigated the relationship between A+ and structural neurodegeneration using structural MRI, amyloid PET, and cognitive data from 872 participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Regional brain volumes (H, IPL and ChP) were classified as abnormal using validated normative thresholds. Diagnostic performance was evaluated using multiple machine-learning approaches, and associations between structural phenotypes, cognitive profiles and cerebrospinal fluid (CSF) biomarkers were examined. Among 872 participants, 411 were amyloid-negative healthy subjects, 339 had amyloid-positive mild cognitive impairment, and 122 had amyloid-positive Alzheimer's disease dementia. At the group level, H and IPL progressively decreased, whereas ChP volume increased across disease stages (all P<0.001). However, 45% of amyloid-positive individuals exhibited no detectable structural neurodegeneration. Consistently, machine-learning models demonstrated limited individual-level diagnostic performance, with a maximum overall accuracy of 0.60 [0.57-0.62]. Hierarchical clustering identified four cognitive phenotypes that mapped onto distinct structural signatures. Amyloid positivity alone does not uniformly predict structural neurodegeneration. The marked dissociation between amyloid pathology and MRI-defined neurodegeneration demonstrates substantial biological heterogeneity within amyloid-positive Alzheimer's disease and supports the concept of distinct structural amyloidopathies. These findings have important implications for biomarker interpretation, patient stratification, and the development of precision therapeutic strategies.