Background and objectives: The amyloid cascade hypothesis posits that Alzheimer disease (AD) progresses from amyloid deposition to tau deposition, neurodegeneration, and eventually cognitive impairment and is the foundation of the revised criteria of Alzheimer's Association Workgroup 2024 (AA-2024). To account for copathologies and cognitive resilience that affect the penetrance of the AD cascade, AA-2024 introduced a 2-dimensional biological-clinical staging framework. We aimed to estimate the proportion of persons along the AD continuum whose biological and clinical trajectories align with the amyloid cascade.

Methods: Cross-sectional data of the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort were tested in the 4 × 4 biological/clinical staging matrix adapted from the AA-2024 criteria. Biological stages were defined by amyloid and tau-PET burden: stage A (amyloid positivity, A+), stage B (medial temporal tau, A+/T_2MTL+), stage C (moderate neocortical tau, A+/T_2MOD+), and stage D (high neocortical tau, A+/T_2HIGH+). Clinical stages were cognitively unimpaired (stage 1), subtle cognitive impairment (stage 2), mild cognitive impairment (stage 3), and dementia (stages 4-6). Tau-PET cutoffs were defined through the implementation of 5 distinct methods. Participants were categorized into (1) compliant with the amyloid cascade (matrix diagonal), (2) resilient (advanced biological stage-early clinical stage), and (3) copathologic (early biological stage-advanced clinical stage). Observed distributions were compared with hypothetical scenarios with zero and high amyloid cascade penetrance using the χ² test, and differences among the 5 methods were tested using the Cochran Q test.

Results: Two-hundred and fifty-six amyloid-positive individuals (mean age: 72.7 years; 51% female) from the ADNI cohort were considered. The proportion of participants compliant with the amyloid cascade was between 31% (95% CI 25%-37%) and 36% (95% CI 30%-42%) depending on the tau-PET cutoff method. The observed number of individuals compliant with the amyloid cascade was higher than in the zero-penetrance scenario but lower than in the high-penetrance distribution (p < 0.01). The proportion of copathologic (17%-63%) and resilient (6%-52%) individuals varied widely by tau-PET cutoff (p < 0.001).

Discussion: Only approximately one-third of persons with an AA-2024 diagnosis of AD complied with the predictions of the amyloid cascade hypothesis. These results suggest the heterogeneity in how clinical symptoms and pathology are coupled along the AD continuum, which has significant implications for interpreting completed antiamyloid clinical trials and designing future studies.