Clinical-biological Alzheimer's disease stage concordance: insights from cohorts and autopsy data

Alzheimer's disease (AD) is defined by its characteristic neuropathological changes, which allow for diagnosis and assessment of severity. Recently, the Alzheimer's Association proposed a framework to stage AD biologically based on tau-PET. Furthermore, the framework hypothesizes a degree of alignment between biological AD severity and clinical symptom severity. We aimed to investigate the concordance between clinical and biological stages of AD and explore factors contributing to discordance using in vivo and post-mortem neuropathological data. Data from 768 amyloid-β-positive individuals were drawn from four observational cross-sectional in vivo cohorts (TRIAD, ADNI, HABS-HD and SCAN) in addition to a post-mortem autopsy dataset from the National Alzheimer's Coordinating Center (NACC; n = 3188). All in vivo participants had tau-PET imaging, clinical diagnosis and neurobehavioural assessments. Participants were assigned a biological AD stage based on their tau-PET scan according to the Alzheimer's Association revised criteria stages. The autopsy dataset included individuals with moderate-to-frequent neuritic plaques (CERAD scores 2-3), along with pre-mortem clinical and neurobehavioural data. Clinical-biological concordance was quantified using squared-weighted Cohen's κ. Ordinal and linear regression models assessed associations between biological stage and clinical severity (Clinical Dementia Rating-Sum of Boxes, Mini-Mental State Examination), adjusting for age, sex and cohort. Post-mortem analyses evaluated the impact of comorbid neuropathologies on clinical-biological discordance using adjusted odds ratios and ordinal regression. Overall concordance between clinical and biological AD staging was moderate (Cohen's κ = 0.52, P < 0.001). Approximately 70% of individuals classified as cognitively unimpaired or with dementia exhibited biological stages consistent with their clinical diagnoses. In contrast, transitional decline and mild cognitive impairment groups were more heterogeneous. Notably, 25% of amyloid-β-positive individuals with mild cognitive impairment demonstrated no detectable tau-PET abnormality. Nonetheless, advanced tau-PET stage was reliably associated with clinical impairment. In the NACC autopsy dataset, nearly all individuals with a more severe clinical stage than their proposed biological stage exhibited comorbid neuropathologies, including frontotemporal lobar degeneration (FTLD)-TDP-43, FTLD-tau, Lewy bodies, limbic-predominant age-related TDP-43 encephalopathy (LATE) and cerebrovascular disease. The number of comorbid pathologies was strongly associated with increased odds of clinical dementia (t = 8.45, P < 0.001). Although there is moderate agreement between clinical and biological stages of AD across the entire disease spectrum, strong agreement is found in clinically unimpaired and dementia stages. Comparison of clinical and biological AD stages provides a framework for understanding the large contributions of non-AD neurodegenerative diseases to dementia in amyloid-β-positive individuals. Our results have important implications for clinical trial recruitment strategies and highlight the urgent need for biomarkers for non-AD pathological processes.