Functional staging of Alzheimer's disease dementia with Addenbrooke Cognitive Examination-III: evidence for subtype-specific cognitive thresholds.

INTRODUCTION: Translating cognitive test results into clinically meaningful measures of dementia severity remains challenging in real-world settings, where biomarkers are limited and functional assessment may occasionally be confounded by non-cognitive factors. Although the Addenbrooke's Cognitive Examination III (ACE-III) is widely used for detecting Alzheimer's disease (AD), its role in severity staging is less established. This study evaluated the usefulness of ACE-III for functionally anchored staging and compared it with the Mini-Mental State Examination (MMSE). METHODS: In a retrospective real-world cohort of 1,170 patients with AD dementia (late-onset, early-onset, and mixed), severity was classified using the Functional Assessment Staging Tool (FAST). Receiver operating characteristic analyses were used to derive cut-offs for differentiating adjacent stages (FAST 4 vs 5; FAST 5 vs 6). Domain-level and subtype-specific analyses were conducted. RESULTS: ACE-III showed excellent discrimination between FAST 4 and 5 (AUC = 0.92; cut-off ≤56) and moderate discrimination between FAST 5 and 6 (AUC = 0.77; cut-off ≤42), outperforming MMSE in later stages (AUC = 0.63). Attention contributed most to earlier transitions, whereas later stages showed broader domain involvement. Cut-offs varied across AD subtypes. CONCLUSION: ACE-III provides clinically meaningful estimates of dementia severity and may help overcome limitations of functional assessment in real-world settings, supporting its use for staging and monitoring in routine clinical practice.