Objective. The current diagnostic criteria for Alzheimer’s disease define discrete biological stages of the disease. While discrete staging simplifies diagnosis, treatment, and participant stratification in clinical trials, discrete disease categories may mask a considerable range of variability in disease severity, even at what is pathologically considered “end stage" disease. We aimed to characterize the contributions of phosphorylated tau neurofibrillary tangle (NFT) pathology and co-pathologies to variability in dementia severity in brain donors diagnosed with a high level of Alzheimer’s Disease neuropathological change (ADNC).

Methods. In a cohort of 197 donors with a primary diagnosis of high ADNC, we computed association between quantitative measures of neocortical NFT load and an informant-based measure of clinical impairment. We examined the impact of co-pathologies on residuals from the canonical pathology-impairment relationship estimated in donors without known significant co-pathologies.

Results. We found a positive association between cortical NFT load and the clinical impairment measure. Clinical impairment was significantly greater than expected for NFT load in presence of co-pathologies, including limbic-predominant age-related TDP-43 encephalopathy (Stage 2 or 3) and cerebrovascular disease. In donors with limbic and neocortical subtypes of Lewy pathology, neocortical -synuclein burden was associated with greater than expected clinical impairment.

Interpretation. Categorizing high ADNC as a single homogeneous category masks a considerable range of pathological severity that contributes to clinical impairment. Complementing discrete staging with continuous measures of tau pathology and probabilistic measures of co-pathologies will improve clinicopathologic studies needed for a precision approach to diagnosis and treatment.