Abstract
BACKGROUND: Abnormal tau proteins are independent contributors to cognitive impairment. Nevertheless, not all individuals exposed to high-level tau pathology will develop cognitive dysfunction. We aimed to construct a model to predict cognitive trajectory for this high-risk population. METHOD: Longitudinal data of 181 non-demented adults (mean age= 73.1; female= 45%), who were determined to have high cerebral burden of abnormal tau by cerebrospinal fluid (CSF) measurements of phosphorylated tau (ptau181) or total tau, were derived from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Cognitive decline was defined as Mini-Mental State Examination scores decline ≥ 3 over three years. A predictive nomogram was constructed using stepwise backward regression method. The discrimination, calibration, and clinical usefulness of the nomogram were evaluated. The model was validated in another 189 non-demented adults via a cross-sectional set (n=149, mean age = 73.9, female = 51%) and a longitudinal set (n= 40, mean age = 75, female = 48%). Finally, the relationships of the calculated risk scores with cognitive decline and risk of Alzheimer's disease were examined during an extended 8-year follow-up. RESULT: Lower volume of hippocampus (odds ratio [OR] = 0.37, p< 0.001), lower levels of CSF sTREM2 (OR = 0.76, p = 0.003), higher scores of Alzheimer's Disease Assessment Scale-Cognitive (OR = 1.15, p = 0.001) and Functional Activities Questionnaire (OR = 1.16, p = 0.016), and number of APOE ε4 (OR = 1.88, p = 0.039) were associated with higher risk of cognitive decline independent of the amyloid status and were included in the final model. The nomogram had an area of under curve (AUC) value of 0.91 for training set, 0.93 for cross-sectional validation set, and 0.91 for longitudinal validation set. Over the 8-year follow-up, the high-risk group exhibited faster cognitive decline (p< 0.001) and a higher risk of developing Alzheimer's dementia (HR= 6.21, 95% CI= 3.61-10.66, p< 0.001 ). CONCLUSION: APOE ε4 status, brain reserve capability, neuroinflammatory marker, and neuropsychological scores can help predict cognitive decline in non-demented adults with high burden of tau pathology, independent of the presence of amyloid pathology.