A combination model of AD biomarkers revealed by machine learning precisely predicts Alzheimer's dementia: China Aging and Neurodegenerative Initiative (CANDI) study

Abstract Introduction To test the utility of the “A/T/N” system in the Chinese population, we study core Alzheimer's disease (AD) biomarkers in a newly established Chinese cohort. Methods A total of 411 participants were selected, including 96 cognitively normal individuals, 94 patients with mild cognitive impairment (MCI) patients, 173 patients with AD, and 48 patients with non‐AD dementia. Fluid biomarkers were measured with single molecule array. Amyloid beta (Aβ) deposition was determined by 18 F‐Flobetapir positron emission tomography (PET), and brain atrophy was quantified using magnetic resonance imaging (MRI). Results Aβ42/Aβ40 was decreased, whereas levels of phosphorylated tau (p‐tau) were increased in cerebrospinal fluid (CSF) and plasma from patients with AD. CSF Aβ42/Aβ40, CSF p‐tau, and plasma p‐tau showed a high concordance in discriminating between AD and non‐AD dementia or elderly controls. A combination of plasma p‐tau, apolipoprotein E ( APOE ) genotype, and MRI measures accurately predicted amyloid PET status. Discussion These results revealed a universal applicability of the “A/T/N” framework in a Chinese population and established an optimal diagnostic model consisting of cost‐effective and non‐invasive approaches for diagnosing AD.