Evaluation of CSF Biomarkers as Predictors of Alzheimer's Disease: A Clinical Follow-Up Study of 4.7 Years

In this study, we determined the diagnostic accuracy of cerebrospinal fluid (CSF) biomarkers to predict development of Alzheimer's disease (AD) within five years in patients with mild cognitive impairment (MCI). To do so, the levels of tau, phosphorylated tau, A beta(42), A beta(40), A beta(38), sA beta PP alpha, and sA alpha PP beta were analyzed in 327 CSF samples obtained at baseline from patients with AD (n = 94), MCI (n = 166), depressive disorder (n = 29), and cognitively healthy controls (n = 38). In the cohort with MCI at baseline, 33% subsequently developed AD and 16% developed other types of dementia; however, 51% were still cognitively stable after a follow-up of 4.7 years (range 3.0-7.2). Optimal cut-offs for each biomarker or combinations of biomarkers were defined in the AD, control, and depressive disorder groups. Several combinations resulted in sensitivity and specificity levels > 85% for differentiation of AD from controls and depressive disorder. Using the previously established cut-offs, a combination of A beta(42) and tau could predict future development of AD in MCI patients with a sensitivity of 88%, specificity 82%, positive predictive value 71%, and negative predictive value 94%. MCI patients with both low A beta(42) and high tau levels had a substantially increased risk of developing AD (OR 20; 95% CI 6-58), even after adjustment for confounding factors. Ultimately, CSF biomarkers can stratify MCI patients into those with very low or high risk for future development of AD. However, the specificities and positive predictive values are still too low to be able to diagnose AD before the patients fulfill the clinical criteria.