Predictive accuracy of the extended Predementia Alzheimer disease Scale (PAS) for Alzheimer-type dementia in subjects with MCI

Background: The Predementia Alzheimer disease Scale (PAS) has been developed to predict Alzheimer disease (AD) in subjects with mild cognitive impairment (MCI). The PAS combines a number of markers for AD: age, the MMSE score, functional impairment, cognitive test impairment, medial temporal lobe atrophy, and the apolipoprotein E genotype. Previous studies showed a high predictive accuracy of the PAS. Objective: To investigate whether the predictive accuracy of the PAS can be increased by addition of a item for the levels of beta amyloid 1-42 and tau protein phosphorylated at threonine 181 in cerebrospinal fluid. This extended version is called the ePAS. Methods: 52 subjects with subjective complaints (n 13) or amnestic MCI (n 39) were selected from the Alzheimer centre Vumc. Outcome measures were AD-type dementia at follow-up and cognitive decline. Cognitive decline was defined as AD at follow-up, or decline of at least 1 standard deviation on neuropsychological tests, or persisting memory impairment. Predictive accuracy was assessed with the Area Under the Curve (AUC) of a ROC curve. Results: Subjects were 69 years old and had an MMSE score of 27 at baseline. After a follow-up of on average 1.7 years 18 subjects had converted to AD-type dementia. Of those without dementia at follow-up, 15 showed cognitive decline, 10 had no decline, and no data on decline were available for 9 subjects. If AD-type dementia was the outcome measure, the AUC of the PAS was 0.78 and of the ePAS 0.82 (difference ns). If cognitive decline was the outcome measure, the AUC of the PAS was 0.85 and of the ePAS 0.86 (difference ns). At the best cut-off, the positive predictive value of the PAS for cognitive decline was 92% with a sensitivity of 70%, and the positive predictive value of the ePAS was 96% with a sensitivity of 67%. Conclusions: Addition of an item for beta-amyloid and tau did not increase the predictive accuracy of the PAS. The good predictive accuracy of the PAS is consistent with previous studies and provides further evidence that the PAS is a useful tool for detecting subjects with predementia AD among subjects with MCI.