Detecting dementia: novel neuropsychological markers of preclinical Alzheimer's disease

Abstract

The results of a previous study have suggested that impaired performance on one neuropsychological test, CANTAB Paired Associates Learning (PAL), may serve as a marker for preclinical Alzheimer's disease (AD). In a group of individuals with 'questionable dementia', the baseline PAL performance was found to correlate significantly with subsequent deterioration in global cognitive function over an 8-month period. The present paper reports diagnostic outcome data for the same individuals 32 months after the first assessment and evaluates the predictive diagnostic utility of baseline neuropsychological measures. Thirty-two months after joining the study, 11 of the 43 'questionable dementia' patients met the criteria for probable AD diagnosis ('converters') and 29 remained free from AD ('non-converters'). Logistic regression analysis revealed that two tests of memory, in combination, could be used to predict a later diagnosis of probable AD with a high level of accuracy [chi(2)(3) = 47.054, p < 0.0001]. As predicted, these tests are measures of visuospatial learning (CANTAB PAL) and, also, semantic memory (Graded Naming Test). These two tests in combination appear to be highly accurate in detecting cognitive dysfunction characteristic of preclinical AD. Using these tests, a simple algorithm is described for calculating, with 100% accuracy for this sample of 40 patients, the probability that an individual with mild memory impairments will go on to receive a diagnosis of probable AD.