Diagnosis and staging of Alzheimer disease

Rather than determining lesions ''threshold'' between ''normal'' cases and patients, we prefer to use clinicopathological correlations, assigning a given intellectual deficit to a given amount of lesions with a chosen level of probability. Because large amounts of A beta diffuse deposits may be found in the absence of dementia, we think advisable not to take them into account for the diagnosis. The diffusion of the neurofibrillary tangles in the paralimbic, limbic and isocortical areas (described by Braak and Braak stages or by the number of areas containing tangles) and the density of isocortical senile plaques (A beta focal deposits) as assessed by the CERAD protocol are both correlated with the intellectual status but give complementary information. They should thus be jointly used. We analyzed the variability of the lesions counts, their coefficients of error, and their causes, as a first step toward standardization. We have shown, however, that semiquantitative estimates ale presently more reproducible than quantitative measures. (C) 1997 Elsevier Science Inc.