The neuropathological diagnosis of Alzheimer's disease: clinical-pathological studies.

The neuropathological diagnosis of Alzheimer's disease currently relies on quantitative or semiquantitative criteria of senile or neuritic plaques that are adjusted for age and for the presence or absence of a clinical history of dementia. Based on clinical-pathological correlation studies, I will argue that neuropathological assessment should stand independently of clinical history and instead should describe brain lesions in the context of the topography and natural history of the disease. Only probabilistic estimates about the presence or absence of dementia can be made from a neuropathological examination, especially in the setting of Alzheimer disease lesions plus other pathological alterations such as Lewy bodies or infarcts. Moreover, I will argue that any neurofibrillary tangles or senile plaques are inherently pathological entities, even if clinically silent and so "incidental" neuropathological findings. Because the intensity and location of neurofibrillary tangles, rather than senile plaques, appears to correlate most closely with clinical symptoms, I suggest using a staging system that highlights this information rather than using absolute numerical cut-offs for diagnostic purposes.