The radiologic prediction of Alzheimer disease: the atrophic hippocampal formation.

PURPOSE To test the hypothesis that atrophy of the hippocampal formation in nondemented elderly individuals would predict subsequent Alzheimer disease. METHOD We studied 86 subjects at two time points, 4 years apart. At baseline all study subjects were nondemented and included 54 control subjects and 32 persons who had memory complaints and minimal cognitive impairments. All subjects received a CT scan using a protocol designed to image the perihippocampal cerebrospinal fluid (HCSF) accumulating in the fissures along the axis of the hippocampal formation. Blind to the clinical evaluations, we subjectively assessed the presence of HCSF at the baseline. Retrospectively, we examined the predicted association between baseline HCSF and clinical decline as determined across the two evaluations. RESULTS At follow-up 25 of the 86 subjects had deteriorated and received the diagnosis of Alzheimer disease. Of the declining subjects, 23 came from the minimally impaired group, and 2 came from the control group. In the minimally impaired group the baseline HCSF measure had a sensitivity of 91% and a specificity of 89% as a predictor of decline. Both control subjects who deteriorated were also correctly identified at baseline. One of these two subjects died, and an autopsy confirmed the presence of Alzheimer disease. M(r) validation studies demonstrated that HCSF is quantitatively related to dilatation of the transverse fissure of Bichat and the choroidal and hippocampal fissures. CONCLUSION Our findings strongly suggest that among persons with mild memory impairments, dilatation of the perihippocampal fissures is a useful radiologic marker for identifying the early features of Alzheimer disease.