The epsilon4 allele of the gene encoding apolipoprotein E (APOE) is strongly associated with Alzheimer's disease, but its value in the diagnosis remains uncertain. We reviewed clinical diagnoses and diagnoses obtained at autopsy in 2188 patients referred to 1 of 26 Alzheimer's disease centers for evaluation of dementia. The sensitivity and specificity of the clinical diagnosis or the presence of an APOE epsilon4 allele were calculated, with pathologically confirmed Alzheimer's disease used as the standard. The added value of the APOE genotype was estimated with pretest and post-test probabilities from multivariate analyses to generate receiver-operating-characteristic curves plotting sensitivity against the false positive rate. Of the 2188 patients, 1833 were given a clinical diagnosis of Alzheimer's disease, and the diagnosis was confirmed pathologically in 1770 patients at autopsy. Sixty-two percent of patients with clinically diagnosed Alzheimer's disease, as compared with 65 percent of those with pathologically confirmed Alzheimer's disease, had at least one APOE epsilon4 allele. The sensitivity of the clinical diagnosis was 93 percent, and the specificity was 55 percent, whereas the sensitivity and specificity of the APOE epsilon4 allele were 65 and 68 percent, respectively. The addition of information about the APOE genotype increased the overall specificity to 84 percent in patients who met the clinical criteria for Alzheimer's disease, although the sensitivity decreased. The improvement in specificity remained statistically significant in the multivariate analysis after adjustment for differences in age, clinical diagnosis, sex, and center. APOE genotyping does not provide sufficient sensitivity or specificity to be used alone as a diagnostic test for Alzheimer's disease, but when used in combination with clinical criteria, it improves the specificity of the diagnosis.
