Background: We aimed to determine the specificity, sensitivity, and predictive value of apolipoprotein E (APOE) genotyping in 67 consecutive patients with clinical diagnoses of sporadic Alzheimer's disease (AD) who underwent necropsy.
Methods: We studied patients who attended the Duke Memory Disorders Clinic and were diagnosed as having probable AD. These patients were followed up until they died. APOE genotyping was done during life in most cases, but in some brain tissue obtained at necropsy was used. Members of known AD families were excluded.
Findings: After neuropathological examination 57 (85%) of 67 of our patients were confirmed as having AD including all 43 who had at least one APOE-epsilon 4 allele. None of the patients found not to have AD carried an epsilon 4 allele. In this series, the specificity of the epsilon 4 allele was 100%, the sensitivity 75%, the positive predictive value 100%, and the negative predictive value 42%. In this necropsy-confirmed series, the epsilon 4/epsilon 4 genotype predicted AD with 100% accuracy. The epsilon 3/epsilon 4 and epsilon 2/epsilon 4 genotypes were also unexpectedly highly specific for AD.
Interpretation: Data from hundreds of necropsy-confirmed non-AD patients in other longitudinal necropsy series will allow the predictive value of APOE genotypes to be assessed with useful confidence limits.