Prediction of APOE genotype by phenotype in Alzheimer's disease

Despite the remarkable success of genetic linkage and association studies in establishing mutations in four different genes which cause or predispose to Alzheimer's disease, little progress has been made in the clinical subtyping of Alzheimer's disease based on genotype. In the current study, we have applied latent class analysis to standardized clinical data on a cohort of over 750 elderly individuals (average age 72 years, 58% female) with or without Alzheimer's disease. Using a split sample analytic strategy, we have found that two to four latent classes can be reliably identified using only 10 questionnaire items from a much larger data set. These latent classes predict clinical disease status (no, mild, moderate, or severe Alzheimer's disease) and number of copies of the ApoE4 allele. There were no significant associations of latent class membership with age, sex, medical problems, blood pressure, other psychiatric syndromes, or neurological disorders. These results suggest that a modest screening effort may be sufficient to identify individuals with Alzheimer's disease and to predict genotype status.