Recent genetic epidemiological studies1-3 have documented a significant association between the apolipoprotein E (ApoE) allele ε4 (ApoE-ε4) and late-onset familial and sporadic Alzheimer's disease (AD). This association is so impressive that one group of investigators concluded that "homozygosity for ApoE-ε4 was virtually sufficient to cause AD by age 80."1 However, the bane of any epidemiological association is bias.4 We suggest that selection bias may have significantly contributed to, or even produced, the observed association between ApoE-ε4 and late-onset AD.
The three common ApoE alleles are ε2, ε3, and ε4. The relative frequencies of these alleles in healthy population samples are about 8%, 76%, and 14%, respectively.5-6 Individuals with ApoE-ε4 have an increased risk of premature coronary artery disease.7-9 As a result of earlier mortality due to heart disease, the relative frequency of the ε4 allele was predicted, and found, to decline with age.10-11 This