Subjects. A total of 1,017 individuals over 70 years of age were randomly selected using district registries in Kadiköy municipality of Istanbul and screened using a validated Turkish version of Mini Mental Status Examina­tion (MMSE) (two different versions for literate, and illit­erate individiuals). All individuals at or below the cut-off score of 24 for literate and 23 for illiterate, and a ran­domly selected 8% of those above the cut-off, were exam­ined by a team of neurologists, and AD was diagnosed according to DSM IV and NINCDS-ADRDA criteria [26].
A total of 281 subjects (223 “screen-positive” and 58 “screen-negative”) underwent detailed clinical, and if deemed necessary, laboratory examination [26]. Of these, 57 subjects were diagnosed as probable AD, and 11 sub­jects as possible AD. These two groups were pooled for analysis. The demographic characteristics are shown in Table 1. Of the remainder, 127 subjects were diagnosed as cognitively normal and constituted the control group, while the rest received diagnoses such as depression, slight cognitive impairment without dementia, vascular dementia and other degenerative dementias and were pooled with the rest of the population (i.e., those who were not clini­cally examined) and labeled as “others”.
ApoE Genotyping. DNA samples were collected from cheek epithelial cells with buccal-swabs. ApoE genotyp­ing was initially performed by the conventional restriction fragment length polymorphism (RFLP) analysis and by the LightCycler-ApoE Mutation Detection System (Roche Molecular Biochemicals, Germany; http://biochem.roche. com). Both methods were used in parallel in the first 226 samples. Once 100% concordance between the methods was confirmed, the remaining samples (n = 791) were analyzed only by the LightCycler method, which is faster and more efficient [27,28].
Restriction Fragment Length Polymorphism Analy­sis. A 227 bp region of the ApoE gene was amplified by polymerase chain reaction (PCR). The PCR products were then digested by the Hin6I restriction enzyme, and the digested products subjected to polyacrylamide gel electro­phoresis (PAGE) and the DNA bands visualized by silver staining [27].
LightCycler-Apo E Mutation Detection System. Dual-color genotyping was performed using the Light Cycler-Apo E Mutation Detection Kit (Roche Molecular Biochemicals, Cat. No. 3004716). Amplification and melting curve analysis protocols were applied as previ­ously described [28].

Statistical Analysis. The λ² test was used for compari­son of allele distributions in the AD and control popula­tions. The strength of association between AD and ApoE genotype, gender, age and education level was described by odds ratios (OR) and 95% confidence intervals (CI). In order to assess the presumed risk constituted by the E4 allele and presumed protective effects of the E2 allele, subjects carrying at least one E4, or at least one E2 allele, were grouped together, while those carrying an E2/E4 genotype were excluded. Multiple logistic regression anal­ysis was used to obtain adjusted effect estimates. The likelihood ratio λ² test was used to analyze the signifi­cance of OR in the multivariate analyses. Significance tests were administered two-sided.

Table 1. Demographic characteristics of the populations studied.