The lipoprotein apolipoprotein E (APOE) is mainly produced in the liver and is involved in cholesterol metabolism. APOE has three isoforms (E2, E3, and Ε4), distinguished by its lipid-binding activity.1 The most common isoform, E3, is considered the normal form. Compared with E3, E2 is defective in receptor-binding activity and is associated with lower low-density lipoprotein (LDL) cholesterol, whereas Ε4 has high binding activity with very-low-density lipoprotein (VLDL) and low binding activity with high-density lipoprotein (HDL). These differences in the function of APOE are determined by two single nucleotide polymorphisms (SNPs) within the APOE genes: rs429358 and rs7412.

APOE polymorphism due to the two SNPs results in three major alleles (E2, E3, and E4) and six genotypes (E2/E2, E2/E3, E2/E4, E3/E3, E3/E4, and E4/E4). Compared with E3 allele carriers, E4 allele carriers have higher risks of dementia2 and cardiovascular disease,3 while the E2 allele may induce hyperlipidemia in conditions such as diabetes and hypothyroidism and are associated with low risks of neurodegenerative disease.24 Furthermore, APOE polymorphism is associated with longevity. In cross-sectional studies,5 the E4 allele frequency decreased with age, while the Ε3 allele frequency increased; centenarian studies demonstrated similar results.6

However, the association between APOE polymorphism and mortality is inconsistent. In many studies, the E4 allele was associated with high mortality,789101112 whereas the E2 allele was associated with low mortality.7911 In other studies,13141516 these associations were not significant. Although several cross-sectional studies have reported associations between APOE polymorphism and longevity,171819 few large-scale cohort studies of Asians have been conducted. Therefore, in this study, the effect of APOE polymorphism on mortality was investigated in two, large population-based cohort studies of Asians.

Table 1 presents the participants' baseline characteristics. The 9,045 participants in the Dong-gu Study were followed for 8.5 ± 1.8 years; among them, 1,070 (11.8%) died during this period. The 10,350 participants in the Namwon Study were followed for 11.5 ± 2.3 years, and 1,399 (13.5%) died during this period. In both studies, the serum triglyceride levels were lowest in participants with the E3 allele, and the serum total cholesterol levels were lowest in participants with the E2 allele. The serum HDL cholesterol level was lowest in participants with the E4 allele. The baseline characteristics of the study population according to the APOE genotypes (E2/E2, E2/E3, E2/E4, E3/E3, E3/E4, and E4/E4) are shown in Supplementary Tables 1 and 2.

Fig. 1 shows the Kaplan–Meier survival curve for the APOE alleles. Compared with the E3 allele, the E2 allele was associated with low mortality, and the E4 allele was non-significantly associated with high mortality.

Table 2 shows the hazard ratios (HRs) and 95% confidence intervals (CIs) according to APOE genotype, with the E3 allele used as the reference. In Model 1, the HR for the E2 allele was 0.86 (95% CI, 0.75–0.99) and that for the E4 allele was 1.05 (95% CI, 0.94–1.16); the respective values were 0.84 (95% CI, 0.74–0.96) and 1.08 (95% CI, 0.97–1.20) in Model 2. To evaluate the mediating effect of lipid profiles, we further adjusted for lipid profiles in Model 3. Consequently, the HR for the E2 allele decreased slightly to 0.80 (95% CI, 0.70–0.92), while that for the E4 allele remained unchanged (HR, 1.08; 95% CI, 0.97–1.20). Compared with the E3/E3 genotype, the HRs of E2/E2, E2/E3, E2/E4, E3/E4, and E4/E4 were 1.00 (95% CI, 0.50–2.01), 0.80 (95% CI, 0.69–0.92), 1.05 (95% CI, 0.74–1.51), 1.09 (95% CI, 0.97–1.21), and 1.01 (95% CI, 0.66–1.55) in Model 3, respectively (Supplementary Table 3).

Table 3 shows the HR and 95% CI according to the subgroups. No significant difference was found among the subgroups in the association between APOE allele and mortality. The difference in HR among the APOE genotypes tended to be greater in the Dong-gu Study than in the Namwon Study. The HRs of the E2 allele were 0.75 (95% CI, 0.61–0.93) and 0.85 (95% CI, 0.71–1.02) in the Dong-gu and Namwon Studies, respectively, while those of the E4 allele were 1.10 (95% CI, 0.94–1.29) and 1.07 (95% CI, 0.93–1.23), respectively. The risk difference according to the APOE alleles tended to be greater in women than in men, and greater in participants younger than 65 years than in older participants. In participants younger than 65 years, the HR of the E2 allele was 0.70 (95% CI, 0.52–0.92) and that of the E4 allele was 1.25 (95% CI, 1.03–1.53), whereas in the older participants, the HRs were 0.85 (95% CI, 0.72–0.99) and 1.03 (95% CI, 0.91–1.17) in the E2 and E4 alleles, respectively. The HRs of the E2 allele were 0.80 (95% CI, 0.68–0.95) and 0.80 (95% CI, 0.64–1.00) in men and women, respectively, and those of the E4 allele were 1.04 (95% CI, 0.90–1.19) and 1.15 (95% CI, 0.97–1.36), respectively.

In our study, the E2 allele was negatively associated with mortality, while the Ε4 allele was positively, but non-significantly, associated with mortality. There was no mediating effect of serum lipids on this association; unexpectedly, the effect of E2 allele increased slightly.

In previous studies, the association between APOE polymorphism and mortality was varied according to the size and age of the study population. The association between the E4 allele and mortality was not significant in small studies or studies with insufficient follow-ups.13141516 In large studies and studies with sufficient follow-ups,781011 the E4 allele was associated with high mortality. Studies with participants with the E4 allele, aged 90 years or older, indicated a significantly higher risk of mortality.91012 However, despite the large sample size in our study, the E4 allele did not confer a significantly higher mortality risk than the E3 allele.

Previous studies have consistently shown that the E2 allele is associated with a lower risk of death than the E3 allele.71011 Rajan et al.11 and Rosvall et al.7 reported that this association was significant. In a meta-analysis of centenarians,10 the probability of surviving more than 105 years, or the oldest 0.1 percentile, was significantly higher in those with the E2 genotype compared with those with the E3 genotype. However, in the Danish 1905 Cohort Survey of participants older than 92 years,12 the E2 allele did not indicate a significantly lower mortality rate than the other alleles. The protective effect of the E2 allele is primarily explained by its association with low LDL cholesterol levels.1 However, in our study, contrary to our expectations, the HR strengthened from 0.84 to 0.80, when we adjusted for the lipid profile. In contrast to a previous study using the same dataset as that used in the present study, we reported an association between carotid atherosclerosis and APOE polymorphism,24 in which adjusting for the lipid profile attenuated this association.

These discrepancies in the association between APOE alleles and mortality in previous studies can be attributed to three factors. First, the genetic influence may depend on age. As mentioned above, the older the study population, the smaller is the effect of E2 and the greater is the effect of E4. Jacobsen et al.9 evaluated changes in the association between the E4 allele and mortality with a follow-up period and reported that the longer the follow-up period, the greater was the HR of the E4 allele. By contrast, in our study, the HR of the E4 allele was higher in the younger groups. Next, the association between APOE polymorphism and mortality can vary depending on the APOE genotype classification method. The APOE genotype can be divided into either E2, E3, and E4 according to the allele type or into E4 carrier and E4 non-carrier according to the E4 carrier type. Previous studies comparing risk differences between E4 carriers and non-carriers8914 overestimated the risk of the E4 allele because the E4 non-carriers included not only E3 carriers but also E2 carriers, who exhibited a lower mortality risk than the E3 carriers. Finally, the frequency of the E4 allele varies by population and is affected by environmental variables such as latitude and temperature.25 The E4 allele frequency of 14.5%–27.9% has been reported, which is higher than 8.6% in our study.7911121415 The statistical power of our study was insufficient to detect an HR of 1.08, which was slightly lower than the HR (1.10 to 1.22) reported in other studies,7101112 because of the low frequency of the E4 allele. Studies based on Asian populations reported a low E4 allele frequency similar to our study, and the association between the E4 allele and mortality was non-significant.1316

This is the largest individual study evaluating the association between APOE genotypes and mortality, which allows us to assess this association according to APOE allele types (E2, E3, or E4). Nevertheless, our study presents several limitations. First, additional genetic information that could affect APOE was not included. Although the rs429358 and rs7412 considered in our study affect the function of APOE, other DNA positions in the APOE may also affect its function.26 Next, because the follow-up period was relatively short in this study, the association between APOE and mortality by cause of death could not be determined. Additionally, interactions with variables such as age and gender could not be obtained.

In conclusion, the death risk differed according to the APOE polymorphism in Korean adults. The E2 allele was associated with a lower risk of mortality than the E3 allele, whereas the E4 genotype was not associated with mortality in Koreans.