Apolipoprotein E epsilon 2 allele and low serum cholesterol as risk factors for gastric cancer in a Chinese Han population

Apolipoprotein E (apoE) mediates lipid metabolism both in peripheral and in the brain. The human APOE gene has three polymorphic alleles that influence the risk for various types of cancer and neurodegenerative diseases. A potential association between APOE allele and the risk for gastric cancer has been implicated, but the specific allele involved and potential associations with the subtype and the grade of cancer malignancy need further clarification. We screened the APOE genotype in 550 gastric cancer patients and 550 non-cancer control individuals and found that the presence of the APOE ε2 and lower serum total cholesterol are associated with an increased risk for gastric cancer (all P ≤ 0.0005). Interestingly, APOE ε2 is also correlated with increased risk for both intestinal and diffuse histotypes but not with TN classification or stage in gastric cancer patients, suggesting that APOE polymorphic alleles are associated with the risk of development but unlikely the progression of gastric cancer. Since ε2 carriers have lower levels of serum total cholesterol than non-ε2 carriers, our findings suggest that the increased risk for gastric cancer by APOE ε2 allele might be mediated through lowered serum total cholesterol levels.

APOE gene polymorphism leads to an alteration in lipid and lipoprotein metabolism 16,17 . In general, compared to the individuals with the ε 3 allele, serum levels of total cholesterol, low-density lipoprotein cholesterol (LDL-cholesterol) and apolipoprotein B (apoB) tend to be lower for those with the ε 2 allele and higher for ε 4 carriers 16,[18][19][20] . Correlation between APOE gene polymorphism and high-density lipoprotein cholesterol (HDL-cholesterol) level was noted in some studies 16,21 , but not in others 20,22 . So far, a large number of cross-sectional and prospective studies have reported that low serum cholesterol levels are associated with higher risk for various cancers including gastric cancer [23][24][25][26] .
In our current study, we also observed a strong inverse association between serum total cholesterol levels and the risk for gastric cancer. In addition, we observed significant lower levels of total cholesterol, LDL-cholesterol and apoB in the ε 2 carriers which are consistent with previous reports. However, we only observed a trend of higher levels of total cholesterol, LDL-cholesterol in the ε 4 carriers. Intriguingly, the presence of the APOE ε 2 allele is also associated with higher risk for both intestinal and diffuse types of gastric cancer. Since ε 2 carriers have lower levels of serum total cholesterol than non-ε 2 carriers, our findings suggested that the increased risk for gastric cancer by APOE ε 2 allele might be mediated through lowered serum total cholesterol levels.

Results
Subject description. A summary of the characteristics for the 550 gastric cancer patients and the 550 cancer-free controls is displayed in Table 1. Median age was 63 years (Range: 18-87 years) in gastric cancer patients and 59 years (Range: 22-84 years) in controls. Male gender was most common in both gastric cancer patients (71.5%) and controls (69.1%). Median BMI was 21.5 (Range: 14.5-32.2) in gastric cancer patients and 24.4 (Range: 17.5-59.7) in controls. In gastric cancer patients, T classification was most commonly either T3 (48.8%) or T4 (33.9%); N classification was fairly evenly distributed but most commonly N3 (40.0%); and stage was most commonly III (61.3%). There are 286 patients (52.0%) and 264 patients (48.0%) for intestinal and diffuse gastric cancer subtype, respectively. Serum lipid parameters and the prevalence of APOE genotypes for both patients and controls are also summarized in Table 1.
Comparisons of demographic variables and serum lipid parameters according to the presence or absence of APOE ε4 and ε2 in all subjects. Comparisons of demographic variables and serum lipid profile according to the presence or absence of the APOE ε 4 and ε 2 alleles in all subjects are displayed in Table 2. After adjustment for multiple testing (P ≤ 0.005 considered significant), there were differences between subjects with and without a copy of ε 4 regarding apoB (Median: 0.99 vs. 0.93, P = 0.0006) and apoA1/apoB ratio (Median: 1.22 vs. 1.31, P = 0.0001). There were significant differences between subjects with and without a copy of ε 2 in the overall sample regarding total cholesterol (Median: 4.72 vs. 5.09, P = 0.0006), LDL-cholesterol (Median: 2.94 vs. 3.35, P < 0.0001), apoB (Median: 0.87 vs. 0.96, P < 0.0001), and apoA1/apoB ratio (Median: 1.47 vs. 1.27, P < 0.0001).
Evaluation of risk factors for gastric cancer. An evaluation of risk factors for gastric cancer is provided in Table 3. In single variable analysis without adjusting for potential confounding variables, there was strong evidence of an association with increased risk of gastric cancer for older age (P < 0.0001) and decreased BMI (P < 0.0001), but no association was evident for gender (P = 0.39). Additionally, risk of gastric cancer was significantly higher for individuals with lower levels of triglycerides (P < 0.0001), total cholesterol (P < 0.0001), HDL-cholesterol (P < 0.0001), LDL-cholesterol (P < 0.0001), apoA1 (P < 0.0001), and apoB (P < 0.0001). There was a trend toward an increased risk of gastric cancer for subjects with a copy of the APOE ε 2 allele; however, this did not survive correction for multiple testing (P = 0.032). There was no statistically significant evidence of an association with risk of gastric cancer for APOE ε 4 (P = 0.87) or apoA1/B ratio (P = 0.087).
In multivariable analysis we adjusted for age, BMI, and other variables (see Table 3 legend for details). In this multivariable analysis, there was still strong evidence of an association with increased risk of gastric cancer for age (P < 0.0001), BMI (P < 0.0001), total cholesterol (P < 0.0001), HDL-cholesterol (P < 0.0001), apoA1 (P < 0.0001), and apoA1/B ratio (P < 0.0001). Additionally, the association between APOE ε 2 and increased risk of gastric cancer strengthened and became significant after adjustment for multiple testing (P = 0.0004). Associations with gastric cancer for triglycerides (P = 0.061), LDL cholesterol (P = 0.35) and apoB (P = 0.48) weakened in multivariable analysis and were no longer even nominally significant, while the lack of association of gender and APOE ε 4 with risk of gastric cancer remained unchanged (P ≥ 0.30). Intriguingly, when examining the association between APOE ε 2 and gastric cancer subtype, an increased risk of gastric cancer was noted in ε 2 carriers for both intestinal gastric cancer (P = 0.017) and diffuse gastric cancer (P = 0.0006) ( Table 4).
Analysis of the associations of APOE genotypes with TN classification and stage in gastric cancer patients. Associations of APOE ε 4 and ε 2 with T classification, N classification, and stage in gastric cancer patients are displayed in Table 5. There was no evidence of an association between ε 4 and T classification (P = 0.38), N classification (P = 0.96), or stage (P = 0.59), and also no evidence of an association between ε 2 and T classification (P = 0.63), N classification (P = 0.78), or stage (P = 0.23).

Discussion
Our case-control study of 550 gastric cancer patients and 550 cancer-free controls evaluated for the first time the effects of APOE gene polymorphism and serum cholesterol levels on the risk of gastric cancer among a Chinese Han population. Our results show that the presence of APOE ε 2 allele and low serum levels of total cholesterol, HDL-cholesterol, apoA1 and apoA1/B ratio are associated with an increased risk of gastric cancer. Interestingly, APOE ε 2 is also correlated with increased risk for both intestinal and diffuse histotypes. However, APOE genotypes do not appear to influence the invasion and metastasis of gastric cancer as reflected by TN classification and stage.
Scientific RepoRts | 6:19930 | DOI: 10.1038/srep19930 In gastric cancer, serum cholesterol levels have been reported to inversely correlate with the risk of disease 25,27,28 . In our current study, we also observed a strong inverse association between serum total cholesterol levels and the risk of gastric cancer. To date, the underlying mechanism for such an association remains unclear. Previous reports have shown that the serum levels of total cholesterol, LDL-cholesterol and apoB levels are associated with APOE genotypes, with individuals carrying the ε 2 allele having lower and those carrying the ε 4 displaying higher levels compared to the more common ε 3 allele 16,[18][19][20] . In our studied population, we observed significant lower levels of total cholesterol, LDL-cholesterol and apoB in the ε 2 carriers which are consistent with previous reports. Additionally, we observed a significant higher level of apoB in the ε 4 carriers. However, we only observed a trend of higher levels of total cholesterol, LDL-cholesterol in the presence of ε 4 allele ( Table 2). Since ε 2 carriers have lower levels of serum total cholesterol than non-ε 2 carriers, our findings suggest that the increased risk effect for gastric cancer by APOE ε 2 allele might be mediated through lowered total cholesterol level.
The effect of APOE genotypes on the risk of numerous cancers has been previously investigated, including breast, prostate, ovarian, colorectal, head and neck cancer [29][30][31][32][33][34][35][36] . However, the results have been conflicting which might be attributed to limited sample sizes or the inherent differences among different ethnic populations. Our current study in a Chinese Han population observed that the presence of ε 2 allele is associated with an increased risk for gastric cancer with OR of 2.34 in multivariable analysis. In contrast, a previous study reported a protective  effect of the APOE ε 2 allele against gastric cancer in an Italian Caucasian population 37 . The discrepancy between these two studies might be attributed to the different ethnic groups that were studied, or potentially due to unmeasured confounding variables. Future extension of these findings is necessary to clarify and further understand the association between APOE ε 2 and gastric cancer risk in both ethnic populations. ApoE has recently been identified as a potential tumor-associated marker in gastric cancer from gene expression analysis 4,38,39 . Higher apoE expression was found in gastric cancer, particularly in advanced T and N grades. We therefore examined the association between APOE genotypes and T classification, N classification, and stage in gastric cancer patients. Our data showed no significant association between APOE genotypes and various classifications or stage of gastric cancer, suggesting that APOE polymorphic alleles are associated with the risk of development but unlikely the progression of gastric cancer.
In summary, our study confirmed the associations of lower levels of serum cholesterol with the incidence of gastric cancer in a Chinese Han population. Importantly, our study reports for the first time that APOE ε 2 as a risk allele for both intestinal and diffuse types of gastric cancer in our studied population, which might be partly attributed to the lower serum total cholesterol of ε 2 carriers compared to those with ε 3 or ε 4 alleles.

Methods
Approval for this study was obtained from the Ethics Committees of the Xiamen University and Fujian Medical University Union Hospital. All experiments were performed in accordance with the approved guidelines. All individuals who participated in this study gave written informed consent.

Study subjects.
A total of 550 gastric cancer patients with histological confirmation and 550 cancer-free controls seen at the Fujian Medical University Union Hospital in China between 2011 and 2014 were included in this case-control study. Information was collected regarding age, gender, body mass index (BMI), and serum lipid parameters which include the levels of triglycerides, total cholesterol, HDL-cholesterol, LDL-cholesterol, apoA1, apoB, and apoA1/apoB ratio. None of the patients underwent pre-operative chemotherapy or radiation therapy. Histopathological evaluations were performed with reference to the Japanese Classification of Gastric Carcinoma, 3 rd English edition 40 . APOE genotyping and quality control. Genomic DNA was extracted from peripheral blood using a DNA extraction kit (Zeesan Biotech, Xiamen). Genotyping of the two APOE SNPs (rs429358:T/C; rs7412:T/C) was carried out using the APOE SNP genotyping kits (Memorigen Biotech, Xiamen, China) and the Applied Biosystems ® 7500 real-time PCR Systems (Applied Biosystems, Foster City, CA). Data analysis was performed by measuring the allele-specific fluorescence. As a measure for quality control, three samples with known APOE genotypes were included in each assay. Additionally, the genotyping analysis was blinded to the subject's case or control status. Finally, 10% of the total samples were randomly selected and retested with 100% concordance. There was no evidence of a departure from Hardy-Weinberg Equilibrium in study controls for either rs429358 (P = 0.79) or rs7412 (P = 0.71).

Statistical analysis.
Continuous variables were summarized with the sample median, minimum, 25 th percentile, 75 th percentile, and maximum. Categorical variables were summarized with number and percentage of patients. Due to the small number of subjects with APOE ε 4/ε 4 and ε 2/ε 2 genotypes, we utilized a dominant model in analyzing both APOE ε 4 (presence vs. absence of the ε 4 allele) and ε 2 (presence vs. absence of the ε 2 allele) in all analysis. Comparisons of demographic variables and serum lipid parameters according to the presence of the APOE ε 4 or ε 2 allele were made using a Wilcoxon rank sum test or Fisher's exact test. Risk factors for  Table 3. Evaluation of risk factors for gastric cancer. OR = odds ratio; CI = confidence interval. ORs, 95% CIs, and p-values result from logistic regression models. Multivariable models were adjusted for age and BMI, as well as other variables that were associated with risk of gastric cancer in single variable analysis with a p-value gastric cancer were evaluated using single variable (i.e. unadjusted) and multivariable logistic regression models. Multivariable models were adjusted for all variables that were associated with risk of gastric cancer in single variable analysis with a p-value of 0.10 or lower, although there were some exceptions to this due to the high degree of correlation between many of the variables of interest and the resulting potential for collinearity (see Table 3 legend for details). Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated. For easier interpretation of results, all continuous variables were categorized based on approximate sample quartiles for use in association analysis. The association between APOE ε 2 and gastric cancer subtype was evaluated using single variable (i.e. unadjusted) and multivariable logistic regression models. Multivariable models were adjusted for all variables that were associated with risk of gastric cancer in single variable analysis with a p-value of 0.10 or lower (see Table 4 of 0.10 or lower. Because many of these other variables were very highly correlated with one another which results in high potential for collinearity, all variables satisfying this criteria could not always be adjusted for in all models, and therefore model adjustments in multivariable analysis were as follows: 1    legend for details). Associations of APOE ε 4 and ε 2 with TN classification and stage in gastric cancer patients were examined using a Wilcoxon rank sum test. In order to adjust for multiple testing, we utilized a Bonferroni correction separately for each group of similar tests. Specifically, p-values ≤ 0.0050 were considered as significant when comparing demographic variables and serum lipid parameters according to presence of ε 4 and ε 2; p-values ≤ 0.0042 were considered as significant when evaluating risk factors for gastric cancer; p-values ≤ 0.0167 were considered as significant when evaluating associations of ε 4 and ε 2 with TN classification and stage in gastric cancer patients; and p-values ≤ 0.025 were considered as significant when evaluating the association between APOE ε 2 and gastric cancer subtype. All statistical analysis was performed using R Statistical Software (version 2.14.0; R Foundation for Statistical Computing, Vienna, Austria).