点击显示 收起
1 From the Departments of Clinical Nutrition and Medicine, University of Kuopio, Kuopio, Finland.
See corresponding editorial on page 669.
ABSTRACT
Background: The E4 allele of the apolipoprotein gene (APOE) is associated with a greater serum cholesterol response to dietary changes in fat and cholesterol. However, less is known about the interaction between APOE polymorphism and other macronutrients in the diet.
Objective: We evaluated the interaction between APOE polymorphism and dietary fat and carbohydrate, particularly sucrose, in relation to serum lipid concentrations.
Design: A total of 284 men and 130 women with coronary artery disease (mean age: 61 y; range: 33–74 y) participated in the cross-sectional EUROASPIRE study. Serum lipids and fatty acids in cholesteryl esters (CEs) were measured and APOE genotypes were determined. Dietary intake was examined by using a 4-d food record.
Results: Patients were grouped by APOE genotype: E2 (E2/E2 and E2/E3; n = 21), E3 (E3/E3; n = 245), and E4 (E4/E2, E4/E3, and E4/E4; n = 148). Patients with the E2 allele had lower LDL-cholesterol concentrations and tended to have higher triacylglycerol concentrations than did patients with the E3 or E4 allele; concentrations were not significantly different between the last 2 groups. In regression analysis, significant predictors of serum triacylglycerol were the interaction between sucrose intake and the E2 allele, proportion of n-3 fatty acids in CEs, body mass index, and diabetes. A high sucrose intake was associated with high triacylglycerol concentrations only in patients with the E2 allele. Interaction between saturated fat intake and the E2 allele, proportion of linoleic acid in CEs, and fiber intake predicted serum cholesterol.
Conclusion: Coronary artery disease patients with the E2 allele will likely have a greater triacylglycerol response to high dietary sucrose intakes than will patients with the E3 or E4 allele.
Key Words: Apolipoprotein E • APOE polymorphism • diet • serum lipids • triacylglycerol • cholesterol • coronary artery disease • EUROASPIRE • European Action on Secondary Prevention through Intervention to Reduce Events
INTRODUCTION
Apolipoprotein (apo) E is a protein associated with the metabolism of triacylglycerol-rich lipoproteins and HDLs and that serves as a ligand for LDL receptors and LDL-receptor-related proteins. Three common alleles of the apolipoprotein gene (APOE) have been described: APOE*E2 (E2), APOE*E3 (E3), and APOE*E4 (E4). In population studies, plasma apo B and total and LDL concentrations are lowest in subjects with the E2 allele, intermediate in those with the E3 allele, and highest in those with the E4 allele (1, 2). High triacylglycerol concentrations are observed in both E2 and E4 carriers (2). The E4 allele is associated with a greater risk of coronary artery disease in both men and women (3).
Several intervention studies were conducted to investigate the interaction between APOE polymorphism and lipid and lipoprotein responses to changes in fat and cholesterol intakes (reviewed by Ordovas; 1). Some investigators also focused on the relation between APOE polymorphism and the effect of fiber intake on serum lipids (4, 5). The main focus of these studies was the response of serum cholesterol; the results were conflicting. Some studies showed that the E4 allele carriers were more responsive to changes in dietary fat and cholesterol than were carriers of the E2 or E3 allele (6–8), whereas other studies showed no differences in response between the allele groups (9, 10).
Few observational studies of diet-lipid associations according to APOE genotype have been conducted. The main finding in a study by Marshall et al (11) was that the positive relation between dietary cholesterol and serum total and LDL cholesterol was strongest in non-Hispanic whites with the E3/E3 genotype. The ratio of linoleic acid to oleic acid in cholesteryl esters (CEs), a biomarker of dietary fat quality, was inversely associated with serum triacylglycerol more strongly in subjects with the E2 allele than in subjects with the E4 allele or the E3/E3 genotype (12). In a population-based sample of free-living children and young adults, the effect of APOE polymorphism on serum lipids was affected by diet in such a way that the serum cholesterol–lowering effect of the E2 allele was larger if the dietary intakes of saturated fat and cholesterol were high (13). In addition, the presence of the E4 allele potentiated the serum cholesterol–raising effect of high dietary intakes of saturated fat and cholesterol (13), in accordance with a dietary intervention study in subjects with different APOE genotypes (8).
Most of the previous observational and interventional studies conducted focused on the effect of the genetic variation in APOE on lipoprotein responses to dietary fat and cholesterol, whereas less is known about the effect of other macronutrients in this respect. We conducted a cross-sectional study to investigate the interaction between APOE polymorphism and dietary intakes of fat and carbohydrate, particularly sucrose, on serum cholesterol and triacylglycerol concentrations. The study was performed in patients with coronary artery disease whose dietary intakes and serum lipid fatty acid profiles were reported previously (14).
SUBJECTS AND METHODS
Subjects
EUROASPIRE (European Action on Secondary Prevention through Intervention to Reduce Events) was a survey on the management of risk factors and the use of cardiovascular drugs in secondary prevention of coronary heart disease (15). According to the EUROASPIRE study protocol, consecutive male and female patients aged <71 y with coronary artery disease were identified from the hospital discharge lists and coronary angiography register of the Kuopio University Hospital from the following 4 diagnostic categories: 1) patients with their first elective or emergency coronary artery bypass grafting (CABG), 2) patients with their first elective or emergency percutaneous transluminal coronary angioplasty (PTCA) but with no previous CABG, 3) patients with their first or recurrent acute myocardial infarction (AMI) but with no previous CABG or PTCA, and 4) patients with acute myocardial ischemia (AMIS) but no evidence of AMI and no previous CABG, PTCA, or AMI.
Altogether, 125 consecutive patients with CABG, PTCA, and AMIS and 156 consecutive patients with AMI were identified. Patients hospitalized before 1 November 1994 were invited for an interview and examination 6 mo after hospitalization. From patient categories 1–4, respectively, 109, 106, 101, and 99 patients participated: 1, 4, 20, and 0 patients died and 15, 15, 35, and 26 patients did not participate (some did not respond to the invitation, some refused to participate, and for some, travel was impractical). The median time intervals between hospital admission and the interview and examination were 1.0 (range: 0.8–1.3), 1.9 (0.9–4.0), 2.3 (0.9–3.7), and 2.2 (1.0–3.8) y in patient groups 1–4, respectively. The study was approved by the Ethics Committee of the University of Kuopio and the Kuopio University Hospital. All patients gave their informed consent.
Methods
Height and weight were measured and body mass index (BMI; in kg/m2) was calculated. Blood pressure was measured with an automatic digital sphygmomanometer. Data on current medication use were obtained in an interview. Patients were classified as diabetic if the diagnosis of diabetes had previously been confirmed by a physician. Blood samples were collected between 0800 and 1000 after the subjects had fasted for 12 h. Lipoproteins were separated by ultracentrifugation for 12 h at a density of 1.006 kg/L to remove VLDL. LDL was precipitated from the infranatant fluid with dextran sulfate–magnesium chloride (16). HDL cholesterol was analyzed from the remaining supernatant fluid. LDL cholesterol was calculated by subtracting the amount of HDL cholesterol from the amount of cholesterol in the infranatant fluid containing both HDL and LDL cholesterol. Cholesterol in the whole serum and in separated lipoproteins and serum triacylglycerols were analyzed by using commercial kits (kits 237574 and 701904; Boehringer GmbH, Mannheim, Germany) and Kone Specific Clinical Analyzer (Kone Ltd, Espoo, Finland).
APOE genotypes were analyzed with use of a slightly modified version of Tsukamoto et al's (17) polymerase chain reaction–restriction fragment length polymorphism method. Serum samples were stored at -70°C before the measurement of fatty acids in serum CEs as described previously (18). Briefly, lipids were extracted from the serum sample with chloroform:methanol (2:1, by vol) and lipid fractions were separated with an aminopropyl column. Fatty acids of CEs were transmethylated with boron trifluoride and analyzed with a gas chromatograph (model 5890 series II; Hewlett-Packard Company, Waldbronn, Germany) equipped with an FFAP column and helium as a carrier gas. Fatty acids are presented as molar percentages of total fatty acids.
Patients completed a 4-d food record (3 weekdays and 1 weekend day) at home according to detailed written instructions. The amount of food was estimated by using a portion-size booklet (19). At the interview, all amounts and qualities of foods in the records were checked by a clinical nutritionist for completion, and missing data were obtained and added. The analyses of nutrients were made by using the MICRO-NUTRICA dietary analysis program (version 2.0; Finnish Social Insurance Institute, Turku, Finland), which is based on the national database of the Finnish Social Insurance Institute.
Statistical methods
Statistical analyses were performed by using the SPSS program (version 8.0; SPSS Inc, Chicago). Normal distribution of variables was checked with the Kolmogorov-Smirnov (Lilliefors) test; those variables that were not normally distributed were log transformed. The differences in basic characteristics, nutrient intakes, and fatty acids in CEs between the APOE groups were analyzed with analysis of variance (ANOVA) and chi-square tests. Post hoc multiple comparisons were performed with Tukey's test after ANOVA and with pairwise chi-square tests followed by Bonferroni correction. Each dietary variable was entered separately in a stepwise multiple regression analysis by using an level of 0.1 for a variable to enter the model. Interaction terms, calculated by multiplying dietary variables by dummy variables of APOE genotype, APOE group, and confounding factors (use of lipid-lowering medication, BMI, and age) were also entered into the regression analyses.
Dietary variables that were tested in the regression analyses of the association with serum cholesterol included total fat, saturated fat, cholesterol, fiber, and palmitic acid, linoleic acid, and n-3 fatty acids in CEs; those tested in the analysis of the association with serum triacylglycerol were total fat, carbohydrate, sucrose, and palmitic acid, linoleic acid, and n-3 fatty acids in CEs. The variables that were independently associated with serum lipids were entered into a full regression model. The full regression model explaining serum triacylglycerol (log transformed) included intake of sucrose and its interaction term with the E2 allele, n-3 fatty acids in CEs, APOE group, diabetes, and relevant confounding factors. The full regression model explaining total cholesterol included intakes of saturated fat, cholesterol and fiber; linoleic acid in CEs; and their interaction terms with the E2 allele, APOE group, and confounding factors. Differences in serum triacylglycerol concentrations according to tertiles of sucrose intake, after stratification by APOE group, were analyzed by the Kruskal-Wallis test because of non-normal distribution and heterogeneous variances of triacylglycerol across the tertiles. For differences in serum cholesterol in tertiles of saturated fat intake, ANOVA was used to adjust for the use of lipid-lowering medications, BMI, and age. The results for continuous variables are expressed as means ± SDs. P values <0.05 (two tailed) were considered statistically significant.
RESULTS
Patients were grouped by APOE genotype: E2 (E2/E2 and E2/E3; n = 21, 5.1%), E3 (E3/E3; n = 245, 59.2%), and E4 (E4/E2, E4/E3, and E4/E4, n = 148, 35.7%) because of similar serum lipid profiles among the genotypes. The allele frequencies were 0.039 for E2, 0.757 for E3, and 0.204 for E4. The basic characteristics of the subjects according to APOE group are given in Table 1. Age, BMI, and blood pressure did not differ significantly between the allele groups. LDL cholesterol was lower and total cholesterol tended to be lower in patients with the E2 allele than in patients with the E3 or E4 allele. Serum triacylglycerol tended to be higher in patients with the E2 allele. The use of lipid-lowering medication was slightly less frequent in patients with the E2 allele, but the difference was not significant. There were more diabetic patients with the E2 allele than with the E3 or E4 allele.
View this table:
TABLE 1. . Basic characteristics of subjects according to APOE group
The intake of cholesterol was higher in patients with the E2 allele than in patients with the E4 allele and the intake of fiber was highest in patients with the E4 allele (Table 2). The intake of saturated fat tended to be higher in patients with the E2 allele, but the difference was not significant. Otherwise, there were no significant differences in the intake of nutrients between the groups. The proportions of myristic, palmitic, palmitoleic, and oleic acids in CEs were higher and the proportion of linoleic acid in CEs was lower in patients with the E2 allele than in patients with the E3 or E4 allele (Table 3).
View this table:
TABLE 2. . Nutrient intake according to APOE group1
View this table:
TABLE 3. . Fatty acid profile of serum cholesteryl esters according to APOE group1
Results of regression analyses with nutrients entered separately as predictors of serum triacylglycerol are presented in Table 4. The interaction term between sucrose intake and the E2 allele was positively associated with serum triacylglycerol (ß = 0.018, P = 0.007). The proportion of palmitic acid in CEs was positively associated and that of linoleic acid and n-3 fatty acids in CEs was inversely associated with serum triacylglycerol. The interaction terms between nutrients and the E4 allele were not related to serum triacylglycerol concentration. In the full regression model, significant predictors of serum triacylglycerol were the interaction between sucrose and the E2 allele, the proportion of n-3 fatty acids in CEs, BMI, and diabetes (Table 5). This model explained 16.8% of the variance in serum triacylglycerol.
View this table:
TABLE 4. . Regression coefficients of dietary variables and fatty acids in serum cholesteryl esters (CEs) and their interaction terms with the E2 allele on serum triacylglycerol and serum total cholesterol concentrations1
View this table:
TABLE 5. . Full regression models for prediction of serum triacylglycerol and serum total cholesterol concentrations1
The intake of dietary cholesterol was independently and positively associated with serum total cholesterol in the regression analysis (Table 4). Fiber intake and the proportion of linoleic acid in CEs were inversely associated with total cholesterol. Interaction terms between the E2 allele and intakes of fat, saturated fat, cholesterol, and fiber were inversely associated with total cholesterol. The interaction terms between nutrients and the E4 allele were not associated with serum total cholesterol. Significant inverse predictors of total cholesterol in the full regression model were the proportion of linoleic acid in CEs, the interaction term between saturated fat and the E2 allele, fiber intake, and the use of lipid-lowering medication (Table 5). The percentage of variance in total cholesterol explained by this model was 12.7%.
A high intake of sucrose was associated with high serum triacylglycerol only in patients with the E2 allele, irrespective of diabetes (Figure 1). Note that the diabetic patients were unevenly distributed across the tertiles of sucrose intake in the E2 group (lowest tertile: 6 diabetic patients and 1 nondiabetic patient; middle tertile: 1 diabetic patients and 6 nondiabetic patients; and highest tertile: 2 diabetic and 5 nondiabetic patients). In addition, the frequency of use of lipid-lowering medications did not differ significantly between the diabetic and nondiabetic patients with the E2 allele (diabetic patients: 1 user and 8 nonusers; nondiabetic patients: 5 users and 7 nonusers).
FIGURE 1. . A: Mean (±SD) serum triacylglycerol concentrations according to tertiles of sucrose intake (, tertile 1; , tertile 2; , tertile 3) after stratification by APOE genotype group; E2 (E2/E2 and E2/E3), E3 (E3/E3), and E4 (E4/E2, E4/E3, and E4/E4). n in brackets. P for trend (Kruskal-Wallis test) = 0.035 for the E2 allele and 0.007 for the E4 allele. B: Association of serum triacylglycerol concentration with sucrose intake in diabetic (solid line, •; n = 9) and nondiabetic (dotted line,; n = 12) patients with the E2 allele.
In patients with the E4 allele, high triacylglycerol concentrations were observed in those with a low intake of sucrose. However, the association between sucrose intake and serum triacylglycerol became nonsignificant after adjustment for BMI in patients with the E4 allele. Serum total cholesterol concentrations tended to be highest in the highest tertile of saturated fat intake only in patients with the E3 allele (NS; data not shown).
DISCUSSION
The purpose of this cross-sectional study was to investigate the interaction between the APOE polymorphism and dietary fat and carbohydrate in modifying serum lipid concentrations. The main result was that a high sucrose intake was associated with a high serum triacylglycerol concentration in coronary artery disease patients with the E2 allele. Only patients with the E3 allele showed a tendency to a cholesterol-raising effect of a high saturated fat intake. It should be emphasized that these patients may have modified their diets before the study.
Consistent with previous findings, serum total and LDL cholesterol were lower and triacylglycerol concentrations were higher in patients with the E2 allele than in patients with the E3 or E4 allele (1, 2). This was observed even though dietary intakes of saturated fat and cholesterol were higher and intakes of fiber were lower in patients with the E2 allele. Patients with the E4 allele did not have higher cholesterol or triacylglycerol concentrations than did patients with the E3 allele in the present study. Besides possible dietary modifications before the study began, another reason for this finding could be that patients with the E4 allele were somewhat more likely to use lipid-lowering medications, although this difference was not significant. APOE allele frequencies did not differ from those reported previously for Finnish patients with coronary artery disease (20) and for unrelated Finnish individuals (21).
To our knowledge, no previous studies reported an interaction between the E2 allele and sucrose intake on serum triacylglycerol. BMI, age, sex distribution, intake of alcohol, and proportion of n-3 fatty acids in CEs did not differ significantly between the tertiles of sucrose intake in patients with the E2 allele; therefore, these confounders did not explain the association. Because type 2 diabetes was shown in the present study and in previous studies (22) to be associated with high VLDL-triacylglycerol concentrations, we examined whether diabetes was associated with the hypertriglyceridemic effect of sucrose. The number of diabetic patients was unevenly distributed across the tertiles of sucrose intake, but, because of a small sample size, separate analyses for diabetic and nondiabetic patients among the E2 allele carriers were not possible. Because higher triacylglycerol concentrations were observed in the nondiabetic patients among the E2 allele carriers, it is unlikely that diabetes accounted solely for the observed association, as indicated in Figure 1B. In addition, differences in the use of lipid-lowering medications between diabetic and nondiabetic patients were not significantly different and thus did not explain the association between the E2 allele and the triacylglycerol-elevating effect of sucrose.
High triacylglycerol concentrations were observed when carbohydrate intakes increased, but triacylglycerol concentrations returned to original or near original concentrations in the long-term (23, 24). Subjects with initially elevated triacylglycerol concentrations or with hyperlipoproteinemia may be more prone to a greater response to dietary sugars. The E2 allele is associated with a delayed catabolism of chylomicron and VLDL remnants, which leads to hypertriglyceridemia (25). The present finding implies that the E2 allele may be a factor modulating the triacylglycerol response to dietary sucrose in the long-term. Consistent with the triacylglycerol concentration, high VLDL-cholesterol (P = 0.029) and high VLDL-triacylglycerol (P = 0.069) concentrations were also associated with a high sucrose intake in patients with the E2 allele (data not shown). In a cross-sectional study, young Finns were divided into 3 groups according to dietary intakes of saturated fat and cholesterol (13). The E2 allele was related to higher triacylglycerol concentrations only in those with low saturated fat and cholesterol intakes and concomitantly high sucrose intakes. E2 allele carriers were shown to have a greater cholesterol response to dietary changes in fiber intake than were noncarriers of the E2 allele (5). Note that the overall intake of sucrose was low in the subjects in the present study.
Overall, the association between dietary saturated fat intake and serum cholesterol in the present study was weak. Only in patients with the E3 allele was there a tendency to higher serum cholesterol concentrations with high saturated fat intakes. This finding is in contrast with many intervention studies in which the E4 allele carriers were more responsive to changes in saturated fat and cholesterol intakes (1, 8). In an observational study by Marshall et al (11), there was no interaction between genetic variation in APOE and saturated fat intake that modified serum lipid concentrations. Instead, high cholesterol intake was associated with high serum cholesterol in non-Hispanics with the E3 allele, but not in the E2 or E4 allele carriers. In cross-sectional studies by Reilly et al (26) and Boer et al (12), the E2 allele carriers deviated from the other allele groups by having stronger associations between adiposity and serum lipid concentrations.
Associations of dietary factors and serum lipid concentrations were in the direction expected on the basis of previous studies. However, the proportion of variance that was explained (R2) by the full regression models was low. The association may be weakened by other factors affecting serum lipids (random variation) and measurement errors in dietary assessment. Four-day diet records are, however, sufficient for gross characterization of usual individual intakes of energy and macronutrients (27). The difference in saturated fat intakes between the groups was reflected in the fatty acid profile of serum CEs: the proportion of saturated fatty acids was higher and the proportion of linoleic acid was lower in patients with the E2 allele, who also tended to have higher intakes of saturated fat. Further studies are needed to clarify whether APOE genotype should be considered when making dietary modifications to reduce serum triacylglycerol concentrations. In conclusion, coronary artery disease patients with the E2 allele will likely have a greater triacylglycerol response to high dietary sucrose intakes than will patients with the E3 or E4 allele.
ACKNOWLEDGMENTS
We thank Minna Kiuttu for skillful technical assistance.
REFERENCES