Dietary Fat Intake Determines the Effect of a Common Polymorphism in the Hepatic Lipase Gene Promoter on High-Density Lipoprotein Metabolism
Evidence of a Strong Dose Effect in This Gene-Nutrient Interaction in the Framingham Study
Background— Gene-nutrient interactions affecting high-density lipoprotein cholesterol (HDL-C) concentrations may contribute to the interindividual variability of the cardiovascular disease risk associated with dietary fat intake. Hepatic lipase (HL) is a key determinant of HDL metabolism. Four polymorphisms in linkage disequilibrium have been identified in the HL gene (LIPC), defining what is known as the −514T allele. This allele has been associated with decreased HL activity and increased HDL-C concentrations. However, the effect is variable among populations.
Methods and Results— We have examined interaction effects between the −514(C/T) LIPC polymorphism, dietary fat, and HDL-related measures in 1020 men and 1110 women participating in the Framingham Study. We found a consistent and highly significant gene-nutrient interaction showing a strong dose-response effect. Thus, the T allele was associated with significantly greater HDL-C concentrations only in subjects consuming <30% of energy from fat (P<0.001). When total fat intake was ≥30% of energy, mean HDL-C concentrations were lowest among those with the TT genotype, and no differences were observed between CC and CT individuals. We found similar gene-nutrient interactions when the outcome variables were HDL2-C (P<0.001), large HDL subfraction (P<0.001), or HDL size (P=0.001). These interactions were seen for saturated and monounsaturated fat intakes (highly correlated with animal fat in this population), but not for polyunsaturated fat.
Conclusions— Dietary fat intake modifies the effect of the −514(C/T) polymorphism on HDL-C concentrations and subclasses. Specifically, in the Framingham Study, TT subjects may have an impaired adaptation to higher animal fat diets that could result in higher cardiovascular risk.
Received July 18, 2002; revision received August 20, 2002; accepted August 21, 2002.
Low plasma high-density lipoprotein cholesterol (HDL-C) concentrations are a major risk factor for coronary artery disease (CAD). In general, HDL-C increases with physical activity, alcohol consumption, and diets high in saturated fatty acids (SFA). However, it is still important to reduce SFA intake because LDL-C increases even more. Conversely, smoking and replacement of dietary fat with carbohydrates decreases HDL-C1. Well-controlled experimental studies have demonstrated heterogeneity of the plasma lipid response to dietary fat and suggested a genetic component.2–4⇓⇓ Thus, gene-nutrient interactions affecting HDL metabolism may contribute to the substantial interindividual variability on the effects of low-fat diets on plasma lipids.2,4⇓ From a public health perspective, and considering that diet is the cornerstone of the prevention and treatment of CAD, in addition to the low-density lipoprotein cholesterol (LDL-C)-lowering factors, it is highly relevant to understand those involved on the dietary regulation of HDL metabolism.1
Hepatic lipase (HL) is a lipolytic enzyme that hydrolyzes triglycerides (TG) and phospholipids on plasma lipoproteins,5 and it has been suggested to be a novel CAD risk factor.6 HL also plays a role as a ligand mediating the binding and uptake of lipoproteins via proteoglycans and/or receptor pathways.7 Thus, both lipolytic and nonlipolytic functions of HL are important for HDL metabolism and ultimately to its association with CAD risk.8 Overall, overexpression of HL decreases HDL-C concentration, whereas HL deficiency increases it.9
A common −514 C to T substitution has been described in the promoter region of the HL gene (LIPC). The T allele is associated with decreased plasma HL activity, increased HDL-C concentrations,10–12⇓⇓ and increased large HDL subfractions.13 Although it is known that diet is an important factor influencing HDL-C concentrations and HDL subclasses, previous studies did not assess potential gene-diet interaction between the −541C/T polymorphism and dietary fat. Thus, the present study was designed to investigate how dietary fat intake, focusing on specific fatty acids and fat sources, interacts with the LIPC polymorphism to determine both HDL-C concentrations and particle size in subjects participating in the Framingham Study.
Subjects and Study Design
The design and methods of the Framingham Offspring Study (FOS) have been presented elsewhere.14 Starting in 1971, 5124 predominantly white subjects were enrolled. Blood samples for DNA were collected between 1987 and 1991. Lipids, genotypes, CHD risk factors, and dietary intake were determined for participants (n=3515) who attended the fifth examination visit, conducted between 1992 and 1995. HL genotyping was performed as previously described.13 Subjects taking lipid-lowering medications (n=166) were excluded from this analysis. The present study describes the results of analysis performed in 1020 men and 1110 women (aged 28 to 79 years) who had complete data for all variables examined (the most limiting variables being the data for HDL subfractions and LIPC genotypes). The sample studied did not differ from total population for the variables of interest.
Plasma Lipid, Lipoprotein, and Apolipoprotein Measurements
Sample collections and biochemical techniques were as described.15 HDL subclass distributions were determined by proton nuclear magnetic resonance (NMR) spectroscopy.16,17⇓ HDL subclass categories were grouped as large HDL (8.8 to 13.0 nm), intermediate HDL (7.8 to 8.7 nm), and small HDL (7.3 to 7.7 nm).13
Dietary intake was estimated with the semiquantitative Willett food-frequency questionnaire (136 food items).18,19⇓ Alcohol consumption was calculated and subjects were also classified as either nondrinkers or drinkers.19
Student’s t test and χ2 tests were applied to test differences in means and in percentages. To evaluate the relationship between the LIPC genotypes and dependent variables, we used analysis of covariance. To test the null hypothesis of no interaction between the LIPC genotypes and dietary fat, interaction terms were included in the models. First, with the use of Proc Mixed in SAS (v.8), we obtained crude results which accounted only for the familial relationships among the members of the study (mostly siblings and cousins).16 Second, we obtained adjusted results after accounting for familial relationships and potential confounding factors (sex, age, body mass index [BMI], smoking, alcohol, estrogens in women, β-blockers, and energy). Men and women were analyzed together because no heterogeneity of allelic effect by sex was detected. To depict the interaction effect between LIPC genotype and fat in determining HDL-C metabolism, predicted values for each dependent variable from the corresponding regression model were computed. As measure of the goodness-of-fit of the models, the square of the correlation coefficient was calculated. In addition, means in fat categories were added to figures as a second measure. All reported probability values were two-sided.
Table 1 shows anthropometric, biochemical, dietary, and genetic data. The prevalence of subjects with diabetes was 5.5%. Genotype frequencies did not deviate from the Hardy-Weinberg equilibrium. HDL-C concentrations were higher and HDL size increased in carriers of the T allele (Table 2), as previously described in this population.13 No significant differences in LDL-C and TG, dietary fat, or alcohol consumption across the LIPC genotypes were detected.
To test if dietary fat modifies the effect of the LIPC polymorphism, we first examined total fat intake categories according to the population mean (30% of energy). We found a strong interaction between the LIPC polymorphism and total fat consumption on HDL-C concentrations (P for interaction <0.001; Figure 1). The T allele was associated with an increase in HDL-C (8.7% in CT and 15.9% in TT; P<0.001) in subjects consuming <30% of energy from fat. However, mean HDL-C concentrations were lowest among those with the TT genotype and total fat intakes ≥30%, (−9.5%; P<0.05), and no differences were observed between CC and CT individuals. We found similar gene-nutrient interactions for HDL2-C (P<0.001), large HDL subfraction (P<0.001), and HDL size (P=0.001).
Furthermore, we examined the effect of total fat and specific fatty acids (FA) as continuous variables. Table 3 displays regression coefficients for main effects and for interaction terms between total fat or specific FA and LIPC genotypes in determining HDL-C concentrations. To increase interpretability, two separate regression models (fat-centered and noncentered) were fitted for each type of fat. We observed significant interaction (P<0.001) between total fat and LIPC genotype. Partial regression coefficients for the interaction effect in CT and TT individuals were both statistically significant as compared with CC individuals. We also found statistically significant interactions with SFA (P=0.002) and monounsaturated fatty acid (MUFA) (P=0.001). However, there was no significant interaction (P=0.36) with polyunsaturated fatty acid (PUFA). According to the noncentered regression model in Table 3, and to depict the interaction effect, Figure 2 shows the modification of the effect of the LIPC polymorphism on HDL-C (A) and HDL size (B) by the amount of total fat consumed. Differences in slope of the regression lines indicate that the effect of total fat on HDL-C concentrations or size depends on the LIPC genotype. Low fat intake was associated with higher HDL-C concentrations in TT individuals. However, in these subjects, HDL-C decreased as total fat intake increased. The opposite effect was found in CC subjects. Moreover, in CC subjects, HDL particle size was associated with higher fat intake, whereas increased fat intake decreased HDL particle size in CT and TT individuals. These effects on HDL metabolism were consistently found on different HDL subfractions even after controlling for potential confounders (Table 4).
As intakes of specific types of fat correlate one with another, we examined this correlation by source of fat (Table 5). MUFA and SFA were highly correlated with animal fat as a result of shared sources of SFA and MUFA in this American population. Consequently, we performed further analysis considering the source of fat (animal or vegetable). Multivariate regression models were fitted for animal and vegetable fat. Models were additionally adjusted for carbohydrates and for vegetable or animal fat. For animal fat, the statistical significance of the interaction terms with the LIPC polymorphism were P<0.001, P=0.003, P<0.001, and P=0.048 in determining HDL-C, HDL particle size, large-HDL subfraction, and intermediate+small HDL subfractions, respectively. The effect of animal fat on these variables was in the same direction as that for total fat. Adjusted slopes (in mmol/L per % of fat) in estimating HDL-C from animal fat were 0.014, P<0.001, in the CC genotype; 0.008, P=0.051 for difference from the CT genotype; and −0.017, P<0.001 for difference from the TT genotype. Figure 3 shows the interaction effect of the LIPC polymorphisms and animal fat on large-HDL subfractions (3A) and on intermediate+small HDL subfractions (3B). However, we did not find statistically significant interactions between vegetable fat and the LIPC polymorphism in determining HDL-C (P=0.400), HDL particle size (P=0.415), large-HDL (P=0.332), and intermediate+small HDL subfractions (P=0.738). Adjusted slopes (in mmol/L per % of fat) in estimating HDL-C from vegetable fat were 0.001, P=0.611, in the CC genotype; −0.004, P=0.161 for difference from the CT genotype; and −0.006, P=0.611 for difference from the TT genotype, all close to 0.
To gain further insights into the mechanisms by which total fat modulates the effect of the LIPC polymorphism on HDL metabolism, this interaction was analyzed for plasma LDL-C, TG, and fasting glucose concentrations. The statistical significance of the interaction terms between the LIPC polymorphism and total fat intake after control for confounders were P=0.944, P=0.154, and P=0.054 in determining LDL-C, TG, and glucose, respectively. No modification of the effect by fat was seen for LDL-C, although TG and glucose were affected. TT subjects showed higher concentrations of glucose (5.73 [0.23] mmol/L versus 4.98 [0.18] mmol/L) and TG (2.0 [1.9] mmol/L versus 1.74 [1.7] mmol/L) when consuming more than and less than 30% fat, respectively. No differences were found in CC (5.19 [0.05] mmol/L versus 5.17 [0.06] mmol/L) or CT subjects (5.20 [0.72] mmol/L versus 5.24 [0.71] mmol/L) for glucose, or TG (1.56 [0.05] mmol/L versus 1.73 [0.054] mmol/L) and (1.63 [0.07] mmol/L versus 1.72 [0.07] mmol/L), when consuming more than and less than 30% fat, respectively. However, when the association of diabetes with the LIPC polymorphism was examined, we found no differences in the percentage of diabetic subjects by genotype (5.67%, 5.44%, and 5.48% for CC, CT, and TT individuals, respectively, P=0.978). Finally, additional control for diabetes as a potential confounder was considered. The adjustment for diabetes had no effect at all on the previously obtained results.
In this population-based study from a well-characterized US cohort, we found a highly significant gene-nutrient interaction. The association of the −514C/T polymorphism with measures of HDL-C metabolism depended on the amount and type of fat consumed. Our results confirm previous findings in this13 and in other populations showing that the T allele is associated with higher HDL-C and HDL2-C concentrations, as well as large particle size10,12,20,21⇓⇓⇓ when no stratification by fat intake was considered. However, we found that the T allele was only associated with higher HDL-C and size in subjects consuming <30% of energy from fat. This gene-diet interaction could help explain some of the reports showing no association between the T allele and HDL-C.22–24⇓⇓ Several studies have reported that the T allele, which is in complete linkage disequilibrium with 3 other promotor polymorphisms (−250 G/A, −710 T/G, and −763 A/C), is associated with markedly decreased plasma HL activity,5,25⇓ which in turn may confer increased HDL-C concentrations and size. Moreover, its functionality has been demonstrated.26 HL is responsible for the lipolysis of both VLDL remnant particles and large LDL, as well as the conversion of HDL2 to HDL3 particles. In some studies,3,27⇓ but not in others,21,22⇓ an increase in HL activity has been related to an increase in small LDL particles. In our study, the increased HDL particle size related to the T allele was not significantly associated with large LDL particles,13 suggesting that other genetic or environmental factors may modulate this association. Moreover, there is contradictory experimental and epidemiological data about the association between HL activity and CAD.5 Dugi et al6 reported that low HL activity is a risk factor for CAD, consistent with the results of Jansen et al,25 which show that patients with genetic HL deficiency develop premature CAD, and with the observation that mice overexpressing HL have reduced atherosclerosis despite decreases in HDL-C.5,9⇓ These associations are paradoxical because it has been described that low HL activity leads to higher HDL-C and a more antiatherogenic lipid profile. This apparent contradiction may be resolved by other consequences of HL activity. Thus, an antiatherogenic role for the HL has also been described because HL activity correlates with postprandial lipid concentrations and the HL deficiency is associated with impaired clearance of lipoprotein remnants.8
We have limited knowledge about the genetic regulation of HL in humans. It has been described that even among −514T homozygotes, HL activity is quite variable, indicating that other factors modulate the allele effects on LIPC expression.27 HL activity is higher in men, in smokers, and in those with diabetes, and it increases with intra-abdominal fat and BMI.22,28⇓ However, information about the effects of dietary fat on HL activity is limited and contradictory. Campos et al29 described increases in lipoprotein lipase (LPL) and HL activities in subjects consuming a high-fat (46% of energy) compared with a low-fat (24% of energy) diet. Conversely, Kasim et al30 observed that LPL, but not HL activity, decreased during dietary fat restriction in women. Dreon et al3 reported that an increase in dietary SFA was associated with decreased HL activity, suggesting that diet-induced changes in HL may contribute to the regulation of large LDL particles. On the basis of our results, we hypothesize that the effect of dietary fat on HL activity depends on the −514C/T polymorphism. An important finding of the present study is the consistency of this interaction in determining several outcome variables related to HDL metabolism and to HL activity after controlling for possible confounders. Moreover, this strong interaction showed a clear dose-response effect. We found a gene-dosage effect when total fat was considered both as categorical and as a continuous variable. Increases in total fat were associated with increases in HDL-C concentrations in CC and CT individuals but with a different slope. In contrast, a similar increase in total fat intake in TT individuals was associated with a decrease in HDL-C. These results are in agreement with the general observation that HDL-C concentrations increase with diets high in fat2,30⇓ because of the high prevalence of CC and CT individuals in the general population. Moreover, we identify a specific subgroup (TT subjects) whose HDL-C may respond differently to changes in dietary fat intake than the majority of the population (CC subjects).
Less information is available about changes in HDL subclasses, which are associated with the amount of fat consumed. We found that higher total fat intake was associated with an increase in HDL particle size only in CC subjects, whereas the opposite was true for CT and TT individuals. Berglund et al2 investigated the effect of reduction of total and SFA on HDL concentration and size in a randomized study, showing that larger-sized HDL subpopulations decreased with the reduction of dietary fat. Considering that CC subjects make up the majority of white populations, their global results are in agreement with our observation in CC individuals.
When specific types of FA were analyzed, we found statistically significant interaction terms with SFA and MUFA, but not for PUFA, consistent with the concept that SFA and MUFA intake are associated with higher HDL-C.1,4⇓ Consistent with the observations that MUFA and SFA were highly correlated as a result of shared sources of these fats (ie, meat and dairy products) in our population, we only found statistically significant interactions for animal fat. The effect of this source of fat paralleled that previously described for total fat, and the effect of animal fat intake on larger- and smaller-sized HDL subpopulations depending on the LIPC genotype was more significant. Thus, in CC subjects, large-sized HDL decreased with the reduction in dietary animal fat, and a corresponding relative increase was seen for the smaller-sized HDL. The opposite was true for CT and TT. However, we cannot conclude that these results are exclusively a result of the fat component of the diet as other constituents of animal foods might play a role, particularly those that are fat soluble. Another issue that needs further consideration and that has not been addressed in the present article relates to the role of trans-fatty acids in the reported interactions.
The design of our study cannot address the mechanism by which dietary fat interacts with the −514C/T polymorphism. It has been reported that high glucose increases LIPC mRNA levels,31 providing a molecular foundation for a role of hyperglycemia in altered HDL metabolism. Previous studies have reported that insulin upregulates the activity of HL via insulin-responsive elements in the LIPC promoter, suggesting that variants in this promoter may affect the ability of insulin to stimulate HL activity.5 Jansen et al25 have reported an association between the LIPC promoter variants and insulin resistance, suggesting some potential mechanisms compatible with our results.
In summary, this highly significant gene-nutrient interaction may help to explain conflicting results with regard to HL activity, the antiatherogenic lipid profile and CAD risk, as well as the intra-individual differences in the plasma lipid response to the dietary fat. Most interesting is the fact that the observed effects are exclusively found for animal fat but not so for vegetable fat. This point requires further investigation with well-controlled experimental studies.
Supported by NIH/NHLBI grant No. HL54776, NIH/NHLBI contract No. 1-38038, and contracts 53-K06-5-10 and 58-1950-9-001 from the US Department of Agriculture Research Service, and grants PR2002-0115 (Dr D. Corella, University of Valencia) and PR2002-0116 (O. Coltell, Jaume I University) from the MECyD, Spain.
- ↵Mensink RP, Katan MB. Effect of dietary fatty acids on serum lipids and lipoproteins: a meta-analysis of 27 trials. Arterioscler Thromb. 1992; 12: 911–919.
- ↵Berglund L, Oliver EH, Fontanez N, et al. HDL-subpopulation patterns in response to reduction in dietary total and saturated fat intakes in healthy subjects. Am J Clin Nutr. 1999; 70: 992–1000.
- ↵Dreon DM, Fernstrom HA, Campos H, et al. Change in dietary saturated fat intake is correlated with change in mass of large-low-density-lipoprotein particles in men. Am J Clin Nutr. 1998; 67: 828–836.
- ↵Dugi KA, Brandauer K, Schmidt N, et al. Low hepatic activity is a novel risk factor for coronary artery disease. Circulation. 2001; 104: 3057–3062.
- ↵Lambert G, Amar MJA, Martin P, et al. Hepatic lipase deficiency decreases the selective uptake of HDL-cholesterol esters in vivo. J Lipid Res. 2001; 41: 667–672.
- ↵Dugi KA, Amar MJA, Haudenschild CC, et al. In vivo evidence for both lipolytic and nonlipolytic function of hepatic lipase in the metabolism of HDL. Arterioscler Thromb Vasc Biol. 2000; 20: 793–800.
- ↵Guerra R, Wang J, Grundy SM, et al. A hepatic lipase (LIPC) allele associated with high plasma concentrations of high density lipoprotein cholesterol. Proc Natl Acad Sci U S A. 1997; 94: 4532–4537.
- ↵Van’t Hooft FM, Lundahl B, Ragogna F, et al. Functional characterization of 4 polymorphisms in promoter region of hepatic lipase gene. Arterioscler Thromb Vasc Bio. 2000; 20: 1335–1339.
- ↵Vega GL, Clark LT, Tang A, et al. Hepatic lipase activity is lower in African American than in white American men: effects of 5′flanking polymorphism in the hepatic lipase gene. J Lipid Res. 1998; 39: 228–232.
- ↵Couture P, Otvos JD, Cupples LA, et al. Association of the C-514T polymorphism in the hepatic lipase gene with variations in lipoprotein subclass profiles. The Framingham Offspring Study. Arterioscler Thromb Vasc Biol. 2000; 20: 815–822.
- ↵Cupples LA, Gagnon DR, Kannel WB. Long- and short-term risk of sudden coronary death. Circulation. 1992; 85: 111–118.
- ↵Otvos JD, Jeyarajah EJ, Bennett DW, et al. Development of a proton nuclear magnetic resonance spectroscopic method for determining plasma lipoprotein concentrations and subspecies distributions from a single, rapid measurement. Clin Chem. 1992; 38: 1632–1638.
- ↵Freeman DS, Otvos JD, Jeyarajah EJ, et al. Relation of lipoprotein subclasses as measured by proton nuclear magnetic resonance spectroscopy to coronary artery disease. Arterioscler Thromb Vasc Biol. 1998; 18: 1046–1053.
- ↵Rimm EB, Giovannucci EL, Stampfer MJ, et al. Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol. 1992; 135: 1114–1126.
- ↵Corella D, Tucker K, Lahoz C, et al. Alcohol drinking determines the effect of the APOE locus on LDL-cholesterol concentrations in men: the Framingham Offspring Study. Am J Clin Nutr. 2001; 73: 736–745.
- ↵Grundy SM, Vega GL, Otvos JD, et al. Hepatic lipase activity influences high density lipoprotein subclass distribution in normotriglyceridemic men: genetic and pharmacological evidence. J Lipid Res. 1999; 40: 229–234.
- ↵Juo SH, Han Z, Smith JD, et al. Promoter polymorphisms of hepatic lipase gene influence HDL(2) but not HDL(3) in African American men: CARDIA study. J Lipid Res. 2001; 42: 258–264.
- ↵Shohet RV, Vega GL, Anwar A, et al. Hepatic lipase (LIPC) promoter polymorphism in men with coronary artery disease: allele frequency and effects on hepatic lipase activity and plasma HDL-C concentrations. Arterioscler Thromb Vasc Biol. 1999; 19: 1975–1978.
- ↵Jansen H, Verhoeven AJ, Weeks L, et al. A common C-to-T substitution at position -480 of the hepatic lipase promoter associated with a lowered lipase activity in coronary artery disease participants. Arterioscler Thromb Vasc Biol. 1997; 17: 2837–2842.
- ↵Deeb S, Peng R. The C-514T polymorphism in the human hepatic lipase gene promoter diminishes its activity. J Lipid Res. 2000; 41: 155–158.
- ↵Zambon A, Deeb S, Hokanson JE, et al. Common variants in the promoter of the hepatic lipase gene are associated with lower levels of hepatic lipase activity, buoyant LDL, and higher HDL2 cholesterol. Arterioscler Thromb Vasc Biol. 1998; 18: 1723–1729.
- ↵Campos H, Dreon DM, Krauss RM. Associations of hepatic and lipoprotein lipase activities with changes in dietary composition and low density lipoprotein subclasses. J Lipid Res. 1995; 36: 462–472.
- ↵Kasim SE, Martino S, Kim P-N, et al. Dietary and anthropometric determinants of plasma lipoproteins during a long-term low-fat diet in healthy women. Am J Clin Nutr. 1993; 57: 146–153.
- ↵Tu A-Y, Albers JJ. Glucose regulates the transcription of human genes relevant to HDL metabolism: responsive elements for peroxisome proliferator-activated receptor are involved in the regulation of phospholipid transfer protein. Diabetes. 2001; 50: 1851–1856.