点击显示 收起
1 From the Departments of Human Nutrition (FLC, CMS, and TJG) and Preventive and Social Medicine (ARG), University of Otago, Dunedin, New Zealand
2 Supported by a research grant from the National Heart Foundation of New Zealand. 3 Reprints not available. Address correspondence to CM Skeaff, Department of Human Nutrition, University of Otago, PO Box 56, Dunedin, New Zealand. E-mail: murray.skeaff{at}stonebow.otago.ac.nz.
ABSTRACT
Background: The results of randomized controlled trials indicate that the amount and type of dietary fat are important predictors of serum cholesterol concentrations. However, the results of observational studies show weak or no association between dietary fat intake and serum cholesterol. Serum fatty acids are valid biomarkers of fat intake and may improve dietary estimates.
Objective: The objective was to ascertain whether serum fatty acids are associated with serum cholesterol concentrations in New Zealand adolescents and adults.
Design: The current study was a cross-sectional, national, population-based survey of 2793 New Zealanders aged 15 y who participated in the 1997 National Nutrition Survey. The fatty acid composition of serum cholesterol esters, phospholipids, and triacylglycerols was measured.
Results: A 1-SD increase in myristic acid (14:0) in serum cholesterol ester, phospholipids, and triacylglycerol corresponded with increases in serum cholesterol of 0.19, 0.13, and 0.10 mmol/L, respectively, after adjustment of the regression analysis for sex, age, body mass index, ethnicity, and smoking. The mean difference in cholesterol concentrations between persons in the highest and the lowest quintiles of serum cholesteryl-myristate was 0.48 mmol/L (P for trend < 0.001). A 1-SD increase in the proportion of linoleic acid (18:2n–6) in serum cholesterol ester, phospholipids, and triacylglycerol corresponded with decreases in serum cholesterol of 0.07, 0.07, and 0.05 mmol/L, respectively. The difference in mean serum cholesterol between the highest and lowest quintiles of cholesteryl-linoleate was 0.18 mmol/L (P for trend = 0.019).
Conclusion: Saturated and polyunsaturated fat intakes, measured by using fatty acid biomarkers, are important predictors of serum cholesterol concentrations in New Zealand.
Key Words: Dietary fats • cholesterol • fatty acids • biological markers • nutrition survey
INTRODUCTION
Serum cholesterol concentrations predict coronary artery disease risk in persons and populations (1, 2). Lowering cholesterol concentrations results in a meaningful reduction in disease risk (3). Dietary approaches to reducing serum cholesterol have proven efficacy (4, 5) and effectiveness (6, 7); a reduction in saturated fat intake is a key feature of these regimens. Results of controlled feeding trials show a strong and independent effect of dietary saturated fat; a reduction of intake by 5% of total energy reduces serum cholesterol by 0.25–0.28 mmol/L (4, 5, 7). A concomitant increase in polyunsaturated fat intake by 5% of total energy provides additive serum cholesterol lowering of 0.12–0.13 mmol/L (4, 5). Individual saturated fatty acids do not all have the same cholesterolemic effects. Myristic acid (14:0), of which dairy products are a rich source, has a strong serum cholesterol–raising effect, as do palmitic (16:0) and lauric (12:0) acids (8, 9).
Despite the proven effect of saturated fat on serum cholesterol concentrations (4, 5, 7) and results from the Seven Countries Study, which showed that a large part of the variation in serum cholesterol concentrations between countries can be explained by the difference in saturated fat consumption (10), the results of most prospective cohort studies have found weak or no association between saturated fat intakes and serum cholesterol concentrations (11-15). The latter observation has been attributed to the imprecision of dietary assessment and a higher intraindividual variation in fat intake relative to the interindividual variation (16, 17). Nevertheless, the inconsistency between the results of intervention and observational studies has led to uncertainty about the relevance of dietary fat as a predictor of serum cholesterol concentrations in populations.
The fatty acid compositions of cholesterol esters, phospholipids, and triacylglycerols are objective biological markers of fat intake (18-20). They reflect actual rather than reported intakes and thus are unaffected by respondents' underreporting or overreporting of fat intakes that is typical of diet records and recalls (21). Moreover, serum fatty acids are particularly good markers of myristic and linoleic acid intakes; in controlled trials, the dietary intakes of these fatty acids are the major dietary predictors of serum cholesterol concentrations (8, 9).
The purpose of the current study was to use serum fatty acid composition as a biomarker of dietary fat intakes to ascertain at a population level the association between dietary fat and serum cholesterol concentrations.
SUBJECTS AND METHODS
Subjects
The 1997 National Nutrition Survey (NNS97) sample was drawn from the 1996/1997 New Zealand Health Survey, a population-based survey that assessed the health status of noninstitutionalized New Zealand adolescents aged 15 y and adults. The detailed methods of the NNS97 were published elsewhere (22-24). Briefly, the survey used an area-based sampling frame with a 3-stage stratified design that consisted of a selection of a set of primary sampling units, households within these populations' sampling units, and a randomly chosen respondent within a household. The response rate for the 1996/1997 New Zealand Health Survey was 73.8% (n = 7862). Of that group, 4636 gave dietary information in the NNS97, and 3223 (69.5%) of that group agreed to give blood samples. The fatty acid results of at least one lipid fraction were available for 2793 participants; thus, the fatty acid analysis represents 35.5% of the original 1996/1997 New Zealand Health Survey.
Written informed consent was obtained from all participants or the guardian of those <18 y old. The 14 ethics committees throughout New Zealand granted ethical approval for the survey.
Methods
The NNS97 was undertaken over a 12-mo period from December 1996 through November 1997. Dietary data were collected from each participant during a home visit by using a computer-assisted 24-h recall. Ethnicity was self-reported and categorized into 3 groups: New Zealand Maori, Pacific People (also known as Pacific Islanders), and New Zealand European and other (NZEO). When the participant stated that he or she belonged to more than one ethnic group, a single ethnic category was assigned to that participant by using a priority system as follows. If Maori was one of the groups reported, the participant was assigned to the Maori group. If any of the Pacific People groups was one of the groups reported, the participant was assigned to the Pacific People group. In this report, the term Pacific People includes those who identify themselves as being Samoan, Tongan, Cook Island Maori, Niuean, Tokelauan, Fijian, or members of other Pacific Ocean–based ethnic groups. All remaining participants were assigned to the NZEO group.
During the home visit, height and weight were measured according to standard techniques; body mass index (BMI; in kg/m2) was calculated. Subjects with a BMI < 25, 25–29.9, and 30 were categorized normal-weight, overweight, and obese, respectively. Participants were asked a set of questions about their smoking habits and categorized as nonsmokers (a group that included former smokers) or current smokers. During the home visit, blood was drawn from subjects, across a range of fasting and postprandial states, from an antecubital vein into vacuum evacuated tubes with no anticoagulant. The tubes were centrifuged and the serum was aliquoted into cryovials for storage at –80 °C. Serum total and HDL-cholesterol concentrations were measured as described by Skeaff et al (25).
Lipids were extracted from 400 µL of serum according to the method of Bligh and Dyer (26). An internal standard was added before extraction. The serum lipids were separated by using thin-layer chromatography with a solvent system of hexane/diethyl ether/acetate (85:15:1 by volume). The lipid bands were visualized under ultraviolet light after the plates were sprayed with a solution of 0.1% (wt:vol) 8-anilio-1-napthalene sulfonic acid. Serum cholesterol ester, triacylglycerol, and phospholipid bands were scraped into glass test tubes and methylated in 6% sulfuric acid in methanol for 2 (for cholesterol esters and triacylglycerols) or 12–16 (phospholipids) h. The samples were then eluted into hexane and stored at –20 °C.
The fatty acid methyl esters were analyzed by using a DB-225 column (30 m x 0.25 mm internal diameter; film thickness, 0.25 µm; J&W Scientific, Deerfield, IL) installed on an HP6890 Series Gas Chromatograph (Agilent, Palo Alto, CA) with flame ionization detection. Blank samples were extracted, separated, and analyzed, and the peak areas were subtracted from the corresponding areas of the sample runs after adjustment for internal standard recovery had been made. Precision of the fatty acids was determined by analyzing pooled plasma in 1 pooled sample for every 20 NNS97 samples. The CVs for the fatty acids myristic, linoleic, and palmitoleic acid in serum cholesterol esters, were 15.0%, 3.0%, and 16.5%, respectively.
Statistical analysis
All statistical analysis was carried out on STATA software (version 8.0; Stata Corp, College Station, TX) by using the survey commands when appropriate to control for the survey design of the study. In the planned analysis, we used multiple linear regression to examine the relation of serum myristic and linoleic acids with serum cholesterol. Another 10 fatty acids were chosen, and their relations with cholesterol concentrations were also examined; a significant association was found only for palmitoleic acid (16:1n–9). The confounding continuous variable, age, was coded into a categorical variable (ages 15–18, 19–24, 25–44, 45–64, and 65 y) and, along with sex, BMI, ethnicity, and smoking status were used as indicator variables in the model to calculate the ß coefficients for the 3 fatty acids (myristic, linoleic, and palmitoleic acids) with serum total and HDL cholesterol. Interactions between sex and age, sex and BMI, and sex and ethnicity were also tested in the models. A P value < 0.05 was considered statistically significant.
Standardized z scores were created for myristic, linoleic, and palmitoleic acids by subtracting the unadjusted mean from each individual value and dividing by the unadjusted SD. These values were then used to calculate standardized ß coefficients, which represent the change in serum cholesterol associated with each 1-SD change in fatty acid composition. Bootstrapping was used to test whether the proportion of myristic acid in serum cholesterol ester, triacylglycerol, or phospholipids was the strongest predictor of total cholesterol. One thousand bootstrap datasets, of a sample size equal to the original, were generated by random selection of participants with replacement. For each dataset, the standardized ß coefficients from 3 fully adjusted models, one for each lipid fraction, were calculated for the relation between myristic acid and total cholesterol. The characteristics and serum lipid concentrations of participants by quintiles of myristic, linoleic, and palmitoleic acid in the 3 lipid fractions were calculated. The linear trends for these variables across the quintiles of fatty acids were tested by using multiple linear regression. Linear prediction was used to calculate the multivariate adjusted values of total and HDL-cholesterol concentrations and the ratio of total to HDL cholesterol for the quintiles of fatty acids.
RESULTS
The fatty acid composition of serum cholesterol esters, triacylglycerols, and phospholipids were completed for 2393, 2402, and 2416, respectively, of the 3223 stored serum samples from the NNS97. The fatty acid composition of 1 serum lipid fraction was analyzed for 2793 participants. There were no significant differences between the age, BMI, ethnicity, dietary fat intakes, or lipid concentrations of participants included in the fatty acid analysis (Table 1) and those of participants in the original NNS97 (23).
View this table:
TABLE 1. Characteristics of participants1
The adjusted ß coefficients and the standardized ß coefficients that describe the relation between serum fatty acids and total cholesterol are presented in Table 2. The proportion of myristic acid in all 3 serum lipid fractions was positively associated with total cholesterol concentrations. A 1-SD increase in cholesteryl-myristate (0.44 mol%) corresponded to a 0.19 mmol/L increase in total cholesterol. The proportion (and 95% CI) of the 1000 bootstrapped datasets in which the lipid fraction had the highest standardized ß coefficient for myristic acid was 89.9% (95% CI: 87.9%, 91.7%) for cholesteryl, 0.2% (0.02%, 0.7%) for triacylglycerol, and 9.9% (8.1%, 11.9%) for phospholipids. The change in the proportion of linoleic acid in cholesterol ester, triacylglycerol, and phospholipids also predicted cholesterol concentrations; the standardized ß coefficients in the 3 fractions varied little and were roughly half those of myristic acid. A 1-SD change in palmitoleic acid in both cholesterol ester and triacylglycerol corresponded to a change of 0.15 mmol/L in serum total cholesterol. There was no significant association between palmitoleic acid in phospholipids and serum cholesterol concentrations.
View this table:
TABLE 2. ß coefficients for the relation between serum lipid fatty acids (mol%) and total cholesterol concentrations (mmol/L)
When participants were categorized by quintile of increasing proportion of myristic acid in serum cholesterol ester, there was a significant trend across the quintiles to greater age, a decreasing proportion of Pacific People, and an increasing proportion of NZEO subjects (Table 3). The proportion of dietary energy (in kJ) as saturated and dairy fat increased by 2.1% and 4.4%, whereas that as polyunsaturated fat decreased by 1.6% from the lowest to the highest quintile of cholesterol ester myristate. Serum cholesterol concentrations increased across the quintiles of cholesterol ester myristate; the mean serum cholesterol concentration was 0.95 mmol/L (15%) higher in the highest quintile than in the lowest quintile. The trend was attenuated but remained significant after adjustment for age, sex, BMI, ethnicity, and smoking; the difference between the 1st and 5th quintiles was 0.48 mmol/L (8%). There was no significant relation between HDL-cholesterol concentrations and cholesteryl-myristate before or after adjustment for confounders. Serum total:HDL cholesterol increased across the quintiles of myristic acid in cholesterol esters and remained significant after adjustment; the mean ratio was 0.47 higher in the highest quintile than in the lowest quintile.
View this table:
TABLE 3. Characteristics of participants in the National Nutrition Survey 1997 by quintile of myristic acid in serum cholesterol esters1
Across the quintiles of serum cholesterol ester linoleate (Table 4), a higher proportion of linoleic acid was associated with lesser age, and participants were more likely to be nonsmokers and either Maori or Pacific People. The trends in dietary fat were for polyunsaturated fat to be an increasing proportion of energy intake in the higher quintiles of cholesterol ester linoleate, whereas saturated and dairy fat intakes decreased from the 1st to 5th quintile by 1.2% and 2.0% of energy, respectively. The results of the univariate analysis showed an inverse association between serum cholesteryl-linoleate and cholesterol concentrations; the difference between the highest and lowest quintiles was 0.46 mmol/L. The difference was attenuated to 0.18 mmol/L after multivariate adjusted analysis. The trend for HDL cholesterol across the quintiles of linoleic acid was not significant before or after adjustment for confounding variables. There was a significant inverse trend for total:HDL cholesterol across the quintiles of linoleic acid; after adjustment, however, it was no longer significant.
View this table:
TABLE 4. Characteristics of participants in the National Nutrition Survey 1997 by quintile of linoleic acid in serum cholesterol esters1
The characteristics of participants by quintile of palmitoleic acid in cholesterol esters are shown in Table 5. The proportion of females increased significantly across the quintiles of cholesteryl-palmitoleate, as did age, BMI, dairy fat, total, HDL, and total:HDL cholesterol. There was an inverse association between polyunsaturated fat intake and palmitoleic acid. No significant trend was observed between saturated fat intake and palmitoleic acid. After adjustment for confounding variables, the associations between cholesteryl-palmitoleate and total and HDL cholesterol weakened slightly but remained significant: the multivariate adjusted difference between quintiles 5 and 1 was 0.41 and 0.11 mmol/L, respectively. After adjustment, there was no significant relation between total:HDL cholesterol and palmitoleic acid.
View this table:
TABLE 5. Characteristics of participants in the National Nutrition Survey 1997 by quintile of palmitoleic acid in serum cholesterol esters1
The quintiles of myristic, linoleic, and palmitoleic acids in the triacylglycerol and phospholipid fractions also showed results similar to those of the cholesterol ester fraction (data not shown).
DISCUSSION
We have taken an original approach in using serum fatty acids as surrogate measures of dietary fat to examine the relation, at a population level, between dietary fat and serum cholesterol concentrations. Many observational studies have failed to show associations between saturated fat intake and total cholesterol concentrations, despite consistent results from randomized controlled trials (11-15). Two elements of the design of our study may have contributed to our detection of an association between dietary fat intake—estimated by using biomarkers—and serum cholesterol concentrations, when others have not been able to do so. First, the ranking of participants according to saturated and polyunsaturated fat intakes is probably more accurate with serum fatty acids than with dietary assessment. The serum biomarkers also have an advantage because they reflect the intake of individual fatty acids (19, 27), some of which have stronger cholesterolemic effects than do others (8, 9). Second, the recruitment of participants by using a population-based sampling technique may have selected a survey population with a wider range of serum cholesterol concentrations and dietary fat intakes than would recruitment via a communitywide, nonselective appeal for volunteers.
The results from the current study show that saturated and polyunsaturated fat intakes, as measured by serum fatty acids, are predictors of serum total cholesterol concentrations in New Zealand adolescents and adults. Specifically, the proportion of myristic acid in serum lipids—cholesterol ester, triacylglycerol, and phospholipids—was positively associated with serum total cholesterol, whereas the proportion of linoleic acid was inversely associated. From the lowest to the highest quintile of serum cholesterol myristate, the mean serum cholesterol concentration increased by 0.48 mmol/L, whereas, across the quintiles of cholesterol, linoleate serum cholesterol decreased by 0.18 mmol/L. On a practical note, our finding that myristic acid in serum cholesterol ester was a stronger predictor of total cholesterol concentrations than was either triacylglycerol or phospholipids suggests that cholesterol ester may be the serum fatty acid biomarker of choice in studies designed to examine the relation between fatty acids and cholesterol.
The associations of serum myristic and linoleic acids with serum cholesterol are in agreement with evidence from controlled trials and community-based interventions showing the effect of these dietary fatty acids on serum cholesterol (8, 9). The consistency of the dietary trial evidence is such that the change in serum cholesterol concentration, at a group level, can be predicted by the change in saturated and polyunsaturated fat intakes (4, 5, 9, 28, 29); increasing polyunsaturated fat intakes produces half the reduction in plasma cholesterol of an equivalent decrease in saturated fat intake. The biomarker results showed a similar pattern: the change in serum cholesterol per unit (SD) change in the proportion of myristic acid was approximately twice that per unit change in linoleic acid. The absence of an association between serum palmitic acid and cholesterol concentrations in our cross-sectional survey contrasts with results from dietary intervention studies that show a cholesterol-raising effect of this fatty acid (9). This contrast probably results from the fact that there is a weak correlation between serum palmitic acid and dietary saturated fat (30, 31). For lauric acid, the proportion of survey samples with undetected amounts was too high to test its association with serum cholesterol.
It is noteworthy that the change in serum cholesterol concentration associated with each 1.0 mol% change in serum cholesteryl-myristate (0.43 mmol/L per 1.0 mol%) in our cross-sectional study is less than that observed per 1.0 mol% when the change in the percentage of myristic acid in serum cholesterol ester is in response to a dietary intervention. For example, the results of an intervention trial (32), in which participants consumed diets of different fat compositions, showed that a 0.64 mol% change in plasma cholesteryl-myristate was associated with a 0.66 mmol/L change in plasma cholesterol concentration (1.03 mmol/L per 1.0 mol%). The larger change in the intervention trial is not surprising because, in controlled trials, the nondietary predictors of serum cholesterol are constant and only dietary fat intake is altered.
The range of fatty acid composition from the lowest to the highest quintile was smaller for myristic acid than for linoleic acid. For example, in cholesterol esters, the difference in myristic acid between the 1st and 5th quintiles was 1.13 mol%, whereas that in linoleic acid was 19.37 mol%. Thus, each unit change in mol% myristic acid produced a change in serum cholesterol of 0.43 mmol/L, which is substantially higher than the per-unit change in linoleic acid: –0.01 mmol/L. It would appear that myristic acid in lipoproteins, or in other tissues for which serum myristic acid is a good marker, has substantially stronger effects on metabolic processes that affect serum cholesterol than does linoleic acid. It is also possible that participants are more correctly classified into quintiles of fat intake, particularly dairy fat, with the use of serum cholesterol myristate than with that of serum cholesterol linoleate, because the range of dairy fat intake across the quintiles of cholesteryl-myristate varied from 6% to 10% of energy, whereas, across the quintiles of cholesteryl-linoleate, the range in polyunsaturated fat intake was 2% of energy. The high proportion of myristic acid in serum lipids was associated with higher dairy and saturated fat intakes and with lower polyunsaturated fat intakes; moreover, there may be associations with other dietary or lifestyle factors that increase serum cholesterol concentrations. Therefore, our results should be interpreted as showing an indirect association. However, the associations we report are consistent with those from dietary intervention trials showing the independent cholesterol-raising effects of myristic acid (8, 9).
We found no evidence that, at a population level, the proportions of myristic and linoleic acid in serum lipids predict serum HDL-cholesterol concentrations. In a review of metabolic ward studies covering a wide range of saturated and polyunsaturated intakes, Clarke et al (4) reported that HDL cholesterol increased by 0.013 mmol/L for each 1.0% increase in energy intake as saturated fat. It is possible that our study was underpowered to detect this small effect.
Serum cholesterol concentrations respond rapidly to changes in dietary fat; the maximum effects were achieved within 10–12 d (33). Serum fatty acids reflect changes in dietary fat over a similar time period (34) and, therefore, are an appropriate biomarker tissue for examining the relation between dietary fat and serum cholesterol. Furthermore, serum lipids are found in lipoproteins, and the fatty acids in these lipoproteins not only reflect dietary intake but are closer to the metabolic sites of cholesterol regulation. A significant proportion of the blood samples in the National Nutrition Survey were collected from participants who were not fasting. At a population level, the fat composition of recent meals is likely be similar to habitual intakes, although the differences between individual persons may be considerable; therefore, the greatest effect of postprandial sampling of blood on serum fatty acids is not on the mean fatty acid composition but on the variability of the estimates. This increased variability would attenuate the true association between serum fatty acids and cholesterol concentrations.
The association of palmitoleic acid with serum total cholesterol was unexpected, largely because there is no evidence from controlled feeding trials, or otherwise, of a cholesterol-raising effect of palmitoleic acid. The association may be a chance finding; however, this seems doubtful, given that a significant relation with serum total cholesterol occurred for palmitoleic acid in cholesterol ester (P < 0.001) and triacylglycerol (P < 0.001), but not in phospholipid (P < 0.078). The most likely explanation appears to lie in the close relation of palmitoleic acid in serum lipids with myristic and linoleic acids. In serum cholesterol esters, palmitoleic acid correlated positively with myristic acid (r = 0.40, P < 0.001) and negatively with linoleic acid (r = –0.63, P < 0.001). Diet is a source of palmitoleic acid (35, 36); however, a major portion of the body pool is produced from the desaturation of palmitic acid by stearoyl-CoA desaturase. The activity of this enzyme is enhanced when a diet high in cholesterol is consumed and is suppressed when polyunsaturated fat intake is high (37). Furthermore, the results of cholesterol-lowering trials showed that the proportions of palmitoleic acid in plasma cholesterol ester and triacylglycerol decrease when polyunsaturated fat replaces saturated fat in the diet (38-41).
The results of the current population-based survey indicate that serum biomarkers of saturated fat intake—particularly dairy fat—are positively associated with serum total cholesterol concentration and total:HDL cholesterol, and that biomarkers of polyunsaturated fat intake are inversely associated with total cholesterol, although to a lesser degree. This evidence, taken together with the substantiated cholesterolemic effects of saturated fats established in controlled trials, indicates that dietary fat is an important predictor of population concentrations of serum total cholesterol. Population approaches to reducing serum cholesterol should continue to focus on decreasing the intakes of saturated fat, particularly dairy fat.
ACKNOWLEDGMENTS
The New Zealand Ministry of Health funded the 1997 National Nutrition Survey. The Life in New Zealand group was responsible for conducting the 1997 National Nutrition Survey. We thank Christian Thoma, Leanne Hodson, Belinda Hunter, and Jody Miller for their technical contributions.
CMS conceived of the study and obtained the funding for the fatty acid work. FC took overall responsibility for producing the fatty acid dataset. AG was the consultant for the statistical analysis. All authors were involved in analyzing the dataset, reviewing and interpreting the results, and writing the manuscript. None of the authors had a personal or financial conflict of interest.
REFERENCES