点击显示 收起
1 From Departments of Nutrition (MT, FBH) and Epidemiology (EC), Harvard School of Public Health, the Channing Laboratory, Boston, and the Division of Preventive Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston (JEM).
See corresponding editorial on page 931.
ABSTRACT
Background: Nutritional therapy is a cornerstone of diabetes management, but no epidemiologic studies have investigated the relation between specific dietary fatty acids and cholesterol and cardiovascular disease (CVD) risk among diabetic patients.
Objective: This study assessed the relation between specific dietary fatty acids and cholesterol and CVD risk among women with type 2 diabetes.
Design: Among 5672 women with type 2 diabetes from the Nurses' Health Study, diet was assessed prospectively and updated periodically. Relative risks of CVD were estimated from Cox proportional hazards analysis after adjustment for potential confounders.
Results: Between 1980 and 1998, we identified 619 new cases of CVD (nonfatal myocardial infarction, fatal coronary heart disease, and stroke). The relative risk (RR) of CVD for an increase of 200 mg cholesterol/1000 kcal was 1.37 (95% CI: 1.12, 1.68; P = 0.003). Each 5% of energy intake from saturated fat, as compared with equivalent energy from carbohydrates, was associated with a 29% greater risk of CVD (RR: 1.29; 95% CI: 1.02, 1.63; P = 0.04). The ratio of polyunsaturated to saturated fat (P:S) was inversely associated with the risk of fatal CVD. We estimated that replacement of 5% of energy from saturated fat with equivalent energy from carbohydrates or monounsaturated fat was associated with a 22% or 37% lower risk of CVD, respectively.
Conclusions: A higher intake of cholesterol and saturated fat and a low P:S were related to increased CVD risk among women with type 2 diabetes. Among diabetic persons, replacement of saturated fat with monounsaturated fat may be more effective in lowering CVD risk than is replacement with carbohydrates.
Key Words: Dietary fat • dietary cholesterol • diabetes mellitus • cardiovascular diseases • epidemiology
INTRODUCTION
Although nutritional therapy is a cornerstone of diabetes management, data on the associations between specific types of fat and cholesterol intake and cardiovascular disease (CVD) risk among persons with type 2 diabetes are lacking. In previous prospective cohort studies among the general population, higher intakes of saturated and trans fats were associated with increased risk of coronary heart disease (CHD), whereas higher intakes of monounsaturated and polyunsaturated fats were associated with decreased risk (1). Most prospective cohort studies did not find a strong association with dietary cholesterol (1, 2)
In an analysis that excluded women who had diabetes at baseline, we previously reported (3) on the association between major types of fat and the risk of CHD in the overall Nurses' Health Study (NHS) cohort. In the current study, we examined the relations between different types of dietary fat and cholesterol and the risk of CVD among women with type 2 diabetes in the NHS, with follow-up extended from 1994 to 1998.
SUBJECTS AND METHODS
The NHS was established in 1976, when 121 700 female nurses who were aged 30–55 y and who resided in 11 large US states completed a mailed questionnaire on their medical history and lifestyle. Follow-up questionnaires were sent every 2 y to update information on potential risk factors and to identify newly diagnosed cases of CVD and other illnesses.
Our study included the 5674 women who reported on any questionnaire from 1976 through 1996 a physician's diagnosis of diabetes mellitus at age =" BORDER="0">30 y. All of these women were sent a supplementary questionnaire on symptoms, diagnosis, and treatment. Cases were considered confirmed if any of the following criteria were met: =" BORDER="0">1 classic symptoms plus an elevated fasting (=" BORDER="0">7.8 mmol/L) or random (=" BORDER="0">11.1 mmol/L) plasma glucose concentration; =" BORDER="0">2 elevated plasma glucose concentrations on separate occasions (fasting =" BORDER="0">7.8 mmol/L) or random concentrations (=" BORDER="0">11.1 mmol/L, or 11.1 mmol/L after 2 h or more on glucose tolerance testing) in the absence of symptoms, or both; or treatment with a hypoglycemic drug (insulin or oral hypoglycemic agent). The validity of this questionnaire has been verified in a subsample of the study cohort (4). Because the cases of diabetes used in this analysis were diagnosed before 1997, we used the National Diabetes Data Group diagnostic criteria that were in place for that period (5), rather than the more recent classifications (6). In the primary analyses, we used subjects' self-reported diabetes to define the analytic cohort. We also conducted secondary analyses including only confirmed diabetics.
Assessment of dietary intake
In 1980, a 61-food item semiquantitative food-frequency questionnaire (FFQ) was mailed to women in the cohort. In 1984, the FFQ was expanded to 116 food items. Dietary intakes were updated in 1986, 1990, and 1994. Details on the nutrient intake calculations can be found elsewhere (3). Both the original and the revised questionnaires provided a reasonable measure of total and specific types of fat, as reflected in the comparison with the multiple 1-wk dietary records: correlation coefficients for total and specific types of fat were assessed by the dietary records, and the FFQs ranged from 0.46 to 0.58 for the 1980 questionnaire and from 0.48 to 0.68 for the longer questionnaire used in 1984, 1986, 1990, and 1994. Cholesterol intakes assessed by using the FFQs had a correlation of 0.61 and 0.73 with the values derived from the dietary record for the original questionnaire and the revised questionnaire, respectively (7). The correlation between the calculated dietary intake of trans unsaturated fatty acids from FFQs and the proportion of trans unsaturated fatty acids in adipose tissue was 0.51 (8).
Endpoints
Total CVD, as a composite of fatal CHD, nonfatal myocardial infarction, and stroke occurring between the return of the 1980 questionnaire and June 1, 1998, was the primary endpoint of this analysis. Self-reported myocardial infarctions were confirmed by reviewing medical records if the subjects met the World Health Organization criteria: ie, characteristic symptoms with either typical electrocardiographic changes or elevations of cardiac enzymes. Stroke was confirmed by medical records according to the criteria of the National Survey of Stroke (9). Physicians who reviewed the records had no knowledge of the subjects' self-reported risk factor status.
Deaths were reported to the NHS group by the the subjects' next of kin, work associates, and postal authorities; in some cases, the National Death Index of the National Center for Health Statistics was used to identify deceased cohort members.A 98% death ascertainment was achieved in this cohort. Fatal CHD was confirmed by a review of medical records or autopsy reports with the permission of the next of kin.
Statistics
Person-months of follow-up were accumulated from the date of return of the 1980 questionnaire for women whose diabetes was diagnosed in or before 1980 or from the return of the questionnaire on which type 2 diabetes was first reported by women whose diabetes was diagnosed after 1980. Accrual of person-months continued until the occurrence of a CVD endpoint, death, or May 31, 1998. We excluded women with a history at baseline of MI, angina, coronary revascularization, stroke, or cancer at baseline. Women who had a CVD event during a previous 2-y interval were excluded from subsequent follow-up.
We used the cumulative average of nutrients and foods from all available questionnaires up to the start of each 2-y year follow-up interval, as described by Hu et al (10). For example, fat intake reported on the 1980 questionnaire was related to the occurrence of events from 1980 through 1984, and the average of fat intake reported on the 1980 and 1984 questionnaires was related to events from 1984 through 1986. Nondietary covariates were updated every 2 y.
Women were categorized by quintiles of each type of fat expressed as percentage of energy, quintiles of cholesterol expressed in mg/1000 kcal, quintiles of the ratio of polyunsaturated fat to saturated fat (P:S), and quintiles of the Keys score:
RESULTS
We examined intakes of saturated fat, polyunsaturated fat, and cholesterol in relation to other potential risk factors for CVD in 1980 (Table 1). Women with a higher intake of saturated fat and women with a higher intake of polyunsaturated fat exercised less, drank less alcohol, and were less likely to use multivitamin and vitamin E supplements. Women with a higher intake of cholesterol tended to have higher BMIs and to drink less alcohol, and they were more likely to use insulin and oral hypoglycemic drugs.
View this table:
TABLE 1. Age-adjusted characteristics of diabetic women (n = 1692) according to quintiles (Q) of fat and cholesterol intakes in 1980
During 57 195 person-years of follow-up, we documented 619 CVD events including 268 nonfatal myocardial infarctions, 183 fatal myocardial infarctions, and 168 strokes. Total fat intake was not associated with risk of CVD in multivariate analyses (Table 2). In age-adjusted analyses, the intakes of animal fat, saturated fat, and cholesterol and the Keys score were each associated with increased risk of CVD. After multivariate adjustment, all these associations were attenuated, but the trends remained significant for animal fat, saturated fat, and cholesterol. Vegetable fat was associated with reduced risk of CVD in age-adjusted analyses, but the relation was attenuated in multivariate analyses. trans Fats and polyunsaturated fats were not significantly associated with risk in multivariate analyses. A higher P:S was significantly associated with decreased risk in age-adjusted analyses, but the association was attenuated after multivariate adjustment. Monounsaturated fat tended to be associated with an increase in CVD risk in age-adjusted analyses. In multivariate analyses including other fats and cholesterol, the direction of association reversed, and there was a nonsignificant decrease in CVD risk across quintiles of monounsaturated fat intake. Dietary cholesterol remained significantly related to CVD risk when controlled for types of fat.
View this table:
TABLE 2. Relative risks (RRs) of cardiovascular disease according to quintiles (Q) of fat intake, cholesterol, and Keys score among diabetic women1
Next, we examined fat and cholesterol intakes as continuous variables entered simultaneously in the multivariate model (Table 3). We observed positive associations for intakes of saturated fat (P = 0.04) and cholesterol (P = 0.003) and the Keys score (P = 0.002). Thus, replacing 5% of energy from carbohydrates with equivalent energy from saturated fat was associated with a 29% increase in risk, each increase of 200 mg cholesterol/1000 kcal was associated with a 37% increase in risk, and each increase of 10 units in the Keys score was associated with a 23% increase in risk. Monounsaturated fat and the P:S tended to be associated with reduced risk. The relation between trans fat and CVD risk was positive but not significant. Additional adjustment for intakes of n–3 fatty acids did not alter these results. Similar results were obtained in secondary analyses among women with confirmed type 2 diabetes.
View this table:
TABLE 3. Relative risks (RRs) of cardiovascular disease according to fat intake, cholesterol, and Keys score among diabetic women1
We estimated the effects of isocaloric substitution of carbohydrates or monunsaturated fat for saturated fat from the multivariate model including saturated fat, polyunsaturated fat, monounsaturated fat, trans fat, cholesterol, protein, total calories, fiber, and nondietary covariates. Replacing 5% of energy from saturated fat with equivalent energy from carbohydrates was associated with a 22% lower risk of CVD (95% CI: 2%, 39%; P = 0.04). The same replacement with monounsaturated fat was associated with a 37% lower risk of CVD (95% CI: 0%, 60%; P = 0.048).
We analyzed separately the relation between types of fat and fatal and nonfatal events (Table 4). The P:S was strongly inversely related to fatal events, but was not related to nonfatal events. In addition, the Keys score was positively related to fatal events, but not to nonfatal events.
View this table:
TABLE 4. Relative risks (RRs) of fatal and nonfatal cardiovascular disease (CVD) events in quintile (Q) 5 and Q1 for types of fat, cholesterol, P:S, and Keys score1
We also assessed separately the relation between types of fat and CVD risk among women with diabetes at baseline (prevalent cases) and among women who became diabetic during the follow-up (incident cases). There were no major differences between the 2 groups of diabetic women. trans Fat intake was associated with a nonsignificant increase in CVD risk among incident cases (RR: 1.49; 95% CI: 0.86, 2.57 for 2% of energy; P = 0.15) but not among prevalent cases (RR: 0.95; 95% CI: 0.59, 1.58; P = 0.84).
DISCUSSION
In this cohort of diabetic women, a diet characterized by higher cholesterol and saturated fat and a lower P:S was associated with increased CVD risk, whereas total fat intake was not related to CVD risk. We estimated that the replacement of saturated fat with monounsaturated fat was associated with a greater risk reduction than was replacement with carbohydrates.
The prospective design of our study reduces the potential for bias from differential reporting of dietary intake between cases and noncases. Furthermore, we incorporated changes in lifestyle through updated measures of dietary intakes and other covariates over the follow-up period. In general, the relations with CVD risk were clearer when different types of fat were introduced as continuous predictors rather than in quintiles. This difference is likely due to improved power and decreased residual confounding in the case of continuous predictors. Although intakes of different types of fat were measured with relatively good validity, some misclassification is inevitable, in which case relative risks would generally tend to be biased toward the null.
In metabolic studies, exchanging saturated fat for carbohydrates increases both LDL and HDL cholesterol, whereas exchanging monounsaturated and polyunsaturated fat for saturated fat lowers LDL cholesterol. Monounsaturated fat apparently does not result in lower HDL cholesterol when it is used to replace saturated fat, whereas the effect of polyunsaturated fat is less clear (7, 12). trans Fatty acids appear to be particularly deleterious because, in addition to increasing LDL cholesterol, they reduce HDL cholesterol (13). In our previous analyses of the overall NHS cohort, higher intakes of trans fat and, to a smaller extent, saturated fat were associated with increased CVD risk, whereas higher intakes of monounsaturated and polyunsaturated fats were associated with decreased CVD risk (3). The positive association between saturated fat and CVD risk was stronger in the subjects in this study than that in the overall NHS cohort. This difference may reflect the adverse effects of saturated fat on lipoproteins as well as other metabolic consequences of insulin resistance among diabetics. In metabolic studies, saturated fat impairs insulin sensitivity, whereas diets containing a larger proportion of unsaturated fat appear to improve glucose metabolism (14). Replacing saturated fat with monounsaturated fat is associated with improvements in lipoprotein and glycemic control in patients with type 2 diabetes (15).
Short-term studies consistently showed that high-monounsaturated fatty acid diets improve lipid profiles and glycemic control more than do high-carbohydrate diets (16). In our data, monounsaturated fat was inversely but nonsignificantly related to the incidence of CVD after adjustment for other types of fat. Although assessment of the independent effects of monounsaturated fat in this population is difficult because of the shared food sources of saturated and monounsaturated fats (dairy and meats), we estimated that the replacement of saturated fat with monounsaturated fat was associated with a greater risk reduction than was replacement with carbohydrates. However, because of the relatively small sample size, the CIs were wide for the estimates. The P:S was strongly related to a lower risk of fatal events. This may be due to a potential antiarrhythmic effect of polyunsaturated fat (2).
Our analyses showed a positive, albeit nonsignificant, relation between trans fat and CVD in diabetic women. This positive association was primarily observed in those women with incident diabetes. It is possible that the adverse effects of trans fat were masked by long duration of diabetes or by diabetes treatment.
We found an increase in CVD risk with higher intakes of cholesterol. We additionally assessed the relation between the consumption of specific foods containing substantial amounts of fat and cholesterol and the risk of CVD. Only the intake of eggs was significantly associated with CVD risk, as previously reported by Hu et al (17). The multivariate relative risks for <1 egg/wk, 1–<3 eggs/wk, 3 eggs/wk, 3–6 eggs/wk, and >6 eggs/wk were 1, 1.10, 1.08, 1.39, and 1.44 (95% CI: 1.05, 1.98; P for trend < 0.001).
In the general population, dietary cholesterol raises concentrations of total and LDL cholesterol, but there is great variability in the lipoprotein response (18–20). Most epidemiologic studies have not found a significant association between dietary cholesterol and CVD risk in the general population (1, 2). However, dietary cholesterol may be more detrimental to diabetics than to nondiabetics because of heightened insulin resistance and dyslipidemia among diabetic patients (21–23).
In conclusion, in this prospective cohort study of diabetic women, we found that higher saturated fat and cholesterol intakes were significantly associated with CVD risk. A higher P:S was associated with a lower risk of fatal events. Thus, among diabetic persons, it is important to reduce intakes of cholesterol and to replace saturated fats with unhydrogenated unsaturated fats.
ACKNOWLEDGMENTS
We are indebted to Walter Willett for his continuous support and invaluable input and to Meir Stampfer for his insightful comments on the manuscript.
REFERENCES
Related articles in AJCN: