Dietary Flavonoid Intake and Risk of Cardiovascular Disease in Postmenopausal Women

Transcription

1 American Journal of Epidemiology Copyright 999 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved Vol. 49, No. 0 Printed in USA. Dietary Flavonoid Intake and Risk of Cardiovascular Disease in Postmenopausal Women Laura Yochum, Lawrence H. Kushi, Katie Meyer, 2 and Aaron R. Folsom Flavonoids, a group of phenolic compounds found in fruits and vegetables, are known to have antioxidant properties. They prevent low density lipoprotein oxidation in vitro and thus may play a role in the prevention of coronary heart disease (CHD). In 986, in a prospective study of 34,492 postmenopausal women in Iowa, the authors examined the association of flavonoid intake with CHD and stroke mortality. Over 0 years of follow-up, 438 deaths from CHD and 3 deaths from stroke were documented. Total flavonoid intake was associated with a decreased risk of CHD death after adjusting for age and energy intake (p for trend = 0.04). This association was attenuated after multivariate adjustment. However, decreased risk was seen in each category of intake compared with the lowest. Relative risks and 95% confidence intervals of CHD death from lowest to highest intake category were, 0.67 (95% confidence interval (Cl) ), 0.56 ( ), (95% Cl -.8), and ( ).There was no association between total flavonoid intake and stroke mortality (p for trend = 0.83). Of the foods that contributed the most to flavonoid intake in this cohort, only broccoli was strongly associated with reduced risk of CHD death. The data of this study suggest that flavonoid intake may reduce risk of death from CHD in postmenopausal women. Am J Epidemiol 999; 49: antioxidants; coronary heart disease; diet; flavonoids; postmenopausal women Oxidation of low density lipoproteins is believed to play an important role in the development of atherosclerosis (,2). Oxidized low density lipoprotein cholesterol (LDL cholesterol) is taken up more readily by macrophages, which leads to the formation of foam cells and atherosclerotic plaques (, 3). Mechanisms that slow or prevent this chain of events may decrease risk of coronary heart disease (CHD) and stroke (4). Flavonoids are a group of phenolic compounds that are found in fruits and vegetables and are known to have antioxidant properties (5). They have been reported to be scavengers of free radicals, including superoxide anions (6), singlet oxygen (7), and lipid peroxy-radicals (8). In addition, flavonoids have been shown to prevent LDL cholesterol oxidation and cytbtoxicity in vitro (9). Epidemiologic studies investigating the relation between flavonoid intake, CHD, and stroke have produced inconsistent results. Some studies have found an Received for publication January 2,998, and accepted for publication September 25, 998. Abbreviations: CHD, coronary heart disease; Cl, confidence interval; LDL cholesterol, low density lipoprotein cholesterol;, relative risk. Division of Epidemiology, University of Minnesota, Minneapolis, MN. 2 Harvard School of Public Health, Department of Epidemiology, Boston, MA. Reprint requests to Dr. Aaron R. Folsom, Division of Epidemiology, University of Minnesota, 300 S. 2nd Street, Suite 300, Minneapolis, MN inverse association between intake and CHD mortality (0-2) and stroke (3), while others have not (4). Another suggested that the potential benefit of flavonoids is limited to men with prevalent CHD (5). To further evaluate the potential effect of dietary flavonoids, we investigated the relation between flavonoid intake, CHD, and stroke mortality in a prospective cohort study of postmenopausal women in Iowa. MATERIALS AND METHODS Study population In January 986, a 6-page questionnaire was mailed to a random sample of 99,826 women aged years who had valid Iowa drivers licenses in 985. The 4,836 women who responded to the questionnaire were enrolled in the Iowa Women's Health Study. Questions related to demographic characteristics, health habits, medical history, and gynecologic history were included in the baseline questionnaire. In addition, diet was assessed through a semiquantitative food frequency questionnaire. Specific questions about weight history, current height and weight, age, education, and marital status were included. Body mass index was then calculated from height and weight information. Waist-to-hip ratio was based on self measurement (a tape measure was 943

2 944 Yochum et al. included with the initial questionnaire) (6). Physical activity (number of times per week participated in moderate and vigorous physical activity), current and past smoking, menopausal status, medication use, and hormone replacement status was assessed through the baseline questionnaire. Participants were also asked if they had a history of diabetes, cancer, heart disease, heart attack, or high blood pressure. Participants were excluded if they had not reached menopause at the time the questionnaire was completed (n = 569), if they skipped more than 30 items on the food frequency questionnaire (n = 2,749), if they had relatively extreme energy intakes (<600 or >5,000 kcal per day) (n = 33), or if they reported having prior angina, heart disease, or heart attack (n = 3,73). After these exclusions, 34,492 women remained. Dietary assessment The participant's diet was assessed using a 27-item semiquantitative food frequency questionnaire similar to that used in the Nurses' Health Study (7). Nutrient values were based primarily on data from the US Department of Agriculture (8). Since the US Department of Agriculture data do not contain information on flavonoids, these values were taken from analyses performed by Hertog et al. (9, 20) in the Netherlands and supplemented with values for additional US foods. These analyses concentrated on five major flavonoids: quercetin, kaempferol, myricetin, apigenin, and luteolin. Intake of individual flavonoids was calculated based on the frequency of consumption multiplied by the flavonoid content of the food. Total flavonoid intake was the sum of the individual flavonoids. Flavonoids can be further classified as flavonols (quercetin, kaempferol, and myricetin) and flavones (luteolin and apigenin) (2). Although we did not validate the ability of the food frequency questionnaire to assess flavonoid intake in this population of Iowa women, validation of the questionnaire for this purpose has been done by others. Feskanich et al. (22) assessed the ability of the questionnaire to measure intake of foods containing the major source of flavonoids. Correlation coefficients for foods contributing the most to flavonoid intake were 0.77 for tea, 0.70 for apples, and 0.46 for broccoli, comparing intake measured by the food frequency questionnaire to intake measured by 28-day diet records in a study of female nurses (22). However, onions and berries, two potentially important sources of flavonoids, were not ascertained. Members of this cohort were followed annually through the State Health Registry of Iowa, which collects information on deaths in Iowa. Deaths were also identified through follow-up questionnaires in 988, 990, and 992 and, for nonresponders, through linkage with the National Death Index. The cause of death was determined using the International Classification of Diseases, Ninth Revision (ICD-9). Death was considered to be coronary heart disease if ICD-9 codes 40 through 44 or were assigned or stroke if ICD-9 codes 430 through 438 were assigned. The cause of death coding was not validated; however, other studies have found the validity of these codes for CHD death to be relatively high (23). Statistical analysis Total person-years of follow-up were calculated for each woman as time from completion of the baseline questionnaire to date of death or December 3, 995. After 0 years of follow-up, 438 deaths from CHD and 3 deaths from stroke had been documented. Our initial statistical analyses examined the relation between total flavonoid intake and CHD mortality, adjusting for age and energy intake. Total flavonoid intake was divided by quintiles; mortality from CHD in higher intake categories was compared with the lowest category of intake using proportional hazards regression analysis. We then performed two subsequent analyses: The first was adjusted for potential nondietary confounders, including high blood pressure, diabetes, body mass index, waist-to-hip ratio, estrogen replacement therapy status, alcohol intake, smoking, physical activity, marital status, and education; and the second was adjusted for these variables as well as for intake of cholesterol, saturated fat, vitamin E, whole grains, and dietary fiber. The stroke analyses were performed in similar manner. However, intakes of beta-carotene, vitamin C, and fish were also adjusted for when stroke mortality was examined. The possible effect of individual flavonoids and of foods high in flavonoid content was also assessed. The associations of quercetin and kaempferol with death from CHD were examined using methods similar to those for total flavonoid intake. Since the proportion of the population that consumed myricetin, luteolin, and apigenin was relatively small, the effects of these flavonoids were examined by comparing the relative risk of CHD death for those estimated to have any intake versus those with no intake. For individual foods, consumption categories were selected to ensure a reasonable distribution of the population across categories of intake. RESULTS As has been shown previously in this cohort, hypertension, diabetes mellitus, cigarette smoking, a higher body mass index, and higher waist-to-hip ratio have been associated with a greater risk of death from CHD

3 Dietary Flavonoid Intake and Risk of Cardiovascular Disease 945 (24). In addition, a lower risk of CHD has been associated with a high level of physical activity (25), use of estrogen replacement therapy (24), and intake of vitamin E from foods (26) in this cohort. Table shows the distribution of these risk factors by quintile of total flavonoid intake. Subjects who had a lower average intake of flavonoids tended to smoke and drink more and exercise less. Mean values of body mass index and waist-to-hip ratio were similar across quintiles. Subjects in the highest quintile of flavonoid intake also had a higher average intake of energy, saturated fat, cholesterol, vitamin E, whole grains, fiber, vitamin C, beta-carotene, and fish. The mean intake of total flavonoids in our cohort was 3.9 mg/day. Mean intakes of flavonols were: quercetin (9.7 mg/day), kaempferol (3.4 mg/day), and myricetin (0.74 mg/day). Mean intakes of flavones were considerably less: luteolin (0.05 mg/day) and apigenin (0.0 mg/day). Individual flavonols were highly correlated (quercetin and kaempferol, r = 0.84; quercetin and myricetin, r = 0.78; and kaempferol and myricetin, r = 0.77), while flavones were not (apigenin and luteolin, r = 0.26). Correlations between individual flavonols and flavones were all below r = Foods inquired about that contributed the most to flavonoid intake were tea (36 percent), apples (7 percent), and broccoli (9 percent). Table 2 shows age and energy-adjusted relative risks of death from CHD by flavonoid intake. Total flavonoid intake adjusted for age and energy intake was associated with a reduced risk of CHD for the highest fifth of intake compared with the lowest (relative risk () = 0.6, 95 percent confidence interval (CI) ). This association was not modified appreciably after adjustment for additional nondietary confounders ( = 0.64, 95 percent CI ). Relative risks for the third versus the first intake category ( = 0.57,95 percent CI ) and the second versus the first ( = 0.67, 95 percent CI ) were also statistically significant. However, the overall test for trend after adjustment for nondietary variables was not statistically significant (p for trend = 0.4), indicating an inconsistent dose-response relation. When these analyses were further adjusted for dietary variables, there was no significant change in the overall test for trend (p for trend = 0.) or the relative risks for any category of flavonoid intake compared with the lowest. There was no association between total flavonoid intake and stroke mortality. Relative risks from the highest intake category compared with the lowest were: =.8, 95 percent CI ; = 0.84, 95 percent CI ; = percent CI ; and = 2, 95 percent CI (p for trend = 0.83). TABLE. Distribution of coronary heart disease risk factors by quintile of flavonoid intake at baseline in 34,492 postmenopausal women, 986 (low) Quintile of total flavonoid intake (high) Median flavonoid intake Range of flavonoid intake (mg/day) Age (years) Body mass index (kg/m 2 ) Waist-to-hip ratio Current smoker (%) Diabetes mellitus (%) Hypertension (%) High level of physical activity (%) Current estrogen replacement therapy (%) Average nutrient intake Total energy (kcal/day) Saturated fat (g/day) Cholesterol (mg/day) Vitamin E (lu/day) Fiber (g/day) Whole grains (servings/week) Alcohol (g/day), , , , , Average servings per week Apples Broccoli Tea

4 946 Yochum et al. TABLE 2. Relative risk of death from coronary heart disease according to quintile of flavonoid intake in 34,492 postmenopausal women, Quintile of flavonoid intake (lowest) * 2 * (highest) p for trend Total flavonoids * Median flavonoid intake (trig/day) Range of flavonoid intake (mg/day) t ^ ib Quercetin Median quercetin intake (mg/day) Range of quercetin intake (mg/day) I Kaempferol Median kaempferol intake (mg/day) Range of kaempferol intake (mg/day) Apigenin Apigenin intake 437 None Luteolin Luteolin intake 33 None Myricetin Myricetin intake 37 None ( *, relative risk; Cl, confidence interval; CHD, coronary heart disease. t Adjusted for age, total energy intake, body mass index squared, waist-to-hip ratio, high blood pressure (yes or no), diabetes (yes or no), estrogen replacement therapy (current, former, or never), alcohol intake (none, <4 g per day, 4-<0 g per day, or >0 g per day), education (no high school diploma, high school diploma, college or vocational school but no degree, or college degree), marital status (currently married, never married, separated or divorced, or widowed), pack-years of smoking (none, -9, 20-39, or >40), and physical activity (low, moderate, or high level). $ Adjusted for above covariates and intake of cholesterol, saturated fat, vitamin E, dietary fiber, and whole grains. When individual flavonoids were examined, the risk estimates for kaempferol and quercetin (two flavonols) suggested an inverse association between intake and death from CHD (table 2). Each category of intake above the first was associated with a decreased risk of death from CHD, although not all estimates reached statistical significance. A dose-response relation was not evident after adjustment for nondietary confounders as the test for trend was not statistically significant for either quercetin (p for trend = 0.25) or kaempferol (p for trend = 0.56). There was also a suggestion of an inverse association between intake of luteolin or myricetin and death from CHD, although neither of these associations was statistically significant after multivariate adjustment. After additional adjustment for dietary variables, there was no signifi-

5 Dietary Flavonoid Intake and Risk of Cardiovascular Disease 947 cant change in any risk estimates for individual flavonoids. The risk of CHD in relation to intake of three specific foods containing flavonoids is presented in table 3. When adjusted for age and energy intake, there was a significant inverse dose-response association of broccoli (p for trend = 0.000) and apple intake (p for trend = 0.0) with risk of death from CHD. After adjustment for nondietary variables, the doseresponse relation for broccoli intake remained (p for trend = 0.000), although the association for apples was attenuated (p for trend = 0.47). Further adjustment for dietary variables had no significant effect on the association between CHD death and broccoli (p for trend = 0.000) or apple (p for trend = 0.42) intake. Consumption of tea was not significantly associated with risk of CHD death (p for trend = 0.2). Intake of broccoli (p for trend = 0.23), tea (p for trend = 0.40), and apples (p for trend = 0.93) was not associated with stroke mortality after multivariate adjustment. DISCUSSION This study of postmenopausal women supports the hypothesis that greater intake of dietary flavonoids is associated with a decreased risk of death from CHD. We observed a statistically significant 38 percent reduction in CHD mortality in the highest category of total flavonoid intake compared with the lowest. Although the risk was decreased in each category of flavonoid intake compared with the lowest, no doseresponse relation was observed. Analysis of individual flavonoids (kaempferol, quercetin, myricetin, and luteolin) also suggested inverse relations consistent with this hypothesis, although not all relative risks were statistically significant. Broccoli, one of three foods contributing a significant amount to flavonoid intake in this population, showed a statistically significant inverse dose-response relation with CHD mortality. There was also a nonsignificant decrease in CHD mortality associated with consumption of apples and tea, two foods that made up most of the remaining flavonoid intake assessed in this population. However, TABLE 3. Relative risk of death from coronary heart disease according to quartile of intake of selected foods containing flavonoids in 34,492 postmenopausal women, Quartile of intake (lowest) * 2 * 3 4 (highest) pfor trend Apples * Range of apple intake (times/week) t ^ Broccoli Range of broccoli intake (times/month) O O Tea Range of tea intake (times/week) *, relative risk; Cl, confidence interval; CHD, coronary heart disease. t Adjusted for age, total energy intake, body mass index squared, waist-to-hip ratio, high blood pressure, diabetes, estrogen replacement therapy, alcohol intake, education, marital status, pack-years of smoking, and physical activity. X Adjusted for above covariates and intake of cholesterol, saturated fat, vitamin E, dietary fiber, and whole grains

6 948 Yochum et al. we found no association between fiavonoid intake and stroke mortality. Our findings are consistent with a growing number of studies that suggest fiavonoid intake may be related to CHD mortality. Table 4 summarizes the prospective epidemiologic studies that have examined this relation to date. In a recently updated analysis of data using 0 years of follow-up from the Zutphen Elderly Study, Hertog et al. () found a 53 percent reduction in risk of death from CHD in the highest category of intake compared with the lowest, which was consistent with their previous findings based on a shorter period of follow-up (0). Knekt et al. (2) found reduced risks of CHD mortality with higher fiavonoid intakes in men ( = 0.67) and women ( = 0.73), although these findings did not reach statistical significance. In addition, Rimm et al. (5) found a statistically nonsignificant decreased risk of CHD mortality with fiavonoid intake, although this association was limited to men with prevalent CHD. In contrast, Hertog et al. (4) recently found a relative risk of.6 for CHD mortality in relation to fiavonoid intake in a population of Welsh men. They hypothesized that this result may have been due to binding of flavonoids in tea with protein from milk, which is frequently added to tea in Wales, thus reducing fiavonoid absorption. We observed a suggestion of an inverse relation between tea consumption and CHD mortality, whereas other investigators have reported results ranging from no effect (5) to decreased risk with increased tea consumption (0). Although our findings are generally consistent with studies that showed an inverse relation between fiavonoid intake and CHD mortality, they did not support those of Keli et al. (3), who found a statistically significant reduced risk of stroke incidence in the highest category of fiavonoid intake compared with the lowest ( = 0.27, 95 percent CI ). The relatively low number of strokes in our study may have contributed to the lack of observed association. Although it is possible that an alternate component of food may be responsible for the observed decrease in CHD mortality in our population, we could not identify any such component. We examined the relation between CHD mortality and fiavonoid intake after adjusting for potential nondietary confounders. When additional dietary variables were added to this model, the relative risk estimates were not substantially modified. A possible mechanism through which flavonoids may help prevent coronary heart disease is their antioxidant properties. Flavonoids are free-radical scavengers (6-8) and can prevent LDL cholesterol oxidation in vitro (9). Since oxidation of LDL cholesterol is thought to promote atherosclerosis, it is plausible that flavonoids may delay the development of atherosclerosis and ultimately decrease CHD mortality. As is common in all epidemiologic studies of diet and disease, the results of our study are limited by misclassification of dietary exposures. We also did not measure any changes in diet that occurred during the follow-up period, as our analyses were based on information from a single food frequency questionnaire administered at the start of our study. In addition, data on the fiavonoid content of foods are primarily limited to analyses conducted in the Netherlands (9, 20), rather than on foods in the U.S. market. The overall fiavonoid intake in this population of postmenopausal Iowa women was lower (3.9 mg/day) than in most studies discussed previously (average intake, mg/day). Part of the reason may be that the food frequency questionnaire did not ask about onions and berries, which are high in flavonoids. Another potential limitation of our data is information on CHD risk factors was based on self-report. However, previous studies have documented the association of these risk factors with CHD in this cohort (24-26). In addition, we did not measure blood cholesterol levels or baseline history of stroke. Finally, we TABLE 4. Prospective epidemiologic studies of fiavonoid intake and occurrence of coronary heart disease and stroke Study (reference no.) Hertog et al. () Hertog et al. (4) Keli et al. (3) Knekt etal. (2) Rimm etal. (5) Present study Population size (no.) 804 men,900 men 552 men 2,748 men 2,385 women 34,789 men 34,492 women Years of follow-up Outcome Death from CHD* Incident fatal Death from CHD Fatal and nonfatal stroke Death from CHD (men) Death from CHD (women) Nonfatal Ml* Death from CHD for men with prevalent CHD Death from CHD ' (high vs. low intake) « *, relative risk; CI, confidence interval; CHD, coronary heart disease; Ml, myocardial infarction. Mean fiavonoid intake (mg/day) (high vs. low intake) >29.9 vs. <9.0 mg/day >34.0 vs. <9.0 mg/day >28.6 vs. <8.3 mg/day >4.8 vs. <2. mg/day >5.5 vs. 2.4 mg/day (men) 40.0 vs. 7. mg/day (women) 32.2 vs. 4.3 mg/day

7 Dietary Flavonoid Intake and Risk of Cardiovascular Disease 949 examined the relation between flavonoid intake, CHD and stroke mortality, not incidence. Our results are potentially biased if flavonoid intake is related differently to incidence as compared with mortality. Overall, our results of a reduced risk of CHD mortality with flavonoid intake are consistent with a growing number of studies and are the most definitive for women to date. The association in this cohort was relatively strong, representing a 38 percent decreased risk of CHD death in the highest consumption category versus the lowest. The role of flavonoids as antioxidants provides a plausible mechanism through which flavonoids may decrease CHD risk. However, given the limitations of the diet assessment and the observational study design, our results can not be considered definitive. Nevertheless, our findings do contribute additional information about a modifiable potential risk factor for CHD. ACKNOWLEDGMENTS Supported by a research grant CA from the National Institutes of Health. The authors recognize Laura Sampson and Dr. Walter Willett for use of the food frequency questionnaire in this study. REFERENCES. Witztum JL, Steinberg D. Role of oxidized low density lipoprotein in atherogenesis. J Clin Invest 99;88: Parthasarathy S, Steinberg D, Witztum JL. The role of oxidized low-density lipoproteins in the pathogenesis of atherosclerosis. Ann Rev Med 992;43: Palinski W, Rosenfeld ME, Yla-Herttuala S, et al. Low density lipoprotein undergoes oxidative modification in vivo. Proc Natl Acad Sci U S A 989,86: Catapano AL. Antioxidant effect of flavonoids. Angiology 997;48: Kandaswani C, Middleton E. Free radical scavenging and antioxidant activity of plant flavonoids. Adv Exp Med Biol 994;336; Robak J, Gryglewski RJ. Flavonoids are scavengers of superoxide anion. Biochem Pharmacol 988;37: Husain SR, Cillard J, Cillard P. Hydroxy radical scavenging activity of flavonoids. Phytochemistry 987;26: Sorata Y, Takahama U, Kimura M. Protective effect of quercetin and rutin on photosensitized lysis of human erythrocytes in the presence of hematoporphyrin. Biochem Biophys Acta 982;799: De Whalley CV, Rankin SM, Hoult JRS, et al. Flavonoids inhibit the oxidative modification of low-density lipoproteins by macrophages. Biochem Pharmacol 990;39: Hertog MGL, Feskens EJM, Hollman PCH, et al. Dietary antioxidant flavonoids and risk of coronary heart disease. The Zutphen Elderly Study. Lancet 993;342:007-ll.. Hertog MGL, Feskens EJ, Kromhout D. Antioxidant flavonols and coronary heart disease risk. (Letter). Lancet 997;349: Knekt P, Jarvinen R, Reunanen A, et al. Flavonoid intake and coronary mortality in Finland: a cohort study. BMJ 996; 32: Keli SO, Hertog ML, Feskens EJM, et al. Dietary flavonoids, antioxidant vitamins, and incidence of stroke. Arch Intern Med 996;56: Hertog MG, Sweetnam PM, Fehily AM, et al. Antioxidant flavonols and ischemic heart disease in a Welsh population of men: the Caerphilly Study. Am J Clin Nutr 997;65: Rimm E, Katan MB, Ascherio A, et al. Relation between intake of flavonoid and risk for coronary heart disease in male health professionals. Ann Intern Med 996;25: Kushi LH, Kaye SA, Folsom AR, et al. Accuracy and reliability of self-measurement of body girths. Am J Epidemiol 988;28: Willett WC, Sampson L, Browne ML, et al. The use of a selfadministered questionnaire to assess diet four years in the past. Am J Epidemiol 988;27: Consumer and Food Economics Institute. Composition of foods. Washington, DC: US Department of Agriculture, Human Nutrition Information Service, Hertog MGL, Feskens EJM, Hollman PCH, et al. Content of potentially anticarcinogenic flavonoids of 28 vegetables and 9 fruits commonly consumed in the Netherlands. J Agric Food Chem 992;40: Hertog MGL, Hollman PC, van de Putte B. Content of potentially anticarcinogenic flavonoids in tea infusions, wines and fruit juices. J Agric Food Chem 993;4: Manach C, Regerat F, Texier O, et al. Bioavailability, metabolism, and physiological impact of 4-oxo-flavonoids. Nutr Res 996;6: Feskanich D, Rimm EB, Giovannucci EL, et al. Reproducibility and validity of food intake measurements from a semiquantitative food frequency questionnaire. J Am Diet Assoc 993;93: Kircher T, Nelson J, Burdo H. The autopsy as a measure of accuracy of the death certificate. N Engl J Med 985; 33: Prineas RJ, Folsom AR, Kaye SA. Central adiposity and increased risk of coronary artery disease mortality in older women. Ann Epidemiol 993;3: Kushi LH, Fee RM, Folsom AR, et al. Physical activity and mortality in postmenopausal women. JAMA 997;277: Kushi LH, Folsom AR, Prineas RJ, et al. Dietary antioxidant vitamins and death from coronary heart disease in postmenopausal women. N Engl J Med 996;334:56-62.