|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2000 American Society for Clinical Oncology Dietary Fat Intake and Endogenous Sex Steroid Hormone Levels in Postmenopausal WomenFrom the Channing Laboratory, Department of Medicine, and Division of Preventive Medicine, Harvard Medical School and Brigham and Women’s Hospital; Department of Obstetrics and Gynecology, Brigham and Women’s Hospital; Departments of Epidemiology, Nutrition, and Biostatistics, Harvard School of Public Health, Boston, MA. Address reprint requests to Michelle D. Holmes, MD, DrPH, Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women’s Hospital, 181 Longwood Ave, Boston, MA 02115.
PURPOSE: To examine the relationship between plasma levels of reproductive sex steroid hormones in postmenopausal women and their reported fat intake. METHODS: We measured plasma sex steroid hormones levels in plasma collected in 1989 and 1990 from 381 healthy postmenopausal women. For each woman, we measured fat intake in 1986 and 1990 by a food-frequency questionnaire. The cross-sectional associations between the percentage of energy from total and specific types of dietary fat intake and plasma hormone levels were assessed by linear regression, controlling for energy intake, obesity, and protein intake. RESULTS: The plasma estradiol level was 4.3% lower (95% confidence limits, -8.3%, -0.2%) for a substitution of 5% of energy from fat intake for an equivalent amount of energy from carbohydrate when adjusted for obesity and other covariates. Estradiol was also inversely associated with all other fat types except trans fat; the inverse associations with vegetable fat and marine omega-3 fats were statistically significant. CONCLUSION: We observed an inverse association between total fat intake averaged over 4 to 5 years and estradiol levels. This result is inconsistent with the hypothesis that fat intake predisposes to breast cancer risk by raising endogenous estrogen levels.
ENDOGENOUS reproductive hormones have been strongly implicated in the etiology of breast cancer; most studies in postmenopausal women show increased risk with elevated estrogen levels.1,2 On the basis of animal studies,3,4 international comparisons,4-6 and a meta-analysis of case-control studies,7 high intake of total dietary fat has been postulated to increase breast cancer risk. The fact that a large randomized trial, the Women’s Health Initiative, is attempting to lower breast cancer risk with a low-fat diet attests to the appeal of this hypothesis. However, mechanisms by which high intake of dietary fat could increase breast cancer risk have not been established. If high dietary fat intake were to increase estrogen levels, this would provide a plausible mechanism. In a recent meta-analysis of metabolic studies, significantly lower estradiol levels were found in the low-fat intervention groups.8 However, most of the component studies in the meta-analysis had no concurrent control group and were confounded by weight loss in the intervention group.9
Prospective studies of dietary fat and breast cancer, which unlike case-control studies are not prone to recall bias, do not support the dietary fat hypothesis.10-12 For a protective effect to be evident, some authors have suggested that fat intake must account for
NHS Subjects and Blood Sample Collection In 1976, the NHS cohort was established when 121,700 female registered nurses from across the United States, aged 30 to 55 years, answered a mailed questionnaire on risk factors for cancer and cardiovascular disease. Every 2 years since, we have sent follow-up questionnaires to NHS participants. For this analysis, information on height and age at menarche was assessed in 1976 and on weight at the age of 18 in 1980. Information on parity and participant’s age at the birth of the participant’s first child was last assessed in 1984, when participants were  38 years of age. Information on age at menopause, smoking status, weight change since the age of 18, family history of breast cancer, and personal history of benign breast disease was assessed in 1990. Physical activity was also measured in 1990 by a validated and reproducible questionnaire13 that requested self-report of weekly leisure time activity estimated for the past year. Scores were calculated in Metabolic Equivalent Task (MET) hours per week, with one MET-hour defined as the energy expended in sitting quietly for one hour. Higher MET values for other activities reflect multiples of energy expenditure needed for sitting quietly; for example, 10 MET-hours would represent 21/2 hours of brisk walking. Blood samples were collected in 1989 and 1990 from 32,826 NHS participants who were from 43 to 69 years of age at the time, as previously detailed.14 Each woman was sent a kit that contained all supplies needed for blood collection, plus a supplemental questionnaire about menopausal status, recent postmenopausal hormone use, time since last meal, and time of day of blood sampling. Participants arranged to have their blood drawn and then mailed the whole-blood sample cooled with an enclosed ice pack via overnight mail. We have previously documented the stability of the hormones during the period of transport.15 After receipt in our laboratory, samples were centrifuged, divided, and frozen in the vapor phase of liquid nitrogen freezers (-130°C or colder). Hormone levels for the plasma samples used in this analysis were assayed in three batches between February and September 1993. This study was approved by the Institutional Review Board of the Brigham and Women’s Hospital, Boston, MA. Women included in this analysis were control subjects in a nested case-control study of plasma hormone levels and breast cancer risk.16 In addition, 49 women who had reported low fat intake (< 25% and < 19% of energy on the 1986 and 1990 food-frequency questionnaires [FFQs], respectively) were added to increase the range of fat intake. These cut points were chosen to maximize the number of women with low fat intakes on both questionnaires, and all women who met these criteria were included in this analysis. These samples were interspersed with the control samples and were assayed at the same time. Women included were all postmenopausal (no menses for at least 12 months before blood sampling) and had not used hormones for at least 3 months before the blood collection. Participants had no previously diagnosed cancer (except nonmelanoma skin cancer) and could not have implausible scores for total energy intake (< 500 kcal or > 3,500 kcal/d).
Semiquantitative FFQs
Laboratory Analysis In each batch of samples, we assessed laboratory precision by the inclusion of replicate samples unidentifiable to laboratory personnel. Within-batch laboratory coefficients of variation ranged from 6% (percentage bioavailable estradiol) to 13.6% dehydroepiandrosterone (DHEA).16
Statistical Analysis
The association between the percentage of energy from fat intake and hormone levels was also assessed by linear regression. The robust variance was used to ensure valid inference even if the regression residuals were not normally distributed.19 Four-knot restricted cubic spline models for the regression of hormone levels on percentage of energy from type of fat, adjusted for other covariates, were used to examine linearity.20 These models were compared with those that assumed a linear association to obtain a likelihood ratio test for nonlinearity. In linear regression, the difference in plasma hormone levels was modeled on the natural logarithm scale. Solving (1-eß A secondary analysis was performed to determine whether results could be attributed to a few points having undue influence. In these linear regressions, dfbetas, which approximate the change in the beta coefficient that would result from deletion of the observation from the data set, were calculated for each observation. Participants were omitted whose dfbeta for the multivariate adjusted slope of hormone levels on percentage of energy from fat type was greater than the absolute value of the median dfbeta plus or minus three times the interquartile range of the dfbetas.21
Characteristics of the 381 women included in the study are listed in Table 1. The mean age was 62.6 years, mean height 64.4 in, and mean body mass index in 1990 25.7 kg/m2. They consumed an average of 1,750 kcal per day, including 6.8 g of alcohol, 74 g of protein, and 20 g of fiber. The average percentage of energy intake from fat was 29% (range, 10% to 46%), which reflects the oversampling of women with low fat intake. Of total fat intake, 55% was from animal fat and 45% from vegetable fat. The mean percentage of energy intake from carbohydrates was 51%. Fourteen percent had a first-degree family history of breast cancer, 36% had a personal history of benign breast disease, and 13% were current smokers.
The 49 women who had been selected for low fat intake had an average of 17% of energy intake from fat. Compared with other women, they consumed more fiber (27 v 19 energy-adjusted grams per day) and more carbohydrates (percentage of energy intake, 65% v 48%). They also had a lower body mass index in 1990 (23.7 v 26.0 kg/m2) and had gained less since the age of 18 (6 v 15 kg). There were no other statistically significant differences between the two groups in other factors listed in Table 1. Median plasma hormone levels, along with the tenth to the ninetieth percentile range, are listed in Table 2.
Table 3 shows the estimated geometric mean of plasma hormone levels across categories of fat intake at the reference level of all covariates. The mean levels of androstenedione and DHEA seemed to increase, and the mean levels of estradiol seemed to decrease in a somewhat linear fashion across categories of increasing fat intake.
In the multivariate models in Tables 4, 5, and 6, the inclusion of total energy and protein intake in each model means that the coefficient for fat can be interpreted as substitution of a percentage of energy from fat for an equal percentage of energy from carbohydrates. Table 4 shows the adjustments for age and energy for each hormone and the multivariate linear relationship between plasma hormone levels and percentage of energy intake from fat. In models adjusted only for age and total energy intake, fat intake was positively associated with higher levels of all hormones and inversely associated with sex hormone–binding globulin (SHBG). Models adjusted only for age were similar.
However, many results were reversed in the multivariate models. The association of fat intake with levels of SHBG was positive although not statistically significant. For a 5% increase in energy from total fat intake, plasma estradiol level was 4.3% lower (95% confidence limits [CL], -8.3%, -0.2%). Bioavailable estradiol level, bioavailable estradiol percentage, free estradiol percentage, estrone sulfate level, and testosterone level were also inversely associated with total fat intake, although the linear associations were not statistically significant. This reversal of association was caused mainly by control for measures of obesity: body mass index at the age of 18 and weight gain since the age of 18. In a multivariate model not adjusted for obesity, a 5% increase in energy from total fat had a nonsignificant positive association with estradiol level, 0.3% (95% CL, -4.2, 4.9). However, in a model that contained as the only covariates age, energy, and obesity, a 5% increase in energy from fat was negatively associated with estradiol level, -2.3% (95% CL, -6.1, 1.6). Results for bioavailable estradiol level, estradiol free percentage, estrone level, SHBG level, and testosterone level were similar. Fiber intake was not directly associated with hormone levels in this study. However, dietary fiber may modify the association between fat intake and hormone levels, with the greatest inverse association seen in women having the lowest fiber intake. The change in estradiol levels associated with a 5% of energy increase in fat intake across tertiles (lowest to highest) of energy-adjusted fiber intake were -14.5% (95% CL, -19.1, -9.7), 1.5% (-6.6, 10.4), and 2.9% (-6.4, 1.3) (P [for the interaction term] = .25). Results were similar for bioavailable estradiol and estrone sulfate levels. Tables 5 and 6 are similar to Table 4 except that the relationship between intake of specific types of fat and plasma hormone levels is examined. In each table, all the major types of fat that make up total fat intake are included in the same model; therefore, the estimated effect of each type of fat can be interpreted as a substitution for an equivalent amount of energy from carbohydrates, holding all other aspects of the data constant. Table 5 shows the relationship of plasma hormone levels with animal and vegetable fat intake. Estradiol level remained inversely associated with both animal and vegetable fat intake, and the linear association with vegetable fat intake was statistically significant (P = .03). SHBG level remained positively associated with animal fat intake (P = .02). In Table 6, which examines saturated, monounsaturated, polyunsaturated, trans-unsaturated, and omega-3 (from fish) fat intake, estradiol level remained inversely associated with intake of all fat types except trans fats; the linear association with omega-3 fats from fish was statistically significant. Estrone level also had a statistically significant inverse association with omega-3 fats from fish. Strong relationships were seen between some types of fat and some of the androgens. Polyunsaturated fat intake was inversely associated with androstenedione, testosterone, DHEA, and dehydroepiandrosterone sulfate (DHEAS): a 5% higher energy intake from polyunsaturated fat was associated with 19% to 46% lower plasma levels of these hormones. In addition, a 5% higher energy intake from monounsaturated fat was associated with a 35% higher plasma level of DHEA and DHEAS. The expression of a linear relationship can be distorted by a few influential points. The linear regressions in Tables 4 and 6 were repeated using the procedure for deletion of influential points reported in Statistical Methods. In each model, an average of 8% of the points (range, 3% to 13%) were removed for undue influence. All results remained substantially the same.
We observed inverse associations between reported dietary intake of total fat averaged over 4 years and five plasma estrogens and testosterone when adjusted for obesity and other covariates; this association was statistically significant for estradiol. Estradiol level, as well as estrone sulfate, estrone, and DHEAS levels, has been positively associated with breast cancer risk in this population.16 Therefore, the present results are contrary to the hypothesis that high fat intake predisposes to breast cancer risk by raising endogenous estrogens. They are, however, consistent with our previously reported finding of a modest inverse relationship between total fat intake and breast cancer risk.12 The inverse association between fat intake and hormone levels that we observed depended primarily on control for obesity. It is important to examine this issue critically. If obesity were an intermediate step between high fat intake and elevated hormone levels (ie, high fat intake leads to obesity, and increased adipose tissue leads to higher hormone levels), then adjustment for obesity would not be appropriate. However, if obesity is independently associated with both fat intake and hormone levels (possibly because health-conscious women restrict both fat intake and total energy) and not an intermediary, then it is a confounder and should be adjusted for. We examined the cross-sectional association between dietary fat intake (assessed in 1980 and updated in 1984, 1986, and 1990) and body mass index from 1980 through 1994 in the NHS. We used regression methods for clustered data, because this analysis was based on 415,077 observations in 85,804 women.22 We adjusted for the following factors found to be associated with obesity in a previous study23: smoking status, race, marital status, level of physical activity, parity, body mass index at age 18, alcohol use, and total energy intake. We additionally adjusted for protein intake to determine the association of body mass index with substituting carbohydrate for fat intake.
Cross-sectionally, women who consumed a higher percentage of total energy from fat in this cohort were slightly (0.08 kg/m2) heavier (95% confidence interval, 0.07 to 0.09; P [for a linear association] We also conducted a longitudinal analysis with 450,013 observations using similar methods in the same women. We examined how the percentage of energy from fat intake, cumulatively averaged and updated over time, predicted the change in body mass index over the next 2-year period. Longitudinally, fat intake did not predict any increase in body mass index. An increment of 5% of energy in dietary fat was associated with -0.003 kg/m2 in body mass index (95% CL, -0.005, -0.001; P [for an inverse linear association] = .02). Thus, this cross-sectional association seen between fat intake and body mass index is probably a result of confounding by health consciousness: women who strive to be lean because they believe it to be healthy also consume a lower fat diet because they have been told that it is healthy. This lack of association between dietary fat and body fat is also consistent with randomized trials that have lasted 1 year or longer.24 Because fat intake does not predict weight gain over time in this cohort, obesity does not seem to be an intermediary between fat intake and hormone levels, and there is no reason not to adjust for obesity. Because obesity is associated cross-sectionally with fat intake in this cohort and is also associated with hormone levels,14 it is a confounder and should be adjusted for. For these reasons, we believe that the multivariate associations of fat intake with hormones, including adjustment for obesity, are the appropriate ones to examine. We note, however, that even when we did not adjust for obesity, the association between dietary fat and estrogen levels was essentially null rather than significantly positive. Previously, the association of dietary fat and endogenous estrogens has been examined in short-term interventions in which women were fed a low-fat diet and their plasma hormones were measured. In a recent meta-analysis of the results from 13 interventions, including four among postmenopausal women, the authors reported a 23% decline (95% CL, -27.7%, -18.1%) in serum estradiol levels among postmenopausal women who consumed 10% to 24% of energy from fat over a period of 3 weeks to 5 months.8 However, the validity of this meta-analysis was compromised by the validity of the component studies. None of the four component studies included a simultaneous control group.25-28 In addition, in three of the four studies, women experienced statistically significant weight loss,25,27,28 and a strong positive association between adiposity and postmenopausal estrogen levels has been well documented.16,29,30 Although small losses in weight are often seen in short-term studies of dietary fat reduction, they are not sustained in longer-term studies.31 Thus, studies that last for only a few weeks or months may prompt misleading conclusions about the long-term effects on blood hormone levels. In our present analysis, we found that control for measures of obesity caused a reversal in the associations seen. Although not included in the meta-analysis because it was conducted among women with breast cancer, one randomized trial among postmenopausal women has been reported. Rose et al32 reported a significant reduction in serum estradiol levels among women with initially higher levels, after an 18-month low-fat diet intervention. However, as described elsewhere,18 the comparisons in this analysis were within the intervention group, not between the intervention and control groups. In addition, women with the lowest initial estradiol levels were excluded, leaving the reduction to be seen only among women with initially high levels. Because a similar reduction in estradiol levels was seen in the control group, the results are consistent with regression toward the mean and no effect of dietary fat. In addition to intervention studies, several cross-sectional studies have examined the relationship between low-fat diet and plasma estrogens. In one of the largest of these, with 325 perimenopausal women, London et al33 reported no association between fat intake and serum estrogen levels. Likewise, a cross-sectional study of 253 postmenopausal women in Wisconsin showed no association between fat intake and serum estrone level.34 In another cross-sectional study of 88 Greek women, no association was noted between fat intake and urinary estrogen levels.35 However, in two cross-sectional studies that included 24 and 93 women, respectively, vegetarian women who consumed a low-fat diet were compared with nonvegetarian women who consumed a higher-fat diet. Both found statistically significantly lower serum or urinary estrogen levels among the vegetarians.36,37 However, in both of these studies, vegetarian women were less obese than nonvegetarian women, and the groups had many dietary differences other than fat intake. In the larger of these two studies, the study by Armstrong et al,36 adjustment for body mass index diminished the difference in urinary estrogens between vegetarian and nonvegetarian women, although it was still statistically significant; the smaller of these two studies, by Barbosa et al,37 did not adjust for obesity at all. In addition, Barbosa et al report a statistically significantly higher intake of fiber among the vegetarian women, but they did not adjust for it in the analysis,37 even though fiber intake has been hypothesized to lower estrogen levels.38 Little has been written about the effect of low-fat diet on hormones other than estrogens. Breast cancer risk has been positively associated with androgens.16,39 Of the interventions previously mentioned, only one, that by Ingram et al,28 examined the effect of a low-fat diet on serum androgens among postmenopausal women. A nonsignificant decrease in testosterone level and essentially no change in DHEAS level were reported.28 Both this study and the intervention by Prentice et al27 report a nonsignificant decrease in SHBG level with low-fat dietary intervention. The cross-sectional study from Wisconsin showed no association of dietary fat intake with serum androgen or SHBG levels.34 In the two cross-sectional studies that compared vegetarian women with nonvegetarian women, one reported no significant differences in plasma testosterone levels,36 the other no significant difference in DHEAS levels,37 and both no statistically significant difference in SHBG levels. Very few cross-sectional studies have examined the correlation between specific types of dietary fat and endogenous sex steroid hormones in postmenopausal women. London et al33 reported no association between serum estrogens and either saturated fat or linoleic acid. Likewise, Katsouyanni et al35 report no association between urinary estrogens and saturated, polyunsaturated, and monounsaturated fats.
Use of diet records to validate the FFQ measure of fat intake in this study has been criticized because of hypothesized correlated error between the two self-reported measures of dietary intake.40 However, we used the same analysis as in Table 3 to predict fasting serum triglyceride levels, which are known to be inversely related to fat intake,41 in 185 women. These women included 84 from the sex steroid hormone analysis for whom we had triglyceride levels, 41 who were diagnosed with breast cancer more than 2 years after dietary assessment, and 60 who were controls in a study of myocardial infarction. All were postmenopausal, nondiabetic, not taking cholesterol-lowering drugs, and had fasted overnight. Controlling for fiber intake and the covariates listed in Table 3, we found the geometric mean serum triglyceride levels (mg/dL) across categories of percentage of energy from fat intake to be as follows: 156, 139, 129, 103, 85, and 70 mg/dL, corresponding to In summary, although a low-fat diet has been hypothesized to reduce levels of endogenous sex hormones, previous intervention studies have been inconclusive because of limitations in design or analysis. Our study was larger than those of exclusively postmenopausal women and had the advantage of averaging dietary intake over 4 to 5 years, which reduced measurement error. Also, we were able to extend the range of fat intake by oversampling women with low fat intake and to cover multiple types of fat as well as multiple hormones. We measured hormones with excellent precision, having previously documented that a single measure reasonably reflects average long-term hormone levels.43 We were also able to control for potential confounding factors, such as obesity and alcohol intake. Limitations of this study include its cross-sectional nature. We were also limited in our ability to examine women with extremely low fat intake, as only 15% of the total population reported 20% or less of total energy from fat. We found no evidence that higher fat intake is associated with higher levels of any reproductive hormones in this group of postmenopausal women. In fact, we found that the levels of serum estradiol, a hormone associated with increased breast cancer risk, were inversely associated with fat intake. Although not consistently statistically significant, this inverse relationship was true for most types of fat, including animal, vegetable, saturated, monounsaturated, polyunsaturated, and omega-3 (from fish) fats; the exception was a nonsignificant positive association with trans-unsaturated fat intake. These results are consistent with our previous finding of a lack of association between low intake of total fat over 14 years of follow-up and a decreased risk of breast cancer.12 In addition, we found that DHEA and DHEAS levels had a strong positive association with monounsaturated fat intake and a strong inverse association with polyunsaturated fat intake; these novel findings need to be duplicated in other studies. Our findings suggest that adoption of a low-fat diet during midlife does not lower long-term endogenous estrogen levels, which has been the hypothesized mechanism by which a low-fat diet might prevent breast cancer.
Supported by the National Institutes of Health grants no. CA40356 and CA49449. D.J.H., G.A.C., and W.C.W. were supported by the Harvard Center for Cancer Prevention, Boston, MA. S.E.H. was partially supported by a career development award DAMD17-96-1-6021 from the United States Army Medical Research and Material Command, Fort Detrick, MD.
1. Bernstein L, Ross K: Endogenous hormones and breast cancer risk. Epidemiol Rev 15: 48-65, 1993 2. Thomas H, Reeves G, Key T: Endogenous estrogen and postmenopausal breast cancer: A quantitative review. Cancer Causes Control 8: 922-928, 1997[Medline] 3. Aylsworth C, Jone C, Trosko J, et al: Promotion of 7,12-dimethylbenz [a] anthracene-induced mammary tumorigenesis by high dietary fat in the rat: Possible role of intercellular communication. J Natl Cancer Inst 72: 637-641, 1984
4.
Prentice R, Kakar F, Hursting S, et al: Aspects of rationale for the Women’s Health Trial. J Natl Cancer Inst 80: 802-814, 1988 5. Carroll K, Khor H: Dietary fat in relation to tumorigenesis. Prog Biochem Pharmacol 10: 308-313, 1975[Medline] 6. Wynder E, Fujita Y, Harris R, et al: Comparative epidemiology of cancer between the United States and Japan: A second look. Cancer 67: 746-763, 1991[Medline]
7.
Howe G, Hirohata T, Hislop T, et al: Dietary factors and risk of breast cancer: Combined analysis of 12 case-control studies. J Natl Cancer Inst 82: 561-569, 1990
8.
Wu A, Pike M, Stram D: Meta-analysis: Dietary fat intake, serum estrogen levels, and the risk of breast cancer. J Natl Cancer Inst 91: 529-534, 1999
9.
Hunter D: Role of dietary fat in the causation of breast cancer. Counterpoint. Cancer Epidemiol Biomarkers Prev 8: 9-13, 1999 10. Willett W, Hunter D, Stampfer M, et al: Dietary fat and fiber in relation to risk of breast cancer. JAMA 268: 2037-2044, 1992[Abstract]
11.
Hunter D, Spiegelman D, Adami H-O, et al: Cohort studies of fat intake and the risk of breast cancer: A pooled analysis. N Engl J Med 334: 356-361, 1996
12.
Holmes M, Hunter D, Colditz G, et al: Association of dietary intake of fat and fatty acids with risk of breast cancer. JAMA 281: 914-920, 1999
13.
Wolf AM, Hunter DJ, Colditz GA, et al: Reproducibility and validity of a self-administered physical activity questionnaire. Int J Epidemiol 23: 991-999, 1994
14.
Hankinson S, Willett W, Manson J, et al: Alcohol, height, and adiposity in relation to estrogen and prolactin levels in postmenopausal women. J Natl Cancer Inst 87: 1297-1302, 1995
15.
Hankinson S, London S, Chute C, et al: Effect of transport conditions on the stability of biochemical markers in blood. Clin Chem 35: 2313-2316, 1989
16.
Hankinson S, Willett W, Manson J, et al: Plasma sex steroid hormone levels and risk of breast cancer in postmenopausal women. J Natl Cancer Inst 90: 1292-1299, 1998
17.
Willett W, Sampson L, Stampfer M, et al: Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 122: 51-65, 1985 18. Willett W. Nutritional Epidemiology. New York, NY, Oxford University Press, 1998 19. White H: A heteroscedasticity-consistent covariance matrix estimator and a direct test for heteroscedasticity. Econometrics 48: 817-838, 1980 20. Durrelman S, Simon R: Flexible regression models with restricted cubic splines. Stat Med 8: 551-561, 1989[Medline] 21. Belsly D, Kuh E, Welsch R: Regression Diagnostics. New York, NY, Wiley, 1980
22.
Liang K-Y, Zeger SL: Longitudinal data analysis using generalized linear models. Biometrika 73: 13-22, 1986 23. Holmes MD, Stampfer MJ, Wolf AM, et al: Can behavioral risk factors explain the difference in body mass index between African-American and European-American women? Ethn Dis 8: 331-339, 1998[Medline] 24. Katan MB, Grundy SM, Willett WC: Should a low-fat, high-carbohydrate diet be recommended for everyone? Beyond low-fat diets. N Engl J Med 337: 563-566, 1997 25. Crighton I, Dowsett M, Hunter M, et al: The effect of a low-fat diet on hormone levels in healthy pre- and postmenopausal women: Relevance for breast cancer. Eur J Cancer 28A: 2024-2027, 1992 26. Heber D, Ashley J, Leaf D, et al: Reduction of serum estradiol in postmenopausal women given free access to low-fat high-carbohydrate diet. Nutrition 7: 137-139, 1991[Medline]
27.
Prentice R, Thompson D, Clifford C, et al: Dietary fat reduction and plasma estradiol concentration in healthy post-menopausal women. The Women’s Health Trial Study Group. J Natl Cancer Inst 82: 129-134, 1990 28. Ingram D, Bennett F, Wilcox D, et al: Effect of low-fat diet on female sex hormone levels. J Natl Cancer Inst 79: 1225-1229, 1987
29.
Kaye S, Folsom A, Soler J, et al: Associations of body mass index and fat distribution with sex hormone concentrations in postmenopausal women. Int J Epidemiol 20: 151-156, 1991 30. Madigan M, Troisi R, Potischman N, et al: Serum hormone levels in relation to reproductive and lifestyle factors in postmenopausal women (United States). Cancer Causes Control 9: 199-207, 1998[Medline] 31. Willett W: Is dietary fat a major determinant of body fat? Am J Clin Nutr 67: 556S-562S, 1998 (suppl)[Abstract] 32. Rose D, Connolly J, Chlebowski R, et al: The effects of a low-fat dietary intervention and tamoxifen adjuvant therapy on the serum estrogen and sex hormone-binding globulin concentrations of postmenopausal breast cancer patients. Breast Cancer Res Treat 27: 253-262, 1993[Medline]
33.
London S, Willett W, Longcope C, et al: Alcohol and other dietary factors in relation to serum hormone concentrations in women at the climacteric. Am J Clin Nutr 53: 166-171, 1991 34. Newcomb P, Klein R, Klein B, et al: Association of dietary and life-style factors with sex hormones in postmenopausal women. Epidemiology 6: 318-321, 1995[Medline] 35. Katsouyanni K, Boyle P, Trichopolous D: Diet and urine estrogens among postmenopausal women. Oncology 48: 490-494, 1991[Medline] 36. Armstrong B, Brown J, Clarke H, et al: Diet and reproductive hormones: A study of vegetarian and nonvegetarian postmenopausal women. J Natl Cancer Inst 67: 761-767, 1981
37.
Barbosa J, Shultz T, Filley S, et al: The relationship among adiposity, diet, and hormone concentrations in vegetarian and nonvegetarian postmenopausal women. Am J Clin Nutr 51: 798-803, 1990 38. Rose D: Dietary fiber, phytoestrogens, and breast cancer. Nutrition 8: 42-51, 1992
39.
Secreto G, Toniolo P, Berrino F, et al: Serum and urinary androgens and risk of breast cancer in postmenopausal women. Cancer Res 51: 2572-2576, 1991
40.
Prentice R: Measurement error and results from analytic epidemiology: Dietary fat and breast cancer. J Natl Cancer Inst 88: 1738-1747, 1996
41.
Mensink R, Katan M: Effect of dietary fatty acids on serum lipids and lipoproteins: A meta-analysis of 27 trials. Arterioscler Thromb 12: 911-919, 1992 42. Knopp R, Walden C, Retzlaff B, et al: Long-term cholesterol-lowering effects of four fat-restricted diets in hypercholesterolemic and combined hyperlipidemic men. The Dietary Alternatives Study. JAMA 278: 1509-1515, 1997[Abstract] 43. Hankinson S, Manson J, Spiegelman D, et al: Reproducibility of plasma hormone levels in postmenopausal women after a 2-3 year period. Cancer Epidemiol Biomarkers Prev 4: 649-654, 1995[Abstract] Submitted December 19, 1999; accepted June 14, 2000. This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
INTRODUCTION
20% of energy. These authors maintain that such a protective effect has not been found in cohort studies because they were conducted in Western populations, in whom the level of fat intake is rarely this low.
) x 100, where ß is the estimated regression slope and 
