AJCN Tufts Nutrition Symposium, Boston Sept 24-26
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Martin, L. J
Right arrow Articles by Boyd, N. F
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Martin, L. J
Right arrow Articles by Boyd, N. F
Agricola
Right arrow Articles by Martin, L. J
Right arrow Articles by Boyd, N. F
American Journal of Clinical Nutrition, Vol. 84, No. 4, 920-928, October 2006
© 2006 American Society for Nutrition

ORIGINAL RESEARCH COMMUNICATION

Intervention with a low-fat, high-carbohydrate diet does not influence the timing of menopause1,2,3

Lisa J Martin, Carolyn V Greenberg, Valentina Kriukov, Salomon Minkin, David JA Jenkins and Norman F Boyd

1 From the Campbell Family Institute for Breast Cancer Research, Ontario Cancer Institute, University Health Network, Toronto, Canada (LJM, CVG, VK, SM, and NFB), and the Department of Nutritional Sciences, University of Toronto, Toronto, Canada (DJAJ)

2 Supported by grants from the Ontario Ministry of Health, the Canadian Breast Cancer Research Alliance, and the American Institute for Cancer Research and by a Postdoctoral Fellowship from the Canadian Institutes for Health Research (to LJM).

3 Reprints not available. Address correspondence to LJ Martin, 9-409, Campbell Family Institute for Breast Cancer Research, Ontario Cancer Institute, 610 University Avenue, Toronto, ON M5G 2M9, Canada. E-mail: lmartin{at}uhnres.utoronto.ca.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Background: Later age at menopause is associated with a greater risk of breast cancer. Dietary factors may at least partially influence breast cancer risk through an effect on the age at menopause.

Objective: We studied the effect of a low-fat, high-carbohydrate (LFHC) dietary intervention on the timing of menopause in women with greater risk of breast cancer.

Design: The study population included participants from an LFHC dietary intervention trial for the prevention of breast cancer in women with extensive mammographic density, a strong risk factor for breast cancer. Women who were premenopausal at baseline (n = 2611) were followed for an average of 7 y for menopause. Survival analysis was used to compare the time to menopause between the LFHC and control groups and to assess other factors associated with age at menopause.

Results: The LFHC intervention did not affect the time to natural menopause overall (P = 0.72 for log-rank test comparing study groups; n = 699 events). An observed interaction between study group and baseline body mass index (BMI; P = 0.01) indicated that the intervention group experienced earlier menopause than did the control group when BMI was low and that a higher BMI was associated with later menopause in the intervention group only. Greater parity, weight, and education were associated with later menopause, and greater age at first birth and baseline smoking were associated with earlier menopause.

Conclusions: Overall, the LFHC dietary intervention did not influence the timing of menopause. Factors associated with age at menopause in this population were consistent with those reported in other populations.

Key Words: Menopause • mammographic density • low-fat diet • dietary fat • dietary intervention • breast cancer


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
International variations in breast cancer rates and increases in breast cancer incidence in immigrants from low-risk to high-risk countries suggest that environmental factors such as diet have a strong influence on breast cancer risk (1). In particular, higher per capita dietary fat intake is strongly associated with higher breast cancer rates across countries (1). Observational studies of the relation of dietary fat to breast cancer incidence according to the measurement of dietary intakes in individual persons have varied; however, our meta-analysis showed that a higher intake of fat is significantly associated with a greater risk of breast cancer (2). Recently, the Women's Health Initiative did not report a significant effect of a low-fat dietary pattern over 8 y on breast cancer incidence in postmenopausal women, but secondary analyses suggested that women who adhered to the study protocol and those with a higher baseline fat intake experienced a reduction in breast cancer risk with the dietary intervention (3).

The main reproductive risk factors for breast cancer are early menarche, late age at first birth, and late age at menopause (4). If dietary fat influences the risk of breast cancer, it may do so, at least partially, through effects on ovarian function. Secular trends (5) and changes in age at menarche in immigrant studies (6) suggest that age at menarche declines as nutrition and body weight increase (7). In prospective studies, higher energy (8) and fat (9) intakes were associated with earlier menarche, and higher fiber intakes were associated with later menarche (10).

In contrast, international variations and secular trends do not provide consistent evidence for an influence of nutrition on the age at menopause (5, 11, 12). Few epidemiologic studies have evaluated the association of age at menopause and dietary intakes measured in individual persons. In a cross-sectional study of Japanese women, higher intakes of dietary fat and cholesterol were associated with later, and soy intake was associated with earlier, menopause (13). In a cross-sectional study of women in the United Kingdom, higher meat and alcohol intakes were associated with later menopause (14). However, when the premenopausal women from those cross-sectional studies were followed prospectively, only alcohol intake was confirmed as being associated with age at menopause (15, 16). Recently, a large cohort study in Germany reported that higher dietary fat intakes were associated with later menopause and higher dietary fiber and carbohydrate intakes were associated with earlier menopause, but the associations were of borderline significance (17).

We are currently carrying out a randomized clinical trial to test whether intervention with a low-fat, high-carbohydrate (LFHC) diet reduces the incidence of breast cancer in women who are at greater risk of the disease because of extensive mammographic density. The purpose of the current study was to determine the effect of the LFHC dietary intervention on the time to menopause. We also examined whether the factors associated with age at menopause in this selected population were similar to those identified in other populations.


   SUBJECTS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
General method
This study involved long-term follow-up of the subset of participants in a multicenter trial of dietary intervention who were premenopausal at entry to that trial. Women in that trial are at increased risk of breast cancer because of the presence of extensive mammographic density (18, 19). The term mammographic density refers to the appearance of the breast on mammography and reflects the amounts of stromal and epithelial tissue in the breast (20). Women were randomly assigned to the LFHC dietary intervention (goal of 15% of energy from fat) or the control (no dietary advice provided) group. Women in both study groups were asked to complete an annual in-person interview to update information on diet and demographic, reproductive, and medical factors.

Subjects
Women were recruited from participating mammography centers and were considered potentially eligible if they had a mammogram showing ≥50% of the breast area to be mammographically dense. Additional criteria for inclusion included age of 30 to 65 y, body mass index (BMI; in kg/m2) of 19 to 27, residence within easy commuting distance of a participating center, and ability to read and write English. Criteria for exclusion were history of cancer (except nonmelanomatous skin cancer), pregnancy (or planning a pregnancy), or breastfeeding, current adherence to a medically prescribed diet for any reason, habitual (ie, ≥4 times/wk) consumption of > 1 meal/d in a restaurant, previous mammoplasty (reduction or augmentation), or previous or current treatment for the reduction of blood lipids. Potentially eligible women attended 2 screening appointments to ensure that they could keep appointments and record food intake with adequate detail and consistency. Eligible women were randomly assigned to the LFHC or the control group. Of the study participants, {approx}94%, 2%, and 2% reported their ethnic background as white, black, and Chinese, respectively.

Written informed consent was obtained from all subjects. The institutional review boards at each participating center approved the study protocol.

Dietary intervention
The goal of the dietary intervention was to reduce dietary fat intake to 15% of energy and to increase carbohydrate intake to 65% of energy without changing protein or total energy intake. On the basis of a dietary assessment conducted at the second screening interview, a diet plan was calculated for each study participant by using a food exchange system. Exchange lists for fat, protein foods, grains, fruit, vegetables, and dairy products were provided to all study participants; they were also given an extensive shopping guide, listings of available low-fat food products, suggestions for eating in restaurants, and {approx}100 low-fat recipes. Photographs of sample meals and portion sizes were also used to visually illustrate the appropriate proportions of foods to be included in the diet. Intervention group subjects were asked to meet individually with a study dietitian on a monthly basis during the first year of the study, quarterly during the second year, and twice yearly thereafter. Control group subjects met with a study dietitian quarterly in the first year, twice in the second year, and annually thereafter. They were not asked to make any dietary changes and were not given any specific dietary advice.

Data collection
Study dietitians collected information on demographic, reproductive, anthropometric, and other health-related factors at baseline and at each annual visit by using a questionnaire designed for this study. Weight was measured to 0.1 kg on a balance scale while the participant was wearing light clothing and no shoes. Height was measured to 0.1 cm by using a stadiometer.

Dietary intake was assessed by using food records collected on 3 nonconsecutive days before each study visit. Study participants were instructed to record, by weight or in household measures, all food and beverages consumed and to provide brand names and recipes where possible. The dietitian reviewed completed food records with each study participant for accuracy and completeness. Details regarding cooking methods, added fat, added sugar, brand names, and portion sizes were verified by the dietitian.

Because of the large number of food records collected and the time required for nutrient analysis, only selected food records collected during the trial underwent nutrient analysis. Food records for randomly selected subsets of study participants were analyzed for overall monitoring of dietary compliance. Food records were also analyzed for subsets of participants included in several substudies, such as the studies on the effect of the dietary intervention on mammographic density (21) and hormone concentrations (22). Food records were analyzed with the use of NUTRIENT DATA SYSTEM software (version 2.4–2.91; Nutrition Coordinating Center, School of Public Health, University of Minnesota, Minneapolis, MN) by trained dietitians who were blinded to study participant and study group. This system was modified to include and accommodate Canadian foods.

Identification of cohort of premenopausal women
A total of 4693 women were recruited to the dietary intervention trial between March 1988 and November 1998. At entry to the trial, women were asked whether they had had a menstrual period in the previous 6 mo and whether they were taking any exogenous sex hormones. Women who had ≥1 menstrual period during the 6 mo before entry to the trial and who were not taking any exogenous sex hormones [hormone replacement therapy (HRT)] or oral contraceptives (OC) were considered premenopausal at baseline (n = 2820; Figure 1Go). Women who did not attend any annual follow-up visits after baseline were excluded, which left 2611 women for analysis. These women were considered to be at risk of menopause until the occurrence of a menopausal event, a censoring event, or the end of follow-up for this study (31 July 2003).


Figure 1
View larger version (25K):
[in this window]
[in a new window]
  		FIGURE 1. Flow chart of identification of menopause events.

 
Definition of menopause and censoring events
Natural menopause
At each annual visit, the women were asked whether they had had a menstrual period in the previous 6 mo, and, therefore, we used ≥6 mo of amenorrhea for the initial definition of menopause. Women who reported that they had not had a menstrual period in the previous 6 mo, had not undergone a hysterectomy or oophorectomy, and were not taking exogenous hormones were classified as experiencing a natural menopause at the date of their last menstrual period (LMP). If women reported further natural menstrual periods at subsequent study visits, they were recoded as premenopausal, or the date of the LMP was changed, or both.

In published reports, it has been common to use 12 mo of amenorrhea to identify natural menopause because, depending on age, a relatively high probability exists that a woman will have another menstrual period after only 6 mo of amenorrhea (23). Although the initial definition of natural menopause in our study was 6 mo of amenorrhea, at the end of follow-up, the mean (±SD) time since LMP for women classified as experiencing a natural menopause was 4.1 ± 2.7 y (range: 0.5–13.4 y). At the end of follow-up, only 74 women (10% of those experiencing natural menopause events) had been observed for <12 mo after their last LMP. The mean age of these women at the time of their LMP (51.4 ± 3.0 y) did not differ significantly from that of women who were observed to have ≥12 mo of amenorrhea (51.4 ± 3.2 y; n = 625). If women who had been observed for <12 mo after their LMP were censored instead of being defined as having experienced natural menopause, the results of all analyses would be virtually unchanged. Therefore, we classified these women as experiencing a natural menopause in all analyses. Women who experienced ≥6 mo of amenorrhea before starting HRT having surgery (oophorectomy or hysterectomy) were also considered to have experienced a natural menopause.

Hormone replacement therapy use
For women who started using HRT before natural menopause, the date of the natural LMP could not be observed. In most previous studies, these women were either excluded (16, 24-26) or censored (27, 28). These strategies may not be appropriate in populations, such as that in the current study, in whom rates of HRT use are relatively high. Women who start HRT before experiencing natural menopause may be approaching menopause (that is, experiencing symptoms) at that time, and therefore, the independence between outcome (menopause) and censoring assumed in the survival analysis may not be valid.

A competing risks framework (29) was used to assess whether censoring for HRT use was informative. Statistical analysis was performed by using 2 outcome definitions: natural menopause alone and combined outcomes of natural menopause and HRT use. In the analysis of natural menopause alone, HRT users were censored at the time of starting HRT, and the model assumes that the risk of menopause for HRT users was the same as that for women censored for other reasons. In contrast, the model using the combined events as the outcome assumes, by definition, that HRT users experienced menopause at the time of starting HRT.

Censored observations
Women who had a natural menstrual period in the 6 mo before their last study visit (n = 989), who dropped out of the trial (n = 114), or who developed a study endpoint such as cancer, other major illness, or death before experiencing 6 mo of amenorrhea (n = 68) were censored at the date of their last study visit. Women who had a surgical menopause (ie, oophorectomy; n = 49) or hysterectomy (n = 51) were censored at the date of surgery. Women who started OC use before natural menopause were censored at the time of starting OC (n = 162). Women who used OC or HRT for <3 mo and resumed natural menstrual cycles were considered to be premenopausal.

Statistical analysis
Effect of the dietary intervention on time to menopause
Baseline demographic characteristics of women in the LFHC dietary intervention and control groups were compared by using 2-sample t tests for continuous variables and chi-square tests for categorical variables. Changes in nutrient intake and weight over time were compared between the intervention and control groups by testing the significance of the study group x time of follow-up interaction by using repeated-measures (mixed model) analysis. Survival analysis was used to assess the effect of diet study group assignment on time from randomization to menopause. The date of randomization was used as time 0 for these analyses because any effect of the dietary intervention would become operative at that point (30). Survivor functions were estimated for each study group by using the Kaplan-Meier method and were compared between the LFHC and control groups by using the log-rank test. Adjustment was made for baseline age and for other baseline characteristics that differed significantly (P ≤ 0.05) between the study groups by using Cox proportional hazards regression. The association of a change in the percentage of energy intake from fat intake with time to menopause was also examined by using proportional hazards regression. Because food record data had not been analyzed for many women for >2 y after randomization, a change in the percentage of energy intake from fat was calculated as the intake at 1 y after randomization (or 2 y after randomization if the year 1 data were missing) minus the baseline intake.

Interaction terms between study group and baseline age, BMI, and the 2 variables most consistently associated with age at menopause in the literature—parity and smoking (31, 32)—were examined. The interaction terms between study group and these variables were simultaneously included in the proportional hazards model, and the partial log likelihood test was used to determine whether the addition of these terms significantly improved the fit over that of the main effects model. If this test was significant (P ≤ 0.05), then the individual interaction terms were examined for their statistical significance. In addition, the interaction term for group and baseline percentage of energy from fat was tested separately because the sample size was reduced for this analysis as a result of the missing food record data.

Association of selected variables with age at menopause
Cox proportional hazards regression was used to examine the associations between selected variables and age at menopause. Age at menopause, rather than time from randomization, was used as the outcome variable for these analyses, because the risk of menopause is expected to change more as a function of age than as a function of the time since randomization for these covariates (33). Models were stratified by study site.

Variables were selected on the basis of evidence in the literature for their association with breast cancer risk (34), with age at menopause (31, 32), or both. Age at menarche, age at first birth, and baseline anthropometric measures were analyzed as continuous measures. Parity was analyzed as a dichotomous variable (nulliparous or parous) and as number of births (none, 1, 2, or >2). Baseline smoking status (never, past, or current smoker), educational level, family history of breast cancer, and past OC use were analyzed as categorical variables. Physical activity was assessed on a 7-point scale based on frequency and intensity of activity. Examples of specific activities considered to be of light, moderate, and heavy intensity were provided. The scale ranged from the lowest category, "taking no deliberate activity to improve physical fitness," to the highest, "taking heavy exercise activity on a regular basis (2–3 sessions/wk)." Annual postrandomization weight measurements were examined as time-dependent variables in 3 ways: the change in weight from baseline to most recent postrandomization weight, the most recent postrandomization weight, and the average of all available weights.

Variables that were associated with age at natural menopause (P ≤ 0.2) in univariate Cox proportional hazards analysis were eligible for inclusion in the multiple regression models. A manual stepwise backward procedure was used to eliminate any covariates that were not significant at P < 0.05. The time-dependent weight variables were then added, one at a time, to the final models to assess whether they were more strongly related to age at menopause than was baseline weight.

Results of the proportional hazards regression models are expressed as hazard ratios (HR) and 95% CIs for a 1-unit difference in continuous variables. For categorical variables, the risk of menopause in each category was compared with the lowest (referent) category (expressed as HR and 95%CI). When appropriate, a P value for linear trend across categories was calculated. For descriptive purposes, estimates of median age at menopause for particular values of a covariate were estimated by using the baseline survivor function from the multiple regression model with other covariates set at the mean for continuous variables and the most common category for categorical variables (35).

The validity of the proportional hazards assumption was checked by simultaneously including interactions between time and each significant variable in the proportional hazards multiple regression models. All statistical analyses were performed by using SAS software (version 8; SAS Institute, Cary, NC).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Descriptive results
The subjects were followed for a mean of 6.9 ± 2.9 y (range: 0.9–14.5 y). We observed 699 events of natural menopause, and 479 women began using HRT before their LMP. The median age at natural menopause, estimated from survival analysis, was 54.0 y (95% CI: 53.4, 54.2 y), and that for combined events (natural menopause and HRT use) was 52.6 y (95% CI: 52.3, 52.8 y).

Selected baseline characteristics of the women by diet study group are shown in Table 1Go. On entry to the study, on average, women were 45 y old and had a BMI of 23. Women in the dietary intervention and control groups did not differ significantly in most characteristics; however, the number of births was slightly (but significantly) higher and age at menarche and age at first child were slightly (but significantly) lower in the intervention group.


View this table:
[in this window]
[in a new window]
  		TABLE 1 Selected baseline characteristics of women in the intervention and control groups1

 
Dietary compliance was assessed by using food record data that had been analyzed as part of the monitoring for the overall dietary intervention trial or for other substudies of the trial. The intakes of selected nutrients and the body weight at baseline and 1, 2, and 5 y after randomization for women who were premenopausal at entry to the trial and for whom at least some dietary records had been analyzed are shown in Table 2Go. Baseline intakes of energy, total fat, carbohydrate, protein, and fiber did not differ significantly between the study groups. After randomization, the proportion of energy from fat decreased from 30% to 20% in the intervention group, but no change was found in the control group. Carbohydrate intake increased significantly and energy intake fell by {approx}100 kcal in the intervention group; no change was found in the control group. The interaction term for change in nutrients between study group and follow-up time was highly significant (P < 0.0001) for all nutrients except percentage of energy from protein (P = 0.08). On average, women in the dietary intervention group lost {approx}1 kg at 1 y after randomization, whereas the control group gained 0.6 kg. After 1 y, participants in both the intervention and control groups tended to gain weight, but on average participants in the intervention group remained lighter than did those in the control group at all time points.


View this table:
[in this window]
[in a new window]
  		TABLE 2 Reported dietary intake of selected nutrients and weight by study group for women who were premenopausal at entry to the dietary intervention trial1

 
Effect of the dietary intervention on time to menopause
The Kaplan-Meier curves for the intervention and control groups for time to natural menopause are shown in Figure 2Go. The estimated median time from randomization to natural menopause in the intervention group was 10.1 y (95% CI: 9.6, 1.1 y) and that in the control group was 9.5 y (95% CI: 9.0, 10.4 y; P = 0.72). After adjustment for baseline age and variables that appeared to be unbalanced between the study groups (ie, number of births, age at menarche, and age at first child) by using proportional hazards regression, the HR for natural menopause between the intervention and control groups was 1.06 (95% CI: 0.91, 1.23; P = 0.44). This analysis was repeated by using the combined events of natural menopause and HRT use as the outcome. The estimated median time to combined events in the intervention group was 7.1 y (95% CI: 6.4, 7.5 y), and that in the control group was 6.9 y (95% CI: 6.3, 7.4 y; P = 0.39 for log-rank test of the Kaplan-Meier curves, data not shown). After adjustment for possible confounders (see above) by using proportional hazards regression, the HR for combined events was 1.01 (95% CI: 0.90, 1.13; P = 0.93).


Figure 2
View larger version (18K):
[in this window]
[in a new window]
  		FIGURE 2. Kaplan-Meier curves for time to natural menopause by diet study group. Survivor functions were estimated for each study group by using the Kaplan-Meier method and were compared between the low-fat, high-carbohydrate diet and control groups by using the log-rank test.

 
Data on the change in percentage of energy from fat between baseline and 1 or 2 y after randomization were available for 85% of the 2611 subjects (n = 1156 and 1055 in the control and intervention groups, respectively). No association was found between the change in percentage of energy from fat and the time to natural menopause (HR for 1-unit difference = 1.00; P = 0.96) or the time to combined events (HR for 1-unit difference = 1.00; P = 0.49).

The overall test for interaction terms between study group and baseline age, BMI, parity, and smoking was significant (P = 0.02). The only significant individual interaction was that of study group x baseline BMI (P = 0.01). The interaction for study group and baseline percentage of energy from fat, which was tested separately because the sample size was different as a result of missing data, was not significant (P = 0.22; n = 2365).

To illustrate the nature of the interaction between study group and BMI, the HR for natural menopause for control compared with intervention group was calculated by using the coefficients from the proportional hazards multiple regression model with BMI set at its 25th and 75th percentile values (21.5 and 24.5, respectively). When BMI was set at 21.5, the intervention group had a significantly higher risk of natural menopause than did the control group (HR = 1.20; 95% CI: 1.00, 1.45), but no effect of study group was found when BMI was set at 24.5 (HR = 0.93; 95% CI: 0.78, 1.10). Higher baseline BMI was associated with a lower risk of natural menopause in the intervention group (HR = 0.92; 95% CI: 0.88, 0.97) but not in the control group (HR = 1.00; 95% CI: 0.96, 1.05). The interaction between baseline BMI and study group was not explained by differences in changes in weight, changes in the percentage of energy from fat, or both, because the interaction term was changed by <5% when these variables were added to the proportional hazards multiple regression model.

Variables associated with age at menopause
In univariate analysis, age at menarche, past OC use, height, and physical activity level were not significantly associated with age at natural menopause or combined events (P > 0.2 for all). Family history (first-degree) of breast cancer was associated with a nonsignificantly lower risk of combined events of natural menopause and HRT use (P = 0.09) but was not associated with natural menopause alone (P = 0.39). Several weight-related variables—baseline weight (P = 0.0006), baseline BMI (P = 0.002), change in weight (P = 0.16), and most recent weight (P = <0.0001)—were positively associated with age at natural menopause in univariate analysis. Only the most recent weight was included in the multiple regression analyses, because it showed the strongest relation with age at menopause and was independent of all other anthropometric measures.

The results of multiple regression analyses of the association of variables with age at natural menopause and with the combined events of natural menopause and HRT use are shown in Table 3Go. The association of variables with age at natural menopause was similar in univariate and multivariate analyses. The results of the multiple regression model showed that women with ≥3 children had a risk of natural menopause 38% lower than that of women with no children (median age: 54.5 and 53.5 y, respectively), and the test for linear trend across categories of number of children was highly significant. Later age at first birth (1-y difference) was associated with a 2% greater risk and greater body weight (1-kg difference) was associated with a 2% lower risk of natural menopause. This effect corresponded to a 0.5-y later age at menopause for a 10-kg increase in weight. Current smokers had a 45% greater risk of menopause than did never and past smokers, and the median age at menopause for current smokers was {approx}1 y earlier than that for nonsmokers (53 and 52 y, respectively). Women with some postgraduate education had a risk of natural menopause 30% lower than that for those with high school education or less, and the trend across categories was significant. Median age at menopause was 1 y later in the highest than in the lowest category of education (54.8 and 53.9 y, respectively).


View this table:
[in this window]
[in a new window]
  		TABLE 3 Association of variables with risk of natural menopause alone and combined events of natural menopause and hormone replacement therapy use1

 
For the combined events of natural menopause and HRT use (see Table 3Go), the associations of all variables except age at first birth were also significant and were similar in magnitude and direction to those for natural menopause alone. The similarity of the results for the outcomes of natural menopause alone and the combined events of natural menopause and HRT use indicates that censoring for HRT use was not strongly informative in this situation.

Sensitivity analyses
Although OCs are sometimes used as low-dose hormone replacement in perimenopausal women, the women in this study who began OC use before their LMP (n = 162) were censored. When OC use was considered as an event in the combined analysis (natural and HRT use), the HRs from the multivariate model were similar. In addition, when oophorectomy was considered as an event of natural menopause, the HRs for the multiple regression model for natural menopause were similar.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
To our knowledge, this is the first study to examine the effect of a dietary intervention on the timing of menopause. In addition to being a randomized clinical trial, this study had several important strengths, including a large sample size, long follow-up period, and regular prospective assessment of menopausal status. Therefore, it provides strong evidence that an LFHC dietary intervention in midlife does not influence the timing of the menopause.

A key issue in interpreting these results is whether a substantial difference in dietary fat intake was maintained between study groups. Dietary compliance, assessed by nutrient analysis of food records, indicated a substantial and persistent reduction in fat intake in the intervention group and almost no change in the control group. However, self-reports of dietary intake are known to have substantial measurement error (36). The difference in body weight between the intervention and control groups provides objective evidence for a sustained dietary difference between study groups. Blood concentrations of HDL cholesterol and triacylglycerol have been used as indicators of dietary fat intake in cross-sectional (37) and intervention (38) studies. The dietary intervention in this study reduced HDL cholesterol and increased nonfasting triacylglycerol concentrations in the intervention group significantly below the concentrations in the control group, independent of changes in body weight (39). The effect of the dietary intervention on these blood lipids persisted over the long term (≥7 y), which provides evidence that a sustained difference in dietary fat and carbohydrate intake exists between the intervention and control groups.

This study focused specifically on the effect of dietary fat reduction on the timing of the menopause. It is possible that other nutrients, such as phytoestrogens, or more comprehensive changes in dietary patterns, including macronutrients, fruit, vegetables, and fiber, may affect the timing of menopause. In addition, the timing of the exposure to a diet may be important. The average baseline age of the women in this study was 45 y, which is slightly younger than the average age at the start of the perimenopausal period ({approx}47–48 y old). Perhaps dietary exposures earlier in life may be more important in determining the time of ovarian failure. For example, the reduction in age at menopause associated with the Dutch famine was primarily observed in women who were aged 2–9 y during the period of severe nutritional deprivation (40). However, Asian women living in Asia, who were likely exposed to a low-fat diet throughout their lives, do not consistently have a lower age at menopause than are Western women living in the West. The median age at menopause of Chinese women (49 y) was lower than that of Australian women (51 y) in one study (41), but other studies (5, 42) have reported a median age at menopause in Asian women that was very similar to that of Western countries (43). Japanese women living in the United States are significantly older at menopause than are white women living in the United States (44).

Although no overall effect of the LFHC dietary intervention on the time to menopause was found, we did observe an interaction between the effect of diet study group and baseline BMI. The LFHC dietary intervention was associated with earlier menopause in women with a low BMI; alternatively, higher BMI was associated with later menopause in women in the LFHC group only. The mechanism for such an interaction is unknown, but it could be related to the effects of BMI and diet on sex hormone concentrations. Lower BMI (45) and reduction in dietary fat intake (46) are both associated with lower blood estrogen concentrations in postmenopausal women, but these associations are less consistent in premenopausal women (22, 46). We did not have an a priori hypothesis regarding this interaction, and it should therefore be interpreted cautiously. This potentially important interaction will be reexamined in the future when more women in this dietary intervention trial have experienced menopause.

The women in this study had extensive mammographic density, which placed them at significantly greater risk of breast cancer than were women without mammographic density (19, 47). Information about the effect of diet on breast cancer risk factors in this population is of particular importance for breast cancer prevention because extensive mammographic density is present in a substantial proportion (ie, 20–25%) of the women undergoing mammography. However, the results of the current study with respect to the effect of the LFHC dietary intervention may not be generalizable to other populations, because women with extensive mammographic density are leaner and have fewer children or are less likely to have had children than are women with little mammographic density (48). The women in the current study also reported a relatively high frequency of first-degree family history of breast cancer (19%), which likely reflects a higher rate of participation in breast screening and breast cancer research by women with a family history of breast cancer, but which also may be related to higher mammographic density in women with a family history of breast cancer (49). The selective nature of this population is also reflected by the estimated median age at natural menopause of 54 y, which is considerably higher than the median age reported in other prospective studies, in which estimates ranged from 50.8 to 51.5 y (24-27, 50). However, the 2 factors most consistently reported to be related to age at menopause in other prospective studies—parity and smoking—were also associated with age at menopause in this population.

Parity is related to a later age at menopause in a fairly consistent manner across prospective studies (16, 24-26, 51, 52). In the current study, a trend for an association of later age at menopause with greater numbers of births was observed, and such an association has been reported by others (26, 53). Pregnancy may increase age at menopause by reducing the lifetime number of ovulatory cycles and thus reducing the loss of follicles, or it may work through long-term effects on endogenous hormones or on the hypothalamic-pituitary-gonadal axis (32).

The smokers in the current study experienced natural menopause an average of 1 y earlier than did never and past smokers, an effect that has been seen across all study designs and populations and that is independent of body weight (27, 32, 44). The polycyclic aromatic hydrocarbons in cigarette smoke may be toxic to ovarian follicles (54), and that toxicity may result in earlier depletion of the oocyte pool. Smoking also may decrease follicle-stimulating hormone concentrations and estrogen production (32, 55), which could impair endometrial growth and thereby affect age at menopause.

Educational level was positively associated with age at menopause in some prospective studies (26, 50), but the association was not always independent of smoking (31, 56). In the current study, educational level was negatively associated with the number of births, age at first birth, and smoking at baseline (data not shown), but the increase in age at menopause observed with increasing level of education was independent of the effects of these covariates. The mechanism for an independent effect of education on age at menopause is unknown. Low educational level may be associated with lower socioeconomic status, which has also been associated with earlier age at menopause (25, 57). Higher childhood cognitive function (58) and higher mental ability in adolescence (56) are associated with later menopause, and they may also influence the level of education.

The later age at menopause observed in the current study is not likely to be due to differences in the distribution of demographic factors related to age at menopause in this population. The estimated median age at menopause was 2–3 y later across all categories of smoking, parity, and education than was seen in other prospective studies (24-26, 31, 50, 52). Because leaner women tend to have an earlier age at menopause than do heavier women (24, 25, 27, 51), the lower body weight of women with extensive mammographic density (see above) would not explain the later age at menopause. This later age at menopause may be related to the higher breast cancer risk associated with mammographic density, and further study of this relation may provide important clues to the causes and prevention of breast cancer.


   ACKNOWLEDGMENTS
 
We acknowledge the work of the highly skilled staff of the Diet and Breast Cancer Prevention Study in carrying out the intervention study and collecting the data. We are also indebted to the dedicated study participants.

LJM, CVG, VK, DJ, SM, and NFB were responsible for the study design; LJM, CVG, and VK were responsible for data collection; LJM performed the data analysis; LJM, SM, DJ, and NFB were responsible for data interpretation; and LJM wrote the manuscript. None of the authors had any personal or financial conflict of interest.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Prentice RL, Sheppard L. Dietary fat and cancer: consistency of the epidemiologic data, and disease prevention that may follow from a practical reduction in fat consumption. Cancer Causes Control 1990; 1: 81–97.[Medline]
  2. Boyd NF, Stone J, Vogt KN, Connelly BS, Martin LJ, Minkin S. Dietary fat and breast cancer risk revisited: a meta-analysis of the published literature. Br J Cancer 2003; 89: 1672–85.[Medline]
  3. Prentice RL, Cann B, Chlebowski RT, et al. Low-fat dietary pattern and risk of invasive breast cancer: the Women's Health Initiative Randomized Controlled Dietary Modification Trial. JAMA 2006; 295: 629–42.[Abstract/Free Full Text]
  4. Pike MC, Pearce CL, Wu AH. Prevention of cancers of the breast, endometrium and ovary. Oncogene 2004; 23: 6379–91.[Medline]
  5. Hoel DG, Wakabayshi T, Pike MC. Secular trends in the distribution of the breast cancer risk factors—menarche, first birth, menopause and weight—in Hiroshima and Nagasaki, Japan. Am J Epidemiol 1983; 118: 78–89.[Abstract/Free Full Text]
  6. Wu AH, Ziegler RG, Pike MC, et al. Menstrual and reproductive factors and risk of breast cancer in Asian-Americans. Br J Cancer 1996; 73: 680–6.[Medline]
  7. Wood JW. Dynamics of human reproduction. New York: Aldine De Gruyter, 1994: 401–39.
  8. Meyer F, Moisan J, Marcoux D, Bouchard C. Dietary and physical determinants of menarche. Epidemiology 1990; 1: 377–81.[Medline]
  9. Merzenich H, Boeing H, Wahrendorf J. Dietary fat and sports activity as determinants for age at menarche. Am J Epidemiol 1993; 138: 217–24.[Abstract/Free Full Text]
  10. Koo MM, Rohan TE, Jain M, McLaughlin JR, Corey PN. A cohort study of dietary fibre intake and menarche. Public Health Nutr 2002; 5: 353–60.[Medline]
  11. Rödstroöm K, Bengtsson C, Milsom I, Lissner L, Sundh V, Bjoürkelund C. Evidence for a secular trend in menopausal age: a population study of women in Gothenburg. Menopause 2003; 10: 538–43.[Medline]
  12. Thomas F, Renaud F, Benefice E, de Meeus T, Guegan J-F. International variability of ages at menarche and menopause: patterns and main determinants. Hum Biol 2001; 73: 271–90.[Medline]
  13. Nagata C, Takatsuka N, Inaba S, Kawakami N, Shimizu H. Association of diet and other lifestyle with onset of menopause in Japanese women. Maturitas 1998; 29: 105–13.[Medline]
  14. Torgerson DJ, Avenell A, Russell IT, Reid DM. Factors associated with onset of menopause in women aged 45–59. Maturitas 1994; 19: 83–92.[Medline]
  15. Nagata C, Takatsuka N, Kawakami N, Shimizu H. Association of diet with the onset of menopause in Japanese women. Am J Epidemiol 2000; 152: 863–7.[Abstract/Free Full Text]
  16. Torgerson DJ, Thomas RE, Campbell MK, Reid DM. Alcohol consumption and age of maternal menopause are associated with menopause onset. Maturitas 1997; 26: 21–5.[Medline]
  17. Nagel G, Altenburg H-P, Nieters A, Boffeta P, Linseisen J. Reproductive and dietary determinants of the age at menopause in EPIC-Heidelberg. Maturitas 2005; 52: 337–47.[Medline]
  18. Boyd NF, Fishell E, Jong R et al. Mammographic densities as a criterion for entry to a clinical trial of breast cancer prevention. Br J Cancer 1995; 72: 476–9.[Medline]
  19. Boyd NF, Byng JW, Jong RA, et al. Quantitative classification of mammographic densities and breast cancer risk: results from the Canadian National Breast Screening Study. J Natl Cancer Inst 1995; 87: 670–5.[Abstract/Free Full Text]
  20. Boyd NF, Lockwood GA, Martin LJ, Byng JW, Yaffe MJ, Tritchler DL. Mammographic density as a marker of susceptibility to breast cancer: a hypothesis. In: Miller AB, Bartsch H, Boffetta P, Dragsted L, Vainio H, eds. Biomarkers in cancer chemoprevention. Lyon, France: International Agency for Research on Cancer, 2001: 163–9.
  21. Boyd NF, Greenberg C, Lockwood G, et al. Effects at two years of a low-fat, high-carbohydrate diet on radiologic features of the breast: results from a randomized trial. Canadian Diet and Breast Cancer Prevention Study Group. J Natl Cancer Inst 1997; 89: 488–96.
  22. Boyd NF, Greenberg C, Martin L, Stone J, Hammond G, Minkin S. Lack of effect of a low-fat-high-carbohydrate diet on ovarian hormones in premenopausal women: results from a randomized trial. IARC Sci Publ 2002; 156: 445–50.[Medline]
  23. Wallace RB, Sherman BM, Bean JA, Treloar AE, Schlabaugh L. Probability of menopause with increasing duration of amenorrhea in middle-aged women. Am J Obstet Gynecol 1979; 135: 1021–4.[Medline]
  24. Kato I, Toniolo P, Akhmedkhanov A, Koenig KL, Shore R, Zeleniuch-Jacquotte A. Prospective study of factors influencing the onset of natural menopause. J Clin Epidemiol 1998; 51: 1271–6.[Medline]
  25. de Vries E, den Tonkelaar I, van Noord PAH, van der Schouw YT, te Velde ER, Peeters PHM. Oral contraceptive use in relation to age at menopause in the DOM cohort. Hum Reprod 2001; 16: 1657–62.[Abstract/Free Full Text]
  26. Do KA, Treloar SA, Pandeya N, et al. Predictive factors of age at menopause in a large Australian twin study. Hum Biol 1998; 70: 1073–91.[Medline]
  27. Palmer JR, Rosenberg L, Wise LA, Horton NJ, Adams-Campbell LL. Onset of natural menopause in African American women. Am J Public Health 2003; 93: 299–306.[Abstract/Free Full Text]
  28. Hardy R, Kuh D. Does early growth influence timing of the menopause? Evidence from a British birth cohort. Hum Reprod 2002; 17: 2474–9.[Abstract/Free Full Text]
  29. Allison PD. Heterogeneity, repeated events, and other topics. In: Survival analysis using the SAS system. Cary, NC: SAS Institute Inc, 1995: 233–52.
  30. Allison PD. Basic concepts of survival analysis. In: Survival analysis using the SAS system. Cary, NC: SAS Institute Inc, 1995: 9–27.
  31. Brambilla DJ, McKinlay SM. A prospective study of factors affecting age at menopause. J Clin Epidemiol 1989; 42: 1031–9.[Medline]
  32. Harlow BL, Signorello LB. Factors associated with early menopause. Maturitas 2000; 35: 3–9.[Medline]
  33. Korn EL, Graubard BI, Midthune D. Time-to-event analysis of longitudinal follow-up of a survey: choice of the time-scale. Am J Epidemiol 1997; 145: 72–80.[Abstract/Free Full Text]
  34. Bernstein L. Epidemiology of endocrine-related risk factors for breast cancer. J Mammary Gland Biol Neoplasia 2002; 7: 3–15.[Medline]
  35. Allison PD. Estimating Cox regression models with PROC PHREG. In: Survival analysis using the SAS system. Cary, NC: SAS Institute Inc, 1995: 111–84.
  36. Prentice RL. Dietary assessment and the reliability of nutritional epidemiology reports. Lancet 2003; 362: 182–3.[Medline]
  37. Willett W, Stampfer M, Chu NF, Spiegelman D, Holmes M, Rimm E. Assessment of questionnaire validity for measuring total fat intake using plasma lipid levels as criteria. Am J Epidemiol 2001; 154: 1107–12.[Abstract/Free Full Text]
  38. Rock CL, Flatt SW, Thomson CA, et al. Plasma triacylglycerol and HDL cholesterol concentrations confirm self-reported changes in carbohydrate and fat intakes in women in a diet intervention trial. J Nutr 2004; 134: 342–7.[Abstract/Free Full Text]
  39. Martin LJ. Methods for the Diet and Breast Cancer Prevention Study. Menopause and breast cancer risk: the influence of dietary fat reduction and breast cancer risk factors on timing of the menopause and change in mammographic density. Doctoral dissertation. University of Toronto, Toronto, Canada, 2005: 41–59.
  40. Elias SG, Van Noord PA, Peeters PH, den Tonkelaar I, Grobbee DF. Caloric restriction reduces age at menopause: the effect of the 1944–1945 Dutch famine. Menopause 2003; 10: 399–405.[Medline]
  41. Morabia A, Costanza MC. World Health Organization Collaborative Study of Neoplasia and Steroid Contraceptives. International variability in ages at menarche, first livebirth, and menopause. Am J Epidemiol 1998; 148: 1195–205.[Abstract/Free Full Text]
  42. Boulet MJ, Oddens BJ, Lehert P, Vemer HM, Visser A. Climacteric and menopause in seven South-east Asian countries. Maturitas 1994; 19: 157–76.[Medline]
  43. Richardson SJ. The biological basis of menopause. Ballieres Clin Endocrinol Metab 1993; 7: 1–16.[Medline]
  44. Gold EB, Bromberger J, Crawford S, et al. Factors associated with age at natural menopause in a multiethnic sample of midlife women. Am J Epidemiol 2001; 153: 865–74.[Abstract/Free Full Text]
  45. Endogenous Hormones and Breast Cancer Collaborative Group. Body mass index, serum sex hormones, and breast cancer risk in postmenopausal women. J Natl Cancer Inst 2003; 95: 1218–26.[Abstract/Free Full Text]
  46. Wu AH, Pike MC, Stram DO. Meta-analysis: dietary fat intake, serum estrogen levels, and the risk of breast cancer. J Natl Cancer Inst 1999; 91: 529–34.[Abstract/Free Full Text]
  47. Byrne C, Schairer C, Wolfe J, et al. Mammographic features and breast cancer risk: effects with time, age, and menopause status. J Natl Cancer Inst 1995; 87: 1622–9.[Abstract/Free Full Text]
  48. Boyd NF, Rommens JM, Vogt K, et al. Mammographic breast density as an intermediate phenotype for breast cancer. Lancet 2005; 6: 798–808.[Medline]
  49. Ziv E, Shepherd J, Smith-Bindman R, Kerlikowske K. Mammographic breast density and family history of breast cancer. J Natl Cancer Inst 2005; 95: 556–8.
  50. Bromberger JT, Matthews KA, Kuller LH, Wing RR, Meilahn EN, Plantinga P. Prospective study of the determinants of age at menopause. Am J Epidemiol 1997; 145: 124–33.[Abstract/Free Full Text]
  51. Willett W, Stampfer MJ, Bain C, et al. Cigarette smoking, relative weight, and menopause. Am J Epidemiol 1983; 117: 651–8.[Abstract/Free Full Text]
  52. Whelan EA, Sandler DP, McConnaughey DR, Weinberg CR. Menstrual and reproductive characteristics and age at natural menopause. Am J Epidemiol 1990; 131: 625–32.[Abstract/Free Full Text]
  53. Weinstein M, Gorrindo T, Riley A, et al. Timing of menopause and patterns of menstrual bleeding. Am J Epidemiol 2003; 158: 782–91.[Abstract/Free Full Text]
  54. Mattison DR, Thorgierssen SS. Smoking and industrial pollution and their effects on menopause and ovarian cancer. Lancet 1978; 1: 187–8.[Medline]
  55. Cooper GS, Baird DD, Hulka BS, Weinberg CR, Savitz DA, Hughes CL. Follicle-stimulating hormone concentrations in relation to active and passive smoking. Obstet Gynecol 1995; 85: 407–11.[Abstract]
  56. Shinberg DS. An event history analysis of age at last menstrual period: correlates of natural and surgical menopause among midlife Wisconsin women. Soc Sci Med 1998; 46: 1381–96.[Medline]
  57. Lawlor DA, Ebrahim S, Smith GD. The association of socio-economic position across the life course and age at menopause: the British Women's Heart and Health Study. BJOG 2003; 110: 1078–87.[Medline]
  58. Richards M, Kuh D, Hardy R, Wadsworth M. Lifetime cognitive function and timing of the natural menopause. Neurology 1999; 53: 308–14.[Abstract/Free Full Text]
Received for publication January 20, 2006. Accepted for publication June 8, 2006.





This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Martin, L. J
Right arrow Articles by Boyd, N. F
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Martin, L. J
Right arrow Articles by Boyd, N. F
Agricola
Right arrow Articles by Martin, L. J
Right arrow Articles by Boyd, N. F

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS