Carbohydrate intake, glycemic index, and glycemic load have been hypothesized to increase risk of breast cancer by raising insulin levels, but these associations have not been studied extensively. The insulin response to dietary carbohydrate is substantially greater among overweight women than among leaner women. Although fiber intake has been hypothesized to reduce the risk of breast cancer, data from early adult life are lacking. We examined dietary carbohydrate, glycemic index, glycemic load, and fiber in relation to breast cancer risk among 90,655 premenopausal women in the Nurses’ Health Study II aged 26–46 years in 1991. Diet was assessed with a food frequency questionnaire in 1991 and 1995. During 8 years of follow-up, we documented 714 incident cases of invasive breast cancer. Dietary carbohydrate intake, glycemic load, and glycemic index were not related to breast cancer risk in the overall cohort. However, the associations differed by body mass index (BMI): among women with BMI < 25 kg/m2, the multivariate relative risks for the increasing quintiles of carbohydrate intake were 1.00 (referent), 0.87, 0.77, 0.66, and 0.62 [95% confidence interval, 0.40–0.97; P, test for trend = 0.02]; and among women with BMI ≥ 25 kg/m2, the corresponding relative risks were 1.00 (referent), 1.30, 1.35, 1.50, and 1.47 (95% confidence interval, 0.84–2.59; P, test for trend = 0.14; P, test for interaction = 0.02). Similar interaction with BMI was observed for glycemic load, but not for glycemic index. Intakes of total fiber and different types of fiber were not appreciably related to breast cancer risk. Our findings suggest that the associations between carbohydrate intake or glycemic load and breast cancer risk among young adult women differ by body weight. Our data do not support a strong association between fiber intake and breast cancer risk.

Dietary glycemic index and glycemic load have been related to elevated risks of adult onset diabetes (1, 2), coronary heart disease (3), colorectal cancer (4), and pancreatic cancer (5), suggesting an important physiological role. Elevation in blood insulin levels has been hypothesized to increase breast cancer risk by depressing levels of IGF7 -binding protein 1, resulting in higher levels of free IGF-I, or by direct mitogenic effects (6). Intake of carbohydrate raises insulin levels, and the magnitude of response is substantially greater in the presence of insulin resistance (7). Overweight and obesity are major determinants of insulin resistance, and we have found that the adverse metabolic response to dietary carbohydrate or glycemic load, manifested by elevated triglycerides and low high-density lipoprotein cholesterol levels, is magnified among overweight men and women (8). In a recent prospective study, serum levels of fasting glucose and IGF-1 were associated with an elevated risk of breast cancer (9), and in one case-control study, both glycemic index and glycemic load were associated with an increased risk of breast cancer (10). On the basis of the metabolic studies, we hypothesized that the positive associations between dietary carbohydrate and glycemic load and incidence of breast cancer would be seen primarily among overweight women.

Dietary fiber intake may protect against breast cancer by inhibiting deconjugation and reabsorption of estrogen from the colon (11). High fiber intake decreased mammary tumor incidence in animal studies (12). Case-control studies have reported an inverse association between dietary fiber intake and breast cancer risk (13, 14, 15, 16, 17, 18), but data from prospective studies have not supported this association (19, 20, 21, 22, 23, 24).

Previous prospective studies on dietary risk factors and breast cancer risk have included primarily postmenopausal women whose diet were measured after the age of 40 years. The magnitude and directions of the association between some risk factors for breast cancer may vary according to age or menopausal status (25). Accordingly, the associations between dietary exposures and breast cancer may be different in younger women. Therefore, we evaluated the associations of dietary carbohydrate, glycemic index, glycemic load, and fiber and the risk of breast cancer in premenopausal women in the NHS II.

Study Population.

The NHS II is a prospective cohort study of 116,671 female registered nurses who were 25–42 years of age and living in 1 of 14 states in the United States when they responded in 1989 to a questionnaire about their medical histories and lifestyles. Follow-up questionnaires have been sent biennially to update information on risk factors and medical events.

For the current analysis, we started follow-up in 1991, when diet was first measured. For the 97,807 women who returned the 1991 dietary questionnaire, we excluded those who had an implausible total energy intake (<800 or >4,200 kcal/day) or who left more than 70 food items blank in the 1991 FFQ (n = 2,361). We also excluded women who reported a diagnosis of cancer, except non-melanoma skin cancer, before returning the 1991 questionnaire (n = 1,325). Because the proportion of postmenopausal women at baseline was small (n = 3,466), we excluded them from this analysis, leaving a total of 90,655 women. Among those who answered the FFQ in 1991, the rate of response to the 8-year follow-up questionnaire was 93%.

The study was approved by the human research committees at the Harvard School of Public Health and the Brigham and Women’s Hospital.

Dietary Assessment.

Semiquantitative FFQs with 133 and 142 food items were sent to women in 1991 and 1995, respectively, to assess usual dietary intake during the past year. Participants were asked how often, on average, they had consumed each type of food or beverage during the past year. The FFQ had nine possible responses, ranging from never or less than once per month to six or more times per day. Nutrient intake per individual was calculated as the sum of the contributions from all foods based primarily on United States Department of Agriculture food composition data (26).

The methods used to assess glycemic indices of individual foods and mixed meals and to calculate dietary glycemic load have been presented elsewhere (1, 27, 28). In brief, dietary glycemic load for a food was calculated by multiplying the carbohydrate content of the food by its glycemic index and its frequency of consumption (servings of the food/day). Dietary glycemic load for a participant was calculated by summing the values of dietary glycemic load for all food items consumed. Each unit of dietary glycemic load represents the equivalent of 1 g of carbohydrate from white bread. Dietary glycemic load thus represents the quality and quantity of dietary carbohydrates and the interaction between the two, given that the product of glycemic index and carbohydrate intake indicates that a higher glycemic index has a greater impact at a higher carbohydrate intake. The overall dietary glycemic index for a participant was calculated by dividing glycemic load by the total amount of carbohydrate intake. The dietary glycemic index represents the overall quality of carbohydrate intake for each participant.

We used the regression-residual method to adjust fiber intake, glycemic index, and glycemic load for total energy intake and used the nutrient-density method to adjust carbohydrate intake for total energy intake (29). We calculated cumulative averaged dietary intakes using the 1991 and 1995 dietary data (30). Specifically, the 1991 value was used for the 1991–1995 follow-up period, and the average of the 1991 and 1995 value was used for 1995–1999 follow-up period. We used cumulative averaged dietary intake in the primary analyses and also examined baseline intake separately.

The reproducibility and validity of carbohydrate and fiber intakes and of individual food items assessed with a similar FFQ have been evaluated in the NHS. The correlation coefficients for energy-adjusted carbohydrate and fiber intakes from the average of four 1-week diet records collected 3–4 years earlier and from the FFQ were 0.61 and 0.56, respectively, after correction for attenuation due to random error in the diet records (31). For individual food items with high glycemic index values, the deattenuated correlation coefficients were 0.71 for white bread, 0.77 for dark bread, 0.66 for potatoes, 0.84 for orange or grapefruit juice, and 0.56 for fruit-flavored punch or noncarbonated beverage (32). Dietary glycemic load was positively related to fasting triacylglycerol concentrations (P < 0.001) and inversely associated with high-density lipoprotein cholesterol levels (P = 0.03; Ref. 8).

Identification of Cases.

Biennial questionnaires mailed between 1993 and 1999 were used to identify newly diagnosed cases of breast cancer. Deaths were documented by responses to follow-up questionnaires by family members or the postal service and by a search of the National Death Index. When a case of breast cancer was reported, we asked the participant (or next of kin for those who had died) for confirmation of the diagnosis and for permission to seek relevant hospital records and pathology reports. Pathology reports confirmed 98% of the self-reports. Because the degree of self-reporting accuracy was high, we included the few self-reported cases for which we could not obtain records.

Statistical Analysis.

Participants contributed person-time from the date of return of the 1991 questionnaire until the date of breast cancer diagnosis, death, or June 1999, whichever came first. Participants were divided into quintiles according to their nutrient intakes, glycemic index, or glycemic load. RRs of breast cancer were calculated as the incidence rate for a given quintile as compared with the rate for the lowest quintile. We used Cox proportional hazards regression to account for potential effects of other risk factors for breast cancer (33). To control as finely as possible for confounding by age, calendar time, and any possible two-way interactions between these two time scales, we stratified the analysis jointly by age in months at start of follow-up and calendar year of the current questionnaire cycle. Multivariate models also were adjusted simultaneously for smoking status, BMI, height, age at menarche, oral contraceptive use, family history of breast cancer, history of benign breast disease, parity and age at first birth, menopausal status, and intakes of calories, animal fat, and alcohol. All covariates except height, age at menarche, and family history of breast cancer were updated in each questionnaire cycle. SAS PROC PHREG (34) was used for all analysis, and the Anderson-Gill data structure (35) was used to handle time-varying covariates efficiently, with a new data record created for every questionnaire cycle at which a participant was at risk and covariates set to their values at the time the questionnaire was returned. For all RRs, 95% CIs were calculated. Tests for trend were conducted using the median value for each quintile as a continuous variable. To examine whether the association between carbohydrate intake, glycemic index, and glycemic load and breast cancer risk was modified by BMI, we included a cross-product term of carbohydrate intake (or glycemic load and glycemic index) and BMI, both expressed as continuous variables, in a multivariate model. P for the tests for interaction was obtained from a likelihood ratio test with 1 degree of freedom. All Ps were two-sided.

We documented 714 cases of incident invasive breast carcinoma during 8 years of follow-up (711,651 person-years) of 90,655 women. The age range of the participants was 26–46 years (mean age, 36 years; SD, 5 years) at baseline. The age range of cases at the time of diagnosis of breast cancer was 26–52 years (mean age, 43 years; SD, 5 years). Table 1 presents the distribution of risk factors for breast cancer by quintiles of carbohydrate and total fiber intakes. Women with a higher intake of dietary carbohydrate or fiber were less likely to smoke, to use oral contraceptives, or to consume alcohol or animal fat. These women were also more likely to have a history of benign breast disease. Women with higher intake of carbohydrate were less likely to have had menarche before age 12 years. In contrast, women with higher intake of dietary fiber were more likely to have had menarche before age 12 years. Because glycemic load was highly correlated with total carbohydrate intake in this cohort (r = 0.92), the distribution of risk factors for breast cancer by glycemic load was similar to that of carbohydrate intake (data not shown).

Carbohydrate intake, glycemic index, and glycemic load were not related to overall breast cancer risk (Table 2). Because we found a positive association between animal fat intake and breast cancer risk (36), we adjusted for animal fat intake in multivariate models. Thus, these RRs for carbohydrate intake imply substituting calories from carbohydrate for the same percentage of calories from protein and vegetable fat. Because adiposity is an important determinant of insulin resistance, we hypothesized that it could modify the relationship between these factors and breast cancer risk. When we examined the associations by BMI (<25 and ≥25 kg/m2), carbohydrate intake was inversely related to breast cancer risk among leaner women but positively related to increased risk among heavier women (P, test for interaction = 0.02; Table 3). Similar associations were observed for glycemic load but not for glycemic index. These results were similar when we examined baseline intake.

Some of the participants who were premenopausal at baseline became postmenopausal during the follow-up period. Thus, about 10% of the cases were postmenopausal at the time of diagnosis. Restricting analyses to those who remained premenopausal throughout follow-up (n = 641 cases of breast cancer) resulted in similar associations. For example, the RRs for the highest quintiles of glycemic load compared with the lowest quintiles were 0.84 (95% CI, 0.55–1.27) among women with BMI of <25 kg/m2 and 1.44 (95% CI, 0.85–2.43) among women with BMI of ≥25 kg/m2.

Fiber intake was minimally related to breast cancer risk (Table 4). The multivariate RR for the highest quintile of intake compared with the lowest was 0.88 (95% CI, 0.67–1.14; P, test for trend = 0.60). We also examined types of fiber intake based on food sources (fiber from cereals, fruits, vegetables, legumes, and cruciferous vegetables) and solubility (soluble and insoluble fiber). None of the subtypes of fiber had a statistically significant relation to a reduced risk of breast cancer. However, there was a weak, nonsignificant inverse association between intake of fiber from legumes and risk of breast cancer; the RR for the fifth quintile of intake was 0.79 (95% CI, 0.62–1.02; P, test for trend = 0.04) compared with the first quintile. The associations were similar in analyses using baseline intake or restricted to women who remained premenopausal during follow-up.

In this prospective study of relatively young women, premenopausal dietary carbohydrate, glycemic index, glycemic load, and fiber were not strongly related to overall breast cancer risk. However, the associations with carbohydrate intake and glycemic load differed by body weight.

Diets with high carbohydrate intake or glycemic load may increase breast cancer risk because raised insulin levels down-regulate IGF-binding protein 1 and thus increase free IGF-I, a strong predictor of breast cancer risk among premenopausal women (6). Fasting blood glucose levels, C-peptide (a marker of insulin resistance) levels, and adult-onset diabetes have been related to elevated risk of breast cancer (9, 37, 38). Positive associations between glycemic load (or glycemic index) and several chronic diseases, including adult-onset diabetes (1, 2), coronary heart disease (3), colorectal cancer (4), and pancreatic cancer (5), have been reported. However, we are aware of only one epidemiological study examining the association between glycemic index or glycemic load and breast cancer risk, a case-control study which found that both glycemic index (RR = 1.4 for the highest versus lowest quintiles) and glycemic load (RR = 1.3 for the highest versus lowest quintiles) were related to an elevated risk of breast cancer (10). Although we did not find any strong overall associations between dietary carbohydrate, glycemic index, and glycemic load and overall breast cancer risk, we hypothesized a priori that the associations may be stronger among heavier women because obesity is an important determinant of insulin resistance that exaggerates adverse metabolic responses related to carbohydrate intake (8). The modest (nonsignificant) positive associations between carbohydrate intake and glycemic load and breast cancer risk among overweight women were consistent with our hypothesis. A metabolic study found that the higher the magnitude of insulin resistance, the higher the plasma insulin levels in response to carbohydrate intake (7); this may be the underlying mechanism mediating carbohydrate intake to breast cancer risk among women with higher BMI.

However, the apparent inverse associations between carbohydrate intake and glycemic load and breast cancer risk among leaner women were not expected and need to be examined further in other studies.

Dietary fiber may promote excretion of estrogen by inhibiting deconjugation and reabsorption of estrogen from the colon (11). In animal studies, high fiber intake reduced mammary tumor development (12), and in humans, a high-fiber diet reduced serum estrogen levels (39). A combined analysis of 10 case-control studies, which included studies conducted before 1986, had reported inverse association between fiber intake and breast cancer risk (13). Some of the later case-control studies (14, 15, 16, 17, 18), but not others (40, 41), conducted after the combined analysis have supported the beneficial role of fiber intake. However, case-control studies may be prone to recall bias. Previous prospective studies have not found a clear association between fiber intake and breast cancer risk (19, 20, 21, 22, 23, 24, 42). Our results among relatively young and mostly premenopausal women are consistent with the results of previous prospective studies.

Among the previous prospective studies, only one study (24) examined types of fiber but did not examine fiber from legumes. Although not statistically significant, we found that fiber intake from legumes was inversely related to breast cancer risk. Only one case-control study examined fiber from legumes and found a suggestion of an inverse association among relatively young women age 20–40 years (41).

This study had several strengths. First, the prospective nature of the study avoided possible recall bias and possible biased control selection in case-control studies, and few participants have been lost to follow-up. Second, we had repeated measures of dietary intake and were able to examine long-term averaged diet as well as baseline intake. Third, we had a wide range of information on potential confounders and adjusted for them in the analysis. As a limitation, the duration of follow-up time and the number of cases were limited, especially in analyses restricted to overweight women.

In conclusion, in this population of mostly premenopausal women, fiber intake was not strongly related to a reduced risk of breast cancer. We found that the associations between carbohydrate intake and dietary glycemic load and breast cancer differed by body weight, with a positive association seen among heavier women. Additional studies are needed to confirm these results and to identify relevant biological mechanisms.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Grant support: Supported by NIH Grant CA50385 and the Breast Cancer Research Foundation.

Requests for reprints: Eunyoung Cho, Channing Laboratory, 181 Longwood Avenue, Boston, Massachusetts 02115. Phone: (617) 525-2091; Fax: (617) 525-2008; E-mail: [email protected]

7

The abbreviations used are: IGF, insulin-like growth factor; BMI, body mass index, CI, confidence interval; NHS, Nurses’ Health Study; FFQ, food frequency questionnaire; RR, relative risk.

Table 1

Age-standardized distributions of potential risk factors for breast cancer according to energy-adjusted carbohydrate and total fiber intakes in 1991 among women 26–46 years of agea

Carbohydrate intake quintileTotal fiber intake quintile
135135
Median (% of energy for carbohydrate and g/day for fiber)b 41.2 50.1 59.4 12.5 17.7 24.8 
Percentage of group       
 Current smokers 18 11 10 19 11 
 Current oral contraceptive use 12 11 11 12 11 10 
 History of benign breast disease 33 34 34 32 34 35 
 Family history of breast cancer in mother and sisters 
 Parity ≥ 3 19 22 18 20 22 18 
 Age at menarche < 12 yrs 26 24 23 23 24 26 
Mean       
 Age (yrs) 36 36 36 35 36 37 
 Alcohol (g/day) 5.3 2.9 1.7 3.8 3.1 2.7 
 BMI (kg/m226 25 23 25 25 24 
 Age at first birth (yrs) 25 26 26 26 26 26 
 Animal fat (% of energy/day) 22 17 13 20 18 14 
 Energy-adjusted fiber (g/day) 15 18 21 12 18 26 
 Carbohydrate (% of energy/day) 40 50 60 47 49 55 
 Energy-adjusted glycemic index 75 77 79 78 77 75 
 Energy-adjusted glycemic load 135 171 214 166 170 187 
Carbohydrate intake quintileTotal fiber intake quintile
135135
Median (% of energy for carbohydrate and g/day for fiber)b 41.2 50.1 59.4 12.5 17.7 24.8 
Percentage of group       
 Current smokers 18 11 10 19 11 
 Current oral contraceptive use 12 11 11 12 11 10 
 History of benign breast disease 33 34 34 32 34 35 
 Family history of breast cancer in mother and sisters 
 Parity ≥ 3 19 22 18 20 22 18 
 Age at menarche < 12 yrs 26 24 23 23 24 26 
Mean       
 Age (yrs) 36 36 36 35 36 37 
 Alcohol (g/day) 5.3 2.9 1.7 3.8 3.1 2.7 
 BMI (kg/m226 25 23 25 25 24 
 Age at first birth (yrs) 25 26 26 26 26 26 
 Animal fat (% of energy/day) 22 17 13 20 18 14 
 Energy-adjusted fiber (g/day) 15 18 21 12 18 26 
 Carbohydrate (% of energy/day) 40 50 60 47 49 55 
 Energy-adjusted glycemic index 75 77 79 78 77 75 
 Energy-adjusted glycemic load 135 171 214 166 170 187 
a

Except for the data on mean age, all data shown are standardized to the age distributions of the cohort in 1991.

b

The range of fiber intake is reduced by energy adjustment because this describes the variation among women with the same energy intake.

Table 2

RR and 95% CIs of breast cancer according to quintiles of cumulative averaged dietary carbohydrate, glycemic index, and glycemic load in women 26–46 years of age at baseline

Risk factorQuintile of intakeP, test for trend
12345
Carbohydrate       
 Median (% of energy) 41.2 46.5 50.1 53.6 59.4  
 Age-adjusted RR (95% CI) 1.00 1.11 (0.89–1.40) 1.06 (0.84–1.33) 0.98 (0.78–1.24) 0.86 (0.67–1.09) 0.13 
 Multivariatea RR (95% CI) 1.00 1.05 (0.82–1.33) 0.98 (0.75–1.29) 0.94 (0.70–1.27) 0.89 (0.63–1.26) 0.42 
Glycemic index       
 Median 70 75 77 79 82  
 Age-adjusted RR (95% CI) 1.00 0.96 (0.77–1.20) 0.91 (0.72–1.14) 0.88 (0.70–1.11) 0.96 (0.76–1.21) 0.48 
 Multivariatea RR (95% CI) 1.00 0.96 (0.77–1.20) 0.93 (0.74–1.16) 0.91 (0.72–1.15) 1.05 (0.83–1.33) 0.97 
Glycemic load       
 Median 138 158 172 187 211  
 Age-adjusted RR (95% CI) 1.00 1.02 (0.82–1.27) 1.02 (0.81–1.28) 0.79 (0.62–1.00) 0.93 (0.74–1.18) 0.19 
 Multivariatea RR (95% CI) 1.00 0.98 (0.77–1.24) 0.99 (0.77–1.28) 0.81 (0.61–1.07) 1.06 (0.78–1.45) 0.96 
Risk factorQuintile of intakeP, test for trend
12345
Carbohydrate       
 Median (% of energy) 41.2 46.5 50.1 53.6 59.4  
 Age-adjusted RR (95% CI) 1.00 1.11 (0.89–1.40) 1.06 (0.84–1.33) 0.98 (0.78–1.24) 0.86 (0.67–1.09) 0.13 
 Multivariatea RR (95% CI) 1.00 1.05 (0.82–1.33) 0.98 (0.75–1.29) 0.94 (0.70–1.27) 0.89 (0.63–1.26) 0.42 
Glycemic index       
 Median 70 75 77 79 82  
 Age-adjusted RR (95% CI) 1.00 0.96 (0.77–1.20) 0.91 (0.72–1.14) 0.88 (0.70–1.11) 0.96 (0.76–1.21) 0.48 
 Multivariatea RR (95% CI) 1.00 0.96 (0.77–1.20) 0.93 (0.74–1.16) 0.91 (0.72–1.15) 1.05 (0.83–1.33) 0.97 
Glycemic load       
 Median 138 158 172 187 211  
 Age-adjusted RR (95% CI) 1.00 1.02 (0.82–1.27) 1.02 (0.81–1.28) 0.79 (0.62–1.00) 0.93 (0.74–1.18) 0.19 
 Multivariatea RR (95% CI) 1.00 0.98 (0.77–1.24) 0.99 (0.77–1.28) 0.81 (0.61–1.07) 1.06 (0.78–1.45) 0.96 
a

The model was stratified by age in months at start of follow-up and calendar year of the current questionnaire cycle and adjusted simultaneously for smoking (never, past <25, past 25+, current <25, and current 25+ cigarettes/day), height (<62, 62–<65, 65–<68, 68+ inches), parity and age at first birth (nulliparous, parity ≤2 and age at first birth <25 years, parity ≤2 and age at first birth 25–<30 years, parity ≤2 and age at first birth 30+ years, parity 3+ and age at first birth <25 years, parity 3+ and age at first birth 25+ years), BMI (<18.5, 18.5–19.9, 20.0–22.4, 22.5–24.9, 25.0–29.9, 30.0+ kg/m2), age at menarche (<12, 12, 13, ≥14 years), family history of breast cancer (yes, no), history of benign breast disease (yes, no), oral contraceptive use (never, past <4 years, past 4+ years, current <8 years, current 8+ years), menopausal status (premenopausal, postmenopausal, dubious, unsure), alcohol intake (nondrinkers, <5, 5–<10, 10–<20, 20+ g/day), energy intake (quintiles), and animal fat intake (quintiles).

Table 3

Multivariate RR and 95% CIs of breast cancer according to quintiles of cumulative averaged dietary carbohydrate, glycemic index, and glycemic load by BMI in women 26–46 years of age at baselinea

Risk factorQuintile of intakeP, test for trendP, test for interaction
12345
Carbohydrate       0.02 
 BMI < 25 kg/m2 1.00 0.87 (0.63–1.20) 0.77 (0.54–1.10) 0.66 (0.45–0.98) 0.62 (0.40–0.97) 0.02  
 BMI ≥ 25 kg/m2 1.00 1.30 (0.89–1.91) 1.35 (0.88–2.07) 1.50 (0.94–2.41) 1.47 (0.84–2.59) 0.14  
Glycemic index       0.60 
 BMI < 25 kg/m2 1.00 1.00 (0.75–1.34) 0.87 (0.64–1.18) 0.92 (0.68–1.26) 1.04 (0.76–1.41) 0.94  
 BMI ≥ 25 kg/m2 1.00 0.92 (0.65–1.31) 1.01 (0.71–1.44) 0.88 (0.61–1.28) 1.05 (0.72–1.52) 0.98  
Glycemic load       0.02 
 BMI < 25 kg/m2 1.00 0.89 (0.65–1.22) 0.87 (0.62–1.21) 0.58 (0.40–0.86) 0.83 (0.56–1.24) 0.19  
 BMI ≥ 25 kg/m2 1.00 1.11 (0.77–1.60) 1.17 (0.79–1.75) 1.22 (0.79–1.88) 1.46 (0.89–2.39) 0.14  
Risk factorQuintile of intakeP, test for trendP, test for interaction
12345
Carbohydrate       0.02 
 BMI < 25 kg/m2 1.00 0.87 (0.63–1.20) 0.77 (0.54–1.10) 0.66 (0.45–0.98) 0.62 (0.40–0.97) 0.02  
 BMI ≥ 25 kg/m2 1.00 1.30 (0.89–1.91) 1.35 (0.88–2.07) 1.50 (0.94–2.41) 1.47 (0.84–2.59) 0.14  
Glycemic index       0.60 
 BMI < 25 kg/m2 1.00 1.00 (0.75–1.34) 0.87 (0.64–1.18) 0.92 (0.68–1.26) 1.04 (0.76–1.41) 0.94  
 BMI ≥ 25 kg/m2 1.00 0.92 (0.65–1.31) 1.01 (0.71–1.44) 0.88 (0.61–1.28) 1.05 (0.72–1.52) 0.98  
Glycemic load       0.02 
 BMI < 25 kg/m2 1.00 0.89 (0.65–1.22) 0.87 (0.62–1.21) 0.58 (0.40–0.86) 0.83 (0.56–1.24) 0.19  
 BMI ≥ 25 kg/m2 1.00 1.11 (0.77–1.60) 1.17 (0.79–1.75) 1.22 (0.79–1.88) 1.46 (0.89–2.39) 0.14  
a

The models were adjusted for the same covariates as the multivariate model in Table 2. Number of cases (person-years); BMI < 25 kg/m2 = 422 (424,644), BMI ≥ 25 kg/m2 = 291 (285,780).

Table 4

RRs and 95% CIs of breast cancer according to quintiles of cumulative averaged fiber intake in women 26–46 years of age at baseline

Risk factorQuintile of intakeP, test for trend
12345
Total fiber       
 Median intake (g/day) 12.5 15.4 17.7 20.2 24.8  
 Age-adjusted RR (95% CI) 1.00 0.93 (0.73–1.17) 0.86 (0.68–1.10) 1.01 (0.81–1.27) 0.84 (0.67–1.07) 0.30 
 Multivariatea RR (95% CI) 1.00 0.89 (0.70–1.13) 0.83 (0.65–1.06) 1.00 (0.78–1.26) 0.88 (0.67–1.14) 0.60 
Fiber from cereals       
 Median intake (g/day) 3.0 4.2 5.2 6.4 8.8  
 Age-adjusted RR (95% CI) 1.00 1.35 (1.06–1.72) 1.30 (1.02–1.65) 1.31 (1.03–1.67) 0.90 (0.69–1.16) 0.12 
 Multivariatea RR (95% CI) 1.00 1.31 (1.03–1.67) 1.27 (0.99–1.63) 1.28 (1.00–1.65) 0.91 (0.69–1.21) 0.21 
Fiber from fruits       
 Median intake (g/day) 1.1 2.0 2.9 4.1 6.2  
 Age-adjusted RR (95% CI) 1.00 1.20 (0.94–1.52) 1.07 (0.84–1.37) 1.15 (0.90–1.46) 1.07 (0.84–1.37) 0.91 
 Multivariatea RR (95% CI) 1.00 1.18 (0.93–1.51) 1.05 (0.82–1.35) 1.15 (0.89–1.47) 1.13 (0.88–1.46) 0.54 
Fiber from vegetables       
 Median intake (g/day) 3.3 4.8 6.1 7.6 10.4  
 Age-adjusted RR (95% CI) 1.00 1.16 (0.92–1.48) 1.10 (0.87–1.40) 1.09 (0.86–1.39) 0.95 (0.74–1.21) 0.37 
 Multivariatea RR (95% CI) 1.00 1.14 (0.90–1.45) 1.07 (0.84–1.36) 1.07 (0.84–1.36) 0.97 (0.75–1.24) 0.52 
Fiber from cruciferous vegetables       
 Median intake (g/day) 0.2 0.5 0.7 1.1 1.8  
 Age-adjusted RR (95% CI) 1.00 1.14 (0.90–1.44) 1.08 (0.85–1.37) 1.03 (0.81–1.32) 0.86 (0.67–1.10) 0.06 
 Multivariatea RR (95% CI) 1.00 1.13 (0.90–1.43) 1.06 (0.84–1.35) 1.02 (0.80–1.31) 0.87 (0.68–1.12) 0.08 
Fiber from legumes       
 Median intake (g/day) 0.1 0.4 0.7 1.1 2.0  
 Age-adjusted RR (95% CI) 1.00 1.01 (0.80–1.28) 0.97 (0.77–1.22) 0.89 (0.71–1.13) 0.78 (0.61–1.00) 0.02 
 Multivariatea RR (95% CI) 1.00 0.97 (0.76–1.23) 0.93 (0.73–1.18) 0.89 (0.70–1.12) 0.79 (0.62–1.02) 0.04 
Soluble fiber       
 Median intake (g/day) 3.8 4.6 5.3 6.1 7.4  
 Age-adjusted RR (95% CI) 1.00 0.97 (0.77–1.23) 0.98 (0.77–1.24) 1.07 (0.85–1.34) 0.84 (0.66–1.06) 0.24 
 Multivariatea RR (95% CI) 1.00 0.95 (0.75–1.21) 0.94 (0.74–1.21) 1.06 (0.83–1.34) 0.87 (0.67–1.13) 0.50 
Insoluble fiber       
 Median intake (g/day) 9.5 11.7 13.5 15.4 19.0  
 Age-adjusted RR (95% CI) 1.00 1.06 (0.84–1.34) 1.02 (0.80–1.29) 1.04 (0.82–1.31) 0.81 (0.63–1.04) 0.07 
 Multivariatea RR (95% CI) 1.00 1.01 (0.80–1.28) 0.96 (0.76–1.23) 1.00 (0.78–1.27) 0.81 (0.62–1.07) 0.14 
Risk factorQuintile of intakeP, test for trend
12345
Total fiber       
 Median intake (g/day) 12.5 15.4 17.7 20.2 24.8  
 Age-adjusted RR (95% CI) 1.00 0.93 (0.73–1.17) 0.86 (0.68–1.10) 1.01 (0.81–1.27) 0.84 (0.67–1.07) 0.30 
 Multivariatea RR (95% CI) 1.00 0.89 (0.70–1.13) 0.83 (0.65–1.06) 1.00 (0.78–1.26) 0.88 (0.67–1.14) 0.60 
Fiber from cereals       
 Median intake (g/day) 3.0 4.2 5.2 6.4 8.8  
 Age-adjusted RR (95% CI) 1.00 1.35 (1.06–1.72) 1.30 (1.02–1.65) 1.31 (1.03–1.67) 0.90 (0.69–1.16) 0.12 
 Multivariatea RR (95% CI) 1.00 1.31 (1.03–1.67) 1.27 (0.99–1.63) 1.28 (1.00–1.65) 0.91 (0.69–1.21) 0.21 
Fiber from fruits       
 Median intake (g/day) 1.1 2.0 2.9 4.1 6.2  
 Age-adjusted RR (95% CI) 1.00 1.20 (0.94–1.52) 1.07 (0.84–1.37) 1.15 (0.90–1.46) 1.07 (0.84–1.37) 0.91 
 Multivariatea RR (95% CI) 1.00 1.18 (0.93–1.51) 1.05 (0.82–1.35) 1.15 (0.89–1.47) 1.13 (0.88–1.46) 0.54 
Fiber from vegetables       
 Median intake (g/day) 3.3 4.8 6.1 7.6 10.4  
 Age-adjusted RR (95% CI) 1.00 1.16 (0.92–1.48) 1.10 (0.87–1.40) 1.09 (0.86–1.39) 0.95 (0.74–1.21) 0.37 
 Multivariatea RR (95% CI) 1.00 1.14 (0.90–1.45) 1.07 (0.84–1.36) 1.07 (0.84–1.36) 0.97 (0.75–1.24) 0.52 
Fiber from cruciferous vegetables       
 Median intake (g/day) 0.2 0.5 0.7 1.1 1.8  
 Age-adjusted RR (95% CI) 1.00 1.14 (0.90–1.44) 1.08 (0.85–1.37) 1.03 (0.81–1.32) 0.86 (0.67–1.10) 0.06 
 Multivariatea RR (95% CI) 1.00 1.13 (0.90–1.43) 1.06 (0.84–1.35) 1.02 (0.80–1.31) 0.87 (0.68–1.12) 0.08 
Fiber from legumes       
 Median intake (g/day) 0.1 0.4 0.7 1.1 2.0  
 Age-adjusted RR (95% CI) 1.00 1.01 (0.80–1.28) 0.97 (0.77–1.22) 0.89 (0.71–1.13) 0.78 (0.61–1.00) 0.02 
 Multivariatea RR (95% CI) 1.00 0.97 (0.76–1.23) 0.93 (0.73–1.18) 0.89 (0.70–1.12) 0.79 (0.62–1.02) 0.04 
Soluble fiber       
 Median intake (g/day) 3.8 4.6 5.3 6.1 7.4  
 Age-adjusted RR (95% CI) 1.00 0.97 (0.77–1.23) 0.98 (0.77–1.24) 1.07 (0.85–1.34) 0.84 (0.66–1.06) 0.24 
 Multivariatea RR (95% CI) 1.00 0.95 (0.75–1.21) 0.94 (0.74–1.21) 1.06 (0.83–1.34) 0.87 (0.67–1.13) 0.50 
Insoluble fiber       
 Median intake (g/day) 9.5 11.7 13.5 15.4 19.0  
 Age-adjusted RR (95% CI) 1.00 1.06 (0.84–1.34) 1.02 (0.80–1.29) 1.04 (0.82–1.31) 0.81 (0.63–1.04) 0.07 
 Multivariatea RR (95% CI) 1.00 1.01 (0.80–1.28) 0.96 (0.76–1.23) 1.00 (0.78–1.27) 0.81 (0.62–1.07) 0.14 
a

The models were adjusted for the same covariates as the multivariate model in Table 2.

We are indebted to Dr. Meir J. Stampfer for valuable advice and Karen Corsano and Susan Malspeis for computer support.

1
Salmeron J., Manson J. E., Stampfer M. J., Colditz G. A., Wing A. L., Willett W. C. Dietary fiber, glycemic load, and risk of non-insulin-dependent diabetes mellitus in women.
J. Am. Med. Assoc.
,
277
:
472
-477,  
1997
.
2
Salmeron J., Ascherio A., Rimm E. B., Colditz G. A., Spiegelman D., Jenkins D. J., Stampfer M. J., Wing A. L., Willett W. C. Dietary fiber, glycemic load, and risk of NIDDM in men.
Diabetes Care
,
20
:
545
-550,  
1997
.
3
Liu S., Willett W. C., Stampfer M. J., Hu F. B., Franz M., Sampson L., Hennekens C. H., Manson J. E. A prospective study of dietary glycemic load, carbohydrate intake, and risk of coronary heart disease in US women.
Am. J. Clin. Nutr.
,
71
:
1455
-1461,  
2000
.
4
Franceschi S., Dal Maso L., Augustin L., Negri E., Parpinel M., Boyle P., Jenkins D. J., La Vecchia C. Dietary glycemic load and colorectal cancer risk.
Ann. Oncol
,
12
:
173
-178,  
2001
.
5
Michaud D. S., Liu S., Giovannucci E., Willett W. C., Colditz G. A., Fuchs C. S. Dietary sugar, glycemic load, and pancreatic cancer risk in a prospective study.
J. Natl. Cancer Inst. (Bethesda)
,
94
:
1293
-1300,  
2002
.
6
Kaaks R. Nutrition, hormones, and breast cancer: is insulin the missing link?.
Cancer Causes Control
,
7
:
605
-625,  
1996
.
7
Jeppesen J., Schaaf P., Jones G., Zhou M. Y., Chen Y. D., Reaven G. M. Effects of low-fat, high-carbohydrate diets on risk factors for ischemic heart disease in postmenopausal women.
Am. J. Clin. Nutr.
,
65
:
1027
-1033,  
1997
.
8
Liu S., Manson J. E., Stampfer M. J., Holmes M. D., Hu F. B., Hankinson S. E., Willett W. C. Dietary glycemic load assessed by food-frequency questionnaire in relation to plasma high-density-lipoprotein cholesterol and fasting plasma triacylglycerols in postmenopausal women.
Am. J. Clin. Nutr.
,
73
:
560
-566,  
2001
.
9
Muti P., Quattrin T., Grant B. J., Krogh V., Micheli A., Schunemann H. J., Ram M., Freudenheim J. L., Sieri S., Trevisan M., Berrino F. Fasting glucose is a risk factor for breast cancer: a prospective study.
Cancer Epidemiol. Biomark. Prev.
,
11
:
1361
-1368,  
2002
.
10
Augustin L. S., Dal Maso L., La Vecchia C., Parpinel M., Negri E., Vaccarella S., Kendall C. W., Jenkins D. J., Francesch S. Dietary glycemic index and glycemic load, and breast cancer risk: a case-control study.
Ann. Oncol
,
12
:
1533
-1538,  
2001
.
11
Gerber M. Fibre and breast cancer.
Eur. J. Cancer Prev.
,
7
:
s63
-s67,  
1998
.
12
Cohen L. A., Zhao Z., Zang E. A., Wynn T. T., Simi B., Rivenson A. Wheat bran and psyllium diets: effects on N-methylnitrosourea-induced mammary tumorigenesis in F344 rats.
J. Natl. Cancer Inst. (Bethesda)
,
88
:
899
-907,  
1996
.
13
Howe G. R., Hirohata T., Hislop T. G., Iscovich J. M., Yuan J. M., Katsouyanni K., Lubin F., Marubini E., Modan B., Rohan T., Toniolo P., Shunzhang Y. Dietary factors and risk of breast cancer: combined analysis of 12 case-control studies.
J. Natl. Cancer Inst. (Bethesda)
,
82
:
561
-569,  
1990
.
14
Freudenheim J. L., Marshall J. R., Vena J. E., Laughlin R., Brasure J. R., Swanson M. K., Nemoto T., Graham S. Premenopausal breast cancer risk and intake of vegetables, fruits, and related nutrients.
J. Natl. Cancer Inst. (Bethesda)
,
88
:
340
-348,  
1996
.
15
De Stefani E., Correa P., Ronco A., Mendilaharsu M., Guidobono M., Deneo-Pellegrini H. Dietary fiber and risk of breast cancer: a case-control study in Uruguay.
Nutr. Cancer
,
28
:
14
-19,  
1997
.
16
La Vecchia C., Ferraroni M., Franceschi S., Mezzetti M., Decarli A., Negri E. Fibers and breast cancer risk.
Nutr. Cancer
,
28
:
264
-269,  
1997
.
17
Challier B., Perarnau J. M., Viel J. F. Garlic, onion and cereal fibre as protective factors for breast cancer: a French case-control study.
Eur. J. Epidemiol.
,
14
:
737
-747,  
1998
.
18
Ronco A., De Stefani E., Boffetta P., Deneo-Pellegrini H., Mendilaharsu M., Leborgne F. Vegetables, fruits, and related nutrients and risk of breast cancer: a case-control study in Uruguay.
Nutr. Cancer
,
35
:
111
-119,  
1999
.
19
Willett W. C., Hunter D. J., Stampfer M. J., Colditz G., Manson J. E., Spiegelman D., Rosner B., Hennekens C. H., Speizer F. E. Dietary fat and fiber in relation to risk of breast cancer. An 8-year follow-up.
J. Am. Med. Assoc.
,
268
:
2037
-2044,  
1992
.
20
Graham S., Zielezny M., Marshall J., Priore R., Freudenheim J., Brasure J., Haughey B., Nasca P., Zdeb M. Diet in the epidemiology of postmenopausal breast cancer in the New York State cohort.
Am. J. Epidemiol.
,
136
:
1327
-1337,  
1992
.
21
Rohan T. E., Howe G. R., Friedenreich C. M., Jain M., Miller A. B. Dietary fiber, vitamins A, C, and E, and risk of breast cancer: a cohort study.
Cancer Causes Control
,
4
:
29
-37,  
1993
.
22
Jarvinen R., Knekt P., Seppanen R., Teppo L. Diet and breast cancer risk in a cohort of Finnish women.
Cancer Lett.
,
114
:
251
-253,  
1997
.
23
Verhoeven D. T. H., Assen N., Goldbohm R. A., Dorant E., van’t Veet P., Sturmans F., Hermus R. J. J., van den Brandt P. A. Vitamins C and E, retinol, β-carotene and dietary fiber in relation to breast cancer risk: a prospective cohort study.
Br. J. Cancer
,
75
:
149
-155,  
1997
.
24
Terry P., Jain M., Miller A. B., Howe G. R., Rohan T. E. No association among total dietary fiber, fiber fractions, and risk of breast cancer.
Cancer Epidemiol. Biomark. Prev.
,
11
:
1507
-1508,  
2002
.
25
Huang Z., Hankinson S. E., Colditz G. A., Stampfer M. J., Hunter D. J., Manson J. E., Hennekens C. H., Rosner B., Speizer F. E., Willett W. C. Dual effects of weight and weight gain on breast cancer risk.
J. Am. Med. Assoc.
,
278
:
1407
-1411,  
1997
.
26
United States Department of Agriculture. .
Composition of Foods–Raw, Processed, and Prepared, 1963–1992, Agricultural Handbook No. 8 Series
, Department of Agriculture, Government Printing Office Washington, DC  
1993
.
27
Foster-Powell K., Miller J. B. International tables of glycemic index.
Am. J. Clin. Nutr.
,
62
:
871S
-893S,  
1995
.
28
Wolever T. M. S., Jenkins D. J., Jenkins A. L., Josse R. G. The glycemic index: methodology and clinical implications.
Am. J. Clin. Nutr.
,
54
:
846
-854,  
1991
.
29
Willett W., Stampfer M. J. Total energy intake: implications for epidemiologic analyses: review.
Am. J. Epidemiol.
,
124
:
17
-27,  
1986
.
30
Hu F. B., Stampfer M. J., Rimm E. B., Ascherio A., Rosner B. A., Spiegelman D., Willett W. C. Dietary fat and coronary heart disease: a comparison of approaches for adjusting for total energy intake and modeling repeated dietary measurements.
Am. J. Epidemiol.
,
149
:
531
-540,  
1999
.
31
Willett W. C., Lenart E. B. Reproducibility and validity of food-frequency questionnaires.
Nutritional Epidemiology
, 2nd ed Oxford University Press New York  
1998
.
32
Salvini S., Hunter D. J., Sampson L., Stampfer M. J., Colditz G. A., Rosner B., Willett W. C. Food-based validation of a dietary questionnaire: the effects of week-to-week variation in food consumption.
Int. J. Epidemiol.
,
18
:
858
-867,  
1989
.
33
Cox D. R., Oakes D. .
Analysis of Survival Data
, Chapman and Hall London  
1984
.
34
SAS/STAT Software. The PHREG Procedure.
Preliminary documentation
, SAS Institute Cary, NC  
1991
.
35
Therneau T. M. Extending the Cox Model Lin D. Y. Fleming T. R. eds. .
Proceedings of the First Seattle Symposium in Biostatistics: Survival Analysis
,
51
-84, Springer Verlag New York  
1997
.
36
Cho E., Spiegelman D., Hunter D. J., Chen W. Y., Stampfer M. J., Colditz G. A., Willett W. C. Premenopausal fat intake and risk of breast cancer.
J. Natl. Cancer Inst.
,
95
:
1079
-1085,  
2003
.
37
Yang G., Lu G., Jin F., Dai Q., Best R., Shu X. O., Chen J. R., Pan X. Y., Shrubsole M., Zheng W. Population-based, case-control study of blood C-peptide level and breast cancer risk.
Cancer Epidemiol. Biomark. Prev.
,
10
:
1207
-1211,  
2001
.
38
Michels K. B., Solomon C. G., Hu F. B., Rosner B. A., Hankinson S. E., Colditz G. A., Manson J. E. Type 2 diabetes and subsequent incidence of breast cancer in the Nurses’ Health Study.
Diabetes Care
,
26
:
1752
-1758,  
2003
.
39
Rose D. P., Goldman M., Connolly J. M., Strong L. E. High-fiber diet reduces serum estrogen concentrations in premenopausal women.
Am. J. Clin. Nutr.
,
54
:
520
-525,  
1991
.
40
Witte J. S., Ursin G., Siemiatycki J., Thompson W. D., Paganini-Hill A., Haile R. W. Diet and premenopausal bilateral breast cancer: a case-control study.
Breast Cancer Res. Treat.
,
42
:
243
-251,  
1997
.
41
Potischman N., Swanson C. A., Coates R. J., Gammon M. D., Brogan D. R., Curtin J., Brinton L. A. Intake of food groups and associated micronutrients in relation to risk of early-stage breast cancer.
Int. J. Cancer
,
82
:
315
-321,  
1999
.
42
Horn-Ross P. L., Hoggatt K. J., West D. W., Krone M. R., Stewart S. L., Anton-Culver H., Bernstein L., Deapen D., Peel D., Pinder R., Reynolds P., Ross R. K. Recent diet and breast cancer risk: the California Teachers Study (USA).
Cancer Causes Control
,
13
:
407
-415,  
2002
.