Low-Fat Dietary Pattern and Risk of Benign Proliferative Breast Disease: A Randomized, Controlled Dietary Modification Trial

Breast cancer incidence rates vary 4- to 5-fold internationally and have been observed to change on migration from low-risk to high-risk countries (1). These observations have led to the hypothesis that modifiable factors, including diet, might influence breast cancer risk. Most of the focus in epidemiologic studies of the association between diet and breast cancer has been on the roles of dietary fat, fiber, and fruit and vegetable intake. In general, cohort studies have not supported an association between dietary fat intake and breast cancer risk (2–5), although several recent cohort studies have reported small increases in risk in postmenopausal women (6–8). Additionally, there is little recent evidence from cohort studies that fruit and vegetable consumption (9, 10) and dietary fiber intake (11) are associated with altered breast cancer risk, although earlier studies provided some support for inverse associations (5).

The Women's Health Initiative (WHI) randomized controlled Dietary Modification trial was the first large-scale randomized trial to test the effect of adoption of a low-fat dietary pattern (decrease in total fat intake, increase in fruit, vegetable, and grain intake) on breast cancer risk (12). After ∼8 years of follow-up, risk in the intervention group was 9% lower than that in the comparison group, but the effect was not statistically significant. Among the possible explanations for this lack of effect is that the follow-up period was too short. If that is the case, it does not preclude the possibility of an effect of the intervention on earlier stages in the natural history of breast cancer in the short term, with a subsequent (and consequent) reduction in breast cancer incidence. Therefore, we used the WHI Dietary Modification trial to test the effect of adoption of a low-fat dietary pattern on risk of benign proliferative breast disease, a condition which is associated with increased risk of, and is considered to be on the pathway to, invasive breast cancer (13–15).

The WHI Dietary Modification trial has been described in detail elsewhere (12, 16, 17). In brief, 48,835 postmenopausal women, ages 50 to 79 y at initial screening, who were likely to reside in the area for 3 y, and who provided written informed consent, were enrolled between 1993 and 1998 at 40 clinical centers throughout the United States. Major reasons for ineligibility for the trial included a history of breast or colorectal cancer, a history of other cancer except nonmelanoma skin cancer in the last 10 y, medical conditions with predicted survival of <3 y, adherence and retention considerations (e.g., alcoholism, dementia), or a diet at baseline with fat intake of <32% of total energy. All participants had a baseline mammogram and clinical breast examination; abnormal findings suggestive of breast cancer required clearance before study entry. Eligible women were randomized either to the dietary modification intervention group (40%; n = 19,541) or to the comparison group (60%; n = 29,294) using a permuted block algorithm with stratification by clinic and age. Dietary modification participants were also eligible to be in the WHI hormone therapy and calcium/vitamin D supplementation trials (16). The WHI and the ancillary study reported here (in which all 40 WHI clinical centers participated) were approved by institutional review boards at all participating institutions.

The dietary intervention was designed to promote dietary change with the goals of reducing intake of total dietary fat to 20% of total energy intake (as a consequence of which, saturated fat intake was expected to be reduced to ∼7% of energy intake), increasing fruit and vegetable intake to ≥5 servings/d, and increasing intake of grain products to at least 6 servings/d (12). Women in the comparison group were not offered a nutrition intervention program because the general strategy for this group was minimum interference with customary diets while collecting nutritional data considered appropriate for comparison with the intervention group. However, they were provided with a copy of the dietary guidelines for Americans. Neither the intervention nor the comparison group was asked to modify health-related behaviors, including use of dietary supplements.

Dietary change was encouraged by providing social support and positive interactions. To this end, women randomized to the intervention arm were assigned to a permanent group of 8 to 15 members led by a designated nutritionist. During the first year of follow-up, the group met weekly for 6 wk, biweekly for 6 wk, and monthly for 9 mo; subsequently, it met quarterly. Each participant had an individual counseling session with her group nutritionist between 12 and 16 wk from the beginning of the intervention sessions.

Comprehensive information on breast cancer risk factors was obtained at baseline by interview (for lifetime hormone use) and by self-report (other covariates) using standardized questionnaires (16).

Study participants were contacted every 6 mo for collection of information on outcomes. Clinic visits were required annually, at which time height and weight were measured using standardized procedures. Mammograms were required every 2 y, and regular clinical breast exams were done as well.

Dietary intake for all participants was monitored using the WHI food frequency questionnaire (12, 18). The food frequency questionnaire was administered to all participants at baseline and 1 y after randomization. Thereafter, about one third of participants completed the food frequency questionnaire annually in a rotating sample. Furthermore, 4-d food records were provided by all women before randomization (12), a 4.6% sample provided further 4-d food records at 1 y after randomization, as well as 24-h dietary recalls at 3 and 6 y postrandomization, and 1% samples of women provided a 24-h dietary recall annually. The dietary data presented here were derived from the baseline food frequency questionnaire.

The outcome of interest for the present study was histologically confirmed incident benign proliferative breast disease with or without atypia (see “Histology”). Clinical events including breast cancers and breast biopsies for noncancerous lesions were initially identified from self-administered questionnaires completed every 6 mo. Breast cancers were confirmed by local and central adjudicators who reviewed medical records and pathology reports and who were blinded both to treatment assignment and to symptoms due to study interventions. For the present study, women who reported breast biopsies that were free of cancer were identified, and clinical centers were sent lists of potentially eligible subjects quarterly. Clinic staff contacted participants to obtain written informed consent to solicit the histologic sections resulting from the biopsies. To investigate the possibility that breast biopsies were missed by using this approach, the charts of 100 randomly selected participants who did not report a breast biopsy were reviewed at one center and none was found to have unreported biopsies.

H&E-stained histologic sections were reviewed by the study pathologist (D.L.P.), who was blinded to the randomization assignment. The benign lesions were classified using well-established criteria as nonproliferative lesions, proliferative lesions without atypia (classified further according to whether they were mild, moderate, or florid in extent), or atypical (ductal/lobular) hyperplasia (19–21).

Incidence rates of benign proliferative breast disease in the dietary modification and comparison groups were compared based on the intention-to-treat principle using time-to-event analyses. The primary analysis used a weighted (two-sided) log-rank test with weight increasing linearly from zero at randomization to a maximum of 1 at 10 y and constant thereafter to enhance statistical power under the design assumptions (12). The time to benign proliferative breast disease was defined as the number of days from the date of randomization to the date of the first postrandomization diagnostic biopsy. Follow-up time was censored at the date of last documented contact, diagnosis of breast cancer, mastectomy, death, or trial close-out (between October 2004 and March 2005), whichever came first. Women who developed a nonproliferative benign breast lesion continued to be followed up because they remained at risk of developing a subsequent proliferative lesion. Event rates over time were summarized using cumulative hazard plots. The intervention effect was summarized using hazard ratios (HR) and 95% confidence intervals (95% CI) estimated from Cox proportional hazards models (22), with stratification by age, prior breast biopsies, and randomization to the WHI hormone therapy and calcium/vitamin D supplementation trials. Stratification was time dependent in the case of the calcium/vitamin D supplementation trial (16). The possibility that the intervention effect differed by levels of other characteristics of the study population was investigated by including product terms between treatment assignment and indicator variables for the subsets of interest in Cox proportional hazards models stratified by age, prior breast biopsies, and randomization to the hormone therapy and calcium/vitamin D trials (16), and was assessed by testing the equality of the product-term coefficients. Given that a total of 22 interactions were tested, approximately one significant test at an α level of 0.05 was expected by chance alone. The proportional hazards assumption, which was tested both by fitting models containing a product term between the intervention and follow-up time and assessing the coefficient of the product term for statistical significance, and by fitting piecewise constant intervention effects on nonoverlapping time intervals and testing for equality of such terms with the intervals chosen in advance, was shown not to be violated. Annualized event rates were calculated for comparisons of absolute disease rates. Results were considered statistically significant when two-sided P ≤ 0.05.

The study groups differed little at baseline with respect to age, ethnicity, breast cancer risk factors, participation in other WHI trials, or intake of energy or selected nutrients and vitamins (Table 1).

Data on follow-up and adherence, as well as change in nutrient and food intake and blood biomarker levels, were reported elsewhere (12). In brief, 4.7% of women in the intervention group and 4.0% of those in the comparison group withdrew from the study or were lost to follow-up. Estimated adherence rates were 87% and 75% in the comparison group and 57% and 31% in the intervention group at years 3 and 6, respectively. Based on estimates from the food frequency questionnaires, the intervention and comparison groups were essentially identical at baseline with respect to intake of fat, fruit and vegetables, and grains (as well as other dietary items). One year after randomization, statistically significant, albeit modest, between-group differences in intake of these dietary items were evident: For the intervention and comparison group, the mean (SD) percent energy from fat, servings of fruit and vegetables per day, and servings of grains per day were 24.3 (7.5) versus 35.1 (6.9), 5.1 (2.3) versus 3.9 (2.0), and 5.1 (2.7) versus 4.2 (2.3), respectively. By year 6, the mean (SD) difference in change (change in intervention group minus change in comparison group) in percent energy from fat between the intervention and comparison groups was –8.1 (7.8), whereas the mean differences in change for servings per day of fruit and vegetables and of grains were 1.1 (2.1) and 0.4 (2.6), respectively; all of these differences were statistically significant. Compared with blood levels in the comparison group, the intervention group showed a greater reduction in blood levels of γ-tocopherol, a greater increase in β-cryptoxanthin, and smaller decreases in α- and β-carotene. Estradiol levels decreased more and sex hormone-binding levels increased more in the intervention group than in the control group.

During follow-up (average duration, 7.7 years), we identified 3,383 potentially eligible biopsies that had been done for benign breast disease. The eligibility of 65 biopsies could not be determined due to lack of consent, hospital refusal, and other reasons. Of the 3,318 biopsies confirmed to be eligible, consent was provided for review of 3,314, and 3,254 histologic sections were obtained. Of the sections reviewed, 172 were from biopsies that occured outside the trial period and 217 had no breast tissue. The remaining 2,865 sections were from 2,610 women. Of these women, 19 were censored (so that the corresponding section was excluded from consideration), 7 had no pathological diagnosis, 1,221 had a nonproliferative lesion (and therefore continued to be eligible to develop a proliferative lesion), and 1,363 had a proliferative lesion.

Overall, 570 cases of benign proliferative breast disease were ascertained in the intervention group and 793 were ascertained in the comparison group. The estimated HR for the association between dietary modification and benign proliferative breast disease was 1.09 (95% CI, 0.98-1.23; Table 2). The intervention was associated with a slight, statistically nonsignificant increase in risk of benign proliferative breast disease without atypia (HR, 1.10; 95% CI, 0.97-1.25) and with a statistically significant increase in risk for either atypical hyperplasia or moderately extensive or florid proliferative disease without atypia (HR, 1.16; 95% CI, 1.02-1.33). Risk of atypical hyperplasia was essentially unaltered by the intervention (HR, 1.05; 95% CI, 0.79-1.39). The Kaplan-Meier estimate of the cumulative hazard of benign proliferative breast disease (all types combined) revealed that there was little difference between the intervention and comparison groups for the first 4 years of follow-up, after which the hazard for the dietary modification group exceeded slightly that for the comparison group (Fig. 1).

The overall dietary modification effect was largely unchanged by exclusion of the first year of follow-up (HR, 1.09; 95% CI, 0.97-1.23), exclusion of women with a breast biopsy before the commencement of the trial (HR, 1.07; 95% CI, 0.93-1.23), or adjustment for annual measures of height and weight (HR, 1.09; 95% CI, 0.98-1.23) or use of prior hormone therapy (estrogen alone or estrogen plus progestin; HR, 1.09; 95% CI, 0.97-1.22).

Risk of benign proliferative breast disease in association with dietary modification varied by levels of baseline total vitamin D intake (from diet and supplements), albeit not substantially and not monotonically. Risk did not vary by levels of other dietary variables (Table 3), and it varied little by levels of most of the other baseline variables examined (Table 4). Although risk varied significantly by levels of estrogen plus progestin use as reported at baseline, the magnitude of the difference in risk between categories of this variable was not large.

To address the possible effect of nonadherence on the study results, we used an inverse probability weighting scheme to estimate a “full adherence” relative risk function for the intervention as described earlier (12). This yielded a HR of 1.37 (95% CI, 0.95-1.99) for all proliferative breast diseases combined.

The WHI study protocol mandated that participants had biennial mammograms and regular clinical breast exams. Compliance with these exams was high, and there was essentially no difference between the intervention and comparison groups with respect to the frequency with which these were done (12). Neither adjustment for the frequency of mammograms (HR, 1.10; 95% CI, 0.98-1.23) nor adjustment for the frequency of clinical breast exams (HR, 1.09; 95% CI, 0.98-1.23) had much impact on the estimated dietary modification effect.

The results of this study suggest that the WHI low-fat dietary intervention, which yielded a statistically significant, albeit modest, decrease in fat intake and increase in intake of fruit, vegetables, and grains, was not associated with altered risk of benign proliferative breast disease after almost 8 years of follow-up. Although there was a small increase in overall risk (9%) that was even larger (37%) when the extent of adherence to the dietary intervention was taken into account, neither of these effect estimates was statistically significant. Furthermore, although the intervention was associated with a statistically significant increase in the risk of the combined outcome category consisting of either atypical hyperplasia or moderately extensive or florid proliferative disease without atypia, the magnitude of this effect was small (16% increase in risk). Similarly, although there were statistically significant interactions between the intervention and baseline total vitamin D intake, postmenopausal hormone use before entry into the trial, and enrollment in the WHI estrogen plus progestin trial, the magnitude of the variations in risk by categories of these variables was not large.

It seems that there have not been any previous randomized trials of the effect of dietary modification on the risk of benign proliferative breast disease. However, the role of diet in the etiology of benign proliferative breast disease has been examined in several case-control (23–27) and cohort studies (28–31). Of the case-control studies, two (24, 25) showed positive associations between saturated fat intake (or indices thereof) and risk of atypical (24) or proliferative forms (25) of benign breast disease, whereas the other studies (23, 26, 27) provided little support for a role for dietary fat. With respect to other nutrients, one case-control study provided some evidence for inverse associations between retinol and β-carotene intake and risk (26) and also showed strong inverse associations for dietary fiber and its components (soluble and insoluble nonstarch polysaccharides and cellulose; ref. 32). These findings were supported to some extent by those of another study (23) in which risk of benign epithelial hyperplasia was reduced in association with consumption of fruit and leafy and orange-red vegetables, both of which contain fiber and micronutrients such as β-carotene. However, in another study (33), carotene and retinol intakes were not associated with risk of atypical or nonatypical forms of benign proliferative breast disease. Findings from cohort studies have been less supportive of associations. Of the four cohort studies to date (28–31), three (29–31) measured diet in adulthood and showed that intakes of dietary fat, carotenoid, fiber (29, 30), and folate (31) were not related to risk of subsequent benign proliferative breast disease, whereas one study (28) showed that vitamin E and fiber intakes during adolescence had inverse associations with risk that were of borderline statistical significance.

Although the dietary intervention was not associated with altered risk of benign proliferative breast disease overall, it was associated with an increase in risk in some subgroup analyses. It is conceivable that the latter represent chance findings due to the many subgroup analyses that were done. However, the dietary intervention was complex, and although it seems unlikely that detrimental changes might have been introduced as a result of the intervention, an alteration in the balance between nutrients that mitigate the risk of cancer and those that increase risk might have occurred.

The mammary gland is particularly susceptible to environmental influences early in life (34, 35). Indeed, early life events are considered to play a role in the etiology of breast cancer, as evidenced by the positive associations of breast cancer risk with birthweight and height (a marker of the adequacy of nutritional status in childhood; ref. 36). Hence, it is conceivable that the essentially null results of this trial (and those of the parent trial; ref. 12) reflect the fact that the intervention was administered well after the critical exposure period.

Limitations of the parent trial were discussed elsewhere (12). In brief, recruitment took longer than expected so that the average follow-up duration was less than had been planned and perhaps insufficiently long to show an effect (in this regard, follow-up of the Dietary Modification trial participants is continuing and an updated report is expected soon); the between-group (intervention versus comparison) difference in percent energy from fat was only ∼70% of the target; relatively few women managed to reduce their fat intake to 20% of energy; self-report methods were used to assess between-group differences in dietary intake, although the observed relative changes in blood levels of γ-tocopherol and carotenoids are consistent with reported differences between the randomization groups with respect to consumption of fats, oils, and fruits and vegetables; and the complexity of the intervention (reduction in fat intake and increase in fruit, vegetable, and grain intake) limits the ability to attribute effects to any specific component of the dietary changes. With respect to the present study, there are two additional potential limitations. First, differential ascertainment of the outcome in the two randomization groups may have occurred, although this seems unlikely given that compliance with the biennial mammograms and regular clinical breast exams was high and essentially the same in both groups, and that adjustment for the frequency of these exams had minimal effect on the estimate of the dietary modification effect (underascertainment in both groups is a possibility given that breast biopsies were not done on all study subjects). Second, it is possible that there was some misclassification of the outcome, although this is likely to have been nondifferential, with consequent biasing of the effect estimate for dietary intervention toward the null (37).

In conclusion, the results of the trial reported here suggest that a modest reduction in fat intake and a modest increase in fruit, vegetable, and grain intake in postmenopausal women do not alter their risk of subsequent benign proliferative breast disease.

No potential conflicts of interest were disclosed.

We thank the WHI investigators and staff for their outstanding dedication and commitment. A list of key investigators involved in this research follows. A full listing of WHI investigators can be found at the following website: http://www.whi.org.

Program Office: National Heart, Lung, and Blood Institute (Bethesda, MD)—Elizabeth Nabel, Jacques Rossouw, Shari Ludlam, Linda Pottern, Joan McGowan, Leslie Ford, and Nancy Geller.

Clinical Coordinating Centers: Fred Hutchinson Cancer Research Center (Seattle, WA)—Ross Prentice, Garnet Anderson, Andrea LaCroix, Charles L. Kooperberg, Ruth E. Patterson, and Anne McTiernan; Wake Forest University School of Medicine (Winston-Salem, NC)—Sally Shumaker; Medical Research Labs (Highland Heights, KY)—Evan Stein; University of California at San Francisco (San Francisco, CA)—Steven Cummings.

Clinical Centers: Albert Einstein College of Medicine (Bronx, NY)—Sylvia Wassertheil-Smoller; Baylor College of Medicine (Houston, TX)—Jennifer Hays; Brigham and Women's Hospital, Harvard Medical School (Boston, MA)—JoAnn Manson; Brown University (Providence, RI)—Annlouise R. Assaf; Emory University (Atlanta, GA)—Lawrence Phillips; Fred Hutchinson Cancer Research Center (Seattle, WA)—Shirley Beresford; George Washington University Medical Center (Washington, DC)—Judith Hsia; Los Angeles Biomedical Research Institute at Harbor-University of California at Los Angeles Medical Center (Torrance, CA)—Rowan Chlebowski; Kaiser Permanente Center for Health Research (Portland, OR)—Evelyn Whitlock; Kaiser Permanente Division of Research (Oakland, CA)—Bette Caan; Medical College of Wisconsin (Milwaukee, WI)—Jane Morley Kotchen; MedStar Research Institute/Howard University (Washington, DC)—Barbara V. Howard; Northwestern University (Chicago/Evanston, IL)—Linda Van Horn; Rush Medical Center (Chicago, IL)—Henry Black; Stanford Prevention Research Center (Stanford, CA)—Marcia L. Stefanick; State University of New York at Stony Brook (Stony Brook, NY)—Dorothy Lane; The Ohio State University (Columbus, OH)—Rebecca Jackson; University of Alabama at Birmingham (Birmingham, AL)—Cora E. Lewis; University of Arizona (Tucson/Phoenix, AZ)—Tamsen Bassford; University at Buffalo (Buffalo, NY)—Jean Wactawski-Wende; University of California at Davis (Sacramento, CA)—John Robbins; University of California at Irvine (Irvine, CA)—F. Allan Hubbell; University of California at Los Angeles (Los Angeles, CA)—Howard Judd; University of California at San Diego (LaJolla/Chula Vista, CA)—Robert D. Langer; University of Cincinnati (Cincinnati, OH)—Margery Gass; University of Florida (Gainesville/Jacksonville, FL)—Marian Limacher; University of Hawaii (Honolulu, HI)—David Curb; University of Iowa (Iowa City/Davenport, IA)—Robert Wallace; University of Massachusetts/Fallon Clinic (Worcester, MA)—Judith Ockene; University of Medicine and Dentistry of New Jersey (Newark, NJ)—Norman Lasser; University of Miami (Miami, FL)—Mary Jo O'Sullivan; University of Minnesota (Minneapolis, MN)—Karen Margolis; University of Nevada (Reno, NV)—Robert Brunner; University of North Carolina (Chapel Hill, NC)—Gerardo Heiss; University of Pittsburgh (Pittsburgh, PA)—Lewis Kuller; University of Tennessee (Memphis, TN)—Karen C. Johnson; University of Texas Health Science Center (San Antonio, TX)—Robert Brzyski; University of Wisconsin (Madison, WI)—Gloria E. Sarto; Wake Forest University School of Medicine (Winston-Salem, NC)–Denise Bonds; Wayne State University School of Medicine/Hutzel Hospital (Detroit, MI)—Susan Hendrix.

The WHI program is funded by the National Heart, Lung and Blood Institute, U.S. Department of Health and Human Services.