A Prospective Study of Multivitamin Supplement Use and Risk of Breast Cancer

Ken Ishitani; Jennifer Lin; JoAnn E. Manson; Julie E. Buring; Shumin M. Zhang


Am J Epidemiol. 2008;167(10):1197-1206. 

In This Article

Materials and Methods

The Women's Health Study was established in 1992 when 39,876 female US health professionals (registered nurses, 75 percent) aged 45 years or older and free of cancer and cardiovascular disease at baseline were enrolled in a randomized trial evaluating the benefits and risks of low-dose aspirin and vitamin E in the primary prevention of cancer and cardiovascular disease.[13,14,15,16] Upon enrollment in the study, all participants completed a baseline questionnaire inquiring about their medical history and lifestyle factors, including potential risk factors for breast cancer. As of the end of the trial (March 31, 2004), the average duration of follow-up was 10 years, and follow-up rates for morbidity and mortality were 97.2 percent and 99.4 percent, respectively.[13,14,16] The current analysis was restricted to 37,920 women after we excluded those who did not provide information on multivitamin supplement use and diet, had implausible total energy intakes (<600 kcal/day or >3,500 kcal/day), or had prerandomization cancers that were reported and confirmed after randomization.

Information about the status (never, past, or current) and duration (0–1, 2–4, 5–9, 10–14, 15–19, or ≥20 years) of multivitamin supplement use was asked on the enrollment questionnaire at baseline. At baseline, 39,345 (98.7 percent) women in the Women's Health Study also completed a 131-item food frequency questionnaire, a format that has been used in the Nurses' Health Study. The questionnaire assessed average consumption over the past year of a specific amount of each food and allowed nine responses, ranging from "never" to "six or more times per day." The food frequency questionnaire also included a section on current use of multivitamin supplements. For current users, women were asked about the exact brand and type of multivitamins they used (write-in format) and how many times they took multivitamins per week (≤2, 3–5, 6–9, or ≥10). These categories of multivitamin supplement use were included in a previous study of colorectal cancer in this cohort.[17] The validity and reliability of the food frequency questionnaire have been assessed in the Nurses' Health Study, which has demographic characteristics similar to those of the Women's Health Study.[18,19,20]

Every 6 months during the first year of follow-up and then annually thereafter, participants were sent questionnaires asking about newly diagnosed diseases, including breast cancer. Deaths of participants were identified through reports from family members, postal authorities, and a search of the National Death Index. We sought medical records and other relevant information, which were reviewed by an endpoints committee consisting of physicians to confirm medical diagnoses. Medical record review confirmed 98 percent of self-reported breast cancer cases in the Women's Health Study.[21] Detailed information on tumor characteristics at diagnosis, including hormone receptor status, tumor size, lymph node metastasis, histology, and histologic grading and differentiation, was also extracted from medical records. Laboratories affiliated with hospitals where breast cancer cases were diagnosed determined hormone receptor status.

During an average of 10 years of follow-up, we ascertained 1,171 confirmed cases of invasive breast cancer, which were included in this analysis. Of the 1,171 invasive cancers, 790 (67.5 percent) were positive for both estrogen receptor (ER) and progesterone receptor (PR) (ER+PR+), 123 (10.5 percent) were positive for ER but negative for PR (ER+PR–), 23 (2.0 percent) were ER–PR+, 165 (14.1 percent) were ER–PR–, and, for 70 (6.0 percent), ER or PR status was unknown. Tumors classified as borderline positive for ER (n = 5) or PR (n = 8) were considered ER+ or PR+ in the analyses. In addition, 848 of the tumors (72.4 percent) were ≤2 cm, 273 (23.3 percent) were >2 cm, five (0.4 percent) were of any size with direct extension to the chest wall or skin, and, for 50 (4.3 percent), tumor size was unknown. Moreover, 828 (70.7 percent) were negative for lymph node metastasis, 279 (23.8 percent) were positive for lymph node metastasis, and 64 (5.5 percent) were unknown for lymph node metastasis. Furthermore, 261 cases (22.3 percent) had well-differentiated tumors, 480 (41.0 percent) had moderately differentiated tumors, 275 (23.5 percent) had poorly differentiated tumors, and 155 (13.2 percent) had tumors whose histologic grading and differentiation were unknown.

We first compared mean values or proportions of baseline risk factors for breast cancer according to status of multivitamin supplement use to evaluate potential confounding by these variables.

Person-years were calculated for each participant, ranging from the date of randomization to the date of confirmed cancer diagnosis, death, or March 31, 2004, whichever occurred first. Cox proportional hazards regression models were used to calculate relative risks and 95 percent confidence intervals.[22] We initially estimated the relative risks according to categories of multivitamin supplement use, with adjustment for age (in years) and randomized treatment assignment (aspirin vs. placebo, vitamin E vs. placebo). We further performed a multivariable analysis that additionally adjusted for known or potential risk factors for breast cancer at baseline, including alcohol intake (none, >0–<10, ≥10–<15, ≥15–<30, or ≥30 g/day), body mass index (<23, ≥23–<25, ≥25–<27, ≥27–<30, or ≥30 kg/m2), family history of breast cancer in a first-degree relative (yes or no), history of hysterectomy (yes or no), bilateral oophorectomy (yes or no), smoking status (never, past, or current), benign breast disease (yes or no), age at menarche (≤11, 12, 13, 14, or ≥15 years), parity (0, 1–2, 3–4, 5, or ≥6), age at first birth (≤24, 25–29, or ≥30 years), physical activity (kcal/week, in quartiles), total energy intake (kcal/day, in quintiles), menopausal status (premenopausal, postmenopausal, or uncertain/unknown), and postmenopausal hormone use (never, past, or current and duration). We conducted an additional multivariable analysis by excluding incident cases of breast cancer diagnosed within the first 2 years of follow-up, with additional adjustments for mammography screening in the past year, which was asked on the 12-month questionnaire.

We also performed an analysis according to combined ER and PR status (ER+PR+, ER+PR–, ER–PR–, and unknown), tumor size (≤2 and >2 cm), lymph node metastasis (with metastasis and without metastasis), and histologic grading and differentiation (well, moderately, and poorly differentiated). Because of limited numbers, ER–PR+ breast cancer cases (n = 23), or cases whose tumors were any size with direct extension to the chest wall or skin (n = 5) or whose tumor size was unknown (n = 50), or cases with unknown lymph node metastasis (n = 64), or cases with unknown histologic grading and differentiation (n = 155) were excluded from the analysis of tumor characteristics. All p values were two sided. Tests for interaction were conducted according to categories of alcohol intake (0, >0–<10, or ≥10 g/day), menopausal status (premenopausal or postmenopausal), randomized vitamin E treatment assignment (vitamin E vs. placebo), and postmenopausal hormone use (never, past, or current) and were performed by log likelihood ratio tests comparing the models with or without interaction terms.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: