Meta-analysis: Lack of evidence for estrogen replacement therapy as a risk for breast cancer
ACP J Club. 1991 May-June;114:82. doi:10.7326/ACPJC-1991-114-3-082
Dupont WD, Page DL. Therapy and breast cancer. Arch Intern Med. 1991 Jan;151:67-72.
To investigate the effects of type, duration, and dosage of estrogen replacement therapy (ERT) on the risk for breast cancer in women.
MEDLINE was used to identify studies reported since 1972 through headings and key words: female and human and breast neoplasms and estrogens, and occurrence or etiology or epidemiology or chemically induced.
Of 571 studies identified, 50 provided estimates of the risk for breast cancer among women who received ERT. Studies that had < 5 patients (6 studies) or too great a proportion of premenopausal women with breast cancer being analyzed with those of postmenopausal women (3) and studies published more than once (13) were excluded, yielding 28 articles for detailed review.
All 28 studies gave explicit criteria for patient selection and made adjustments for confounding variables. Not all studies maximized follow-up of patients, confirmed cases of cancer histologically, excluded patients with previous breast cancer from the control group in case-control studies, or excluded women with premenopausal cancer. Adjusted log-relative risks were calculated from each study.
The overall relative risk for breast cancer associated with ERT was 1.07. However, relative risks from these studies differed (P < 0.001). The overall risk for women who took a daily dose ≥ 1.25 mg/d did not provide consistent estimates of risk. Women with a history of benign breast disease who had received ERT had an overall relative risk for breast cancer of 1.16 (CI 0.89 to 1.5).
The overall effects of the duration, dosage, or type of estrogen on the risk for breast cancer found in reported studies varies. Consistent evidence indicates that a dose of ≤ 0.625 mg/d of conjugated estrogens does not appreciably increase the risk for breast cancer, even among women with a history of benign breast disease.
Source of funding: National Cancer Institute.
Address for article reprint: Dr. W.D. Dupont, Department of Preventive Medicine, A-1128 Medical Center North, Vanderbilt University School of Medicine, Nashville, TN 37232-2637, USA.
The risk-benefit equation for the long-term use of postmenopausal ERT is complicated. ERT is beneficial in preventing osteoporosis, in inducing favorable changes in lipoproteins, and perhaps in reducing the incidence of cardiovascular death. Although ERT is known to increase the risk for endometrial cancer, the role of ERT in breast cancer is uncertain. In this meta-analysis, Dupont and Page try to clarify the conflicting conclusions of studies that were done in the past 20 years.
Although the methods were rigorous, the presentation would have been improved by including a table of the criteria used to assess studies for inclusion. Second, the results of case-control and cohort studies of ERT and breast cancer were reported together. Whether a trend based on study design exists would be of interest and could be shown graphically.
Dupont and Page found no consistent relations among the duration, dose, and type of ERT and breast cancer. Yet they did not definitely rule out an association. Sacks (1) and others have proposed that nonrandomized studies should be included in a medical meta-analysis. In many of the studies included in Dupont and Page's meta-analysis, biases could have influenced whether ERT was given and could have distorted the relation between ERT and breast cancer. Because no randomized trial of ERT is in progress, large prospective cohort studies, such as the Nurses' Health Study (2) will provide the best answers. Even these studies will leave room for doubt.
Bruce E. Hillner, MD
Medical College of VirginiaRichmond, Virginia, USA