Causing and Preventing Breast Cancer

By Leon Speroff, MD

In the past few months, we have been told that estrogen agonists/antagonists (tamoxifen and raloxifene) can prevent breast cancer. And, of course, we are all aware that some studies have indicated that long-term postmenopausal hormone therapy moderately (or slightly) increases the risk of breast cancer (although more studies have not). I have been wondering whether a somewhat similar phenomenon explains these observations.

The Tamoxifen Preventive Trial

Women at increased risk for breast cancer participated in a breast cancer prevention trial initiated in the United States in 1992. The study was intended to compare two groups of women, one treated with placebo and one with 20 mg tamoxifen daily for five years. Early in 1998 (after about 4 years of follow-up), the study was unblinded because there were 45% fewer cases of invasive breast cancer in the tamoxifen-treated arm of the study.1 This outcome was not without risk. There was a 2.4-fold increase in postmenopausal endometrial cancer,1 a 2.8-fold increase in pulmonary embolism, and a 1.6-fold increase in venous thrombosis. (See Table 1.)

Table 1

The Tamoxifen Preventive Trial

Breast Cancer
85/6600 (1.3%)
154 (2.3%)
Endometrial Cancer 
33 (0.05%)
14 (0.2%)
Pulmonary Embolism
Venous Thromboembolism

Results with Raloxifene

A greater understanding of the estrogen receptor mechanism allows us to comprehend how mixed estrogen agonists-antagonists can have selective actions on specific target tissues. New agents are being developed in an effort to isolate desired actions from unwanted side effects. In relatively short-term studies, raloxifene exerts no proliferative effect on the endometrium but produces favorable responses in bone and lipids.2,3 Early results (2-3 years of treatment) indicate that women receiving raloxifene have a reduction in breast cancer incidence. The results were reported at the recent meeting of the American Society of Clinical Oncologists. (See Table 2.)

Short-term clinical trial data indicate that raloxifene has an effect on bone that is not as great as estrogen, the effect on lipids may not be sufficient to protect against cardiovascular disease, and raloxifene increases hot flushing.4 Long-term clinical trial data will be necessary to determine the ultimate effect on clinical events, specifically fractures, coronary heart disease, stress incontinence, endometrial cancer, and cognition. In a two-year randomized trial in monkeys, raloxifene exerted no protection against coronary artery atherosclerosis despite changes in circulating lipids similar to those achieved in women.5 In a rabbit model, raloxifene did inhibit aortic atherosclerosis but not as effectively as estrogen treatment.6

Table 2

Results with Raloxifene

    At the end of 33 months, 7705 women with osteoporosis showed the following results:

    Placebo group 1% incidence of breast cancer

    Raloxifene group 0.25-0.30% incidence of breast cancer

    Invasive breast cancer 22 cases in placebo group

      13 cases in raloxifene group

      Effect observed only in estrogen

      receptor positive tumors

      (70% reduction): 54% reduction overall, including all tumors.

Estrogen and the Risk of Breast Cancer

Many of the studies that have examined the mortality rates of women who were taking estrogen at the time of breast cancer diagnosis have documented improved survival rates.7-9 For example, the American Cancer Society nine-year, prospective follow-up documented a 16% reduced risk of fatal breast cancer.9 Undoubtedly, this reflects earlier diagnosis in users because the greater survival rate in current users is associated with a lower frequency of late stage disease.10-13 There is also evidence to suggest that estrogen users develop better differentiated tumors, and that detection/surveillance bias is not the only explanation for better survival.12,14 This implies that hormone treatment accelerates the growth of a malignant locus already in place, and it presents clinically at a less virulent and aggressive stage. This conclusion is consistent with the fact that virtually all the studies find that any increase in risk disappears within five years of discontinuing hormone therapy.

Increased use of mammography by hormone users is a well-recognized phenomenon.15 Indeed, when corrected for use of mammography, an apparent increase in breast cancer in long-term estrogen users in a retrospective cohort study lost its statistical significance.16 A greater frequency of mammography and breast examinations among hormone users introduces detection/surveillance bias into all observational studies. The mortality data support the contention that accelerated growth of a pre-existing tumor and detection/surveillance bias are influencing the results of positive observational studies.

Is It Prevention or Deceleration/Acceleration?

The doubling time of breast cancer is variable, but, in general, a tumor doubles in size every 100 days. Thus, it takes a single malignant cell approximately 10 years to grow to a clinically detectable 1 cm mass.17 Given the length of time it takes for a malignancy to become clinically detectable, how is it possible to demonstrate hormonal effects within several years? Do tamoxifen and raloxifene prevent breast cancer, and does estrogen increase the risk of cancer-or are we only changing the time (age) of diagnosis? Instead of prevention/causation, are we observing deceleration/acceleration of pre-existing tumors? The correct answers await the passage of time. Meanwhile, I offer the following clinical observations:

1. The decision to take tamoxifen to prevent breast cancer is not easy. Many women must be treated, especially women not at high risk, in order to gain a possible effect. The side effects of endometrial cancer and venous thromboem-bolism are serious considerations.

2. Raloxifene should not be viewed as a substitute for estrogen, because raloxifene probably does not match estrogen's broad spectrum of effects.

3. Observational studies lack the ability to overcome recognized and unrecognized biases, unless the studied effect is large. A large effect yields uniformity and consistency of results with case-control and cohort studies (good examples are the benefits of a reduction in the risks of endometrial and ovarian cancer with the use of oral contraception). Therefore, any effect of postmenopausal hormone therapy on the risk of breast cancer is unlikely to be great; otherwise, the observational studies would have achieved uniformity and consistency of results. The results with postmenopausal hormone therapy and the risk of breast cancer indicate either a small effect of estrogen use or the effect of biases that can only be eliminated by large, randomized trials such as the ongoing Women's Health Initiative in the United States (to be completed in 2008) and the WISDOM trial (Women's International Study of Long-Duration Oestrogen use after Menopause) in the United Kingdom (to be completed in 2011).


    1. National Cancer Institute. NCI CANCER TRIALS, 1998.

    2. Draper MW, et al. J Bone Miner Res 1996;11: 835-842.

    3. Boss SM, et al. Am J Obstet Gynecol 1997;177: 1458-1464.

    4. Delmas PD, et al. N Engl J Med 1997;337:1641-1647.

    5. Clarkson TB, et al. J Clin Endocrinol Metab 1998;83: 721-726.

    6. Bjarnason NH, et al. Circulation 1997;96:1964-1969.

    7. Hunt K, et al. Br J Obstet Gynaecol 1990;94:620-635.

    8. Persson I, et al. Int J Cancer 1996;67:327-332.

    9. Willis DB, et al. Cancer Causes Control 1996;7: 449-457.

    10. Hunt K, et al. Br J Obstet Gynaecol 1990;97:1080.

    11. Strickland DM, et al. Obstet Gynecol 1992;80:400.

    12. Bonnier P, et al. Obstet Gynecol 1995;85:11.

    13. Grodstein F, et al. N Engl J Med 1997;336:1769-1775.

    14. Magnusson C, et al. Breast Cancer Res Treat 1996; 38:325-334.

    15. LaCroix AZ, Burke W. Lancet 1997;350:1042-1043.

    16. Ettinger B, et al. Menopause 1997;4:125-129.

    17. Wertheimer MD, et al. JAMA 1986;255:1311.