Postmenopausal Hormone Therapy and the Risk of Breast Cancer


Synopsis: Until answers to this debate are available from the ongoing randomized clinical trials, subjective reviews such as this by Colditz are not helpful for patients or clinicians.

Source: Colditz GA. J Nat Cancer Inst 1998;90:814-823.

Colditz believes that the long-term use of postmenopausal hormone therapy increases the risk of breast cancer, basing his belief upon the following criteria:

· Consistency. Colditz believes that an increased risk of breast cancer is reported consistently with increasing duration of hormone use by postmenopausal women.

· Dose-Response. A dose-response relationship (a relationship that would support causality) is derived in this review from the above conclusion-that duration of exposure is a dose-response relationship.

· Biologic Plausibility. Supporting biologic plausibility are the relationships between breast cancer and estrogen receptors, tamoxifen treatment, reproductive experience, endogenous estrogen levels, increased bone density, and postmenopausal obesity.

· Strength of Association. The association is with age of menopause: the putative increased risk of breast cancer is a predicted magnitude that is comparable to delaying the age of menopause.

· Relationship in Time. Colditz states that the increased risk remains increased five or more years after discontinuing therapy.


Although it seems like I review the topic of breast cancer and postmenopausal hormone therapy frequently, the importance of the issue and the publicity that followed publication of this review motivate me to put my thoughts in print once again.

The strongest support for a relationship between breast cancer risk and postmenopausal hormone therapy is the biologic plausibility. There are problems with each of the other criteria. First and foremost is the fact that in my view, Colditz is guilty of what I call "selective reporting." He concludes that an increased risk with increasing duration of use is consistently reported and bases his first two criteria on that conclusion. He neglects the many reports in which no link with duration was observed.1,2 Another instance of selective reporting is the referral by Colditz to previous reports as supporting his conclusions, and neglecting to inform the reader that these reports were not statistically significant.

A dose-response relationship with actual doses has never been established, but probably never can be because of the small numbers of women using doses other than standard doses. The comparison of risk magnitude to delaying the age of menopause is an attractive exercise of logic. However, the steady exposure to postmenopausal estrogen is not exactly the same as extended exposure to cyclic ovarian function. The relationship with time argument is just plain incorrect. Indeed, even Colditz's own Nurses' Health Study and all other positive studies, including the world reanalysis, found that the risk returns to baseline within five years after discontinuing treatment. Repeatedly in this review by Colditz, cause and promotion are intertwined. Sometimes, Colditz decides estrogen is a promoter and then he decides it is etiologic. He cannot have it both ways.

An issue of major concern is his contention that the risk of breast cancer has been underestimated because estrogen users are at lower risk (by virtue of being leaner, having more hot flushing, and lower bone densities). A critical unanswerable question is whether these differences are sufficient to influence the risk of breast cancer. The Nurses' Health Study has concluded that the health differences in the hormone users are not sufficiently different enough when compared with nonusers to explain the magnitude of protection against coronary heart disease by hormone therapy. Furthermore, any lowering of risk might be balanced by factors in hormone users that increase the risk of breast cancer: more mammograms and breast exams, more alcohol consumption, and higher socio-economic status.

There are so many unanswerable questions that it is impossible to be definitive. Certainly, it is not possible to be as definitive as Colditz. It is more appropriate to assess the entire literature (and not be a selective reporter), and indicate to clinicians and patients that the lack of a definitive answer (the lack of uniformity, consistency, and agreement in more than 50 case-control and cohort studies) suggests that any effect of hormone therapy on the risk of breast cancer is likely to be small.

And, there are other possible explanations for the positive studies. Many studies have documented that breast cancer diagnosed in postmenopausal hormone users is associated with a decreased risk of mortality. This is an apparent paradox: increased risk and decreased mortality. There is an explanation for this apparent paradox. A combination of detection/surveillance bias and accelerated growth of pre-existing tumors leads to earlier diagnosis of less aggressive, less virulent tumors with a better outcome. This is supported by the decreased risk of mortality, the total lack of metastatic disease in the world reanalysis,3 and the return of risk to baseline after discontinuing therapy in the positive studies.

Until answers to this debate are available from the ongoing randomized clinical trials, subjective (selective reporting) reviews such as that by Colditz are not helpful for patients or clinicians. The following is my best effort in producing an objective summary of the literature.

Some epidemiologic case-control and cohort studies conclude that long-term (5 or more years) use of postmenopausal hormone therapy is associated with a slight increase in the risk of breast cancer-a risk that is less than that associated with postmenopausal obesity or daily alcohol consumption. This conclusion might be due to confounding biases, particularly detection/surveillance bias, and accelerated growth of a pre-existing malignancy.

Many observational studies have failed to develop evidence that long-term postmenopausal hormone therapy increases the risk of breast cancer.

All epidemiologic studies fail to find an increased risk of breast cancer associated with short-term (< 5 years) use or past use of postmenopausal hormone therapy.

The epidemiologic data agree that the addition of a progestin to the treatment regimen neither increases nor decreases the risk observed in individual studies.

The epidemiologic data indicate that a positive family history of breast cancer should not be a contraindication to the use of postmenopausal hormone therapy.

Women who develop breast cancer while using postmenopausal hormone therapy have a reduced risk of dying from breast cancer. This is because of two factors: 1) increased surveillance and early detection; and 2) acceleration of preexisting tumor growth so that tumors appear at a less virulent and aggressive stage. (Dr. Speroff is Professor of Obstetrics and Gynecology, Oregon Health Sciences University, Portland, OR.)


    1. Sellers TA, et al. Ann Intern Med 1997;127:973-980.

    2. Newcomb PA, et al. Am J Epidemiol 1995;142: 788-795.

    3. Collaborative Group on Hormonal Factors in Breast Cancer. Lancet 1997;350:1047.