The role of estrogen in the promotion of breast cancer is difficult to quantitate. Isolated measurement of serum estrogens may not reflect long-term exposure. Studies linking estrogen replacement therapy to breast cancer are subject to bias, since most of them have been case-control studies.
Bone density in women is related, in part, to estrogen exposure. Thus, if long-term estrogen exposure increases bone density and also increases the risk of breast cancer, there should be a positive correlation between bone density and the incidence of breast cancer. This study was designed to look at this issue, using data from the Framingham study.
The Framingham study is a population-based study begun in 1948 in Framingham, Massachusetts. The original cohort included 2873 women 28 to 62 years old. Subjects were evaluated clinically approximately every two years.
Between 1967 and 1970, 1760 women were seen at biennial visits. Of these 1760 women, 1394 underwent posteroanterior hand radiography. The second metacarpal bone was measured for cortical external width (R) and medullary width (r) and the relative metacarpal cortical area was calculated as an indicator of bone mass (100 x (R2 - r2) / R2).
Women were stratified into 2-year age groups and each woman was assigned to one of four quartiles of bone mass, depending on her metacarpal cortical area and the distribution of metacarpal cortical areas in that age group.
Breast cancer cases were sought by interview at the biennial examinations, by telephone and mail contact for those who missed examinations, by review of the local hospital records, death records and the National Death Index. Pathology reports were available for all identified cases.
Other risk factors for breast cancer were analyzed, including: age at time of x-ray, years of education, height, weight, parity, age at first pregnancy, age at menopause, years of post-menopausal estrogen use (after the time of the hand x-ray), cigarette consumption, alcohol consumption and physical activity level.
Of the 1394 women who had x-rays, 21 already had a history of breast cancer and were not analyzed. Among the remaining 1373 women, breast cancer developed in 91, at a median age of 74 and at a median of 22 years after the x-ray.
Comparing age-specific quartiles of bone density 1 through 4 (increasing bone density), there were 12 women with breast cancer in quartile 1, 17 in quartile 2, 18 in quartile 3 and 44 in quartile 4.
Comparing quartile 4 to quartile 1, the incidence rate was 7.03 cases/1000 person-years vs. 2.02 cases/1000 person-years. The rate ratio, quartile 4 vs. quartile 1 was 3.5, both before and after multivariate adjustment for the major risk factors for breast cancer. This significant increase in the risk of breast cancer in the highest quartile of bone density (but not in quartiles 2 and 3) was clearly seen in the Kaplan-Meier survival curves plotting the cumulative incidence rate of breast cancer over time.
Comparing women with and without breast cancer, the only analyzed risk factor that was significantly different in this study was age at first pregnancy (27.5 in women with breast cancer, vs. 25.9; p=0.02).
Women in the higher quartiles of bone density had a higher body-mass index, were older at menopause, had achieved a higher level of education and used post-menopausal hormone replacement therapy for a longer time.
Looking at each of the 11 risk factors that were studied, a high bone mass remained a significant risk factor for breast cancer in all of the subgroups. For example, the relative risk of breast cancer in quartile 4 (highest bone mass) compared to quartile 1 was:
The authors conclude that bone mass in middle aged and elderly women (measured here by metacarpal cortical thickness) is a strong predictor of postmenopausal breast cancer. They speculate that cumulative exposure to circulating estrogens may be the causal link, since estrogen is known to favorably influence bone mass and may play a role in the development of breast cancer (exposure to circulating androgens could also play a similar role).
They note that some, but not all, studies examining the relationship between serum estrogen levels and breast cancer found a positive association; similarly for studies of hormone replacement therapy and breast cancer.
Two studies have shown an inverse relation between risk of fractures and breast cancer, and a recent study examining the relation between bone density and breast cancer found a similar effect.
The relative risk associated with the highest bone density quartile was surprisingly high (3.5). In studies looking at the effect of estrogen replacement therapy, the relative risk associated with current estrogen use is usually around 1.5 or less.
Although this study is unlikely to have a significant impact on clinical practice, it does provide another piece of evidence linking higher estrogen exposure to breast cancer. It also raises a number of questions which may ultimately be of clinical importance. In particular, is bone density a particularly sensitive marker for estrogen effect? Are women with higher bone densities at higher risk for estrogen-therapy related cancer? Are women with lower bone densities relatively immune from this problem?
If the theory that the highest bone density patients have circulating estrogens is correct, then perhaps these patients should be studied for the incidence of ischemic heart disease. The model would predict a lower rate of coronary artery disease provided other risk factors such as smoking and diabetes are held constant.
Further study might also determine whether ERT in patients with high bone density is contraindicated.
Date: Sat, 19 Apr 1997
From: firstname.lastname@example.org (David K Winsemius)
The careful generalist clinician must guard against focusing on a single disease as the pivotal factor in decisions concerning hormone replacement therapy (HRT). The potential for decreased coronary artery disease, increased breast cancer, and decreased osteoporosis must be weighed with appropriate attention to all relevant risk factors and patient values. In most women the decison to use HRT will be life-prolonging, even after considering the potential for increasing breast cancer risk. Cardiac disease remains a higher risk than breast cancer.
It is also important to consider quality of life in making the decision with your patient. Many women report important improvements in sleep pattern, emotional state, and sexual functioning while using post-menopausal HRT. There are probably a minority of women who will have a net decrease in life-expectancy if HRT is chosen. Concern for their safety should not blind physicians to the important cardiac risks of the majority of women who would derive a net longevity benefit from estrogen replacement.
It also seems possible that the link between bone density and breast cancer demonstrated in this study identifies a genetic-cellular trait that is more important than the use of physiologic estrogen replacement, i.e., they are more likely to get breast cancer regardless of whether HRT is or was used. The article indicated that the link persisted even after controllong for HRT. Some women's estrogen receptors in bone may be more sensitive or active reflecting similar activity in breast tissue. We can hope that tailored estrogens could then be formulated that would spare the breast receptors while delivering cardioprotection and desirable neurocognitive benefits.
Follow this link to obtain citations documenting the cardioprotective effects of estrogens: Hormone replacement therapy and cardiovascular disease.
You may also want to look at a the lead article in this week's issue of the American Journal of Cardiology that documents an impressive mortality difference for women who are estrogen users following their bypass operations. The difference is huge. Makes you wonder if the well-described difference in survival benefits between men and women from cardiac procedures represents a failure to appropriately use estrogens.
David Winsemius, MD, MPH
Date: Tue, 29 Apr 1997
From: Mario Ruiz <email@example.com>
I have just a few questions regarding this study:
1. Does the fact that these women were initially assessed from 1967-1970 and thus were at a higher dose of estrogen (most likely) present a confounding variable? It could be that a cumulative dose of estrogen is important here. Thus at a lower dose we may in fact preferentially treat for osteoporosis and coronary artery disease without necessarily increasing the risk of breast cancer.
2. If in fact these ladies were exposed to higher doses of estrogen they should have a much higher incidence of endometrial cancer possibly (depending on how many had hysterectomies).
3. This study failed to control for 2 major risk factors. Mainly, a family history and age of menarche. They attempt to brush off the importance of a family history with one paper (unfortunately, I have not had a chance to look this one over just yet)...but even so it does present, I believe, a major confounding variable to this study.
What are your thoughts?
The definition of HRT use in this study does not include women who had used estrogens prior to the bone density determination. Thus, we cannot totally exclude exogenous estrogen exposure as the main cause of the linkage between bone density and breast cancer found in this study. But, again, distinguishing between exogenous and endogenous estrogen exposure was not the primary focus of this paper.
2. If higher bone density is a marker for increased estrogen effect, then we might indeed expect a correlation between bone density and the incidence of endometrial cancer.
3. Although the authors did not control for age at menarche and family history of breast cancer, and both of these are related to the risk of breast cancer, it is hard to see how these two factors could also be related to bone density. Such a relation would be necessary for them to act as confounding variables.