Hormone Replacement Therapy

An understanding of the studies on hormone replacement therapy (HRT) and breast cancer risk is essential to help women make informed decisions about hormone use and prophylactic oophorectomy. To achieve this understanding, they need information about epidemiologic studies, sample size, and statistical and clinical significance. In my practice, much of the information about HRT and breast cancer risk is transmitted in one session, with follow-up discussions during subsequent visits as needed.

First, patients benefit by learning that epidemiologic studies are studies of association, not cause and effect. To avoid being misled by spurious differences between groups, factors such as consistency among studies, strength of association, and reduction of risk with removal of the agent are assessed (for a discussion see reference 1).

In determining the relevance of a relative risk or odds ratio, statistical significance and clinical relevance are considered. Statistical significance means only that the difference between two groups is greater than expected due to chance alone. Clinical relevance is present when the risk is large enough to be important to the individual patient. Differences due to the construction of the study or to confounding variables-variables influencing risk that are present because the investigator is unaware of their influence-must be excluded. For example, if investigators were unaware that cigarette smoking increased the risk of lung cancer, they might be misled by the finding that lung cancer risk increases with the number of matches found in an individual's pocket. Since many risk factors for breast cancer have not yet been identified, confounding variables are likely to be present. For this reason an assessment of confounding variables is most important. Generally speaking, relative risks of 0.5 or less and relative risks of 3.0 or more are required to rule out confounding variables (17).

Most studies on the risk of breast cancer in HRT users have found no statistically significant or clinically relevant increase in breast cancer risk when dose is increased (18,19), when estrogen is used for up to 20 years (19-21), or in those who have a family history of breast cancer or a personal history of benign breast disease compared to nonusers with similar risk factors (20,22-24). When hormone use stops, studies generally find no decrease in breast cancer risk, as would be expected with a causative agent (18,25). In a number of studies, women who were HRT users at the time of their breast cancer diagnosis, and who were therefore using hormones while their breast cancer was present, were found to have a more favorable prognosis than nonusers diagnosed with breast cancer (26-28). Users' breast cancers do not differ significantly in size and differentiations from those of nonusers.

A recent study by Grodstein et al. (29) contained reassuring findings about the use of HRT and breast cancer risk. Current HRT users were less likely to die of breast cancer (relative risk = 0.8) and statistically less likely to die of cardiovascular disease (relative risk = 0.5) (Table 3). With 10 or more years of HRT use, risk of death was significantly less than that of never users (Table 4). When HRT use was discontinued, risk of death increased (Table 5).

Strangely, the study itself and various reports on the study failed to make these reassuring findings clear. Instead, the study pointed to an increase in death rate among longer-term HRT users-an increase from 0.6 to 0.8. With only 163 women with breast cancer in one group and 181 in another, the difference between a relative risk of 0.6 and 0.8 may well be due to chance fluctuations. Even if chance were excluded, long-term HRT users still have a risk of death that is less than that of nonusers (see Table 4).

Grodstein et al. (30) cite a previous study that found '' 43% increase in death due to breast cancer,'' and appear to conclude that the 0.6-0.8 change in the death rate in the 1997 study among longer-term users of HRT is due to breast cancer. Inspection of the earlier study reveals that the 1.43 relative risk to current users was barely statistically significant, with a wider confidence interval than other subgroups, suggesting a smaller sample size that is more subject to random fluctuation. This relative risk was not adjusted for age or seven additional variables used in reporting other risks in the same study, casting further doubt on its validity. Finally, in this study, no increase in breast cancer death rates to past HRT users was present, a finding that is in agreement with at least four other studies that also find no increase in breast cancer death rates among HRT users (26,31-33). In summary, most studies find that HRT use is not associated with a statistically significant or clinically relevant increase in risk of death due to breast cancer or to an increased incidence of this disease.

Supportive Cancer Care: by Ernest H. Rosenbaum, MD & Isadora R. Rosenbaum, MA
www.CancerSupportiveCare.com Modules | Home | Search Our Website | Site Map | Contact Us | Top: www.CancerSupportiveCare.com | Phone : 510-649-8177 | Fax: 510-649-8276

Cause of death	Number	Risk+
All	2,051	0.6#
CHD	289	0.5#
Stroke	91	0.7
All cancer	1,103	0.7#
Breast cancer	246	0.8

Years of Use *	Number	Risk+
Less than 5	215	0.6#
5-9	163	0.6#
10+	181	0.8#

Years since last use	Number	Risk*
Less than 3	173	0.6+
3-4.9	115	0.8+
5 or more	618	1.2+