The trusted source for
healthcare information and
Breast Cancer Risk in Finnish Women Using Estrogen-Only Therapy
Abstract & Commentary
By Leon Speroff, MD, Editor, Professor of Obstetrics and Gynecology, Oregon Health and Science University, Portland, is Editor for OB/GYN Clinical Alert.
Synopsis: A cohort study from Finland concludes that long-term users of oral and transdermal estradiol have an increased risk of breast cancer.
Source: Lyytinen H, et al. Breast cancer risk in postmenopausal women using estrogen-only therapy. Obstet Gynecol. 2006;108:1354-1360.
Lyytinen and colleagues from Helsinki reported the results of a cohort study assessing the risk of breast cancer in postmenopausal Finnish women using oral, transdermal, and vaginal estrogen products containing either estradiol or estriol.1 The women were identified as those who purchased any type of estrogen in 1994-2001. Women who used conjugated equine estrogens (387) and women who used estrogen for less than 6 months (172,309) were excluded, leaving 110,984 women in the final cohort. Breast cancer was recorded to the end of 2002. The use of estradiol for 5 or more years was associated with an increased risk of 1.44 (CI = 1.29-1.59). The risk was similar comparing oral and transdermal estradiol therapy. An increase was observed in both localized and metastatic disease. A statistically significant increase was noted with carcinoma-in-situ, 2.43 (CI = 1.66-3.42).
Only 7% of Finnish postmenopausal women use hormone therapy for more than 5 years. This study reports an increase in breast cancer risk in this long-term user group of women. No increase in breast cancer risk was detected either in association with estriol given orally or with vaginal estrogen products. It is inappropriate to conclude, as the authors do, that these formulations can be used without risk. To make this conclusion, users and nonusers of these formulations would have to be identical in terms of breast cancer risk factors, and to be comparable in terms of bioequivalent blood levels of estrogen. Only then could a valid comparison be made. This study cannot and does not adjust for these factors.
The use of estradiol patches was associated with a statistically significant increase with doses of 30 to 60 µg per day, used for 5 years or more. However, neither lower doses nor higher doses demonstrated a statistically significant change. Because there were only 599 lower-dose users and 611 higher-dose users compared with 6845 using the standard dose, the power was not sufficient to reveal a dose-response relationship. The authors interpreted their findings as indicating that both oral and transdermal routes of administration shared an increased risk of breast cancer in long-term users. There is a major strength in this study. The use of postmenopausal hormone therapy in Finland can be accurately recorded because all treatments must be prescribed and then paid for by the National Social Insurance Institution. However, the study is affected by an overwhelming problem: the results are questionable because of an inability to control for confounders.
This is the first epidemiologic study, to my knowledge, to report a statistically significant increase in the risk of in-situ breast cancers. But before we accept that conclusion, consider that there were only 141 in-situ cancers in the cohort, and the increased ratio of 2.43 (CI=1.66-3.42) reported for use of 5 years or more was based on only 13 cases. The authors themselves caution us that hormone users visited physicians at a greater rate and more regularly, and thus, the increase in in-situ disease may reflect a detection bias. In this paragraph, they raise the real possibility of confounders in the hormone users, but then they argue to the contrary in another paragraph. You can't have it both ways.
The major problem with this study is that the risk was expressed as incidence ratios, calculated by dividing the observed number of cases by the numbers expected (based on the general statistics in Finland). Therefore, the study could not be controlled for confounders. It is well demonstrated that hormone users differ when compared to non-users in terms of recognized risk factors for breast cancer. The differences include a greater prevalence of mammography among hormone users. A good example can be found in the report from the Nurses' Health Study that, like the cohort from Finland, indicated an increased risk of breast cancer with long-term users of estrogen-only.2 The long-term users in the Nurses' Health Study had more bilateral salpingo-oophorectomies, more nulliparity, more benign breast disease, greater alcohol consumption, and they were thinner—all factors that make a comparison of users to nonusers very difficult.
The authors of the Finnish report argue that "there are no socioeconomic differences between postmenopausal hormone therapy users and the general population in Finland," citing a previous report. It is a bit mind boggling that this citation is not totally accurate. The report in 1999 was based on population surveys and measured only two things: length of education and rural vs. urban living.3 The authors concluded that a lack of socioeconomic differences was present in Finnish women under the age of 55, but older postmenopausal had more years of education. In addition, there were regional differences at all ages, with the current use of hormone therapy being most common in the Helsinki area (especially among older women). Therefore, the 1999 study does not imply a lack of differences in hormone users in Finland; in fact, just the opposite. Age information is not provided in the current report, but I would expect the longer-term hormone users to be an older group of women, and according to the 1999 Finnish report, they do differ when compared to the general population of Finland. Remember, this cohort study is not comparing users with nonusers. It compares users to general population statistics. Therefore, we cannot know whether the results of this study reflect long-term use of estrogen, or whether the results of this study reflect a greater prevalence of risk factors and mammography in the hormone-using group.
The accompanying editorial is written by a distinguished statistician.4 He reviews the results in the Women's Health Initiative (WHI) and, to my surprise and disappointment, he offers no criticisms of the current report from Finland. He concludes that "longer use of combined estrogen-progestin therapy undeniably increases" breast cancer risk "likely to a greater extent than exposure to estrogen alone." He further argues that estrogen-progestin therapy "causes" (his word) more breast cancers than the number of endometrial cancers prevented.
In the latest report from the estrogen-progestin arm of the WHI, the overall risk of breast cancer in the treated estrogen-progestin group was the same as previously reported by the WHI (1.24; CI=1.02-1.50).5,6 However, after adjusting for the multiple factors recognized to influence the risk of breast cancer, the hazard ratio was 1.20, and no longer statistically significant (CI=0.94-1.53). The WHI results are not consistent with a large effect, and the results are finding it hard to escape the influence of differences in risk factors and personal characteristics. This further emphasizes the weakness in the current Finnish cohort study: the inability to control for confounding risk factors. Thus, in my view, the article and the editorial both overstate the case, and we still don't know whether hormone therapy is associated with a small risk of breast cancer or whether the epidemiologic data reflect an impact on pre-existing tumors.