The trusted source for
healthcare information and
Postmenopausal Hormone Therapy and Ovarian Cancer
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: The Million Women Study reported a small increase in the risk of ovarian cancer in current users of postmenopausal hormone therapy.
Source: Million Women Study Collaborators. Ovarian cancer and hormone replacement therapy in the Million Women Study. Lancet. (Online). April 19, 2007.
The Million Women Study reported a small increase in the incidence of ovarian cancer and ovarian cancer mortality in current users of postmenopausal hormone therapy.1 Overall the relative risk of ovarian cancer in current users was 1.20 (CI = 1.09-1.32) and in past users 0.98 (CI = 0.88-1.11). There was a trend for increasing risk with increasing duration of use. The increase was observed only with epithelial tumors, and only serous cancers had a significant increase. There were no significant differences with different estrogens or different progestins, or between oral and transdermal administration. Adjustment for various confounding factors detected no influences.
If the results of the Million Women Study are accurate, the relative risks translate into the following absolute risks:
|Incidence||Fatal Ovarian Cancer|
|U.K. population||2.2/1000 women/5 years||1.3/1000 women/5 years|
All tables adapted by L. Speroff
There have been over 20 case-control and cohort studies assessing the relationship between postmenopausal hormone therapy and the risk of ovarian cancer. The relative risks have encompassed a wide range from below 1.0 to greater than 1.0. Because this is a relatively infrequent cancer, all studies have been hampered by relatively small numbers.
The canceled estrogen-progestin arm of the Women's Health Initiative reported an increase in ovarian cancer that was not statistically significant (Hazard ratio = 1.58, CI = 0.77-3.24), prompting the authors to say: "The possibility of an increased risk of ovarian cancer incidence and mortality remains worrisome and needs confirmation."2 The Kaplan-Meier curves suggested an increasing effect over time, but this, too, was not statistically significant. There were no differences reported in histologic type, stage, or grade (but the small numbers made it essentially impossible to assess subcategories).
The studies have found it difficult to control for all of the factors that influence the risk of ovarian cancer. This is because there are multiple factors and information regarding each factor is not readily available.
Factors That Decrease the Risk of Ovarian Cancer
Factors That Increase the Risk of Ovarian Cancer
Mixed Reports on Decreased Risk
Because of the many factors that influence the risk of ovarian cancer, it has been essentially impossible in case-control and cohort studies to match cases and controls. Hormone users typically have used more oral contraceptives, have had fewer children, and are more educated and thinner. Adjustments have been made only for major factors, such as oral contraceptive use. The technique of meta-analysis is especially hampered by these confounding issues. The authors of the published meta-analyses17-19 have inappropriately assumed that controlling for risk factors was uniformly accomplished in all studies.
The Million Women Study is a cohort of 948,576 women who have provided information through a baseline questionnaire and a follow-up questionnaire about 3 years later (64% responded to the request for follow-up information; thus, the accuracy of the data can legitimately be questioned). The current report, however, is impressive in the relatively large number of cases (obtained by assessing the National Health Service Central Registers in the U.K.) and the multiple adjustments for confounding factors. The investigators adjusted their analyses for the following:
But remember this information was derived from 2 questionnaires, and 36% of the women did not return the follow-up questionnaire. When relative risks are in the range as reported from this Million Women Study, a shift of a small number of cases can change the statistical conclusions. And although this report covers more confounding factors than any other previous study, nevertheless, some influencing factors remained unadjusted.
A major problem in previous studies has been the impact of endometrioid cancers, an ovarian cancer that logically can be expected to be influenced by estrogen therapy. In many of the studies, the overall results are swayed by the increase in endometrioid cancers, a cancer that could originate in hormonally-stimulated endometriosis. An accurate analysis requires a separate consideration of endometrioid cancers, but this is difficult because the small numbers do not allow effective sub-categorization. The number of endometrioid cancers in the Million Women Study totaled 72 cases in current users and 114 in never users for a relative risk of 1.05 (CI = 0.77-1.43). An important finding in the Million Women Study, if it is accurate, is that the increase was in serous cancers, not endometrioid cancers.
Why might postmenopausal hormone therapy increase the risk of ovarian cancer whereas steroid contraception reduces the risk? One can only speculate, but the most obvious explanation is inhibition of multiple ovulations (which increase the possibility of harmful mutations), an effect only present in premenopausal ovaries.
The authors present in their discussion an appropriate warning that comparing various reports on this subject is difficult because of differing populations, small numbers, and variances in classifications and adjustments. This doesn't stop them, however, from adding a meta-analysis of 9 studies (theirs included) with an overall increased relative risk. They further dramatize their report by producing a figure that combines ovarian, endometrial, and breast cancer. Of course the impact of hormone therapy is impressive because of the much greater prevalence of breast cancer. But a similarity between ovarian cancer and breast cancer is apparent in the data from the Million Women Study. The risk of the two cancers is reported to be increased only in current users. The risk returns to that of never users rapidly after discontinuation. Women in the Million Women Study who developed ovarian cancer were diagnosed from 0.8 to 4 years after an average use of hormone therapy for 7.7 years was last reported. As with breast cancer, the data suggest that hormone therapy is influencing pre-existing tumors. In other words, it is not certain these are new cancers. Cancers originate in mutations, a point of origin not known to be influenced by hormones.