The most award winning
healthcare information source.
TRUSTED FOR FOUR DECADES.
Abstract & Commentary
Rodriguez and colleagues from the american cancer Society examined the association of postmenopausal estrogen use and ovarian cancer mortality in a prospective cohort study. The American Cancer Society Cancer Prevention Study II enrolled 676,306 postmenopausal women by a baseline questionnaire in 1982. Deaths in this cohort through 1996 accounted for 107,810 (15.9%) of the original group. After exclusions (premenopausal, unavailable information, hysterectomy, ovarian surgery), 211,581 postmenopausal women were left for analysis, with a total of 1497 ovarian cancer deaths. Estrogen use (ever use, past use, current use) was based on responses to the baseline questionnaire. The risk ratio for ovarian cancer mortality was adjusted for age, race, oral contraceptive use, number of live births, body mass index, age of menarche and menopause, and tubal ligation. The statistically significant increased adjusted risk ratios are presented in the Table.
|No. of Deaths||Rate Ratio (similar to Relative Risk)|
|Ever use||255||1.23 (1.06-1.43)|
|>10 years of use||31||2.20 (1.53-3.17)|
These numbers indicated 64.4 ovarian cancer deaths per 100,000 users of estrogen for 10 or more years, compared with 26.4 for never users. Rodriguez et al further concluded that some risk persisted for up to 29 years after discontinuing estrogen. Rodriguez et al considered a possible mechanism for their conclusion, suggesting that ovarian cancer is more affected by lower gonadotropin levels than higher levels (this would not be consistent with the protective effect associated with oral contraceptives, or that estrogen directly stimulates ovarian cellular proliferation). (Rodriguez C, et al. JAMA. 2001;285:1460-1465).
COMMENT BY LEON SPEROFF, MD
There are so many things about this epidemiologic report that remind one of the similar circumstances surrounding the issue of postmenopausal hormone therapy and the risk of breast cancer.
Rodriguez et al argue that one of the reasons the results suggest causality is that the findings are similar to previously published case-control studies.1-3 In fact, one of their references, that I did not cite, is a book chapter; in 2 of their cited studies the conclusions for long-term users were not statistically significant; and in the 1 statistically significant result, the finding applied only to serous carcinomas. I am not impressed with the quality of their reporting. Another example of selective reporting is the statement in the introduction of the current report that breast cancer incidence increases only after long-duration estrogen use, citing the Nurses’ Health Study report in 1995, and neglecting to point out that the results of the American Cancer Society study with these same authors failed to support the Nurses’ Health Study conclusion.4
One case-control study that examined long-term use did not find an increased risk.5 The pooled analysis by Whittemore and colleagues of the 12 case-control studies up to 1992 could find no evidence for an association between ovarian cancer and estrogen therapy.6 A meta-analysis in 1998 concluded that there was a 27% increase in risk of ovarian cancer with more than 10 years of estrogen use, but among the 6 studies included in this analysis, only 1 reported a statistically significant increase in risk with 10 or more years of therapy; and by definition, even the meta-analysis conclusion of a 27% increase in risk did not reach statistical significance (CI = 1.00-1.61).7 In a more recent meta-analyis of 15 case-control studies, in the year 2000, Coughlin and associates could not find an association of estrogen therapy with ovarian cancer and no evidence of an effect with increasing duration of use.8 To be complete, I will add to this list a re-analysis of 4 European case-control studies that found a statistically significant increased risk of ovarian cancer with estrogen use, but responsibly noted that it is essentially impossible to control in observational studies (case-control and cohort) for the possibility that hormone users and never users have different risks for ovarian cancer because they are not identical populations.9
It should also be noted that retrospective analyses have not detected any detrimental effect on prognosis after surgery for ovarian cancer in patients subsequently treated with hormones.10,11
The weakest link in the American Cancer Society Study is the fact that information regarding estrogen use was obtained from the single self-administered questionnaire in 1982. Despite the fact that it is touted as a large prospective study, the conclusion with some strength of association (the increase with users of 10 or more years) was based on 31 cases. But most of all, the results of the study do not represent further information added to a uniform, strong, and consistent story in the literature on this subject. Instead, the subject is similar to that of breast cancer risk and hormone therapy—many negative studies with some positive studies. The positive studies exhibit a strength of association that is not huge and could be due to the problems of small numbers and the attempt to compare 2 groups that may be basically dissimilar.
Epidemiologic evaluation of the effects of postmenopausal combined estrogen-progestin therapy will not be forthcoming for several years because of the relative recency of combined regimens. Pointing out that a combination of estrogen and progestin may prove to protect against ovarian cancer (similar to the results seen with oral contraceptives) is justified, but I would not make that the major part of my response to this current epidemiologic report. I would rather emphasize the small numbers, the weak associations, and the mixed story in the observational studies—all indicating either a very small or no effect.
1. Cramer D, et al. J Natl Cancer Inst. 1983;71:711-716.
2. Kaufman D, et al. Am J Epidemiol. 1989;130: 1442-1151.
3. Risch H. Gynecol Oncol. 1996;63:254-257.
4. Willis DB, et al. Cancer Causes Control. 1996;7:
5. Hempling R, et al. Obstet Gynecol. 1997;89:1012-1016.
6. Whittemore A, et al. Am J Epidemiol. 1992;136:
7. Garg P, et al. Obstet Gynecol. 1998;92:472-479.
8. Coughlin S, et al. J Clin Epidemiol. 2000;53:367-375.
9. Negri E, et al. Int J Cancer. 1999;80:848-851.
10. Eeles R. BMJ. 1991;302:259-262.
11. Ursic-Vrscaj M, et al. Menopause. 2001;8:70-75.