Effects of Estrogen Plus Progestin on Gynecologic Cancers and Associated Diagnostic Procedures the Women’S Health Initiative Randomized Trial

ORIGINAL CONTRIBUTION

Effects of Estrogen Plus Progestin on Gynecologic Cancers and Associated Diagnostic Procedures The Women’s Health Initiative Randomized Trial

Garnet L. Anderson, PhD Context The effects of continuous combined hormone therapy on gynecologic can- Howard L. Judd, MD cers have not been investigated previously in a randomized trial setting. Andrew M. Kaunitz, MD Objective To determine the possible associations of estrogen plus progestin on gynecologic cancers and related diagnostic procedures. David H. Barad, MD, MS Design, Setting, and Participants Randomized, double-blind, placebo- Shirley A. A. Beresford, PhD controlled trial of 16608 postmenopausal women, who had not had a hysterectomy Mary Pettinger, MS at baseline and who had been recruited from 40 US clinical centers between Septem- ber 1993 and October 1998 (average follow-up, 5.6 years). James Liu, MD Intervention One tablet per day containing 0.625 mg of conjugated equine estro- S. Gene McNeeley, MD gens plus 2.5 mg of medroxyprogesterone acetate (n=8506) or placebo (n=8102). Ana Maria Lopez, MD Main Outcome Measure Incident invasive cancer of the ovary and endometrium. for the Women’s Health Initiative Results In 5.6 years of follow-up, there were 32 cases of invasive ovarian cancer, 58 Investigators cases of endometrial cancer, 1 case of nonendometrial uterine cancer, 13 cases of cervical cancer, and 7 cases of other gynecologic cancers. The hazard ratio (HR) for inva- OR YEARS, THERE HAS BEEN CON- sive ovarian cancer in women assigned to estrogen plus progestin compared with pla- cern about possible associations cebo was 1.58 (95% confidence interval [CI], 0.77-3.24). The HR for endometrial cancer of gynecologic malignancies with was 0.81 (95% CI, 0.48-1.36). No appreciable differences were found in the distribu- postmenopausal hormone tions of tumor histology, stage, or grade for either cancer site. The incidence of other therapy.F The development of endome- gynecologic cancers was low and did not differ by randomization assignment. More women trial hyperplasia and endometrial can- taking estrogen plus progestin required endometrial biopsies (33% vs 6%; PϽ.001). cer with unopposed estrogen is well Conclusions This randomized trial suggests that continuous combined estrogen plus recognized. To reduce or avoid this com- progestin therapy may increase the risk of ovarian cancer while producing endome- plication, progestin has been added,1-3 trial cancer rates similar to placebo. The increased burden of endometrial biopsies re- although results from randomized trials quired to assess vaginal bleeding further limits the acceptability of this regimen. These are extremely limited. These concerns data provide additional support for caution in the use of continuous combined hormones. have created a need for reasonable moni- JAMA. 2003;290:1739-1748 www.jama.com toring guidelines to follow-up women who experience vaginal bleeding while controlled setting. The trial was stopped on the basis of an increased risk of breast taking estrogen plus progestin. early at the recommendation of the inde- cancer supported by a summary mea- The Women’s Health Initiative (WHI) pendent data and safety monitoring board sure of effects indicating risks exceeded trial of estrogen plus progestin provides Author Affiliations: Division of Public Health Sci- Western Reserve University, Cleveland, Ohio (Dr the first opportunity to examine pos- ences, Fred Hutchinson Cancer Research Center, Liu); Department of Obstetrics and Gynecology, Seattle, Wash (Drs Anderson and Beresford and Ms Wayne State University, Detroit, Mich (Dr sible associations of gynecologic malig- Pettinger); Department of Epidemiology, University McNeeley); and Department of Internal Medicine, nancies with continuous combined post- of Washington, Seattle (Dr Beresford); Department University of Arizona Cancer Center, Tucson (Dr menopausal hormone therapy in a large, of Obstetrics and Gynecology, University of Califor- Lopez). nia, Los Angeles (Dr Judd); Department of Obstetrics WHI Investigators, Centers, and Financial Disclo- randomized, double-blind, placebo- and Gynecology, University of Florida Health Science sures are listed at the end of this article. Center, Jacksonville (Dr Kaunitz); Department of Corresponding Author and Reprints: Garnet L. Ander- Obstetrics and Gynecology and Women’s Health, son, PhD, Fred Hutchinson Cancer Research Center, See also p 1729. Albert Einstein College of Medicine, Bronx, NY (Dr 1100 Fairview Ave N, MP-1002, Seattle, WA 98109 Barad); Department of Reproductive Biology, Case (e-mail: [email protected]).

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013 GYNECOLOGIC CANCERS IN THE WHI

benefits over an average of 5.2 years of gen plus progestin trial prior to random- discontinuation of study pills unless fur- follow-up. The initial report provided ization. A 5% cohort of women was ran- ther evaluation ruled out malignancy. interim, locally adjudicated outcomes for domly selected at enrollment to undergo Cervical Cytology. In all partici- endometrial cancer reported through routine biopsies at follow-up years 3, 6, pants, Papanicolaou tests were per- April 30, 2002. This article summarizes and 9. This cohort was intended to pro- formed at 3-year intervals in conjunc- centrally coded gynecologic cancer out- vide a valid comparison of the rates of tion with routine pelvic examinations. comes and related diagnostic proce- endometrial pathological findings. Absent an a priori hypothesis of asso- dures occurring prior to the announce- The WHI procedures called for endo- ciation between estrogen plus proges- ment of the trial closure and participant metrial biopsies to be performed by WHI tin and cervical cancer, these tests were unblinding (July 8, 2002). trained and certified staff, who were li- performed by WHI staff as a courtesy, censed physicians, nurse practitioners, in conjunction with a required pelvic METHODS or physician assistants, using plastic en- examination. Pathological analysis was Study Population, Randomization, dometrial suction curettes.6,7 Readings obtained locally. During follow-up, the and Intervention were obtained from local pathologists protocol was modified to accept Papa- Women were recruited at 40 clinical cen- blinded to randomization assignment. nicolaou test results from a partici- ters in the United States between Sep- During follow-up, women with per- pant’s health care clinician. tember 1993 and October 1998, largely sistent or heavy bleeding were evalu- Abnormal test results were referred to through direct mail. Eligibility required ated by the clinic gynecologist. If a non- the participant’s health care clinician for women to be between age 50 and 79 routine biopsy was indicated, study further diagnostic evaluation and treat- years, to be postmenopausal, and to pro- guidelines permitted the clinic gynecolo- ment, except when a change in use of vide written informed consent. Women gist to then be unblinded to assist in fur- study medicines was required. were excluded if they had preexisting ther safety evaluation. Women selected conditions that contraindicated use of for routine biopsies who had a biopsy in Outcomes hormones, had health conditions that the last 12 months to evaluate bleeding Outcome ascertainment procedures have suggested a predicted survival of less than problems were not required to repeat the been described.5 Briefly, semiannual self- 3 years, or were considered likely to be procedure. When biopsies could not be reports of new diagnoses were re- poor adherers to the study protocol. Only accomplished, vaginal ultrasounds of the corded and all associated medical re- women who had not had a hysterec- endometrium were performed. cords were obtained from local health tomy were considered for this trial. Eli- At baseline, biopsy evidence of endo- care clinicians and classified by blinded gible women were randomized in equal metrial cancer, complex or adenoma- physician adjudicators at each clinical proportions using a stratified permuted tous hyperplasia, or atypia disqualified center. The documents from all gyneco- block algorithm to either placebo or to women from participating, whereas find- logic cancer cases were forwarded to the 0.625 mg/d of conjugated equine estro- ings of simple hyperplasia resulted in WHI clinical coordinating center for cen- gens plus 2.5 mg/d of medroxyproges- temporary exclusion pending resolu- tralized review by cancer coding special- terone acetate, which was administered tion. During study follow-up, evidence ists who were also blinded to random- in a single tablet (Prempro, Wyeth, St of endometrial cancer, complex or ad- ization assignment and reported Davids, Pa). Women who had a prior enomatous hyperplasia, or atypia re- symptoms. Histological codes were based hysterectomy were randomized to a par- quired permanent discontinuation of on the International Classification of Dis- allel trial of estrogen alone and are not study medicines. Simple hyperplasia eases for Oncology, 2nd Edition.9 Histo- included in these results. Study design identified during follow-up led to un- logical analysis was available from pa- details have been published.4,5 blinding of the consulting gynecolo- thology reports for all ovarian and gist. Those participants assigned to endometrial cancers. Stage and grade Data Collection placebo had study medications discon- were coded using Surveillance, Epide- All WHI participants provided demo- tinued and were referred to their health miology, and End Results guidelines.10 graphic, medical, reproductive, and care clinicians for further management. In this system, local-stage ovarian can- family history information using self- Those assigned to active therapy contin- cer refers to a tumor confined to the administered questionnaires at base- ued study medications supplemented ovary with no tumor on the ovarian sur- line. Prior postmenopausal hormone with 20 mg/d of medroxyprogesterone face. A regional stage ovarian cancer rep- use was ascertained through a struc- acetate. In such women, the biopsy was resents a tumor that is present on the tured interview asking women to de- repeated in 3 to 6 months. ovarian surface, or has evidence of spread scribe the strength, schedule, and du- An endometrial wall thickness of to ascites, peritoneal washings, or other ration of each hormone preparation. greater than 0.5 cm on vaginal ultra- locations within the pelvis. Tumors that Endometrial Evaluation. For safety, sound was considered evidence of an spread beyond the pelvis are consid- endometrial biopsies were performed on endometrial pathological finding8 and ered distant stage. For endometrial every woman interested in the estro- resulted in exclusion (at baseline) or cancers, local stage refers to cancers

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013 GYNECOLOGIC CANCERS IN THE WHI

confined to the endometrium or myo- Table 1. Baseline Risk Factors for Gynecologic Cancers by Randomization Assignment metrium/serosa. Regional-stage tumors No. (%) of Women have spread only to the pelvis, vagina, and/or the wall of the rectum or blad- Estrogen + Progestin Placebo der. Information could not be obtained (n = 8506) (n = 8102) Age at screening, y for ovarian cancer stage in 1 woman and 50-59 2839 (33.4) 2683 (33.1) ovarian cancer grade in 9 women. Be- 60-69 3853 (45.3) 3657 (45.1) cause the number of primary peritoneal 70-79 1814 (21.3) 1762 (21.8) and fallopian tube cancers was small, Ethnicity these are reported together as other gy- White 7140 (83.9) 6805 (84.0) necologic cancers. Black 549 (6.5) 575 (7.1) Hispanic 472 (5.5) 416 (5.1) Statistical Analyses American Indian 26 (0.3) 30 (0.4) Tests for differences between groups in Asian/Pacific Islander 194 (2.3) 169 (2.1) the distribution of pathological fea- Unknown 125 (1.5) 107 (1.3) tures of diseases, and rates and results Body mass index, kg/m2 Ͻ of endometrial biopsies and Papanico- 25 2579 (30.3) 2479 (30.6) laou tests are based on Fisher exact 25-29 2992 (35.2) 2834 (35.0) Ն tests. When a diagnostic procedure was 30 2899 (34.1) 2737 (33.8) performed multiple times for a woman, Unknown 36 (0.4) 52 (0.6) Smoking her most severe result is reported. Never 4178 (49.1) 3999 (49.3) Cancer incidence rate comparisons are Past 3362 (39.5) 3157 (39.0) presented as hazard ratios (HRs) and Current 880 (10.3) 838 (10.3) 95% confidence intervals (CIs) from Cox Unknown 86 (1.0) 108 (1.3) proportional hazards models, stratified Hypertension* by age and randomization to the WHI Never 5332 (62.7) 5277 (65.1) dietary trial. Kaplan-Meier estimates of Not treated 635 (7.5) 631 (7.8) cumulative hazards are shown as 1 mi- Treated 1640 (19.3) 1631 (20.1) nus the disease-specific failure time es- Unknown 899 (10.6) 563 (6.9) timates. For the primary results, ad- History of ovarian cancer justed 95% CIs are corrected for the 7 No 8428 (99.1) 8034 (99.2) Ͻ outcomes that contributed to the global Yes 6 (0.1) 2 ( 0.1) index, which is consistent with our ini- Unknown 72 (0.8) 66 (0.8) 5 Female relatives with breast cancer tial report. This adjustment may not None 6770 (79.6) 6486 (80.1) completely account for the variability in 1 927 (10.9) 816 (10.1) these estimates, given the multiple out- 2 75 (0.9) 69 (0.9) comes that will be examined in this trial. Ն3 7 (0.1) 10 (0.1) No adjustments were made for mul- Unknown 727 (8.5) 721 (8.9) tiple tests over time because these out- Female relatives with ovarian cancer comes make only a minor contribution None 7704 (90.6) 7332 (90.5) to early stopping considerations under Ն1 186 (2.2) 172 (2.1) WHI monitoring guidelines.11 Unknown 616 (7.2) 598 (7.4) The primary analyses are based on the First-degree relative with breast/ovarian cancer intent-to-treat principle. In the original None 6937 (81.6) 6679 (82.4) design, women who had not had a hys- 1 1054 (12.4) 949 (11.7) terectomy were randomized to either 2 101 (1.2) 83 (1.0) unopposed estrogen, estrogen plus pro- Ն3 11 (0.1) 15 (0.2) gestin, or placebo. After information from Unknown 403 (4.7) 376 (4.6) the Postmenopausal Estrogen/Proges- Relatives with colorectal cancer tin Interventions (PEPI) trial12 indi- None 6631 (78.0) 6183 (76.3) cated that it was not feasible to include 1 832 (9.8) 891 (11.0) an unopposed estrogen group in a long- 2 88 (1.0) 105 (1.3) term prevention study in women who Ն3 18 (0.2) 23 (0.3) had not had a hysterectomy, the estro- Unknown 937 (11.0) 900 (11.1) gen group was closed and the 331 women (continued)

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013 GYNECOLOGIC CANCERS IN THE WHI

previously randomized to unopposed Table 1. Baseline Risk Factors for Gynecologic Cancers by Randomization Assignment (cont) estrogen were unblinded and switched No. (%) of Women to combined hormone therapy. Mean Estrogen + Progestin Placebo (SD) duration of exposure to unop- (n = 8506) (n = 8102) posed estrogen in these women was 5.4 Age at menopause, y Ͻ40 195 (2.3) 189 (2.3) (3.1) months. When appropriate, analy- 40-44 677 (8.0) 632 (7.8) ses were performed separating out the 45-49 1943 (22.8) 1996 (24.6) women randomized during this phase to 50-54 3629 (42.7) 3506 (43.3) account for the exposure to unopposed Ն55 1235 (14.5) 1186 (14.6) estrogen. Several other sensitivity analy- Unknown 827 (9.7) 593 (7.3) ses were conducted. In per protocol Age at menarche, y analyses, events and follow-up time Յ11 1725 (20.3) 1670 (20.6) occurring more than 6 months after the 12 2141 (25.2) 2095 (25.9) 13 2437 (28.7) 2239 (27.6) participant became nonadherent to study Ն14 2182 (25.7) 2061 (25.4) medicines or initiated use of nonstudy Unknown 21 (0.2) 37 (0.5) hormones were censored. This analysis Parity preserves the randomization assign- Never pregnant 856 (10.1) 832 (10.3) ment and attempts to limit the dilution 1 690 (8.1) 661 (8.2) of effects that nonadherence may entail. 2 1908 (22.4) 1708 (21.1) As treated analyses (attributing events to 3 2020 (23.7) 1952 (24.1) the woman’s use of hormones, either 4 1416 (16.6) 1412 (17.4) study pills or those provided by her health Ն5 1575 (18.5) 1500 (18.5) care clinician, 6 months prior to the Unknown 41 (0.5) 37 (0.5) Oral contraceptive use event) were also performed. Analyses Never 4811 (56.6) 4655 (57.5) censoring women at the time of surgical Ever 3695 (43.4) 3447 (42.5) removal of the organs of interest (total Duration, y hysterectomy for endometrial cancer and Ͻ 5 1982 (23.3) 1781 (22.0) bilateral oophorectomy for ovarian can- Ͻ 5to 10 825 (9.7) 808 (10.0) cer) were also conducted. 10 to Ͻ15 598 (7.0) 585 (7.2) Potential effect modification with Ն15 288 (3.4) 270 (3.3) Unknown 2 (Ͻ0.1) 3 (Ͻ0.1) known gynecologic cancer risk fac- Recency of use tors was assessed in expanded propor- Ͻ20 y ago 748 (8.8) 705 (8.7) tional hazards models that included the 20 to Ͻ30 y ago 2133 (25.1) 2013 (24.8) designated risk factor and randomiza- Ն30 y ago 807 (9.5) 718 (8.9) tion assignment as main effects and the Unknown 7 (0.1) 11 (0.1) interaction between these. Partici- Infertility† No 7123 (83.7) 6897 (85.1) pants with missing values were ex- Yes 1301 (15.3) 1145 (14.1) cluded only from the analyses using the Unknown 82 (1.0) 60 (0.7) relevant variables. Oopherectomy status Throughout this article, we report None 8083 (95.0) 7705 (95.1) unadjusted 2-sided P values to indicate Partial 347 (4.1) 339 (4.2) the relative strength of evidence in these Bilateral 29 (0.3) 24 (0.3) secondary analyses. Seventeen tests for Unknown 47 (0.6) 34 (0.4) interactions with selected baseline char- Unopposed estrogen use Never 7603 (89.4) 7237 (89.3) acteristics were examined and accord- Ever 903 (10.6) 865 (10.7) ingly 1 test would be expected to be sig- Duration, y nificant at the P=.05 level by chance Ͻ5 677 (8.0) 659 (8.1) alone. All analyses were performed using 5toϽ10 134 (1.6) 109 (1.4) SAS statistical software (version 8.2, SAS Ն10 92 (1.0) 97 (1.2) Institute Inc, Cary, NC). Recency of use Within past 5 y 204 (2.4) 194 (2.4) RESULTS 5toϽ10 y ago 141 (1.7) 154 (1.9) Ն10 y ago 555 (6.5) 513 (6.3) Many subject characteristics have been 5 Unknown 3 (Ͻ0.1) 4 (Ͻ0.1) presented. Use of unopposed estro- (continued) gen prior to enrollment was limited

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013 GYNECOLOGIC CANCERS IN THE WHI

(11%), was generally of short dura- for these tests was limited because the years.13 A small, nonsignificant reduction (Ͻ5 years), and had not taken place number of events was small. None of tion in endometrial cancer risk was recently (TABLE 1). Prior use of com- the women reporting a history of ovar- observed with estrogen plus progestin use bined hormones was more common ian cancer at baseline experienced a new (27 vs 31; HR, 0.81; 95% CI, 0.48-1.36 (18%) and also more recent. diagnosis of ovarian cancer. Ovarian [adjusted 95% CI, 0.40-1.64]). A simi- The average follow-up time for this cancer was the reported cause of death lar reduction was observed in cancers report is 5.6 years. At the time of our in 9 women taking estrogen plus pro- arising from the epithelium (all reported interim report, 42% of women random- gestin and 3 women taking placebo histological classes except stromal sar- ized to estrogen plus progestin and 38% (HR, 2.70; 95% CI, 0.73-10.0). coma and mixed mullerian) (23 vs 30; of women randomized to placebo had The observed incidence rate of endo- HR, 0.71; 95% CI, 0.41-1.22). Kaplan stopped taking their study medica- metrial cancer was 62 per 100000 per- Meier estimates of the cumulative haz- tions.5 A total of 485 deaths (3%) oc- son-years, which was also lower than the ards reveal no differences in rates curred before July 8, 2002 (FIGURE 1). Surveillance, Epidemiology, and End throughout follow-up beyond what could The observed annual incidence rate of Results rate of 83 per 100000 person- be readily explained by chance alone ovarian cancer was 34 per 100000 person-years, somewhat lower than the population-based rate of 45 per 100000 Table 1. Baseline Risk Factors for Gynecologic Cancers by Randomization Assignment (cont) person-years reported by the Surveil- No. (%) of Women lance, Epidemiology, and End Results for Estrogen + Progestin Placebo women of this age distribution.13 The rate (n = 8506) (n = 8102) Estrogen + progestin use in the estrogen plus progestin group was Never 6990 (82.2) 6706 (82.8) elevated (20 vs 12; HR, 1.58; 95% CI, Ever 1516 (17.8) 1396 (17.2) 0.77-3.24 [adjusted 95% CI, 0.59-4.23]), Duration, y but not statistically significant. Limit- Ͻ5 1050 (12.3) 997 (12.3) ing the analyses to invasive epithelial can- 5toϽ10 315 (3.7) 258 (3.2) cers did not change the results substan- Ն10 151 (1.8) 141 (1.7) tially (HR, 1.64; 95% CI, 0.78-3.45). The Recency of use possibility of an increasing effect over Within past 5 y 1129 (13.3) 1019 (12.6) 5toϽ10 y ago 247 (2.9) 225 (2.8) time is suggested by the Kaplan-Meier Ն10 y ago 138 (1.6) 151 (1.9) estimates of cumulative hazards Unknown 2 (Ͻ0.1) 1 (Ͻ0.1) (FIGURE 2A) but likewise did not reach Abnormal Papanicolaou test result in last 3 y statistical significance. Controlling for Never had test 101 (1.2) 115 (1.4) family history of colorectal cancer, the No abnormal result 6748 (79.3) 6674 (82.4) only baseline risk factor exhibiting a no- Abnormal result 341 (4.0) 328 (4.0) ticeable degree of imbalance, produced Unknown 1316 (15.5) 985 (12.2) a modest increase in the estimated effect *Defined as self-report of physician diagnosis. †Defined as unable to conceive after a year of unprotected intercourse. (HR, 2.11; 95% CI, 0.96-4.60). No substantial changes were found in analyses conducted per protocol (HR, 1.51; 95% Figure 1. Profile of the Estrogen Plus Progestin Component of the Women’s Health Initiative CI, 0.64-3.55), as treated (HR, 1.76; 95% as of July 7, 2002 CI, 0.87-3.55), or censoring at the time of bilateral oophorectomy (HR, 1.59; 373 092 Women Initiated Screening 95% CI, 0.78-3.25). 18 845 Provided Consent and Reported No Hysterectomy There was no evidence of a difference between treatment groups in the distribution of histological classes, mor- 16 608 Randomized phological grade, or stage of disease at + diagnosis (TABLE 2). No significant in- 8506 Assigned to Receive Estrogen Progestin 8102 Assigned to Receive Placebo teractions were found with age, race/ Status on July 7, 2002 Status on July 7, 2002 ethnicity, body mass index, family his- 7962 Alive and Outcomes Data Submitted in 7620 Alive and Outcomes Data Submitted in tory of breast or ovarian cancer, family Last 18 mo Last 18 mo 296 Unknown Vital Status 245 Unknown Vital Status history of colorectal cancer, prior use 248 Deceased 237 Deceased of oral contraceptives, prior exposure to unopposed estrogen, or prior use of 8506 Included in Analysis 8102 Included in Analysis combined hormones. However, power

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013 GYNECOLOGIC CANCERS IN THE WHI

(Figure 2B). Three (0.9%) endometrial The numbers of borderline ovarian tu- Among women selected for usual cancers were diagnosed among the 331 mors (1 vs 3) and other gynecologic can- care, the fraction of women taking com- women originally randomized to estro- cers (6 vs 1) were too small to provide bined hormones and requiring diag- gen alone compared with 3 (0.5%) cases meaningful comparisons. Using a com- nostic biopsies increased more than among the 573 women randomized to bined outcome of invasive ovarian, pri- 5-fold over placebo (33% vs 6%; combined hormones and 2 (0.4%) cases mary peritoneal, and fallopian tube can- PϽ.001) and twice as many women re- among the 522 women randomized to cers resulted in an estimated HR that was quired multiple biopsies (38% vs 17%; placebo during this same period. Remov- slightly higher than invasive ovarian can- PϽ.001). Among women having suc- ing these cases had a limited impact on cer alone (26 vs 13; HR, 1.92; 95% CI, cessful biopsies, a higher proportion the results (21 vs 29; HR, 0.72; 95% CI, 0.99-3.74). We note that one primary taking estrogen plus progestin had nor- 0.41-1.26). The per protocol and as peritonealcancerintheestrogenpluspro- mal findings (85% vs 68%), reflecting treated analyses for endometrial cancer gestin group was detected by an ultra- increased bleeding not arising from ma- also did not yield important differences sound performed to evaluate the endo- lignant or premalignant lesions. While in overall findings (HR, 0.83; 95% CI, metrium. One leiomyosarcoma of the the proportion of biopsies yielding ab- 0.42-1.64 and HR, 0.80; 95% CI, 0.45- uterus was also reported. No difference normal findings in women taking es- 1.43, respectively). Censoring at the time was detected in cervical cancer incidence trogen plus progestin was low, the pro- of hysterectomy provided a similar result (8 vs 5; HR, 1.44; 95% CI, 0.47-4.42). portion of women having these (HR, 0.81; 95% CI, 0.49-1.36). Estrogen plus progestin reduced the abnormalities in the estrogen plus pro- There was no evidence of a differ- percentage of unsuccessful biopsies rela- gestin group was not reduced relative ence in the distributions of histologi- tive to placebo in the cohort randomly to placebo. More simple and adenoma- cal class, morphological grade, or stage selected for routine surveillance (21% vs tous hyperplasias and atypias were at diagnosis of endometrial cancer by 36%, PϽ.001), but no difference was ob- found in the estrogen plus progestin randomization assignment. No signifi- served in the distribution of findings group, although this increase may be cant interactions were found with age, among women with results available an artifact of the higher biopsy rate. race/ethnicity, body mass index, hy- (P=.28) (TABLE 3). This comparison may More women in the estrogen plus pro- pertension, smoking status, pack- be confounded by a higher proportion gestin group were examined with ultra- years of smoking, prior use of unop- of women having multiple biopsies in the sound; the fraction with repeat ultra- posed estrogen, or prior use of estrogen estrogen plus progestin group (47% vs sounds was also elevated (TABLE 4). No plus progestin. The possibility of an in- 27%; PϽ.001) because it is impossible significant differences were found in the teraction with diabetes could not be to accurately attribute biopsies in these endometrial findings of these examina- tested because of the sparseness of these women to those conducted in response tions, but some small differences were data. One death in the placebo group to reports of bleeding vs those done rou- noted in the proportion with other pel- was attributed to endometrial cancer. tinely. vic abnormalities.

Figure 2. Kaplan-Meier Estimates of Cumulative Hazards by Randomization Assignment

A Invasive Ovarian Cancer B Endometrial Cancer 0.010 Estrogen + Progestin Placebo

0.005 Cumulative Hazard

0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Time, y Time, y No. at Risk Estrogen + Progestin 8506 8408 8339 8263 8060 5876 3103 1360 8506 8408 8339 8263 8051 5867 3100 1357 Placebo 8102 8024 7951 7879 7698 5527 2763 1005 8102 8022 7945 7868 7684 5511 2755 1002

A, The hazard ratio is 1.58 (95% confidence interval, 0.77-3.24; adjusted 95% confidence interval, 0.59-4.23). B, The hazard ratio is 0.81 (95% confidence interval, 0.48-1.36; adjusted 95% confidence interval, 0.40-1.64).

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013 GYNECOLOGIC CANCERS IN THE WHI

Follow-up Papanicolaou test re- gen alone (OR, 1.5; 95% CI, 0.91-2.4). case-control study21 (the only cur- sults were available for 94% of trial par- That study did not address the sched- rently available study known to distin- ticipants (TABLE 5). The distribution of ule of progestin use. A recent Swedish guish between continuous combined findings varied significantly (PϽ.001) with the estrogen plus progestin group Table 2. Gynecologic Cancer Incidence and Distribution of Tumor Characteristics by yielding slightly more mild dysplasia, Randomization Assignment* low grade squamous intraepithelial le- No. (%) of Cancers sions, or atypia than the placebo group (7.8% vs 5.5%) and fewer normal re- Estrogen + Progestin Placebo Type of Cancer (n = 8506) (n = 8102) HR (95% CI) sults (92% vs 94%). Invasive ovarian (annualized %)† 20 (0.04) 12 (0.03) 1.58 (0.77-3.24) Histology COMMENT Serous papillary 11 (55.0) 7 (58.3) During 5.6 years of follow-up, 111 Adenocarcinoma (not otherwise 4 (20.0) 3 (25.0) women were diagnosed as having specified carcinoma) invasive gynecologic cancers (invasive Clear cell 2 (10.0) 1 (8.3) 1.64 (0.78-3.45)‡ ovarian, 32; endometrial, 58; nonen- Endometroid 2 (10.0) 0 (0) dometrial uterine, 1; cervical, 13; and Embryonal 1 (5.0) 0 (0) other gynecologic, 7). In women ran- Mixed mullerian 0 (0) 1 (8.3) Tumor stage domized to estrogen plus progestin, a Localized§ 3 (15.0) 0 (0) nonsignificant 1.58-fold increase in Regional ࿣ 4 (20.0) 3 (25.0) ovarian cancer and a nonsignificant Distant¶ 12 (60.0) 9 (75.0) 19% reduction in endometrial cancer Missing 1 (5.0) 0 (0) were observed, relative to placebo. Tumor grade Data for other gynecologic malignan- Moderately differentiated 5 (25.0) 1 (8.3) cies were too sparse to provide mean- Poorly differentiated 6 (30.0) 8 (66.7) ingful comparisons but were included Undifferentiated 1 (5.0) 2 (16.7) for completeness. Women randomized Missing 8 (40.0) 1 (8.3) Borderline ovarian (annualized %)† 1 (Ͻ0.01) 3 (Ͻ0.01) to continuous combined hormones Endometrial (annualized %)† 27 (0.06) 31 (0.07) 0.81 (0.48-1.36) were subjected to more endometrial Histology biopsies and vaginal ultrasounds and Endometroid 14 (51.9) 17 (54.8) were more frequently found to have Adenocarcinoma (not otherwise 8 (29.6) 9 (29.0) mild abnormalities in routine Papani- specified carcinoma) colaou tests. Mucinous 1 (3.7) 1 (3.2) Squamous cell 0 (0) 1 (3.2) 0.71 (0.41-1.22)‡ Ovarian Cancer Clear cell 0 (0) 1 (3.2) In this trial, women taking estrogen plus Serous papillary 0 (0) 1 (3.2) Stromal sarcoma 2 (7.4) 0 (0) progestin were diagnosed as having in- Mixed mullerian 2 (7.4) 1 (3.2) vasive ovarian cancer at a rate of 42 per Tumor stage 100000 person-years, 15 per 100000 Localized# 23 (85.2) 26 (83.9) person-years more than the placebo Regional** 4 (14.8) 5 (16.1) group rate. The possibility of an in- Tumor grade crease in ovarian cancer mortality was Well differentiated 9 (33.3) 6 (19.4) also noted. Many, but not all, observa- Moderately differentiated 10 (37.0) 15 (48.4) tional studies have found a modest in- Poorly differentiated 5 (18.5) 7 (22.6) creased risk of ovarian cancer or ovar- Undifferentiated 3 (11.1) 3 (9.7) Nonendometrial uterine (annualized %)† 1 (Ͻ0.01) 0 (0) ian cancer mortality associated with Cervical (annualized %)† 8 (0.02) 5 (0.01) 1.44 (0.47-4.42) postmenopausal estrogen use,14-19 but Other gynecologic (annualized %)† †† 6 (Ͻ0.01) 1 (Ͻ0.01) few studies have reported results spe- Abbreviations: CI, confidence interval; HR, hazard ratio. cifically on combined hormones. In a re- *The mean (SD) follow-up time was 67.8 (16.2) for the estrogen plus progestin group and 66.8 (15.2) for the placebo 20 group. cently reported US cohort study, an as- †Reflects rate per person-year. sociation was found with estrogen alone ‡Among epithelial histologies. §Confined to ovary with no tumor on the surface of ovary. (odds ratio [OR], 1.6; 95% CI, 1.2-2.0) ࿣Confined to the pelvis. ¶Implants beyond the pelvis. but not with combined hormones (OR, #Confined to endometrium, myometrium/serosa. 1.1; 95% CI, 0.64-1.7), except possibly **Confined to pelvis, vagina, or wall of rectum or bladder. ††Includes fallopian tube and primary peritoneum cancers. in women previously exposed to estro-

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013 GYNECOLOGIC CANCERS IN THE WHI

Table 3. Proportion With Follow-up Endometrial Biopsies and Distribution of Findings by Randomization Assignment* No. (%) of Women Selected for Routine Biopsy No. (%) of Women Selected for Usual Care

Estrogen + Progestin Placebo P Estrogen + Progestin Placebo P (n = 482) (n = 453) Value† (n = 7693) (n = 7649) Value† No. of women with biopsies 412 (85.5) 368 (81.2) .09 2569 (33.4) 439 (5.7) Ͻ.001 No. of women with multiple biopsies 194 (40.2) 98 (21.6) Ͻ.001 979 (12.7) 72 (0.9) Ͻ.001 Findings‡ No endometrial tissue/insufficient specimen 74 (18.0) 105 (28.5) Ͻ.001 270 (10.5) 98 (22.3) Ͻ.001 Normal 264 (64.1) 180 (48.9) 2004 (78.0) 267 (60.8) Simple hyperplasia 3 (0.7) 0 (0) 53 (2.1) 12 (2.7) Adenomatous hyperplasia 1 (0.2) 1 (0.3) 10 (0.4) 3 (0.7) .28§ Ͻ.001§ Atypia 3 (0.7) 5 (1.4) 16 (0.6) 6 (1.4) Cancer 0 (0) 0 (0) 7 (0.3) 7 (1.6) Unknown 67 (16.3) 77 (20.9) 259 (10.1) 46 (10.5) *Excludes the 331 women switched from estrogen alone to estrogen plus progestin. †Fisher exact test. ‡Most severe result for women with multiple biopsies. §Among women with a sufficient specimen sample. have been introduced and allow for an Table 4. Proportion With Transvaginal Uterine Ultrasound Examinations and Findings by increased risk with hormones through Randomization Assignment an indirect pathway. The WHI data do No. (%) of Women not relate directly to the incessant ovu- Estrogen + Progestin Placebo P lation hypothesis28 because all partici- (n = 8506) (n = 8102) Value* pants were menopausal at entry. Ͻ No. of women with ultrasound examinations 1089 (12.8) 331 (4.1) .001 Progestins have been hypothesized to No. of women with multiple ultrasound 299 (3.5) 60 (0.7) Ͻ.001 examinations have a favorable effect on ovarian can- 22 Findings† cer incidence based on generally con- Unable to evaluate thickness/perform 31 (2.8) 8 (2.4) .85 sistent findings of lower risk associated examination with increasing parity and the use of oral Endometrial thickness 29 Յ5 mm 771 (70.8) 248 (75.0) contraceptives, and on animal data de- .16‡ Ͼ5 mm 277 (25.4) 72 (22.0) scribing a role for progestins in promot- 30 Unknown 10 (0.9) 3 (0.9) ing apoptosis. The current WHI trial Pelvic pathological result (ever)§ cannot address this question directly, but Polyps 97 (9.1) 17 (5.1) .03 the eventual comparison with the par- Uterine mass 229 (21.0) 52 (15.7) .04 allel trial of estrogen alone in women Pelvic fluid 11 (1.0) 3 (0.9) Ͼ.99 with prior hysterectomy may provide Ovarian mass 77 (7.1) 29 (8.8) .28 some insight. The WHI data suggest that Other 65 (6.0) 14 (4.2) .27 the continuous combined estrogen and *Fisher exact test. progestin preparation examined in the †Most severe result for women with multiple ultrasound examinations. ‡Among women with evaluable thickness. trial will have no role in ovarian cancer §Data for other pelvic pathologies available on 98% of women with ultrasound. prevention.

hormone therapy and sequential pro- While the etiologies of ovarian can- Endometrial Cancer gestin treatment) found an increased risk cer are poorly understood, a role for estro- In women taking estrogen plus proges- associated with use of unopposed estrogen and progestin is biologically plau- tin, the incidence of endometrial can- gen (OR, 1.43; 95% CI, 1.02-2.00) and sible.22 For example, the gonadotropin cer during the 5.6 years of follow-up sequential preparations (OR, 1.54; 95% hypothesis asserts that the many repro- was 56 per 100000 person-years or 13 CI, 1.15-2.05), but not with use of con- ductive history risk factors having a mod- fewer cases per 100000 person-years tinuous progestin regimens (OR, 1.02; est association with ovarian cancer risk than observed in women taking pla- 95% CI, 0.73-1.43). However, the ma- act indirectly by increasing exposure of cebo. This difference cannot be distin- jority of women in this latter study used the ovarian epithelium to estrogen23,24 and guished from chance. a progestin derived from 19-nortestoste- consequently to proliferation and malig- This is the first randomized, double- rone rather than the 17-hydroxyproges- nant transformation.25 Other hypoth- blind, placebo-controlled trial to dem- terone derivative medroxyprogester- eses regarding inflammation of the ovar- onstrate that endometrial cancer rates for one acetate used in the current trial. ian epithelium26 and retrograde bleeding27 women taking continuous combined

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013 GYNECOLOGIC CANCERS IN THE WHI

hormones are similar to placebo group Table 5. Proportion With Follow-up Papanicolaou Tests and Findings by Randomization rates, indicating that progestin protects Assignment against the increased risk of endome- No. (%) of Women trial cancer associated with unopposed Estrogen + Progestin Placebo P estrogen. Several reasonably sized, ran- (n = 8506) (n = 8102) Value* domized, placebo-controlled trials of No. of women with Papanicolaou test 7950 (93.5) 7599 (93.8) .39 combined hormones have reported a No. of women with multiple Papanicolaou tests 6072 (71.4) 5671 (70.0) .05 substantial reduction in the formation of Findings† endometrial hyperplasia, the presumed Insufficient specimen or slide damaged 1 (Ͻ0.1) 5 (0.1) .12 precursor to many endometrial carcino- Normal result 7300 (91.8) 7135 (93.9) mas compared with women given estro- Mild dysplasia/LGSIL, atypia‡ 619 (7.8) 420 (5.5) gen-only therapy.12,31,32 The results of the Moderate to severe dysplasia/HGSIL 25 (0.3) 29 (0.4) Ͻ.001 present study are also consistent with one Cancer 2 (Ͻ0.1) 3 (Ͻ0.1) smaller randomized trial33 and 3 obser- Unknown 3 (Ͻ0.1) 7 (Ͻ0.1) vational studies reporting no increase in Abbreviations: HGSIL, high-grade squamous intraepithelial lesion; LGSIL, low-grade squamous intraepithelial lesion. *Fisher exact test. endometrial cancer risk associated with †Most severe result for women with multiple Papanicolaou tests. ‡Includes atypical squamous cells of undetermined significance and atypical glandular cells of uncertain significance/ continuous combined estrogen plus pro- atypical glandular cells of undetermined significance. gestin therapy of at least 3 years’ duration.1-3 Uterine bleeding was a frequent ad- constrains our ability to provide de- these medicines. These data provide fur- verse effect of this regimen, leading to tailed subclassification of disease. As gy- ther support for the recently revised much more frequent biopsies and ultra- necologic cancers were secondary end guidelines for the use of continuous com- sounds in women taking combined hor- points of the trial, issues of multiple test- bined estrogen plus progestin therapy.34,35 mones than taking placebo. This in- ing apply. Adjusted CIs are provided to Author Contributions: Dr Anderson, as co-principle crease in bleeding and the gynecologic suggest the additional conservatism investigator of the Women’s Health Initiative Clinical procedures needed to diagnose and re- needed in interpreting these results. The Coordinating Center, had full access to all study data and takes responsibility for the integrity of the data solve the ensuing concerns (including trial was stopped early, which limited the and the accuracy of the data analyses. hysterectomy) continue to be a major precision of these results and pre- Study concept and design: Anderson, Judd, Barad, Beresford. drawback of combined hormone therapy, cluded the examination of longer-term Acquisition of data: Anderson, Barad, Beresford, including continuous progestins. exposure. Finally, this trial tested one McNeeley. Analysis and interpretation of data: Anderson, Judd, means of administering a single hor- Kaunitz, Barad, Pettinger, Liu, Lopez. Cervical Cancer mone regimen. We do not know the ex- Drafting of the manuscript: Anderson, Judd, Barad. The WHI data on cervical cancer are too tent to which these results apply to other Critical revision of the manuscript for important in- tellectual content: Anderson, Judd, Kaunitz, Barad, limited to suggest there is any associa- postmenopausal hormones. The effects Beresford, Pettinger, Liu, McNeeley, Lopez. tion between incidence and estrogen of estrogen alone will be examined in the Statistical expertise: Anderson, Pettinger. Obtained funding: Anderson. plus progestin therapy. The statisti- parallel trial that is ongoing. Administrative, technical, or material support: Judd, cally significant increase in mild ab- Beresford, Liu, McNeeley, Lopez. Implications Study supervision: Anderson, Judd, Liu. normalities detected on Papanicolaou WHI Investigators and Clinical Centers: Alabama: Uni- tests is interesting and warrants fur- In assessing the overall merit of estro- versity of Alabama at Birmingham, Cora E. Lewis, Albert ther investigation. gen plus progestin therapy, the low rates Oberman, Mona N. Fouad, James M. Shikany, Delia Smith West. Arizona: Tucson/Phoenix: University of Ari- of gynecologic cancers in the popula- zona, Tamsen Bassford, John Mattox, Marcia Ko, Timo- Limitations tion and the limited precision in the es- thy Lohman. California: University of California at Los Angeles, Howard Judd, David Heber, Robert Elashoff; The WHI estrogen plus progestin trial is timated effects from this trial suggest that Oakland: Kaiser Permanente Division of Research, Bette the largest, randomized, double-blind, these results should not have an appre- Caan, Stephen Sidney, Geri Bailey, Jane Hirata; Orange: University of California at Irvine, Allan Hubbell, Gail Frank, placebo-controlled trial of continuous ciable influence on most women’s deci- Nathan Wong, Nancy Greep, Bradley Monk; Sacra- combined hormones that has been con- sion making when seeking relief for mod- mento: University of California at Davis, John Robbins, S. Yasmeen, Karen Lindfors, Judith Stern; University of ducted, yet some limitations must be ac- erate to severe vasomotor symptoms, nor California at San Diego, Robert D. Langer, Michael H. knowledged. The number of gyneco- can they resolve questions of etiology. Criqui, Gregory T. Talavera, Cedric F. Garland, R. Elaine logic cancers observed was small, The possibility of an increased risk of Hanson; Stanford Center for Research in Disease Pre- vention, Stanford University, Marcia L. Stefanick, Mark yielding wide CIs for the overall effects ovarian cancer incidence and mortality A. Hlatky, Bertha Chen, Randall S. Stafford, Linda C. Giu- and limited power to examine possible remains worrisome, however, and needs dice; Torrance: Harbor-UCLA Research and Education Institute, Rowan Chlebowski, Robert Detrano, Anita Nel- differential effects in disease subtypes or confirmation. The increased need for di- son, James Heiner, John Marshall. District of Colum- in subgroups of women. Use of commu- agnostic procedures in response to bleed- bia: Washington: MedStar Research Institute/Howard University, Barbara V. Howard, Lucile Adams- nity-based diagnoses rather than cen- ing is an added burden and could rea- Campbell, Maureen Passaro, Monique Rainford, Tanya tral reading of pathological specimens sonably affect a woman’s decision to use Agurs-Collins; George Washington University, Judith

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013 GYNECOLOGIC CANCERS IN THE WHI

Hsia, Nancy Gaba, Joao Ascensao, Somchia Laowatt- Coordinating Center Subcontractors: California: Uni- 12. Effects of hormone replacement therapy on en- ana. Florida: Gainesville/Jacksonville: University of Florida, versity of California at San Francisco, Steven Cum- dometrial histology in postmenopausal women. JAMA. Marian Limacher, Michael Perri, Andrew Kaunitz, R. Stan mings, Michael Nevitt, Maurice Dockrell. Kentucky: 1996;275:370-375. Williams, Yvonne Brinson; University of Miami, Mary Highland Heights: Medical Research Labs, Evan Stein, 13. National Cancer Institute. Surveillance, Epidemi- Jo O’Sullivan, Linda Parker, R. Estape, Diann Fernan- Peter Laskarzewski. Maryland: Rockville: McKesson ology, and End Results (SEER)*Stat Database dez. Georgia: Atlanta: Emory University, Lawrence Phil- BioServices, Frank Cammarata, Steve Lindenfelser. (1973-2000). Available at: http://www.seer.cancer lips, Margaret Pedersen, Ora Strickland, Margaret Huber, Minnesota: Minneapolis: University of Minnesota, Lisa .gov. Accessed April 17, 2003. Vivian Porter. Hawaii: Honolulu: University of Hawaii, Harnack. North Carolina: Winston-Salem: Wake For- 14. Weiss NS, Lyon JL, Rishnamurthy S, Dietert SE, David Curb, Helen Petrovitch, Beatriz Rodriguez, Kamal est University School of Medicine, Sally Shumaker, Liff JM, Daling J. Noncontraceptive estrogen use and the occurrence of ovarian cancer. J Natl Cancer Inst. Masaki, Santosh Sharma. Illinois: Chicago: Northwest- Pentti Rautaharju, Ronald Prineas, Michelle Naugh- ton. Washington: University of Washington, Seattle, 1982;68:95-98. ern University, Linda Van Horn, Philip Greenland, Janar- 15. Bruce Psaty, Susan Heckbert. Booth M, Beral V, Smith P. Risk factors for ovar- dan Khandekar, Kiang Liu, Carol Rosenberg; Rush- ian cancer. Br J Cancer. 1989;60:592-598. Program Office: National Heart, Lung, and Blood In- Presbyterian-St Luke’s Medical Center, Henry Black, 16. Whittemore AS. Personal characteristics relating stitute, Bethesda, Md, Barbara Alving, Jacques Ros- Lynda Powell, Ellen Mason. Iowa: Davenport: Univer- to risk of invasive epithelial ovarian cancer in older sity of Iowa, Robert Wallace, James Torner, Susan souw, Linda Pottern, Shari Ludlam, Joan McGowan. women in the United States. Cancer. 1993;72:558- Johnson, Linda Snetselaar, Bradley VanVoorhis. Mas- Financial Disclosures: Dr Kaunitz has received fund- 565. sachusetts: Boston: Brigham and Women’s Hospital, ing for clinical trials from Barr Laboratories, Berlex, Ga- 17. Purdie D, Green A, Bain C, et al. Reproductive and JoAnn Manson, Julie Buring, J. Michael Gaziano, Kathryn len, Johnson & Johnson, and Pharmacia (Pfizer); has other factors and risk of epithelial ovarian cancer. Int Rexrode, Claudia Chae; Worcester: University of Mas- continuing medical education presentations and pub- J Cancer. 1995;62:678-684. sachusetts/Fallon Clinic, Judith Ockene, Milagros Rosal, lications funded by Aventis, Ortho-McNeil, Pharma- 18. Risch HA. Estrogen replacement therapy and risk Ira Ockene, Robert Yood, Patricia Aronson. Michigan: cia (Pfizer), and Wyeth-Ayerst; has been a consul- of epithelial ovarian cancer. Gynecol Oncol. 1996; Detroit: Wayne State University School of Medicine/ tant for Aventis, Barr Laboratories, Berlex, Johnson & 63:254-257. Hutzel Hospital, Susan Hendrix, Michael Simon, Gene Johnson (Ortho-McNeil), and Pharmacia (Pfizer); and 19. Rodriguez C, Patel AV, Calle EE, Jacobs EJ, Thun McNeeley, Pamela Gordon, Paul Makela. Minnesota: holds stock in Aventis and Johnson & Johnson. Dr Liu MJ. Estrogen replacement therapy and ovarian can- Minneapolis: University of Minnesota, Richard H. Grimm, has been on the speakers bureau with Merck Phar- cer mortality in a large prospective study of US women. Kathleen M. Hall, Donald B. Hunninghake, June LaVal- maceuticals and Wyeth; has been a consultant for Wat- JAMA. 2001;285:1460-1465. leur, Karen L. Margolis. Nevada: Reno: University of son Pharmaceuticals, Barr Laboratories, UMD Inc, 20. Lacey JV, Mink PJ, Lubin JH, et al. Menopausal Nevada, Robert Brunner, Sachiko St Jeor, William Graet- Merck Pharmaceuticals, and Solvay Pharmaceuticals; hormone replacement therapy and risk of ovarian cancer. JAMA. 2002;288:334-341. tinger, Vicki Oujevolk. New Jersey: Newark: University and has received research grants from Procter & Gamble Pharmaceuticals, Aventis, and Pfizer. 21. Riman T, Dickman PW, Nilsson S, et al. Hor- of Medicine and Dentistry of New Jersey, Norman Lasser, mone replacement therapy and the risk of invasive epi- Norman Hymowitz, Vera Lasser, Monika Safford, John Funding/Support: The WHI program is funded by the National Heart, Lung, and Blood Institute, US Depart- thelial ovarian cancer in Swedish Women. J Natl Can- Kostis. New York: Bronx: Albert Einstein College of Medi- cer Inst. 2002;94:497-504. cine, Sylvia Wassertheil-Smoller, William Frishman, Judith ment of Health and Human Services. Wyeth-Ayerst Re- search provided the study medication (active and placebo). 22. Risch HA. Hormonal etiology of epithelial ovar- Wylie-Rosett, David Barad, Ruth Freeman; State Uni- ian cancer, with a hypothesis concerning the role of versity of New York at Buffalo, Maurizio Trevisan, Jean Acknowledgment: We appreciate the WHI participants and staff for their efforts and in particular to Al- androgens and progesterone. J Natl Cancer Inst. 1998; Wactawski-Wende, Susan Graham, June Chang, Ellen 90:1774-1786. lison Weber for cancer coding and Noel Weiss for his Smit; State University of New York at Stony Brook, Doro- 23. Gardner WU. Tumorigenesis in transplanted ir- review of this article. thy Lane, Iris Granek, William Lawson, Gabriel San radiated and non-irradiated ovaries. J Natl Cancer Inst. Roman, Catherine Messina. North Carolina: Chapel Hill: 1961;26:829-854. University of North Carolina, Gerardo Heiss, Pamela REFERENCES 24. Stadel BV. The etiology and prevention of ovarian Haines, David Ontjes, Carla Sueta, Ellen Wells; Winston- cancer. Am J Obstet Gynecol. 1975;123:772-774. Salem: Wake Forest University School of Medicine, Greg 1. Pike MC, Peters RK, Cozen W, et al. Estrogen- 25. Cramer DW, Welch WR. Determinants of ovar- Burke, Robin Crouse, Lynne Parsons, Mara Vitolins. Ohio: progestin replacement therapy and endometrial can- ian cancer risk, II: inferences regarding pathogenesis. University of Cincinnati, Margery Gass, Suzanne Wernke, cer. J Natl Cancer Inst. 1997;89:1110-1116. J Natl Cancer Inst. 1983;71:717-721. Nelson Watts; Columbus: Ohio State University, Rebecca 2. Weiderpass E, Adami HO, Baron JA, et al. Risk of 26. Ness RB, Cottreau C. Possible role of ovarian epi- Jackson, Randall Harris, David Frid, W. Jerry Mysiw, endometrial cancer following estrogen replacement thelial inflammation in ovarian cancer. J Natl Cancer with and without progestins. J Natl Cancer Inst. 1999; Inst. 1999;91:1459-1467. Michael Blumenfeld. Oregon: Portland: Kaiser Perma- 91:1131-1137. 27. Cramer DW, Xu H. Epidemiologic evidence for nente Center for Health Research, Cheryl Ritenbaugh, 3. Hill DA, Weiss NS, Beresford SAA, et al. Continu- uterine growth factors in the pathogenesis of ovarian Barbara Valanis, Patricia Elmer, Victor Stevens, Njeri ous combined hormone replacement therapy and risk cancer. Ann Epidemiol. 1995;5:310-314. Karanja. Pennsylvania: University of Pittsburgh, Lewis of endometrial cancer. Am J Obstet Gynecol. 2000; 28. Fathalla MF. Incessant ovulation: a factor in ovar- Kuller, Arlene Caggiula, Jane Cauley, Sarah Berga, N. 183:1456-1461. ian neoplasia? Lancet. 1971;2:163. Carole Milas. Rhode Island: Providence: Brown Univer- 4. Women’s Health Initiative Study Group. Design of 29. Ness RB, Grisso JA, Klapper J, et al. Risk of ovar- sity, Annlouise R. Assaf, Richard Carleton†, Carol the Women’s Health Initiative clinical trial and obser- ian cancer in relation to estrogen and progestin dose Wheeler, Charles Eaton, Michelle Cyr. Tennessee: vational study. Control Clin Trials. 1998;19:61-109. and use characteristics of oral contraceptives. Am J Epi- Memphis: University of Tennessee, Karen C. Johnson, 5. Rossouw JE, Anderson GL, Prentice RL, et al. Risks demiol. 2000;152:233-241. Suzanne Satterfield, Raymond W. Ke, Jere Vile, Fran and benefits of estrogen plus progestin in healthy post- 30. Rodriguez GC, Walmer DK, Cline M. Effect of pro- Tylavsky; Texas: Houston: Baylor College of Medi- menopausal women. JAMA. 2002;288:321-333. gestin on the ovarian epithelim of macaques. J Soc Gy- cine, Jennifer Hays, Ronald Young, Jill Anderson, Sandy 6. Kaunitz AM, Masciello A, Ostrowski M, Rovira EZ. necol Investig. 1998;5:271-276. Lithgow, Paul Bray; San Antonio: University of Texas Comparison of endometrial biopsy with the endome- 31. Woodruff JD, Pickar JH, et al. Incidence of en- Health Science Center, Robert Brzyski, Robert Schen- trial pipelle and vabra aspirator. J Reprod Med. 1988; dometrial hyperplasia in postmenopausal women tak- ken, Jose Trabal, Mercedes Rodriguez-Sifuentes, 33:427-430. ing conjugated estrogens (Premarin) with medroxy- Charles Mouton. Washington: Seattle: Fred Hutchin- 7. Stovall TG, Photopulos GJ, Poston WM, Ling FW, progesterone acetate or conjugated estrogens alone. son Cancer Research Center, Shirley A. A. Beresford, Sandles LG. Pipelle endometrial sampling in patients Am J Obstet Gynecol. 1994;170:1213-1223. Vicky M. Taylor, Nancy F. Woods, Maureen Hender- with known endometrial carcinoma. Obstet Gyne- 32. Speroff L, Rowan J, Symons J, et al. The com- col. son, Mark Kestin. Wisconsin: Madison: University of 1991;77:954-956. parative effect on bone density, endometrium and lip- 8. Smith-Bindman R, Kerlikowske K, Feldstein VA, et ids of continuous hormones as replacement therapy Wisconsin, Catherine Allen, Douglas Laube, Patrick al. Endovaginal ultrasound to exclude endometrial can- (CHART study). JAMA. 1996;276:1397-1403. McBride, Julie Mares-Perlman, Barbara Loevinger; Mil- cer and other endometrial abmormalites. JAMA. 1998; 33. Hulley S, Furberg C, Barrett-Connor E, et al. Non- waukee: Medical College of Wisconsin, Jane Morley 280:1510-1517. cardiovascular disease outcomes during 6.8 years of Kotchen, Vanessa Barnabei, Theodore A. Kotchen, 9. Percy C, Van Holten V, Muir C, eds. International hormone therapy. JAMA. 2002;288:58-66. Mary Ann C. Gilligan, Joan Neuner. Classification of Diseases for Oncology, 2nd Edition. 34. American College of Obstetricians and Gynecolo- †Deceased. Geneva, Switzerland: World Health Organization; 1990. gists. Hormone therapy: questions and answers from Clinical Coordinating Center: Fred Hutchinson Can- 10. The SEER Program Code Manual, Revised Edi- ACOG physicians. Available at: http://www.acog.org cer Research Center, Seattle, Wash: Ross Prentice, Gar- tion. Bethesda, Md: Cancer Statistics Branch, Na- /from_home/publications/press_releases/nr08-30-02 net Anderson, Andrea LaCroix, Ruth Patterson, Anne tional Cancer Institute; 1992. .cfm. Accessed March 31, 2003. McTiernan, Barbara Cochrane, Julie Hunt, Lesley Tinker, 11. Freedman LS, Anderson GL, Kipnis V, et al. Ap- 35. US Preventive Services Task Force. Postmeno- Charles Kooperberg, Martin McIntosh, C. Y. Wang, proaches to monitoring the results of long-term dis- pausal hormone replacement therapy for primary pre- Chu Chen, Deborah Bowen, Alan Kristal, Janet Stan- ease prevention trials. Control Clin Trials. 1996;17: vention of chronic conditions: recommendations and ford, Nicole Urban, Noel Weiss, Emily White. 509-525. rationale. Ann Intern Med. 2002;137:834-839.

Downloaded From: http://jama.jamanetwork.com/ by a Universidade de Antioquia User on 06/11/2013