{"paper_id":"7d5e0af7-f404-4136-8743-d1c981384250","body_text":"Unopposed oestrogen replacement therapy (ORT) is an established\nrisk factor for endometrial adenocarcinoma (Grady et al, 1995).\nThe relation between hormone replacement therapy (HRT) and\novarian cancer remains uncertain, however. Since cancer of the\novary shares some risk factors with cancer of the endometrium,\nsuch as low parity (Whittemore et al, 1992) and obesity (Purdie et\nal, 1995), post-menopausal oestrogen therapy may also have a role\nin its aetiology. This may particularly be so for the endometrioid\nepithelial form, which is histologically similar to adenocarcinoma\nof the endometrium (Russel, 1994). Some studies have shown\nunopposed oestrogen therapy for at least 1 year to be associated\nwith an increase in risk of endometrioid tumours predominantly\n(La Vecchia et al, 1982; Weiss et al, 1982; Cramer et al, 1983;\nRisch, 1996) while others have suggested either no effect (Booth\net al, 1989; Kaufman et al, 1989; Whittemore et al, 1992) or an\ninverse association (Hartge et al, 1988; Hempling et al, 1997)\nbetween oestrogens and this type of ovarian cancer. Some of these\nstudies, however, have involved small numbers of exposed cases\n(La Vecchia et al, 1982; Cramer et al, 1983; Kaufman et al, 1989),\nhave not distinguished unopposed oestrogen from oestrogen given\nin combination with progestogen (Booth et al, 1989; Hempling et\nal, 1997), or have used hospital controls (Hartge et al, 1988; Booth\net al, 1989; Kaufman et al, 1989; Hempling et al, 1997) with\npossible overestimation of general exposure rates (Rodriguez et al,\n1995). Hence, the issue is currently unresolved.\nWe report here the results of a large Australian case–control\nstudy in which the relation between HRT, and in particular unop-\nposed ORT, and epithelial ovarian cancer, both overall and among\nhistological types, was explored. We also examined any potential\nmodification of the effect of ORT on ovarian cancer risk by tubal\nsterilization and hysterectomy.\nMETHODS\nHistologically confirmed incident cases of primary epithelial\novarian cancer registered in major gynaecological-oncology treat-\nment centres in three Australian states were ascertained. Cases\ndiagnosed in 1991 and 1992 in New South Wales and Victoria,\nand from August 1990 to the end of 1993 in Queensland were\nrecruited. The histological types of all cancers of women included\nin the study were reviewed by an independent gynaecological\npathologist in each state, as described in detail elsewhere (Purdie\nHormone replacement therapy and risk of epithelial\novarian cancer\nDM Purdie1, CJ Bain1, V Siskind1, P Russell2, NF Hacker3, BG Ward4, MA Quinn5 and AC Green6\n1Department of Social and Preventive Medicine, The University of Queensland, Medical School, Herston Road, Herston, Queensland 4006, Australia;\n2Department of Anatomical Pathology, Royal Prince Alfred Hospital, Missenden Road, Camperdown, New South Wales 2050, Australia; 3Gynaecological\nCancer Centre, Royal Hospital for Women, 188 Oxford Street, Paddington, New South Wales 2021, Australia; 4Department of Obstetrics and Gynaecology,\nThe University of Queensland, Royal Brisbane Hospital, Herston, Queensland 4029, Australia; 5Gynaecologic Oncology/Dysplasia Unit, The Royal Women’s\nHospital, 132 Grattan Street, Carlton, Victoria 3053, Australia; 6Queensland Institute of Medical Research, Royal Brisbane Hospital, Herston, Queensland 4029,\nAustralia\nSummary It has been suggested that oestrogen replacement therapy is associated with risk of epithelial ovarian cancer of the endometrio id\ntype. Using data from an Australian population-based case–control study, the relation between unopposed oestrogen replacement t herapy\nand epithelial ovarian cancer, both overall and according to histological type, was examined. A total of 793 eligible incident cases of epithelial\novarian cancer diagnosed from 1990 to 1993 among women living in Queensland, New South Wales and Victoria were identified. These were\ncompared with 855 eligible female controls selected at random from the electoral roll, stratified by age and geographic region.  Trained\ninterviewers administered standard questionnaires to obtain detailed reproductive and contraceptive histories, as well as detai ls about\nhormone replacement therapy and pelvic operations. No clear associations were observed between use of hormone replacement thera py\noverall and risk of ovarian cancer. Unopposed oestrogen replacement therapy was, however, associated with a significant increase in risk of\nendometrioid or clear cell epithelial ovarian tumours (odds ratio (OR) 2.56; 95% confidence interval (CI) 1.32–4.94). In additi on, the risk\nassociated with oestrogen replacement therapy was much larger in women with an intact genital tract (OR 3.00; 95% Cl 1.54–5.85)  than in\nthose with a history of either hysterectomy or tubal ligation. Post-menopausal oestrogen replacement therapy may, therefore, be a risk factor\nassociated with endometrioid and clear cell tumours in particular. Additionally, the risk may be increased predominantly in wom en with an\nintact genital tract. These associations could reflect a possible role of endometriosis in the development of endometrioid or clear cell ovarian\ntumours. © 1999 Cancer Research Campaign\nKeywords: ovarian neoplasms; case–control study; oestrogen replacement therapy\n559\nBritish Journal of Cancer (1999) 81(3), 559–563\n© 1999 Cancer Research Campaign\nArticle no. bjoc.1999.0731\nReceived 5 October 1998\nRevised 19 March 1999\nAccepted 24 March 1999\nCorrespondence to: DM Purdie, Department of Social and Preventive\nMedicine, 4th Floor – North Wing, Diamantina House, Princess Alexandra\nHospital, Ipswich Road, Woolloongabba, Queensland 4102, Australia\n\net al, 1995). Briefly, ovarian cancer patients aged 18–79 years at\ndiagnosis who were capable of completing the questionnaire were\neligible to participate. A control series was selected at random\nfrom the electoral roll (enrolment to vote is compulsory in\nAustralia) frequency matched to cases on age (20-year bands) and\nbroad geographic region. Women with a history of ovarian cancer\nor bilateral oophorectomy, were not eligible to be controls; cases\nnot on the electoral roll were excluded from the analysis.\nIdentically trained interviewers administered a standard ques-\ntionnaire to each woman in a face-to-face interview to obtain\npersonal details including education, height and weight, smoking\nhistory, family history of ovarian and other cancers, full reproduc-\ntive and contraceptive histories and age at menopause. Questions\nwere also asked about tubal sterilization and hysterectomy, and\nwith the women’s written consent, confirmation of these pro-\ncedures was sought from relevant medical practitioners as\ndescribed previously (Green et al, 1997a).\nInformation was gathered about use of HRT, specifically the\ntype of hormones, doses, age at first use and duration of use were\nrecorded for up to three different types of treatment. Any HRT\nused by cases after diagnosis was not considered; and those using\nHRT at diagnosis were considered to be current users. HRT use in\ncontrols was treated similarly.\nRelative risks of ovarian cancer, both overall and within subcat-\negories, were estimated using multiple logistic regression to\nsimultaneously adjust for age (in years), level of education,\nresidential location, parity, duration of oral contraceptive pill use,\ntalc use in the perineal region, body mass index (BMI), smoking\nstatus, hysterectomy and tubal sterilization (where appropriate)\nand a family history of breast or ovarian cancer (Purdie et al,\n1995). Women who had ever used unopposed ORT for at least\n1 month were compared with those who had never used HRT, so\nwomen who had used other forms of HRT (63 cases and 74\ncontrols) were excluded from the unopposed oestrogen analyses.\nSimilar analyses were performed among women with and without\na history of tubal sterilization or hysterectomy to assess potential\neffect modification from these procedures. Interactions between\nfactors were assessed using the product term in the logistic\nregression models. All analyses were carried out using the SAS\nstatistical package.\nRESULTS\nThere were 793 eligible cases and 855 eligible controls in the\nstudy representing response rates of 90% and 73% respectively.\nOvarian cancer was subdivided into five histological groups:\nserous (415, 52%); endometrioid and the related clear cell (164,\n21%); mucinous (114, 14%); poorly differentiated adenocarci-\nnomas (54, 7%); and mixed epithelial tumours with malignant\nmixed mesodermal/Müllerian tumours (44, 6%). Two cases with\ntransitional cell (malignant Brenner) tumours were not included in\nthe subtype analyses.\nOverall, 131 cases (16.5%) and 149 controls (17.4%) reported\never taking any form of HRT for menopause-related reasons,\nwhich is consistent with cross-sectional data from a similar aged\npopulation-based Australian study (Dennerstein et al, 1994). Of\nthese, 68 cases (8.6%) and 75 controls (8.8%) reported ever using\nunopposed ORT. After adjustment for potential confounders, there\nwere no significant associations found between occurrence of\novarian cancer and ever use of HRT, either overall or in its various\nforms (Table 1). There were also no clear trends in risk for\nincreased duration of HRT use or time since last use, either overall\nor for the most common forms (Table 2).\nThere was, however, higher use of ORT among cases with\nendometrioid or clear cell cancers than among controls or other\ncases and, after adjustment for potential confounders, a marked\nand significant elevation in risk among this group of cases was\nobserved (Table 3). A comparison of the effect in endometrioid or\nclear cell tumours with the effect in other tumour types approached\nsignificance (P = 0.06). The separate risks for endometrioid and\nfor clear cell tumours were nearly identical.\nWomen who reported having either a hysterectomy or tubal\nsterilization had higher usage of HRT, in particular unopposed\noestrogen: 30% of women with a hysterectomy and 15% of women\nwith a tubal sterilization had ever used ORT, compared with only\n6% of women without such surgery. These women were also at\nlower risk of ovarian cancer (Green et al, 1997 b). We therefore\nexamined whether hysterectomy or tubal sterilization, as well as\nbeing confounders, modified the relationship between ever use of\nORT and ovarian cancer risk. For all cases, an inverse association\nbetween unopposed oestrogen use and ovarian cancer was seen in\nwomen with a prior hysterectomy, and no relationship was seen in\nthose with tubal sterilization alone. In contrast, women who\nreported neither hysterectomy nor tubal sterilization exhibited a\nmoderately strong and significant increase in risk associated with\nORT (Table 4). The interaction effects between hysterectomy and\nORT and between tubal sterilization and ORT among those women\nwithout a hysterectomy were significant ( P = 0.006 and 0.04\nrespectively). Relative risks of unopposed oestrogen therapy were\nalso higher for endometrioid and clear cell tumours than for other\ntumours among women with a prior hysterectomy or tubal steriliza-\ntion and strikingly so among those without surgery (Table 4).\n560 DM Purdie et al\nBritish Journal of Cancer (1999) 81(3), 559–563 © 1999 Cancer Research Campaign\nTable 1 Ever use of hormone replacement therapy among all cases of epithelial ovarian cancer and controls with crude and adjusted odds ratios\nHRT use Cases Controls Crude OR Adjusted OR b\n(n = 793)a (n = 855)a (95% CI) (95% CI)\nNonec 83.5% 82.6% 1.00 1.00\nAny HRT 16.5% 17.4% 0.94 (0.72–1.21) 1.20 (0.90–1.60)\nUnopposed oestrogen 8.6% 8.8% 0.97 (0.68–1.36) 1.27 (0.86–1.88)\nUnopposed progestogen 1.8% 1.3% 1.36 (0.61–3.01) 2.18 (0.91–5.20)\nOestrogen and progestogen in combination 4.9% 5.4% 0.90 (0.58–1.40) 1.34 (0.83–2.17)\nNot known 2.9% 2.7% 1.07 (0.59–1.92) 1.05 (0.55–2.01)\naThe combined total number from treatment groups is greater than the total number of HRT users as 13 cases and six controls reported use of different\nhormone treatments at different times. bORs adjusted for age, education, area of residence, BMI, hysterectomy, tubal sterilization, talc use in perineal region,\nsmoking status, duration of OCP use, parity and a family history of breast or ovarian cancer. cReference category for all forms of HRT.\n\nThe effect of ORT was essentially uniform across the different\nlevels of duration of use, irrespective of surgical history or histo-\nlogical type of tumour. There was, however, a distinct decrease in\nrisk of ovarian cancer seen with increasing time since last use\namong women with an intact reproductive system. In this group,\ncompared with never-users of HRT, current ORT users were at\nhighest risk (OR 3.92; 95% CI 1.32–11.6); women who had ceased\nuse within the last 5 years were at intermediate risk (OR 2.09; 95%\nCI 0.90–4.89); and women who had last used oestrogen more than\n5 years ago were at lowest risk (OR 1.45; 95% CI 0.90–2.32). No\nsimilar trend in risk with time since last use of ORT was apparent\nin those with a previous hysterectomy or tubal sterilization.\nOestrogen doses were too poorly reported by subjects for a useful\nanalysis.\nNo associations similar to those seen with unopposed oestrogen\nwere found for any other form of HRT.\nDISCUSSION\nA non-significant 20% increase in risk of ovarian cancer was found\nto be associated with ever use of HRT, which is consistent with\nsummary estimates reported in a recently published meta-analysis\non the subject (Garg et al, 1998). Ovarian carcinomas are, however,\na heterogeneous collection, morphologically, pathogenically and\nHRT and epithelial ovarian cancer 561\nBritish Journal of Cancer (1999) 81(3), 559–563© 1999 Cancer Research Campaign\nTable 2 Duration of use and time since last use of hormone replacement therapy, overall and specifically for unopposed oestrogen and oestrogen in\ncombination with progestogen, among all cases of epithelial ovarian cancer and controls with crude and adjusted odds ratios\nHRT use Cases Controls Crude OR Adjusted OR a\n(n = 793) ( n = 855) (95% CI) (95% CI)\nNever usedb 83.5% 82.6% 1.00 1.00\nUsed any HRT forc\n<1 year 3.5% 5.1% 0.68 (0.42–1.10) 0.70 (0.41–1.19)\n1–3 years 5.3% 4.9% 1.07 (0.69–1.66) 1.11 (0.88–1.41)\n>3 years 6.3% 6.3% 0.99 (0.66–1.47) 1.13 (0.91–1.41)\nLast used any HRT\nCurrent users 8.6% 11.0% 0.77 (0.55–1.07) 0.95 (0.67–1.36)\n1–5 years ago 3.5% 2.7% 1.30 (0.74–2.28) 1.28 (0.95–1.72)\n>5 years ago 3.0% 2.7% 1.11 (0.62–1.99) 1.03 (0.75–1.42)\nUsed unopposed oestrogen for\n< 1 year 2.4% 2.6% 0.92 (0.49–1.72) 0.93 (0.47–1.85)\n1–3 years 3.4% 3.0% 1.11 (0.64–1.92) 1.13 (0.85–1.50)\n>3 years 2.7% 3.2% 0.83 (0.46–1.48) 0.92 (0.67–1.25)\nLast used unopposed oestrogen\nCurrent users 3.5% 4.3% 0.81 (0.49–1.33) 0.94 (0.56–1.60)\n1–5 years ago 2.3% 1.8% 1.28 (0.64–2.56) 1.19 (0.82–1.72)\n>5 years ago 2.6% 2.7% 0.97 (0.53–1.78) 0.94 (0.67–1.30)\nUsed oestrogen and progestogen in combination for\n<1 year 1.1% 1.6% 0.69 (0.29–1.59) 0.80 (0.33–1.97)\n1–3 years 1.9% 1.6% 1.14 (0.55–2.39) 1.36 (0.92–2.00)\n>3 years 1.8% 1.6% 1.07 (0.50–2.25) 1.33 (0.88–2.00)\nLast used oestrogen and progestogen in combination\nCurrent users 3.9% 4.6% 0.85 (0.52–1.37) 1.24 (0.73–2.09)\nPast users 0.9% 0.4% 2.49 (0.64–9.66) 2.02 (0.99–4.14)\naORs adjusted for age, education, area of residence, BMI, hysterectomy, tubal sterilization, talc use in perineal region, smoking status, duration of OCP use,\nparity and a family history of breast or ovarian cancer. bReference category for all forms of HRT. cEleven cases and nine controls did not report on duration of\nHRT use (and hence time since last use could also not be calculated).\nTable 3 Ever use of unopposed oestrogen replacement therapy among women with various histological types of ovarian cancer and among controls\nHistological type ORT users Never used Crude OR Adjusted OR b\nHRTa (95% Cl) (95% CI)\nControls 75 (8.8%) c 706 (82.6%) 1.00 1.00\nAll cancers 68 (8.6%) 662 (83.5%) 0.97 (0.69–1.36) 1.27 (0.86–1.88)\nSerous 36 (8.7%) 340 (81.9%) 1.00 (0.66–1.51) 1.26 (0.78–2.03)\nEndometrioid/clear cell 18 (11.0%) 132 (80.5%) 1.28 (0.74–2.22) 2.56 (1.32–4.94)\nMucinous 6 (5.3%) 106 (93.0%) 0.53 (0.23–1.25) 0.85 (0.32–2.23)\nUndifferentiated 4 (7.4%) 45 (83.3%) 0.84 (0.29–2.39) 0.51 (0.14–1.89)\nMixed epithelial/mesodermal 4 (9.1%) 38 (86.4%) 0.99 (0.34–2.85) 1.15 (0.35–3.80)\naExcludes women who used another form of HRT (63 cases, 74 controls). bORs adjusted for age, education, area of residence, BMI, hysterectomy, tubal\nsterilization, talc use in perineal region, smoking status, duration of OCP use, parity and a family history of breast or ovarian cancer. cPercentages are based on\nrow totals (including users of other forms of HRT).\n\naetiologically (Russell, 1994). Many epidemiological studies have\nassumed homogeneity for the purposes of studying general aetio-\nlogical factors. An important outcome has been revealed in the\nhistological subtype analysis of this case–control study, namely, the\nmoderately strong association found between ever use of unop-\nposed oestrogen and endometrioid or, the closely related, clear cell\ntumours. This parallels the findings of a number of certain other\nstudies that looked specifically at endometrioid ovarian cancers (La\nVecchia et al, 1982; Weiss et al, 1982; Cramer et al, 1983; Risch,\n1996), although not all (Hartge et al, 1988; Booth et al, 1989;\nKaufman et al, 1989; Whittemore et al, 1992; Hempling et al,\n1997). Endometrioid and clear cell tumours are broadly regarded as\nthe equivalent of the same cancers in the endometrium (showing\nproliferative, or ‘pregnancy-related’ secretory changes), whereas\nserous tumours are neoplastic analogues of the fallopian tube\nmucosa and mucinous tumours are similar to those of the endo-\ncervix or colon (Scully, 1977). An analogy can therefore be drawn\nbetween our findings and those in endometrial cancer in which the\nrelationship with oestrogens is most marked for the endometrioid\nand least strong for serous papillary carcinomas (Bokham, 1983;\nSmith and McCartney, 1985).\nIt is clear that the aetiology of endometrial adenocarcinomas is\nlinked to the role of oestrogens (IARC, 1987). That endometrial\nadenocarcinoma is a ‘generic diagnosis’ and represents a heteroge-\nneous group of neoplasms is also clear, as it is for ovarian cancers\n(Greven and Corn, 1997).\nA finding of interest was that unopposed ORT was inversely\nassociated with all ovarian cancers in women with a prior\nhysterectomy (although not significantly), but positively associ-\nated in women without either hysterectomy or tubal sterilization.\nIn women with tubal sterilization no association was observed.\nWhile this is not the first study to report a lower risk of ovarian\ncancer associated with ORT among women with a hysterectomy\n(Whittemore et al, 1992; Hartge et al, 1988), others have reported\na larger effect of ORT in this group (Booth et al, 1989; Kaufman et\nal, 1989) or no difference (Risch, 1996). However, no previous\nstudy has examined the effect of ORT stratified by hysterectomy\nand tubal sterilization. The association seen here may imply that\nnon-contraceptive-oestrogen is a risk factor specifically for\nwomen with an intact genital tract. In all categories of surgical\nhistory, the endometrioid/clear cell group of tumours retained their\nelevated risk relative to other tumour types. In addition, among\nwomen with an intact genital tract, there was also a noticeable\ndecrease in risk with increasing time since last use of unopposed\noestrogen, consistent with the type of association with endometrial\ncarcinomas found by some (Green et al, 1996), though not all\nstudies (Paganini-Hill et al, 1989).\nThe positive links seen with ORT may reflect the suggested role\nof ovarian endometriosis in the aetiology of endometrioid or clear\ncell ovarian cancer (Sainz de la Cuesta et al, 1996). Post-\nmenopausal ORT could cause proliferation of the uterine\nendometrium, and in women with an intact genital tract, could\ntherefore increase the risk of retrograde menstruation (Metzger\nand Haney, 1989). This, in combination with the increased levels\nof circulating oestrogens from the oestrogen therapy, could\npromote ovarian endometriosis (Faulkner and Riemenschneider,\n1945). Women without hysterectomy or tubal sterilization who are\nprescribed unopposed oestrogen may therefore have an increased\nrisk of subsequent ovarian endometriosis and consequently have a\nhigher risk of endometrioid or clear cell ovarian cancer (Sainz de\nla Cuesta et al, 1996; Paulson, 1997).\nContrary to what might be expected given the well-established\nassociation between oestrogen and endometrial adenocarcinoma,\nunopposed oestrogen was prescribed for some women with an\nintact uterus in our study until quite recently. Year of first use is,\ntherefore, unlikely to be a major confounding factor. Further,\namong the 73 women who both used unopposed oestrogen and\nhad a hysterectomy, only three began use of oestrogen before the\noperation and all three continued use after surgery.\nWe do not believe that selection or recall biases could have\naccounted for our finding (Purdie et al, 1995). It is unlikely the\nresults could be explained by incorrect reporting of operations\nsince self-reported histories of hysterectomy and tubal sterilization\nwere checked against available medical reports and 96% and 88%\nrespectively were confirmed (Green et al, 1997 a). Furthermore,\nrecall of unopposed ORT would hardly differ from other forms of\n562 DM Purdie et al\nBritish Journal of Cancer (1999) 81(3), 559–563 © 1999 Cancer Research Campaign\nTable 4 Ever use of unopposed oestrogen replacement therapy in women with and without hysterectomy and tubal sterilization\nSubgroup of women ORT users Never used Crude OR Adjusted OR b\nHRTa (95% CI) (95% CI)\nWith hysterectomy:\nControls 51 (29.8%) c 114 (66.7%) 1.00 1.00\nAll cases 22 (19.5%) 80 (70.8%) 0.61 (0.35–1.09) 0.65 (0.34–1.24)\nEndometrioid/clear cell 5 (33.3%) 9 (60.0%) 1.24 (0.40–3.89) 1.75 (0.43–7.14)\nOther cases 17 (17.4%) 71 (72.4%) 0.53 (0.29–1.00) 0.57 (0.29–1.15)\nWithout hysterectomy but with tubal sterilization:\nControls 9 (5.8%) 123 (79.3%) 1.00 1.00\nAll cases 5 (5.7%) 74 (84.1%) 0.92 (0.30–2.86) 1.03 (0.27–3.96)\nEndometrioid/clear cell 2 (12.5%) 13 (81.3%) 2.10 (0.41–10.8) 2.72 (0.27–27.3)\nOther cases 3 (4.2%) 61 (84.7%) 0.67 (0.18–2.57) 0.84 (0.17–4.12)\nWithout hysterectomy or tubal sterilization:\nControls 15 (2.8%) 469 (88.7%) 1.00 1.00\nAll cases 41 (6.9%) 508 (85.8%) 2.52 (1.38–4.62) 3.00 (1.54–5.85)\nEndometrioid/clear cell 11 (8.3%) 110 (82.7%) 3.13 (1.40–7.00) 4.29 (1.67–11.1)\nOther cases 30 (6.5%) 398 (86.7%) 2.36 (1.25–4.44) 2.66 (1.31–5.37)\naExcludes women who used another form of HRT (63 cases, 74 controls). bORs adjusted for age, education, area of residence, BMI, talc use in perineal region,\nsmoking status, duration of OCP use, parity and a family history of breast or ovarian cancer. cPercentages are based on row totals (including users of other\nforms of HRT).\n\nHRT, which did not show similar effects. However, many of the\nassociations observed in this investigation of HRT were based on\nsmall numbers: only 18 cases of endometrioid or clear cell ovarian\ncancer had ever used unopposed ORT. When further subdivided by\nhistory of gynaecological surgery, this problem was even more\npronounced.\nPrior efforts to evaluate the effects of menopausal oestrogenic\nagents on ovarian carcinomas as a whole have been inconsistent,\nsome studies showing an increase in risk (Weiss et al, 1982) some\na decrease (Hartge et al, 1988; Hempling et al, 1997) and others no\nsignificant effect (La Vecchia et al, 1982; Kaufman et al, 1989;\nPurdie et al, 1995). These inconsistencies may be due to the histo-\nlogical type-specific associations observed in these data, as well\nas to the effect modification observed with certain genital tract\nsurgery. Those studies that examined the effect of menopausal\noestrogen separately for the different types of ovarian cancer have\nalso yielded contradictory findings. Those not finding a positive\nassociation more often used hospital based controls (Hartge et al,\n1988; Booth et al, 1989; Kaufman et al, 1989; Hempling et al,\n1997), which may reflect an excess of HRT use among hospital-\nized women (Rodriguez et al, 1995).\nIn conclusion, our results are consistent with the possibility that\nsome utero-ovarian carcinogenic influence, such as endometriosis,\nis promoted by ORT, increasing the risk of endometrioid and clear\ncell ovarian carcinomas.\nACKNOWLEDGEMENTS\nFunding for this study was provided by the Australian National\nHealth and Medical Research Council and the Queensland Cancer\nFund. We acknowledge the support and assistance of the Survey\nof Women’s Health study research group, namely the following\ncollaborators: Sèamus Campbell, Chris Dalrymple, Arthur Day,\nAlan Ferrier, Keith Free (deceased), Peter Grant, Paul Harnett, Phil\nHarvey, Roger Houghton, Tom Jobling, Peter MacCullum Cancer\nInstitute, Robert Planner, Tony Proietto, Robert Rome, John\nSolomon, Beatrice Susil, Gerry Wain, Gordon Wright; and\nresearch assistants: Carol Birks, Patrica Brisley, Paula Candlish,\nSuzanne Clarke, Lorraine Dommett, Kerri-Ann Lockwood,\nLauretta Luck, Helen Merry, Teresa Pangan, Roslyn Patterson,\nLouise Potter, Ann Ward, Mignon Watson, Mary-Ellen Yarker.\nThanks, of course, also to the doctors who allowed us to interview\ntheir patients and to all the women who participated in the study.\nREFERENCES\nBokhman JV (1983) Two pathogenetic types of endometrial carcinoma. Gynecol\nOncol 15: 10\nBooth M, Beral V and Smith P (1989) Risk factors for ovarian cancer: a\ncase–control study. Br J Cancer 60: 592–598\nCramer DW, Hutchinson GB, Welch WR, Scully RE and Ryan KJ (1983)\nDeterminants of ovarian cancer risk. I. Reproductive experiences and family\nhistory. J Natl Cancer Inst 71: 711–716\nDennerstein L, Shelly J, Smith AMA and Ryan M (1994) Hysterectomy experience\namong mid-aged Australian women. Med J Aust 161: 311–313\nFaulkner RL and Riemenschneider EA (1945) Reactivation of endometriosis by\nstilbestrol therapy. Am J Obstet Gynecol 105: 560\nGarg PP, Kerlikowske K, Subak L and Grady D (1998) Hormone replacement\ntherapy and risk of epithelial ovarian carcinoma: a meta-analysis. Obstet\nGynecol 92: 472–479\nGrady D, Gebretsadik T, Kerlikowske K, Ernster V and Petitti D (1995) Hormone\nreplacement therapy and endometrial cancer risk: a meta-analysis. Obstet\nGynecol 85: 304–313\nGreen A, Purdie D, Green L, Dick M-L, Bain C and Siskind V (1997a) Validity of\nself-reported hysterectomy and tubal sterilisation. Aust N Z J Public Health 21:\n337–340\nGreen A, Purdie D, Bain C, Siskind V , Russell P, Quinn M and Ward B (1997b)\nTubal sterilisation, hysterectomy and decreased risk of ovarian cancer. Int J\nCancer 71: 948–951\nGreen PK, Weiss NS, McKnight B, V oigt LF and Beresford SAA (1996) Risk of\nendometrial cancer following cessation of menopausal hormone use\n(Washington, United States). Cancer Causes Control 7: 575–580\nGreven KM and Corn BW (1997) Endometrial cancer. Curr Probl Cancer 21:\n65–127\nHartge P, Hoover R, McGowan L, Lesher L and Norris HJ (1988) Menopause and\novarian cancer. Am J Epidemiol 127: 990–998\nHempling RE, Wong C, Piver MS, Natarajan N and Mettlin CJ (1997) Hormone\nreplacement therapy as a risk factor for epithelial ovarian cancer: results of a\ncase-control study. Obstet Gynecol 89: 1012–1016\nInternational Agency for Research on Cancer (1987) Overall Evaluations of\nCarcinogenicity: an Updating of IARC Monographs, V ol 1–42. IARC Monogr\nEval Carcinog Risk Chem Hum (Suppl 7), pp 280–283. IARC: Lyon\nKaufman DW, Kelly JP, Welch WR, Rosenberg L, Stolley PD, Warshauer ME,\nLewis J, Woodruff J and Shapiro S (1989) Noncontraceptive estrogen use and\nepithelial ovarian cancer. Am J Epidemiol 130: 1142–1151\nLa Vecchia C, Liberati A and Franceschi S (1982) Noncontraceptive estrogen use\nand the occurrence of ovarian cancer. J Natl Cancer Inst 69: 1207\nMetzger DA and Haney AF (1989) Etiology of endometriosis. Obstet Gynecol Clin\nNorth Am 16: 1–14\nPaganini-Hill A, Ross RK and Henderson BE (1989) Endometrial cancer and\npatterns of use of oestrogen replacement therapy: a cohort study. Br J Cancer\n59: 445–447\nPaulson RJ (1997) Fertility drugs and ovarian epithelial cancer: the endometriosis\nhypothesis. J Assist Reprod Genet 14: 228–230\nPurdie D, Green A, Bain C, Siskind V , Ward B, Hacker N, Quinn M, Wright G,\nRussell P and Susil B (1995) Reproductive and other factors and risk of\nepithelial overian cancer: an Australian case-control study. Int J Cancer 62:\n678–684\nRisch HA (1996) Estrogen replacement therapy and risk of epithelial ovarian cancer.\nGynecol Oncol 63: 254–257\nRodriguez C, Calle EE, Coates RJ, Miracle-McMahill HL, Thun MJ and Health CW\n(1995) Estrogen replacement therapy and fatal ovarian cancer. Am J Epidemiol\n141: 828–835\nRussell P (1994) Surface epithelial-stromal tumors of the ovary. In: Blaustein’ s\nPathology of the Female Genital Tract, 4th edn, Kurman RJ (ed), pp. 705–782.\nSpringer-Verlag: New York\nSainz de la Cuesta R, Eichhorn JH, Rice LW, Fuller AF Jr, Nikrui N and Goff BA\n(1996) Histologic transformation of benign endometriosis to early epithelial\novarian cancer. Gynecol Oncol 60: 238–244\nScully RE (1977) Ovarian tumors. Am J Pathol 87: 686–720\nSmith M and McCartney AJ (1985) Occult, high-risk endometrial cancer. Gynecol\nOncol 22: 154\nWeiss NS, Lyon JL, Krishnamurthy S, Dietert SE, Liff JM and Daling JR (1982)\nNoncontraceptive estrogen use and the occurrence of ovarian cancer. J Natl\nCancer Inst 68: 95–98\nWhittemore AS, Harris R, Itnyre J and The Collaborative Ovarian Cancer Group\n(1992) Characteristics relating to ovarian cancer risk: collaborative analysis of\n12 US case-control studies. II. Invasive epithelial ovarian cancers in white\nwomen. Am J Epidemiol 136: 1184–1203\nHRT and epithelial ovarian cancer 563\nBritish Journal of Cancer (1999) 81(3), 559–563© 1999 Cancer Research Campaign","source_license":"CC0","license_restricted":false}