Risk factors of epithelial ovarian carcinomas among women with endometriosis: a systematic review

Transcription

1 AOGS SYSTEMATIC REVIEW Risk factors of epithelial ovarian carcinomas among women with endometriosis: a systematic review LINE H. THOMSEN 1, TINE H. SCHNACK 2, KRISTINA BUCHARDI 3, LONE HUMMELSHOJ 4, STACEY A. MISSMER 5,6, AXEL FORMAN 1 & JAN BLAAKAER 1 1 Department of Gynecology and Obstetrics, Aarhus University Hospital, Aarhus, Denmark, 2 Gynecologic and Obstetric Department, Rigshospitalet University Hospital of Copenhagen, Copenhagen, 3 Oncologic Department, Aarhus University Hospital, Aarhus, Denmark, 4 World Endometriosis Research Foundation, London, UK, 5 Department of Obstetrics, Gynecology and Reproductive Biology College of Human Medicine, Michigan State University, Boston, MA, and 6 Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA Key words Endometriosis, epidemiology, ovarian cancer, risk factor, systematic review Correspondence Line H. Thomsen, Department of Anesthesiology, Randers Regional Hospital, Skovlyvej 1, indgang 1, plan 3, 8930 Randers NØ, Denmark. lineholdgaard@gmail.com Conflict of interest The authors have stated explicitly that there are no conflicts of interest in connection with this review. Please cite this article as: Thomsen LH, Schnack TH, Buchardi K, Hummelshoj L, Missmer SA, Forman A, et al. Risk factors of epithelial ovarian carcinomas among women with endometriosis: a systematic review. Acta Obstet Gynecol Scand 2017; 96: Received: 15 February 2016 Accepted: 23 August 2016 DOI: /aogs Abstract Introduction. The objective of this review was to evaluate the published literature on epidemiologic risk factors for epithelial ovarian cancer among women with a diagnosis of endometriosis. Material and methods. A systematic literature search was conducted in PubMed and Scopus. Studies comparing epidemiologic risk factors of epithelial ovarian cancer among women with endometriosis were included. A quality assessment was conducted using the Newcastle-Ottawa Scale. Results. Eight of 794 articles met the inclusion criteria. A lower risk of epithelial ovarian cancer was observed in women with documented complete surgical excision of endometriotic tissue and suggested among women with unilateral oophorectomy. The use of oral contraceptives ( 10 years) may be associated with a lower risk of epithelial ovarian cancer among women with endometriosis, whereas older age at endometriosis diagnosis ( 45 years, pre- or postmenopausal), nulliparity, hyperestrogenism (endogenous or exogenous), premenopausal status at endometriosis diagnosis, solid compartments as well as larger size of endometrioma ( 9 cm in diameter at endometriosis diagnosis) were all associated with an increased risk of ovarian cancer. Conclusions. A subgroup of women with endometriosis characterized by endometriosis observed through surgery or imaging after the age of 45 years, nulliparity, postmenopausal status at endometriosis diagnosis, larger size of endometrioma (>9 cm) at endometriosis diagnosis, hyperestrogenism (endogenous or exogenous) and/or cysts with solid compartments may have an elevated risk of epithelial ovarian cancer. However, due to the limited number and size of studies in this area we cannot draw definitive conclusions. Further research into a risk factor profile among women with endometriosis is needed before clear recommendations can be made. Abbreviations: 95% CI, 95% confidence interval; BMI, body mass index; EOC, epithelial ovarian cancer; HR, hazard ratio; NOS, Newcastle Ottawa Scale; OC, oral contraceptives; OR, odds ratio; SIR, standardized incidence ratio. Introduction Endometriosis, estimated to affect 6 10% of women of reproductive age and 25 50% of women with infertility, is defined as endometrial-like tissue outside the uterine cavity often associated with pelvic pain and infertility (1). Endometriosis is a benign disease, although it shares similarities with malignant diseases such as attachment, damage, and invasion of target tissue and local and distant foci formation (2,3). ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

2 Endometriosis and ovarian cancer risk L.H. Thomsen et al. Epithelial ovarian cancer (EOC) is the deadliest of all gynecological cancers, often diagnosed beyond cure due to the late onset of symptoms and lack of good screening options (4 6). Several histological subtypes (serous, clear cell, endometrioid, and mucinous) exist, which are believed to develop through distinct pathological pathways (7). Endometriosis has consistently been associated with an increased risk of developing ovarian clear cell, low-grade serous, and ovarian endometrioid adenocarcinomas (8 12). A large pooled analysis among women with a selfreported diagnosis of endometriosis found a 46% increased risk (OR = 1.46; 95% CI , p < ) compared with women without endometriosis. The risk of EOC among women with endometriosis might, however, be higher, because a recent study found a stronger relation with EOC according to the strength of evidence behind the definitive endometriosis diagnosis (verified with histology). This finding suggests an impact of endometriosis misclassification that attenuated the observed association (13). There is pathological evidence suggesting that a subtype of endometrioid and clear cell carcinomas may develop from endometriotic precursor lesions within the ovaries (14,15). The endometriotic cysts are characterized by cyclic hemorrhage into the cysts leading to accumulation of blood components, for example iron, known to induce inflammation by oxidative stress. This inflammatory microenvironment may cause DNA damage, possibly leading to malignant transformation (16 18). Findings of subtype-specific molecular alterations within these tumors and endometriotic precursor lesions support this theory, for example somatic gene mutations in ARID1A, PIK3CA, b-catenin, and PTEN and loss of heterozygosity (19 22). Low-grade serous ovarian carcinoma is hypothesized to arise from endosalpingiosis or by shedding of normal fallopian tube epithelium into inclusion cysts at time of ovulation (7,14,15), but the role of this mechanism in endometriosis-related ovarian cancer is unknown. Evidence of the causality between endometriosis and ovarian cancer has not yet been established. The purpose of this review was to critically review the literature regarding EOC risk factors among women with endometriosis to identify potential subgroups of women with endometriosis at high and/or low risk of EOC (Figure 1). Material and methods Inclusion criteria were specified before the literature search and constituted retrospective and prospective cohort and case control studies written in English, Danish, Swedish, or Norwegian and published after 1990 EOC Endometriosis associated EOC Endometriosis Figure 1. We wish to compare the risk factor profile among women with endometriosis vs. women with endometriosis associated epithelial ovarian cancer (EOC). [Color figure can be viewed at wileyonlinelibrary.com] examining any risk factor of developing EOC among women with a diagnosis of ovarian endometrioma, where the primary population of included women constituted women with a diagnosis of endometriosis. The two electronic databases PubMed and Scopus were searched using a detailed search string combining the words: endometriosis, risk factor, epidemiology, complication, clear cell, endometrioid adenocarcinoma, serous, and EOC. The search strategy was prepared in collaboration with two librarians and conducted by one of the authors (LHT) on 9 December For the detailed search strategy see Appendix S1. Two investigators independently screened title and abstracts for eligibility; any discrepancies resulted in retrieval of the full text, disagreements among full-text versions were resolved by consensus or by involving a third party. A further analysis of the references cited in included articles and relevant reviews was performed, leading to the inclusion of two additional articles. Study characteristics (design, inclusion period, number of cases and controls, age, diagnosis and location of endometriosis, ovarian cancer histology, statistical analysis and confounding variables) (Table 1) and risk estimates on all published variables were extracted by two of the authors Key Message A subgroup of women with endometriosis [nulliparous, 45 years with visualized endometriosis, or with a large endometrioma (>9 cm)], could be offered increased surveillance for ovarian cancer development. However, more research is needed to make clear recommendations with regards to systematic screening or counseling of this population. 762 ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

3 L.H. Thomsen et al. Endometriosis and ovarian cancer risk Table 1. Summary of included articles. Author (publication year) Country Study design Inclusion criteria Kadan et al. (2015) (28) United States Retrospective, case control study a Women who underwent surgery between 2004 and 2013 with a prediagnosed pelvis mass or ovarian cyst Kobayashi et al. (2008) (29) Japan Retrospective and prospective components, cohort study (register based) a Ultrasound verified ovarian endometrioma between 1985 and 1995 Melin et al. (2013) d (26) Sweden Retrospective, case control study (registerbased) a All women discharged from a Swedish hospital with an endometriosis diagnosis registered in the NSIR between 1969 and 2007 Reference standard for endometriosis/ location of endometriosis Ovarian cancer histology Number of cases Risk factors explored Main study findings c Surgical/Ovarian 14 clear cell 17 endometrioid 7 mucinous 6 serous 1 other-squamous high grad 1 undifferentiated Clinical by vaginal ultrasound/ Ovarian Surgical/372 ovarian 129 peritoneal or other organs 135 adenomyosis 18 clear cell 16 endometrioid 4 mucinous 5 serous 3 unknown 42 Age, BMI, menopause, symptoms, CA 125, cyst size, solid components, cyst complexity, free fluid and laterality 46 Age, menopause, parity, marital status, use of hormones, smoking, family history of cancer, diameter of endometrioma EOC 220 Hysterectomy, one-sided oophorectomy, sterilization, complete removal of endometriosis, use of combined oral contraceptives, gestagens, danocrine, gonadotropin-releasing hormone agonists and hormone replacement therapy Age: Per 5 year increase OR 2.17 ( ) Tumor size: Per 1 cm increase OR 1.21 ( ) Solid complexity OR 23.7 ( ) Age 45 years HR 8.12 ( ), Menopause HR 8.68 ( ) Parity (yes vs. no) HR 2.17 ( ) Tumor size HR 13.5 ( ) One-sided oophorectomy OR 0.10 ( ) Complete extirpation of endometriosis (no vs. yes) OR 0.29 ( ) ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

4 Endometriosis and ovarian cancer risk L.H. Thomsen et al. Table 1. Continued Author (publication year) Country Study design Inclusion criteria Rossing et al. (2008) (10) Melin et al. (2006) d (9) United States Retrospective, case control study Sweden Retrospective, cohort study (registerbased) a Cases: Women diagnosed with primary invasive or borderline epithelial ovarian tumor from 2002 to 2005 Controls: RDD sampled in five-year categories, oneyear calendar interval and two county strata 2:1 All Swedish women with a overnight stay in a public hospital and a first-time discharge diagnosis of endometriosis between 1969 and 2000 Reference standard for endometriosis/ location of endometriosis Ovarian cancer histology Number of cases Risk factors explored Main study findings c Self-reported (interview)/not published 31 serous 26 endometrioid/clear cell 7 other, excluding mucinous 64 Ovarian surgery No significant results Mainly surgical (81%)/Any location Ovarian cancer 122 Hysterectomy No significant results 764 ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

5 L.H. Thomsen et al. Endometriosis and ovarian cancer risk Table 1. Continued Author (publication year) Country Study design Inclusion criteria Modugno et al. (2004) b (24) United States Retrospective case control study, based on 4 population based case control studies Ness (31): Cases: women age with EOC within 6 months prior to interview between May 1994 to July Controls: 65 years RDD Goodman (30): Cases: all patients diagnosed with primary epithelial cancer between July and June on Oahu or Los Angeles. Controls: women interviewed by the Health Surveillance Program without ovarian cancer Moorman (32): Cases: women with newly diagnosed EOC in the North Carolina Ovarian Cancer Study, aged years, no prior history of ovarian cancer and resided in a 48- country arear. Controls: RDD and Health Care Financing Administration Kupelian (33): EOC diagnosed between June to January , in residents of Cook County age 18 74, Controls: community controls, not specified further Reference standard for endometriosis/ location of endometriosis Ovarian cancer histology Number of cases Risk factors explored Main study findings c Self-reported medical diagnosis (interview)/not published EOC 177 Births, use of oral contraceptives, tubal ligation, hysterectomy. Oral contraceptives: 10 years 0.21 ( ) Births: 1 2 births OR 0.31 ( ), 3 births OR 0.22 ( ) ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

6 Endometriosis and ovarian cancer risk L.H. Thomsen et al. Table 1. Continued Author (publication year) Country Study design Inclusion criteria Reference standard for endometriosis/ location of endometriosis Ovarian cancer histology Number of cases Risk factors explored Main study findings c Cottreau et al. (2003) b (25) Zanetta et al. (2000) (35) United States Retrospective, case control study. Pooled from two studies United States Retrospective case control study a Ness (31): Cases: women age with EOC within 6 months prior to interview between May 1994 to July Controls: 65 years RDD Goodman (34): Cases: omen of Caucasian, Asian, native Hawaiian or Samoan origin, between 18 and 84 years old diagnosed with primary epithelial cancer between July and June in any major hospital center on Oahu. Controls: women interviewed by the Health Surveillance Program without ovarian cancer Cases: Patients with malignant tumors arising from endometriosis, surgically verified at the Mayo Clinic between 1986 and Controls: surgical and pathological verified endometrioses during the same study interval Self-reported medical diagnosis (questionnaire)/ Not published Pathological via surgery/cases: at cancer location. Controls: 36 ovarian, 26 other location EOC 120 Use of oral contraceptives, infertility drug, Danazol or Lupron, hysterectomy 16 endometrioid 4 clear cell 2 serous 1 mucinous 8 other 31/62 Parity, menopause, smoking, history of endometrioses, operation for endometriosis, hormones, use of unopposed estrogens, personal history of cancer, BMI Oral contraceptives (never vs. ever) OR 0.5 ( ) Unopposed estrogen or BMI 27: p = 0.05 OR and 95% CI not published BMI, body mass index; EOC, epithelial ovarian cancer; OC, oral contraceptive; NSR, National Swedish Inpatient register; RDD, random digit dialing. a Retrospective studies, but data collected prospective without knowledge of the current studies. b Both studies use data from Ness et al. in their analysis and both publications by Goodman are based on the same population and inclusion period. c Only significant results included. d Overlap in population, both studies have used Swedish registers. 766 ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

7 L.H. Thomsen et al. Endometriosis and ovarian cancer risk and divided into three categories: hormonal, surgical, and other. The search resulted in a total of 741 articles after removing duplicates. Seventy-nine articles were excluded having been published either before 1990 or in languages other than English, Danish, Swedish, or Norwegian. Of the remaining 662 studies causes for exclusion included reviews and case reports: 133 were neither regarding ovarian cancer nor endometriosis; 158 were only regarding endometriosis or ovarian cancer; 24 were about adenomyosis; 144 described endometriosis (with and without malignant transformation) and cancers in other locations than the ovaries; and 100 articles on endometriosis and ovarian cancer did not examine risk factors). One hundred and three articles were identified for full text review, a detailed description of exclusion criteria can be seen in the Supporting Information (Appendix S2) (one was neither regarding ovarian cancer nor endometriosis, 18 were only regarding endometriosis or ovarian cancer, 76 articles were about endometriosis and ovarian cancer but were not primary research on risk factors including reviews, case reports, published comments, editorials, etc.). In total, eight studies met the inclusion criteria from this literature search (Figure 2). All included studies were retrospective case control or cohort studies, relying on databases, interviews or questionnaires. A quality assessment was conducted using the Newcastle Ottawa Scale (NOS) for nonrandomized studies (23) (Tables 2 and 3). NOS assesses the quality of a case control study based on selection of cases and controls, comparability of the groups, and the ascertainment of the exposure or outcome of interest and the quality of of a cohort study based on selection of the cohort, comparability on the basis of design/analysis and outcome. For comparability, age was chosen as the most important factor, with oral contraceptive (OC) use, family history of cancer, and gravidity as additional factors. Quantifying response rate, a point was given if response rate among both cases and controls was above 60%. Regarding cohort studies an adequate follow-up was defined as 10 years among a cohort of women with an average age of 40 years. All studies were given a score between 0 and 9 points by two of the investigators. In this systematic review a score of 7 points equals a high score (i.e. a high quality in methodology), a score between 4 and 6 was of moderate quality, and 0 3 constituted low quality. Two of the included case control studies pooled data from two or more previous research articles. In such cases, each individual study was scored, and a total score was given by the lowest common denominator. Other study limitations are described in the Discussion section. Data published by Rossing et al. (10) were used to calculate the effect of ovarian surgery in relation to the risk 53 duplicates 79 published before 1990 or in other language 133 Neither regarding ovarian cancer nor endometriosis 158 Regarding endometriosis or ovarian cancer 24 adenomyosis 144 Endometriosis in other location then the ovaries 100 Endometriosis and ovarian cancer but not primary research on risk factors 1 Neither regarding ovarian cancer nor endometriosis 16 Regarding endometriosis or ovarian cancer 76 Endometriosis and ovarian cancer but not primary research on risk factors 1 Endometriosis in other location then the ovaries 3 in other language than English, Danish, Swedish or Norwegian 9th December published studies from PubMed and Scopus full text required 8 primary studies included: 6 from literature search and 2 retrieved from reference list Figure 2. Flowchart of literature search conducted 9 December, of EOC (logistic regression) and when studies were based on the same population of women and variables were coinciding, data from the largest study was used (9,24 26). The diversity and limitations of the included studies did not make it reasonable to conduct a meta-analysis (Table 1). A descriptive approach was adopted concerning all published risk estimates (odds ratio, standardized ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

8 Endometriosis and ovarian cancer risk L.H. Thomsen et al. Table 2. Quality assessment by the Newcastle Ottawa Scale (NOS); case and control studies. Selection Comparability Exposure Author (publication year) Adequate case definition Representativity of the cases Selection of controls Definition of controls Comparability of cases and controls on the basis of study design Comparability of cases and controls on the basis of analysis Ascertainment of exposure Same method of ascertainment of exposure case control Nonresponse rate Total Kadan et al. (2015) (28) Rossing et al. (2008) (10) Melin et al. (2013) (26) Modugno et al. (2004) (24) Cottreau et al. (2003) (25) Ness et al. (31) Goodman et al. (30) Moorman et al. (32) Kupelian et al. (33) Ness et al. (31) Goodman et al. (34) Zanetta et al. (2000) (35) the feature is present, the feature is absent. For comparability age was chosen as the most important factor, and oral contraceptives use, family history and gravidity as additional factors. 768 ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

9 L.H. Thomsen et al. Endometriosis and ovarian cancer risk incidence ratio, and hazard ratio). The PRISMA Statement for systematic reviews was followed (27). Results Table 1 summarizes the characteristics of the included studies, and Tables 2 and 3 show in detail the NOS quality assessment score. For a detailed outline of published data, Tables 4 6 summarize some of the published data on nonhormonal and nonsurgical risk factors, surgical treatment, as well as data on hormonal risk factors, respectively. In a recent case control study by Kadan et al. (28) (42 cases/96 controls) age and solid complexity were significantly associated with ovarian cancer arising in an endometriotic cyst (OR 2.17; 95% CI , p = and OR 23.72; 95% CI , p < , respectively) in the multivariate analysis (Table 4). Regarding age, being 49 years at diagnosis of endometriosis was associated with ovarian malignancy (specificity 80.6%; 95% CI %, and sensitivity 82.9%; 95% CI %) (Table 4). Cyst size, CA 125, body mass index (BMI), ascites, laterality, and symptoms at diagnosis (except weight changes) did not differ between the two groups. Among the subgroup of women, who had an observational period before surgery, larger cyst size was associated with malignancy compared with smaller cyst size (4.2 cm vs. 1.0 cm, p = 0.02) (not shown in Table 4). These findings are in line with the results of a registerbased multicenter cohort study by Kobayashi et al. (29), which included a total of 6398 Japanese women with a diagnosis of ovarian endometrioma. At the time of endometriosis diagnosis, older age (<44 years vs. 45 years) (HR 8.12; 95% CI ), postmenopausal status (HR 8.68; 95% CI ), and greater diameter (<9 cm vs. 9 cm) of endometrioma (HR 13.5; 95% CI ) were all associated with a significant increased risk of EOC (Table 4). Interestingly, the incidence of EOC was 0%, 35%, and 65% in women with ovarian endometrioma <6, 6 9, and 9 cm in diameter, respectively. This study also found a trend towards increased risk of EOC among nulliparous women (HR 2.17; 95% CI , p = 0.212) (Table 4). No significant association between marital status, use of hormones (users vs. nonusers; not stratifying on type i.e. unopposed estrogen, estrogen progesterone combination) (Table 6), family history of cancer, or current or previous smoking history at diagnosis of ovarian endometrioma and ovarian cancer were observed. In a Swedish register-based study, Melin et al. found unilateral oophorectomy or complete extirpation of endometriotic tissue (regardless of location in the pelvic region) to have a strong protective effect on the risk of Table 3. Quality assessment by the Newcastle Ottawa Scale (NOS) cohort studies. Selection Comparability Exposure Total Score Adequacy of follow up of cohorts Was follow up long enough for outcomes to occur Assessment of outcome Comparability of cases and controls on design or analysis Comparability of cases and controls on design or analysis Demonstration that outcome of interest was not present at start of study Ascertainment of exposure Selection of the non exposed cohort Representativeness of the exposed cohort Author Kobayashi et al. (2008) (29) Melin et al. (2006) (9) + the feature is present, the feature is absent. For comparability age was chosen as the most important factor, and oral contraceptives use, family history and gravidity as additional factors. ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

10 Endometriosis and ovarian cancer risk L.H. Thomsen et al. Table 4. Risk of ovarian cancer in relation to nonhormonal and other nonsurgical investigated risk factors among women with endometriosis. b Author (publication year) Age Menopause (yes vs. no) Parity Solid complexity vs. not solid Tumor size Kadan et al. (2015) a (28) Kobayashi et al. (2008) (29) Modugno et al. (2005) (24) Zanetta et al. (2000) (35) Per 5 year increase OR 2.17 ( ) <44 years c (reference) 45 years c HR 8.12 ( ) NS OR 23.7 ( ) Per 1 cm increase OR 1.21 ( ) c Yes vs. no c <9 cm c HR 8.68 ( ) HR 2.17 ( ) (reference) 9 cm c HR 13.5 ( ) 0 births (reference) 1 2 births OR 0.31 ( ) 3 births OR 0.22 ( ) NS 0 births (reference) Odds ratio: OR (95% CI), Hazard ratio: HR (95% CI). NS: not significant result (OR, 95% CI and/or p-value not published). a Multivariate analysis. b CA 125: please see text, not in Table. c At endometriosis diagnosis. EOC (OR 0.10; 95% CI and OR 0.29; 95% CI , respectively) (Table 5) (26). Regarding unilateral oophorectomy, a numbers-needed-to-treat analysis showed that 62 women would have to go through the surgical procedure to prevent one case of ovarian cancer. Hormonal treatment was not associated with EOC risk in this study (Table 5). In a very small study by Rossing et al. (10) the association between ovarian surgery (10 cases/20 controls) and EOC (OR 0.69; 95% CI , p = 0.19) was not significant (Table 5). Melin et al. (9) included women with a first time hospital discharge diagnosis of endometriosis, of whom 122 were diagnosed with ovarian cancer. They found a lower standard incidence ratio (SIR) for the development of ovarian cancer among women, who had a hysterectomy at the time of or before the diagnosis of endometriosis (SIR 1.05; 95% CI ), compared with women who did not (SIR 1.54; 95% CI ) (Table 5). It was unclear whether the corresponding women never had a hysterectomy or could have had a hysterectomy after receiving the diagnosis of endometriosis. Modugno et al. (24) pooled data from four US population-based case control studies (30 33). Data were stratified on endometriosis and adjusted for age, parity, OC use, tubal ligation, family history of ovarian cancer, and study site. Associations, although nonsignificant, between tubal ligation (no vs. yes) or hysterectomy (no vs. yes) and the risk of developing ovarian cancer were observed with an OR of 0.7 (95% CI ) and 0.69 (95% CI ), respectively. Furthermore, a significant inverse association between OC use (ever vs. never users) and risk of ovarian cancer was found with an OR of 0.58 (95% CI ) for <10 years of contraceptive use and an OR of 0.21 (95% CI ) for 10 years of contraceptive use) (Table 6). The strength of the inverse association significantly increased with the number of years of contraceptive use (p-value for trend = 0.003). Furthermore, a significant trend was seen for number of births (p < 0.001), with a decrease in the risk of ovarian cancer with an increasing number of childbirths: OR 0.31 (95% CI ) for 1 2 births and OR 0.22 (95% CI ) for 3 births, compared with nulliparous women (Table 4). In the small study by Cottreau et al. (25) using data pooled from two case control studies (31,34) the use of Danazol showed an increased risk of ovarian cancer with an OR of 2.9 (95% CI ) (Table 6). No associations between the use of leuprolide/nafarelin (9 cases/8 controls), OC use, hysterectomy, or infertility drug use and risk of EOC were observed (Tables 5 and 6). However, factors indicating selection of Danazol vs. these other treatments were not evaluated. The study by Zanetta et al. (35) included women with histologically verified cancer anywhere in the pelvic region (i.e. not only restricted to ovarian cancer) arising in close physical proximity to the histologically verified endometriotic lesions, compared with age-matched women with endometriosis without a cancer diagnosis. Cancer locations were: ovaries (15 cases; 48.4%), serosa of the bowel (five cases; 16.1%), cul-de-sac (four cases; 12.9%), or other locations (seven cases; 22.6%). Women who were either obese (BMI > 27 kg/m 2 ) or who had 770 ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

11 L.H. Thomsen et al. Endometriosis and ovarian cancer risk Table 5. Risk of ovarian cancer in relation to previous surgical procedures among women with endometriosis. Author (publication year) Hysterectomy no vs. yes Sterilization/tubal ligation No vs. yes Complete extirpation of endometriosis No vs. yes Other non-radical procedures Melin et al. (2013) a (26) Rossing et al. (2008) (10) Melin et al. (2006) (9) Modugno et al. (2004) (24) Cottreau et al. (2003) (25) Zanetta et al. (2000) (35) OR 1.63 ( ) OR 0.76 ( ) b OR 0.29 ( ) a OR 0.10 ( ) a SIR 1.05 ( ) vs. SIR 1.54 ( ) d OR 0.69 ( ) OR 0.7 ( ) OR 0.6 ( ) NS One-sided oophorectomy OR 0.68 ( ) Ovarian surgery after endometriosis diagnosis c (yes vs. no) NS Non-radical extirpation Odds ratio: OR (95% CI), Hazard ratio: HR (95% CI); NS: not significant result, OR, 95% CI and/or p-value not published. a Multivariate analysis. b Only data published including women with adenomyosis. c Includes unilateral oophorectomy, excision of a cyst or part of an ovary. d Hysterectomy at the same time or before a diagnosis of endometriosis had a lower SIR 1.05 ( ) than women who did not have this surgical procedure in close relation or before endometriosis diagnosis 1.54 ( ). used unopposed estrogens (21 cases/28 controls) had 4.25 times greater risk of developing malignancy (p = 0.05) (Table 5). Furthermore, a trend towards a greater risk was observed among users of unopposed estrogen and a BMI > 27 (Table 6). No significant relation between menopausal status, parity, smoking, history of endometriosis, personal history of cancer, previous treatments with hormones or surgery and the risk of EOC were observed (some are shown in Tables 4 and 6). None of the analyses adjusted for potential confounding. Discussion To our knowledge, this systematic review is the first summary of available data regarding risk factors for EOC among women with endometriosis. Despite a systematic identification of pertinent publications and wide inclusion criteria, only eight studies met the inclusion criteria. In addition, most of the studies were based on small numbers of cases and controls (Table 1), which are prone to type two errors yielding insignificant and inconclusive results. Finally, four of eight studies (10,24,25,29) relied on self-reported or clinically verified/non-surgical diagnosis of endometriosis, both of which may be highly impacted by misclassification (36). Four studies (9,26,28,35) included (mainly) surgical diagnosis of endometriosis (Table 1), but could be temporally impacted by incidental observation of endometriosis concurrent with ovarian cancer discovery. Despite our initial goals, the small number of eligible published studies made it untenable to exclude publications based on study size or endometriosis location, and also resulted in inclusion of studies where the primary goal was not to evaluate risk factors among women with endometriosis or studies that did not distinguish between subtypes of ovarian cancer (Table 1). However, the number of non-eoc in these studies are few. We therefore consider this misclassification of minor importance. The above described aspects limit clear conclusions from this review. This being said, the study provides important information regarding risk factors of EOC among women with endometriosis, which is of interest in a clinical setting. The average score of the studies according to the NOS was 7 for the two cohort studies and 6.66 for the six case control studies (Tables 2 and 3), indicating a high quality/low risk of bias in methodology of the included cohort studies and a moderate quality/moderate risk of bias of the included case control studies. None of the studies reached a maximum of 9 stars, indicating the results could be affected by bias. As the studies did not examine the same risk factors it was not feasible to conduct a sub-analysis on the studies with a high NOS score ( 7) evaluating the reliability of our results. Two included studies investigated the role of age and found a significant predictive correlation between older age ( 45 years and >49 years, respectively) at diagnosis of endometriomas and the risk of ovarian cancer (28,29) ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

12 Endometriosis and ovarian cancer risk L.H. Thomsen et al. Table 6. Risk of ovarian cancer in relation to previous hormonal treatment among women with endometriosis. Author (publication year) Hormone replacement treatment (HRT) with estrogen BMI Oral contraceptives Danazol/Danocrine Kadan et al. (2015) (28) Melin et al. (2013) a (26) Modugno et al. (2004) (24) Cottreau et al. (2003) (25) Zanetta et al. (2000) (35) Never users (reference) 1 6 months OR 0.65 ( ) >6 months OR 2.06 ( ) NS Unopposed estrogen or BMI 27 p = 0.05 OR and 95% CI not published NS <27 vs. 27 NS Never users (reference) 1 12 months OR 0.81 ( ) b >12 months OR 0.96 ( ) b Never users (reference) <10 years 0.58 ( ) 10 years 0.21 ( ) Never users (reference) Never users (reference) 1 6 months OR 1.02 ( ) b >6 months OR 1.32 ( ) b Neither medications (reference) OR 0.5 ( ) OR 2.9 ( ) NS Odds ratio: OR (95% CI), Hazard ratio: HR (95% CI). NS: not significant result, OR, 95% CI and/or p-value not published. a Multivariate analysis. b Including women with adenomyosis. (Table 4). A recent retrospective Taiwanese cohort study by Wang et al. (37) found an increase in incidence rate of EOC with advancing age among women with and without a diagnosis of endometriosis (surgically pathologically verified). Interestingly, a significant increased risk of EOC among women with endometriosis aged <30 or 50 years was found when comparing the incidence rate of EOC with age-matched controls without endometriosis (HR 9.63, 95% CI and HR 4.97, 95% CI ), respectively). Several studies have shown that endometriosis-associated cancers develop at younger ages, although the risk still appears to be extremely low especially for younger women with endometriosis (age <45) (9,38,39). This is supported by Erzen et al. who found a higher percentage of ovarian cancer arising in ovarian endometriosis among women >50 years of age compared with younger women (40). The genetic precancerous mutations within endometriotic lesions, most likely caused by continuous inflammation in the microenvironment combined with the increased gonadotropin levels, could make perimenopausal women with endometriosis susceptible to EOC earlier in life, explaining why some women with endometriosis may be younger at ovarian cancer diagnosis. Only one of three studies found a significant association between postmenopausal status and ovarian cancer; however, this study did not adjust for age, leaving the possibility that the increased risk could be a function of age and not menopause (28,29,35) (Table 4). A diagnosis of endometriosis among postmenopausal women is rare due to reduced endogenous estrogen levels and thereby a decrease in symptoms (41,42). Endometriosis is an enigmatic disease where the severity of symptoms and the extent of disease based on surgical visualization (rasrm staging system) are not necessarily related (43). Furthermore, there is a well-known diagnostic delay for women with endometriosis (44). Hence, older age at endometriosis diagnosis is not necessarily equivalent with late onset of endometriosis nor does it enable us to draw any conclusions with regards to severity of symptoms or disease stage of endometriosis and risk of EOC. Among the general population, postmenopausal women have a twofold increased risk of ovarian cancer with a peak incidence after menopause (45,46). One theory suggests that the elevated level of gonadotropins possibly stimulates, and thereby leads to, the development of ovarian cancer (47,48). 772 ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

13 L.H. Thomsen et al. Endometriosis and ovarian cancer risk A theory behind the observed increased risk of ovarian cancer with older age and menopause among women with a diagnosis of endometriosis may be explained by the fact that women diagnosed with endometriosis late in life have not had any previous treatment aimed at reducing symptoms by reducing endometriotic activity and thereby impacting the inflammatory microenvironment. Alternatively, women diagnosed with late-onset endometriosis may indeed represent a subgroup who have a greater risk of malignant transformation because the natural course of endometriosis is regression rather than progression after the menopause. In summation, older age or menopausal status at endometriosis diagnosis may be suggestive of an increased risk of malignancy. Therefore, close monitoring of women with a late diagnosis of endometriosis, as well as women with continuous signs or symptoms of endometriosis after menopause, may be considered. Three studies investigated the association between parity and ovarian cancer among women with endometriosis (24,29,35). Two studies (24,35) found a protective effect with increasing number of childbirths although only one reached a significant level (24) (Table 4). Endometriosis is associated with infertility, though causality has not been established. The etiology may be multifactorial, including altered inflammatory environment, anatomical abnormalities, immunologic, hormonal and genetic aspects (41,49 52). In a cohort of women seeking treatment for infertility, Brinton et al. found that women with endometriosis had the highest risk of developing ovarian cancer (SIR 2.48; 95% CI ) compared with the general population of women. This estimate was strengthened when analyses were restricted to women with endometriosis and primary infertility (SIR 4.19; 95% CI ), whereas the risk of EOC among women with endometriosis, who had conceived a child previously, was similar to that in the general population (SIR 1.05; 95% CI ) (53). Additionally, Stewart et al. found that women with endometriosis had an increased risk of ovarian cancer (HR 2.33; 95% CI ) compared with women without endometriosis in a cohort of women seeking fertility treatment. Among women with endometriosis who remained childless, this risk was increased threefold (HR 3.11; 95% CI ), whereas women with endometriosis who succeeded in conceiving a child had a lower and not significantly increased risk of EOC (HR 1.52; 95% CI ) (54). These findings, supported by Ness et al. (55), underpin a link between endometriosis, infertility, and ovarian cancer, suggesting that nulliparous women with endometriosis have an elevated risk of developing ovarian cancer. When assessing the risk of malignancy after the discovery of an ovarian cyst in an outpatient setting, it is recommended to use the Risk of Malignancy Index (RMI), where imaging plays an important role, along with menopausal status and CA 125 (56,57). Ultrasonographic findings of solid components in an ovarian cyst, symptomatic or not, are known to strongly indicate malignancy among the general population of women (58,59). The study by Kadan et al. confirms that this also applies for women with endometriomas (28) (Table 4). None of the reviewed studies found a statistically significant effect of hysterectomy and sterilization on the risk of EOC among women with endometriosis (Table 5) (9,24 26). A hypothesis could be that endometriosis has already caused cellular damage initiating the inflammatory response leading to malignant transformation, making potential prevention (hysterectomy or tubal ligation implied) of further endometriotic implantation futile. However, hysterectomy and tubal ligation (especially before the age of 35 years) have in some, albeit not all, previous studies been shown to be protective interventions of EOC among the general population (60 63). The effect of hysterectomy (on nonmalignant indication), or of tubal ligation in relation to women with endometriosis, may be reduction in the development of new endometriotic foci in premenopausal women by blocking retrograde menstrual flow, a reduction of the blood supply to the ovaries thereby reducing estrogen production and/or a reduction of anti-muc-1 antibodies due to tubal ligation (61,64). Two studies investigated the relation between complete surgical extirpation of endometriosis and the risk of EOC. Zanetta et al. did not find a significant association (35); however, the largest and best designed study found a significant correlation between complete excision of endometriosis, or unilateral oophorectomy, and a lower risk of EOC (26). Two smaller studies investigating the effect of ovarian surgery and the risk of EOC found no significant association (10,35) (Table 5). Interestingly, excision of endometriosis located in extra-ovarian pelvic areas also seems to decrease the risk of ovarian cancer (26). This diminishes the strength of the data suggesting that EOC is the result of malignant transformation of only ovarian endometriomas. However, radical excision of endometriosis in the pelvic region can be difficult and carries the risk of severe complications, and it does not reduce the risk of small (microscopic) foci to be overlooked nor does it prevent possible recurrence of endometriosis. Full laparoscopic excision of an endometrioma, including the cyst wall, has shown significant benefits regarding endometrioma recurrence, reduced risk of recurrent dysmenorrhea, dyspareunia, non-menstrual pelvic pain, and further surgical requirements in the future (65,66). Surgical intervention may also reduce the cancer risk, as ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

14 Endometriosis and ovarian cancer risk L.H. Thomsen et al. removal of all visible endometriotic foci possibly decreases the risk of malignant transformation (67 69). Furthermore, extensive surgery may reduce the inflammatory responses and level of mediators known to be involved in the development of ovarian cancer (70,71). Results are conflicting regarding surgery of endometriosis and the risk of EOC, making a clear conclusion or recommendation impossible. The risks of radical surgery should be weighed against the benefits in terms of a possible lowered risk of malignant transformation and the decrease in the woman s symptoms before an operative procedure. Two studies, including the best designed study, did not find a significant association between exogenous hyperestrogenism (hormone replacement therapy) and EOC among women with endometriosis (26,35). Regarding endogenous hyperestrogenism (high BMI) Kadan et al. and Zanetta et al. found no significant association in relation to ovarian cancer risk (28,35) (Table 6). However, after pooling women with unopposed estrogen (hormone replacement therapy) or BMI > 27 kg/m 2, Zanetta et al. found a significant increase in risk of ovarian cancer (Table 6). Unopposed estrogen is thought to stimulate endometriotic growth and maybe recurrence or reactivation of endometriotic foci after surgery (72 74) causing hyperplasia, atypical endometriosis and, at worst, leading to malignant transformation (75,76). Estrogen has shown EOC cell growth stimulation and higher invasion rates after estrogen application in vitro (77). Known risk factors for the development of ovarian cancer among the general population of women is associated with high levels of estrogen such as long menstrual life span, early age at menarche, late age at menopause (78 81) and use of HRT with estrogen-only substitution (82 84). Three studies [not counting Cottreau et al. (25) because of the coinciding study population with Modugno et al. (24)], investigated the effect of OC use among women with endometriosis. Only one found a significant association (24,26,35). The discrepancy may result from differences in the cut-off point (Table 6); as the effect of OCs may be related to long-term usage. Indeed the study by Modugno et al. compared <10 years vs. 10 years of OC use, whereas the cut-offs in the other studies were shorter. In the study by Kobayashi et al. the use of hormones (yes vs. no) was examined. Due to few hormone users in the case and the control groups (seven estrogen progesterone users) the result was difficult to interpret (29). Oral contraceptive use and pregnancy, especially longterm use (>10 years) and increasing number of childbirths, respectively, are both known protectors of ovarian cancer among women in the general population (85 88). Continuous OC use is often a first-line treatment in endometriosis, reducing pain and, possibly, further progression (89). The use of OCs and pregnancy lead to reduced ovulatory cycles contributing to the incessant ovulatory theory described above. Pregnancy suppresses gonadotropin levels, increases estrogen and progesterone levels, and stops the possible retrograde transportation of exogenous substances through the fallopian tube (7,90). Progesterone is hypothesized to be protective of ovarian cancer by reducing gonadotropin and androgen levels, which is in alignment with the observation that progesterone-only OCs also seem to be protective of ovarian cancer (91). Androgens may result in the malignant transformation of ovarian surface epithelium cells as they have been shown to stimulate growth (92). The very small sample size in the study by Cottreau et al. (13 cases/5 controls) does not enable us to justify any conclusion based on the borderline increased risk of ovarian cancer (25), and Melin et al. found no significant association between Danazol use and EOC (26). Moreover, in clinical practice Danazol has now largely been replaced by medical therapies with more favorable adverse effects (Table 5). In summary, risk factors for the development of EOC among women with endometriosis seem to include older age at the time of diagnosis of endometriosis ( 45 years or 49 years) (28,29), cysts with solid compartments (28), postmenopausal status at the time of diagnosis of endometriosis (29), larger size of endometrioma ( 9 cm) at the time of diagnosis of endometriosis (29), nulliparity, and hyperestrogenism (endogenous and/or exogenous) (35). Use of OCs may be preventive of EOC among women with endometriosis, although a significant association was observed in only one study, as was a protective effect of greater parity (24). One study found a significant protective effect of surgical intervention with unilateral oophorectomy and/or macroscopically complete excision of endometriosis, including foci in other locations than the ovaries (26). Kadan et al. found a low-risk group of ovarian malignancy among women who were under 49 years, had CA 125 <43 U/mL, cysts without solid compartments, and cyst size <11 cm (OR 0; 95% CI , p = 0.008). Conclusion A subgroup of women characterized by a diagnosis of endometriosis after the age of 45 years, nulliparity, postmenopausal status, larger size of endometrioma (>9 cm), hyperestrogenism (endogenous or exogenous) and/or cysts with solid compartments at the time of diagnosis may have an elevated risk of EOC, though this subgroup has yet to be identified clearly. It is possible that this subgroup should be offered regular screenings with CA 125 and ultrasound. However, the limited number of published studies, the small study 774 ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)

15 L.H. Thomsen et al. Endometriosis and ovarian cancer risk sample sizes herein and the different study designs do not allow us to draw definitive conclusions and consequent recommendations. Further robust and larger studies are needed to confirm whether these initial findings are truly significant in women with endometriosis (93 96). Funding No specific funding. References 1. Giudice LC. Endometriosis. N Engl J Med. 2010;362: Nezhat F, Datta MS, Hanson V, Pejovic T, Nezhat C, Nezhat C. The relationship of endometriosis and ovarian malignancy: a review. Fertil Steril. 2008;90: Swiersz LM. Role of endometriosis in cancer and tumor development. Ann N Y Acad Sci. 2002;955: Jemal A, Bray F, Ferlay J. Global cancer statistics. CA Cancer J Clin. 2011;61: Smith LH, Morris CR, Yasmeen S, Parikh-Patel A, Cress RD, Romano PS. Ovarian cancer: can we make the clinical diagnosis earlier? Cancer. 2005;104: U. S. Preventive Task Force. Screening for ovarian cancer: recommendation statement. Ann Fam Med. 2004;2: Kurman RJ, Shih I-M. The origin and pathogenesis of epithelial ovarian cancer: a proposed unifying theory. Am J Surg Pathol. 2010;34: Pearce CL, Templeman C, Rossing MA, Lee A, Near AM, Webb PM, et al. Association between endometriosis and risk of histological subtypes of ovarian cancer: a pooled analysis of case control studies. Lancet Oncol. 2012;13: Melin A, Sparen P, Persson I, Bergqvist A. Endometriosis and the risk of cancer with special emphasis on ovarian cancer. Hum Reprod. 2006;21: Rossing MA, Cushing-Haugen KL, Wicklund KG, Doherty JA, Weiss NS. Risk of epithelial ovarian cancer in relation to benign ovarian conditions and ovarian surgery. Cancer Causes Control. 2008;19: Chang W-H, Wang K-C, Lee W-L, Huang N, Chou Y-J, Feng R-C, et al. Endometriosis and the subsequent risk of epithelial ovarian cancer. Taiwan J Obstet Gynecol. 2014;53: Sampson JA. Metastatic or embolic endometriosis, due to the menstrual dissemination of endometrisl tissue into the venous circulation. Am J Pathol. 1927;15: Lee W-L, Chang W-H, Wang K-C, Guo C-Y, Chou Y, Huang N, et al. The risk of epithelial ovarian cancer of women with endometriosis may be varied greatly if diagnostic criteria are different. Medicine (Baltimore). 2015;94: Veras E, Mao T-L, Ayhan A, Ueda S, Lai H, Hayran M, et al. Cystic and adenofibromatous clear cell carcinomas of the ovary. Distinctive tumors that differ in their pathogenesis and behavior: a clinicopathologic analysis of 122 cases. Am J Surg Pathol. 2009;33: Vang R, Shih L-M, Kurman RJ. Ovarian low-grade and high-grade serous carcinoma: pathogenesis, clinicopathologic and molecular biologic features, and diagnostic problems. Adv Anat Pathol. 2009;16: Yamaguchi K, Mandai M, Toyokuni S, Hamanishi J, Higuchi T, Takakura K, et al. Contents of endometriotic cysts, especially the high concentration of free iron, are a possible cause of carcinogenesis in the cysts through the iron-induced persistent oxidative stress. Clin Cancer Res. 2008;14: Reuter S, Gupta SC, Chaturvedi MM, Aggarwal BB. Oxidative stress, inflammation, and cancer: how are they linked? Free Radic Biol Med. 2011;49: Shigetomi H, Tsunemi T, Haruta S, Kajihara H, Yoshizawa Y, Tanase Y, et al. Molecular mechanisms linking endometriosis under oxidative stress with ovarian tumorigenesis and therapeutic modalities. Cancer Invest. 2012;30: Matsumoto T, Yamazaki M, Takahashi H, Kajita S, Suzuki E, Tsuruta T, et al. Distinct beta-catenin and PIK3CA mutation profiles in endometriosis-associated ovarian endometrioid and clear cell carcinomas. Am J Clin Pathol. 2015;144: Sato N, Tsunoda H, Nishida M, Morishita Y, Takimoto Y, Kubo T, et al. Loss of heterozygosity on 10q23.3 and mutation of the tumor suppressor gene PTEN in benign endometrial cyst of the ovary: possible sequence progression from benign endometrial cyst to endometrioid carcinoma and clear cell carcinoma of the ovary. Cancer Res. 2000;60: Wiegand KC, Shah SP, Al-Agha OM, Zhao Y, Tse K, Zeng T, et al. ARID1A mutations in endometriosis-associated ovarian carcinomas. N Engl J Med. 2010;363: Munksgaard PS, Blaakaer J. The association between endometriosis and gynecological cancers and breast cancer: a review of epidemiological data. Gynecol Oncol. 2011;123: oxford.asp. The Newcastle-Ottawa Scale (NOS). 24. Modugno F, Ness RB, Allen GO, Schildkraut JM, Davis FG, Goodman MT. Oral contraceptive use, reproductive history, and risk of epithelial ovarian cancer in women with and without endometriosis. Am J Obstet Gynecol. 2004;191: Cottreau CM, Ness RB, Modugno F, Allen GO, Goodman MT. Endometriosis and its treatment with danazol or lupron in relation to ovarian cancer. Clin Cancer Res. 2003;9: Melin AS, Lundholm C, Malki N, Swahn M-L, Sparen P, Sparen P, et al. Hormonal and surgical treatments for ª 2016 Nordic Federation of Societies of Obstetrics and Gynecology, Acta Obstetricia et Gynecologica Scandinavica 96 (2017)