Human Reproduction, Vol.30, No.1 pp. 222–231, 2015 Advanced Access publication on November 5, 2014 doi:10.1093/humrep/deu297

ORIGINAL ARTICLE Reproductive genetics

Risk for borderline ovarian tumours after exposure to fertility drugs: results of a population-based cohort study 1 2

Virus, Lifestyle and Genes, Danish Cancer Society Research Center, Strandboulevarden 49, DK-2100 Copenhagen Ø, Denmark Department of Gynaecology, Rigshospitalet, University of Copenhagen, DK-2100 Copenhagen Ø, Denmark

*Correspondence address. Virus, Lifestyle and Genes, Danish Cancer Society Research Center, Strandboulevarden 49, DK-2100 Copenhagen Ø, Denmark. Tel: +45-35-25-76-93; E-mail: [email protected]

Submitted on July 17, 2014; resubmitted on September 15, 2014; accepted on October 16, 2014

study question: Do fertility drugs increase the risk for borderline ovarian tumours, overall and according to histological subtype? summary answer: The use of any fertility drug did not increase the overall risk for borderline ovarian tumours, but an increased risk for serous borderline ovarian tumours was observed after the use of progesterone.

what is known already: Many epidemiological studies have addressed the connection between fertility drugs use and risk for ovarian cancer; most have found no strong association. Fewer studies have assessed the association between use of fertility drugs and risk for borderline ovarian tumours, and the results are inconsistent.

study design, size, duration: A retrospective case–cohort study was designed with data from a cohort of 96 545 Danish women with fertility problems referred to all Danish fertility clinics in the period 1963–2006. All women were followed for first occurrence of a borderline ovarian tumour from the initial date of infertility evaluation until a date of migration, date of death or 31 December 2006, whichever occurred first. The median length of follow-up was 11.3 years. participants/materials, setting, methods: Included in the analyses were 142 women with borderline ovarian tumours (cases) and 1328 randomly selected sub-cohort members identified in the cohort during the follow-up through 2006. Cases were identified by linkage to the Danish Cancer Register and the Danish Register of Pathology by use of personal identification numbers. To obtain information on use of fertility drugs, hospital files and medical records of infertility-associated visits to all Danish fertility clinics were collected and supplemented with information from the Danish IVF register. We used case–cohort techniques to calculate rate ratios (RRs) and corresponding 95% confidence intervals (CIs) for borderline ovarian tumours, overall and according to histological subtype, associated with the use of any fertility drug or five specific groups of fertility drugs: clomiphene citrate, gonadotrophins (human menopausal gonadotrophins and follicle-stimulating hormone), gonadotrophin-releasing hormone analogues, human chorionic gonadotrophins and progesterone. main results and the role of chance: Analyses within the cohort showed that the overall risk for borderline ovarian tumours was not associated with the use of any fertility drug (RR 1.00; 95% CI 0.67 –1.51) or of gonadotrophins (RR 1.32; 95% CI 0.81 –2.14), clomiphene citrate (RR 0.96; 95% CI 0.64– 1.44), human chorionic gonadotrophins (RR 0.91; 95% CI 0.61 –1.36) or gonadotrophin-releasing hormone analogues (RR 1.10; 95% CI 0.66– 1.81). Furthermore, no associations were observed between the risk for borderline ovarian tumours and these groups of fertility drugs according to the number of cycles of use, length of follow-up or parity. In contrast, use of progesterone increased the risk for borderline ovarian tumours, particularly serous tumours, for which statistically significantly increased risks were observed with any use of progesterone (RR 1.82; 95% CI 1.03– 3.24), among women treated with ≥4 cycles of progesterone (RR 2.63; 95% CI 1.04 –6.64) and for all women followed up for ≥4 years after their first treatment with progesterone.

limitations, reasons for caution: Although we tried to minimize the effects of the underlying infertility, the severity of infertility might have affected our risk estimates, as women with more severe fertility problems may receive more treatment. The results from the subgroup analyses, e.g. the findings of an elevated risk for borderline ovarian tumours associated with increased time since first use of progesterone and with increased number of treatment cycles, should be interpreted with caution as these analyses are based on a limited number of women with borderline ovarian tumours. & The Author 2014. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. For Permissions, please email: [email protected]

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

Sarah Marie Bjørnholt1, Susanne Kru¨ger Kjaer 1,2, Thor Schu¨tt Svane Nielsen 1, and Allan Jensen 1,*

223

Fertility drugs and risk for borderline ovarian tumours

wider implications of the findings: Although this study, which is the largest to date, provides reassuring evidence that there is no strong link between the use of fertility drugs and risk for borderline ovarian tumours, the novel observation of an increased risk for serous tumours after use of progesterone should be investigated in large epidemiological studies. The results of the present study provide valuable knowledge for clinicians and other health care personnel involved in the diagnosis and treatment of fertility problems. study funding/competing interest(s): No conflict of interest was reported. S.M.B. was supported by a research scholarship from the Danish Cancer Society. Key words: infertility / fertility drugs / borderline ovarian tumours / population-based cohort study / Denmark

Introduction

Materials and Methods Study population The Danish Infertility Cohort was established in 1997 and has been described in detail elsewhere (Baldur-Felskov et al., 2013). In brief, the cohort included all women referred to public gynaecological hospital departments or private fertility clinics for fertility problems in the period 1963 – 1998. In addition, we included all women with a diagnosis of infertility (ICD-8 code 628 or ICD-10 code N97) recorded in the National Patient Register in the period 1977 – 2009 and all women recorded in the Danish IVF register in the period 1994 – 2005. The National Patient Registry has registered virtually all people admitted to Danish hospitals for somatic conditions since 1977 and outpatient contacts since 1995 (Lynge et al., 2011). The Danish IVF Register contains compulsory information on all IVF treatments in Denmark since 1994. The present study comprised all 96 545 women evaluated for fertility problems who were included in the Danish Infertility Cohort from 1 September 1963 to 31 December 2006. All data were entered into a single database, with one record for each woman, including the initial date of infertility evaluation, the clinic’s name and the woman’s personal identification number. The unique Danish personal identification number is assigned to all Danish citizens and is registered in the computerized Danish Civil Registration System. It is included in all Danish health registries and ensures accurate linkage of data among registries. We linked the Danish Infertility Cohort to the Danish Civil Registration System to determine dates of migration, death or disappearance.

Ascertainment of cases Cases of borderline ovarian tumours diagnosed in women after enrolment in the Danish Infertility Cohort were identified by linking the cohort to both the Danish Cancer Register and the Danish Register of Pathology by use of personal identification numbers. We used both registers in order to increase the completeness of our data. Reporting of borderline ovarian tumours to the Danish Cancer Register has never been mandatory; consequently, these tumours are probably underreported in this register (Kjaerbye-Thygesen

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

Infertility is a common condition, which affects 10 –15% of all couples in western countries (Juul et al., 1999). During the past three decades, the use of fertility treatments, including intrauterine insemination, in vitro fertilization (IVF) and intracytoplasmic sperm injection has increased constantly in western countries (Ferraretti et al., 2013). The increased use of fertility drugs has given rise to studies of their potential adverse health effects and, especially, their potential to cause cancer. A number of epidemiological studies addressed the potential connection between the use of fertility drugs and risk for ovarian cancer (Rizzuto et al., 2013; Diergaarde and Kurta, 2014). Some studies, especially in the early 1990s, showed an increased risk for ovarian cancer associated with use of fertility drugs (Whittemore et al., 1992; Rossing et al., 1994), but most of the more recent studies have not observed a strong link (Ness et al., 2002; Rossing et al., 2004; Jensen et al., 2009; Kurta et al., 2012; Asante et al., 2013). Since the 1970s, borderline ovarian tumours have been recognized as a unique group of non-benign ovarian tumours (Scully, 1975), as they are considered an intermediate between malignant and benign tumours. They can metastasize and recur, but, in contrast to ovarian cancer, they do not grow invasively (Tavassoli and Devilee, 2003). Borderline ovarian tumours are characterized by an earlier age at diagnosis and a more favourable prognosis than ovarian cancers (Skirnisdottir et al., 2008; Hannibal et al., 2014). In contrast to the relatively large number of studies on use of fertility drugs and risk for ovarian cancer, fewer studies with borderline ovarian tumours as the outcome of interest have been performed (Harris et al., 1992; Shushan et al., 1996; Mosgaard et al., 1998; Parazzini et al., 1998; Cusido et al., 2007; Sanner et al., 2008; van Leeuwen et al., 2011; Yli-Kuha et al., 2012; Stewart et al., 2013). In contrast to the largely reassuring results concerning use of fertility drugs and risk for ovarian cancer, the majority of studies on borderline ovarian tumours have found an increased risk associated with use of fertility drugs (Harris et al., 1992; Shushan et al., 1996; Parazzini et al., 1998; Sanner et al., 2008; van Leeuwen et al., 2011; Stewart et al., 2013), although other studies have found no convincing association (Mosgaard et al., 1998; Cusido et al., 2007; Yli-Kuha et al., 2012). This literature was recently summarized in a Cochrane review, which concluded that there might be an increased risk for borderline ovarian tumours, especially after IVF (Rizzuto et al., 2013), indicating the need for further research in this field. Many of the previous studies on borderline ovarian tumours were constrained by methodological limitations, including low statistical power due to small sample size, short and incomplete follow-up, lack of control for potentially important confounders and inability to assess the potentially different effects of various types of fertility drug on the

risk for borderline ovarian tumours. In addition, in only a few of the earlier studies could the effects of fertility drugs be distinguished from the underlying causes of infertility. We compiled a large cohort of .100 000 women with fertility problems referred to Danish hospitals and private fertility clinics in the period 1963 –2009. To our knowledge, this cohort contains the most cases of borderline ovarian tumours in an infertility cohort to date. Furthermore, we had extensive information on the specific fertility drug used and on reproductive factors. We report here the results of a case –cohort study conducted to evaluate the effects of different types of fertility drug on the risk for borderline ovarian tumour, overall and according to histological subtype.

224

Identification of the sub-cohort We used a case– cohort design, in which the exposure of cases is compared with that of a randomly selected sub-cohort, as described by Prentice (1986). The sub-cohort consisted of 1458 women who were randomly selected from the Danish Infertility Cohort and stratified by age at cohort entry (18– 26, 27 – 30, 31 – 36 and .36 years) and in seven strata of calendar year of cohort entry (1963– 1977, 1978 –1984, 1985 – 1989, 1990– 1994, 1995 – 1998, 1999 – 2003 and 2004 – 2006), resulting in 28 strata.

Ascertainment of fertility drug use and potential confounders Hospital files and medical records for all medical visits for infertility to public hospitals and private fertility clinics were collected for all women in the cohort with borderline ovarian tumours and for all women in the sub-cohort. Information was abstracted by medical doctors and included data on medical interventions for infertility, with the types of fertility drugs prescribed and the number of cycles of use. For each treatment cycle, we abstracted the first and last treatment day to determine the window of exposure. Trained abstractors entered the data into computers, using standardized software. We supplemented the information from hospital files and medical records with information from the Danish IVF register. From the Danish National Patient Register and the Danish IVF Register, we also obtained information about the causes of infertility. For 50 cases and for 87 sub-cohort members, the hospital records could not be found, and no information on these persons was identified in the IVF register. We further excluded 8 sub-cohort

members because their diagnosis of infertility could not be confirmed, and 2 cases and 35 sub-cohort members were excluded from the analysis because their cause of infertility was previous sterilization, leaving 142 women with borderline ovarian tumours (73%) and 1328 women in the sub-cohort (91%) for analysis. Borderline ovarian tumours were diagnosed in three women in the sub-cohort during follow-up, and these women were therefore included both as cases and as sub-cohort members in the analyses. To obtain information on the parity of all infertile women with borderline ovarian tumours and for those in the sub-cohort, we linked the Danish Infertility Cohort to the Danish Civil Registration System and the National Danish Birth Register by the personal identification numbers. The population-based National Birth Register contains information on all births in Denmark since 1973. Thus, we obtained information on reproductive history from 1973 onwards from this register and reproductive history before 1973 from the Civil Registration System, which allows linkage between parents and children.

Statistical analysis Because of the sampling strategy, an unweighted case– cohort approach (Barlow, 1994; Barlow et al., 1999) was used to estimate rate ratios (RRs) for borderline ovarian tumours in a Cox proportional hazard regression model. The model was stratified according to the sampling strata (age and year of enrolment in the Danish Infertility Cohort). Age was used as the timescale to ensure that estimates were made from comparisons of women of the same age. All analyses were corrected for delayed entry to ensure that women were considered at risk only from their age at first infertility assessment. The RRs were calculated as suggested by Prentice (1986), in which all women in the sub-cohort contributed to all relevant risk sets until the end of follow-up due to a diagnosis of borderline ovarian tumour, death, migration or censoring, whereas case women not included in the sub-cohort contributed only their own risk sets. We evaluated the effects on overall risk for borderline ovarian tumour and risks for different histological subtypes according to any use of fertility drugs and for the following types of fertility drugs: human menopausal gonadotrophin (hMG), follicle-stimulating hormone (FSH), clomiphene citrate, human chorionic gonadotrophin (hCG), gonadotrophin-releasing hormone (GnRH) analogues and progesterone. In the analyses, we pooled the two gonadotrophins, FSH and hMG, into one group called gonadotrophins since, as their physiological modes of operation are identical. The potential effects of all types of fertility drugs were analysed as ‘ever’ use (ever/never), number of cycles of use (1–3 and .4) and time since first use (,4, 4–7, 8–11 and ≥12 years). The 142 epithelial ovarian borderline tumours analysed were subdivided into the histological subtypes of the ICD-O for the Danish Cancer Registry and of the ICD (seventh revision) for the Danish Registry of Pathology: serous (n ¼ 100), mucinous (n ¼ 36) and other types of borderline ovarian tumours (n ¼ 6). The potential confounder variables investigated were parity status (any childbirth) and number of additional births. All variables were entered as time-dependent covariates, which changed values at the ages at which a new event occurred (e.g. birth of child or start of new treatment cycle). All statistical analyses were carried out with SAS version 9.3 (SAS Institute, Cary, NC, USA).

Research ethics and data protection The study protocol was approved by the Danish Scientific Ethical Committee and the Danish Data Protection Agency.

Results Characteristics of the Danish Infertility Cohort The Danish Infertility Cohort accrued 96 545 women between 1963 and 2006. The median age at first fertility evaluation and hence entry into the cohort was 30.3 years, and the median age at the end of follow-up was

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

et al., 2007). Furthermore, registration of tumour cases in the Danish Register of Pathology was not complete before 1997. The nationwide Danish Cancer Register has recorded all incident malignancies since 1943 and contains information on cancer type, topography and morphology and is supplemented by linkage to the Causes of Death Register and the National Patient Register to ensure complete registration (Gjerstoff, 2011). Since 1978, the Danish Cancer Register has coded cases by topography and morphology according to the International Classification of Disease for Oncology (ICD-O). The Danish Register of Pathology is a national databank established in 1997 that contains the topography and morphology of all pathology specimens from all Danish pathology departments coded according to the Systematised Nomenclature of Medicine (SNOMED) as well as the date of diagnosis. All private and public pathology departments in Denmark are obliged to report to the registry, which is updated daily, through an online reporting system. The Danish Register of Pathology contains data back to 1978 but is not complete before 1997 (Bjerregaard and Larsen, 2011). Women in the Danish Infertility Cohort who were evaluated for infertility during 1963 – 2006 (n ¼ 96 545) were followed for the occurrence of borderline ovarian tumours from the initial date of infertility evaluation until a date of migration, date of death or 31 December 2006, whichever occurred first. In the Danish Cancer Register, borderline ovarian tumours were identified from the ICD-O3 topography code C569 and morphology codes beginning with M8 or M9 (behaviour code 1 or 2). Borderline ovarian tumours were identified in the Danish Register of Pathology by the SNOMED topography codes T87, T86910, T86920, T86921, T86922 and relevant morphology codes (all codes starting with M8 or M9 and behaviour codes 1 or 2). All borderline ovarian tumours were further classified into the following histological subtypes: serous borderline ovarian tumours (morphology codes M84A01, M84411, M84601, M84611, M90141), mucinous borderline ovarian tumours (morphology codes M84701, M84702, M84711, M84801, M90151) or other histological types (all other morphology codes encountered). If a woman was recorded as having more than one borderline ovarian tumour, we used only the first recorded incident tumour. An epithelial borderline ovarian tumour was diagnosed in 194 women during the follow-up period, and these were hence included in the analyses.

Bjørnholt et al.

225

Fertility drugs and risk for borderline ovarian tumours

Overall borderline ovarian tumours After adjustment for parity status, use of any fertility drug did not significantly affect the overall risk for borderline ovarian tumours (RR 1.00; 95% CI 0.67 –1.51) (Table II). When we assessed the risk for borderline ovarian tumours according to type of fertility drug, the risk was not significantly affected by any of the five evaluated types of fertility drug: progesterone (RR 1.51; 95% CI 0.90 –2.55), gonadotrophins (RR 1.32; 95% CI 0.81 –2.14), clomiphene citrate (RR 0.96; 95% CI 0.64 –1.44), hCG (RR 0.91; 95% CI 0.61 –1.36) or GnRH (RR 1.10; 95% CI 0.66 – 1.81). However, an increased risk for borderline ovarian tumours was observed with increasing time since first use of progesterone; this was statistically significant for women 4– 7 years after first use (RR 2.14; 95% CI 1.09 –4.22) and for women ≥12 years after first use of progesterone (RR 6.26; 95% CI 2.20 –17.76). The number of cycles of progesterone did not significantly affect the risk for borderline ovarian tumours. For the other four types of fertility drug, the risk for borderline ovarian tumours did not differ markedly according to number of cycles of use or time since first use.

Table I Rate ratios (RRs) for borderline ovarian tumours according to reproductive factors. Variable

No. of cases/no. in sub-cohort

RR (95% CI)

........................................................................................ Parity Nulliparous

82/419

1.00

Parous

60/909

0.37 (0.26– 0.54)

The risk for borderline ovarian tumours after use of fertility drugs was also analysed according to parity status (Table III). A significantly increased risk was observed for parous women who took progesterone (RR 2.09; 95% CI 1.03 –4.25), but not for nulliparous women (RR 1.12; 95% CI 0.57 –2.18); however, the interaction term was not statistically significant. The risk for borderline ovarian tumours associated with the other types of fertility drug was not markedly affected by parity status, and none of the interaction terms were statistically significant.

Histological subtypes of borderline ovarian tumours The risks for serous and mucinous borderline ovarian tumours after use of fertility drugs were analysed separately (Tables IV and Supplementary Table SI). The risk for serous tumours was not affected by any use or any number of cycles of use of any fertility drug, but women who were followed up for 4–7 years after their first use of any fertility drug had a statistically significantly increased risk (RR 2.15; 95% CI 1.04 –4.46). The risk for serous tumours was significantly increased by any use of progesterone (RR 1.82; 95% CI 1.03– 3.24), primarily among women who had ≥4 treatment cycles (RR 2.63; 95% CI 1.04 –6.64) and with increasing time since first use of progesterone; the increase in risk was statistically significant for all groups of women who had been followed for .4 years since first use of progesterone and most pronounced for women who were followed up for ≥12 years, who had a more than 10-fold increase in risk (RR 10.23; 95% CI 3.23 –32.34). No marked associations were found between any use of clomiphene citrate, gonadotrophins, hCG or GnRH analogues and risk for serous tumours; however, one additional cycle of gonadotrophins was shown to reduce the risk (RR 0.83; 95% CI 0.69 –1.00) (Table IV). The risk for mucinous borderline ovarian tumours was not significantly affected by use of any fertility drug or any specific type of fertility drug and was not substantially affected by the number of cycles of use or time since first use (Supplementary Table SI). Information on causes of infertility was available for the subset of women (89% of cases and 77% of sub-cohort members) who were recorded in the Danish National Patient Register and in the Danish IVF Register. To evaluate the effects of causes of infertility (categorical variable: (i) female infertility factor only [anovulation, tubal disease, endometriosis, uterine disorder, cervical disorder, other cause of female infertility, unknown cause of female infertility], (ii) male infertility factor only or (iii) both female and male infertility factors) as a potential confounder, all the analyses were adjusted for this variable in the subset of women for whom this information was available. However, none of the risk estimates changed noticeably after adjustment for causes of infertility (data not shown).

No. of births 1

34/423

0.39 (0.25– 0.61)

2

21/365

0.36 (0.22– 0.61)

≥3

5/121

0.30 (0.12– 0.79)

–

0.88 (0.63– 1.24)

Per additional birth among parous women

All analyses stratified according to calendar year (in categories) and age at start of follow-up (in categories). Rate ratios for no. of births are adjusted for parity (nulliparous/parous). CI, confidence interval.

Discussion In this study of the largest number of cases of borderline ovarian tumours in infertile women to date, we found no convincing association between use of any fertility drug and overall risk for these tumours. With respect to type of fertility drug, our results indicate that use of progesterone increases the risk for borderline ovarian tumours and especially for serous tumours, the risk for which was increased by almost 2-fold; the risk was further increased by four or more treatment cycles and with increasing time since first use. The presence of both latency

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

42.5 years. The median length of follow-up was 11.3 years, with a maximum of 49 years. The 96 545 women contributed a total of 1 222 252 person-years of observation. Age at diagnosis of borderline ovarian cancer ranged from 21.7 to 65.1 years, with a median of 40.2 years. The median time between entry into the cohort and diagnosis was 10.7 years. Fertility drugs were used by 89 of the 142 women (63%) with borderline ovarian tumours and by 683 of the 1328 women (51%) in the sub-cohort. The most commonly used fertility drug was hCG, used by 63 cases (44%) and 488 (37%) women in the subcohort, followed by clomiphene citrate (cases: 39%; sub-cohort members: 33%), gonadotrophins (cases: 39%; sub-cohort members: 19%), GnRH analogues (cases: 28%; sub-cohort members: 14%) and progesterone (cases: 27%; sub-cohort members: 12%). Parous woman had a significantly lower risk for borderline ovarian tumours than nulliparous women (RR, 0.37; 95% CI, 0.26 –0.54), but no further decrease in risk was observed with additional number of births (Table I).

226

Table II Rate ratios (RRs) of borderline ovarian tumours associated with use of fertility drugs. Any fertility drug

Gonadotrophins†

Clomiphene citrate

hCG

No. of cases/no. in sub-cohort

No. of cases/no. in sub-cohort

No. of cases/no. in sub-cohort

No. of cases /no. in sub-cohort

................................... RR (95% CI)

................................... RR (95% CI)

................................... RR (95% CI)

................................... RR (95% CI)

GnRH

...................................

No. of cases /no. in sub-cohort

RR (95% CI)

Progesterone

...................................

No. of cases /no. in sub-cohort

RR (95% CI)

.......................................................................................................................................................................................................................................................... Use Never

53/645

1.00

87/1072

1.00

86/888

1.00

77/840

1.00

102/1148

1.00

103/1175

1.00

Ever

89/683

1.00 (0.67 –1.51)

55/256

1.32 (0.81 –2.14)

56/440

0.96 (0.64 –1.44)

65/488

0.91 (0.61 –1.36)

40/180

1.10 (0.66 – 1.81)

39/153

1.51 (0.90 –2.55)

1–3

40/263

1.12 (0.70 –1.80)

35/138

1.51 (0.88 –2.60)

25/208

0.86 (0.52 –1.42)

34/241

0.88 (0.54 –1.44)

26/121

1.03 (0.59 – 1.81)

32/122

1.47 (0.85 –2.53)

≥4

49/420

0.91 (0.57 –1.45)

20/118

1.04 (0.57 –1.92)

31/232

1.07 (0.66 –1.76)

31/247

0.94 (0.57 –1.54)

14/59

1.28 (0.64 – 2.57)

7/31

1.73 (0.70 –4.28)

No. of cycles

Per additional cycle

0.96 (0.91 –1.01)

0.87 (0.76 –1.01)

0.99 (0.92 –1.06)

0.98 (0.91 –1.06)

0.97 (0.80 – 1.17)

1.09 (0.87 –1.36)

Time since first use (years) ,4

24/16

1.16 (0.60 –2.28)

16/16

0.85 (0.44 –1.62)

12/4

0.90 (0.45 –1.76)

15/14

0.62 (0.33 –1.15)

13/14

0.82 (0.43 – 1.56)

10/15

0.67 (0.32 –1.37)

4–7

28/34

1.48 (0.76 –2.87)

21/34

1.59 (0.84 –3.01)

15/12

0.99 (0.49 –2.01)

21/31

1.10 (0.59 –2.06)

19/30

1.76 (0.90 – 3.42)

18/32

2.14 (1.09 –4.22)

8–11

15/65

0.97 (0.50 –1.91)

10/69

1.51 (0.71 –3.19)

12/32

0.97 (0.51 –1.86)

16/72

1.37 (0.74 –2.54)

6/71

1.04 (0.40 – 2.68)

7/74

2.22 (0.94 –5.24)

≥12

22/568

0.73 (0.40 –1.31)

8/137

2.18 (0.82 –5.80)

17/392

1.00 (0.53 –1.89)

13/371

0.83 (0.42 –1.64)

2/65

0.58 (0.10 – 3.52)

4/32

6.26 (2.20 –17.76)

All analyses stratified according to calendar year (in categories) and age at start of follow-up (in categories). All rate ratios adjusted for parity status (nulliparous/parous). CI, confidence interval; hCG, human chorionic gonadotrophins; GnRH, gonadotrophins-releasing hormone analogues. † Follicle-stimulating hormone and human menopausal gonadotrophins.

Bjørnholt et al.

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

227

Fertility drugs and risk for borderline ovarian tumours

Table III Rate ratios (RRs) for borderline ovarian tumours in nulliparous and parous women according to use of fertility drugs. Fertility drug and use

Nulliparous women

Parous women

......................................................

......................................................

No. of cases /no. in sub-cohort

No of cases /no. in sub-cohort

RR (95% CI)

Interaction (P-value)

RR (95% CI)

............................................................................................................................................................................................. All Never

32/192

1.00

21/453

1.00

Ever

50/227

0.95 (0.57 –1.58)

39/456

1.07 (0.59 –1.95)

0.75

†

Gonadotrophins

52/346

1.00

35/726

1.00

30/73

1.22 (0.67 –2.21)

25/183

1.45 (0.76 –2.78)

0.66

Clomiphene citrate Never

48/271

1.00

38/617

1.00

Ever

34/148

1.06 (0.64 –1.74)

22/292

0.85 (0.47 –1.52)

Never

48/270

1.00

29/570

1.00

Ever

34/149

0.83 (0.49 –1.39)

31/339

1.03 (0.58 –1.82)

0.55

hCG 0.54

GnRH Never

62/362

1.00

40/786

1.00

Ever

20/57

0.85 (0.44 –1.62)

20/123

1.52 (0.77 –3.02)

0.19

Progesterone Never

65/380

1.00

38/795

1.00

Ever

17/39

1.12 (0.57 –2.18)

22/114

2.09 (1.03 –4.25)

0.17

All analyses stratified according to calendar year (in categories) and age at start of follow-up (in categories). All rate ratios adjusted for parity status (nulliparous/parous). CI, confidence interval; hCG, human chorionic gonadotrophins; GnRH, gonadotrophins-releasing hormone analogues. † Follicle-stimulating hormone and human menopausal gonadotrophins.

and dose–response effects may support the likelihood of a causal association, but these results need to be interpreted with caution as they are based on a limited number of women with borderline ovarian tumours. In contrast, none of the ovulation-stimulating drugs (clomiphene citrate, gonadotrophins and hCG) or GnRH analogues affected the risk for borderline ovarian tumours. None of the risk estimates was markedly changed when the results were stratified by parity status. The results of previous studies on the association between use of fertility drugs and risk for borderline ovarian tumours have been inconsistent, due partly perhaps to substantial differences in study design, definitions of exposure and outcome measures as well as methodological limitations, including imprecise risk estimates due to small sample size and short or incomplete follow-up. Our finding of no association between use of any fertility drug and risk for borderline ovarian tumours is in line with both previous case –control (Mosgaard et al., 1998; Rossing et al., 2004; Cusido et al., 2007; Asante et al., 2013) and cohort studies (Brinton et al., 2004; Sanner, et al., 2008; Yli-Kuha et al., 2012). In a Swedish cohort study by Sanner et al. (2008), however, of 2768 women assessed for infertility problems between 1961 and 1975, those treated with fertility drugs had a more than 3-fold higher risk for borderline ovarian tumours than women in the general Swedish population but not when they were compared with untreated women with a diagnosis of infertility. Increased overall risks for borderline ovarian tumours associated with use of any fertility drug were

observed especially in early case –control studies (Harris et al., 1992; Rossing et al., 1994; Shushan et al., 1996; Ness et al., 2002), and van Leeuwen et al. (2011) and Stewart et al. (2013) also observed an increased risk in cohort studies; however, only the effect of any IVF was studied and not associations with use of fertility drugs. Therefore, these results are not directly comparable with ours. With regard to type of fertility drug, we found no convincing associations with use of any of the ovulation-stimulating drugs (clomiphene citrate, gonadotrophins and hCG) or GnRH analogues. No previous study has analysed the potential association with use of GnRH analogues, and only six other studies have assessed potential associations with specific types of ovulation-stimulating fertility drugs (Shushan et al., 1996; Mosgaard et al., 1998; Brinton et al., 2004; Cusido et al., 2007; Sanner et al., 2008; van Leeuwen et al., 2011). In line with our results, most of these studies showed no significantly increased risk with specific types of fertility drugs, including clomiphene citrate (Mosgaard et al., 1998; Brinton et al., 2004; Cusido et al., 2007; Sanner et al., 2008), gonadotrophins (Cusido et al., 2007; Sanner et al., 2008) and hMG (Mosgaard et al., 1998; van Leeuwen et al., 2011). Only the results of a small case–control study by Shushan et al. (1996) showed an increased risk for borderline ovarian tumours after treatment with a specific type of ovulationstimulating fertility drug, with an almost 10-fold increase among women who had used hMG (OR 9.38; 95% CI 1.66 –52.08). This result should, however, be interpreted with caution, as only 6 of the 36 case women had ever used this drug.

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

Never Ever

228

Table IV Rate ratios (RRs) for serous borderline ovarian tumours associated with use of fertility drugs. Any fertility drug

Gonadotrophins†

Clomiphene citrate

hCG

No. of cases/ no. in sub-cohort

No. of cases/ no. in sub-cohort

No. of cases/ no. in sub-cohort

No. of cases/ no. in sub-cohort

.................................... RR (95% CI)

.................................... RR (95% CI)

.................................... RR (95% CI)

.................................... RR (95% CI)

GnRH

....................................

No. of cases/ no. in sub-cohort

RR (95% CI)

Progesterone

....................................

No. of cases/ no. in sub-cohort

RR (95% CI)

.......................................................................................................................................................................................................................................................... Use Never

37/645

1.00

61/1072

1.00

61/888

1.00

54/840

1.00

70/1148

1.00

71/1175

1.00

Ever

63/683

1.10 (0.69 – 1.75)

39/256

1.42 (0.82 –2.45)

39/440

1.02 (0.64 –1.62)

46/488

0.97 (0.61 –1.55)

28/180

1.11 (0.62 –1.98)

29/153

1.82 (1.03 –3.24)

1–3

28/263

1.21 (0.70 – 2.08)

26/138

1.71 (0.93 –3.14)

15/208

0.79 (0.43 –1.44)

24/241

0.91 (0.52 –1.60)

19/121

1.08 (0.57 –2.05)

23/122

1.68 (0.92 –3.08)

≥4

35/420

1.02 (0.59 – 1.74)

13/118

1.02 (0.51 –2.06)

24/232

1.27 (0.73 –2.23)

22/247

1.03 (0.58 –1.83)

9/59

1.20 (0.52 –2.78)

6/31

2.63 (1.04 –6.64)

No. of cycles

Per additional cycle

0.96 (0.91 – 1.02)

0.83 (0.69 –1.00)

1.01 (0.94 –1.09)

0.98 (0.91 –1.06)

0.92 (0.74 –1.14)

1.10 (0.87 –1.39)

Time since first use (years) ,4

14/16

1.05 (0.47 – 2.36)

11/16

0.9 (0.46 –1.92)

6/4

0.73 (0.28 –1.85)

9/14

0.55 (0.26 –1.20)

8/14

0.72 (0.34 –1.50)

7/15

0.71 (0.33 –1.53)

4–7

23/3

2.15 (1.04 – 4.46)

16/34

2.00 (1.00 –4.03)

14/12

1.56 (0.76 –3.20)

16/31

1.32 (0.66 –2.62)

14/30

2.08 (0.99 –4.40)

13/32

2.63 (1.24 –5.58)

8–11

11/65

1.15 (0.53 – 2.52)

7/69

1.47 (0.61 –3.55)

8/32

0.95 (0.45 –2.01)

12/72

1.50 (0.75 –3.16)

5/71

1.18 (0.40 –3.49)

5/74

2.77 (1.03 –7.44)

≥12

15/568

0.73 (0.37 – 1.44)

5/137

1.78 (0.54 –5.81)

11/392

0.94 (0.44 –2.00)

9/371

0.82 (0.37 –1.84)

1/65

0.37 (0.04 –3.80)

4/32

10.23 (3.23 –32.34)

All analyses stratified according to calendar year (in categories) and age at start of follow-up (in categories). All rate ratios adjusted for parity status (nulliparous/parous). CI, confidence interval; hCG, human chorionic gonadotrophins; GnRH, gonadotrophins-releasing hormone analogues. † Follicle-stimulating hormone and human menopausal gonadotrophins.

Bjørnholt et al.

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

Fertility drugs and risk for borderline ovarian tumours

Our study has several strengths. First and perhaps most importantly, in comparison with previous cohort studies, we included a large number of women with borderline ovarian tumours (n ¼ 142), of whom 89 were treated with fertility drugs. This increased the statistical precision of our risk estimates over that of previous studies, in which the number of women with borderline ovarian tumours ranged from 5 to 53 and of whom 4 to 27 were treated with fertility drugs (Harris et al., 1992; Rossing et al., 1994, 2004; Shushan et al., 1996; Mosgaard et al., 1998; Parazzini et al., 1998; Brinton et al., 2004; Cusido et al., 2007; Sanner et al., 2008; van Leeuwen et al., 2011; Yli-Kuha et al., 2012; Asante et al., 2013; Stewart et al., 2013). However, for subgroups of fertility drugs users, our study may still have been somewhat underpowered and limited by risk estimates of relatively low precision. This especially applies for the findings of a significantly increased risk for borderline ovarian tumours associated with increasing time since first use of progesterone and with increasing number of treatment cycles, as these analyses are based on a limited number of women with borderline ovarian tumours. Therefore, these findings need to be interpreted with caution and should be further investigated in large epidemiological studies. Secondly, as we had detailed information on the types of fertility drugs used; from both the Danish IVF register and the women’s medical records, we were able to evaluate their potentially different effects on risk for borderline ovarian tumour. Only three cohort studies previously assessed the effects of different types of fertility drug (Brinton et al., 2004; Sanner et al., 2008; van Leeuwen et al., 2011), and none examined the separate effects of progesterone and GnRH. Thirdly, a major problem in this field of research is the disentangling the effects of fertility treatment from the potential effects of the underlying infertility (i.e. of biological or genetic origin). This means that if negative health effects of fertility treatment are found, it is uncertain whether the effect is associated with the treatment itself or the underlying infertility. In the present study, we estimated the effect of fertility drugs on the risk for borderline ovarian tumours solely among women with fertility problems; thereby minimizing the effect of the underlying infertility. Previously, only the cohort study by Sanner et al. (2008) had a suitable reference group of infertile women not treated with fertility drugs and could thereby separate the effect on risk for borderline tumours due to infertility from that of fertility drugs. van Leeuwen et al. (2011) and Stewart et al. (2013) also used a reference group of infertile women not treated with IVF, but, as previously mentioned, their main exposure was to IVF, and the results are therefore not directly comparable to ours. Fourthly, the detailed histological information in the Danish Cancer Registry and the Danish Registry of Pathology enabled us to evaluate the risks for serous and mucinous borderline ovarian tumours separately, which was previously done in only two studies (Ness et al., 2002; van Leeuwen et al., 2011). Other important strengths of our study include the lack of recall bias, as information on exposure was abstracted from medical files and registers, the virtual absence of loss to follow-up, because of accurate linkage of the Danish Infertility Cohort to the Danish Civil Registration System, and the complete ascertainment of borderline ovarian tumours through linkage to the Danish Cancer Register and the Danish Register of Pathology. Some limitations of our study should also be noted. Although we were able to adjust for the potentially confounding variables parity status, the number of additional births and causes of infertility (the latter only in a subset analyses, though), we were not able to adjust for other potential confounders such as use of oral contraceptives that may slightly have affected our results. By studying the effects of fertility

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

The associations between use of fertility drugs and risk for borderline ovarian tumours in our study were not markedly changed when the results were stratified by parity status. This finding is in line with results of one study (Rossing et al., 2004) but contrasts with those of two others suggesting a higher risk for borderline ovarian tumours among women who had received fertility treatment and remained nulliparous (Ness et al., 2002; van Leeuwen et al., 2011). Stewart et al. (2013) observed a significantly increased risk associated with IVF among parous women and not among nulliparous women. Hence, the effect of parity status on the pattern of risk for borderline ovarian tumours and use of fertility drugs warrants further investigation. The relatively large number of borderline ovarian tumours included in this study and the detailed histological information available from the Danish Cancer Register and the Danish Register of Pathology enabled us to analyse differences in risk for the two commonest histological subtypes of borderline ovarian tumour: serous and mucinous. Similar to overall borderline ovarian tumours, no convincing association was found between use of any fertility drug or of ovulation-stimulating drugs or GnRH and risks for serous and mucinous tumours. The risk estimates for serous and mucinous tumours were virtually the same as or similar to those observed for overall borderline ovarian tumours. Only two previous studies have evaluated the risks for serous and mucinous tumours separately (Ness et al., 2002; van Leeuwen et al., 2011). van Leeuwen et al. (2011) found an increased risk for serous tumours among women treated with IVF, and Ness et al. (2002) found that fertility drugs increased the risk for nulligravid women only. Neither study found an increased risk for mucinous borderline ovarian tumours. No previous studies have evaluated the separate effect of progesterone on the risk for borderline ovarian tumours. We found that use of progesterone may increase the risk for serous tumours, as any use of progesterone increased the risk almost 2-fold and the risk was further increased by four or more treatment cycles and with increasing time since first use. In contrast, the risk for mucinous tumours was not affected by use of progesterone; however, the results for mucinous tumours should be interpreted with care owing to the small number of these tumours included. In fertility treatment, progesterone is used routinely in most IVF protocols. The rationale for using progesterone during the luteal phase in IVF treatment cycles is due to the GnRH agonist dysregulation of luteal pituitary LH pulses which normally support the corpus luteum. The biological mechanisms that could link progesterone to risk for borderline ovarian tumours are not clear, but there are several possibilities. Sex hormones, including progesterone, are considered to be some of the most potent carcinogenic hormones (Franks and Teich, 2003), and progesterone is classified as ‘reasonably anticipated to be human carcinogen’ on the basis of studies in experimental animals (National Toxicology Program, 2011). Other studies have shown that hormone therapy regimens containing both progestin (synthetic compounds with similar modes of action as progesterone) and estrogen increase the risks for both ovarian cancer (Beral et al., 2007) and borderline ovarian tumours (Morch et al., 2012) to a greater extent than regimes with estrogen only. The results of a recent study showed significantly greater expression of progesterone receptors in serous than in mucinous borderline ovarian tumours, which might explain why we observed an increased risk primarily for serous tumours (Sallum et al., 2013). Additional large epidemiological studies with appropriate reference groups are clearly needed to confirm or reject our findings on progesterone and risk for serous borderline ovarian tumours.

229

230

Supplementary data Supplementary data are available at http://humrep.oxfordjournals.org/.

Acknowledgements The authors are truly grateful to the staff at the participating private fertility clinics and public gynaecological hospital departments for providing access to their patients’ medical files. Without these people, this study would not have been possible.

Authors’ roles S.M.B., S.K.K. and A.J. designed the study and interpreted the data. S.M.B. drafted the manuscript. S.K.K. and A.J. critically revised the manuscript for important intellectual content. T.S.S.N. analysed and interpreted the data and critically revised the manuscript for important intellectual content. All authors approved the final version of the manuscript.

Funding This study was supported by a scholarship grant from the Danish Cancer Society.

Conflict of interest None declared.

References Asante A, Leonard PH, Weaver AL, Goode EL, Jensen JR, Stewart EA, Coddington CC. Fertility drug use and the risk of ovarian tumors in infertile women: a case-control study. Fertil Steril 2013; 7:2031 – 2036. Baldur-Felskov B, Kjaer SK, Albieri V, Steding-Jessen M, Kjaer T, Johansen C, Dalton SO, Jensen A. Psychiatric disorders in women with fertility problems: results from a large Danish register-based cohort study. Hum Reprod 2013;3:683 –690. Barlow WE. Robust variance estimation for the case-cohort design. Biometrics 1994;4:1064– 1072. Barlow WE, Ichikawa L, Rosner D, Izumi S. Analysis of case-cohort designs. J Clin Epidemiol 1999;12:1165 – 1172. Beral V, Bull D, Green J, Reeves G. Ovarian cancer and hormone replacement therapy in the Million Women Study. Lancet 2007;9574: 1703 – 1710. Bjerregaard B, Larsen OB. The Danish Pathology Register. Scand J Public Health 2011;39:72 – 74. Brinton LA, Lamb EJ, Moghissi KS, Scoccia B, Althuis MD, Mabie JE, Westhoff CL. Ovarian cancer risk after the use of ovulation-stimulating drugs. Obstet Gynecol 2004;6:1194 – 1203. Cusido M, Fabregas R, Pere BS, Escayola C, Barri PN. Ovulation induction treatment and risk of borderline ovarian tumors. Gynecol Endocrinol 2007;7:373 – 376. Diergaarde B, Kurta ML. Use of fertility drugs and risk of ovarian cancer. Curr Opin Obstet Gynecol 2014;3:125 – 129. Ferraretti AP, Goossens V, Kupka M, Bhattacharya S, de MJ, Castilla JA, Erb K, Korsak V, Nyboe Andersen A, ; European IVF-Monitoring (EIM) Consortium for the European Society of Human Reproduction and Embryology (ESHRE). Assisted reproductive technology in Europe, 2009: results generated from European registers by ESHRE. Hum Reprod 2013;9:2318 –2331. Franks LM, Teich NM. Introduction to the Cellular and Molecular Biology of Cancer, 3rd edn. New York: Oxford University Press, 2003. Gjerstoff ML. The Danish Cancer Registry. Scand J Public Health 2011; 39:42 – 45. Hannibal CG, Huusom LD, Kjaerbye-Thygesen A, Tabor A, Kjaer SK. Trends in incidence of borderline ovarian tumors in Denmark 1978– 2006. Acta Obstet Gynecol Scand 2011;4:305 – 312. Hannibal CG, Vang R, Junge J, Frederiksen K, Kjaerbye-Thygesen A, Andersen KK, Tabor A, Kurman RJ, Kjaer SK. A nationwide study of serous ‘borderline’ ovarian tumors in Denmark 1978 – 2002: Centralized pathology review and overall survival compared with the general population. Gynecol Oncol 2014;2:267–273. Harris R, Whittemore AS, Itnyre J. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 US case-control studies. III. Epithelial tumors of low malignant potential in white women. Collaborative Ovarian Cancer Group. Am J Epidemiol 1992;10:1204 –1211. Jensen A, Sharif H, Frederiksen K, Kjaer SK. Use of fertility drugs and risk of ovarian cancer: Danish Population Based Cohort Study. BMJ 2009; 338:b249. Juul S, Karmaus W, Olsen J. Regional differences in waiting time to pregnancy: pregnancy-based surveys from Denmark, France, Germany, Italy and Sweden. The European Infertility and Subfecundity Study Group. Hum Reprod 1999;5:1250– 1254.

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

drugs on borderline ovarian tumours solely among women with fertility problems, however, we minimized any effect of the underlying infertility. However, even though we observed no substantial impact on the results when adjusting for causes of infertility, we cannot rule out the possibility that the severity of infertility affected our risk estimates, as women with more severe fertility problems might have received more treatment than women with less severe fertility problems. Of the women in our study who were treated with progesterone, almost all of these women had received progesterone as a part of their IVF treatment regimen (98%). This almost complete overlap means that the potential effect of progesterone and of the IVF procedure cannot analytically be disentangled. Hence, we cannot rule out that the increased risk for borderline ovarian tumours is to some degree due to unmeasured factors associated with the IVF treatment regimen, and not solely due to the use of progesterone. Further limitations of the study were that no information on the dosage of the fertility drugs was available, and further no information on the type of progestin used for fertility treatment was available. Also, although the follow-up period was relatively long (median, 11.3 years), the median age at the end of follow-up was only 42.5 years, which is below the peak age for diagnosis of borderline ovarian tumours in Denmark (52 years) (Hannibal et al., 2011). This might have weakened our risk estimates. In conclusion, this nationwide study of the largest number of borderline ovarian tumours in a cohort of infertile women to date provides reassuring evidence that there is no strong link between use of fertility drugs and risk for these tumours. The novel observation of an increased risk for serous borderline ovarian tumours after progesterone treatment should be further investigated in large epidemiological studies. In interpreting the results of the present study, any unfavourable effects of a fertility drug, including a small increased risk for borderline ovarian tumours with a good prognosis, would have to be weighed against the physical and psychological benefits of having a child, which is made possible for an increasing number of women only with the use of fertility drugs.

Bjørnholt et al.

Fertility drugs and risk for borderline ovarian tumours

Sallum LF, Sarian LO, Lucci De Angelo AL, Vassallo J, Soares FA, Pinto GA, Ferreira PA, Derchain S. Survival of women with ovarian carcinomas and borderline tumors is not affected by estrogen and progesterone receptor status. J Gynecol Oncol 2013;2:167 – 176. Sanner K, Conner P, Bergfeldt K, Dickman P, Sundfeldt K, Bergh T, Hagenfeldt K, Janson PO, Nilsson S, Persson I. Ovarian epithelial neoplasia after hormonal infertility treatment: long-term follow-up of a historical cohort in Sweden. Fertil Steril 2008;4:1152 – 1158. Scully RE. World Health Organization classification and nomenclature of ovarian cancer. Natl Cancer Inst Monogr 1975;42:5 – 7. Shushan A, Paltiel O, Iscovich J, Elchalal U, Peretz T, Schenker JG. Human menopausal gonadotropin and the risk of epithelial ovarian cancer. Fertil Steril 1996;1:13 – 18. Skirnisdottir I, Garmo H, Wilander E, Holmberg L. Borderline ovarian tumors in Sweden 1960 – 2005: trends in incidence and age at diagnosis compared to ovarian cancer. Int J Cancer 2008;8:1897– 1901. Stewart LM, Holman CD, Finn JC, Preen DB, Hart R. In vitro fertilization is associated with an increased risk of borderline ovarian tumours. Gynecol Oncol 2013;2:372 – 376. Tavassoli FA, Devilee P (eds). World Health Organization Classification of Tumours. Pathology and Genetics, Tumours of the Breast and Female Genital Organs. Lyon: IARC Press, 2003. van Leeuwen FE, Klip H, Mooij TM, van de Swaluw AM, Lambalk CB, Kortman M, Laven JS, Jansen CA, Helmerhorst FM, Cohlen BJ et al. Risk of borderline and invasive ovarian tumours after ovarian stimulation for in vitro fertilization in a large Dutch cohort. Hum Reprod 2011; 12:3456– 3465. Whittemore AS, Harris R, Itnyre J. Characteristics relating to ovarian cancer risk: collaborative analysis of 12 US case-control studies. II. Invasive epithelial ovarian cancers in white women. Collaborative Ovarian Cancer Group. Am J Epidemiol 1992;10:1184 –1203. Yli-Kuha AN, Gissler M, Klemetti R, Luoto R, Hemminki E. Cancer morbidity in a cohort of 9175 Finnish women treated for infertility. Hum Reprod 2012; 4:1149 – 1155.

Downloaded from http://humrep.oxfordjournals.org/ at University of California, San Francisco on April 10, 2015

Kjaerbye-Thygesen A, Huusom LD, Frederiksen K, Kjaer SK. Primary ovarian cancer. A comparison of registrations in the Danish Cancer Registry and the Pathology Data Bank. Ugeskr Laeger 2007;1:50 – 54. Kurta ML, Moysich KB, Weissfeld JL, Youk AO, Bunker C, Edwards RP, Modugno F, Ness RB, Diergaarde B. Use of fertility drugs and risk of ovarian cancer: results from a US-based case-control study. Cancer Epidemiol Biomarkers Prev 2012;8:1282 – 1292. Lynge E, Sandegaard JL, Rebolj M. The Danish National Patient Register. Scand J Public Health 2011;7:30 – 33. Morch LS, Lokkegaard E, Andreasen AH, Kjaer SK, Lidegaard O. Hormone therapy and ovarian borderline tumors: a national cohort study. Cancer Causes Control 2012;1:113– 120. Mosgaard BJ, Lidegaard O, Kjaer SK, Schou G, Andersen AN. Ovarian stimulation and borderline ovarian tumors: a case-control study. Fertil Steril 1998;6:1049 – 1055. National Toxicology Program. 12th Report on Carcinogens (Rep Carcinog 2011). National Institute of Environmental Health Standards, Research Triangle Park, North Carolina. NessRB, CramerDW, GoodmanMT, KjaerSK, Mallin K, Mosgaard BJ, Purdie DM, Risch HA, Vergona R, Wu AH. Infertility, fertility drugs, and ovarian cancer: a pooled analysis of case-control studies. Am J Epidemiol 2002;3:217–224. Parazzini F, Negri E, La Vecchia C, Moroni S, Polatti A, Chiaffarino F, Surace M, Ricci E. Treatment for fertility and risk of ovarian tumors of borderline malignancy. Gynecol Oncol 1998;3:226 – 228. Prentice RL. A case-cohort design for epidemiological cohort studies and disease prevention trials. Biometrika 1986;73:1 – 11. Rizzuto I, Behrens RF, Smith LA. Risk of ovarian cancer in women treated with ovarian stimulating drugs for infertility. Cochrane Database Syst Rev 2013; 8:CD008215. Rossing MA, Daling JR, Weiss NS, Moore DE, Self SG. Ovarian tumors in a cohort of infertile women. N Engl J Med 1994;12:771 – 776. Rossing MA, Tang MT, Flagg EW, Weiss LK, Wicklund KG. A case-control study of ovarian cancer in relation to infertility and the use of ovulationinducing drugs. Am J Epidemiol 2004;11:1070– 1078.

231