Risk Factors for Endometriosis in a German Case–Control Study

!

The etiology of endometriosis is still a research field in which few consistent data are available. Large case –control studies or even co- hort studies are rare, and most of the published data are conflicting. The aim of the present study was therefore to examine common epidemiologi- cal and endometriosis-specific risk factors in a German case–control study. Design: From 2001 to 2010, a pool of 595 laparos- copically confirmed cases and 475 controls were recruited in a hospital-based setting. After match- ing for age, 298 cases and 300 controls remained in the pool. Age at menarche, menstrual cycle length, duration of menstrual bleeding, number of pregnancies, live births, miscarriages, use of contraceptive pills, body mass index (BMI), and smoking status were analyzed with logistic re- gression models predicting endometriosis case – control status.

Menstrual cycle length, duration of men- strual bleeding, number of pregnancies, number of miscarriages, and smoking status, as relevant predictors for endometriosis case –control status, were identified as risk factors for endometriosis. Other factors such as age at menarche, number of live births, ever having used contraceptive pills, and BMI were not predictive.

This hospital-based case –control study reproduced most of the familiar risk factors. Comparison of this study with others reveals a wide variety of effect sizes and directions of asso- ciation with risk factors and may increase the in- formation available about the characteristics of the patient population being treated in the rele- vant hospital setting. Zusammenfassung ! Einleitung: Die Ätiologie der Endometriose bleibt ein Forschungsgebiet mit spärlichen Erkenntnis- sen. Große Fall-Kontroll-Studien oder Kohorten- studien sind selten, und die publizierten Daten widersprechen sich auf den ersten Blick zum größten Teil. Ziel unserer Studie war es deswegen, in einer deutschen Fall-Kontroll-Studie Risikofak- toren zu identifizieren und zu beschreiben. Methoden: Von 2001 bis 2010 wurde ein Pool von 595 laparoskopisch diagnostizierten Endometrio- sepatientinnen und 475 in einem krankenhausba- sierten Design rekrutiert. Von diesem Pool wurden 298 Fälle und 300 Kontrollen altersgematched. Ergebnisse: Alter bei der Menarche, Zyklusdauer, Dauer der Periodenblutung, Anzahl der Schwan- gerschaften und Lebendgeburten, Fehlgeburten, Pilleneinnahme, Body-Mass-Index (BMI) und Raucherstatus wurden in logistischen Regres- sionsmodellen auf ihre Prädiktion für den Endo- metriose-Fall-Kontroll-Status untersucht. Dauer der Periodenblutung, Dauer des Zyklus, Anzahl der Schwangerschaften, Anzahl der Fehlgeburten und Raucherstatus waren relevante und signifi- kante Risikofaktoren für die Endometriose. Die an- deren Faktoren wie Alter bei Menarche, Anzahl der Lebendgeburten, Pillennutzung und BMI waren im multivariaten Modell nicht prädiktiv in Bezug auf den Fall-Kontroll-Status. In dieser krankenhaus- basierten Fall-Kontroll-Studie konnten die meis- ten, bekannten Risikofaktoren für eine Endome- trioseerkrankung reproduziert werden. Schlussfolgerung: Vergleicht man die Studie mit anderen, so fällt auf, dass die Variabilität der Ef- fektstärke und der Richtung der Prädiktion für die Risikofaktoren in den meisten Studien unter- schiedlich ist. Die Analyse der Risikofaktoren im eigenen Kollektiv könnte helfen, die Population der Endometriosepatientinnen besser zu verste- hen, die im entsprechenden Krankenhaus behan- delt werden. Risk Factors for Endometriosis in a German Case –Control Study Risikofaktoren für Endometriose in einer deutschen Fall-Kontroll-Studie Authors S. Burghaus 1, P. Klingsiek 1, P. A. Fasching 1, A. Engel 1, L. Häberle 1, P. L. Strissel 1, M. Schmidt 1, K. Jonas 1, J. D. Strehl 2, A. Hartmann 2, J. Lermann 1, A. Boosz 1, F. C. Thiel 1,A .M ü l l e r1, M. W. Beckmann 1, S. P. Renner 1 Affiliations 1 Department of Gynecology and Obstetrics, Erlangen University Hospital, Friedrich-Alexander University Erlangen-Nuremberg, Erlangen 2 Institute of Pathology, Erlangen University Hospital, Friedrich-Alexander University Erlangen-Nuremberg, Erlangen Schlüsselwörter l" Endometriose l" Risiko l" Fall‑Kontroll‑Studie l" Epidemiologie l" Genetik Key words l" endometriosis l" risk factor l" case‑control study l" epidemiology l" genetics received 28. 10. 2011 revised 7. 11. 2011 accepted 7. 11. 2011

DOI http://dx.doi.org/ 10.1055/s-0031-1280436 Geburtsh Frauenheilk 2011; 71: 1073–1079 © Georg Thieme Verlag KG Stuttgart · New York · ISSN 0016‑5751 Correspondence Peter A. Fasching Department of Gynecology and Obstetrics Erlangen University Hospital Friedrich-Alexander University of Erlangen-Nuremberg Universitätsstraße 21–23 91054 Erlangen [email protected] 1073 Burghaus S et al. Risk Factors for … Geburtsh Frauenheilk 2011; 71: 1073 –1079 Original Article

! Endometriosis is a chronic disease that affects 4 –30 % of all wom- en of reproductive age [1 –3]. It is also one of the most frequent gynecological diseases. The prevalence of endometriosis in wom- en who present with infertility problems is even higher (up to 50 %) [3–6]. However it can reasonably be assumed that the prev- alence is about 10 % [3]. Pelvic pain during menstruation is the main symptom in patients with endometriosis. The disease is characterized by the presence of endometrial cells outside the uterus. They are located mainly in the rectouterine pouch, but can also be found in the vesicouterine pouch, abdominal wall, ovaries, and uterus [7 –9], and may also be more widespread at more unusual locations [10]. The increased rate of the disease among first-degree relatives of patients with endometriosis may also suggest a genetic predis- position [11]. Due to the recurrent nature of the condition, which hinders women in their occupational and private lives, endome- triosis creates substantial health-care costs, which have been es- timated as exceeding $ 22 billion in the United States in 2002 [12] and totaling approximately € 2 billion (2 000 000 000) in Ger- many [13]. Treatment options mainly consist of medication and surgical therapy. Surgical removal of the lesion is often the first-line ther- apy. Surgical treatment using minimally invasive approaches is usually preferred, but more extensive surgery may become nec - essary in cases of deeply infiltrating endometriosis [14, 15]. About 19 –45 % of all endometriosis patients suffer one or more recurrences within 5 years, leading to repeated operations in many patients [16–20]. Medication includes treatment with non- steroidal anti-inflammatory drugs (NSAIDs), antihormonal treat- ment with gonadotropin-releasing hormone (GnRH) analogs, or aromatase inhibitors [21 –23]. Some forms of treatment, such the selective estrogen receptor modulator raloxifene, even short- en the time to recurrence in endometriosis patients [24]. The pathogenesis of endometriosis is considered to be complex. Historically a metaplastic transformation of peritoneal cells or the still favourably discussed retrograde menstruation of endo- metrial cells through the tubes into the peritoneal cavity [26]. On a molecular level different pathways such as the estrogen and progesterone pathway, vasculogenesis, sphingolipids, pros- taglandins, and cytokines appear to be involved [27 –35]. Some information has been obtained about the etiology, but there is a lack of large case –control studies on the disease and especially of studies using a population-based design. There have been several reports linking characteristics of the menstrual cycle with endometriosis, such as age at menarche, duration of menstruation, and length of the cycle [2, 36 –42]. It is thought that an increased frequency of and duration of menstru- ations is associated with endometriosis. This applies to risk fac - tors such as early menarche, long duration of menstruation, short menstrual cycles, and lower parity. However, the data are not consistent with regard to risk estimates or the direction of the risk. Similarly inconsistent data have been reported for other risk fac - tors such as body mass index (BMI), physical activity, and smok- ing. There are some data suggesting an inverse relationship be- tween BMI and endometriosis, but many studies have not found this association [43]. The Nurses ʼ Health Study observed women prospectively and found no clear association between physical activity and endometriosis [44]. There are also inconsistent data with regard to smoking. It has been reported that smoking re- duces the estrogen level and thus the risk of endometriosis [40], but other studies did not identify any influence on the risk. There are also no differences between active and former smokers [2, 45]. Although the pathogenesis and etiology are poorly understood, there have been several studies that strengthen the hypothesis that endometriosis has complex genetic traits [46 –48]. Familial clustering has been described in several studies [49, 50]. A large linkage analysis including more than 1100 families with at least two cases of endometriosis in the family identified two loci, one on chromosome 10q26 and another on chromosome 20p13 [51]. However, follow-up candidate gene studies in these regions did not identify a specific gene or genetic variation [52]. A genome-wide association study (GWAS) in Japan identified a single nucleotide polymorphism (SNP) in CDKN2B‑AS and an SNP close to WNT4 at the genome-wide significance level [53]. A GWAS conducted in Australia, the United States, and the United Kingdom also identified an SNP in proximity to NFE2L3 and HOXA10. As the genetic causes of endometriosis are being investigated in more detail, the aim of the present investigation was to conduct a case–control study with epidemiological data to allow analysis not only of the genetic causes of the disease, but also of gene-en- vironment interactions. This study reports on the clinical data from this case –control study. Patients and Methods ! This analysis is based on 298 endometriosis cases and 300 con- trols, who were recruited between 2001 and 2010 in a pool of 595 endometriosis patients and 475 healthy controls as part of a hospital-based case –control study, the Bavarian Endometriosis Study (BENS). Women were eligible if they were aged at least 18 and were willing to complete an epidemiological questionnaire and provide a blood sample for genetic analysis. They were re- cruited as endometriosis patients if the disease was confirmed by surgery, either macroscopically (n = 18) or with histological examination (n = 281). Controls were eligible if they reported no previous abdominal surgery and no pelvic pain syndrome. All of the women provided written informed consent, and the ethics committee of the institution ʼs medical faculty approved the study. All of the women completed a structured and assisted question- naire designed to provide comprehensive epidemiological risk factor data on their previous reproductive history, menstrual cycle characteristics, previous medical history, family history, and lifestyle factors. The questionnaire has previously been used in other case –control studies on breast cancer [54 –56]. In addi- tion to these parameters, specific questions were asked about en- dometriosis, such as previous operations and pelvic pain charac - teristics (lower bowel pain, dyschezia, dysuria, and dyspareunia). Endometriosis had to be confirmed either histologically or mac - roscopically. The information was obtained either from the origi- nal surgery reports or the pathological reports from the patient charts. Statistical considerations The cases and controls were matched at a ratio of 1 : 1 by age at the time of diagnosis and interview, respectively, within deciles. Characteristics of cases and controls are presented as means and standard deviations, or counts and percentages. For each charac - teristic, a simple logistic regression model was used to calculate 1074 Burghaus S et al. Risk Factors for … Geburtsh Frauenheilk 2011; 71: 1073 –1079 Original Article unadjusted odds ratios (OR) and the corresponding 95 % confi- dence intervals. Multifactorial logistic regression analyses were carried out with the predictors age at menarche, menstrual cycle length, duration of menstrual bleeding, number of pregnancies, number of live births, number of miscarriages, use of contraceptive pills at any time, BMI, and smoking status, categorized as shown in l " Table 1 to identify a set of predictors that were together associated with endometriosis case –control status. Five hundred bootstrap sam- ples of the same size as the data set were selected with replace- ment. For each bootstrap sample, a stepwise backward logistic model selection procedure starting with all predictors was car- ried out to obtain the best model according to the Akaike infor- mation criterion. The retained predictors from each bootstrap sample were recorded, and a final variable selection was made by applying a procedure proposed by Sauerbrei and Schumacher [57] to our setting. In this procedure, the most frequent (> 70 %) predictors were selected, and, because of correlation, the predic - tor with the larger frequency out of each highly frequent predic - tor pair (> 90 %) was chosen. A multifactorial logistic regression model using these finally selected predictors was fitted to calcu- late adjusted ORs with its 95 % confidence intervals. Repetitive variable selections were carried out to stabilize the stepwise re- gression results [58]. The predictive ability of the final model was measured by the area under the curve (AUC) of the receiver operating characteris- tics. The AUC ranges from 0.5 (random prediction) to 1 (perfect prediction). In relation to overfitting, the AUC was evaluated with 10-fold cross-validation with 20 repetitions and with the 0.632+ bootstrapping method with 500 bootstrap samples [59, 60]. All of the tests were two-sided, and a p value < 0.05 was regarded as statistically significant. Calculations were carried out using the R system for statistical computing (version 2.13.1; R Develop- ment Core Team, Vienna, Austria, 2011).

! From a pool of 595 endometriosis patients and 475 controls, 298 endometriosis patients and 300 controls remained for the final analysis after matching for age. The average age of cases was 37.2 (± 8.7), while that of controls was 37.3 (± 9.4). Common patient characteristics for cases and controls are presented in l " Table 1. In the univariate analysis, age at menarche, menstrual cycle length, parity, number of miscarriages, and smoking status were found to differ significantly between cases and controls. Women with a higher age at menarche seemed to have a lower likelihood of being in the group of cases. The OR for women aged 15 or older at the time of menarche was 0.44 (95 % CI, 0.23 to 0.87) in com- parison with women who had the menarche at age 11 or earlier. Similarly, the groups of women who had the menarche at ages 12 and 14 had a lower likelihood of being in the case group ( l " Table 2). Women with a longer menstrual cycle ( ≥ 29 vs. ≤ 27 days) were more likely to be endometriosis patients (OR 1.73; 95 % CI, 1.09–2.75). Women who had a longer duration of menstrual bleeding had an OR of 1.48 (95 % CI, 0.96 –2.29), but this was not significant in the univariate analysis ( l " Table 2 ). Numbers of pregnancies and numbers of live births were distributed in a clearly ordinal way. The more pregnancies or live births were re- ported, the less likely the woman was to be in the group of endo- metriosis patients ( l " Table 2). Women with three or more preg- nancies had an OR of 0.30 (95 % CI, 0.18 –0.49). Use of oral contra- ceptives was also seen less often among cases (OR 0.48; 95 % CI, 0.29–0.78) and smoking was more prevalent in cases (OR 2.09; 95 % CI, 1.48–2.95). BMI and number of miscarriages were not as- sociated with case–control status (l " Table 2). However, the num- ber of miscarriages is not independent of the number of pregnan- cies. With regard to the multifactorial analysis, the variable selection process described above in the statistical methods section identi- fied menstrual cycle length, duration of menstrual bleeding, number of pregnancies, number of miscarriages, and smoking status as relevant predictors for endometriosis case –control sta- tus ( l " Table 2 ). The other predictors considered – age at men- arche, number of live births, use of contraceptive pills at any time, and BMI – were not selected for the final logistic regression model. This means that their predictive values appeared to be irrelevant, or that they can already be explained by the selected predictors. Menstrual cycle length and duration of menstrual bleeding were positively associated with endometriosis and had adjusted ORs of 2.33 (95 % CI, 1.34 –4.04) and 1.86 (95 % CI, 1.07 – 3.24), respectively. The inverse association with the number of pregnancies was even stronger, with an adjusted OR of 0.17 (95 % CI, 0.08–0.40) among women with three or more pregnan- cies, in comparison with the univariate analysis. Having one or more miscarriages in the medical history was identified as a sta- tistically significant risk factor (adjusted OR 2.76; 95 % CI, 1.28 – Table 1 Patient characteristics of endometriosis cases and healthy controls. Patient characteristics Cases Controls Age (years) Mean (SD) 37.2 (± 8.7) 37.3 (± 9.4) Age at menarche (years) ≤ 11 48 (16.5 %) 26 (8.9 %) 12 79 (27.1 %) 85 (29.1 %) 13 84 (28.9 %) 72 (24.7 %) 14 49 (16.8 %) 71 (24.3 %) ≥ 15 31 (10.7 %) 38 (13.0 %) Menstrual cycle length (days) ≤ 27 71 (31.8 %) 96 (35.0 %) 28 79 (35.4 %) 121 (44.2 %) ≥ 29 73 (32.7 %) 57 (20.8 %) Duration of menstrual bleeding (days) ≤ 4 78 (31.2 %) 123 (43.9 %) 5 71 (28.4 %) 80 (28.6 %) ≥ 6 101 (40.4 %) 77 (27.5 %) Number of pregnancies 0 157 (53.2 %) 107 (35.8 %) 1 61 (20.7 %) 55 (18.4 %) 2 47 (15.9 %) 69 (23.1 %) ≥ 3 30 (10.2 %) 68 (22.7 %) Number of live births 0 181 (61.4 %) 130 (43.6 %) 1 57 (19.3 %) 67 (22.5 %) 2 44 (14.9 %) 70 (23.5 %) ≥ 3 13 (4.4 %) 31 (10.4 %) Number of miscarriages 0 253 (86.3 %) 252 (84.6 %) ≥ 1 40 (13.7 %) 46 (15.4 %) Use of contraceptive pill, ever 0 52 (18.4 %) 29 (9.7 %) 1 231 (81.6 %) 269 (90.3 %) Body mass index (kg/m 2) ≤ 20 31 (12.3 %) 36 (12.3 %) 20–25 163 (64.4 %) 188 (64.2 %) 25–30 45 (17.8 %) 51 (17.4 %) > 30 14 (5.5 %) 18 (6.1 %) Smoking no 102 (41.6 %) 176 (59.9 %) yes 143 (58.4 %) 118 (40.1 %) 1075 Burghaus S et al. Risk Factors for … Geburtsh Frauenheilk 2011; 71: 1073 –1079 Original Article 5.93) independent of the number of pregnancies, for instance. Smoking status continued to show a significant association, with an adjusted OR of 2.16 (95 % CI, 1.39 –3.36). l " Fig. 1 provides an overview of the distribution of the risk factors finally selected in this case–control study. The apparent AUC value of the final regression model was 0.72, while the bootstrap-validated and cross-validated AUCs were both 0.68.

! Certain menstrual cycle characteristics, pregnancies, miscar- riages, and smoking were identified as risk factors associated with endometriosis in this case –control study in Germany. Body mass index did not show any association. Most of the results are in line with previously published studies, although some are not. With regard to menstrual cycle character- istics, it has been hypothesized that increasing exposure to men- strual bleeding is a risk factor for endometriosis [2, 38 –41], im- plying that early menarche, short cycles, longer duration of men- strual bleeding, and fewer pregnancies are risk factors. The present study confirms the duration of menstrual bleeding as a risk factor, but showed conflicting results with regard to the length of the menstrual cycle. Cases had a longer menstrual cycle, but the categorization differed from some other studies, as most of the participants in this study had fairly regular cycles, with most women having a cycle length of 28 days. Other studies have constructed categories ranging from ≤ 24 days to ≥ 31 days, re- sulting in categories with very small sample sizes and wide con- fidence intervals [42]. Another study has reported no association and show differences depending on which controls were used for the comparison [2]. A cohort study with data for cycle length in 688 women showed no association between cycle length and en- dometriosis in the overall group of women, but did find an asso- ciation in the group of women with no history of infertility [40]. The present study clearly shows an inverse association between the number of pregnancies and endometriosis, and a positive as- sociation between the number of miscarriages and endometrio- sis. This has also been observed in most of the published studies on the etiology of endometriosis. The Nurses ʼ Health Study II re- ported that the risk of endometriosis declines with an increasing number of pregnancies. This effect was clearly evident in women with no history of infertility and somewhat weaker in women with previous or concurrent infertility [40]. The relative risk in women with no pregnancies was 1.4 (95 % CI, 1.2–1.6) in compar- ison with women with two pregnancies. The effect size in the present study is comparable to that in other case –control studies [41]. The present study is also more consistent with other reports de- scribing no association between endometriosis and BMI [2, 5, 61]. However, there have been several studies reporting an associa- Table 2 Odds ratio (OR) for endometriosis cases and healthy controls in relation to patient characteristics (predictors).The unadjusted and adjusted OR, w ith 95 % confidence intervals in brackets, and the corresponding p values are shown. Predictor Values OR unadjusted 1 p value OR adjusted 2 p value Age at menarche (years) ≤ 11 1.00 (reference) –– 3 12 0.50 (0.29, 0.89) 0.02 13 0.63 (0.36, 1.12) 0.12 14 0.37 (0.21, 0.68) < 0.01 ≥ 15 0.44 (0.23, 0.87) 0.02 Menstrual cycle length (days) ≤ 27 1.00 (reference) – 1.00 (reference) – 28 0.88 (0.58, 1.34) 0.56 0.71 (0.43, 1.18) 0.19 ≥ 29 1.73 (1.09, 2.75) 0.02 2.33 (1.34, 4.04) < 0.01 Duration of menstrual bleeding (days) ≤ 4 1.00 (reference) – 1.00 (reference) – 5 0.71 (0.47, 1.10) 0.12 0.84 (0.48, 1.45) 0.53 ≥ 6 1.48 (0.96, 2.29) 0.08 1.86 (1.07, 3.24) 0.03 Number of pregnancies 0 1.00 (reference) – 1.00 (reference) – 1 0.76 (0.49,1.18) 0.21 0.71 (0.40, 1.28) 0.25 2 0.46 (0.29,0.72) < 0.001 0.48 (0.27, 0.88) 0.02 ≥ 3 0.30 (0.18,0.49) < 0.00001 0.17 (0.08, 0.40) < 0.0001 Number of live births 0 1.00 (reference) –– 3 1 0.61 (0.40,0.93) 0.02 2 0.45 (0.29,0.70) < 0.001 ≥ 3 0.30 (0.15,0.60) < 0.001 Number of miscarriages 0 1.00 (reference) – 1.00 (reference) – ≥ 1 0.87 (0.55,1.38) 0.54 2.76 (1.28, 5.93) < 0.01 Use of contraceptive pill, ever no 1.00 (reference) –– 3 yes 0.48 (0.29,0.78) 30 0.90 (0.39, 2.11) 0.81 Smoking no 1.00 (reference) – 1.00 (reference) – yes 2.09 (1.48,2.95) < 0.0001 2.16 (1.39, 3.36) < 0.001 1 OR estimated with simple logistic regression models; one model per predictor. 2 OR estimated by a multifactorial logistic regression model with variable selection, as described in the statistical methods section. ORs are adjust ed for all other predictors. 3 The predictors age at menarche, number of live births, use of contraceptive pills ever, and body mass index were dropped during the variable selection process. 1076 Burghaus S et al. Risk Factors for … Geburtsh Frauenheilk 2011; 71: 1073 –1079 Original Article tion between low BMI and endometriosis [36, 40, 62, 63]. In the Nursesʼ Health Study II cohort, physical activity was not strongly associated with the endometriosis risk [44]. In the present study, smoking was a clear risk factor in both the univariate analysis and the multivariate model. As with all the other risk factors described, published studies have both identi- fied and also failed to identify associations between smoking and endometriosis. Three larger case –control studies reported no difference between cases and controls with regard to smoking status [61, 62, 64]. Two earlier case–control studies also reported no relationship [2, 36]. The present study showed a positive asso- ciation between smoking (past and current) and endometriosis. The Nursesʼ Health Study II reported a more complex relationship between endometriosis and smoking status. In women who re- ported never having had an infertility problem, there was a pos- itive correlation between smoking status and endometriosis, whereas the association was inverse in the group of patients with an infertility history [65]. This study has several limitations and strengths. In endometrio- sis case –control studies, not only the selection of controls but also the selection of cases is critical with regard to the possible results. Case selection in a hospital-based study may be biased by the variety of health care that is provided in the hospital con- cerned. Women seeking help for pelvic pain might increase the numbers of cases of pain-inducing endometriosis, and a hospital with specialized health care for infertility patients might have larger numbers of patients in that subgroup. In the present study, the hospital is a center for all types of treatment, so that a bias for one of these groups seems unlikely. No previous abdominal sur- gery was required for the controls, in order to exclude any bias regarding the reason for surgery. This group might have larger numbers of women with less pelvic pain. However, a clear limita- tion is that the study is not population-based, so that any selec - tion bias could influence the results. One strength of the study is the robust selection of the variables during the bootstrap validation procedure, which yielded stable variable selection results. The final model predicted case –control status quite well, with a validated AUC of 0.68. In conclusion, the findings of the present study show both consis- tencies and inconsistencies with other published studies for al- most every risk factor. It is difficult for both patients and physi- cians to use the available information [66]. Only the number of pregnancies and the duration of menstrual bleeding appear to be consistent throughout the published studies and in this Ger- man case –control study. Assessing risk estimates in the popula- tion as determined in a hospital setting might improve our understanding of the nature of the condition for treatment of en- dometriosis patients and might also help identify differences be- tween clinical varieties of endometriosis. Duration of menstrual bleeding (days) Number of miscarriages Number of pregnancies Smoking status Menstrual cycle length (days) Number o fp a t i e n t s( % ) Number o fp a t i e n t s( % ) Number o fp a t i e n t s( % ) Number o fp a t i e n t s( % ) Number o f patients (%) ≤ 4 00 2 ≥ 3N o ≤ 4≥ 6 ≥ 65 ≥ 11 Y e s 5 50 40 30 20 10 0 100 80 60 40 20 0 60 50 40 30 20 10 0 80 60 40 20 0 50 40 30 20 10 0 Control Case Control Case Control Case Control Case Control Case Fig. 1 Distribution of risk factors between cases and controls. 1077 Burghaus S et al. Risk Factors for … Geburtsh Frauenheilk 2011; 71: 1073 –1079 Original Article Conflict of Interest ! None.

1 Bulun SE. Endometriosis. N Engl J Med 2009; 360: 268 –279 2 Signorello LB, Harlow BL, Cramer DW et al. Epidemiologic determinants of endometriosis: a hospital-based case-control study. Ann Epidemiol 1997; 7: 267 –274 3 Vigano P, Parazzini F, Somigliana E et al. Endometriosis: epidemiology and aetiological factors. Best Pract Res Clin Obstet Gynaecol 2004; 18: 177–200 4 Mahmood TA, Templeton A. Prevalence and genesis of endometriosis. Hum Reprod 1991; 6: 544 –549 5 Moen MH, Schei B. Epidemiology of endometriosis in a Norwegian county. Acta Obstet Gynecol Scand 1997; 76: 559 –562 6 Witz CA, Burns WN. Endometriosis and infertility: is there a cause and effect relationship? Gynecol Obstet Invest 2002; 53 (Suppl. 1): 2 –11 7 Mechsner S, Infanger M, Ebert AD et al. Abdominal wall endometriosis – diagnosis and therapeutic management. Geburtsh Frauenheilk 2010; 70: 207–214 8 Wobser M, Seitz CS, Becker JC et al. Umbilical endometriosis. Geburtsh Frauenheilk 2009; 69: 331 –333 9 Krentel H, Hucke J. Disseminated hormone-producing leiomyomatosis after laparoscopic supracervical hysterectomy: a case report. Geburtsh Frauenheilk 2010; 70: 894 –897 10 Goppel K, Becker K, Schmalfeldt B et al. Leiomyomatosis peritonealis dis- seminata – four case reports of a rare disease. Geburtsh Frauenheilk 2009; 69: 945 –951 11 Simpson JL, Elias S, Malinak LR et al. Heritable aspects of endometriosis. I. Genetic studies. Am J Obstet Gynecol 1980; 137: 327 –331 12 Simoens S, Hummelshoj L, D ʼHooghe T. Endometriosis: cost estimates and methodological perspective. Hum Reprod Update 2007; 13: 395 – 404 13 Brandes I, Kleine-Budde K, Mittendorf T. Cost of illness of endometriosis. Geburtsh Frauenheilk 2009; 69: 925 –930 14 Gardanis K, Zdichavsky M, Kramer M et al. Minimally invasive surgery for deep infiltrating endometriosis in the rectovaginal septum. Ge- burtsh Frauenheilk 2010; 70: 194 –200 15 Weichbrodt M, Gericke J, Riedlinger WF et al. Is sentinel lymph node ex- amination in patients with deep infiltrating rectovaginal endometrio- sis beneficial? Geburtsh Frauenheilk 2010; 70: 391 –396 16 Evers JL, Dunselman GA, Land JA et al. Is there a solution for recurrent endometriosis? Br J Clin Pract Suppl 1991; 72: 45–50; discussion 51-43 17 Redwine DB. Conservative laparoscopic excision of endometriosis by sharp dissection: life table analysis of reoperation and persistent or re- current disease. Fertil Steril 1991; 56: 628 –634 18 Renner SP, Ekici AB, Maihofner C et al. Neurokinin 1 receptor gene poly- morphism might be correlated with recurrence rates in endometriosis. Gynecol Endocrinol 2009; 25: 726 –733 19 Renner SP, Rix S, Boosz A et al. Preoperative pain and recurrence risk in patients with peritoneal endometriosis. Gynecol Endocrinol 2010; 26: 230–235 20 Wheeler JM, Malinak LR. Recurrent endometriosis: incidence, manage- ment, and prognosis. Am J Obstet Gynecol 1983; 146: 247 –253 21 Allen C, Hopewell S, Prentice A. Non-steroidal anti-inflammatory drugs for pain in women with endometriosis. Cochrane Database Syst Rev 2005; 4: CD004753 22 Hoshiai H, Terakawa N. Novel therapeutic targets for GnRH analogues in the treatment of endometriosis and current approaches to optimiz- ing GnRH analogue therapy. Introduction. Gynecol Obstet Invest 2008; 66 (Suppl. 1): 1 –2 23 Nawathe A, Patwardhan S, Yates D et al. Systematic review of the effects of aromatase inhibitors on pain associated with endometriosis. Bjog 2008; 115: 818 –822 24 Stratton P, Sinaii N, Segars J et al. Return of chronic pelvic pain from en- dometriosis after raloxifene treatment: a randomized controlled trial. Obstet Gynecol 2008; 111: 88 –96 25 Meyer R. Ueber den Stand der Frage der Adenomyositis, Adenomyome im Allgemeinen und insbesondere über Adenomyositis seroepithelialis und Adenomyometritis sarcomatosa. Zentralbl Gynakol 1919; 36: 745–750 26 Sampson J. Peritoneal endometriosis due to menstrual dissemination of endometrial tissue into the peritoneal cavity. Am J Obst Gynecol 1927; 14: 442 –469 27 Tseng JF, Ryan IP, Milam TD et al. Interleukin-6 secretion in vitro is up- regulated in ectopic and eutopic endometrial stromal cells from wom- en with endometriosis. J Clin Endocrinol Metab 1996; 81: 1118 –1122 28 Kao LC, Germeyer A, Tulac S et al. Expression profiling of endometrium from women with endometriosis reveals candidate genes for disease- based implantation failure and infertility. Endocrinology 2003; 144: 2870–2881 29 K a oL C ,T u l a cS ,L o b oSe ta l .Global gene profiling in human endometri- um during the window of implantation. Endocrinology 2002; 143: 2119–2138 30 Noble LS, Simpson ER, Johns A et al. Aromatase expression in endo- metriosis. J Clin Endocrinol Metab 1996; 81: 174 –179 31 Bulun SE, Zeitoun K, Takayama K et al. Estrogen production in endo- metriosis and use of aromatase inhibitors to treat endometriosis. En- docr Relat Cancer 1999; 6: 293 –301 32 Noble LS, Takayama K, Zeitoun KM et al. Prostaglandin E2 stimulates aromatase expression in endometriosis-derived stromal cells. J Clin Endocrinol Metab 1997; 82: 600 –606 33 Burney RO, Hamilton AE, Aghajanova L et al. MicroRNA expression profiling of eutopic secretory endometrium in women with versus without endometriosis. Mol Hum Reprod 2009; 15: 625 –631 34 Burney RO, Talbi S, Hamilton AE et al. Gene expression analysis of endo- metrium reveals progesterone resistance and candidate susceptibility genes in women with endometriosis. Endocrinology 2007; 148: 3814 – 3826 35 Gaetje R, Holtrich U, Engels K et al. Does sphingosine kinase 1 (SPHK1) play a role in endometriosis? Geburtsh Frauenheilk 2009; 69: 935 –939 36 Berube S, Marcoux S, Maheux R; Canadian Collaborative Group on Endo- metriosis. Characteristics related to the prevalence of minimal or mild endometriosis in infertile women. Epidemiology 1998; 9: 504 –510 37 Cramer DW, Wilson E, Stillman RJ et al. The relation of endometriosis to menstrual characteristics, smoking, and exercise. JAMA 1986; 255: 1904–1908 38 Eskenazi B, Warner ML. Epidemiology of endometriosis. Obstet Gynecol Clin North Am 1997; 24: 235 –258 39 Metzger DA, Haney AF. Etiology of endometriosis. Obstet Gynecol Clin North Am 1989; 16: 1 –14 40 Missmer SA, Hankinson SE, Spiegelman D et al. Reproductive history and endometriosis among premenopausal women. Obstet Gynecol 2004; 104: 965–974 41 Parazzini F, Ferraroni M, Fedele L et al. Pelvic endometriosis: reproduc - tive and menstrual risk factors at different stages in Lombardy, north- ern Italy. J Epidemiol Community Health 1995; 49: 61 –64 42 Treloar SA, Bell TA, Nagle CM et al. Early menstrual characteristics asso- ciated with subsequent diagnosis of endometriosis. Am J Obstet Gyne- col 2010; 202: 534, e531 –e536 43 Lafay Pillet MC, Schneider A, Borghese B et al. Deep infiltrating endome- triosis is associated with markedly lower body mass index: a 476 case – control study. Hum Reprod, advance online publication 24 October 2011; doi: 10.1093/humrep/der346 44 Vitonis AF, Hankinson SE, Hornstein MD et al. Adult physical activity and endometriosis risk. Epidemiology 2010; 21: 16 –23 45 Heilier JF, Donnez J, Nackers F et al. Environmental and host-associated risk factors in endometriosis and deep endometriotic nodules: a matched case-control study. Environ Res 2007; 103: 121 –129 46 Simpson JL, Bischoff F. Heritability and candidate genes for endometrio- sis. Reprod Biomed Online 2003; 7: 162 –169 47 Kennedy S. Genetics of endometriosis: a review of the positional clon- ing approaches. Semin Reprod Med 2003; 21: 111 –118 48 Giudice LC. Genomicsʼ role in understanding the pathogenesis of endo- metriosis. Semin Reprod Med 2003; 21: 119 –124 49 Kennedy S, Mardon H, Barlow D. Familial endometriosis. J Assist Reprod Genet 1995; 12: 32 –34 50 Stefansson H, Geirsson RT, Steinthorsdottir V et al. Genetic factors con- tribute to the risk of developing endometriosis. Hum Reprod 2002; 17: 555–559 51 Treloar SA, Wicks J, Nyholt DR et al. Genomewide linkage study in 1,176 affected sister pair families identifies a significant susceptibility locus for endometriosis on chromosome 10q26. Am J Hum Genet 2005; 77: 365–376 52 Treloar SA, Zhao ZZ, Le L et al. Variants in EMX2 and PTEN do not con- tribute to risk of endometriosis. Mol Hum Reprod 2007; 13: 587 –594 1078 Burghaus S et al. Risk Factors for … Geburtsh Frauenheilk 2011; 71: 1073 –1079 Original Article 53 Uno S, Zembutsu H, Hirasawa A et al. A genome-wide association study identifies genetic variants in the CDKN2BAS locus associated with en- dometriosis in Japanese. Nat Genet 2010; 42: 707 –710 54 Fasching PA, Loehberg CR, Strissel PL et al. Single nucleotide polymor- phisms of the aromatase gene (CYP19A1), HER2/neu status, and prog- nosis in breast cancer patients. Breast Cancer Res Treat 2008; 112: 89 – 98 55 Schrauder M, Frank S, Strissel PL et al. Single nucleotide polymorphism D1853 N of the ATM gene may alter the risk for breast cancer. J Cancer Res Clin Oncol 2008; 134: 873 –882 56 Heusinger K, Loehberg CR, Haeberle L et al. Mammographic density as a risk factor for breast cancer in a German case-control study. Eur J Cancer Prev 2011; 20: 1 –8 57 Sauerbrei W, Schumacher M. A bootstrap resampling procedure for model building: application to the Cox regression model. Stat Med 1992; 11: 2093 –2109 58 Simon R, Altman DG. Statistical aspects of prognostic factor studies in oncology. Br J Cancer 1994; 69: 979 –985 59 Steyerberg EW, Harrell jr. FE, Borsboom GJ et al. Internal validation of predictive models: efficiency of some procedures for logistic regres- sion analysis. J Clin Epidemiol 2001; 54: 774 –781 60 Adler W, Lausen B. Bootstrap estimated true and false positive rates and ROC curve. Comput Stat Data An 2009; 53: 718 –729 61 Hemmings R, Rivard M, Olive DL et al. Evaluation of risk factors associ- ated with endometriosis. Fertil Steril 2004; 81: 1513 –1521 62 Matalliotakis IM, Cakmak H, Fragouli YG et al. Epidemiological charac - teristics in women with and without endometriosis in the Yale series. Arch Gynecol Obstet 2008; 277: 389 –393 63 Parazzini F, Cipriani S, Bianchi S et al. Risk factors for deep endometrio- sis: a comparison with pelvic and ovarian endometriosis. Fertil Steril 2008; 90: 174 –179 64 Chapron C, Souza C, de Ziegler D et al. Smoking habits of 411 women with histologically proven endometriosis and 567 unaffected women. Fertil Steril 2010; 94: 2353 –2355 65 Missmer SA, Hankinson SE, Spiegelman D et al. Incidence of laparoscopi- cally confirmed endometriosis by demographic, anthropometric, and lifestyle factors. Am J Epidemiol 2004; 160: 784 –796 66 Zimmermann A, Brandes I, Babitsch B. Information needs of women with endometriosis within the scope of healthcare provision. Geburtsh Frauenheilk 2010; 70: 568 –573 1079 Burghaus S et al. Risk Factors for … Geburtsh Frauenheilk 2011; 71: 1073 –1079 Original Article