Human Reproduction, Vol.29, No.8 pp. 1666 –1676, 2014 Advanced Access publication on June 4, 2014 doi:10.1093/humrep/deu128

ORIGINAL ARTICLE Gynaecology

A clinical score can predict associated deep infiltrating endometriosis before surgery for an endometrioma M.C. Lafay Pillet 1,6,*, C. Huchon 5,7, P. Santulli 1,3,4, B. Borghese 1,2,3, C. Chapron 1,2,3,†, and A. Fauconnier 5,7,† 1

*Correspondence address. E-mail: [email protected]

Submitted on April 29, 2013; resubmitted on April 8, 2014; accepted on April 16, 2014

study question: Is it possible to detect associated deep infiltrating endometriosis (DIE) before surgery for patients operated on for endometriomas using a preoperative clinical symptoms questionnaire? summary answer: A diagnostic score of DIE associated with endometriomas using four clinical symptoms defined a high-risk group where the probability of DIE was 88% and a low-risk group with a 10% probability of DIE. what is known already: Many clinical symptoms are already known to be associated with DIE but they have not yet been used to build a clinical prediction model.

study design, size, duration: We built a diagnostic score of DIE based on a case control study of 326 consecutive patients operated on for an endometrioma between January 2005 and October 2011: 164 had associated DIE (DIE+) and 162 had no DIE (DIE2). We derived the score on a training sample obtained from a random selection of 2/3 of the population (211 patients, 101 DIE+, 110 DIE2), and validated the results on the remaining third (115 patients, 63 DIE+, 52 DIE2). The gold standard for the diagnosis of DIE was based on surgical exploration and histological diagnosis. participants/materials, setting, methods: Participants were consecutive patients aged 18 –42 years who underwent surgery for an endometrioma with histological confirmation and complete treatment of their endometriotic lesions: data for these women were extracted from a prospective database including a standardized preoperative questionnaire. On the training dataset, variables associated with DIE in a univariate analysis were introduced in a multiple logistic regression and selected by a backward stepwise procedure and a Jackknife procedure. A diagnostic score of DIE was built with the scaled/rounded coefficients of the multiple regression. Two cut-off values delimitated a high and a low risk group, and their diagnostic accuracy was tested on the validation dataset. main results and the role of chance: Four variables were independently associated with DIE: visual analogue scale of gastrointestinal symptoms ≥5 or of deep dyspareunia .5 (adjusted diagnostic odds ratio (aDOR) ¼ 6.0, 95% confidence interval (CI) [2.9 –12.1]), duration of pain greater than 24 months (aDOR ¼ 3.8, 95% CI [1.9– 7.7]), severe dysmenorrhoea (defined as the prescription of the oral contraceptive pill for the treatment of a primary dysmenorrhoea or the worsening of a secondary dysmenorrhoea) (aDOR ¼ 3.8, 95% CI [1.9– 7.6]) and primary or secondary infertility (aDOR ¼ 2.5, 95% CI [1.2 –4.9]). The sum of these variables weighted by their rounded/scaled coefficients constituted the score ranging from 0 to 53. A score ,13 defined a low-risk group where the probability of DIE was 10% (95% CI [7–15] with a sensitivity of 95% (95% CI [89 –98]) and a negative likelihood ratio of 0.1 (95% CI [0.0 –0.3]). A score ≥35 defined a high-risk group where the probability of DIE was 88% (95% CI [83–92%]), with a specificity of 94% (95% CI [87 –97]), and a positive likelihood ratio of 8.1 (95% CI [3.9 –17.0]). The performance of the score was confirmed on the validation dataset with 11% of DIE+ patients having a score ,13 (sensibility: 95%) and 90% of DIE+ patients having a score ≥35 (specificity: 94%). †

C.C. and A.F. contributed equally to the direction of the study.

& 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 Serials Department on October 26, 2014

Universite´ Paris Descartes, Sorbonne Paris Cite´, Faculte´ de Me´decine, Assistance Publique – Hoˆpitaux de Paris (APHP), Groupe Hospitalier Universitaire (CHU) Cochin, Department of Gynaecology II and Reproductive Medicine Paris, Paris, France 2Institut Cochin, Universite´ Paris Descartes, CNRS (UMR 8104), Paris, France 3Inserm, Unite´ de Recherche U1016, Paris, France 4Universite´ Paris Descartes, Sorbonne Paris Cite´, Faculte´ de Me´decine, AP-HP, Hoˆpital Cochin, Laboratoire d’immunologie, EA 1833, 75679 Paris, France 5Service de Gyne´cologie Obste´trique, CHI Poissy Saint-Germain en Laye, Poissy, Universite´ Versailles- Saint Quentin en Yvelines, Versailles, France 6INSERM, Unite´ de Recherche U953, Paris, France 7EA 7285 Risques cliniques et se´curite´ en sante´ des femmes, Universite´ Versailles-Saint-Quentin en Yvelines, Versailles, France

1667

A score to predict DIE associated with endometrioma

limitation, reasons for caution: This study was performed in a department specialized in DIE management. Score accuracy could be different in less specialized centres.

wider implications of the findings: This score could have a major clinical impact on the time of diagnosis, the management of DIE and could reduce the cost of investigations by helping to identify high-risk patients, while preserving the quality of care. study funding/competing interest(s): The authors have no competing interests to declare. No grant supported the study. Key words: deep endometriosis / endometrioma / chronic pelvic pain / examination / clinical prediction rule

Introduction

Methods This was a single centre prospective study in a gynaecology department of a teaching hospital that specialized in endometriosis.

Patients The population of patients was extracted from a prospective database and included all patients 18 – 42 years old that underwent gynaecological surgery by laparotomy or laparoscopy from January 2005 to October 2011 (excluding surgeries for cancer, pregnancy and prolapse). We extracted the patients from the database according to histological and surgical criteria: histological diagnosis of endometrioma and complete excision of the endometriotic lesions: –

–

Inclusion criteria were patients with histological diagnosis of endometriosis lesions that had, according to the findings of the operator, a complete treatment of endometriotic lesions. Exclusion criteria were incomplete surgical exploration that did not allow the exclusion of DIE.

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

Endometriosis is a common gynaecological pathology that can cause disabling pain (Fauconnier and Chapron, 2005) and reproductive failure (de Ziegler et al., 2010). It is defined as the presence of endometrial cells that have grafted and grown outside the uterine cavity (Sampson, 1927), reaching not only the reproductive system, but also the neighbouring organs of the abdominopelvic cavity (gastrointestinal system and urinary tract). Infiltration of cells to these further systems defines the development of deep infiltrating endometriosis (DIE). Different forms of endometriosis have been described (Chapron et al., 2006), which raises different problems for both the diagnosis and treatment of this disease: (i) superficial endometriosis, which implants in the area of the peritoneum, (ii) endometriosis characterized by deep penetration of lesions underneath the peritoneum, where the lesion is often multifocal and reaches several organs and (iii) the ovarian cyst or endometrioma. All of these forms are characterized by the same histological lesion, which involves the penetration of the stroma and glands to various depths within the tissues (Kamergorodsky et al., 2009). The diagnosis of endometrioma is easily performed using a transvaginal ultrasound, which has a sensitivity of 90% and a specificity of 97% (Guerriero et al., 2007). DIE is difficult to diagnose, and there is an important diagnosis delay for endometriosis (Hadfield et al., 1996; Husby et al., 2003), especially when the disease is severe (Matsuzaki et al., 2006). This form requires more specialized radiological assessments and a combinatorial approach (ultrasonography, magnetic resonance imaging (MRI) and rectal endoscopic ultrasonography) (Kinkel et al., 1999; Abrao et al., 2007; Piketty et al., 2009) to detect the DIE lesions preoperatively. These tests are expensive and sometimes poorly accepted by patients. In practice, there is a wide disparity in the radiological and endoscopic exploration of patients that undergo surgery for endometriosis cysts. In non-specialized centres, endometriomas are often operated on without any of these examinations, and DIE lesions are discovered intra-operatively, which does not allow for optimal treatment. On the other hand, in specialized centres, all of these imaging tests are often routinely ordered, resulting in an excess of unnecessary radiological examinations in patients at low risk of DIE. Both forms of endometriosis (endometrioma and DIE) are often associated with one another (Somigliana et al., 2004; Chapron et al., 2009), and only simultaneous excision of both types of lesions is effective in pain removal and a low rate of recurrence. In a study of 100 patients that had complete excision of the DIE lesions, the recurrence rate was 2% at 5 years post-surgery (Dousset et al., 2010). The rate of recurrence is due to incomplete excision (Abbott et al., 2003) and the growth of DIE lesions after surgery (Fedele et al., 2005), which contribute to persistent or recurrent pain and impact the quality of life (Garry et al., 2000).

Therefore, it is important to preoperatively diagnose patients that are at risk of having DIE lesions associated with endometrioma and to perform the appropriate radiologic imaging to guide the best surgical management (Chapron et al., 2012). Analysis of preoperative clinical data to determine which characteristics are associated with DIE has been the subject of only a few studies and failed to demonstrate a good diagnostic performance in DIE detection. However, several symptoms have been shown to be associated with DIE. Pain has been recognized by several studies as the main symptom of DIE lesions (Porpora et al., 1999; Fauconnier et al., 2002) and there is a strong association between DIE and the intensity of the pain (Koninckx et al., 1991; Chapron et al., 2003; Chopin et al., 2006), as well as the association of different types of pain (dyspareunia, dysmenorrhoea and pelvic pain) with increased risk of DIE (Sinaii et al., 2008). Some of the pain symptoms are related to specific anatomic localizations (Fauconnier and Chapron, 2005). The association of infertility with pain has also been related to the diagnosis of DIE (Forman et al., 1993). In addition, the clinical symptoms and medical history during adolescence differ for those found to have DIE during surgery compared with those who did not (Chapron et al., 2011). However, all of the clinical characteristics associated with DIE have never been combined to enable prediction of the presence of DIE. The aim of our study was to build a clinical model for the preoperative prediction of the association of DIE with endometrioma using a questionnaire focused on standardized preoperative symptoms and risk factors of the disease. The definition of this model will aid in the detection of DIE lesions in high-risk patients before surgery for endometrioma and will also decrease the use of unnecessary examinations in low-risk patients.

1668

Lafay Pillet et al.

Gold standard

Data collection The aim of the study was to build a score that could predict DIE based on clinical symptoms; the score was built using variables from the preoperative questionnaire independently of other possible preoperative tests ordered later by the surgeon. The answers from the standardized preoperative questionnaires were prospectively collected. The patient was referred by a physician or a gynaecologist with an ultrasound from a non-specialized radiologic centre and a diagnosis of an ovarian cyst. At the first preoperative visit, the patient’s questionnaire was completed. At that time, the surgeon did not know whether the diagnosis was an endometrioma or another kind of ovarian cyst and whether there was an associated DIE or not. We developed a database with the collected answers to three standardized questionnaires: (i) a preoperative questionnaire of all patients who had procedures by laparotomy or laparoscopy, regardless of the surgical indication (apart from cancer, pregnancy and prolapse), (ii) a post-operative questionnaire completed by the surgeon performing the procedure, including a description of the macroscopic surgical endometriotic lesions (lesion assessment, staging and samples taken) and (iii) a questionnaire that included the findings from the pathological examination of the surgical excisions. (i) The patient’s questionnaire was a preoperative questionnaire completed at the first preoperative consultation by the surgeon in charge of the patient. It contained 57 variables, including: – –

–

–

General data: age, BMI, family history, surgical history, obstetrical history and smoking habits. Gynaecological data: characteristics about the menstrual cycle (length and regularity) and menstruation (menorrhagia and total length of menstruation), the existence of a primary or secondary dysmenorrhoea, and the existence of primary or secondary infertility. History of symptoms and treatments during adolescence: age at menarche, primary dysmenorrhoea and its consequences on life (absenteeism from school and loss of consciousness), medication by analgesics and whether non-steroidal anti-inflammatory drugs and the oral contraceptive pill (OCP) were necessary, duration of treatments, and age at prescription. ‘OCP prescription for severe primary dysmenorrhoea’ refers to the history of prescription use during adolescence. The characteristics of all pains, both menstrual and non-menstrual: the presence or absence of pain, its duration and the worsening of a secondary dysmenorrhoea. Five painful symptoms were measured

– –

Other questions about possible symptoms related to endometriosis: rectal bleeding and haematuria. Finally, details of treatments prescribed were collected: mode of contraception, treatment for endometriosis (progestins, OCP, GnRH agonist, etc.) and their details of prescription (current or past use, age at the time of prescription, and duration).

A study of the correlation between the variables and their localizing value which had been previously demonstrated (Fauconnier and Chapron, 2005) led us to create a composite variable from the VAS of GI symptoms and the VAS of dyspareunia. Two variables were combined to define severe dysmenorrhoea: ‘OCP prescription for treatment of severe primary dysmenorrhoea’ and ‘worsening of secondary dysmenorrhoea. (i) This questionnaire was completed post-operatively by the surgeon using a form for a standardized description of the lesions visualized, their size, their location, their number, their character (deep versus superficial), the existence of adhesions, and whether the pouch of Douglas was obliterated. The completeness of the excision, staging of the lesions, samples taken and their location were all specified. These data were used for the macroscopic diagnosis of the lesions. (ii) This questionnaire was completed post-operatively and included histological findings and confirmation of endometriosis within the surgically excised lesions. Pathologists were unaware of the clinical symptoms. These data were used for the final diagnosis and classification of patients. Written consent was obtained from the included patients, and the constitution of the database was authorized by the ethics committee (Advisory Committee on Protection of Persons in Biomedical Research) of Cochin Hospital, Paris, France.

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

The diagnosis and classification of the lesions were based on the macroscopic surgical results and a histological examination of the lesions where an excision was performed. A macroscopic anatomical classification was carried out according to a previously described method (Chapron et al., 2006). The histological diagnosis was based on characteristics previously defined in a standardized histological description, comprising the combination of endometrial glands and stroma (Kamergorodsky et al., 2009). The combination of macroscopic and histological data served as a reference for the definition of the lesions and the classification of patients (endometrioma and associated DIE lesions). Patients were defined to be DIE+ if they met the criteria of the gold standard (macroscopic lesions and histological diagnosis according to the previous description), regardless of the side or the number of endometriomas. The patients included in the study were divided into two groups: patients that underwent surgery for an endometrioma associated with DIE (DIE+) and patients that underwent surgery and were not associated with DIE lesions (DIE2).

by five visual analogue scales (VAS) from 0 to 10 (Peveler et al., 1996). They related to primary or secondary dysmenorrhoea, dyspareunia, pain of gastrointestinal (GI) origin, pain from urinary tract symptoms and noncyclic chronic pelvic pain. Patients were asked at the first preoperative visit about their worst pain in the last 6 months. For patients using hormonal treatment before surgery, VAS pain scores were evaluated regardless of any treatment. If patients had no menstruation for .6 months, the dysmenorrhoea VAS score was 0, as well as other variables defined by cyclic variations. If patients had no sexual activity, the dyspareunia VAS score was 0. The variable ‘pain’ was defined by the presence of at least 6 months of dysmenorrhoea, and/or intermenstrual pelvic pain, and/or dyspareunia (Fedele et al., 2007). Primary dysmenorrhoea was defined as painful menstruations occurring shortly after menarche, though it could also occur as late as 1 year after menarche (Dawood, 2006). Secondary dysmenorrhoea was defined as painful menstruations occurring later. ‘Worsening of a secondary dysmenorrhoea’ was defined according to the patient’s evaluation, as long as she was able to give the duration of the worsening symptoms. Infertility was defined by at least 12 months of unprotected intercourse in which pregnancy was not achieved (Marcoux et al., 1997). Patients were coded as non-infertile if they did not try to get pregnant. The GI symptoms were defined as pain when defecating at the time of menstruation or intestinal cyclic pain. If the patient had no menstruation for .6 months, the GI symptoms could not be evaluated during menstruation or during the cycle, and the VAS score was then coded 0. Chronic pelvic pain was defined as noncyclic pain located in the pelvic area, apart from dyspareunia. Menorrhagia was defined as ‘heavy bleeding’ during menstruations according to the patient’s evaluation (Warner et al., 2004). The ‘length of menstruation’ was another variable indicating the number of days of menstruation, which was independent of the patient’s opinion.

1669

A score to predict DIE associated with endometrioma

Statistical methods

Results For our study, 343 patients met the inclusion criteria; however, 17 patients were excluded because surgical exploration was considered incomplete due to partial or complete obliteration of the pouch of Douglas. Without complete exploration, it would not be possible to perform the complete excision of the lesions. Of the remaining 326 patients, 162 patients showed no associated DIE lesions (DIE2) and 164 patients had DIE lesions associated with endometrioma (DIE+), based on the histological criteria of the surgical excision of the organs (Table I). After a random selection, the training sample included 211 patients: 101 patients DIE+ and 110 patients DIE2; the validation sample included 115 patients: 63 patients DIE+ and 52 patients DIE2. The main characteristics of the DIE+ and DIE2 patients from the derivation dataset are summarized in Table II. A univariate analysis comparing the DIE+ and DIE2 patients did not reveal any significant differences in the general variables (age, BMI, gravidity, parity and family history). Indeed, the variables significantly associated with DIE in the univariate analysis were all variables related to pain (presence of pain, duration of pain and intensity of the five types of pain evaluated by VAS), infertility, symptoms/treatment history during adolescence, the severity of primary dysmenorrhoea as assessed by the need to prescribe estrogenprogestin, and the severity of secondary dysmenorrhoea as evaluated by the notion of worsening. Prior history of surgery for endometrioma and/or endometriosis was more frequent in DIE+ patients, but we did not find a significant interaction of past surgery on the relationship between VAS scores and DIE (data shown in Supplementary data, Table SI). There were significantly more DIE+ patients using hormonal treatments at the time of surgery, but the VAS scores were not lower for the patients using treatment (data shown in Supplementary data, Tables SII and SIII).

Table I Description of the population of patients with endometrioma (n 5 326). Endometrioma

....................................................................... Right

Left

Bilateral

Types of associated DIE according to localizationa

Number of patients with endometrioma

............................................................................................................................................................................................. Total

93

143

90

326

Isolated

53

71

38

162

Associated with DIE

40

72

52

Bowel

27

53

41

121

Bladder

5

5

9

19

164

Uretere

3

10

5

18

Vagina

15

40

30

85

6

7

9

22

US Right Left Bilateral

6

20

3

29

15

24

26

65

DIE, deep infiltrating endometriosis; US, utero-sacral ligaments. a Several localizations can be associated in the same patient, the diagnosis is based on histological findings of surgical specimen, and the lumen of the organs was entered.

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

A random sample of two-thirds of the patients was obtained prior to the analysis to develop a clinical prediction model (training sample), leaving one-third of the patients for validation (validation sample). We first performed a univariate analysis comparing DIE+ and DIE2 patients from the training sample for each variable of the preoperative questionnaire, using a quantitative (Student’s t-test) or qualitative (x 2 test) test as appropriate. Continuous variables associated with the presence of a DIE at a threshold of P , 0.10 were then converted into dichotomous variables using receiver operating characteristic (ROC) curves. Crude diagnostic odds ratios (OR) were calculated for each variable significantly associated with the diagnosis of a DIE. The diagnostic performance of each variable was assessed using sensitivity, specificity, positive likelihood ratio (LR+), negative likelihood ratio (LR2) and the C-index. We then used a multiple logistic regression to select the best combination of variables that was independently associated with the diagnosis of DIE (P , 0.05). Variables were selected by a backward, stepwise procedure from those significantly associated with DIE in the univariate analysis at a threshold of P , 0.10. Adjusted diagnostic odds ratios (aDOR) were calculated with their 95% confidence intervals (95% CI). A jackknife procedure was applied to the model to detect variables potentially responsible for instability in the model and that had to be removed (Efron and Gong, 1983). The performance of the final model in the prediction of DIE was specified by calculating its sensitivity, specificity and area under the curve. The calibration of the model was evaluated using the Hosmer– Lemeshow test (Lemeshow and Hosmer, 1982). We then built a score by rounding up the b coefficients from the multivariate analysis (Coste et al., 1997). The C-index of the logistic regression model and score model were compared to verify that the rounded coefficients did not change the results from the model. We classified patients as high risk or low risk by choosing two threshold values of the score to make two classifications, one with sensitivity high enough to provide an overall LR+ of ≥4.0 and the other maximizing specificity to allow a LR2 of ≤0.25 (Buckley et al., 1998). Finally, the score was applied to the validation sample. The sensitivity and specificity of the score, as well as the probabilities of DIE, in the high-risk and the low-risk groups were compared with those obtained from the training sample.

Statistical analyses were performed using STATA statistical software version 12.1 (Stata Corp.; College Station, TX, USA).

1670

Lafay Pillet et al.

Table II Continued

Table II Preoperative characteristics of DIE1 and DIE2 patients in the training sample. DIE2 patients N 5 110 mean + SD or n (%)b

DIE1 patients N 5 101 mean + SD or n (%)b

DIE2 patients N 5 110 mean + SD or n (%)b

P-value

DIE1 patients N 5 101 mean + SD or n (%)b

P-value

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

Age (year)

31.5 + 5.5

32.2 + 5.0

0.36

Height (cm)

165.3 + 6.8

166 + 6.5

0.42c

Rectal bleeding

8 (7.3)

17 (16.8)

0.032

Weight (kg)

59.9 + 10.2

63.0 + 12.9

0.042c

Haematuria

5 (4.5)

3 (3.0)

0.56d

BMI (kg/m2)

22.0 + 4.0

22.8 + 4.1

0.13c

Menorrhagia

46(41.8)

64(63.4)

0.002d

Length of menstruation (days)

4.9 + 0.16

5.9 + 0.25

0.002c

c

Gravida (number of pregnancies) 77 (70.0)

77 (76.2)

1

22 (20.0)

16 (15.8)

≥1

11 (10.0)

7 (6.9)

0.47d

Never

25 (22.7)

6 (5.9)

Ever

61 (55.5)

81 (80.2)

Current

24 (21.8)

13 (12.9)

Preoperative hormonal treatments (GnRHa, progestins, OCP, else)

56 (50.9)

84 (83.2)

,0.001d

0.51d

Prior surgery for endometriosis

28 (25.5)

68 (67.3)

,0.001d

0.005d

Prior surgery for endometrioma

18 (16.4)

41 (40.6)

,0.001d

Parity (number of deliveries) 0

95 (86.4)

85 (84.2)

1

9 (8.2)

12 (11.9)

≥1

6 (5.5)

3 (3.0)

13 (11.8)

15 (14.8)

Family history of endometriosis Infertility

d

0.57

None

79 (71.8)

Primary

22 (20.0)

40 (39.6)

Secondary

8 (7.3)

11 (10.9)

Duration of infertility (months)

12.2 + 30.1

25.2 + 34.0

Paina

75 (68.2)

90 (89.1)

,0.001d

Duration of paina (months)

33.5 + 50.0

73.9 + 68.8

,0.001c

50 (49.5)

15 (13.6)

2 (2.0)

0.004c

Dysmenorrhoea No

,0.001d

OCP

0.008d

Yes primary

53 (48.2)

48 (47.5)

Yes secondary

41 (37.3)

50 (49.5)

VAS dysmenorrhoea (n ¼ 211)

6.1 + 2.8

7.7 + 2.0

,0.001c

VAS deep dyspareunia (n ¼ 201)

3.4 + 3.2

4.8 + 3.4

,0.001c

VAS NCPP (n ¼ 211)

2.4 + 3.1

3.8 + 3.0

,0.001c

VAS GI symptoms (n ¼ 211)

2.0 + 2.9

5.8 + 3.1

,0.001c

VAS urinary tract pain (n ¼ 211)

0.3 + 1.4

1.5 + 2.8

,0.001c

OCP for severe primary dysmenorrhea

12 (10.9)

23 (22.8)

0.021d

Worsening of secondary dysmenorrhea

15 (13.6)

31 (30.7)

0.003d

Age at first menstruations 13.0 + 0.2

12.8 + 0.2

0.43

Absence from school because of primary dysmenorrhoea

40 (39.6)

0.0847d

Infertility: at least 12 months of unprotected intercourse which did not achieve pregnancy. DIE, deep infiltrating endometriosis; DIE+, associated with DIE; DIE2, no association with DIE; VAS: Visual Analogue Scale; NCPP: non-cyclic chronic pelvic pain; GI symptoms ¼ pain when defecating at the time of menstruation or intestinal cyclic pain; OCP: oral contraceptive pill; GnRHa: GnRH agonist; primary dysmenorrhea: menstruations occurring shortly after menarche (up to 1 year), secondary dysmenorrhoea occurs later; menorrhagia: heavy bleeding according to the patient. a Presence of at least 6 months of dysmenorrhoea and/or dyspareunia and/or non-chronic pelvic pain and/or gastro-intestinal (GI) pain. b The percentages are calculated on the total number of patients despite some missing values. c P: Student t-test. d P: chi2 test.

VAS pains scores

28 (25.5)

Continued

Most of the VAS scores were complete, except for 10 missing values for dyspareunia that were input as absent. There were very few missing values for other variables (,5% for all variables), and no significant correlation was found between a class of missing values and the variable of interest (DIE); all missing values were coded 0 by simple imputation. Dichotomous categorical variables significantly associated in the univariate analysis with DIE (P , 0.10) were evaluated for their diagnostic performance. Continuous variables significantly associated in the univariate analysis with DIE (P , 0.10) were dichotomized using ROC curves, and their diagnostic performance was then also evaluated; their sensitivity, specificity, LR+ and LR2 are shown in Table III. The composite variable (GI symptoms ≥5 or deep dyspareunia .5) was statistically different between DIE+ and DIE2 patients (P , 0.001; Table III). The variable ‘severe dysmenorrhoea’ previously defined was statistically different between DIE+ and DIE2 patients (P , 0.001; Table III). The multiple logistic regression analysis from the 14 variables introduced into the model selected by a backward stepwise procedure identified six variables independently significantly associated with the

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

0

23 (22.8)

0.06d

Loss of consciousness due 15 (13.6) to dysmenorrhea

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

1671

A score to predict DIE associated with endometrioma

Table III Diagnostic performances of dichotomized variables associated with DIE1 according to univariate analysis. Variables

Population total (n)

Se (%)

Sp (%)

LR1

LR2

C index

Crude OR

P

............................................................................................................................................................................................. Infertility (primary or secondary)

210

50.5

72.5

1.84

1.0

0.68

1.5

0.003

Pain

207

92.3

31.8

1.4

0.2

0.62

6.0

,0.001

Duration of pain .24 months

199

62.0

81.3

3.3

0.5

0.72

7.1

,0.001

211

81.2

49.1

1.6

0.4

0.65

4.2

,0.001 ,0.001

VAS pains scores VAS dysmenorrhoea ≥7

211

66.3

61.8

1.5

0.7

0.60

2.2

200

69.0

59.0

1.7

0.5

0.64

3.2

0.007

VAS GI symptoms ≥5

211

75.2

75.5

3.1

0.3

0.75

9.3

,0.001

VAS urinary tract pain.1

211

25.7

93.6

4.0

0.8

0.60

5.1

,0.001

Absence from school or loss of consciousness during menstruations

209

44.0

69.7

1.4

0.8

0.57

1.8

0.04

OCP for severe primary dysmenorrhoea

207

23.5

89.0

2.1

0.9

0.56

2.5

0.014

Worsening of secondary dysmenorrhoea

207

63.3

62.5

1.7

0.6

0.63

2.9

0.003

Age ,25 years

211

93.1

16.4

1.1

0.4

0.55

2.6

0.034

Age at first menstruation ,12 years

210

46.0

68.2

1.4

0.8

0.57

1.8

0.035

Length of menstruation .6 days

202

37.4

78.6

1.7

0.8

0.58

2.2

0.012

Severe dysmenorrhoea

211

54.5

74.5

2.1

0.6

0.65

3.5

,0.001

VAS deep dyspareunia .5 or VAS GI pain ≥5

211

80.2

62.7

2.2

0.3

0.71

3.5

,0.001

Rectal bleeding

211

16.8

92.7

2.3

0.9

0.55

2.6

0.032

Menorrhagia

209

64.0

57.8

1.52

0.6

0.61

2.4

0.002

Infertility: at least 12 months of unprotected intercourse which did achieve pregnancy coded ‘no’ if the patient does not desire to get pregnant ; pain: presence of at least 6 months of dysmenorrhoea and/or dyspareunia and/or NCPP and/or GI pain; GI symptoms ¼ pain when defecating at the time of menstruation or intestinal cyclic pain; severe dysmenorrhoea ¼ OCP for primary dysmenorrhoea or worsening of secondary dysmenorrhoea; primary dysmenorrhoea: menstruations occurring shortly after menarche (up to 1 year), secondary dysmenorrhoea occurs later; menorrhagia: heavy bleeding according to the patient. Se, sensibility; Sp, specificity; OR, odds ratio; LR, likelihood ratio.

diagnosis of DIE, namely duration of pain, severe dysmenorrhoea, VAS GI pain or dyspareunia, infertility, VAS urinary symptoms and menorrhagia (Table IV). Among these variables, two (VAS urinary symptoms and menorrhagia) were unstable according to the jackknife procedure, and we therefore removed them from the model. A simplified model with four variables was finally selected and had a diagnostic performance close to the full model (Table IV). The C-index of this model was 0.84 (95% CI 0.79 –0.90) and the Hosmer –Lemeshow test confirmed the suitability of the model for the data (P ¼ 0.23). We used the b coefficients from the logistic regression and rounded and multiplied by 10 for easier use. The DIE score (Table V) was then given by the following equation: Score ¼ [(dyspareunia VAS .5 or GI symptoms VAS ≥5) × 18] + (pain duration .24 months × 13) + (severe dysmenorrhoea × 13) + (existence of infertility × 9). There were no significant losses of fit due to rounding the coefficient, as shown by the C-index score of 0.84 (95% CI 0.79 –0.90), which was similar to the C-index of the logistic regression model (P ¼ 0.74). We then defined a low- and high-risk group of DIE by choosing two cut-off points of the score, with an LR2 ≤0.25 and an LR+ ≥4: –

Patients with a score of ,13 were defined as a low-risk group, where the probability of DIE was 10% (95 CI 7– 15). A threshold score value of 13 produced a sensitivity of 95% (95% CI 89–98)

–

and a specificity of 42% (95% CI 32–52), with LR2 equal to 0.1 (95% CI 0.0 –0.3). Patients with a score ≥35 were defined as a high-risk group, where the probability of DIE was 88% (95% CI 83– 92). A threshold value of 35 produced a specificity of 94% (95% CI 87– 97) and a sensitivity of 51% (95% CI 41–62), with a LR+ of 8.1 (95% CI 3.9 –17).

Next, the score was calculated for the patients in the validation sample (Fig. 1). The observed percentages of DIE were in the expected range: 11% in the low-risk group and 90% in the high-risk group. The C-index of the score for the prediction of DIE (0.84, 95% CI 0.76 –0.91) was not significantly different from the C-index of the training sample (0.84, 95% CI 0.79 –0.90; P ¼ 0.99) (Fig. 2).

Discussion In this study, we developed a prediction score for the presence of DIE associated with endometriotic cysts based on a simple preoperative questioning of the patient with four variables: dyspareunia VAS score .5 or GI symptoms ≥5 (coefficient: 18); duration of pain for over 24 months (coefficient: 13); prescription of OCP for severe primary dysmenorrhoea or worsening of secondary dysmenorrhoea (coefficient: 13); and primary or secondary infertility (coefficient: 9). Their value are 0 if absent or 1 if present, and they are multiplied by their coefficients

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

VAS NCPP ≥3 VAS deep dyspareunia .5

1672

Lafay Pillet et al.

Table IV Factors associated with DIE in the multiple logistic regression models (full and simplified models after downward stepwise multiple regression analysis and Jackknife procedure). aOR

95% CI

P jackknife

AUC

............................................................................................................................................................................................. Full model (14 variables)

0.87

Model selected by downward stepwise regression analysis (6 variables)

0.86

Duration of pain .24 months

3.3

1.6– 7.1

0.002 0.000

VAS dyspareunia .5 or GI symptoms ≥5

2.2

2.6– 12.5

Severe dysmenorrhoea

3.7

1.7– 7.9

0.001

VAS urinary tract symptoms.1

2.9

0.8– 11.4

0.118

Infertility (primary or secondary)

2.4

1.1– 5.0

0.002

Menorrhagia

2.0

9.6– 4.2

0.065

Duration of pain .24 months

3.8

1.9– 7.7

0.000

VAS dyspareunia .5 or GI symptoms ≥5

6.0

2.9– 12.1

0.000

Severe dysmenorrhoea

3.8

1.9– 7.6

0.000

Infertility (primary or secondary)

2.5

1.2– 4.9

0.015

Simplified model (4 variables)

0.84

and added to get the score ranging from 0 (0 + 0 + 0 + 0) to 53 (18 + 13 + 13 + 9). The choice of two cut-off points to define a population at low risk (score ,13) and a population at high risk (score ≥35) provided a clinical prediction rule with a good diagnostic performance: the low-risk cut-off point defined a model with a sensitivity of 95%, and the high-risk cut-off point defined a model with a specificity of 94%. This classification offers the advantage of avoiding unnecessary radiological investigations in patients at low risk of DIE, while tracking the largest number of patients that require specific surgical management for associated DIE. The choice between the two score cut-off points (with high sensitivity or with high specificity) to order radiological investigations depends on the main objectives set in advance: the clinical and economic context or the personal choice of the patient. On the whole, the high sensitivity model would have avoided unnecessary radiological explorations for 30% of the controls, while 97% of DIE would have been detected. This model should be used by specialized surgical services, where examinations are usually almost always undertaken. This will target the patient population where it is not necessary to use a complete radiological investigation. The high specificity model would have avoided unnecessary radiological explorations for 91% of the controls, while 51% of DIE would have been detected. This model should be used by less specialized teams for whom investigations are in practice rarely ordered. This will target the high-risk patients for whom assessment is essential for proper surgical management. We did not build a single score for every organ; instead, we defined a population at high risk of DIE who would need a complete radiologic evaluation, including at least an MRI and computed tomography scan and, based on imaging results and clinical symptoms, more specific explorations (rectal endoscopic ultrasonography and uro-CT scan).

Table V Number of DIE score points contributed by each factor and clinical prediction rule. Variables

Score points

........................................................................................ VAS dyspareunia .5 or GI symptoms ≥5 Yes

18

No

0

Duration of pain .24 months Yes

13

No

0

Severe dysmenorrhoea Yes

13

No

0

Infertility (primary or secondary) Yes

9

No

0 Score ¼ sum of points

Predicted risk [95% CI]

,13: low risk

10% [7–15]

≥35: high risk

88% [83–92]

GI symptoms ¼ pain when defecating at the time of menstruation or intestinal cyclic pain; pain ¼ presence of at least 6 months of dysmenorrhoea and/or dyspareunia and/or NCPP and/or GI pain; severe dysmenorrhoea ¼ OCP prescription for severe primary dysmenorrhoea or worsening of a secondary dysmenorrhoea; primary dysmenorrhoea: menstruations occurring shortly after menarche, secondary dysmenorrhoea occurs later; infertility : at least 12 months of unprotected intercourse which did not achieve pregnancy coded ‘no’ if the patient does not desire to get pregnant.

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

Infertility: at least 12 months of unprotected intercourse which did not achieve pregnancy coded ‘no’ if the patient does not desire to get pregnant; pain: presence of at least 6 months of dysmenorrhoea and/or dyspareunia and/or chronic pelvic pain and/or GI pain; GI symptoms ¼ pain when defecating at the time of menstruation or intestinal cyclic pain; severe dysmenorrhoea ¼ OCP for primary dysmenorrhoea or worsening of secondary dysmenorrhoea; primary dysmenorrhoea: menstruations occurring shortly after menarche (up to 1 year), secondary dysmenorrhoea occurs later; menorrhagia: heavy bleeding according to the patient. AUC, area under the receiver operating characteristic curve; aOR, adjusted odd ratio; CI, confidence interval.

A score to predict DIE associated with endometrioma

1673

This classification is easy and helpful for the surgeon, gynaecologist or general practitioner to decide whether to order radiological investigations and whether to refer the woman to a specialized surgical centre. Our study confirms previously published studies on pain, infertility and disease history. Indeed, the variables in the final model have been the subject of previous studies, whose findings are in agreement with our results: pain is recognized as a key symptom of the disease, particularly for DIE (Vercellini et al., 1996, 2007). Different types of pain have been shown to correspond to specific locations of the disease. For example, dyspareunia and pain with defecation are correlated and significantly related to the presence of a nodule in the rectovaginal septum (Fauconnier and Chapron, 2005), which could justify the creation of a composite variable from these two parameters. The combination of endometrioma and pain is significantly related to the simultaneous presence of DIE (Chopin et al., 2006; Chapron et al., 2012). Infertility, when combined with pelvic pain, has been shown to be often associated with DIE (Koninckx et al., 1991). The history of patients at the time of adolescence has revealed that some events or symptoms in early menstruation are statistically more frequently associated with a later surgical diagnosis of DIE (Chapron et al., 2011). The duration of pain and other related symptoms are also often associated with the existence of DIE (Hadfield et al., 1996; Husby et al., 2003). Our study confirmed all of these findings and in combining them, provided a good prediction of DIE in a group of patients that underwent surgery for an endometrioma.

The strength of this study lies in its prospective assessment of pain using a standardized questionnaire that employed visual scales validated for each type of pain known to be associated with endometriosis. Furthermore, the diagnosis of DIE was based on strict criteria: all patients underwent complete surgical exploration, and diagnosis was always confirmed histologically. The choice of a population with a surgical indication related to the presence of an endometrioma minimized the selection bias of DIE due to surgical recruitment; in most studies on endometriosis, the main problem with selecting patients based on a histological diagnosis is that only patients having surgery are included in the study. The indication for surgery for DIE depends on the characteristics of DIE, such as the intensity of pain or specific localization, creating a bias of selection. However, most ovarian cysts are an indication for surgery independently of associated DIE characteristics, thereby minimizing the bias selection for DIE patients. The population of 163 patients with DIE and associated endometrioma allowed us to use a validation sample without any prejudice for the power of the study. Internal validation of the model limits the possibility of a model overfitting the data. Random selection of one-third of the population for validation of the model built on the other two-thirds enabled this validation. In addition, the estimate of the variance of the coefficients by the jackknife method avoided the use of variables responsible for instability of the model, which might have affected the results of the logistic regression model (Huchon et al., 2010). These two methods provided two strong ways of validation. The results from the validation

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

Figure 1 Repartition of DIE+ and DIE2 patients according to their score values on the training and validation sample. The histogram on the left shows the results of the score derived on the training sample with deep infiltrating endometriosis (DIE) patients with the highest scores and controls with the lowest scores, and the histogram on the right shows the results of the score applied in a second step to the validation sample showing similar results. The score is calculated for every patient according to the following equation obtained from the logistic regression on the training sample: Score ¼ [(VAS dyspareunia .5 or GI symptoms ≥ 5) × 18] + [(pain duration .24 months) × 13] + [(severe dysmenorrhoea) × 13] + [(infertility) × 9]. The four variables dyspareunia VAS score .5 or GI symptoms . or ¼ 5 (coefficient: 18); duration of pain for over 24 months (coefficient: 13); prescription of OCP for severe primary dysmenorrhoea or worsening of secondary dysmenorrhoea (coefficient: 13); and primary or secondary infertility (coefficient: 9) take the value 0 if absent or 1 if present; they are multiplied by their coefficients and added to get the score ranging from 0 (0 + 0 + 0 + 0) to 53 (18 + 13 + 13 + 9). Values on the x-axis are all the possible values of the score. VAS, visual analogue scale; GI, gastrointestinal; OCP, oral contraceptive pill. GI symptoms ¼ pain when defecating at the time of menstruation or intestinal cyclic pain; pain ¼ presence of at least 6 months of dysmenorrhoea and/or dyspareunia and/or non-chronic pelvic pain and/or gastro-intestinal pain; severe dysmenorrhoea ¼ OCP prescription for severe primary dysmenorrhoea or worsening of a secondary dysmenorrhoea; primary dysmenorrhoea: menstruations occurring shortly after menarche, secondary dysmenorrhoea occurs later; infertility: at least 12 months of unprotected intercourses which did not allow to achieve pregnancy coded ‘no’ if the patient does not desire to get pregnant.

1674

sample were in perfect agreement with those from the training sample and had very good diagnostic performance. However, our study may have several limitations that arise from the evaluation of the four variables selected to determine the score. Using a duration of pain of 24 months might seem paradoxical, as ideally we want to make an early diagnosis. One could argue that according to our score, a DIE patient considered at high risk because of duration of pain for .24 months would be classified as low risk a few months earlier with the same disease. However, this duration was based on the population sample using the ROC curve to maximize sensitivity and specificity to detect DIE. According to many previous studies, there is a significant delay in the diagnosis of endometriosis that averages 7–12 years despite many complaints of patients seeking medical assistance throughout the world (Ballard et al., 2006). This is why a period of 24 months, even though it may appear long, will allow a much earlier diagnosis than before. For patients with no desire to get pregnant, we could not determine whether they were infertile. Therefore, we considered them as fertile for the analysis to potentially minimize the difference between DIE+ and DIE2 patients for this parameter. As a consequence, this minimized the coefficient and the aOR of the infertility variable. The coefficient of the infertility variable was 9, which could have been an underestimate. Patients that were using treatment before surgery might have little or no pain, particularly if they have no menstruation using hormonal treatments and are coded as no pain for dysmenorrhoea and for all cyclic pains. However, even if they were underestimated, patients undergoing treatment with or without amenorrhoea still had significantly higher pain scores than those without treatment, both in the DIE+ and in DIE2 patients. This reflects the severity of the disease more than the result of their treatment (Supplementary data, Tables SII and SIII).

Our study could also be subject to bias. During questioning at the first preoperative visit, the surgeon may have had a high suspicion of the diagnosis because of the typical symptoms described by the patient or be aware of the diagnosis in cases of prior surgery. This could influence the questioning of the patient and create an ascertainment bias. The difference between DIE+ and DIE2 could therefore be artificially increased. For this reason, we were careful to check that a previous surgery did not change the OR of DIE for pain symptoms. In addition, patients were recruited from a specialized centre where they could be specifically referred for surgical treatment of DIE. Patients in this series were referred initially for an ovarian cyst or an endometrioma. The selection of a population with endometrioma minimized the selection bias for indication of surgery because almost all endometriomas are treated with surgery. However, in this series, there is a high prevalence of associated DIE, especially GI tract lesions, which could influence the OR of predicted variables by increasing the difference between cases and controls and increasing the coefficients of the score. Finally, misclassifications could have occurred. The description of the localization of DIE lesions in Table I is based on histological findings after the surgical removal of the lesions to provide the most reliable diagnostic test. The gold standard for diagnosis was then based on two steps, which included a macroscopic surgical diagnosis by palpation and visualization and a histological confirmation of the diagnosis on the surgical lesions removed to minimize any bias. However, in spite of the experience of the surgeon, some lesions could have potentially gone undetected. This would result in a misclassification of the DIE lesions and/or a misclassification of the DIE+ patients as DIE2, which would have minimized the difference between the cases and controls. In summary, all of these biases could modify the score by increasing or decreasing the OR of the DIE, modifying the choice of variables, and modifying the coefficients of the variables. We evaluated these potential consequences and tried to minimize them as much as possible: we used a jackknife procedure and a validation sample, and we evaluated the influence of a previous surgery and of treatments with no significant interaction on the relationship between DIE and symptoms. We reduced the sampling bias of the surgical indication by specifically using a population of patients with endometrioma. The high prevalence of DIE, especially localized in the GI, could have influenced the choice of variables, but the application of this score on this population should predict at least the patients with the same characteristics with the same sensibility and specificity. Despite our efforts to analyse and reduce biases, an external validation on a population with a different prevalence of DIE and a different distribution of lesions will be necessary before generalization of the score. This is the first clinical model proposed to determine the likelihood of having DIE endometriosis from the ultrasound diagnosis of an ovarian cyst. The model defined with our four variables is simple to use in clinical practice and has good diagnostic performance. We postulate that DIE+ patients without endometrioma could be detected by the same score, as long as the presence of an endometrioma does not affect the clinical symptoms of DIE (Chapron et al., 2012). This hypothesis should be tested in further studies.

Conclusion We built a predictive score of associated DIE in patients that underwent surgery for an endometriosis cyst with a good diagnostic predictive

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

Figure 2 Receiver operating characteristic curves of the score on the training (red line) and validation (blue line) samples. The two curves of the sensitivity and (1-specificity) of every value of the score on the training and validation sample to predict DIE are shown on this figure. The area under the curve of the score derived on the training sample (0.84, 95% confidence interval (CI): 0.79– 0.90) shows a good performance of the score for the diagnostic of DIE. The area under the curve of the score applied on the validation sample (0.84, 95% CI: 0.76 – 0.91) was not significantly different from the area under the curve of the training sample (P ¼ 0, 99) and validated the results of the score.

Lafay Pillet et al.

A score to predict DIE associated with endometrioma

performance. External validation of these results is necessary in less specialized departments for the management of DIE.

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

Acknowledgements The authors thank researchers of unite 953 INSERM (Pr Franc¸ois Goffinet) for their expert advice, surgeons from the department for their expert assistance with data collection. The authors also acknowledge Nathalie Girma for continuously managing the patients’ database.

C.C., A.F. and M.C.L.P. conceived and designed the study. M.C.L.P., A.F. and C.H. analysed and interpreted the data. C.C., A.F., C.H. and M.C.L.P. wrote the manuscript. C.C., B.B., M.C.L.P. and P.S. contributed to data collection and/or performed surgical procedures. All the authors contributed to write the manuscript. All the authors approved the final version of the manuscript.

Funding No grant supported the study.

Conflict of interest None declared.

References Abbott JA, Hawe J, Clayton RD, Garry R. The effects and effectiveness of laparoscopic excision of endometriosis: a prospective study with 2 – 5 year follow-up. Hum Reprod 2003;18:1922 – 1927. Abrao MS, Goncalves MO, Dias JA Jr, Podgaec S, Chamie LP, Blasbalg R. Comparison between clinical examination, transvaginal sonography and magnetic resonance imaging for the diagnosis of deep endometriosis. Hum Reprod 2007;22:3092 – 3097. Ballard K, Lowton K, Wright J. What’s the delay? A qualitative study of women’s experiences of reaching a diagnosis of endometriosis. Fertil Steril 2006;86:1296 – 1301. Buckley RG, King KJ, Disney JD, Ambroz PK, Gorman JD, Klausen JH. Derivation of a clinical prediction model for the emergency department diagnosis of ectopic pregnancy. Acad Emerg Med 1998;5:951– 960. Chapron C, Fauconnier A, Dubuisson JB, Barakat H, Vieira M, Breart G. Deep infiltrating endometriosis: relation between severity of dysmenorrhea and extent of disease. Hum Reprod 2003;18:760 –766. Chapron C, Chopin N, Borghese B, Foulot H, Dousset B, VacherLavenu MC, Vieira M, Hasan W, Bricou A. Deeply infiltrating endometriosis: pathogenetic implications of the anatomical distribution. Hum Reprod 2006;21:1839 – 1845. Chapron C, Pietin-Vialle C, Borghese B, Davy C, Foulot H, Chopin N. Associated ovarian endometrioma is a marker for greater severity of deeply infiltrating endometriosis. Fertil Steril 2009;92:453– 457. Chapron C, Lafay-Pillet MC, Monceau E, Borghese B, Ngo C, Souza C, de Ziegler D. Questioning patients about their adolescent history can

identify markers associated with deep infiltrating endometriosis. Fertil Steril 2011;95:877 – 881. Chapron C, Santulli P, de Ziegler D, Noel JC, Anaf V, Streuli I, Foulot H, Souza C, Borghese B. Ovarian endometrioma: severe pelvic pain is associated with deeply infiltrating endometriosis. Hum Reprod 2012; 27:702– 711. Chopin N, Ballester M, Borghese B, Fauconnier A, Foulot H, Malartic C, Chapron C. Relation between severity of dysmenorrhea and endometrioma. Acta Obstet Gynecol Scand 2006;85:1375 – 1380. Coste J, Bouyer J, Job-Spira N. Construction of composite scales for risk assessment in epidemiology: an application to ectopic pregnancy. Am J Epidemiol 1997;145:278 – 289. Dawood MY. Primary dysmenorrhea: advances in pathogenesis and management. Obstet Gynecol 2006;108:428 – 441. de Ziegler D, Borghese B, Chapron C. Endometriosis and infertility: pathophysiology and management. Lancet 2010;376:730– 738. Dousset B, Leconte M, Borghese B, Millischer AE, Roseau G, Arkwright S, Chapron C. Complete surgery for low rectal endometriosis: longterm results of a 100-case prospective study. Ann Surg 2010; 251:887 – 895. Efron B, Gong G. A leisurely look at the bootstrap, the jackknife and cross validation. Am Stat 1983;37:36 – 48. Fauconnier A, Chapron C. Endometriosis and pelvic pain: epidemiological evidence of the relationship and implications. Hum Reprod Update 2005; 11:595– 606. Fauconnier A, Chapron C, Dubuisson JB, Vieira M, Dousset B, Breart G. Relation between pain symptoms and the anatomic location of deep infiltrating endometriosis. Fertil Steril 2002;78:719 – 726. Fedele L, Bianchi S, Zanconato G, Berlanda N, Borruto F, Frontino G. Tailoring radicality in demolitive surgery for deeply infiltrating endometriosis. Am J Obstet Gynecol 2005;193:114– 117. Fedele L, Bianchi S, Carmignani L, Berlanda N, Fontana E, Frontino G. Evaluation of a new questionnaire for the presurgical diagnosis of bladder endometriosis. Hum Reprod 2007;22:2698 – 2701. Forman RG, Robinson JN, Mehta Z, Barlow DH. Patient history as a simple predictor of pelvic pathology in subfertile women. Hum Reprod 1993; 8:53 – 55. Garry R, Clayton R, Hawe J. The effect of endometriosis and its radical laparoscopic excision on quality of life indicators. BJOG 2000;107:44–54. Guerriero S, Ajossa S, Gerada M, D’Aquila M, Piras B, Melis GB. ‘Tenderness-guided’ transvaginal ultrasonography: a new method for the detection of deep endometriosis in patients with chronic pelvic pain. Fertil Steril 2007;88:1293– 1297. Hadfield R, Mardon H, Barlow D, Kennedy S. Delay in the diagnosis of endometriosis: a survey of women from the USA and the UK. Hum Reprod 1996;11:878 – 880. Huchon C, Staraci S, Fauconnier A. Adnexal torsion: a predictive score for pre-operative diagnosis. Hum Reprod 2010;25:2276 – 2280. Husby GK, Haugen RS, Moen MH. Diagnostic delay in women with pain and endometriosis. Acta Obstet Gynecol Scand 2003;82:649 – 653. Kamergorodsky G, Ribeiro PA, Galvao MA, Abrao MS, Donadio N, Lemos NL, Aoki T. Histologic classification of specimens from women affected by superficial endometriosis, deeply infiltrating endometriosis, and ovarian endometriomas. Fertil Steril 2009;92:2074– 2077. Kinkel K, Chapron C, Balleyguier C, Fritel X, Dubuisson JB, Moreau JF. Magnetic resonance imaging characteristics of deep endometriosis. Hum Reprod 1999;14:1080 – 1086. Koninckx PR, Meuleman C, Demeyere S, Lesaffre E, Cornillie FJ. Suggestive evidence that pelvic endometriosis is a progressive disease, whereas deeply infiltrating endometriosis is associated with pelvic pain. Fertil Steril 1991;55:759 – 765.

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

Authors’ roles

1675

1676

Sampson JA. Metastatic or Embolic Endometriosis, due to the Menstrual Dissemination of Endometrial Tissue into the Venous Circulation. Am J Pathol 1927;3:93 –110.43. Sinaii N, Plumb K, Cotton L, Lambert A, Kennedy S, Zondervan K, Stratton P. Differences in characteristics among 1,000 women with endometriosis based on extent of disease. Fertil Steril 2008;89:538 – 545. Somigliana E, Infantino M, Candiani M, Vignali M, Chiodini A, Busacca M. Association rate between deep peritoneal endometriosis and other forms of the disease: pathogenetic implications. Hum Reprod 2004;19:168–171. Vercellini P, Trespidi L, De Giorgi O, Cortesi I, Parazzini F, Crosignani PG. Endometriosis and pelvic pain: relation to disease stage and localization. Fertil Steril 1996;65:299– 304. Vercellini P, Fedele L, Aimi G, Pietropaolo G, Consonni D, Crosignani PG. Association between endometriosis stage, lesion type, patient characteristics and severity of pelvic pain symptoms: a multivariate analysis of over 1000 patients. Hum Reprod 2007;22:266 – 271. Warner PE, Critchley HO, Lumsden MA, Campbell-Brown M, Douglas A, Murray GD. Menorrhagia I: measured blood loss, clinical features, and outcome in women with heavy periods: a survey with follow-up data. Am J Obstet Gynecol 2004;190:1216 – 1223.

Downloaded from http://humrep.oxfordjournals.org/ at Serials Department on October 26, 2014

Lemeshow S, Hosmer DW Jr. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemiol 1982; 115:92–106. Marcoux S, Maheux R, Berube S. Laparoscopic surgery in infertile women with minimal or mild endometriosis. Canadian Collaborative Group on Endometriosis. N Engl J Med 1997;337:217– 222. Matsuzaki S, Canis M, Pouly JL, Rabischong B, Botchorishvili R, Mage G. Relationship between delay of surgical diagnosis and severity of disease in patients with symptomatic deep infiltrating endometriosis. Fertil Steril 2006;86:1314 – 1316; discussion 1317. Peveler R, Edwards J, Daddow J, Thomas E. Psychosocial factors and chronic pelvic pain: a comparison of women with endometriosis and with unexplained pain. J Psychosom Res 1996;40:305 – 315. Piketty M, Chopin N, Dousset B, Millischer-Bellaische AE, Roseau G, Leconte M, Borghese B, Chapron C. Preoperative work-up for patients with deeply infiltrating endometriosis: transvaginal ultrasonography must definitely be the first-line imaging examination. Hum Reprod 2009;24:602–607. Porpora MG, Koninckx PR, Piazze J, Natili M, Colagrande S, Cosmi EV. Correlation between endometriosis and pelvic pain. J Am Assoc Gynecol Laparosc 1999;6:429– 434.

Lafay Pillet et al.