The Value of Pelvic MRI in the Diagnosis of Posterior Cul-De-Sac Obliteration in Cases of Deep Pelvic Endometriosis

Transcription

1 Women s Imaging Original Research Macario et al. MRI of Deep Pelvic Endometriosis Women s Imaging Original Research Sabrina Macario 1 Madleen Chassang 1 Sebastien Novellas 1 Guillaume Baudin 1 Jerome Delotte 2 Olivier Toullalan 3 Patrick Chevallier 1 Macario S, Chassang M, Novellas S, et al. Keywords: deep pelvic endometriosis, Douglas pouch, obliteration, pelvic MRI DOI: /AJR Received September 13, 2011; accepted after revision May 17, Service d Imagerie Diagnostique et Interventionnelle, Centre Hospitalier Régional et Universitaire de Nice, Hôpital Archet 2, 151 Route de Saint Antoine de Ginestière, BP 3079, Nice Cedex 3, France. Address correspondence to S. Macario (Macario.s@chu-nice.fr). 2 Service de Gynécologie-Obstétrique, Centre Hospitalier Régional et Universitaire de Nice, Hôpital Archet, Nice, France. 3 Service de Gynécologie-Obstétrique, Hôpital de Cannes, Cannes, France. AJR 2012; 199: X/12/ American Roentgen Ray Society The Value of Pelvic MRI in the Diagnosis of Posterior Cul-De-Sac Obliteration in Cases of Deep Pelvic Endometriosis OBJECTIVE. The objective of our study was to define relevant MRI signs allowing preoperative diagnosis of posterior cul-de-sac obliteration in patients with deep pelvic endometriosis. MATERIALS AND METHODS. This retrospective study included patients who underwent pelvic MRI completed by a laparoscopic examination. Three radiologists performed the MRI review blinded and recorded the following signs: sign 1, retroflexed uterus; sign 2, retrouterine mass; sign 3, displacement of intraperitoneal fluid; sign 4, elevation of the fornix; and sign 5, adherence of bowel loops. Laparoscopic results provided the criterion standard for diagnosis of posterior cul-de-sac obliteration. The performance of MRI was evaluated by calculating the average sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of MRI results of the two more experienced radiologists for each sign and for combinations of signs. Interobserver agreement for each sign and impression for posterior cul-de-sac obliteration were calculated for all radiologists. RESULTS. Sixty-three patients were included in the study. Posterior cul-de-sac obliteration was diagnosed in 43 patients at laparoscopy. The mean sensitivity, specificity, and accuracy of each sign and impression of posterior cul-de-sac obliteration were, respectively, as follows: sign 1, 24.4%, 77.5%, 41.3%; sign 2, 97.1%, 83.7%, 92.8%; sign 3, 95.0%, 88.7%, 93.1%; sign 4, 30.2%, 97.5%, 51.6%; sign 5, 83.7%, 91.2%, 86.1%; and impression of posterior cul-de-sac obliteration, 91.9%, 91.2%, 91.7%. Interobserver concordance varied from 0.26 to 0.81 with best results obtained with the combination of signs 2, 3, and 5. Best concordances for junior radiologist evaluations were obtained with assessment of sign 3. CONCLUSION. MRI allows posterior cul-de-sac obliteration diagnosis. Pelvic fluid displacement may be the sign with greatest utility when considering both diagnostic accuracy and interobserver agreement. D eep pelvic endometriosis is defined histologically as subperitoneal lesions infiltrating at least 5 mm into the walls of the pelvic organs [1]. It may result in various symptoms including dysmenorrhea, dyspareunia, chronic pelvic pain, and infertility that diminish the quality of life of the patient. Treatment with curative intent is complete surgical excision of the symptomatic lesions [2 7]. MRI has been shown to play a major role in the detection and evaluation of lesions of deep pelvic endometriosis [8 10]. The posterior pelvic compartment is the most commonly affected region and may result in posterior cul-de-sac obliteration via adhesions between the rectum and the uterus [11]. Diagnostic laparoscopy in these cases of obliteration and adhesions without presurgical radiologic evaluation may result in an in- creased procedural complication rate including rectal or ureteral lacerations as well as rectovaginal fistulas [12]. Preprocedural detection of posterior cul-de-sac obliteration may be helpful in determining surgical options and minimizing complications [2, 10]. Few studies have tried to delineate the role of MRI in the diagnosis of posterior cul-de-sac obliteration [13, 14]. The purpose of our study was to define relevant MRI signs that allow preoperative diagnosis of posterior cul-de-sac obliteration in patients with deep pelvic endometriosis while using laparoscopy as a reference. Materials and Methods Population This single-center retrospective study adhered to the Declaration of Helsinki and our institution does not require approval through the ethics committee 1410 AJR:199, December 2012

2 MRI of Deep Pelvic Endometriosis for this type of study. During the study period between January 2007 and December 2009, 150 patients were evaluated for deep pelvic endometriosis via pelvic MRI. Inclusion criteria were adult patients with a diagnosis of deep pelvic endometriosis who underwent pelvic MRI and subsequently underwent laparoscopy less than 2 months after MRI evaluation. Exclusion criteria were any previous abdominal or pelvic surgery, the absence of exploratory surgery, a delay of more than 2 months between pelvic MRI and intervention, or a current gynecologic infection. Of the 150 patients initially selected, 63 patients fit the criteria for inclusion. Of the excluded patients, 60 patients had not undergone surgical exploration, 15 patients had prior abdominal or pelvic surgery, 10 patients had a delay of more than 2 months between MRI and surgical exploration, and two patients suffered from a gynecologic infection resulting in the formation of a pelvic abscess or pelvic peritonitis. The average age of the study group was 34.4 years (minimum age, 23 years; maximum age, 54 years). Among the patients fitting the inclusion criteria, 43 patients, which represents 68.25% of the population, presented a posterior cul-de-sac obliteration at laparoscopic exploration. Fig year-old woman who presented with dysmenorrhea. Posterior cul-de-sac obliteration was detected at laparoscopy. T2-weighted sagittal image shows uterine retroflexion. Sign 1 is present with α angle of > 180 (lines). MRI Technique and Protocol Pelvic MRI was preferentially performed at the beginning of the patient s menstrual cycle to best detect any small foci of endometriosis. The protocol traditionally used at our institution included the patient s fasting at least 3 hours before the study along with completing a colonic preparation. Peripheral venous access was obtained for injection of an antispasmodic (1 mg/ml GlucaGen [glucagon, rdna origin], Novo Nordisk) before patient placement on the examination table and subsequent administration of 0.1 mmol/kg gadolinium (Dotarem [gadoterate dimeglumine], Guerbet). A physician or technician administered contrast media rectally and vaginally to improve MRI sensitivity in the detection of deep endometrial implants of the vagina and rectovaginal wall [15, 16]. Vaginal opacification used on average 60 ml of sterile ultrasound gel performed through a Foley catheter, and rectal opacification followed the same protocol with an average of 100 ml of ultrasound gel used. An abdominal support belt was placed around the patient as well as presaturation bands on MRI by which good positioning was verified. The images were obtained using a 1.5-T MRI unit (Achieva, Philips Healthcare) with a phasedarray antenna. T2-weighted turbo spin-echo sequences were performed in the sagittal and axial planes. T1-weighted images were obtained initially without fat suppression in the sagittal plane and then in the axial plane after fat suppression before and after the injection of gadolinium. The sequence parameters were as follows for T1-weighted imaging: TR/TE, 600/7; slice thickness, 3 mm; interslice space, 1 mm; FOV, 320 mm; and matrix, pixels. The parameters for T2-weighted imaging were TR/TE, 3500/90; slice thickness, 3 mm; interslice space, 1 mm for the axial and sagittal images with an acquisition time of 3 minutes each; FOV, 230 mm; and matrix, pixels. Analysis of the MRI Data The examinations were blinded and data were transferred on a digital console. Each examination was interpreted by three independent radiologists: readers 1, 2, and 3 with 6 years, 2 years, and 6 months of experience, respectively, in the interpretation of pelvic MR images. Because the studies were blinded, the radiologists had no prior knowledge of the laparoscopic results. Each radiologist performed two readings from the MRI database at a minimum interval of 15 days without the possibility of reviewing prior interpretations. After a review of the literature and from our prior experience, we chose the five signs for evaluation. For each sign, a binary analysis was performed. Sign 1: uterine retroflexion Uterine flexion is the angle formed between the intersection of the axis of the uterine cervix and the axis of the uterine body. Retroflexion of the uterus corresponds to backward displacement of this angle such that the body of the uterus is toward the rectum rather than the bladder. The vertex of the angle is situated at the level of the uterine isthmus. Sign 1 evaluated the respective presence or absence of retroflexion of the uterus if the angle is higher or lower than 180 (Fig. 1). Sign 2: retrouterine fibrous mass Sign 2 was defined as a mass in T2 hyposignal and T1 iso- or hyposignal along with possible small foci appearing in T1 hypersignal or T2 hypersignal. The mass extends from the uterus to the rectosigmoid junction (Fig. 2). Sign 3: displacement of intraperitoneal fluid Sign 3 took into account the position of any fluid within the pelvic cavity. Posterior cul-de-sac obliteration prevents fluid accumulation in this dependent area where free peritoneal fluid usually collects. The sign was positive if the fluid was located uniquely outside the posterior cul-de-sac and negative if it was visualized within the posterior culde-sac (Fig. 3). A particular score was assigned in the absence of any fluid, which decreased the number of patients evaluated for this sign only. Sign 4: elevation of the fornix Sign 4 evaluated for the elevation of the posterior vaginal fornix (Fig. 4). We define the location of this anatomic site as the superior end of the posterior vaginal cul-de-sac, specifically with regard to its location relative to the uterine isthmus. The sign was considered positive when the fornix was found to be superior to the uterine isthmus. Sign 5: adherence of bowel loops Sign 5 was considered positive if there were small-bowel loops adherent to the posterior surface of the uterus or angulation of the anterior surface of the rectosigmoid colon adherent to the posterior surface of the uterus (Fig. 5). At the conclusion of the analysis for these five signs, each reader decided whether the posterior cul-de-sac was obliterated. Laparoscopy In our study, all patients had undergone diagnostic or therapeutic laparoscopy performed by one of two experienced gynecologists. Each procedure was carried out independently by one of the two Fig year-old woman who presented with infertility. Posterior cul-de-sac obliteration was diagnosed at laparoscopy. T2-weighted sagittal image shows (sign 2). There is retrouterine fibrous mass present: endometrial lesion of torus (arrow) with involvement of anterior wall of rectum. AJR:199, December

3 Macario et al. Fig year-old woman who presented with chronic pelvic pain. Posterior cul-de-sac obliteration was noted at laparoscopy. Sign 3 displacement of intraperitoneal fluid can be seen on T2-weighted sagittal image: displaced fluid in vesicouterine pouch (white arrow) with posterior cul-de-sac free of fluid with fatty signal (black arrow). gynecologists with a maximum delay of 2 months after pelvic MRI. Diagnostic abdominal laparoscopy and subsequent pelvic laparoscopy were carried out rigorously and systematically. Exploration of the abdominal cavity included evaluation of the parietocolic gutters, domes of the diaphragm, liver, omentum, digestive tract, and appendix, and pelvic cavity exploration included the vesicouterine pouch, uterosacral ligaments, uterus, ovarian tubes, ovaries, and ureters. A report with specifications in accordance with the American Society for Reproductive Medicine (ASRM) was generated to identify any possible case of obliteration of the posterior cul-de-sac [17]. According to the ASRM, the posterior cul-de-sac may be classified as nonobliterated, partially obliterated, or totally obliterated. For our purposes, cases of partial obliteration were considered nonobliterated. Statistical Analysis MRI performance was evaluated by calculating sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy, all expressed as percentages, for each sign as well as the impression for posterior culde-sac obliteration. These values were calculated for each radiologist from their two readings. We chose to calculate and present the average results of the data of the two more experienced readers regarding the accuracy of signs (individually and in combination) to determine the best performance by MRI. Data from the most inexperienced reader were used solely to evaluate inter- and intraobserver variability. The data of each of the two other readers results were expressed as a percentage rounded to the nearest decimal with a CI of 95%. Fig year-old woman who presented with infertility. No posterior cul-de-sac obliteration was detected at laparoscopy. T2-weighted sagittal image shows sign 4, elevation of fornix (arrow). Interobserver variability and intraobserver variability were quantified for each sign and for the impression for the obliteration of the posterior cul-de-sac by means of the coefficient kappa. The concordance value was defined according to Landis and Koch [18]. Kappa values were interpreted as follows: less than 0.00 represented poor agreement; , slight; , fair; , moderate; , substantial; and , almost perfect agreement. Differences were considered statistically significant at a threshold of α < Data were analyzed using statistical software (Stata, version 9, StataCorp). Results MRI Performance in the Evaluation of Posterior Cul-De-Sac Obliteration Performance by sign The results presented were calculated starting with the average of the two more experienced readers data. Best performances were seen with signs 2 (retrouterine fibrous mass), 3 (displacement of intraperitoneal fluid), and 5 (adherence of bowel loops) with sensitivity, specificity, and accuracy of 97.1%, 83.7%, 92.8%, respectively, for sign 2; 95.0%, 88.7%, 93.1% for sign 3; and 83.7%, 91.2%, 86.1% for sign 5 (Table 1). Performance based on sign association The best results were obtained with the following sign associations calculated starting with the average of the two more experienced readers data: signs 2 and 3 (retrouterine fibrous mass and displacement of intraperitoneal fluid); signs 2 and 5 (retrouterine fibrous mass and adherence of bowel loops); signs 3 and 5 (displacement of intraperitoneal fluid Fig year-old woman who presented with secondary infertility. Posterior cul-de-sac obliteration was diagnosed at laparoscopy. T2- weighted axial image reveals sign 5, attraction of bowel loops. Attraction of anterior wall of rectum (arrow) to posterior wall of uterus is shown. and adherence of bowel loops); and signs 2, 3, and 5 (retrouterine fibrous mass, displacement of intraperitoneal fluid, and adherence of bowel loops). The sensitivity, specificity, and accuracy were, respectively, 96.8%, 88.0%, and 94.0% for signs 2 and 3; 95.4%, 72.3%, and 86.1% for signs 2 and 5; 97.7%, 69.4%, and 85.8% for signs 3 and 5; and 97.8%, 69.4%, and 85.8% for signs 2, 3, and 5 (Table 2). Concerning the global impression for posterior cul-de-sac obliteration, which was also calculated with the average of the two more experienced readers data, the sensitivity, specificity, PPV, NPV, and accuracy were as follows: 91.9%, 91.2%, 95.8%, 72.3%, and 91.7%, respectively. Inter- and Intraobserver Concordance by Sign and Impression for Posterior Cul-De-Sac Obliteration Interobserver concordance The best concordance was seen for signs 2 (retrouterine fibrous mass), 3 (displacement of intraperitoneal fluid), and 5 (adherence of bowel loops) for the two more experienced radiologists (readers 1 and 2). Agreement was almost perfect for sign 2 (κ = 0.81) and agreement was substantial for signs 3 and 5 (κ = 0.75 and 0.76, respectively). Uniform and almost equivalent substantial concordance by all readers including the junior radiologist with regard to sign 3 (displacement of intraperitoneal fluid) was found (Table 3). Intraobserver concordance The best intraobserver concordance was found for readers 1 and 2 and for signs 1 (uterine retroflexion), 1412 AJR:199, December 2012

4 MRI of Deep Pelvic Endometriosis TABLE 1: Performances by MRI Sign and for the Global Impression for Posterior Cul-De-Sac Obliteration MRI Sign (Description of Sign) Sensitivity (%) Specificity (%) Average a (95% CI) Positive Predictive Value (%) Negative Predictive Value (%) Accuracy (%) Sign 1 (uterine retroflexion) 24.4 ( ) 77.5 ( ) 70.0 ( ) 32.3 ( ) 41.3 ( ) Sign 2 (retrouterine mass) 97.1 ( ) 83.7 ( ) 92.8 ( ) 93.1 ( ) 92.8 ( ) Sign 3 (displacement of fluid) 95.0 ( ) 88.7 ( ) 95.0 ( ) 88.7 ( ) 93.1 ( ) Sign 4 (elevation of fornix) 30.2 ( ) 97.5 ( ) 96.3 ( ) 39.4 ( ) 51.6 ( ) Sign 5 (attraction of bowel loops) 83.7 ( ) 91.2 ( ) 95.4 ( ) 72.3 ( ) 86.1 ( ) Impression for posterior cul-de-sac obliteration a Combined results of the two more experienced radiologists ( ) 91.2 ( ) 95.8 ( ) 72.3 ( ) 91.7 ( ) TABLE 2: MRI Performance in the Diagnosis of Posterior Cul-De-Sac Obliteration by Association of Radiologic Signs MRI Signs Sensitivity (%) Specificity (%) Average a (95% CI) Positive Predictive Value (%) Negative Predictive Value (%) Accuracy (%) 1 and ( ) 37.5 ( ) 24.4 ( ) 97.5 ( ) 47.6 ( ) 1 and ( ) 37.2 ( ) 25.5 ( ) ( ) 48.3 ( ) 1 and ( ) 35.1 ( ) 13.9 ( ) ( ) 41.3 ( ) 1 and ( ) 35.9 ( ) 19.2 ( ) 97.5 ( ) 44.0 ( ) 2 and ( ) 88.0 ( ) 94.4 ( ) 93.0 ( ) 94.0 ( ) 2 and ( ) 39.7 ( ) 30.2 ( ) 98.7 ( ) 52.0 ( ) 2 and ( ) 72.3 ( ) 83.7 ( ) 91.2 ( ) 86.1 ( ) 3 and ( ) 38.9 ( ) 31.7 ( ) 98.6 ( ) 52.1 ( ) 3 and ( ) 69.4 ( ) 81.4 ( ) 95.8 ( ) 85.8 ( ) 4 and ( ) 38.7 ( ) 27.3 ( ) 98.7 ( ) 50.0 ( ) 1, 2, and ( ) 37.2 ( ) 25.5 ( ) ( ) 48.3 ( ) 1, 2, and ( ) 35.1 ( ) 13.9 ( ) ( ) 41.3 ( ) 1, 2, and ( ) 35.9 ( ) 19.2 ( ) 97.5 ( ) 44.0 ( ) 1, 3, and ( ) 34.1 ( ) 14.9 ( ) ( ) 40.9 ( ) 1, 3, and ( ) 35.5 ( ) 19.9 ( ) ( ) 44.4 ( ) 1, 4, and ( ) 34.5 ( ) 11.6 ( ) ( ) 39.7 ( ) 2, 3, and ( ) 38.9 ( ) 31.7 ( ) 98.6 ( ) 52.1 ( ) 2, 3, and ( ) 69.4 ( ) 81.4 ( ) 95.8 ( ) 85.8 ( ) 2, 4, and ( ) 38.7 ( ) 27.3 ( ) 98.7 ( ) 50.0 ( ) 3, 4, and ( ) 37.8 ( ) 28.6 ( ) 98.6 ( ) 50.0 ( ) 1, 2, 3, and ( ) 34.1 ( ) 14.9 ( ) ( ) 40.9 ( ) 1, 2, 3, and ( ) 35.5 ( ) 19.9 ( ) ( ) 44.4 ( ) 1, 2, 4, and ( ) 34.5 ( ) 11.6 ( ) ( ) 39.7 ( ) 1, 3, 4, and ( ) 33.5 ( ) 12.4 ( ) ( ) 39.2 ( ) 2, 3, 4, and ( ) 37.8 ( ) 28.6 ( ) 98.6 ( ) 50.0 ( ) 1, 2, 3, 4, and ( ) 33.5 ( ) 12.4 ( ) ( ) 13.0 ( ) Note Boldface shows combinations of signs with best performance. a Combined results of the two more experienced radiologists. AJR:199, December

5 Macario et al. TABLE 3: Interobserver Concordance by Sign and for the Global Impression for Posterior Cul-De-Sac Obliteration κ MRI Sign (Description of Sign) Readers 1 and 2 Readers 1 and 3 Readers 2 and 3 Sign 1 (uterine retroflexion) Sign 2 (retrouterine mass) Sign 3 (displacement of fluid) Sign 4 (elevation of fornix) Sign 5 (attraction of bowel loops) Impression for posterior cul-de-sac obliteration (retrouterine fibrous mass), 3 (displacement of intraperitoneal fluid), and 5 (adherence of bowel loops). Agreement was almost perfect, with kappa values of between 0.85 and 1.00 (Table 4). Discussion Our study revealed that MRI has excellent ability for detecting posterior cul-de-sac obliteration, with an accuracy of 92%,that is superior to those given in the literature [13, 14]. In a series comparable to our study in terms of population and methodology, Kataoka et al. [13] reported that the precision of MRI in this setting was only 72% [13]. This discordance may be explained by the absence of vaginal opacification in the study of Kataoka and colleagues, which would limit the analysis of certain signs such as elevation of the fornix [13]. Along this line of reasoning, one recent study showed that opacification allowed better detection of subperitoneal implants [15]. Another reason for the discordance between our results and those of Kataoka et al. may be that Kataoka et al. did not evaluate all the possible signs notably, the presence of intraperitoneal fluid. Kikuchi et al. [14] evaluated different signs with less rigorous methods than we used. In effect, one sole senior radiologist evaluated all the radiographic data. Our methodology allowed us to substantiate our performance, as we now describe in detail, sign by sign. The presence of a retroflexed uterus has been advanced in the literature as a predictor for posterior cul-de-sac obliteration [13, 14]. However, this radiographic sign appears to have only modest prognostic significance because most cases of obliteration are caused by adherence of freely mobile digestive tract to the denser uterus that is fixed within the pelvis. Only the most bulky retracted lesions, which are rare, would lead to a retroflexed uterus. Moreover, the retroflexed uterus is a normal physiologic variant in roughly 20% of patients. The presence of a retrouterine mass, sign 2 in our study, is defined as a posterior lesion connecting the uterus to the rectum and constitutes an excellent prognostic sign. However, relatively small fibrous bands are difficult to discern for a less experienced radiologist, as we noted in the statistical analysis of the intraobserver concordance value for reader 3 and the weaker interobserver concordance values for readers 1 and 2. Kataoka et al. [13] reported that the sensitivity of this sign was only 50%. We believe that the source of this suboptimal result was an imprecise definition of a retrouterine lesion that included proximal involvement of the uterosacral ligaments not affecting the rectum and not sufficiently obliterating the posterior cul-de-sac. Analysis of the position of intraperitoneal fluid is the best sign indicating posterior culde-sac obliteration. This sign had excellent accuracy, and inter- and intraobserver concordance in our study were independent of the level of experience of the radiologist. Fluid was almost universally present in the pelvis: Only two patients with posterior cul-desac obliteration presented without pelvic fluid. Because of its bright signal on T2-weighted imaging, even small amounts of fluid are easily visualized by an inexperienced radiologist. The displacement of pelvic fluid is most often found in an anterior position, around the bladder, or in contact with the uterine body. Kataoka et al. [13] did not evaluate this sign, whereas Kikuchi et al. [14] reported that the specificity of this sign was weak. In their work [14], patients who did not have fluid in the posterior cul-de-sac were considered to have an obliterated posterior cul-de-sac, resulting in an elevated sensitivity and correspondingly lower specificity with regard to this sign. Elevation of the fornix in our study had low diagnostic value. We were unable to find in the literature a precise anatomic definition for the position of the normal fornix on MRI. Moreover, it has not been shown that vaginal opacification with ultrasound gel smoothes out the vaginal fornix in a systematic and uniform manner. A study showed that the vaginal culde-sac does not distend in a uniform manner in a significant percentage of patients without an otherwise apparent causative disease or abnormality [19]. In addition, the position and depth of the fornix are dependent on the degree of uterine flexion. This sign thus had the worst interobserver concordance because of the difficulties in defining the position of the fornix. It is also a poor prognosticator in cases in which a posterior mass invades and pushes back the vaginal cul-de-sac that, thus, conversely appears lower or contracted. For Kataoka et al. [13], the accuracy of this sign is equally weak, approaching 50%. Kikuchi et al. [14] achieved excellent sensitivity with this sign, which is unexpected given the imprecise definition as discussed in this study. They also discussed the shape of the posterior fornix, specifying a beaklike characteristic as highly suggestive of posterior cul-de-sac obliteration [14]. However, posterior cul-de-sac obliteration regularly occurs in cases of a mass between the superior part of the rectum and the uterus. In these cases, the fornix is at a distance from the principal lesion of the deep pelvic endometriosis causing the obliteration of the posterior cul-de-sac. It thus cannot be affected because it is not connected to the endometrial mass. TABLE 4: Intraobserver Concordance by Sign and for the Global Impression for Posterior Cul-De-Sac Obliteration κ MRI Sign (Description of Sign) Reader 1 Reader 2 Reader 3 Sign 1 (uterine retroflexion) Sign 2 (retrouterine mass) Sign 3 (displacement of fluid) Sign 4 (elevation of fornix) Sign 5 (attraction of bowel loops) Impression for posterior cul-de-sac obliteration AJR:199, December 2012

6 MRI of Deep Pelvic Endometriosis The last sign studied was that of the adherence of bowel loops, which had an accuracy of 86%. However, the weak intraobserver concordance values observed with the junior radiologist indicate a certain level of difficulty in detecting this sign for the novice reader. The value of this sign is explained by the routine involvement of the digestive tract during posterior cul-de-sac obliteration. We should note, however, that this sign could involve various regions of the digestive tract. Classically, it is the junction between the superior and middle thirds of the rectum that is involved. Retrouterine lesions that are superior in location attract the rectosigmoid or the sigmoid colon alone. Last, a series of bowel loops adherent within the pelvis and fixed by a lesion of deep pelvic endometriosis could also be included within the definitions of this sign. A detailed analysis of these five signs reveals the potential role of MRI in detecting posterior cul-de-sac obliteration. Our work, however, presents certain limitations, as we now discuss. This study was retrospective and thus may be at increased risk for bias and requires less evidence than a prospective study. In our work, certain data relating to the symptoms of these patients could not be collected. It would be of interest to characterize the pain and fertility state of these patients presenting with an obliteration of the posterior cul-de-sac. Moreover, there exists a dissociation between the number of MRI examinations performed and the remaining population of 63 patients in our study. This divergence is partly explained by the fact that a number of patients had undergone pelvic MRI to evaluate for deep pelvic endometriosis as part of a workup for infertility. In these cases, priority would have been given to techniques to improve fertility before resorting to extensive surgical resection. Another limitation of our study is the chosen delay of 2 months maximum between pelvic MRI examination and laparoscopic exploration. Because the lesions of deep pelvic endometriosis evolve over time, a shorter interval between these two studies would have allowed greater correlation between the data. This time interval was less than 1 month in the study of Kataoka et al. [13], whereas the time interval was unspecified by Kikuchi et al. [14]. Another potential point of discussion involves the laparoscopic protocol. Diagnostic laparoscopy was performed by a single gynecologist for all the patients, thus obviating any analysis of interobserver concordance. Moreover, using diagnostic laparoscopy as a criterion standard should be undertaken with caution because a review of the literature shows that interoperator discordance with diagnostic laparoscopy in the evaluation of the lesions of deep pelvic endometriosis is significant [20, 21]. This may be partially explained by the experience of the operator as well as the classification scheme proposed by the ASRM, which for a large number of gynecologists appears incomplete with an imperfect correlation between the visual and histologic data for atypical lesions. Conclusion Preoperative radiographic mapping of the lesions of deep pelvic endometriosis and the assessment for posterior cul-de-sac obliteration are essential to the surgeon. MRI allows preoperative posterior cul-de-sac obliteration diagnosis with an accuracy approaching 90%. The presence of a retrouterine mass and the displacement of intraperitoneal fluid are the best performing signs. Analysis of the position of pelvic peritoneal fluid is the simplest and most useful sign in the evaluation of deep pelvic endometriosis. References 1. 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: Chapron C, Fauconnier A, Vieira M, et al. Anatomical distribution of deeply infiltrating endometriosis: surgical implications and proposition for a classification. Hum Reprod 2003; 18: Redwine DB, Wright JT. Laparoscopic treatment of complete obliteration of the cul-de-sac associated with endometriosis: long-term follow up of en bloc resection. Fertil Steril 2001; 76: Abbott J, Hawe J, Hunter D, Holmes M, Finn P, Garry R. Laparoscopic excision of endometriosis: a randomized, placebo-controlled trial. Fertil Steril 2004; 82: Panel P, Chis C, Gaudin S, et al. Laparoscopic surgery of deep endometriosis: about 118 cases [in French]. Gynecol Obstet Fertil 2006; 34: Dubernard G, Piketty M, Rouzier R, Houry S, Bazot M, Darai E. Quality of life after laparoscopic colorectal resection for endometriosis. Hum Reprod 2006; 21: Canis M, Matsuzaki S, Jardon K, et al. Yes, patients with deep infiltrating endometriosis should be operated on! Prefer optimistic will to pessimistic intelligence! [in French]. Gynecol Obstet Fertil 2008; 36: Kinkel K, Frei KA, Balleyguier C, Chapron C. Diagnosis of endometriosis with imaging: a review. Eur Radiol 2006; 16: Roy C, Balzan C, Thoma V, Sauer B, Wattiez A, Leroy J. Efficiency of MR imaging to orientate surgical treatment of posterior deep pelvic endometriosis. Abdom Imaging 2009; 34: Bazot M, Daraї E. Evaluation of pelvic endometriosis: the role of MRI [in French]. J Radiol 2008; 89: Bazot M, Daraї E, Hourani R, et al. Deep pelvic endometriosis: MR imaging for diagnosis and prediction of extension of disease. Radiology 2004; 232: Golfier F, Sabra M. Surgical management of endometriosis [in French]. J Gynecol Obstet Biol Reprod (Paris) 2007; 36: Kataoka ML, Togashi K, Yamaoka T, et al. Posterior cul-de-sac obliteration associated with endometriosis: MR imaging evaluation. Radiology 2005; 234: Kikuchi I, Takeuchi H, Kuwatsuru R, et al. Diagnosis of complete cul-de-sac obliteration (CCDSO) by the MRI jelly method. J Magn Reson Imaging 2009; 29: Chassang M, Novellas S, Bloch-Marcotte C, et al. Utility of vaginal and rectal contrast medium in MRI for the detection of deep pelvic endometriosis. Eur Radiol 2010; 20: Takeuchi H, Kuwatsuru R, Kitade M, et al. A novel technique using magnetic resonance imaging jelly for evaluation of rectovaginal endometriosis. Fertil Steril 2005; 83: [No authors listed]. Revised American Fertility Society classification of endometriosis: Fertil Steril 1985; 43: Landis JR, Koch GG. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics 1977; 33: Hornstein MD, Gleason RE, Orav J, et al. The reproducibility of the revised American Fertility Society classification of endometriosis. Fertil Steril 1993; 59: Buhmann-Kirchhoff S, Lang R, Kirchhoff C, et al. Functional cine MR imaging for the detection and mapping of intraabdominal adhesions: method and surgical correlation. Eur Radiol 2008; 18: Kirchhoff S, Ladurner R, Kirchhoff C, Mussack T, Reiser MF, Lienemann A. Detection of recurrent hernia and intraabdominal adhesions following incisional hernia repair: a functional cine MRI-study. Abdom Imaging 2010; 35: AJR:199, December