Radiol med DOI 10.1007/s11547-013-0336-1
ABDOMINAL RADIOLOGY
How to improve MRI accuracy in detecting deep infiltrating colorectal endometriosis: MRI findings vs. laparoscopy and histopathology Anna Lia Valentini • Benedetta Gui • Maura Micco` • Maria Carla Mingote • Valeria Ninivaggi • Maurizio Guido • Gian Franco Zannoni • Eleonora Marrucci • Lorenzo Bonomo
Received: 21 January 2013 / Accepted: 20 May 2013 Ó Italian Society of Medical Radiology 2013
Abstract Objective To verify whether the capability of magnetic resonance imaging (MRI) in diagnosing deep infiltrating colorectal endometriosis (DICE) is improved using an association of MRI findings. Methods and materials The imaging database of our Institute of Radiology was retrospectively reviewed to identify patients subjected to MRI for a suspicion of deep infiltrating endometriosis. Medical history was then investigated and only patients who were also subjected to laparoscopy (LA) were included. Absence of LA represented the exclusion criterion. Images were evaluated twice by two radiologists using two different diagnostic criteria for an abnormal result: the contemporary presence of nodules or hypointense plaque-like lesions in the adjacent fat plane and bowel wall thickness, without (first criterion) or with (second criterion) semicircular shape (i.e. ‘‘radial and retracting shape’’). Radiologists worked in consensus evaluating images in two separate sessions, using the first criterion in This paper was presented at ECR 2013 as a scientific paper (Control number: 2377). A. L. Valentini (&) B. Gui M. Micco` M. C. Mingote V. Ninivaggi L. Bonomo Department of Radiological Sciences, Institute of Radiology, Catholic University of Rome, Policlinico A. Gemelli, L.go A. Gemelli 8, 00168 Rome, Italy e-mail: [email protected] M. Guido Department of Obstetrics and Gynecology, Catholic University of Rome, Policlinico A. Gemelli, L.go A. Gemelli 8, 00168 Rome, Italy G. F. Zannoni E. Marrucci Department of Pathology, Catholic University of Rome, Policlinico A. Gemelli, L.go A. Gemelli 8, 00168 Rome, Italy
the first section and the second criterion in the second one. MRI results were compared with LA or histopathology as the gold standard by 2 9 2 tables and statistically analyzed (k statistics). Likelihood-ratio test was also performed, being independent from the prevalence of the disease. Results By consulting case sheets, 33/50 females (ranging age 24–39 years, mean age 32.2 years) who were subjected to MRI also underwent LA. Intestinal resection for DICE was performed in 11/33 patients; in 22/33 superficial intestinal foci, adhesions/nodules in the fat plane were simply removed. When the first criterion was applied, MRI agreement with histopathology or LA was poor (51.5 %) (k value = 0.20; p \ 0.055), while it was improved (96.9 %) when using the second diagnostic criterion (k value = 0.93; p \ 0.0000). Likelihood ratio was 1.375 (95 % CI 0.69–2.72) using the first and 22 (95 % CI 20.08–24.1) using the second criterion. Conclusion The second criterion, or the joint presence of nodules or hypointense plaque-like lesions in the adjacent fat plane and bowel wall thickness showing ‘‘radial and retracting shape’’, improves MRI capability in DICE diagnosis. It can be considered an effective indicator of DICE on T2-weighted images at 1.5-T MRI, and can ensure the correct preoperative assessment of the disease for the best therapeutic procedure and treatment planning. Keywords Magnetic resonance imaging Deep endometriosis Pelvic pain
Introduction The reported rate of intestinal endometriosis is 4–37 % in patients with deep pelvic disease [1], in most cases (75–90 %) related to colorectal involvement [2].
123
Radiol med
Although the gold standard for this diagnosis is laparoscopy (LA), which also allows for an operative approach, a precise imaging evaluation of the lesion site and extension can be useful preoperatively, since only the infiltrating type needs bowel resection. Among the imaging techniques employed in the preoperative evaluation of deep infiltrating colorectal endometriosis (DICE), transvaginal or transrectal ultrasound as well as computed tomography enteroclysis has been used [3, 4]. However, they are limited by the small field of view and radiation exposure, respectively. Magnetic resonance imaging (MRI) is employed to investigate chronic pelvic pain in women [5] and is currently used in the assessment of deep endometriosis [6–8]. Double contrast barium enema (DCBE) is performed for preoperative DICE diagnosis, with 99 % reported overall accuracy [9]. The only report comparing DCBE and MRI in the diagnosis of colorectal endometriosis [7] states a better accuracy for DBCE (86.6 vs. 74.6 %). The diagnosis of DICE is considered possible on MRI when one or more of the following findings are identified: (1) hypointense nodular or plaque-like bowel wall thickening or irregular wall thickening, sometimes associated with haemorrhagic foci, showing enhancement after gadolinium injection [4, 10]; (2) loss of the fat-tissue plane between the intestinal loop and the uterus or other adjacent organs [6, 10]; (3) abnormal angulations of bowel loops [4, 6, 10]; (4) the ‘‘mushroom cap’’ sign, or rather the umbrella-like head of a mushroom image, which has been recently described as a typical MRI finding for DICE at 3.0-T MRI [2]; (5) ‘‘fan shaped’’ configuration of the bowel wall [11]. The aim of this study was to investigate whether MRI capability in DICE diagnosis could be improved by using an association of two selected MRI findings, namely: (a) thickened bowel wall with ‘‘radial and retracting shape’’ associated with (b) nodules or plaque-like lesions in the adjacent fat-tissue plane.
Materials and methods Study population and design This was a retrospective, single-centre study including patients suffering from chronic pelvic pain, referred to the Dysfunctional Gynaecological Unit of our university (Endometriosis Clinic) for imaging. The institutional review board granted permission for this study, and the requirement for informed consent was waived. The RIS-PACS system of our Institute of Radiology (radiological information system: Imagoweb-El.Co. S.r.l., Savona, Italy; picture archiving and communication system: Carestream Health, Genoa, Italy) was used for the search. Applied filters were: interval time, 2007–2012;
123
technique, MRI; UDC (cost-centre diagnostic unit), Gynaecology; ULD (logistic diagnostic unit), Dysfunctional Gynaecological Unit (Endometriosis Clinic). In this way, women who underwent MRI within the past 5 years because of suspected deep endometriosis could be identified. Then, the medical records were assessed for diagnostic or operative LA, and histopathology in the case of bowel resection. Inclusion criteria were: (a) MRI examination for suspected deep endometriosis and (b) diagnostic or operative LA. The absence of LA was the exclusion criterion. The database search, medical record assessment and histopathological correlation were carried out by a radiologist not involved in the assessment of the MRI findings. The radiologists assessing the MRI data were not aware of the patient’s clinical symptoms or outcome. MRI technique The MRI protocol applied in our department for the investigation of suspected pelvic endometriosis involves the use of a 1.5-T superconducting magnet (Vectra, GE Medical Systems, Milwaukee, USA) with a phased-array coil and no rectal distension. Patients are requested to fast for 6 h before the examination. To limit intestinal peristalsis and to attenuate uterine contractions, patients receive an intramuscular injection of 1 mg of hyoscine Nbutylbromide (Buscopan, Schering, Berlin, Germany) before the examination and are always studied after the 8th day of the menstrual cycle, to better identify hyperintense bleeding lesions. Axial T1-weighted spin echo images, axial coronal and sagittal T1-weighted gradient-echo images with fat suppression (to better evaluate hyperintense bleeding lesions), axial, coronal and sagittal T2-weighted fast spin echo images and axial T2-weighted fast spin echo images extending to the upper abdomen (to evaluate hydronephrosis) are always obtained. Dynamic axial and delayed phase sagittal images acquired with LAVA (liver acquisition with volume acquisition) sequence (3D T1-weighted gradient echo) after intravenous injection of gadoliniumbased contrast material are employed in selected cases. In particular, LAVA sequences with fat saturation are usually employed when endometriomas are investigated, to exclude degeneration, or also in the case of adenomyosis to verify the grade of infiltration of the myometrial surface. This study was retrospective and images after contrast medium injection (gadobenate dimeglumine, MultiHance 0.5 M, Bracco, Milan, Italy) were not always available. For this reason and since no significant benefit of intravenous gadolinium has been reported in the literature for deep pelvic endometriosis [8], the MRI evaluation was performed only on the T2-weighted images. The technical details are summarised in Table 1.
Radiol med Table 1 Magnetic resonance imaging (MRI) sequence characteristics Sequences
TR/TE (ms)
ETL
ST (mm)
MATRIX
FOV (cm)
NEX
IG
COIL
3D localiser
Minimum
–
10
256 9 128
36
2
2
Body
Axial T1W SE
640/1.8
–
4
288 9 256
24
1.5
0.5
Body
Axial T2W FSE
4,800/85
12
4
320 9 256
24
2
0.5
Body
Sagittal T2W FSE
4,550/102
10
4
384/234
24
2
0.5
Body
Coronal T2W FSE
5,200/85
13
4
320x256
30
2
0.5
Body
Axial T1W GRE FS
390/1.8
–
4
320 9 256
24
2
0.5
Body
Sagittal T1W GRE FS
460/4.2
–
4
320 9 256
24
2
0.4
Body
Coronal T1W GRE FS
390/3.3
–
4
320/224
30
2
0.5
Body
Coronal T2 FSE FS
5,175/85
11
4
320 9 224
30
2
0.5
Body
Axial T2W FSE Body
2,200/84
9
6
256 9 224
40
1
1
Body
Axial LAVA
Minimum
3
256 9 192
40
1
–
Body
–
SE spin echo, FSE fast spin echo, FS fat saturated, TR/TE repetition time ms/echo time ms, ETL echo train length, ST section thickness, FOV field of view, NEX number of excitations, IG intersection gap
MRI data MR images were assessed as showing normal (i.e. no DICE) or abnormal findings (i.e. DICE) using two different diagnostic criteria based on T2-weighted morphological images: –
–
The absence or presence of generally thickened intestinal wall surrounded by fat-plane interface represents a normal result; the contemporary presence of nodules or hypointense plaque-like lesions in the fat plane and regularly thickened intestinal wall is an abnormal result. Absence or presence of generally thickened intestinal wall with variable fat-plane interface represents a normal result; the contemporary presence of nodules or hypointense plaque-like lesions in the fat plane and thickened intestinal wall with ‘‘radial and retracting shape’’ is an abnormal result.
The terms ‘‘nodules’’ and ‘‘plaque-like lesions’’ are used to identify lesions with or without ‘‘mass effect’’, respectively. Both types of lesions can be found in the fat plane adjacent to the bowel wall in the case of deep endometriosis, such as fresh nodules in the recto-vaginal septum or fibrotic lesions at the uterine torus. Bowel wall thickness was not evaluated in terms of millimetres, as it is a relative rather than absolute evaluation, related to the morphology of the near proximal and distal portion of the same intestinal segment. The focal thickness of the bowel wall without semicircular shape was considered a ‘‘regular thickness’’, while a ‘‘radial and retracting shape’’ was assessed when a semicircular shape of the bowel was found. The term ‘‘radial and retracting shape’’ can be considered a 1.5-T version of the ‘‘mushroom cap’’ sign [2] and resembles the ‘‘fan shape’’ configuration of the
bowel wall, as recently described at 1.5-T imaging for the diagnosis of DICE [11]. To assess what association of signs improves DICE diagnosis, MRI evaluation was carried out twice by two radiologists with 10 and 15 years of experience in female pelvic MRI, who reviewed the images in consensus in two separate sessions. In the first evaluation (first diagnostic criterion), any general thickness of the bowel wall combined with nodules or plaque-like lesions in the adjacent fatty plane was considered abnormal, while in the second evaluation (second diagnostic criterion), only the characteristic ‘‘radial and retracting shape’’ of the thickened bowel wall was considered abnormal when associated with nodules or plaque-like lesions in the adjacent fatty plane. When no nodule or plaque-like lesion was identified in the adjacent fatty plane, the presence of bowel thickness, even without adjacent fat-plane interface, was always considered a normal result. LA or histopathology specimens in the case of surgical resection were taken as the gold standard. Reference standard and statistical analysis LA was always performed at the hospital by a single skilled gynaecological surgeon specialised in endometriosis who was assisted by an intestinal surgeon in the case of bowel resection. Superficial, or serosal, involvement of the bowel was directly diagnosed by LA visualisation and was treated by superficial excision, serosal shaving and adhesion removal. DICE was suspected in the event of large or disc-like implants retracting the wall and was treated by segmental bowel resection. All surgical specimens were sent to the hospital’s histopathology service and were fixed in 10 % formalin for 24 h. The entire lesion was routinely processed and
123
Radiol med
embedded in paraffin. From each paraffin block, two 3-lmthick histological sections, cut at different levels and stained with haematoxylin–eosin, were obtained and analysed. The histopathology specimens were evaluated by a single pathologist with 20 years’ experience in female pelvic pathology. The diagnosis of endometriosis was made on the basis of the presence of endometrial-type epithelium and endometrial stroma on bowel resection specimens. The MRI results in DICE diagnosis obtained using the two previously described criteria (see ‘‘MRI data’’) were compared by 2 9 2 tables with the gold standard histopathology in the case of bowel resection for DICE. Comparison of MRI results with LA was used in patients not subjected to bowel resection because of the absence of disease or presence of superficial foci (no DICE) and were statistically analysed (k statistic). The strength of agreement was assessed as follows: k values less than or equal to 0.20 indicated poor agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, good agreement; and 0.81–1.0, excellent agreement. Values of p \ 0.05 were considered to be statistically significant. The likelihood ratio test was also applied, being independent of the prevalence of disease.
Results Thirty-three patients (age range 24–39 years; mean age 32.2) met the inclusion criteria. Bowel resection for DICE
Fig. 1 Bar chart showing the capability of magnetic resonance imging (MRI) in the assessment of deep infiltrating type of colorectal endometriosis. Histograms show a high diagnostic precision of MRI vs. histopathology or laparoscopy when the second criterion is applied, with a markedly decreased rate of false-positive results. First criterion: absence or presence of generally thickened intestinal wall surrounded by fat-plane interface represents a normal result (no deep infiltrating colorectal endometriosis); the simultaneous presence of nodules or hypointense plaque-like lesions in the fat plane and
123
was performed in 11/33 patients (rectosigmoid nodule ablation in 3/11 cases; recto-sigmoid resection in 6/11 cases; sigmoid tract resection in 2/11 cases), whereas in the remaining 22/33 patients, superficial foci of intestinal endometriosis (7/22), adhesions between the intestinal serosa and surrounding structures (9/22) and nodules of the rectovaginal septum (6/22) were diagnosed and removed at LA. Using the first MRI diagnostic criterion (Fig. 1) for an abnormal result (i.e. the contemporary presence of nodules or hypointense plaque-like lesions in the adjacent fattissue plane and generally thickened intestinal wall), a diagnosis of DICE was established by MRI in 27/33 cases, but it was confirmed at histopathology only in 11 cases, since either no disease or superficial disease was found at LA in 16 cases (Table 2). The agreement with the gold standards histopathology or LA was poor (k = 0.20; p \ 0.055) with an unacceptable rate of falsepositive results (72.7 %). Using the second diagnostic criterion (Fig. 1) for an abnormal result (i.e. the contemporary presence of nodules or hypointense plaque-like lesions in the adjacent fat-tissue plane and thickened intestinal wall with characteristic ‘‘radial and retracting shape’’), the overall accuracy of MRI in DICE diagnosis was improved (96.9 %) (Table 2), with the false-positive results decreasing from 72.7 % to 4.5 %. The k value was 0.93, and the p value became significant (p \ 0.0000). The likelihood ratio test was 1.375 (95 % CI 0.69–2.72) using the first and 22 (95 % CI 20.08–24.1) using the second diagnostic criterion.
generally thickened intestinal wall is an abnormal result (deep infiltrating colorectal endometriosis). Second criterion: absence or presence of generally thickened intestinal wall with variable fat-plane interface represents a normal result (no deep infiltrating colorectal endometriosis); the simultaneous presence of nodules or hypointense plaque-like lesions in the fat plane and thickened intestinal wall with a ‘‘radial and retracting pattern’’ is an abnormal result (deep infiltrating colorectal endometriosis)
Radiol med Table 2 2 9 2 tables: MRI vs. the gold standard histopathology (in bowel resection for DICE) or laparoscopy (for superficial disease or no disease)
Histopatology (?) DICE
Laparoscopy (-) no disease or superficial disease
Total
11
16
27
First diagnostic criterion MRI MRI? nodules or hypointense plaque-like lesions in the adjacent fat-tissue plane associated with regularly thickened intestinal wall is considered DICE MRI- no thickness or generally thickened intestinal wall surrounded by fat plane
0
6
6
11
22
33
11
1
12
0
21
21
11
22
33
Second diagnostic criterion MRI MRI? nodules or hypointense plaque-like lesions in the adjacent fat-tissue plane associated with thickened intestinal wall showing ‘‘radial and retracting shape’’ is considered DICE MRI- no thickness or generally thickened intestinal wall surrounded or not by fat-tissue plane interface represents No DICE
Discussion These results show that the simultaneous presence of nodules or hypointense plaque-like lesions in the fat-tissue plane and thickened intestinal wall characteristically showing ‘‘radial and retracting shape’’ (i.e. the second diagnostic criterion) improves the MRI diagnosis of DICE (Figs. 2, 3). Using this association to define an abnormal result, the MRI capability in the diagnosis of DICE increased, with an excellent agreement with histopathology (97 %) and a greatly decreased rate of false-positive diagnoses (4.5 vs. 72.7 %). The better diagnostic performance of the second diagnostic criterion was also confirmed with the likelihood ratio test. To our knowledge, the overall accuracy of T2-weighted MRI is higher in this study (97 %) than in other published reports on the same subject (75–96 %) [4, 6, 7, 10, 11]. Some reports identified a typical MRI morphology of the intestinal wall thickness in the case of DICE: the ‘‘mushroom cap sign’’ was first reported at 3.0-T magnet imaging [2], but it was not statistically validated because of the small number of patients in that series. The ‘‘fan shape’’ configuration of the bowel wall was reported with 1.5-T magnet imaging [11], showing 96 % accuracy. A prior study suggested the same typical feature using transrectal and transvaginal ultrasound [12]. Taking into consideration a combination of signs, or the contemporary presence of nodules or plaque-like lesions in the adjacent fat-tissue plane and the typical bowel thickness morphology for the correct assessment of DICE, as this study suggests, can be justified by the course of the disease. Even if retracting adhesions and muscle layer invasion are responsible for producing the protrusion which results in a typical
semicircular morphology of the involved intestinal tract— variously known as the ‘‘mushroom cap sign’’, the ‘‘fan shape image’’ or the ‘‘radial and retracting shape’’—it should be noted that DICE is improbable when no nodules or plaque-like lesions are detectable in the fat-tissue plane closely adjacent to the bowel wall. DICE is in fact a result of endometriotic foci gradually infiltrating, in an externalto-internal direction, the colorectal wall. The results of this study confirm the usefulness of this association of signs. Colorectal endometriosis can cause severe intestinal symptoms, including pelvic pain, rectal bleeding and problems with defaecation, but needs to be surgically treated by intestinal resection only in the deeply infiltrating type, since fibrosis invading the intestinal wall muscularis does not respond to medical therapy and advanced disease can be complicated by lumen obstruction [8]. Superficial endometriosis lesions located on the serosa surface are not easily evaluated by imaging [2], but are negligible since they do not require bowel resection and are usually diagnosed and simply removed during LA. The gold standard for this diagnosis is LA, which directly visualises suspicious lesions and also represents the most common operative approach to pelvic disease. However, a precise preoperative assessment by imaging of location and extent of the lesions can be helpful to ensure the best therapeutic procedure and treatment planning. In the past, the imaging evaluation of colorectal involvement due to endometriosis was especially assessed by DCBE, which seems to correlate with the patients’ clinical history and clinical findings and was shown to be capable of detecting bowel wall involvement requiring intestinal surgery [9]. Recently, some reports have demonstrated the ability of MRI to detect DICE using T2-weighted images without or with air,
123
Radiol med
Fig. 2 Deep infiltrating endometriosis of the rectosigmoid junction in a 32-year-old patient suffering from chronic pelvic pain and defaecation problems. MRI and histopathology. T2-weighted sagittal MRI a shows thickened rectosigmoid junction wall showing ‘‘radial and retracting shape’’ (a white arrows) and adhesions (a void arrow) which converge on the hypointense plaque-like lesion at the uterine torus, as shown on the T2-weighted axial fat-saturated image (b arrow). Histopathology specimen after bowel resection (c) confirms the infiltrating type of colorectal endometriosis showing endometrial typical glands (white arrow) and stroma (white star) within the muscular layer (void star) of the large bowel. The intestinal submucosa (black star) and mucosa (void arrow) are also visible
123
Fig. 3 Deep infiltrating endometriosis of the rectosigmoid junction in a 36-year-old patient suffering from chronic pelvic pain and rectal bleeding. MRI and laparoscopy. T2-weighted sagittal MRI a shows a thickened colorectal junction wall typically showing a ‘‘radial and retracting shape’’ (white arrows) associated with hypointense plaquelike lesion at the uterine torus (black arrow). At laparoscopy (b), adhesions between the uterus and rectosigmoid junction are visible (arrows). After adhesion removal (c), a ‘‘disc-like’’ implant of endometriosis is visible (arrow). U uterus, RSJ rectosigmoid junction
Radiol med
water or gel distension of the rectum [2, 7, 13, 14], employing contrast-enhanced MRI [8] or MR colonography [4] or also using 3.0-T superconductive magnets [2, 13]. According to the literature, neither rectal distension by gel nor gadolinium-enhanced T1-weighted imaging seems to improve detection of deep endometriosis by MRI [8, 13]. Some authors state that MR colonography improves the detection of colorectal endometriosis [4], since in their series accuracy increased to 97 (vs. 91 % of T2-weighted MR images). However, MR colonography is a complex procedure requiring administration of contrast medium and complete colonic distension, which is potentially discomforting for patients. MR colonography is also time-consuming, since examination time is about 35 min, resulting in a reduction of patient tolerance. By contrast, MRI is an ‘‘all-in-one examination’’ able to detect endometriotic foci in different sites, including the colorectal site [4, 6, 7, 10, 11], as well as remaining the examination of choice for the detection of deep pelvic implants [4]. Patient management is improved by using a preoperative assessment of the disease by MRI, as decisions to perform intestinal surgery will involve the need for a colorectal surgeon. The retrospective analysis of data could be considered a limitation of this study. However, it should be noted that MRI and LA, as well as the histopathological evaluation, were always performed at the hospital, that MRI was always carried out by the same pool of radiologists using the same study protocol, and that the MR images were reviewed by the same two radiologists with 10 and 15 years of experience in female pelvic MRI, who were not aware of the patients’ specific symptoms and outcome. In conclusion, the contemporary presence of nodules or hypointense plaque-like lesions in the fat-tissue plane and thickened intestinal wall characteristically showing a ‘‘radial and retracting shape’’ improves MRI capability in DICE diagnosis. The association can be considered an effective indicator of DICE on T2-weighted images at 1.5T MRI, and can ensure the correct preoperative assessment of the disease for the best therapeutic procedure and treatment planning.
Gian Franco Zannoni, Eleonora Marrucci and Lorenzo Bonomo declare no conflict of interest.
References 1. Remorgida V, Ferrero S, Fulcheri E et al (2007) Bowel endometriosis: presentation, diagnosis, and treatment. Obstet Gynecol Surv 62:461–470 2. Yoon JH, Choi D, Jang K-T et al (2010) Deep recto sigmoid endometriosis: ‘‘mushroom cap’’ sign on T2-weighted MR imaging. Abdom Imaging 35:726–731 3. Bergamini V, Ghezzi F, Scarperi S et al (2010) Preoperative assessment of intestinal endometriosis: a comparison of transvaginal sonography with water-contrast in the rectum, transrectal sonography, and barium enema. Abdom Imaging 35:732–736 4. Scardapane A, Bettocchi S, Lorusso F et al (2011) Diagnosis of colorectal endometriosis: contribution of contrast enhanced MR colonography. Eur Radiol 21:1553–1563 5. Valentini AL, Gui B, Basilico R et al (2012) Magnetic resonance imaging in women with pelvic pain from gynecological causes: a pictorial review. Radiol Med 117:575–592 6. Bazot M, Daral E, Hourany N et al (2004) Deep pelvic endometriosis: MR imaging for diagnosis and prediction of extension of disease. Radiology 232:379–389 7. Faccioli N, Foti G, Manfredi R et al (2010) Evaluation of colonic involvement in endometriosis: double-contrast barium enema vs. magnetic resonance imaging. Abdom Imaging 35:414–421 8. Bazot M, Gasnera A, Lafonta C et al (2011) Deep pelvic endometriosis: limited additional diagnostic value of post-contrast in comparison with conventional MR images. Eur J Radiol 80:331–339 9. Landi S, Barbieri F, Fiaccavento A et al (2004) Preoperative double-contrast barium enema in patients with suspected intestinal endometriosis. J Am Assoc Gynecol Laparosc 11:223–228 10. Marchal L, Nothft MA, Coelho F, Choi K (2010) Deep pelvic endometriosis: MR imaging. Abdom Imaging 35:708–715 11. Busard MP, Van der Houwen LEE, Bleeker Maaike C et al (2012) Deep infiltrating endometriosis of the bowel: MR imaging as a method to predict muscular invasion. Abdom Imaging 37:549–557 12. Koga K, Osuga Y, Yano T et al (2003) Characteristic images of deeply infiltrating rectosigmoid endometriosis on transvaginal and transrectal ultrasonography. Hum Reprod 18:1328–1333 13. Hottat N, Larrousse C, Anaf V et al (2009) Endometriosis: contribution of 3.0-T pelvic MR imaging in preoperative assessment—initial results. Radiology 253:126–134 14. Loubeyre P, Petignat P, Jacob S et al (2009) Anatomic distribution of posterior deeply infiltrating endometriosis on MRI after vaginal and rectal gel opacification. AJR 192:1625–1631
Conflict of interest Anna Lia Valentini, Benedetta Gui, Maura Micco`, Maria Carla Mingote, Valeria Ninivaggi, Maurizio Guido,
123
ABDOMINAL RADIOLOGY
How to improve MRI accuracy in detecting deep infiltrating colorectal endometriosis: MRI findings vs. laparoscopy and histopathology Anna Lia Valentini • Benedetta Gui • Maura Micco` • Maria Carla Mingote • Valeria Ninivaggi • Maurizio Guido • Gian Franco Zannoni • Eleonora Marrucci • Lorenzo Bonomo
Received: 21 January 2013 / Accepted: 20 May 2013 Ó Italian Society of Medical Radiology 2013
Abstract Objective To verify whether the capability of magnetic resonance imaging (MRI) in diagnosing deep infiltrating colorectal endometriosis (DICE) is improved using an association of MRI findings. Methods and materials The imaging database of our Institute of Radiology was retrospectively reviewed to identify patients subjected to MRI for a suspicion of deep infiltrating endometriosis. Medical history was then investigated and only patients who were also subjected to laparoscopy (LA) were included. Absence of LA represented the exclusion criterion. Images were evaluated twice by two radiologists using two different diagnostic criteria for an abnormal result: the contemporary presence of nodules or hypointense plaque-like lesions in the adjacent fat plane and bowel wall thickness, without (first criterion) or with (second criterion) semicircular shape (i.e. ‘‘radial and retracting shape’’). Radiologists worked in consensus evaluating images in two separate sessions, using the first criterion in This paper was presented at ECR 2013 as a scientific paper (Control number: 2377). A. L. Valentini (&) B. Gui M. Micco` M. C. Mingote V. Ninivaggi L. Bonomo Department of Radiological Sciences, Institute of Radiology, Catholic University of Rome, Policlinico A. Gemelli, L.go A. Gemelli 8, 00168 Rome, Italy e-mail: [email protected] M. Guido Department of Obstetrics and Gynecology, Catholic University of Rome, Policlinico A. Gemelli, L.go A. Gemelli 8, 00168 Rome, Italy G. F. Zannoni E. Marrucci Department of Pathology, Catholic University of Rome, Policlinico A. Gemelli, L.go A. Gemelli 8, 00168 Rome, Italy
the first section and the second criterion in the second one. MRI results were compared with LA or histopathology as the gold standard by 2 9 2 tables and statistically analyzed (k statistics). Likelihood-ratio test was also performed, being independent from the prevalence of the disease. Results By consulting case sheets, 33/50 females (ranging age 24–39 years, mean age 32.2 years) who were subjected to MRI also underwent LA. Intestinal resection for DICE was performed in 11/33 patients; in 22/33 superficial intestinal foci, adhesions/nodules in the fat plane were simply removed. When the first criterion was applied, MRI agreement with histopathology or LA was poor (51.5 %) (k value = 0.20; p \ 0.055), while it was improved (96.9 %) when using the second diagnostic criterion (k value = 0.93; p \ 0.0000). Likelihood ratio was 1.375 (95 % CI 0.69–2.72) using the first and 22 (95 % CI 20.08–24.1) using the second criterion. Conclusion The second criterion, or the joint presence of nodules or hypointense plaque-like lesions in the adjacent fat plane and bowel wall thickness showing ‘‘radial and retracting shape’’, improves MRI capability in DICE diagnosis. It can be considered an effective indicator of DICE on T2-weighted images at 1.5-T MRI, and can ensure the correct preoperative assessment of the disease for the best therapeutic procedure and treatment planning. Keywords Magnetic resonance imaging Deep endometriosis Pelvic pain
Introduction The reported rate of intestinal endometriosis is 4–37 % in patients with deep pelvic disease [1], in most cases (75–90 %) related to colorectal involvement [2].
123
Radiol med
Although the gold standard for this diagnosis is laparoscopy (LA), which also allows for an operative approach, a precise imaging evaluation of the lesion site and extension can be useful preoperatively, since only the infiltrating type needs bowel resection. Among the imaging techniques employed in the preoperative evaluation of deep infiltrating colorectal endometriosis (DICE), transvaginal or transrectal ultrasound as well as computed tomography enteroclysis has been used [3, 4]. However, they are limited by the small field of view and radiation exposure, respectively. Magnetic resonance imaging (MRI) is employed to investigate chronic pelvic pain in women [5] and is currently used in the assessment of deep endometriosis [6–8]. Double contrast barium enema (DCBE) is performed for preoperative DICE diagnosis, with 99 % reported overall accuracy [9]. The only report comparing DCBE and MRI in the diagnosis of colorectal endometriosis [7] states a better accuracy for DBCE (86.6 vs. 74.6 %). The diagnosis of DICE is considered possible on MRI when one or more of the following findings are identified: (1) hypointense nodular or plaque-like bowel wall thickening or irregular wall thickening, sometimes associated with haemorrhagic foci, showing enhancement after gadolinium injection [4, 10]; (2) loss of the fat-tissue plane between the intestinal loop and the uterus or other adjacent organs [6, 10]; (3) abnormal angulations of bowel loops [4, 6, 10]; (4) the ‘‘mushroom cap’’ sign, or rather the umbrella-like head of a mushroom image, which has been recently described as a typical MRI finding for DICE at 3.0-T MRI [2]; (5) ‘‘fan shaped’’ configuration of the bowel wall [11]. The aim of this study was to investigate whether MRI capability in DICE diagnosis could be improved by using an association of two selected MRI findings, namely: (a) thickened bowel wall with ‘‘radial and retracting shape’’ associated with (b) nodules or plaque-like lesions in the adjacent fat-tissue plane.
Materials and methods Study population and design This was a retrospective, single-centre study including patients suffering from chronic pelvic pain, referred to the Dysfunctional Gynaecological Unit of our university (Endometriosis Clinic) for imaging. The institutional review board granted permission for this study, and the requirement for informed consent was waived. The RIS-PACS system of our Institute of Radiology (radiological information system: Imagoweb-El.Co. S.r.l., Savona, Italy; picture archiving and communication system: Carestream Health, Genoa, Italy) was used for the search. Applied filters were: interval time, 2007–2012;
123
technique, MRI; UDC (cost-centre diagnostic unit), Gynaecology; ULD (logistic diagnostic unit), Dysfunctional Gynaecological Unit (Endometriosis Clinic). In this way, women who underwent MRI within the past 5 years because of suspected deep endometriosis could be identified. Then, the medical records were assessed for diagnostic or operative LA, and histopathology in the case of bowel resection. Inclusion criteria were: (a) MRI examination for suspected deep endometriosis and (b) diagnostic or operative LA. The absence of LA was the exclusion criterion. The database search, medical record assessment and histopathological correlation were carried out by a radiologist not involved in the assessment of the MRI findings. The radiologists assessing the MRI data were not aware of the patient’s clinical symptoms or outcome. MRI technique The MRI protocol applied in our department for the investigation of suspected pelvic endometriosis involves the use of a 1.5-T superconducting magnet (Vectra, GE Medical Systems, Milwaukee, USA) with a phased-array coil and no rectal distension. Patients are requested to fast for 6 h before the examination. To limit intestinal peristalsis and to attenuate uterine contractions, patients receive an intramuscular injection of 1 mg of hyoscine Nbutylbromide (Buscopan, Schering, Berlin, Germany) before the examination and are always studied after the 8th day of the menstrual cycle, to better identify hyperintense bleeding lesions. Axial T1-weighted spin echo images, axial coronal and sagittal T1-weighted gradient-echo images with fat suppression (to better evaluate hyperintense bleeding lesions), axial, coronal and sagittal T2-weighted fast spin echo images and axial T2-weighted fast spin echo images extending to the upper abdomen (to evaluate hydronephrosis) are always obtained. Dynamic axial and delayed phase sagittal images acquired with LAVA (liver acquisition with volume acquisition) sequence (3D T1-weighted gradient echo) after intravenous injection of gadoliniumbased contrast material are employed in selected cases. In particular, LAVA sequences with fat saturation are usually employed when endometriomas are investigated, to exclude degeneration, or also in the case of adenomyosis to verify the grade of infiltration of the myometrial surface. This study was retrospective and images after contrast medium injection (gadobenate dimeglumine, MultiHance 0.5 M, Bracco, Milan, Italy) were not always available. For this reason and since no significant benefit of intravenous gadolinium has been reported in the literature for deep pelvic endometriosis [8], the MRI evaluation was performed only on the T2-weighted images. The technical details are summarised in Table 1.
Radiol med Table 1 Magnetic resonance imaging (MRI) sequence characteristics Sequences
TR/TE (ms)
ETL
ST (mm)
MATRIX
FOV (cm)
NEX
IG
COIL
3D localiser
Minimum
–
10
256 9 128
36
2
2
Body
Axial T1W SE
640/1.8
–
4
288 9 256
24
1.5
0.5
Body
Axial T2W FSE
4,800/85
12
4
320 9 256
24
2
0.5
Body
Sagittal T2W FSE
4,550/102
10
4
384/234
24
2
0.5
Body
Coronal T2W FSE
5,200/85
13
4
320x256
30
2
0.5
Body
Axial T1W GRE FS
390/1.8
–
4
320 9 256
24
2
0.5
Body
Sagittal T1W GRE FS
460/4.2
–
4
320 9 256
24
2
0.4
Body
Coronal T1W GRE FS
390/3.3
–
4
320/224
30
2
0.5
Body
Coronal T2 FSE FS
5,175/85
11
4
320 9 224
30
2
0.5
Body
Axial T2W FSE Body
2,200/84
9
6
256 9 224
40
1
1
Body
Axial LAVA
Minimum
3
256 9 192
40
1
–
Body
–
SE spin echo, FSE fast spin echo, FS fat saturated, TR/TE repetition time ms/echo time ms, ETL echo train length, ST section thickness, FOV field of view, NEX number of excitations, IG intersection gap
MRI data MR images were assessed as showing normal (i.e. no DICE) or abnormal findings (i.e. DICE) using two different diagnostic criteria based on T2-weighted morphological images: –
–
The absence or presence of generally thickened intestinal wall surrounded by fat-plane interface represents a normal result; the contemporary presence of nodules or hypointense plaque-like lesions in the fat plane and regularly thickened intestinal wall is an abnormal result. Absence or presence of generally thickened intestinal wall with variable fat-plane interface represents a normal result; the contemporary presence of nodules or hypointense plaque-like lesions in the fat plane and thickened intestinal wall with ‘‘radial and retracting shape’’ is an abnormal result.
The terms ‘‘nodules’’ and ‘‘plaque-like lesions’’ are used to identify lesions with or without ‘‘mass effect’’, respectively. Both types of lesions can be found in the fat plane adjacent to the bowel wall in the case of deep endometriosis, such as fresh nodules in the recto-vaginal septum or fibrotic lesions at the uterine torus. Bowel wall thickness was not evaluated in terms of millimetres, as it is a relative rather than absolute evaluation, related to the morphology of the near proximal and distal portion of the same intestinal segment. The focal thickness of the bowel wall without semicircular shape was considered a ‘‘regular thickness’’, while a ‘‘radial and retracting shape’’ was assessed when a semicircular shape of the bowel was found. The term ‘‘radial and retracting shape’’ can be considered a 1.5-T version of the ‘‘mushroom cap’’ sign [2] and resembles the ‘‘fan shape’’ configuration of the
bowel wall, as recently described at 1.5-T imaging for the diagnosis of DICE [11]. To assess what association of signs improves DICE diagnosis, MRI evaluation was carried out twice by two radiologists with 10 and 15 years of experience in female pelvic MRI, who reviewed the images in consensus in two separate sessions. In the first evaluation (first diagnostic criterion), any general thickness of the bowel wall combined with nodules or plaque-like lesions in the adjacent fatty plane was considered abnormal, while in the second evaluation (second diagnostic criterion), only the characteristic ‘‘radial and retracting shape’’ of the thickened bowel wall was considered abnormal when associated with nodules or plaque-like lesions in the adjacent fatty plane. When no nodule or plaque-like lesion was identified in the adjacent fatty plane, the presence of bowel thickness, even without adjacent fat-plane interface, was always considered a normal result. LA or histopathology specimens in the case of surgical resection were taken as the gold standard. Reference standard and statistical analysis LA was always performed at the hospital by a single skilled gynaecological surgeon specialised in endometriosis who was assisted by an intestinal surgeon in the case of bowel resection. Superficial, or serosal, involvement of the bowel was directly diagnosed by LA visualisation and was treated by superficial excision, serosal shaving and adhesion removal. DICE was suspected in the event of large or disc-like implants retracting the wall and was treated by segmental bowel resection. All surgical specimens were sent to the hospital’s histopathology service and were fixed in 10 % formalin for 24 h. The entire lesion was routinely processed and
123
Radiol med
embedded in paraffin. From each paraffin block, two 3-lmthick histological sections, cut at different levels and stained with haematoxylin–eosin, were obtained and analysed. The histopathology specimens were evaluated by a single pathologist with 20 years’ experience in female pelvic pathology. The diagnosis of endometriosis was made on the basis of the presence of endometrial-type epithelium and endometrial stroma on bowel resection specimens. The MRI results in DICE diagnosis obtained using the two previously described criteria (see ‘‘MRI data’’) were compared by 2 9 2 tables with the gold standard histopathology in the case of bowel resection for DICE. Comparison of MRI results with LA was used in patients not subjected to bowel resection because of the absence of disease or presence of superficial foci (no DICE) and were statistically analysed (k statistic). The strength of agreement was assessed as follows: k values less than or equal to 0.20 indicated poor agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, good agreement; and 0.81–1.0, excellent agreement. Values of p \ 0.05 were considered to be statistically significant. The likelihood ratio test was also applied, being independent of the prevalence of disease.
Results Thirty-three patients (age range 24–39 years; mean age 32.2) met the inclusion criteria. Bowel resection for DICE
Fig. 1 Bar chart showing the capability of magnetic resonance imging (MRI) in the assessment of deep infiltrating type of colorectal endometriosis. Histograms show a high diagnostic precision of MRI vs. histopathology or laparoscopy when the second criterion is applied, with a markedly decreased rate of false-positive results. First criterion: absence or presence of generally thickened intestinal wall surrounded by fat-plane interface represents a normal result (no deep infiltrating colorectal endometriosis); the simultaneous presence of nodules or hypointense plaque-like lesions in the fat plane and
123
was performed in 11/33 patients (rectosigmoid nodule ablation in 3/11 cases; recto-sigmoid resection in 6/11 cases; sigmoid tract resection in 2/11 cases), whereas in the remaining 22/33 patients, superficial foci of intestinal endometriosis (7/22), adhesions between the intestinal serosa and surrounding structures (9/22) and nodules of the rectovaginal septum (6/22) were diagnosed and removed at LA. Using the first MRI diagnostic criterion (Fig. 1) for an abnormal result (i.e. the contemporary presence of nodules or hypointense plaque-like lesions in the adjacent fattissue plane and generally thickened intestinal wall), a diagnosis of DICE was established by MRI in 27/33 cases, but it was confirmed at histopathology only in 11 cases, since either no disease or superficial disease was found at LA in 16 cases (Table 2). The agreement with the gold standards histopathology or LA was poor (k = 0.20; p \ 0.055) with an unacceptable rate of falsepositive results (72.7 %). Using the second diagnostic criterion (Fig. 1) for an abnormal result (i.e. the contemporary presence of nodules or hypointense plaque-like lesions in the adjacent fat-tissue plane and thickened intestinal wall with characteristic ‘‘radial and retracting shape’’), the overall accuracy of MRI in DICE diagnosis was improved (96.9 %) (Table 2), with the false-positive results decreasing from 72.7 % to 4.5 %. The k value was 0.93, and the p value became significant (p \ 0.0000). The likelihood ratio test was 1.375 (95 % CI 0.69–2.72) using the first and 22 (95 % CI 20.08–24.1) using the second diagnostic criterion.
generally thickened intestinal wall is an abnormal result (deep infiltrating colorectal endometriosis). Second criterion: absence or presence of generally thickened intestinal wall with variable fat-plane interface represents a normal result (no deep infiltrating colorectal endometriosis); the simultaneous presence of nodules or hypointense plaque-like lesions in the fat plane and thickened intestinal wall with a ‘‘radial and retracting pattern’’ is an abnormal result (deep infiltrating colorectal endometriosis)
Radiol med Table 2 2 9 2 tables: MRI vs. the gold standard histopathology (in bowel resection for DICE) or laparoscopy (for superficial disease or no disease)
Histopatology (?) DICE
Laparoscopy (-) no disease or superficial disease
Total
11
16
27
First diagnostic criterion MRI MRI? nodules or hypointense plaque-like lesions in the adjacent fat-tissue plane associated with regularly thickened intestinal wall is considered DICE MRI- no thickness or generally thickened intestinal wall surrounded by fat plane
0
6
6
11
22
33
11
1
12
0
21
21
11
22
33
Second diagnostic criterion MRI MRI? nodules or hypointense plaque-like lesions in the adjacent fat-tissue plane associated with thickened intestinal wall showing ‘‘radial and retracting shape’’ is considered DICE MRI- no thickness or generally thickened intestinal wall surrounded or not by fat-tissue plane interface represents No DICE
Discussion These results show that the simultaneous presence of nodules or hypointense plaque-like lesions in the fat-tissue plane and thickened intestinal wall characteristically showing ‘‘radial and retracting shape’’ (i.e. the second diagnostic criterion) improves the MRI diagnosis of DICE (Figs. 2, 3). Using this association to define an abnormal result, the MRI capability in the diagnosis of DICE increased, with an excellent agreement with histopathology (97 %) and a greatly decreased rate of false-positive diagnoses (4.5 vs. 72.7 %). The better diagnostic performance of the second diagnostic criterion was also confirmed with the likelihood ratio test. To our knowledge, the overall accuracy of T2-weighted MRI is higher in this study (97 %) than in other published reports on the same subject (75–96 %) [4, 6, 7, 10, 11]. Some reports identified a typical MRI morphology of the intestinal wall thickness in the case of DICE: the ‘‘mushroom cap sign’’ was first reported at 3.0-T magnet imaging [2], but it was not statistically validated because of the small number of patients in that series. The ‘‘fan shape’’ configuration of the bowel wall was reported with 1.5-T magnet imaging [11], showing 96 % accuracy. A prior study suggested the same typical feature using transrectal and transvaginal ultrasound [12]. Taking into consideration a combination of signs, or the contemporary presence of nodules or plaque-like lesions in the adjacent fat-tissue plane and the typical bowel thickness morphology for the correct assessment of DICE, as this study suggests, can be justified by the course of the disease. Even if retracting adhesions and muscle layer invasion are responsible for producing the protrusion which results in a typical
semicircular morphology of the involved intestinal tract— variously known as the ‘‘mushroom cap sign’’, the ‘‘fan shape image’’ or the ‘‘radial and retracting shape’’—it should be noted that DICE is improbable when no nodules or plaque-like lesions are detectable in the fat-tissue plane closely adjacent to the bowel wall. DICE is in fact a result of endometriotic foci gradually infiltrating, in an externalto-internal direction, the colorectal wall. The results of this study confirm the usefulness of this association of signs. Colorectal endometriosis can cause severe intestinal symptoms, including pelvic pain, rectal bleeding and problems with defaecation, but needs to be surgically treated by intestinal resection only in the deeply infiltrating type, since fibrosis invading the intestinal wall muscularis does not respond to medical therapy and advanced disease can be complicated by lumen obstruction [8]. Superficial endometriosis lesions located on the serosa surface are not easily evaluated by imaging [2], but are negligible since they do not require bowel resection and are usually diagnosed and simply removed during LA. The gold standard for this diagnosis is LA, which directly visualises suspicious lesions and also represents the most common operative approach to pelvic disease. However, a precise preoperative assessment by imaging of location and extent of the lesions can be helpful to ensure the best therapeutic procedure and treatment planning. In the past, the imaging evaluation of colorectal involvement due to endometriosis was especially assessed by DCBE, which seems to correlate with the patients’ clinical history and clinical findings and was shown to be capable of detecting bowel wall involvement requiring intestinal surgery [9]. Recently, some reports have demonstrated the ability of MRI to detect DICE using T2-weighted images without or with air,
123
Radiol med
Fig. 2 Deep infiltrating endometriosis of the rectosigmoid junction in a 32-year-old patient suffering from chronic pelvic pain and defaecation problems. MRI and histopathology. T2-weighted sagittal MRI a shows thickened rectosigmoid junction wall showing ‘‘radial and retracting shape’’ (a white arrows) and adhesions (a void arrow) which converge on the hypointense plaque-like lesion at the uterine torus, as shown on the T2-weighted axial fat-saturated image (b arrow). Histopathology specimen after bowel resection (c) confirms the infiltrating type of colorectal endometriosis showing endometrial typical glands (white arrow) and stroma (white star) within the muscular layer (void star) of the large bowel. The intestinal submucosa (black star) and mucosa (void arrow) are also visible
123
Fig. 3 Deep infiltrating endometriosis of the rectosigmoid junction in a 36-year-old patient suffering from chronic pelvic pain and rectal bleeding. MRI and laparoscopy. T2-weighted sagittal MRI a shows a thickened colorectal junction wall typically showing a ‘‘radial and retracting shape’’ (white arrows) associated with hypointense plaquelike lesion at the uterine torus (black arrow). At laparoscopy (b), adhesions between the uterus and rectosigmoid junction are visible (arrows). After adhesion removal (c), a ‘‘disc-like’’ implant of endometriosis is visible (arrow). U uterus, RSJ rectosigmoid junction
Radiol med
water or gel distension of the rectum [2, 7, 13, 14], employing contrast-enhanced MRI [8] or MR colonography [4] or also using 3.0-T superconductive magnets [2, 13]. According to the literature, neither rectal distension by gel nor gadolinium-enhanced T1-weighted imaging seems to improve detection of deep endometriosis by MRI [8, 13]. Some authors state that MR colonography improves the detection of colorectal endometriosis [4], since in their series accuracy increased to 97 (vs. 91 % of T2-weighted MR images). However, MR colonography is a complex procedure requiring administration of contrast medium and complete colonic distension, which is potentially discomforting for patients. MR colonography is also time-consuming, since examination time is about 35 min, resulting in a reduction of patient tolerance. By contrast, MRI is an ‘‘all-in-one examination’’ able to detect endometriotic foci in different sites, including the colorectal site [4, 6, 7, 10, 11], as well as remaining the examination of choice for the detection of deep pelvic implants [4]. Patient management is improved by using a preoperative assessment of the disease by MRI, as decisions to perform intestinal surgery will involve the need for a colorectal surgeon. The retrospective analysis of data could be considered a limitation of this study. However, it should be noted that MRI and LA, as well as the histopathological evaluation, were always performed at the hospital, that MRI was always carried out by the same pool of radiologists using the same study protocol, and that the MR images were reviewed by the same two radiologists with 10 and 15 years of experience in female pelvic MRI, who were not aware of the patients’ specific symptoms and outcome. In conclusion, the contemporary presence of nodules or hypointense plaque-like lesions in the fat-tissue plane and thickened intestinal wall characteristically showing a ‘‘radial and retracting shape’’ improves MRI capability in DICE diagnosis. The association can be considered an effective indicator of DICE on T2-weighted images at 1.5T MRI, and can ensure the correct preoperative assessment of the disease for the best therapeutic procedure and treatment planning.
Gian Franco Zannoni, Eleonora Marrucci and Lorenzo Bonomo declare no conflict of interest.
References 1. Remorgida V, Ferrero S, Fulcheri E et al (2007) Bowel endometriosis: presentation, diagnosis, and treatment. Obstet Gynecol Surv 62:461–470 2. Yoon JH, Choi D, Jang K-T et al (2010) Deep recto sigmoid endometriosis: ‘‘mushroom cap’’ sign on T2-weighted MR imaging. Abdom Imaging 35:726–731 3. Bergamini V, Ghezzi F, Scarperi S et al (2010) Preoperative assessment of intestinal endometriosis: a comparison of transvaginal sonography with water-contrast in the rectum, transrectal sonography, and barium enema. Abdom Imaging 35:732–736 4. Scardapane A, Bettocchi S, Lorusso F et al (2011) Diagnosis of colorectal endometriosis: contribution of contrast enhanced MR colonography. Eur Radiol 21:1553–1563 5. Valentini AL, Gui B, Basilico R et al (2012) Magnetic resonance imaging in women with pelvic pain from gynecological causes: a pictorial review. Radiol Med 117:575–592 6. Bazot M, Daral E, Hourany N et al (2004) Deep pelvic endometriosis: MR imaging for diagnosis and prediction of extension of disease. Radiology 232:379–389 7. Faccioli N, Foti G, Manfredi R et al (2010) Evaluation of colonic involvement in endometriosis: double-contrast barium enema vs. magnetic resonance imaging. Abdom Imaging 35:414–421 8. Bazot M, Gasnera A, Lafonta C et al (2011) Deep pelvic endometriosis: limited additional diagnostic value of post-contrast in comparison with conventional MR images. Eur J Radiol 80:331–339 9. Landi S, Barbieri F, Fiaccavento A et al (2004) Preoperative double-contrast barium enema in patients with suspected intestinal endometriosis. J Am Assoc Gynecol Laparosc 11:223–228 10. Marchal L, Nothft MA, Coelho F, Choi K (2010) Deep pelvic endometriosis: MR imaging. Abdom Imaging 35:708–715 11. Busard MP, Van der Houwen LEE, Bleeker Maaike C et al (2012) Deep infiltrating endometriosis of the bowel: MR imaging as a method to predict muscular invasion. Abdom Imaging 37:549–557 12. Koga K, Osuga Y, Yano T et al (2003) Characteristic images of deeply infiltrating rectosigmoid endometriosis on transvaginal and transrectal ultrasonography. Hum Reprod 18:1328–1333 13. Hottat N, Larrousse C, Anaf V et al (2009) Endometriosis: contribution of 3.0-T pelvic MR imaging in preoperative assessment—initial results. Radiology 253:126–134 14. Loubeyre P, Petignat P, Jacob S et al (2009) Anatomic distribution of posterior deeply infiltrating endometriosis on MRI after vaginal and rectal gel opacification. AJR 192:1625–1631
Conflict of interest Anna Lia Valentini, Benedetta Gui, Maura Micco`, Maria Carla Mingote, Valeria Ninivaggi, Maurizio Guido,
123
