JOURNAL OF MAGNETIC RESONANCE IMAGING 9:373 377 (1999) Original Research Pelvic Floor Descent in Females: Comparative Study of Colpocystodefecography and Dynamic Fast MR Imaging D. Vanbeckevoort, MD, 1 * L. Van Hoe, MD, 1 R. Oyen, MD, 1 E. Ponette, MD, 1 D. De Ridder, MD, 2 and J. Deprest, MD 3 The purpose of this study was to compare fast dynamic magnetic resonance imaging (MRI) with colpocystodefecography (CCD) in the evaluation of pelvic floor descent in women. Thirty-five women with clinical evidence of pelvic floor descent were studied. A fast single-shot MR sequence was performed in the supine position during pelvic floor relaxation and during maximal pelvic strain. On the same day, a dynamic CCD was performed with the patient seated on a stool-chair. The degree of descent of the bladder, vagina, and anorectal junction was evaluated as the vertical distance between the pubococcygeal line and the bladder base, the vaginal vault, and the anorectal junction, respectively. A bulge of more than 3 cm measured as the distance between the extended line of the anterior border of the anal canal and the tip of the rectocele was interpreted as a rectocele. MRI was compared with CCD during maximal pelvic strain (CCD I) and during voiding and defecation (CCD II). CCD was considered as the gold standard. Compared with clinical examination, CCD I showed a larger number of involved compartments, except for the middle compartment. CCD II was superior to clinical examination in all cases. In comparison with CCD I and especially CCD II, MRI had a lower sensitivity, especially for the anterior and middle compartment. Even four enteroceles seen on CCD II were not detected by MRI. When CCD I and CCD II were compared, a cystocele, a vaginal vault prolapse, an enterocele, and a rectocele were more readily seen on CCD II than with CCD I. When compared with CCD, supine dynamic MRI is unreliable, especially in the anterior and middle compartment. Even in the detection of enteroceles CCD was superior to MRI. In general, the best results with MRI can be expected for evaluation of the rectum. J. Magn. Reson. Imaging 1999;9:373 377. 1999 Wiley-Liss, Inc. Index terms: pelvic organs, abnormalities; pelvic organs, MR 1 Department of Radiology, University Hospitals, B-3000 Leuven, Belgium. 2 Department of Urology, University Hospitals, B-3000 Leuven, Belgium. 3 Department of Obstetrics and Gynecology, University Hospitals, B-3000 Leuven, Belgium. *Address reprint requests to: D.V., Department of Radiology, U.Z. Gasthuisberg, Herestraat 49, B-3000 Leuven, Belgium. Received July 6, 1998; Accepted October 20, 1998. 1999 Wiley-Liss, Inc. 373 THE PREVALENCE OF DEFECTS of pelvic support in older multipara women, especially after hysterectomy, is reported to be as high as 16% (1). In many centers, physical examination is the method of choice for evaluating pelvic descent. This method is not always reliable, is difficult to quantify, and is prone to significant interobserver variability (2). Since Bethoux and Bory and colleagues (3) introduced colpocystodefecography (CCD), a variety of imaging techniques have become available to physicians caring for women with pelvic support defects. The development of fast sequences in MRI provides a new approach for the simultaneous evaluation of pelvic visceral prolapse and movements. The purpose of this study was to assess the accuracy of MRI in the evaluation of pelvic descent. MRI was compared with CCD, the latter considered as the gold standard. In the current literature, only CCD at maximal strain has been used as a reference (4,5), although the pelvic floor reaches its maximal state of relaxation only during defecation and voiding. Therefore, in this study, CCD during voiding and defecation was used as the gold standard. MATERIALS AND METHODS Subjects Thirty-five women (mean age 65.4 years; range 44 83 years) with clinical evidence of pelvic floor descent were included in the study. Nine patients had had a hysterectomy. All patients were examined clinically by an experienced gynecologist (J.D.). Two patients were excluded because of their inability to strain adequately. All patients gave informed consent after explanation of the complete examination procedure. Imaging Techniques All MR imaging studies were performed on a 1.5 T system (Magnetom Vision, Siemens Medical Systems, Erlangen, Germany) with a gradient switching capability of 25 mt/m in a rise time of 600 msec. All patients were imaged with a body-phased-array receiver coil. After an initial localizer in three different planes, half- Fourier single-shot turbo spin-echo (HASTE) images were obtained in the sagittal plane during pelvic floor relaxation and during maximal pelvic strain. The rectum was filled with 100 ml of aqueous sonographic gel. No opacification of the bladder, the vagina, or small bowel was used. The HASTE sequence is a T2-weighted acquisition in which all radiofrequency (RF) refocused echoes are
374 Vanbeckevoort et al. obtained after a single excitation. The following parameters were used: time interval between subsequent echoes 4.2 msec; effective TE 60 msec; flip angle 160 ; number of excitations 1; matrix 160 256. Only half of the k-space was measured (echo train length 88); the k-space was then expanded with the half-fourier method to 160 lines. Slice thickness was 5 mm, and 20 slices were obtained in one acquisition with a distance between the measured slices of 5 mm and during quiet breathing. The field of view (FOV) was 300 320 mm (with rectangular FOV 6/8 if possible) and receiver bandwidth was 650 Hz/pixel. On the same day, a dynamic CCD was performed with the patient seated on a stool-chair. Opacification included the bladder (Telebrix), the vagina (Hytrast), and the rectum (barium). The small bowel was opacified by a barium meal 90 minutes prior to the CCD. Lateral images (100 mm camera) were obtained with conventional X-ray equipment (Diagnost 75, Philips-Fluorospot Siemens) at rest, during maximal pelvic strain, and during voiding and defecation. The process of voiding and defecation was recorded on videotape. Image Analysis All personal information was removed from the radiological images. The images were then independently assessed by two experienced observers (D.VB. and L.V.H). The line extending from the most inferior portion of the symphysis pubis to the tangent of the sacro-coccygeal joint (pubococcygeal line) acted as a reference line on both CCD and MRI (5). The diagnosis of descent of the bladder, vagina, and rectum was based on measurement of the vertical distance between the pubococcygeal line and the bladder base, the vaginal vault, and the anorectal junction, respectively. The advantage of this reference line is that it can easily be drawn and reproduced and that it is independent of the pelvic tilt. Furthermore, components of the pelvic floor support, such as the pubococcygeal muscle, puborectal muscle, and pubovesical ligament, attach along this line. Cystocele was defined as descent of the bladder base below the pubococcygeal line. Rectal descent was diagnosed when the anorectal junction was located more than 2.5 cm below the pubococcygeal line. Enterocele was diagnosed when bowel loops or peritoneal fat were interposed between the upper vagina and adjacent rectum, or when bowel loops prolapsed below the pubococcygeal line. Rectocele was defined as an outpouching of the anterior wall more than 3 cm during straining down (6). Data Analysis Data evaluation was done as follows. The two observers independently determined the presence or absence of descent in each compartment on both MRI and CCD. In case of disagreement the final diagnosis was made by consensus (Table 1). Statistical analysis was performed on these data. RESULTS Clinical Examination vs. CCD I and vs. CCD II In comparison with CCD I, three cystoceles, one enterocele, and six rectal descents were missed by clinical Table 1 Percentages of Agreement and Disagreement in Each Compartment* Diagnosis Agreement Disagreement Cystocele 100 (34/34) 0 (0/0) Vaginal vault 91 (21/23) 9 (2/23) Enterocele 100 (9/9) 0 (0/0) Descent rectum 97 (32/33) 3 (1/33) *Absolute numbers are between slashes (in parentheses). examination. In 14 patients, however, clinical examination revealed an additional descent of the middle compartment, not seen on CCD. The high number of falsepositive results of clinical examination in the middle compartment (specificity 53%) when compared with CCD I is most likely explained by an imperfect gold standard. Indeed, CCD I detected only 5 of 23 vaginal vault prolapses when compared with CCD II (sensitivity of CCD I 22%). In contrast, CCD II was superior to clinical examination in all cases. Ten additional cystoceles, four descents of the middle compartment, five enteroceles, and ten rectal descents could only be ascertained by CCD II. The sensitivity and specificity of clinical examination vs. CCD I and vs. CCD II for each compartment are listed in Table 2. MRI vs. CCD I and CCD II In comparison with CCD I, MRI showed less organ descents in all compartments (Fig. 1). The diagnosis of enteroceles, however, could always be achieved with MRI: five enteroceles with small bowel loops were both detected by MRI and CCD I. A minor rectal descent in one case, seven cystoceles, and two vaginal vault prolapses were not seen at MRI. When MRI and CCD II were compared, CCD II showed more compartments involved (Figs. 2, 3). CCD II revealed either an additional cystocele (14 cases), a vaginal vault prolapse (20 cases), an enterocele (4 cases) (Fig.. 4), a rectocele (13 cases), and a rectal descent (5 cases). Sensitivity and specificity of MRI vs. CCD I and vs. CCD II split for each compartment are shown in Table 3. CCD I vs. CCD II Cystocele, vaginal vault prolapse, rectocele, and rectal descent were more readily seen on CCD II than with CCD I. Even enteroceles were best seen on CCD II. Sensitivity and specificity of CCD I vs. CCD II for each compartment are shown in Table 4. Table 2 Sensitivity and Specificity of Clinical Examination Versus CCD I and CCD II* Clinical examination Diagnosis Sensitivity vs Specificity vs CCDI CCD II CCD I CCD II Cystocele 89 (24/27) 71 (24/34) 100 (8/8) 100 (1/1) Vaginal vault 100 (5/5) 83 (19/23) 53 (30/44) 100 (12/12) Enterocele 80 (4/5) 44 (4/9) 100 (30/30) 100 (26/26) Rectal descent 79 (23/29) 70 (23/33) 100 (6/6) 100 (2/2) Data are percents, with absolute numbers of diagnoses in each compartment are between slashes (in parentheses).
Comparative Study of Pelvic Floor Descent 375 of contrast material into the vagina, small bowel, or bladder. In 1991, Yang et al (4) had already proposed dynamic fast MR imaging for the assessment of pelvic prolapse. A gradient-recalled acquisition in a steady-state pulse sequence was used in the sagittal plane. No opacification of pelvic organs was used. Delemarre et al (9) acquired one image during squeezing and pushing in patients with rectoceles and pointed out that MRI in the prone position is unsuitable for detecting rectoceles (9). Figure 1. MR image of a 59-year-old woman demonstrating a small rectocele (a). CCD during evacuation showed not only a moderate rectocele but also an additional small cystocele (b). DISCUSSION MRI has gained in value as a non-invasive imaging technique, and it challenges CCD in the diagnosis of pelvic floor descent. The advantage of MRI is that it permits complete analysis of the three pelvic compartments in a single procedure without exposure to ionizing radiation. The high signal intensity of the urinary bladder and of the gel-filled rectum allows assessment of all compartments, spaces, and adjacent structures and does not require additional application Figure 2. In a 66-year-old patient with a moderate descent of bladder and rectum on MRI (a), CCD additionally showed an enterocele (b).
376 Vanbeckevoort et al. Healy et al (10) studied patients with obstructed defecation using a fat gradient echo sequence. They demonstrate marked pelvic visceral and levator muscle descent on MRI, despite normal evacuation proctography and anorectal physiology. Goodrich el al (11) employed fast gradient-recalled acquisition at steady state (GRASS) sequences with acquisition times of 3 10 seconds per image and both sagittal and coronal spatial orientation. In this study, quantitative results showed widening of the levator hiatus and more vertical position of the levator plate postoperatively. They also confirmed the clinical relevance of dynamic MRI in analyzing and assessing pelvic floor relaxation and in understanding anatomical changes occuring before and after surgical repair of uterovaginal and vaginal vault prolapse. Figure 3. A mild cystocele and rectocele was present on both MRI (a) and CCD (b). In addition, CCD demonstrated a marked prolapse of the vaginal vault (arrowheads), not seen on the MR examination. Figure 4. Demonstration of a large enterocele on CCD (a), not seen on the MR examination (b).
Comparative Study of Pelvic Floor Descent 377 Table 3 Sensitivity and Specificity of MRI Versus CCD I and CCD II* MRI Diagnosis Sensitivity (%) vs Specificity (%) vs CCD I CCD II CCD I CCD II Cystocele 74 (20/27) 59 (20/34) 100 (8/8) 100 (1/1) Vaginal vault 60 (3/5) 13 (3/23) 100 (30/30) 100 (12/12) Enterocele 100 (5/5) 56 (5/9) 100 (30/30) 100 (26/26) Rectocele 62 (8/13) 38 (8/21) 100 (22/22) 100 (14/14) Rectal descent 97 (28/29) 85 (28/33) 100 (6/6) 100 (2/2) *Data are percents, with absolute numbers of diagnoses in each compartment between slashes (in parentheses). Table 4 Sensitivity and Specificity of CCD I Versus CCD II* CCD I Diagnosis Sensitivity Specificity Cystocele 79 (27/34) 100 (1/1) Vaginal vault 22 (5/23) 100 (12/12) Enterocele 56 (5/9) 100 (26/26) Rectocele 62 (13/21) 100 (14/14) Rectal descent 88 (29/33) 100 (2/2) *Data are percents, with absolute numbers of diagnosis in each compartment between slashes (in parentheses). Osaza et al (12) studied patients with uterine prolapse using T1- and T2-weighted spin-echo sequences. Their results support the notion that topographical changes involving the levator ani muscle and the vagina occur in association with uterine prolapse. Therefore, MRI may be exploitable for the assessment of the efficacy of surgical prolapse repair. Further studies are essential to evaluate the exact role of MRI in influencing management and predicting successful surgical outcome in patients with pelvic floor descent. This study focused on the comparison between MRI and CCD in the diagnosis of descent in each compartment. The data suggest that dynamic single-shot MRI in the supine position is not as accurate as CCD in the evaluation of pelvic floor descent. The high number of false-negative MRI studies in the anterior and middle compartment (and to a lesser extent the posterior compartment) is likely to be explained by the position in which this examination is performed. Maximal strain of the pelvic floor is expected to be less when patients are examined in the supine position (MRI) compared with a seated position (CCD). In addition, the pelvic floor muscles reach their greatest degree of relaxation during defecation. This is probably the physiologic basis for the greater sensitivity of CCD in the detection of pelvic floor defects compared with MRI. The results in the present study are contrary to those of a recent study by Lienemann et al (5). In this comparative study of 44 patients, however, MRI was clearly superior to CCD and accurately depicted pelvic floor descent and prolapse in women. A possible explanation for the good performance of MRI in this study could be the less rigorous gold standard. Indeed, physical examination and peroperative findings were used as key references. Furthermore, these authors did not compare MRI with CCD during voiding and defecation. Some preliminary studies of MRI of the pelvic floor in the erect position show posture-related quantitative significant differences and suggest that erect MRI will certainly increase the number of diagnoses of pelvic floor descent (13). The major limitation of this study is that appropriate control of the straining maneuvers of the patient is not feasible. Therefore, a) patients practiced the maneuvers prior to the MRI and CCD; b) a patient was included in the study only when the movement of the pelvic floor was at least 2 cm between rest and straining; and c) patients with less movement of the pelvic floor were excluded. In conclusion, the results show that CCD is superior to dynamic single-shot MRI in the supine position for the evaluation of pelvic floor descent. MRI in the erect position in an open system should reveal new perspectives. REFERENCES 1. Cruikshank SH. Preventing posthysterectomy vaginal vault prolapse and enterocele during vaginal hysterectomy. Am J Obstet Gynecol 1987;156:1433 1440. 2. Altringer WE, Saclarides TJ, Dominguez JM, Brubaker LT, Smith CS. Four-contrast defecography: pelvic floor-scopy. Dis Colon Rectum 1995;38:695 699. 3. Bethoux A, Bory S, Huguier M, Sheao SL. Le colpocystogramme. J Chir (Paris) 1965;8:809 828. 4. Yang A, Mostwin JL, Rosenshein NB, Zerhouni EA. Pelvic floor descent in women : dynamic evaluation with fast MR imaging and cinematic display. Radiology 1991;179:25 33. 5. Lienemann A, Anthuber A, Baron A, Kohz P, Reiser M. Dynamic MR colpocystorectography assessing pelvic-floor descent. Eur Radiol 1997;7:1309 1317. 6. Yoshioka K, Matsui Y, Yamada O. Physiologic and anatomic assessment of patients with rectocele. Dis Colon Rectum 1991;34:704. 7. Siproudis L, Ropert A, Vilotte J, Bretagne JF, Heresbach D, Raoul JL, Gosselin M. How accurate is clinical examination in diagnosing and quantifying pelvirectal disorders? A prospective study in a group of 50 patients complaining of defecatory difficulties. Dis Colon Rectum 1993;36:430. 8. Hock D, Lombard R, Jehaes C, et al. Colpocystodefecography. Dis Colon Rectum 1993;36:1015 1021. 9. Delemarre JBVM, Kruyt RH, Doornbos J, et al. Anterior rectocele: assessment with radiographic defecography, dynamic magnetic resonance imaging, and physical examination. Dis Colon Rectum 1997;37:249. 10. Healy JC, Halligan S, Reznek RH, et al. Magnetic resonance imaging of the pelvic floor in patients with obstructed defaecation. Br J Surg 1997:84:1555 1558. 11. Goodrich MA, Webb MJ, King BF, Bampton AE, Campeau NG, Riedere SJ. Magnetic resonance imaging of the pelvic floor relaxation: dynamic analysis and evaluation of patients before and after surgical repair. Obstet Gynecol 1993;82:883 891. 12. Osaza H, Mori T, Togashi K. Study of uterine prolapse by magnetic resonance imaging: topographical changes involving the levator ani muscle and the vagina. Gynecol Obstet Invest 1992;34:43 48 13. Versi E, Griffiths DJ, Fielding J, Mulkern R, Lehner M, Jolesz F. In: Proceedings of the International Continence Society 26th Annual Meeting, Athens, Greece, 1996.