The Urethra and Its Supporting Structures in Women with Stress Urinary Incontinence: MR Imaging Using an Endovaginal Coil

Jeong Kon Kim 1 Yong Jae Kim 2 Myoung Soo Choo 2 Kyoung-Sik Cho 1 Received August 2, 2002; accepted after revision September 18, 2002. 1 Department of Radiology, Asan Medical Center, University of Ulsan, 388-1 Poongnap-dong, Songpa-gu, Seoul, 138-736, South Korea. Address correspondence to K-S Cho. 2 Department of Urology, Asan Medical Center, University of Ulsan, Seoul, 138-736, South Korea. AJR 2003;180:1037 1044 0361 803X/03/1804 1037 American Roentgen Ray Society The Urethra and Its Supporting Structures in Women with Stress Urinary Incontinence: MR Imaging Using an Endovaginal Coil OBJECTIVE. The objective of this study was to evaluate the urethra and its supporting structures in patients with stress urinary incontinence using MR imaging with an endovaginal coil. SUBJECTS AND METHODS. We reviewed MR images obtained using an endovaginal coil in 63 patients with stress urinary incontinence and in 16 continent women. We compared the two groups for the thickness of the striated muscle, smooth muscle, and mucosa submucosa of the urethra; degree of asymmetry of the puborectalis muscle; frequency of distortion in the periurethral, paraurethral, and pubourethral ligaments; degree of the vesicourethral angle; and dimension of the retropubic space. Using the status of the urethra and its supporting structures as our basis, we scored the risk of stress urinary incontinence for each woman on a scale of 0 5. RESULTS. The striated muscle layer of the urethra was thinner in the group with stress urinary incontinence (mean ± SD, 1.9 ± 0.5 mm) than that in the continent group (2.6 ± 0.4 mm) (p < 0.001). A high degree of asymmetry of puborectalis muscle (>1.5) was more frequent in the group with stress urinary incontinence (29%) than in the continent group (0%) (p = 0.015). Supporting ligaments were more frequently distorted in the incontinent group than in the continent group. Distorted periurethral ligaments were found in 56% of the patients with stress urinary incontinence versus 13% of the women who were continent; distorted paraurethral ligaments were found in 83% of the patients with stress urinary incontinence versus 19% of the women who were continent; and distorted pubourethral ligaments were found in 54% of the patients with stress urinary incontinence versus 19% of the women who were continent (p < 0.05). The group with stress urinary incontinence had a greater vesicourethral angle (148 vs 125 ) and larger retropubic space (7.5 vs 5.1 mm) than did the women who were continent (p < 0.05). The score for the risk of stress urinary incontinence was higher in the group with stress urinary incontinence (3.3 ± 1.4) than in the women who were continent (1.0 ± 1.2) (p < 0.001). CONCLUSION. MR imaging with an endovaginal coil revealed significant morphologic alterations of the urethra and supporting structures in patients with stress urinary incontinence. S tress urinary incontinence involuntary urine loss without detrusor contraction is a common social and hygienic problem of middle-aged women. In the urinary continence system, closure of the bladder outlet is one of the core processes and is chiefly provided by the urethra and its supporting structures. Therefore, functional or morphologic impairment of the urethra and its supporting structures is regarded as the major cause of stress urinary incontinence. The inadequate coaptation of the urethra (the so-called intrinsic sphincter deficiency) and the downward displacement of the urethra due to weak urethra-supporting structures (the so-called urethral hypermobility) result in urine leakage when there is an increase in intraabdominal pressure [1 3]. A myriad of treatments have been proposed for stress urinary incontinence, including pelvic floor muscle exercises, pharmacotherapy, and various surgical procedures. The expected efficacy of those treatments can be divided into two basic groups: reinforcement of the urethral coaptation and replacement of the function of the urethra-supporting structures. Therefore, precise visualization of the urethra and its supporting structures is important in selecting treatments and estimating their efficacy. Many investigators have attempted to evaluate the urethra and its supporting structures in patients with stress urinary incontinence [1 22]. However, to our knowledge, there has not been a large study comparing the morphology of the urethra and its supporting structures in incontinent and continent women. AJR:180, April 2003 1037

Kim et al. MR imaging has been widely used to evaluate various pelvic diseases because it has the advantages of multiplanar depiction and high resolution of soft tissue. Furthermore, additional use of an intracavitary coil, which is placed closer to the target organ and allows higher resolution and signal-to-noise ratio than a body coil, can provide more detailed visualization of minute structures. We believed that the application of an endovaginal coil in MR imaging would be helpful in the evaluation of the urethra and its supporting structures. The purpose of this study was to compare the morphology of the urethra and its supporting structures in patients with stress urinary incontinence with that in continent women using MR imaging with an endovaginal coil. Ultimately, we sought to assess the morphology of the urethra and its supporting structures in patients with stress urinary incontinence. Subjects and Methods Patient Population The institutional review board approved our study, and written informed consent was obtained from all patients. Between July 2000 and February 2001, 63 women with stress urinary incontinence (mean age ± SD, 54 ± 9 years) and 16 continent volunteers (mean age, 40 ± 12 years) were enrolled in our study. Using the definition of the condition devised by Sarker and Ritch [22], urologists diagnosed genuine stress urinary incontinence on the basis of the history, physical examination findings, and urodynamic evaluation of each patient. All patients with stress urinary incontinence had delivered at least one child vaginally (range, 1 5; median, 3), and seven patients (11%) had histories of previous pelvic surgery, including cesarean delivery (n = 4) and salpingo-oophorectomy (n = 3). For our population of continent women, we selected 16 women among patients who were scheduled to undergo MR imaging with an endovaginal coil for evaluation of uterine cervix cancer; in our institution, MR imaging with an endovaginal coil is a standard imaging modality for evaluation of uterine cervix cancer. The criteria for this group were that the tumor was shown on MR imaging as confined in the uterine cervix, that the patient had no history of pelvic surgery or radiation, and that a urologist confirmed urinary continence on the basis of the patient s medical history and findings at physical examination. All patients of this group also had had at least one vaginal delivery (range, 1 4; median, 2). Endovaginal Coil Design and MR Imaging Examination The endovaginal coil consisted of a fixed, tuned, elliptical copper ring made in the Asan Institute for Life Sciences (Seoul, Korea) (Fig. 1). The diameter of the coil ring was 2.5 5.5 cm, and the thickness of the coil wire was 0.3 cm. The coil, wrapped in a supple rubber, was inserted into the vagina so that the entire urethra could be imaged. In the women who were continent, after MR imaging for the uterine cervix cancer was completed, the coil was relocated to the same level as the coil in the group with stress urinary incontinence. All examinations were performed on a 1.5-T MR system (CV/i; General Electric Medical Systems, Milwaukee, WI). Off-axis axial and sagittal T2-weighted fast spin-echo sequences (TR/TE, 3800/95; echo-train length, 9) were obtained in planes perpendicular and parallel to the long axis of the urethra. Sixteen signals in each plane were obtained with a 3-mm section thickness without an interslice gap. The field of view was 160 160 mm, and the matrix size was 512 256. MR Image Analysis Fig. 1. Photograph and diagram of endovaginal coil. Coil, wrapped in supple rubber, was inserted into vagina so that entire urethra was covered. Inset diagram shows dimensions of coil. Two radiologists who were unaware of the clinical data independently interpreted the MR images in random order. MR images were evaluated at a PACS (picture archiving and communication system) (Radpia; Hyundai Information & Technology, Seoul, Korea). To ensure that the reviewers were unaware of whether the patients were in the incontinent or continent group, we did not include the MR images of the uterine cervix in the set of images reviewed by the radiologists, and no information regarding the uterine cervix cancer was given. If the two radiologists had differing interpretations of an image, the final judgment was reached by consensus. Measurement of a given structure was obtained by one radiologist at the anatomic site that had been determined by consensus of the two radiologists. We evaluated the urethra and its supporting structures the periurethral, paraurethral, and pubourethral ligaments as well as the puborectal sling. In addition, the vesicourethral angle and dimension of the retropubic space were measured. To standardize the measurements and decisions, we determined the location of a given structure on the basis of the percentile of the length of the urethra in which the structure appeared; the internal urethral meatus was considered as the zero point, and the external meatus as the 100th percentile. The urethra histologically consists of the outer striated muscle, inner smooth muscle, and mucosa submucosa layers [4 8]. On MR images, the striated muscle layer is observed as an outer ring of low signal intensity, the smooth muscle layer as a ring of middle intermediate signal intensity, and the mucosa submucosa layer as a central zone of high signal intensity [5 9] (Fig. 2). On axial MR images, we measured the thickness of these three layers at the level of approximately the 30th percentile of the urethra. The puborectal sling was also evaluated on axial MR images. We measured the thickness of the bilateral limbs of the puborectal sling in the 4- and 8-o clock directions from the urethra at the level of approximately the 50th percentile of the urethra (Fig. 2C). Thereafter, the mean thickness (thickness of the thicker limb plus thickness of the thinner limb divided by two) and the degree of asymmetry (thickness of the thicker limb divided by thickness of the thinner limb) were calculated. The periurethral, paraurethral, and pubourethral ligaments were also evaluated on axial MR images. If these ligaments maintained tightness and could be traced along the entire length without discontinuity, we considered them to be normal. On the other hand, if any discontinuity or fluttering of these ligaments was seen, we considered them to be distorted. The periurethral and paraurethral ligaments were evaluated at the level of approximately the 30th percentile of the urethra on axial MR images. The periurethral ligament was observed as a thin hypointense structure originating from the medial aspects of the puborectal sling and coursing ventrally to the urethra [8, 9] (Fig. 2A). The paraurethral ligament appeared as a slightly oblique, hypointense thin structure connecting the lateral wall of the urethra to the periurethral ligament [8, 9] (Fig. 2A). The pubourethral ligament was evaluated at the level of approximately the 50th percentile of the urethra. This ligament appeared as a hypointense structure connecting the lateral aspect of the urethra and the arcus tendineus fasciae pelvis [10 14] (Fig. 2C). The vesicourethral angle was evaluated on sagittal MR images. Two lines were drawn, one through the long axis of the urethra and one parallel to the bladder base. The intersection of these lines determined the vesicourethral angle (Fig. 2E). We also used sagittal MR images to measure 1038 AJR:180, April 2003

MR Imaging of the Urethra A C B D Fig. 2. Normal anatomy of urethra and its supporting structures in 34-year-old continent woman. A, Axial T2-weighted fast spin-echo MR image obtained at level of approximately 30th percentile of urethra shows normal striated muscle, smooth muscle, and mucosa submucosa layers of urethra. Normal periurethral and paraurethral ligaments, which maintain their tightness and can be traced along entire length without discontinuity, are shown. Symmetric puborectal sling is also noted. B, Drawing corresponds to A. SP = symphysis pubis, V = vagina, C = endovaginal coil, R = rectum, pr = puborectal sling, pa = paraurethral ligament, pe = periurethral ligament, st = striated muscle layer, sm = smooth muscle layer, m = mucosa submucosa layer. C, Axial MR image obtained at approximately 50th percentile of urethra shows normal pubourethral ligament connecting urethra to arcus tendineus fasciae pelvis and caudal portion of periurethral ligament. Thickness of puborectalis sling, measured in 4- and 8o clock directions from urethra, is 4.2 mm in right limb and 3.8 mm in left limb. Therefore, mean thickness of puborectalis muscle is 4.0 mm and degree of asymmetry is 1.1. D, Drawing corresponds to C. SP = symphysis pubis, V = vagina, C = endovaginal coil, R = rectum, ATFP = arcus tendineus fasciae pelvis, pu = pubourethral ligament, pe = periurethral ligament, pr = puborectal sling. E, Sagittal T2-weighted fast spin-echo MR image obtained through long axis of urethra shows 117 vesicourethral angle (ag, intersecting lines) and retropubic space measuring 2.2 mm (r, arrows). BLD = urinary bladder, SP = symphysis pubis. E AJR:180, April 2003 1039

Kim et al. TABLE 1 Urethral Layer (mm) Mean Thickness a of Three Urethral Layers in Women With and Without Stress Urinary Incontinence Stress Urinary Incontinence Present Yes (n = 63) Note. NS = not statistically significant. a Mean ± SD. No (n = 16) Striated muscle 1.9 ± 0.5 2.6 ± 0.4 <0.001 Smooth muscle 3.3 ± 0.5 3.4 ± 0.6 NS Mucosa submucosa 2.5 ± 0.5 2.6 ± 0.6 NS the dimension of the retropubic space from the posterior wall of the symphysis pubis to the anterior urethral wall at approximately the 50th percentile of the urethra (Fig. 2E). Data Analysis To compare the data for the two groups of patients, we applied the Student s t test to evaluate differences in the thickness of each layer of the urethra, the mean thickness and degree of asymmetry of the puborectal sling, the vesicourethral angle, and the dimension of the retropubic space. We evaluated the interobserver agreement (κ statistics) for judging the distortion of the periurethral, paraurethral, and pubourethral ligaments; a κ value of less than 0.20 was considered to indicate poor agreement; 0.21 0.40, fair agreement; 0.41 0.60, moderate agreement; 0.61 0.80, good agreement; and 0.81 1.00, excellent agreement. The chi-square test was used to compare the frequency of distortion of these ligaments as well as the frequency of a degree of asymmetry in the puborectal sling exceeding 1.5. On the basis of the morphologic status of the urethra and its supporting structures, we scored the risk of stress urinary incontinence. For the striated muscle of the urethra, we assigned a score of 0 if the thickness was equal to or greater than 2.0 mm (which was the 50-percentile value for all subjects) and 1 if the thickness was less than 2.0 mm. For the asymmetry of puborectal sling, the score assigned was 0 if the degree of asymmetry was less than 1.5 and 1 if the angle was greater than 1.5. A separate score was determined TABLE 2 p for each of the ligaments (periurethral, paraurethral, and pubourethral); 0 indicated a normal appearance and 1 indicated distortion of the ligament. By integrating the scores for all parameters, we determined a score for the risk of stress urinary incontinence, ranging from 0 to 5. The Student s t test was used to compare the score between the two groups. To evaluate the diagnostic validity of this scoring system in differentiating between the group with stress urinary incontinence and the group without the condition, we obtained the receiver operating characteristic curves. Results Incidence of Distorted Urethra-Supporting Ligaments in Women With and Without Stress Urinary Incontinence Stress Urinary Incontinence Present Distorted Ligament Yes (n = 63) No (n = 16) p No. % No. % Periurethral 35 56 2 13 < 0.01 Paraurethral 52 83 3 19 <0.001 Pubourethral 34 54 3 19 <0.001 The length and width of the endovaginal coil were sufficient for the evaluation of the urethra throughout its entire length. The striated muscle, smooth muscle, and mucosa submucosa layers were easily identified in all patients. Table 1 shows the mean thickness of each layer of the urethra in the stress-urinaryincontinent and the continent groups. The striated muscle layer was significantly thinner in the group with stress urinary incontinence (range, 1.0 2.8 mm) than in the patients who were continent (range, 2.2 3.4 mm) ( p < 0.001), although in the other layers, similar thickness was found in the two groups (p > 0.05) (Table 1 and Figs. 2A and 3A). The thickness of the thicker limb of the puborectalis muscle was 5.0 ± 1.8 mm in the group with stress urinary incontinence and 5.1 ± 1.4 mm in the continent group (p = 0.904), and the thickness of the thinner limb was 3.8 ± 1.6 mm in the group with stress urinary incontinence and 4.5 ± 1.6 mm in the continent group (p = 0.134). The mean thickness of the puborectal sling was also similar in the group with stress urinary incontinence (4.4 ± 1.6 mm) and in the continent group (4.8 ± 1.5 mm) (p = 0.408) (Figs. 2C and 3B). The mean degree of asymmetry of the puborectal sling was not significantly different between the two groups (group with stress urinary incontinence, 1.4 ± 0.5; continent group, 1.1 ± 0.1) (p = 0.055). However, we found a difference in the distribution: in the group with stress urinary incontinence, the frequency of a degree of asymmetry exceeding 1.5 was greater (n = 18, 29%) than in the continent group (n = 0) (p = 0.015) (Figs. 4 and 5). The periurethral, paraurethral, and pubourethral ligaments were more frequently distorted in the group with stress urinary incontinence than in the continent group (p = 0.01 for the periurethral ligament, and p < 0.001 for the paraurethral and pubourethral ligaments) (Table 2 and Figs. 2 and 3). Distortion of at least one ligament was more frequently noted in the group with stress urinary incontinence (n = 56, 89%) than in the continent group (n = 5, 31%) (p < 0.001). Distortion of all three ligaments was also more frequent in the stress-urinary-incontinent group (n = 21, 33%) than in the continent group (n = 0) (p < 0.001). Interobserver agreement between the two reviewers for evaluation of the periurethral, paraurethral, and pubourethral ligaments was generally good: the kappa value was 0.796 for the paraurethral ligament, 0.731 for the periurethral ligament, and 0.743 for the pubourethral ligament. The two radiologists disagreed on the evaluation of periurethral ligaments in seven (10%) of 70 patients, paraurethral ligaments in six patients (9%), and pubourethral ligaments in nine patients (13%). Those 22 cases of disagreement occurred in judgments on whether the ligaments were normal or fluttering (n = 20) and whether the ligaments were normal or discontinuous (n = 2). The distribution and receiver operating characteristic curve for the score of the risk of stress urinary incontinence are illustrated in Figures 6 and 7. The area under the curve (A z value) for this scoring system in the differentiation between the group with stress urinary incontinence and the continent group was 0.930 (95% confidence interval, 0.850 0.975). The risk score was significantly higher in the group with stress urinary incontinence (3.8 ± 1.3) than in the continent group (0.8 ± 1.1) (p < 0.001). In the group of 16 continent women, the risk score was 0 in seven (44%) and equal to or greater than 2 in three women (19%). In contrast, in the group of 63 women who had stress urinary incontinence, no patient had a score of 0, and 44 patients (70%) had a risk score equal to or greater than 2. With the score of 1 as the cutoff, the sensitivity of the risk score in differentiating the two groups was 81.3%, and its specificity was 95.2% (Fig. 6). The vesicourethral angle was greater in the group with stress urinary incontinence (148 ± 13 ) than in the continent group (122 ± 11 ) 1040 AJR:180, April 2003

MR Imaging of the Urethra ( p < 0.001) (Figs. 2E and 3C). The retropubic space was also greater in the group with stress urinary incontinence (7.5 ± 1.6 mm) than in the continent group (5.1 ± 1.1 mm) ( p < 0.001) (Figs. 2E and 3C). Discussion Several methods have been used to treat stress urinary incontinence [23 26]. Basically, the goal of treatment of stress urinary incontinence is reinforcement of the urethra and its supporting structures, which are the main functional elements in the urinary continence mechanism. Therefore, direct visualization of the morphology of theses structures is important in deciding treatment options. For example, pelvic floor exercises, a commonly applied conservative treatment, can contribute to controlling the urethral hypermobility by strengthening the pelvic floor muscles; however, this method is less effective in patients with intrinsic sphincter deficiency. Our results showed that the striated muscle layer was significantly thinner in the group with stress urinary incontinence than in the continent group, although the other layers of the two groups had similar thickness. The striated muscle layer of the urethra functions as a sphincter and provides more than 50% of the static urethral resistance. With aging, the striated muscle A undergoes marked morphologic changes; the volume decreases and the muscle bulk is replaced by connective tissue [3 6]. These changes are closely related to stress urinary incontinence because the striated muscle atrophy leads to a drop in intraurethral pressure [5]. The role of the asymmetry of the puborectal sling has been the source of some confusion and controversy. Although many authors have emphasized the importance of asymmetry of the puborectal sling as a contributing factor in stress urinary incontinence, some studies have found little difference in the asymmetry of this structure between stress-urinary-incontinent and continent women [12 14]. In our study, B Fig. 3. Morphologic alteration of urethra and its supporting structures in 57year-old woman with stress urinary incontinence. A, Axial T2-weighted fast spin-echo MR image obtained at level of approximately 30th percentile of urethra shows that striated muscle layer of urethra (arrowheads) is thin compared with normal muscle layer seen in woman imaged in Figure 2A. Fluttering periurethral ligament (straight arrows) and discontinuous paraurethral ligament (curved arrows) are evident. R = rectum, SP = symphysis pubis, V = vagina. B, Axial MR image obtained at level of approximately 50th percentile of urethra shows discontinuous pubourethral ligament (arrows). R = rectum, SP = symphysis pubis, V = vagina. C, Sagittal T2-weighted fast spin-echo MR image shows vesicourethral angle of 152 and retropubic space of 11 mm. BLD = urinary bladder, SP = symphysis pubis. C AJR:180, April 2003 1041

Kim et al. Degree of Asymmetry 3.5 3.0 2.5 2.0 1.5 1.0 Continent Group Group with Stress Urinary Incontinence Fig. 4. Dot graph illustrates different distribution of degree of asymmetry of puborectal sling in continent group and in group with stress urinary incontinence. Most patients in both groups showed degree of asymmetry of less than 1.5. However, patients with degree of asymmetry equal to or greater than 1.5 are found only in group with stress urinary incontinence. Risk Score 6 5 4 3 2 1 0 Continent Group Group with Stress Urinary Incontinence the group with stress urinary incontinence had high-degree asymmetry in puborectal sling (degree of asymmetry > 1.5) more frequently than did the continent group, although the mean degree of asymmetry was similar for the two groups. However, high-degree asymmetry was only specific, not sensitive, for differentiating between the two groups; only 29% of patients with stress urinary incontinence showed high-degree asymmetry. The main function of the urethra-supporting ligaments is to anchor the urethra to the structures of the pelvic wall such as puborectalis muscle or arcus tendineus fasciae pelvis. As described by Tan et al. [9], the periurethral and paraurethral ligaments link the proximal urethra to the puborectal sling. The connection between the proximal urethra and the puborectal sling provides a vital framework for urethral immobilization against downward force exerted by increased abdominal pressure; contraction of the puborectal sling results in the elevation and constriction of the urethrovesical neck by means of its attachment to the proxi- Fig. 5. Axial T2-weighted fast spin-echo MR image obtained at level of approximately 50th percentile of urethra shows asymmetric puborectal sling (arrows) in 49-year old woman with stress urinary incontinence. SP = symphysis pubis, U = urethra, V = vagina, R = rectum. True-Positive Rate 100 80 60 40 20 0 0 20 40 False-Positive Rate 60 80 100 Fig. 6. Dot graph reveals different distribution of scores for risk of stress urinary incontinence between continent group and group with stress urinary incontinence. Cutoff value = 1, sensitivity = 81.3%, and specificity = 95.2%. Fig. 7. Graph shows receiver operating characteristic curve for risk score of stress urinary incontinence. Area under curve = 0.930 (95% confidence interval, 0.850 0.975). 1042 AJR:180, April 2003

MR Imaging of the Urethra mal urethra [3, 9]. In our study, distortion of the periurethral and paraurethral ligaments was frequently noted in the patients with stress urinary incontinence, suggesting that a defect of connection between the urethra and the puborectal sling is one of the principal causes of urethra hypermobility. The pubourethral ligament runs anterolaterally from approximately the 20th to the 60th percentile of the urethra and attaches at the arcus tendineus fasciae pelvis [2, 6, 14]. Together with the anterior vaginal wall, the pubourethral ligament contributes to the hammock structure, first suggested by DeLancey [13]. The hammock structure is formed by the lateral attachment of the urethra and vagina to the pelvic side wall and provides stabilization of the urethra against increased abdominal pressure. Many authors [1, 7, 10, 11 15] have reported that a deficiency in the hammock structure is closely related to stress urinary incontinence; our results confirm these previous reports because we found that the pubourethral ligament was more frequently distorted in the patients with stress urinary incontinence. The vesicourethral angle and the dimension of the retropubic space are closely related to the urethra-supporting structures. Contraction of the puborectal sling elevates the bladder neck and decreases the vesicourethral angle, and normally the hammock maintains the angle by sustaining the urethra toward the symphysis pubis [3, 9]. A damaged hammock may cause posterior displacement of the urethra, leading to an increase in the retropubic space. Therefore, we suggest that defective urethra-supporting structures may increase the vesicourethral angle and the retropubic space. In spite of statistically significant differences, we also found some overlap in the thickness of the striated muscle layer of the urethra and the degree of asymmetry in the puborectalis muscle between the group with stress urinary incontinence and the continent group; the standard deviation in the thickness of the striated muscle layer was large and the mean degree of asymmetry in the puborectalis sling was similar between the two groups. We suggest that the urethra and its supporting structures work as a consolidated unit, not as separate organs. The morphologic status of each structure may vary, even among patients with stress urinary incontinence. For example, a patient could have normal thickness in the striated muscle in the urethra but could also have severely distorted urethra-supporting ligaments and could experience stress urinary incontinence. For this reason, we applied a scoring system for the risk of stress urinary incontinence that integrated the status of all the functional elements in the urinary continence mechanism. The risk score in our study showed high diagnostic validity in differentiating the patients with stress urinary incontinence from the patients who were continent. The A z value was 0.930, the sensitivity was 95.2%, and the specificity was 81.3%. Many coil systems, such as the phased array coil and endovaginal coil, have been used to improve the quality of MR imaging in depicting the pelvic floor structures. In this study, we used an endovaginal coil because we expected that this coil system could provide a more detailed depiction of minute structures (although a recent report showed that MR imaging with only the phased array coil also could depict ligamentous structures in the pelvis [27]). The difference between our study and previous studies that used an endovaginal coil is that we did not use a hard cylindrical coil-holder but rather a supple rubber wrap. We were concerned about the possible risk of a hard coil-holder widening the flat vaginal lumen and displacing the urethra and its supporting structures; the morphology of minute structures may be easily affected by even a light compression. A major limitation of our study is the lack of matching for various patient factors such as age, body weight, and hormonal status between the patients in the two groups. The group with stress urinary incontinence (mean age, 54 years) was significantly older than the continent group (mean age, 40 years), and this age gap might partly have contributed to the morphologic difference in the urethra and its supporting structures between the two groups. Obesity and hormonal status are also important risk factors for stress urinary incontinence. Being overweight may increase intraabdominal pressure and the dimension of the retropubic space, and estrogen affects the urinary tract at multiple levels, including epithelial, connective, muscular, and vascular tissues of the urethra and bladder. Another potential limitation of our study is that MR imaging was performed with the patient resting in a supine position; therefore, dynamic changes of the urethra, retropubic space, and vesicourethral angle under increased abdominal pressure could not be evaluated in a more natural position. Our study is also potentially limited by the nomenclature. Confusion and controversy on the definition and nomenclature for the urethra-supporting ligaments still exist. We based our study on the work of Tan et al. [9], who first described the periurethral and paraurethral ligaments, because we also confirmed the presence of these structures on MR imaging in most continent women. However, there is still a lack of supporting studies for this concept, and some authors still used the term endopelvic fasciae for these structures [27]. We hope that in the future, serious discussions based on high-quality images will result in a true consensus on the definition and nomenclature for these minute structures. In conclusion, MR imaging using an endovaginal coil shows significant morphologic alterations of the urethra and supporting structures in women with stress urinary incontinence. This imaging modality may contribute to evaluating the cause of disease and to planning treatment in patients with stress urinary incontinence. References 1. Klutke CG, Golomb J, Bararic Z, Shlom R. The anatomy of stress incontinence: magnetic resonance imaging of the female bladder neck and urethra. J Urol 1990;143:563 566 2. Klutke CG, Siegel CL. Functional female pelvic anatomy. Urol Clin North Am 1995;22:487 497 3. Strohbehn K. Normal pelvic floor anatomy. Obstet Gynecol Clin North Am 1998;25:683 705 4. Colleselli K, Stenzl A, Eder R, Strasser H, Poisel S, Bartsch G. The female urethral sphincter: a morphological and topographical study. J Urol 1998;160:49 54 5. Strohbehn K, Quint LE, Prince MR, et al. Magnetic resonance imaging anatomy of the female urethra: a direct histologic comparison. Obstet Gynecol 1996;88:750 756 6. Carlile A, Davies I, Rigby A, Brocklehurst JC. Age changes in the human female urethra: a morphometric study. J Urol 1988;139:532 535 7. Cruikshank SH, Kovac SR. The functional anatomy of the urethra: role of the pubourethral ligaments. Am J Obstet Gynecol 1997;176:1200 1205 8. Stoker J, Halligan S, Bartman CI. Pelvic floor imaging. Radiology 2001;218:621 641 9. Tan IL, Stoker J, Zwamborn AW, Entius KAC, Calame JJ, Lameris JS. Female pelvic floor endovaginal MR imaging of normal anatomy. Radiology 1998;206:777 783 10. Aronson MP, Susan M, Jacoby AF, Chelmow D, Sant G. Periurethral and paravaginal anatomy: an endovaginal magnetic resonance imaging study. Am J Obstet Gynecol 1995;173:1702 1710 11. DeLancey JO, Starr RA. Histology of the connection between the vaginal and levator ani muscles: implication for urinary tract function J Reprod Med 1990;35:765 771 12. Strohbehn K, Ellis JH, Strohbehn JA, DeLancey JO. Magnetic resonance imaging of the levator ani with anatomic correlation. Obstet Gynecol 1996;87:277 285 AJR:180, April 2003 1043

Kim et al. 13. DeLancey JO. Structural support of the urethra as it related to stress urinary incontinence: the hammock hypothesis. Am J Obstet Gynecol 1994;170:1713 1723 14. DeLancey JO. Correlative study of paraurethral anatomy. Obstet Gynecol 1986;68:91 97 15. Nichols DH, Milley PS. Identification of pubourethral ligaments and their role in transvaginal surgical correction of stress incontinence. Am J Obstet Gynecol 1973;68:123 125 16. DeLancey JO. Structural aspects of the extrinsic continence mechanism. Obstet Gynecol 1988;72:296 301 17. Milley PS, Nichols DH. The relationship between the pubourethral ligaments and the urogenital diaphragm in the human female. Anat Rec 1996;163:433 451 18. Zivkovic F, Tamussino K, Hass J. Contribution of the posterior compartment to the urinary continence mechanism. Obstet Gynecol 1998;91:22 233 19. Unterweger W, Marincek B, Gottstein-Aalame N, et al. Ultrafast MR imaging of the pelvic floor. AJR 2001;176:959 963 20. Fielding JR, Versi E, Mulkern RV, Lerner MH, Griffiths DJ, Jolesz FA. MR imaging of the female pelvic floor in the supine and upright positions. J Magn Reson Imaging 1996;6:961 963 21. Fielding JR, Griffiths DJ, Versi E, Mulkern RV, Lee MT, Jolesz FA. MR imaging of pelvic floor continence mechanisms in the supine and sitting positions. AJR 1998;171:1607 1610 22. Sarker PK, Ritch AE. Management of urinary incontinence. J Clin Pharm Ther 2000;25:251 263 23. Haab F, Zimmern PE, Leach GE. Female stress urinary incontinence due to intrinsic sphincteric deficiency: recognition and management. J Urol 1996;156:3 17 24. Portera JC, Summitt RL Jr. Common operations for stress urinary incontinence: selecting the correct operation. Clin Obstet Gynecol 1998;41:712 718 25. Lightner DJ, Itamo NM. Treatment options for women with stress urinary incontinence. Mayo Clin Proc 1999;74:1149 1156 26. Richardson AC, Edmonds PB, Williams NL. Treatment of stress urinary incontinence due to paravaginal fascial defect. Obstet Gynecol 1981; 57:357 362 27. Tunn R, Delancy JO, Quint EE. Visibility of pelvic organ support system structures in magnetic resonance images without an endovaginal coil. Am J Obstet Gynecol 2001;184:1156 1163 The full text and images from the American Journal of Roentgenology may also be viewed online at www.arrs.org or www.ajronline.org. 1044 AJR:180, April 2003