Diagnostic accuracy and potential limitations of transvaginal sonography for bladder endometriosis

Authors


Abstract

Objectives

To evaluate the accuracy and the potential limitations of transvaginal sonography (TVS) in the preoperative evaluation of women with clinically suspected bladder endometriosis and to describe the sonographic features of the pathological condition in cases in which it was confirmed.

Methods

In the period between 2001 and 2006, we operated on 490 patients with clinically/sonographically suspected endometriosis. In 41 cases, bladder endometriosis was diagnosed at surgery and confirmed at histopathological examination. All patients underwent TVS in a standardized manner not more than 1 month before surgery. Findings at preoperative TVS were described and compared with those at laparoscopy in order to evaluate the sensitivity, specificity, and positive and negative predictive values of TVS.

Results

Bladder endometriosis was correctly identified at TVS in 18/41 cases (43.9%) while 23/41 (56.1%) patients had a negative preoperative sonogram. The sensitivity, specificity and positive and negative predictive values of TVS for bladder endometriosis were 44% (18/41), 100% (449/449), 100% (18/18) and 95% (449/472), respectively, and the total accuracy was 95% (467/490). The detection rate was strongly related to mean lesion diameter as measured by the pathologist (mean ± SD, 42.5 ± 22.1 mm in the nodules detected vs. 28.9 ± 14.8 mm in the nodules missed; P = 0.029) and to a history of previous surgery for endometriosis (70.6% vs. 25.0%; P = 0.005). At TVS, the nodule was hypoechogenic, its morphology was either elongated (‘comma-shaped’: 12/18, 66.7%) or spherical (6/18, 33.3%), and the site involved was the dome (11/18, 61.1%) or the base (7/18, 38.9%) of the bladder. Small anechogenic cystic areas within the nodule were seen in five of the 18 patients (27.8%) and a bright hyperechogenic rim was seen in 10 (55.6%).

Conclusions

The detection rate of bladder endometriosis by TVS depends on the size of the endometriotic nodules, with detected nodules being larger than those that were missed. A history of previous surgery for endometriosis increases the likelihood of bladder endometriosis being detected on ultrasound examination. Copyright © 2009 ISUOG. Published by John Wiley & Sons, Ltd.

INTRODUCTION

Involvement of the urinary tract occurs in approximately 1–2% of patients with endometriosis and involves the bladder in 90% of these cases1. Deep infiltrating endometriosis is a particular form of endometriosis which penetrates > 5 mm under the peritoneal surface2, 3.

The anatomical classification of deep infiltrating endometriosis proposed by Chapron et al. divides such lesions into two groups, defined by location in the anterior or posterior compartment4. Anterior deep infiltrating endometriosis corresponds to bladder endometriosis, in which the endometrial glands and stroma infiltrate the bladder muscularis propria5. We have recently demonstrated that anterior deep infiltrating endometriosis is a specific entity responsible for pelvic pain and dysuria6.

Once considered a rare pathological condition, bladder endometriosis is probably underdiagnosed owing to non-specific symptoms, often mimicking recurrent cystitis, such as dysuria, urgency, frequency, suprapubic pain, vesical tenesmus, incontinence and hematuria3, 7. The non-specific presentation and insidious onset can delay diagnosis, leading to increased morbidity and incorrect treatment7; the efficacy of surgery depends on the radicality of the operation8 and so an accurate preoperative assessment of disease extension is required for planning complete surgical excision9.

Only a few studies on a handful of cases have evaluated the role of transvaginal sonography (TVS) in the preoperative diagnosis of bladder endometriosis and they have reported conflicting results, but TVS performed better than transabdominal ultrasound imaging10–13. The aim of this retrospective study was to determine the diagnostic accuracy and potential limitations of TVS in diagnosing bladder endometriosis in a large population of women, and to describe the sonographic features of this specific location of endometriosis.

METHODS

All patients who underwent laparoscopic treatment for suspected pelvic endometriosis at the Center of Reconstructive Pelvic Endosurgery, Bologna University Hospital, between January 2001 and December 2006 were included. After retrospective analysis of the clinical, surgical, histological and sonographic information, we identified 490 cases with detailed medical records who underwent a comprehensive and documented transvaginal sonographic examination at our institution before surgery. The severity of preoperative dysuria was evaluated in all women during an outpatient visit by means of a visual analog scale14.

All TVS scans were performed not more than a month before surgery by one of three investigators (L.S., L.M., P.P.) aware of the patients' clinical and surgical history, symptoms and the results of a physical examination. Digital and/or printed images were available and were reviewed in each case. The indication for TVS was any of the symptoms suggestive of endometriosis: dysmenorrhea, dyspareunia, chronic pelvic pain, dysuria, urgency, frequency, suprapubic pain, vesical tenesmus and infertility. The ultrasound examinations were performed in a standardized manner using a Sonoline Elegra (Siemens, Erlangen, The Netherlands) ultrasound machine equipped with a 5.0–8.0-MHz vaginal probe.

First, a detailed examination of the pelvis was undertaken to evaluate the anatomy of the uterus and ovaries. The transducer was then positioned in the anterior vaginal fornix and tilted upwards to visualize the vesicouterine space and the bladder, in longitudinal and transverse sections. In these planes, the bladder wall can easily be visualized if a moderate amount of urine is present. When required, according to the examiner's judgment, a transabdominal scan with a 3.5–5.0-MHz transducer was performed.

Consensus regarding the diagnostic criteria suggestive of a bladder endometriotic nodule was reached before starting the study on the basis of the sonographic features described in the literature: the presence of a hypoechogenic or isoechogenic nodule in the bladder wall12, or a nodule with a heterogeneous echostructure containing numerous anechogenic (‘bubble-like’) areas10.

Whenever a bladder endometriotic nodule was suspected at TVS, its location, shape, mean diameter, echogenicity, mobility and associated pain on pressure with the probe were recorded. Moreover, vascularization of the nodule was evaluated by means of power Doppler imaging, with a pulse repetition frequency of 500 Hz and a wall filter of 50 Hz, to detect low-velocity blood flow.

All 490 women underwent laparoscopic treatment as described previously6. In 31 cases conversion to laparotomy was needed to complete the intervention. In the presence of a lesion involving the bladder, complete excision of the nodule was attempted. In the case of a solitary implant of the vesical dome, appearing as a protruding hard fixed nodule, a partial-thickness laparoscopic resection of the muscularis was performed, sparing the unaffected mucosa if possible. When base involvement was documented intraoperatively, with fusion of the bladder to the uterine isthmus owing to retroperitoneal fibrosis, full-thickness dissection of the vesicouterine space with separation of the bladder from the uterus was performed followed by a partial cystectomy. Data regarding the definitive location of the endometriotic lesion, either base or dome, were collected from surgical reports. The diagnosis of bladder endometriosis was confirmed at histopathology when endometrial glands and stroma were found infiltrating the bladder muscularis propria5. The pathologist responsible for the histological examination after surgical removal measured three orthogonal diameters of the nodule three times using a ruler and calculated the mean.

Statistical analysis

All continuous variables are expressed in terms of mean ± SD and range. All categorical variables are expressed in terms of frequency and percentage.

For continuous variables non-parametric tests were used owing to the non-normal distribution of the variables as assessed using the Kolmogorov–Smirnov test. The statistical significance of the differences between patients with detected vs. missed endometriotic bladder nodules was determined using the Mann–Whitney U-test. The mean diameters of the nodules at TVS and at histological examination were compared using the paired Wilcoxon test.

Pearson's chi-square test, calculated using the Monte Carlo method for small samples, was used to investigate differences in categorical characteristics between patients with detected and those with missed endometriotic bladder nodules. Sensitivity, specificity, and positive and negative predictive values were calculated to investigate the accuracy of TVS in predicting bladder endometriosis. For all tests, P < 0.05 was considered significant. Statistical analysis was carried out using SPSS version 14.1 (SPSS Inc., Chicago, IL, USA).

RESULTS

Of the 490 women who satisfied our inclusion criteria, 41 had bladder endometriosis diagnosed at surgery and confirmed at histology. The clinical characteristics of the patients are shown in Table 1. The overall mean size of the bladder endometriotic nodules at histological examination was 34.88 ± 19.35 (range, 10–92) mm; at surgery the lesion was located in the bladder base in 17/41 patients (41.5%) and in the bladder dome in 24/41 patients (58.5%).

Table 1. Characteristics of the study population
ParameterMean ± SD or n (%)
  1. R-AFS, Revised American Fertility Society classification of endometriosis (1985)17.

Number of patients490
Age (years)33.08 ± 5.97
Previous surgery for endometriosis140 (28.6)
Previous surgery (other than for endometriosis) 
 Laparoscopic myomectomy18 (3.7)
 Cesarean section15 (3.1)
Endometriotic location at surgery467 (95.3)
 Ovarian250 (51.0)
 Posterior compartment160 (32.7)
 Ureteral (unilateral or bilateral)18 (3.7)
 Superficial peritoneal implants306 (62.4)
 Bladder41 (8.4)
Endometriosis stage (R-AFS) 
 I80 (16.3)
 II35 (7.1)
 III271 (55.3)
 IV104 (21.2)
Vesical lesion diameter at histology (mm)34.9 ± 19.4
Location of vesical nodule (according to surgeon) 
 Base17 (41.5)
 Dome24 (58.5)

Preoperative TVS yielded a diagnosis of bladder endometriosis in 18/41 cases (43.9%), whereas in 23/41 (56.1%) the presence of the endometriotic nodule was missed. The sonographic characteristics of endometriotic nodules visualized at TVS are given in Table 2. In all cases in which bladder endometriosis was detected, a hypoechogenic nodule was seen attached to the bladder wall. The lesion was located in the bladder base, close to the ureteral ostia in 7/18 (38.9%) cases (Videoclip S1, Figures 1–3) and in the bladder dome in 11/18 (61.1%) patients (Videoclip S2, Figures 4 and 5). No discrepancy was found between TVS and surgery as far as the location of the nodule was concerned.

Figure 1.

Sagittal transvaginal sonographic image of the vesicouterine space. A comma-shaped endometriotic nodule on the bladder base is seen located between the bladder (distended by a moderate amount of urine) and the uterus. Note the bright hyperechogenic rim of the tissue surrounding the nodule.

Figure 2.

Transvaginal sagittal section of a spherical endometriotic nodule of the bladder base. Note the presence of a cystic space inside the nodule.

Figure 3.

Transvaginal sagittal section of a comma-shaped endometriotic nodule. Power Doppler imaging reveals the presence of very few blood vessels within the implant.

Figure 4.

Transvaginal transverse section of the pelvis, showing a hypoechogenic discrete nodule attached to the dome of the bladder at a level above the vesicouterine pouch.

Figure 5.

Color Doppler image showing the presence of a ureteral jet in the same patient as shown in Figure 4. Note that the ureteral ostium is located at the apex of the cone produced by the color pixels, distant from the nodule.

Table 2. Transvaginal sonographic characteristics of endometriotic nodules detected in 18 patients
ParameterMean ± SD or n (%)
Lesion diameter (mm)20.22 ± 8.82
Hypoechogenic18/18 (100)
Location 
 Base7/18 (38.9)
 Dome11/18 (61.1)
Morphology 
 Comma-shaped12/18 (66.7)
 Spherical6/18 (33.3)
Presence of a bright rim10/18 (55.6)
Obliteration of vesicouterine pouch15/18 (83.3)
Pain on pressure with probe18/18 (100)

The mean diameter of the endometriotic nodule measured at TVS was 20.2 ± 8.8 mm, but it was significantly larger at histological examination in the same 18 patients (42.5 ± 22.1 mm; P < 0.001). In 12/18 women (66.7%) the nodule had a ‘comma shape’ (Figures 1 and 3), whereas in 6/18 women (33.3%) it had a spherical shape (Figure 2). Small anechogenic cystic areas within the nodule were seen in 5/18 patients (27.8%) (Figures 2 and 4). In 10/18 patients (55.6%) a bright hyperechogenic rim was present, probably due to the presence of adipose tissue and inflammation around the implant (Figures 1 and 2). In addition, all patients had pain on pressure with the probe and in 15/18 women (83.3%), the vesicouterine pouch was obliterated without sliding of the uterus along the urinary bladder. Power Doppler imaging invariably revealed the presence of very few blood vessels within the mass (Figure 3). The sensitivity, specificity, positive and negative predictive values, and overall accuracy of TVS for the diagnosis of bladder endometriotic nodules were 44% (18/41), 100% (449/449), 100% (18/18), 95% (449/472) and 95% (467/490), respectively.

In patients correctly classified, the mean lesion diameter at histological examination was 42.5 ± 22.1 (range, 10–92) mm, whereas in the cases missed the mean lesion diameter was significantly smaller (28.9 ± 14.8 (range, 10–60) mm; P = 0.029). In particular, all nodules with a mean diameter > 35 mm were correctly identified and 11 of 14 patients (78.6%) who needed bladder segmental resection for removal of the nodule had the nodule detected at ultrasound examination. Figure 6 illustrates the sonographic and laparoscopic findings in a case of bladder base endometriosis which was undetected at TVS.

Figure 6.

(a) Transvaginal sonographic longitudinal section of the vesicouterine space in a woman with a thick fibrotic plaque due to endometriosis of the bladder dome, which was misdiagnosed at transvaginal sonography (TVS). The absence of a bulky nodule prevented the correct preoperative sonographic diagnosis. (b) Laparoscopic image in the same patient, obtained a week after TVS. The dome of the bladder shows fibrosis and the endometrial plaque, which extends deeply beneath the peritoneum, towards the bladder muscle.

Age and severity of preoperative dysuria did not differ between patients in whom the bladder endometriosis was detected and those in whom it was missed at ultrasound examination. However, a significant relationship with previous surgery for endometriosis (any type of surgery) was found; 12 of 18 patients (66.7%) in whom TVS suggested a diagnosis of bladder endometriosis had previously undergone surgery for endometriosis, compared with five of 23 (21.7%) cases in which bladder endometriosis was not detected by ultrasound imaging (P = 0.005) (Table 3). This corresponded to a detection rate for bladder endometriosis of 70.6% (12/17) for patients with a history of surgery for endometriosis, and 25.0% (6/24) for those without (P = 0.005).

Table 3. Characteristics of patients with bladder endometriosis nodules detected and those missed at transvaginal sonography (TVS)
ParameterCases diagnosed at TVSCases missed at TVSP
  • Values are mean ± SD, n, n (%) or median (range). VAS, visual analog scale.

  • *

    Mann–Whitney U-test.

  • Chi-square test.

Number of patients18/41 (43.9)23/41 (56.1) 
Age (years)33.39 ± 4.3832.83 ± 5.440.83*
Previous surgery for endometriosis12/18 (66.7)5/23 (21.7)0.005
Lesion diameter at histology (mm)42.5 ± 22.128.9 ± 14.80.029*
Mean diameter of nodules at histology (mm)   
 < 2018 
 20–3048 
 > 30137 
Severity of dysuria according to a VAS3.50 (0–10)0 (0–10)0.13*

DISCUSSION

Accurate preoperative mapping of all endometriotic lesions is required for planning complete excision, for properly counseling patients and for estimating the duration of surgery15. Our data show that bladder endometriotic nodules can be seen at TVS as discrete hypoechogenic lesions in the bladder wall. The most frequently diagnosed site was the bladder dome, close to the vesicouterine pouch. This space is easily investigated by ultrasound imaging if the bladder is distended by a small amount of urine; we therefore recommend asking patients not to empty their bladder before the sonographic examination. A moderate amount of urine, creating an anechogenic acoustic window, could facilitate the detection of nodules along the bladder wall, but excessive distension of the bladder pushes the dome away from the tip of the vaginal probe and this may make it more difficult to detect a nodule in the dome of the bladder.

The ultrasound morphology of bladder endometriotic nodules is fairly consistent. They are either spherical or comma-shaped, and their borders are regular and may exhibit a bright rim, possibly explained by the presence of adipose tissue. We found that such nodules display very few blood vessels at power Doppler evaluation, which is consistent with the results reported by Brosens et al.16.

The mean lesion diameter at TVS was 20.22 ± 8.82 mm, significantly smaller than the diameter of the nodule measured at histological examination (42.50 ± 22.07 mm; P < 0.001). It should be kept in mind that the ultrasound examiner and the pathologist used different instruments to measure the nodules (the first used an ultrasound machine and took an in vivo measurement, whereas the second used a scale to measure the specimen ex vivo). Obviously, the two measurements are not directly comparable. A strength of our study is the large number of patients, all evaluated by TVS in a standardized manner and subsequently undergoing surgery at a single institution with physicians with a special interest in endometriosis.

In our series, TVS showed a high overall accuracy of 95% in diagnosing bladder endometriosis but the sensitivity was low (44%). This is in agreement with a previous study by Bazot et al. on seven patients12; the diagnostic characteristics of TVS in their study were similar to ours (sensitivity, 71%; specificity, 100%; positive predictive value, 94%; negative predictive value, 87%; accuracy, 91%). Interestingly, Bazot et al. also found that detected endometriotic bladder nodules were larger than those missed12. It is reasonable to think that the bigger the nodule the easier the diagnosis; thus, both the physician performing the preoperative ultrasound examination and the surgeon managing the patient should be aware that small endometriotic nodules can be missed at TVS. We had the strong impression that implants forming a bulky nodule (either comma-shaped or spherical) are clearly detectable at TVS, whereas those forming a fibrous plaque along the bladder wall can easily be missed if the sliding of the cervix along the bladder is not systematically sought. Moreover, as most nodules, regardless of their morphology, obliterate the vesicouterine space, we suggest evaluating the sliding of the cervix along the bladder by gently pushing the vaginal probe while looking for the presence of an endometriotic implant. The pain produced by pressing with the probe, due to the fibrosis and obliteration of the vesicouterine space, can be considered as a further ‘soft marker’ for bladder endometriosis. Given the fairly low sensitivity of TVS, especially in small endometriotic bladder implants, the managing clinician should be aware that, in cases with negative preoperative sonographic findings but symptoms such as dysuria, urgency, frequency, suprapubic pain or vesical tenesmus, a magnetic resonance imaging scan can be useful in reaching a correct diagnosis11.

We recommend careful consideration of the medical/surgical history before evaluating a patient with suspected pelvic endometriosis by means of TVS; previous surgery for endometriosis increases the likelihood of detecting bladder endometriosis.

SUPPORTING INFORMATION ON THE INTERNET

The following supporting information may be found in the online version of this article:

Videoclip S1 Endometriotic comma-shaped nodule on the bladder base. The small endometriotic nodule is visible on a longitudinal section of the vesicouterine space close to the tip of the vaginal probe, located between the bladder and the isthmic portion of the uterus.

Videoclip S2 Endometriotic spherical nodule on the bladder dome. The endometriotic implant is seen on a transverse section of the pelvis as a hypoechogenic nodule attached to wall of the bladder, at a level above the fundal part of the uterus.

Ancillary