To assess the value of three-dimensional (3D) vs two-dimensional (2D) ultrasonography (US) in the diagnostic evaluation of the urinary bladder in patients with haematuria.
To assess the value of three-dimensional (3D) vs two-dimensional (2D) ultrasonography (US) in the diagnostic evaluation of the urinary bladder in patients with haematuria.
In all, 42 patients with painless haematuria and/or irritative voiding symptoms were examined with 2D- and 3D-US. US was done with an Acuson Sequoia unit (Siemens Medical Sol. Mountain View, CA, USA) and the Perspective® 3D technique, to assess the presence of bladder lesions, including bladder cancer, bladder wall hypertrophy with trabeculation and diverticula, mucosal bladder folds or re-growth of the prostate mimicking a bladder tumour. The imaging findings were compared with cystoscopy and/or bladder biopsy.
In 21 of the 42 patients (50%) cystoscopy with bladder biopsy revealed bladder cancer. Overall, 3D-US gave a correct diagnosis for 36 of 42 patients (86%). All 21 bladder cancers were correctly diagnosed, and 15 (71%) of the 21 benign bladder lesions were correctly identified. By contrast, 2D-US findings gave suspected bladder cancer in all patients.
3D-US is significantly more accurate than standard 2D-US in the diagnostic evaluation of patients with haematuria. Thus, this diagnostic technique might be useful for routine evaluation of the urinary bladder.
region of interest
two (three)- dimensional
Almost all of the 54 000 new patients with bladder cancer detected each year in the USA are treated by urologists because of symptoms such as painless haematuria in the absence of UTI and/or irritative voiding symptoms . This mode of detection might be acceptable for the 70–80% of new patients who have a low-grade tumour. Low-grade tumours rarely progress (≈ 2% progression rate) and remain confined to the mucosa (stage Ta). However, at first diagnosis, almost 20% of patients with high-grade bladder cancer have muscle-invasive disease, which is the most common cause of death [2–4]. Thus, to improve the prognosis, the early detection of bladder cancer is critical.
The current mode of evaluation for initial diagnosis traditionally involves IVU with two-dimensional ultrasonography (2D-US), CT, MRI and final cystoscopy with eventual biopsy, of which the latter is an invasive and relatively expensive procedure.
Transabdominal 2D-US is often used for examining patients when a bladder tumour is suspected. Most exophytic tumours can be detected, but especially small papillary tumours, flat ‘lawn-like’ tumours and those on the dome of the bladder are hardly detectable or cannot be differentiated from benign lesions.
The recently developed three-dimensional (3D)-US acquires a series of slices of the region of interest (RoI) at slightly different orientations, and forms a volume dataset. It contains all the echo information of the scanned areas. By translation and rotation across the volume dataset, arbitrary planes of the RoI can be presented . 3D-US provides a plastic image of the RoI by surface and volume rendering. To date, 3D-US has been of considerable benefit in evaluating conditions such as congenital heart diseases, congenital uterine malformations and fetal malformations [5–8].
Because the urine-filled bladder has a good penetrability for ultrasound, it is particularly suitable for 3D-US rendering. In the present study we investigated whether 3D-US provides more diagnostic clues, and might be applicable as a new diagnostic technique in the diagnosis of bladder diseases. Recently, the use of 3D-US in the bladder was confined to evaluating the bladder volume or emptying function [9,10]. In this study we compared the value of 3D-US with 2D-US in the diagnostic evaluation of the urinary bladder in patients with haematuria.
Between April 2001 and May 2002, 42 patients (33 men and nine women, mean age of 52 years, range 46–73) were referred to our department of urology with painless haematuria in the absence of UTI; 36 of them also had irritative voiding symptoms, 32 (76%) presented with intermittent gross haematuria and 10 (24%) had microscopic haematuria. Written informed consent was obtained from all patients. All 42 patients were examined with 2D-US, followed by 3D-US and consecutive flexible or rigid cystoscopy with eventual bladder wall biopsy or after transurethral resection of the lesion.
A commercially available Acuson Sequoia 512 US scanner (Siemens Medical Sol. Mountain View, CA, USA) and a transabdominal transducer (4C1 or 6C2) with a frequency range of 1.5–6.0 MHz, were used. The US examinations were done by one investigator; all patients had 2D-US before 3D-US. For optimized scanning, the bladder must be filled to the maximum capacity that could be tolerated by the patient; the volume had to be ≥200 mL.
The device parameter settings were optimized to ensure high-quality 2D images. The grey-scale US gain was carefully adjusted to display the lesion distinctly, to eliminate noise, and to reduce a high posterior acoustic enhancement, which might reduce the imaging quality of the posterior bladder portion. The focal zone was adjusted to the level of the RoI. 2D-US was done in the transverse and sagittal plane, and the bladder was carefully evaluated for the presence of abnormalities. 2D-US was used to assess the inner surface of the bladder wall, the thickness of the bladder wall, and intravesical pathologies such as bladder tumours. The size and location of the lesion was evaluated. Subsequently, 3D-US of the bladder was used to evaluate the whole bladder, using surface, volume and multiplanar rendering modes. The US data were digitally stored and special RoI evaluated.
Briefly, the imaging used three types of 3D displaying algorithms. Surface and volume rendering were used to obtain direct 3D views of bladder disease, and the multiplanar mode to observe the lesion from various orientations and to obtain arbitrary sectional images of the lesion. The Perspective® 3D technology (Siemens Medical Sol.) allows a freehand 3D untracked US sweep. This technology enables imaging acquisition through data volume records with almost simultaneous real-time 3D reconstruction. By means of a cutting knife the pathological processes can be further evaluated. Two to four volume datasets (mean, three sets) were obtained for each patient. The time needed for the whole 3D imaging process was 8 (5–10) min.
The 2D images, volume dataset, and rendered images of each patient were saved on a magneto-optical disk (640 mB; Sony Corporation, Tokyo, Japan), and interpreted in consensus by two experienced investigators (F.F. and L.P.). The 3D images were compared with the 2D images to evaluate the diagnostic accuracy.
The number, exact location, ability to visualize the ureteric orifices and other incidental abnormalities in the bladder were collated. During the image analysis, the diagnosis of a bladder cancer was based on the typical US features (i.e. a superficial lesion as hyperechoic papillary object in the bladder cavity, in the case of bladder wall infiltration as heterogenic hypoechogenic pattern with disruption of the uniformity of the bladder wall echoes). In bladder carcinoma, the findings of 3D-US were compared with those of 2D-US and surgical-pathological examination for evaluating the accuracy in defining the locations and staging the tumours. The staging was based on the TNM classification of bladder cancer according to the Union Internationale Contre le Cancer (2002).
All patients had flexible or rigid cystoscopy with bladder biopsy or transurethral resection of bladder tumour, and the histopathological results were then reviewed. The cystoscopy was taken as the reference standard and the findings of 2D- and 3D-US were compared with it. Statistical analysis was by Pearson’s correlation coefficient, with differences considered significant at a two-tailed P < 0.05.
In 21 patients (50%), cystoscopy and biopsy revealed bladder cancer; in the other 21, cystoscopy showed a benign lesion of the bladder, and in six (14%) a bladder wall biopsy was taken, which proved the lesion to be benign. The histopathological results showed a TCC of the bladder in all 21 patients; 11 were histologically classified as pTa, four were pT1 and six >pT1. One patient had carcinoma in situ in addition to a pT1 G3 tumour; the grade and stage are shown in Table 1. The tumour size varied from 3 mm to occupying more than a quarter of the bladder, and in 12 patients the bladder tumour was multifocal. In all, the 21 patients with bladder tumour presented with 37 intravesical lesions.
|Patient||Number of tumours||3D-US stage||Pathology|
Of the 21 patients with benign lesions, 10 had bladder wall hypertrophy with trabeculation and diverticula, five had mucosal bladder folds, and of six patients with bladder biopsy, histology confirmed three cases of cystitis cystica, two cases of re-growth of the prostate mimicking a bladder tumour, and one case of malakoplakia.
2D-US gave a suspicion of bladder cancer in all patients. Furthermore, 2D-US could not correctly define the locations of the 37 intravesical lesions in 11 cases (30%) and tumour extension in 12 (57%) of 21 cases with pathologically confirmed bladder cancer. Of 37 bladder lesions, 15 were < 5 mm, and eight of these were missed on 2D-US, whereas 3D-US correctly located all lesions. The Pearson correlation coefficient for the correlation between 2D-US findings and in situ situation as assessed by cystoscopy was calculated at r 2 = 0.64, showing a high and statistically significant (P = 0.002) correlation.
3D-US yielded an overall correct diagnosis in 86% (36 of 42) of the patients, providing better diagnostic features than 2D-US, with 100% sensitivity for malignant bladder lesions (21 of 21) and 71% (15 of 21) for benign bladder changes. Also the locations of all 37 intravesical lesions were correctly described at 3D-US when compared with cystoscopy. Furthermore, in nine patients with a suspicion of bladder cancer after 2D-US, 3D-US revealed a correct diagnosis of bladder wall trabeculation. Comparing the in situ situation and 3D-US diagnosis, the Pearson correlation coefficient was 1.0, i.e. a perfect correlation with a two-tailed P < 0.001. Compared with 2D-US, 3D-US provides superior diagnosis (r2 = 0.64; P = 0.002). Additionally, 3D-US allowed a better and systematic visualization especially of the inner bladder wall and of the three different layers: the hyperechogenic layer of the urothelium, the less-echogenic layer representing the lamina propria and a further hyperechogenic pattern of the bladder muscle. Therefore a staging of the bladder tumours was done and correctly assessed for seven of 11 pTa tumours, three of four pT1 tumours and the six tumours >pT1.
In particular, selective visualization of the RoI was very helpful. The tumour-volume measurement by 3D-US indicated an over-estimation in nine of 21 patients (43%) in comparison to the cystoscopy. Of the 42 patients, only half of the diseases were correctly diagnosed by 2D-US, whereas 86% were correctly diagnosed by 3D-US. Therefore, 2D-US overestimated diagnosis with false-positive findings in half, but overlooked 30% of intravesical lesions, especially those <5 mm.
The presence of painless haematuria and/or irritative voiding symptoms warrants investigation of both the upper and lower urinary tracts. IVU, 2D-US, CT and MRI have been used, but the ideal and cost-effective imaging method remains to be determined . Traditionally, IVU was used because it is easily available and cheap. 2D-US is more sensitive for renal parenchymal lesions, but less sensitive for ureteric or bladder lesions. The sensitivity of transabdominal US depends on operator experience as well as tumour size and location, and the degree of bladder distension. The normal bladder wall returns uniformly intense echoes that are disrupted by an infiltrating bladder tumour. Strong echogenic foci are not uncommonly identified on the tumour surface, representing small, encrusted calculi or tumour calcification. Papillary tumours as small as 2–3 mm can be detected, but there is less sensitivity for flat anomalies and those on the dome of the bladder. Deeper tumours are staged more accurately, with a tendency to over-stage superficial disease. Tumours can be mimicked by thrombus, BPH, cystitis and bladder trabeculae, whilst oedema, intravesical clot, bladder wall hypertrophy and tumour calcification all contribute to overstaging [12,13].
TRUS has the potential advantage of being able to distinguish stage T1 and T2 tumours separately, although differentiating between mucosa and lamina propria is not possible. There are particular difficulties with large and echodense carcinomas. Staging accuracy is reported as 78–93% overall and unlike transabdominal scanning, the transurethral approach has greatest accuracy for superficial tumours . The technique is invasive and requires anaesthesia, often in conjunction with cystoscopy.
Transvaginal US might be used as an alternative to transrectal and transurethral techniques, and is thought to be very useful when tumours are localized on the neck, base and front wall of the bladder . CT is the best method for evaluating urinary stones, renal and perirenal infections, and is similar to IVU for upper tract TCC . CT urography is being used increasingly, but the radiation dose, potential allergies to contrast medium, the lack of biopsy specimens, and the reduced sensitivity in detecting tumours of 0.5–1.0 cm remain the primary drawbacks of this procedure [15,16]. MRI of the bladder showed decreased sensitivity and specificity in the detection of polyps of <1 cm, and the entire process is expensive and protracted .
None of the imaging methods is sensitive enough to evaluate the lower urinary tract, and so cystoscopy is needed for most of these patients . Although conventional cystoscopy is well-tolerated by patients, the main drawbacks are the failure to evaluate adjacent structures, a 5–15% risk of UTI, and patient’s discomfort and anxiety. Iatrogenic injury to the urethra and bladder might also occur [18,19].
There would be considerable benefit in examining the bladder internally without using the above methods as the first screening step. US technology has advanced dramatically in recent years, and the development of transabdominal 3D-US has led to its use in this area. The virtual computer-based 3D-US volume-rendering technique allows systematic visualization of the bladder wall and its layers. Recently, Wagner et al. assessed the value and limitations of 3D-US rendering to distinguish invasive from noninvasive bladder cancers. In all, 63 patients with intravesical tumours had 3D-US of the bladder before transurethral resection or radical cystectomy. Superficial (pTa) carcinomas were correctly staged in 66% by 3D-US. Lamina propria infiltrating (pT1) were correctly staged in 83%, and the quota of correct staging of infiltrating carcinomas (>pT1) by 3D rendering was 100%. The overall accuracy was 79%. Thus they showed that 3D-US rendering is most valuable for discriminating between superficial stages of < pT1 and muscle-invasive carcinoma of >pT1.
The present study investigated the clinical value of 3D-US as a screening tool in diagnosing bladder diseases. By comparison with 2D-US, 3D-US was superior. The visualization and selective examination of the suspected lesions in surface- and volume-rendered mode, as well as in arbitrarily selectable levels was very helpful for the diagnosis. (Figs 1–3). It is also useful in areas of the bladder that are difficult to assess, e.g. the anterior bladder neck and tumour within a narrow-mouthed diverticulum. It is possible to view the lesion from many directions and accurately locate and measure it. The data are stored electronically and can be repeated at any time and by different people in different places.
Although the 3D images are acquired from the 2D dataset, it seems that the 3D images in the volume- or surface-rendered mode improves the anatomical orientation and the detection of abnormalities. As multiple views can be taken, interpretation and making a definite diagnosis are easier in 3D, as shown by other studies comparing 2D- and 3D-US.
A limitation is that this technique is a little more time consuming and needs special hardware and software, which results in higher costs (for the US unit and the investigator). Furthermore we have not compared different 3D techniques, so we are only confident that these data can be obtained with the same equipment. As the findings were interpreted by consensus only, we have no data on inter- or intra-observer variability.
3D-US of the bladder proved to be relatively simple and feasible in clinical application. This technology gives additional anatomical information and seems to allow better differentiation between benign and malignant bladder diseases. Further advantages include the costs, no radiation exposure and the good visualization of small lesions and sessile lesions. However, this technology cannot replace pathological staging and there is still a need to further improve the technology of 3D-US to assess mucosal abnormalities better, and studies in normal control subjects to assess the specificity. In conclusion, 3D-US is better than 2D-US for the diagnostic evaluation of suspected bladder lesions.