| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Diagnostic Radiology, Keio University School of Medicine, 35
Shinanomachi, Shinjuku-ku, Tokyo, Japan 160-8582.
2 Department of Urology, Keio University School of Medicine, Tokyo, Japan.
3 Present address: Department of Urology, Tokyo Medical University, Toyko,
Japan.
4 Department of Radiology, Brigham and Women's Hospital, Boston, MA.
Received April 11, 2006;
accepted after revision August 31, 2006.
Address correspondence to M. Jinzaki
(jinzaki{at}sc.itc.keio.ac.jp).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Thirty-six patients (30 with 59 cystoscopy-proven bladder cancers and six with normal bladders) underwent dynamic contrast-enhanced MDCT of the pelvis and abdomen. Images were obtained from the symphysis pubis to the diaphragm 70 seconds after injection of 100 mL of contrast medium. McNemar test was used to compare sensitivity per patient, segment, and tumor and specificity per patient and segment for each of three reconstruction methods: 5-mm sections with no overlap (i.e., 5-mm axial images), 2.5-mm sections with 1.25-mm overlap (i.e., thin-section axial images), and 2.5-mm sections with 1.25-mm overlap and multiplanar reformation (MPR) (i.e., thin-section axial images with MPR).
RESULTS. MDCT with a combination of thin, overlapped sections and MPR depicted all but one of 47 bladder tumors larger than 5 mm but only five of 12 tumors 5 mm or smaller. There were no false-positive findings. Per-tumor sensitivity was significantly better with thin-section images with MPR (90%) and thin-section images alone (86%) than with 5-mm axial images (80%) (p < 0.05). Per-segment sensitivity was significantly better with thin-section images with MPR (95%) and thin-section axial images alone (87%) than with 5-mm axial images (79%) (p < 0.05). Per-patient sensitivity and per-patient and per-segment specificity did not differ with the three methods.
CONCLUSION. Dynamic contrast-enhanced MDCT of the pelvis shows promise for the detection of bladder tumors. Use of thin-section images with MPR and thin-section axial images alone had a significantly better rate of detection of bladder tumors than use of 5-mm axial images.
Keywords: bladder cancer • CT urography • dynamic CT • genitourinary tract imaging • MDCT • urinary tract
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
MDCT yields images that are high in spatial resolution and can be viewed in multiple planes. As a result, MDCT has been used to examine the entire urinary tract, including the intrarenal collecting systems, ureters, and bladder [16-19]. These protocols, referred to as CT urography, have included CT scans of the abdomen and pelvis before and after administration of contrast material. To date, published protocols have described scanning the abdomen first and then scanning the pelvis during the excretory phase. Because bladder tumors become more enhanced than the adjacent normal bladder wall during dynamic contrast-enhanced CT of the pelvis [20], we sought to determine whether detection of bladder tumors would improve if the pelvis were scanned first. In this study, we assessed whether early-phase dynamic contrast-enhanced MDCT can be used to detect bladder tumors and whether thin reconstruction would improve the detection rate.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Cystoscopic examination within 8 days of MDCT revealed 59 bladder tumors in 30 patients; all tumors were pathologically confirmed to be bladder cancer. The bladder was divided into six segments: 10 tumors were located in the anterior segment, 17 tumors in the posterior segment, eight in the right side, 10 in the left side, 10 in the dome, and four in the neck. In one patient, one focus of carcinoma in situ (CIS) was detected among eight random biopsy specimens from the right side of the bladder. Cystoscopic examination for evaluation of synchronous tumors revealed no synchronous tumors in six patients with upper urinary tract tumors.
CT Examination
MDCT examinations were performed with a four-channel data acquisition
system (LightSpeed QXi, GE Healthcare). Patients were instructed not to void
for at least 1 hour before MDCT. MDCT scans began in the pelvis and were
obtained from the symphysis pubis to the diaphragm 70 seconds after injection
of 100 mL of 300 mg I/mL iohexol (Omnipaque 300, Daiichi Pharmaceutical) at a
rate of 3 mL/s. MDCT technique included a gantry rotation speed of 0.6 second,
tube voltage of 120 kVp, tube current of 180-250 mA, collimation of 2.5 mm,
and pitch of 0.75. Images of the abdomen and pelvis reconstructed as 5-mm
sections at 5-mm intervals were used for clinical interpretation. In addition,
images of the pelvis were reconstructed as both 5-mm sections with no overlap
(i.e., 5-mm axial images) and 2.5-mm sections with 1.25-mm overlap (i.e.,
thin-section axial images) for this study. Radiation dose was evaluated with
the weighted CT dose index defined by the International Electrotechnical
Commission [21]. Values of
weighted CT dose index ranged from 18 to 25 mGy.
Image Analysis
Two radiologists, each with at least 15 years of experience in interpreting
abdominal CT scans and blinded to history and radiologic and cystoscopic
findings, independently interpreted images of the pelvis acquired in 36 MDCT
examinations. Each scan was presented as three reconstruction methods: 5-mm
axial images alone, thin-section axial images alone, and thin-section axial
images combined with multiplanar reformation (MPR) (i.e., thin-section images
with MPR) on an independent workstation (Advantage Windows 4.1, GE
Healthcare). When MPR was available, reviewers were free to choose any planes
(e.g., coronal, sagittal, and oblique planes) to supplement the axial
plane.
Reviewers were asked to identify and record the size, number, and location of bladder tumors, including bladder segment, using the definitions used at cystoscopy. Masses that were polyplike and protruded into the bladder lumen and areas of focal wall thickening more enhanced than the adjacent bladder wall were classified as bladder tumors. MDCT scans were presented in a random manner to each reviewer at each of three sessions. The 5-mm axial images were assessed alone for each patient during the first review session. After a 3-week delay, thin-section axial images were assessed alone for each patient. After an additional 3 weeks, thin-section images with MPR were assessed for each patient. Differences in interpretation were resolved by consensus.
The kappa statistic, calculated to determine inter-observer agreement, was classified as follows: a kappa value less than 0.20 indicated poor agreement; 0.21-0.40, fair; 0.41-0.60, moderate; 0.61-0.80, good; and 0.81-1.00, very good agreement [22].
Data Analysis
Reviewers' findings on MDCT scans were compared with cystoscopic findings.
Tumors not detected prospectively with MDCT were retrospectively reevaluated
for visibility by two radiologists on three data sets. Differences in
detection rates for each of three data sets (5-mm axial images, thin-section
axial images, and thin-section images with MPR) were assessed with McNemar
test. Statistical significance was considered to be present at p
<0.05.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Tumor Detection Stratified by Patient
Table 1 shows the
per-patient results of the reviewers' interpretation using the three
reconstruction methods for bladder tumor detection. For thin-section images
with MPR, sensitivity and specificity for identification of bladder lesions
both were 100%. Per-patient sensitivity and specificity did not differ among
the three reconstruction methods.
|
Tumor Detection Stratified by Bladder Segment
Table 2 shows the results of
the three re-construction methods for lesion detection based on bladder
segment. For thin-section images with MPR, the sensitivity and specificity for
tumor detection per bladder segment were 95% and 100%, respectively.
Per-segment sensitivity was significantly better for thin-section axial images
alone and thin-section images with MPR than for 5-mm axial images (p
< 0.05). Per-segment sensitivity for thin-section axial images alone was
not significantly different from that of per-segment sensitivity for
thin-section images with MPR (p > 0.05). Per-segment specificity
was the same for the three reconstruction methods.
|
Tumor Detection Stratified by Tumor
Detection rates for each of the three reconstruction methods according to
tumor size are shown in Table
3. Fifteen (25.5%) of 59 tumors were larger than 20 mm, 15 (25.5%)
of 59 tumors were larger than 10 mm but were 20 mm or smaller
(Fig. 1), 17 (29%) of 59 tumors
were larger than 5 mm but were 10 mm or smaller
(Fig. 2), and 12 (20%) of 59
tumors measured 5 mm or smaller (Fig.
3). Thin-section images with MPR were used to detect all but one
of 47 bladder tumors larger than 5 mm but only five of 12 tumors 5 mm or
smaller. For thin-section images with MPR, the overall per-tumor sensitivity
was 90%, including 79% for tumors 1 cm or smaller and 58% for tumors 5 mm or
smaller. Per-tumor sensitivity was significantly better for thin-section
images with MPR (90%) and thin-section axial images alone (86%) than for 5-mm
axial images (80%) (p < 0.05). Per-tumor sensitivity for
thin-section axial images alone was not significantly different from per-tumor
sensitivity for thin-section images with MPR (p >0.05).
|
|
|
|
There were no false-positive findings. The sensitivity for detecting bladder tumors 5 mm or smaller was significantly better for thin-section images with MPR and thin-section axial images (both, 58%) than for 5-mm axial images (25%) (p < 0.05). Use of thin-section axial images improved the detection rate only for tumors smaller than 5 mm (Table 3). Of the four tumors detected on thin-section images, three were located in the bladder dome or neck. MPR improved the detection of two bladder lesions larger than 5 mm adjacent to normal anatomic structures (Table 3). One tumor was located in the anterior aspect of the bladder contiguous with the symphysis pubis (Fig. 4A, 4B, 4C, 4D), and the other was in the bladder neck contiguous with the prostate (Fig. 5A, 5B). One 8-mm tumor adjacent to a 15-mm tumor (Fig. 6) was not detected with any reconstruction method. These two tumors were considered to be one lesion by both reviewers.
|
|
|
|
|
|
|
The sole tumor classified as CIS and found only at random biopsy was not detected prospectively with MDCT. One reviewer detected mucosal enhancement of the bladder wall in the right side of the bladder, the same segment that yielded CIS. The other reviewer agreed that mucosal enhancement was present, but both reviewers prospectively considered the abnormality not to represent tumor.
Interobserver Agreement
Table 4 shows the
interobserver agreement for per-tumor detection by the two reviewers. When
5-mm axial images were evaluated alone, both reviewers initially agreed on the
identification of 43 tumors in 30 patients and of no tumor in six patients. By
consensus, the reviewers decided that 47 (80%) of 59 bladder tumors were
present in 30 patients and that no lesion was present in six patients. The
calculated kappa value was 0.76.
|
When thin-section axial images were evaluated alone, both reviewers initially agreed on the identification of 49 lesions in 30 patients and of no lesion in six patients. By consensus, the reviewers decided that 51 (86%) of 59 bladder tumors were present in 30 patients and that no lesion was present in six patients. The kappa value for agreement was 0.80.
When thin-section images with MPR were evaluated, both reviewers initially agreed on the identification of 51 lesions in 30 patients and of no lesion in six patients. By consensus, the reviewers decided that 53 (90%) of 59 bladder tumors were present in 30 patients and that no lesion was present in six patients. The kappa value for agreement was 0.84.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Per-tumor sensitivity achieved with thin-section images combined with MPR was 90%, including 79% for bladder lesions 1 cm or smaller, and 58% for lesions of 5 mm or smaller. These data are comparable with those for virtual cystoscopy [12-15]. Virtual cystoscopy with an air-filled bladder technique was used to detect 90% of bladder lesions, including 60% of lesions smaller than 5 mm in one study [12] and 77% of lesions smaller than 1 cm in another [13]. In a third study [15], investigators evaluated virtual cystoscopy with contrast material. The overall rate of detection of bladder tumors was 95%, including 88% for tumors smaller than 5 mm. The detection rate for tumors smaller than 5 mm was better than ours (58%), likely because those investigators used thinner collimation (1.25 mm). However, their specificity was only 87% because there were several false-positive findings.
Virtual cystoscopy shows promise. However, catheterization of the bladder is required for performing virtual cystoscopy with an air-filled-bladder technique, which results in a prolonged examination time. Furthermore, when IV contrast material is used to fill the bladder, patients are required to wait for more than 1.5 hours and to take the supine and prone position several times before scanning to obtain adequate mixing of contrast material and urine [15]. Virtual cystoscopy with an air-filled bladder also requires the patient to be scanned twice, in the supine and the prone positions, to be sure that air or contrast medium interfaces with each segment of the bladder [12, 14]. On the other hand, dynamic contrast-enhanced MDCT requires only one scan and no catheterization. MDCT does, however, require the patient not void for at least 1 hour.
With dynamic contrast-enhanced MDCT, only one of 47 tumors larger than 5 mm was missed. In the case of an 8-mm mass contiguous with a 15-mm mass, both lesions were interpreted as a single mass. Five of the masses 5 mm and smaller were missed. However, additional tumors were detected in every patient in whom a bladder tumor was missed. Therefore, at least one bladder tumor was detected in every patient in our study population.
Reconstructing the data in thin (2.5-mm) overlapped images and in multiple
planes is an important component of our technique. An additional 10% of tumors
would not have been detected had we used only 5-mm axial sections. The benefit
of thin sections was particularly evident in the cases of small (
5 mm)
tumors. MPR images were helpful for detecting tumors in the anterior aspect of
the bladder contiguous with the symphysis pubis and in the bladder neck
contiguous with the prostate. MPR did not improve the rate of detection of
tumors smaller than 5 mm, but it was helpful for differentiating bladder
tumors from normal anatomic structures.
A theoretic limitation of dynamic contrast-enhanced MDCT relative to cystoscopy is its inability to depict lesions (e.g., CIS) that manifest at cystoscopy only with changes in color and texture. One case of CIS in our study may have been represented by focal hyperenhancement of the bladder wall because it was seen in the same segment in which CIS was found. Focal inflammation can manifest as focal hyperenhancement of the bladder wall. The relation between focal hyperenhancement and CIS or focal inflammation needs further investigation.
We selected a 70-second delay. The results of a previous study [11] in which attenuation values of bladder tumors were measured on CT images obtained with delays of 40, 60, 80, and 100 seconds suggested that the attenuation value of bladder tumors was significantly greater when the bladder was imaged with a 60- or 80-second delay than with a 100-second delay. No significant difference was found between 60 and 80 seconds. Thus, a 70-second delay should be suitable for obtaining sufficient contrast between a bladder tumor and normal bladder wall.
There were limitations to our study. First, it was a pilot study, so the number of subjects was small. Second, we used 2.5-mm sections with 1.25-mm overlap. Recent advances in MDCT enable use of thinner sections, which may further improve the rate of detection of bladder tumors smaller than 5 mm. Cystoscopy was used as the reference standard in this study. Cystoscopy, however, is imperfect; tumors can be missed, and normal structures can be mistaken for bladder tumors [23, 24].
Adding an excretory phase to our protocol may aid in detection of both bladder and upper urinary tract cancers. For example, during contrast-enhanced MDCT of both the abdomen and pelvis, the pelvis can be scanned first. This method would allow scanning of the bladder during the early phase and of the kidneys during the nephrographic phase, the ideal time for imaging renal masses. Scanning during the excretory phase would help detect upper urinary tract cancers [25]. Adding an excretory phase to our protocol also would aid in staging of both bladder and upper urinary tract cancers. Multiphasic MDCT has been reported to be accurate for staging of upper urinary tract cancers [26]. Early dynamic contrast-enhanced CT of the bladder has been reported to be useful for local staging of bladder cancer, that is, for determining whether the tumor has extended into the perivesical fat [11]. One limitation to this technique is that the radiation dose would be increased if full abdominal and pelvic CT scans were obtained during both the early dynamic contrast-enhanced and excretory phases. Adding the excretory phase to our technique would result in a single comprehensive MDCT examination for patients with hematuria, and the findings could be used to detect and stage urinary tract cancer.
In conclusion, early-phase dynamic contrast-enhanced MDCT was used to detect 90% of bladder tumors in this pilot study. Diagnostic accuracy improved when the data were reconstructed as thin sections with MPR. In particular, thin axial sections were used to improve the detection of bladder tumors smaller than 5 mm, and MPR was used to improve the detection of bladder tumors in areas contiguous with other anatomic structures and difficult to analyze. Dynamic contrast-enhanced MDCT is a promising technique for the detection of bladder tumors and deserves further study. A CT technique that can be used effectively to detect bladder tumors may ultimately be used instead of cystoscopy in selected patients.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  R. H. Cohan, E. M. Caoili, N. C. Cowan, A. Z. Weizer, and J. H. Ellis MDCT Urography: Exploring a New Paradigm for Imaging of Bladder Cancer Am. J. Roentgenol., June 1, 2009; 192(6): 1501 - 1508. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. G. Silverman, J. R. Leyendecker, and E. S. Amis Jr What Is the Current Role of CT Urography and MR Urography in the Evaluation of the Urinary Tract? Radiology, February 1, 2009; 250(2): 309 - 323. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Tsampoulas, A. C. Tsili, D. Giannakis, Y. Alamanos, N. Sofikitis, and S. C. Efremidis 16-MDCT Cystoscopy in the Evaluation of Neoplasms of the Urinary Bladder Am. J. Roentgenol., March 1, 2008; 190(3): 729 - 735. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
