| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Technical Developments |
1 From the Departments of Radiology (H.J.K., A.Y.K., S.S.H., J.H.B., H.J.W., Y.M.S., P.N.K., H.K.H., M.G.L.) and Internal Medicine (M.H.K.), Asan Medical Center, University of Ulsan College of Medicine, 388-1 Pungnap-dong, Songpa-gu, Seoul 138-736, Korea; and Department of Diagnostic Radiology, Kyung Hee University Hospital, Seoul, Korea (H.J.K.). Received November 10, 2004; revision requested January 13, 2005; revision received January 30; accepted February 28. Address correspondence to A.Y.K. (e-mail: aykim{at}amc.seoul.kr).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Nevertheless, biliary ductal evaluation in patients with hilar cholangiocarcinoma is difficult, particularly when performed with two-dimensional computed tomography (CT), because of the longitudinal growth pattern of the tumor along the axis of the bile duct (6–10). This is why direct cholangiography has been performed in addition to CT, although the primary imaging modality for patients suspected of having this cancer is ultrasonography (US) or CT. To overcome this limitation of CT, the potential usefulness of CT cholangiography has been investigated extensively; however, diagnostic yields have not been satisfactory (11–20). With CT cholangiography performed with oral or intravenous contrast agents, reliable biliary excretion of contrast material cannot be guaranteed in patients with high-grade bile duct obstruction (14), and CT cholangiography performed without biliary contrast agent does not enable detailed three-dimensional (3D) evaluation of the biliary tree.
The advent of multi–detector row CT—owing to rapid scanning, thinner (1-mm) sections, and newer workstations—has enabled the evolution from section-based to volume-based techniques. Therefore, we assumed that by using a combination of direct cholangiography and multi–detector row CT—that is, 3D direct multi–detector row CT cholangiography—we could overcome the fundamental limitations of CT in the examination of patients with hilar cholangiocarcinoma. To our knowledge, however, no investigators have previously investigated the feasibility and clinical performance of 3D direct 16–detector row CT cholangiography in this population of patients. Thus, the purpose of our study was to assess the feasibility of 3D direct multi–detector row CT cholangiography for determining the extent of biliary ductal invasion by hilar cholangiocarcinoma by using pathologic and surgical results as the reference standard.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
For the 11 patients who underwent surgery, the interval between direct multi–detector row CT cholangiography and surgery ranged from 12 to 33 days (mean, 19 days). Curative (n = 7) or palliative (n = 1) surgery was performed in eight of these patients. In the remaining three patients, tumor extension could be identified, but resection was not attempted owing to vascular invasion (n = 2) or advanced tumor extension (n = 1). Twelve patients underwent conservative management owing to intrahepatic metastasis to the contralateral segment (n = 2), advanced tumor involvement in both liver lobes (n = 3), vascular invasion (n = 1), small liver volume (n = 1), or poor health status (ie, not able to tolerate extensive surgery) (n = 5). Consequently, surgical or pathologic confirmation of bile duct involvement by hilar cholangiocarcinoma was possible in 11 patients (eight men, three women; age range, 39–69 years; mean age, 57 years), and these patients constituted the final study population for imaging analysis.
Three-dimensional Direct Multi–Detector Row CT Cholangiography
In the 11 patients, all CT images were obtained with a 16–detector row CT scanner (Somatom Sensation 16; Siemens, Forchheim, Germany). Patients were instructed to fast for at least 6 hours before the examination. After the function of the preexisting biliary drainage tube(s) was checked, bile within the biliary tree was drained by using a syringe and mild negative pressure. Then, with the patient supine, a 1:6 dilution of nonionic contrast material (iopromide, Ultravist 370; Schering, Berlin, Germany) mixed with normal saline was gently injected through an ENBD or PTBD tube to fill the bile duct. Injection of contrast material was stopped when the patient reported having discomfort in the right upper quadrant. Then, the biliary drainage tubes were clamped to prevent injected contrast material from escaping from the bile duct.
While in the left decubitus position, the patient was shaken to and fro approximately 15 times at about a 30° angle to completely fill the bile duct—especially the left medial segmental and left lateral inferior subsegmental ducts—with contrast material. Then, with the patient supine, a scout image was obtained to identify bile duct distention with contrast material before full-scale CT scanning. Other contrast materials, such as oral or intravenous agents, were not used in these patients.
CT images of the area from the dome of the diaphragm to the inferior edge of the liver were obtained during a single breath hold. The CT parameters used included a detector-row configuration of 0.75 x 16.0 mm, a gantry rotation speed of 0.5 second, a table feed of 12 mm per gantry rotation, 120 kVp, and 150 mA.
After CT scanning, the clamp was removed and the injected contrast material was allowed to drain from the bile duct. Patients were transferred to the ward, where physicians and nurses who were in charge of monitoring the status of each patient checked their bile drainage amount and vital signs for 48 hours to identify any postprocedural complications.
All direct multi–detector row CT cholangiographic procedures were performed in the CT suite by one radiologist (H.J.K.). The radiologist recorded the amount of diluted contrast agent injected, the weighted CT dose index, and the dose-length product. He also checked the patients' medical records and communicated with the physician in charge of monitoring the patients to learn of any postprocedural complications.
Image Processing
The 1-mm transverse CT sections were reconstructed at 0.7-mm intervals, and then postprocessing was performed at a commercially available workstation (Advantage Window 4.2; GE Medical Systems, Milwaukee, Wis). Reformation of the source images was performed by one radiologist (H.J.K., 4 years experience in CT reformation). Targeted maximum intensity projection images were generated to visualize the primary and secondary confluence levels of the bile duct and the relationship between missing (ie, non–contrast material–filled) and contrast material–filled bile ducts. Mainly a 5-cm slab thickness and oblique coronal or transverse projections were used to obtain the maximum intensity projection images.
Primary confluence level of the bile duct was defined as the confluence of the right and left main hepatic ducts. Secondary confluence level of the bile duct was defined as the confluence of the second-order biliary ducts (ie, between right anterior and right posterior segmental ducts or between left medial and left lateral segmental ducts). Three-dimensional CT images were generated by using a volume-rendering technique. Volume-rendering parameters were subjectively selected for optimal visualization of the bile duct. The volume-rendered images were magnified and projected at the appropriate sizes and viewing angles, which were similar to the sizes and angles of the maximum intensity projection images used to identify the primary and secondary confluence levels of the bile duct and bile duct variation.
Image Analyses
All images were interpreted at an independent workstation (Advantage Window 4.2). Two certificated abdominal radiologists who had 7 years (J.H.B.) and 11 years (H.J.W.) experience in interpreting hepatobiliary images analyzed all of the CT images in consensus. The radiologists were aware of the main clinical indication for performing direct multi–detector row CT cholangiography (ie, suspected hilar cholangiocarcinoma) but unaware of the results of the other diagnostic examinations. The complete direct multi–detector row CT cholangiography volumetric data sets, including the source images, were evaluated.
First, the maximum intensity projection and volume-rendered images were reviewed for assessment of bile duct opacification, with a focus on secondary biliary confluence because proper opacification of the secondary confluence level was a prerequisite for evaluation of the extent of ductal involvement by the tumor. If the bile ducts were well opacified to the level of the fourth-order branch, the ipsilateral secondary confluence level was judged to have excellent opacification. If the ducts were well opacified to the level of the third-order branch, the ipsilateral secondary confluence level was judged to have good opacification. If the third-order ductal branches were not properly opacified despite the injection of contrast material through a biliary drainage tube, the ipsilateral secondary confluence level was judged to have poor opacification. If the bile duct could not be opacified owing to complete obstruction of the primary confluence level and/or absence of a biliary drainage tube, the ipsilateral secondary confluence level was judged to be nonopacified at cholangiography. On the targeted maximum intensity projection images, we also determined the presence or absence of respiration or stair-step artifact.
The radiologists then determined the extent of biliary ductal invasion by hilar cholangiocarcinoma at all identifiable primary and secondary confluence levels of the bile duct. Complete ductal occlusion (ie, a focal nonopacified ductal segment), abrupt ductal narrowing, or focal tubular ductal narrowing was considered to indicate tumor invasion. Also, if there was a nonopacified or incompletely opacified segmental or subsegmental duct—that is, a missing duct—the ipsilateral secondary confluence level was considered to have tumor invasion.
Finally, the anatomy of the second-order biliary tract was classified as conventional or variant. Conventional biliary tract anatomy was defined as a configuration in which the right posterior duct (which drains liver segments VI and VII) drained into the right anterior duct (which drains liver segments V and VIII) to form a right hepatic duct that then joined the left hepatic duct (formed by ducts draining liver segments II, III, and IV) (21).
CT Cholangiographic Results Compared with Surgical and Pathologic Findings
One author (H.J.K., 7 years experience in hepatobiliary CT scanning), who was not involved in the image analyses, collected information regarding the standard of reference. In eight patients, the surgical and pathologic specimen findings were considered the reference standard, and in three patients, only the surgical findings were the reference standard. In the surgical and pathologic reports, the extent of tumor involvement (ie, tumor extent) in the following regions was recorded: common bile duct, cystic duct, gallbladder, common hepatic duct, first-order branches (left and right hepatic ducts), and second-order branch (segmental duct). In seven patients who underwent left or right lobectomy, tumor involvement in the second-order branch could be identified only in the pathologic specimens.
Three radiologists—two who reviewed the direct multi–detector row CT cholangiographic findings and one who collected information for the reference-standard data—in consensus compared the direct multi–detector row CT cholangiographic findings with the reference-standard findings. Comparisons were made according to the primary and secondary confluence level findings—not according to the first- and second-order branch findings—because direct multi–detector row CT cholangiography depicts the stenosis or obstruction next to the tumor rather than the tumor itself.
Statistical Analyses
With regard to ductal tumor involvement, the sensitivity (SEN), specificity (SPEC), positive predictive value (PPV), and negative predictive value (NPV) of direct multi–detector row CT cholangiography focused on the secondary biliary confluence level were calculated as follows: SEN = (TPCTC/PP) · 100, where TPCTC is the number of correct CT cholangiography–based diagnoses of ductal tumor invasion and PP is the number of proved cases of ductal invasion. SPEC = (TNCTC/PN) · 100, where TNCTC is the total number of correct CT cholangiography–based diagnoses of negative ductal tumor invasion and PN is the number of proved cases of intact bile duct without tumor invasion. PPV = (TPCTC/CTCINV) · 100, where CTCINV is the total number (correct and incorrect) of CT cholangiography–based diagnoses of ductal tumor invasion. NPV = (TNCTC/CTCINT) · 100, where CTCINT is the total number (correct and incorrect) of CT cholangiography–based diagnoses of intact bile duct without tumor invasion.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Injected Contrast Agent Concentrations and CT Radiation Doses
In the 11 patients in whom comparisons between the CT cholangiographic results and the surgical and/or pathologic findings were possible, the amount of diluted contrast agent injected through the biliary drainage tube was variable: In some patients, diluted contrast agent flowed into the duodenum through an incompletely obstructed common hepatic duct or side holes in the ENBD tube. The amount of contrast agent injected ranged from 15 to 60 mL (mean, 31 mL). The weighted CT dose index and the mean dose-length product were 11 mGy and 262.6 mGy · cm ± 46.6 (standard deviation) (range, 203–332 mGy · cm), respectively.
Opacification
With regard to biliary tree opacification, two (9%) of the total 22 secondary confluence levels were nonopacified at direct multi–detector row CT cholangiography. Direct multi–detector row CT cholangiography could be performed in 20 of the 22 secondary confluence levels and yielded excellent opacification (Fig 1) in most (n = 18, 90%) of them. Good (Fig 2) and poor opacification was noted at one secondary confluence each (two of 20, 10%). Thus, three secondary confluence levels were judged to be nonopacified (n = 2) or have poor opacification (n = 1) and were excluded from the image analysis performed to determine the extent of tumor involvement at the secondary confluence level. No respiration or stair-step artifact caused disruption of the targeted maximum intensity projection image (Fig 2a).
|
|
|
|
|
|
|
|
|
Ductal Variation
Bile duct variation was detected at CT cholangiography and confirmed at surgery in three patients. These variations included the left medial segmental duct draining into the left lateral superior subsegmental duct, the right posterior segmental duct draining into the left hepatic duct (Fig 1), and an accessory hepatic duct (Fig 3).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
CT cholangiographic procedures can be divided into two groups: those in which the bile duct is accentuated as a low-attenuating structure on minimum intensity projection or multiplanar reformatted images (27–29) and those in which the bile duct is highlighted as a high-attenuating structure with use of oral cholangiographic contrast agent (11–13), intravenous cholangiographic contrast agent (14–20), or direct cholangiography (26,30). However, most CT cholangiographic methods may have major clinical limitations—especially for bile duct evaluation in patients with hilar cholangiocarcinoma—and thus have not been widely applied in clinical practice.
Our study results demonstrate the feasibility of 3D direct multi–detector row CT cholangiography as a reliable diagnostic tool for evaluating biliary ductal involvement by hilar cholangiocarcinoma. The described technique yielded excellent opacification of the biliary tree to the fourth-order branch (in 18 [90%] of 20 secondary confluence levels), was tolerable in most patients, and caused no major complications. It is also interesting that nonopacification or poor opacification was noted at only three (14%) of the 22 secondary biliary confluences at CT cholangiography. Considering the number of bile-draining tubes in each patient (one tube each in four patients, two tubes each in seven patients), we believe that these outcomes represented fewer undesirable results compared with the outcomes of conventional direct cholangiography (31).
A well-known major disadvantage of direct cholangiography in the assessment of biliary ductal invasion is failed bile duct opacification after five or six punctures (32). Our better results can be attributed to the improved ductal patency that resulted from the preexisting biliary decompression. With emergent bile drainage, infected bile is removed and ductal epithelial inflammation and swelling are diminished. Accordingly, we think that 3D direct multi–detector row CT cholangiography is more useful than conventional direct cholangiography in the diagnosis of bile duct involvement by hilar cholangiocarcinoma. Also, all of the nonopacified or poorly opacified secondary biliary confluences were pathologically proved to have tumor invasion. Therefore, if this finding is noted at 3D direct multi–detector row CT cholangiography, tumor invasion at this level should be suspected and further careful evaluation, such as comparison with cross-sectional image findings, will lead to more accurate tumor staging.
Our evaluation of potential biliary ductal invasion by hilar cholangiocarcinoma yielded correct determinations of tumor extent in 10 of 11 patients; these data are superior to previously reported CT results (22,23). Similar to our group, Furukawa et al (26) reported 3D direct CT cholangiography to have high diagnostic accuracy (100%) in five patients with hilar cholangiocarcinoma. Although their diagnostic yield seems to be superior to ours, their results could not be directly compared with ours because they evaluated bile duct involvement by using rotating cine cholangiography. In their study, stair-step artifacts on reformatted images were inevitable owing to the use of a single–detector row CT scanner, so the extent of bile duct cancer invasion could not be assessed with 3D CT cholangiography. However, our direct multi–detector row CT cholangiography examinations revealed no artifacts because we used thin collimation and a thin reconstruction interval, which were generated by the 16–detector row CT scanner.
In this study, 3D direct multi–detector row CT cholangiography yielded one false-positive diagnosis of the tumor extent in the biliary tree. Retrospectively, this overestimation proved to have been the result of a complex anatomic configuration—specifically, an accessory hepatic duct and the drainage tube itself. Since an accessory hepatic duct is usually smaller than a normal duct and the tip of the drainage tube also was located within the bile duct in this case, it is not difficult to suppose that these overlapping findings were likely to mimic ductal narrowing. Therefore, the potential for a false-positive finding, which can be caused by the biliary drainage tube itself or a ductal variation, should be considered when interpreting direct multi–detector row CT cholangiograms.
Compared with the performance values of other imaging modalities, the high positive and negative predictive values (90% and 100%, respectively) of direct multi–detector row CT cholangiography achieved in the current study are encouraging. In assessments of the extent of hilar cholangiocarcinoma involvement, US and helical CT generally tend to yield underestimations (7–10,33,34) because of failure to depict nonunited intrahepatic ductal branches. Magnetic resonance (MR) cholangiography has good diagnostic accuracy in defining the extent of hilar or perihilar biliary ductal involvement compared with direct cholangiography (2,23,24,35,36). However, MR cholangiography also tends to generate underestimated data because of failure to depict obstructed noncommunicating hepatic ducts (2,24,35). Zidi et al (2) reported that MR cholangiography enabled the correct diagnosis of biliary ductal involvement in 14 (78%) of 18 patients and led to the underestimation of tumor extent in four (22%) patients. The use of recently introduced technical improvements such as parallel imaging and prospective acquisition correction may further increase the diagnostic value of MR cholangiography (37). However, the section thickness in MR cholangiography still cannot be decreased to less than 2 mm with use of these new technical improvements.
The ability to depict missing ducts easily and clearly is thought to be an apparent advantage of 3D direct multi–detector row CT cholangiography. Although the concept of a missing duct is not uncommon in direct cholangiography, it is not easy to establish with confidence that a small duct, such as the left medial segmental duct or subsegmental duct, is missing with this examination. Also, this concept cannot be used as easily when interpreting MR cholangiographic findings. Another advantage of 3D direct multi–detector row cholangiography is that reformatted images of variable projections and volumes can be generated without additional examinations if one complete CT examination is performed. Conversely, the reformation or change of a projection is not impossible with direct cholangiography after completion of the examination.
This study had several limitations. First, to use the described CT technique, the preliminary biliary drainage—whether endoscopic or percutaneous transhepatic—must be performed without stent displacement. Second, the risk of secondary infections associated with this technique, such as cholangitis or sepsis, should be considered. Because of these two major drawbacks, we do not think that this CT technique should replace noninvasive imaging modalities such as MR cholangiography.
Because biliary drainage to decompress the bile duct is an inevitable procedure in some patients with hilar cholangiocarcinoma and direct cholangiography is the best method of demonstrating the biliary tree anatomy (12,24,25), direct multi–detector row CT cholangiography might have a role in the evaluation of bile duct involvement by hilar cholangiocarcinoma. Although one patient in this study developed mild cholangitis, it was well controlled with antibiotic therapy. This potential complication may be prevented with gentle and slow injection of the sterile contrast material and continuous monitoring of the patient. The risk of this complication might not be a concern if diluted contrast material mixed with intravenous antibiotics is injected into the biliary tract (26).
Third, the additional value of multi–detector row CT in tumor staging was not considered in this study. The role of multi–detector row CT itself in staging hilar cholangiocarcinoma was not evaluated. Finally, we did not compare CT cholangiography with MR cholangiography because patients with preliminary biliary drainage were enrolled in this study. However, further comparative studies are needed to establish which examination is the most cost-effective, diagnostic, and therapeutic.
Although our study was preliminary and limited by the small number of patients examined, 3D direct multi–detector row CT cholangiography was feasible for defining the extent of ductal involvement in patients with hilar cholangiocarcinoma. We believe that this technique may be a promising diagnostic tool for situations in which either PTBD or ENBD is frequently performed for various purposes in these patients.
|
FOOTNOTES |
|---|
Abbreviations: ENBD = endoscopic nasobiliary drainage • PTBD = percutaneous transhepatic biliary drainage • 3D = three-dimensional
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, A.Y.K.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, H.J.K., A.Y.K., S.S.H., Y.M.S.; clinical studies, H.J.K., M.H.K., J.H.B., H.J.W.; statistical analysis, H.J.K.; and manuscript editing, H.J.K., A.Y.K., P.N.K., H.K.H., M.G.L.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. Hoeffel, L. Azizi, M. Lewin, V. Laurent, C. Aube, L. Arrive, and J.-M. Tubiana Normal and Pathologic Features of the Postoperative Biliary Tract at 3D MR Cholangiopancreatography and MR Imaging RadioGraphics, November 1, 2006; 26(6): 1603 - 1620. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
