| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Seoul National University Hospital and College of
Medicine, 28, Yongon-dong, Chongno-gu, Seoul, 110-744, Republic of
Korea.
2 Institute of Radiation Medicine, Seoul National University College of
Medicine, Seoul, Korea.
3 Present address: Department of Radiology, Yonsei University College of
Medicine, Seoul, Korea.
4 Department of Radiology, Konkuk University College of Medicine, Seoul,
Korea.
5 Department of Radiology, Seoul Medical Center, Seoul, Korea.
Received March 24, 2007;
accepted after revision August 16, 2007.
Address correspondence to J. M. Lee
(leejm{at}radcom.snu.ac.kr).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. From January 2003 to March 2006, 27 patients (18 men, nine women; mean age, 60.8 years; range, 43–80 years) with surgically proven hilar cholangiocarcinoma or common bile duct (CBD) cancer who had undergone preoperative 2D and 3D MRCP with gadolinium-enhanced MRI and triple-phase MDCT with direct cholangiography (ERCP or percutaneous transhepatic cholangiography) were included in this retrospective study. Two experienced radiologists independently reviewed the two image sets. These readers evaluated the longitudinal extent of the tumor for involvement of the secondary confluence of both intrahepatic ducts and the intrapancreatic CBD, vascular involvement of the tumor, lymph node metastasis, and tumor resectability. The radiologists' performance was evaluated by calculation of sensitivity, specificity, and overall accuracy. Correlation was made with the resected specimens or findings at surgical exploration.
RESULTS. For each reviewer, the overall accuracy rates for predicting involvement of the bilateral secondary biliary confluences and the intrapancreatic CBD were 90.7% and 87.0% for MRI with MRCP and 85.1% and 87.0% for MDCT with direct cholangiography. The differences were not statistically significant for either image set for either reviewer (p > 0.05). In the assessment of vascular involvement, lymph node metastasis, and tumor resectability, the readers' diagnostic performance using MRI with MRCP was similar to that with MDCT with direct cholangiography (p > 0.05).
CONCLUSION. In the diagnosis of bile duct cancer with a noninvasive procedure, the information regarding tumor extent and resectability obtained with contrast-enhanced MRI combined with MRCP is comparable with that obtained with MDCT with direct cholangiography.
Keywords: bile duct neoplasm • MDCT • MR cholangiopancreatography • MRI
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Sonography and CT traditionally have been used as noninvasive imaging techniques to evaluate patients with suspected malignant disease of the biliary tract [3]. Despite its invasiveness, direct cholangiography, including ERCP and percutaneous transhepatic cholangiography (PTC), also is used to visualize the presence and level of biliary obstruction and to suggest the cause of the obstruction [3]. MR cholangiopancreatography (MRCP) is the most recently developed noninvasive imaging technique for evaluating biliary disease [4, 5], whereas cross-sectional MRI provides important diagnostic information on tumor characteristics, vascular invasion, and metastasis [6]. Advances in MRI technique, such as 3D MRCP and the 3D gradient-recalled echo (GRE) sequence, have resulted in improved spatial resolution and better image quality [7–11].
In one study [12], the investigators found that combined interpretation of CT and direct cholangiographic images had favorable accuracy (74.5%) in assessment of the resectability of hilar cholangiocarcinoma. Direct cholangiography has several limitations, however, such as invasiveness, lack of depiction of the entire biliary tree, and need for contrast material. MRCP can overcome these limitations of direct cholangiography. If it can be proved to perform similarly to combined CT and direct cholangiography, the combination of MRCP and contrast-enhanced MRI is expected to be an acceptable alternative in the evaluation of bile duct cancer. The purpose of our study was to compare the diagnostic performance of MRCP combined with contrast–enhanced cross-sectional MRI with that of MDCT combined with direct cholangiography in the assessment of tumor extent and resectability in the evaluation of hilar cholangiocarcinoma and distal CBD cancer. The reference standard was surgical and histopathologic findings.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CT Technique
All patients underwent triple-phase CT, which consisted of unenhanced early
arterial, late arterial, and portal venous phases. All CT scans were obtained
with one of the following commercially available MDCT scanners: LightSpeed (GE
Healthcare) (n = 13), MX 8000 (Marconi Medical Systems) (n =
8), Sensation 16 (Siemens Medical Solutions) (n = 6). The scanning
parameters for each of CT scanner are shown in
Table 1.
|
For unenhanced imaging, sections 2.5–5 mm thick were acquired. A dose of 120 mL of nonionic contrast material (iopromide, Ultravist 370, Bayer HealthCare) was administered with a power injector (Multilevel CT, Medrad) at a rate of 3 mL/s through an 18-gauge plastic IV catheter placed in an antecubital vein. The contrast material was followed by a 20-mL flush of sterile saline solution. For early arterial phase imaging, the scanning delay was determined with an automatic bolus tracking technique provided by the manufacturer of the CT system. Contrast enhancement was automatically calculated by placement of the region-of-interest cursor over the vessel of interest, that is, the abdominal aorta, and the level of the trigger threshold was set at an increase of 100 H. Early arterial phase scans were obtained automatically 6 seconds after the trigger threshold was reached. The early and late arterial phase images were acquired separately during each breath-hold with a minimum interscan delay of 5–9 seconds. The mean scanning delay was 23 seconds for the early arterial phase and 37–45 seconds for the late arterial phase. Hepatic venous phase scans were obtained 70 seconds after triggering. For this triple-phase abdominal CT, the estimated effective dose was 13–14 mSv.
Direct Cholangiographic Technique
Direct cholangiographic images were available for 27 patients (PTC,
n = 18; ERCP, n = 9). These cholangiographic techniques were
used for therapeutic drainage rather than for diagnostic purposes. ERCP was
performed by one of two attending gastroenterologists using standard
techniques and fiberoptic endoscopes. The patients were in the left-sided
position under mild IV analgesia and sedation with 5–10 mg of midazolam
and 30 of mg pentazocine. The papilla of Vater was cannulated with a 5-French
catheter, and 10–15 mL of contrast medium was injected. After
cannulation, the patient was placed in a supine position to allow more
accurate evaluation of the intrapancreatic CBD. Four patients underwent
endoscopic retrograde biliary drainage, two of them after CT and before MRI
and the other two after all imaging.
PTC and biliary drainage procedures were performed by two experienced radiologists in an angiography suite; these procedures were guided by both sonography and fluoroscopy. The patients underwent local anesthesia and IV sedation with fentanyl. A 21-gauge Chiba needle was used for cholangiography. A 0.018-inch guidewire was inserted into the bile duct through the Chiba needle. Placement of a 5-French yellow sheath and a 0.035-inch guidewire was followed by insertion of an 8.5-French biliary drainage catheter. Biliary stents were placed in two of the patients. The estimated radiation dose for direct cholangiography varied according to the type of procedure and the operators' experience; it ranged from 0.5 to 2.0 mSv.
MRI Technique
All MRI was performed on 1.5-T super-conducting systems (Magnetom Vision
Plus, Siemens Medical Solutions, n = 4; Sonata, Siemens Medical
Solutions, n = 31) with a four-channel phased-array torso coil. MRCP
examinations were performed with thick-slab T2-weighted turbo spin-echo (TSE)
and thin-slab multisection HASTE sequences. Thick-slab T2-weighted TSE MRCP
images were obtained in the coronal plane, and thin-slab T2-weighted HASTE
MRCP images were obtained in both the coronal and transverse planes. At least
five thick-slab T2-weighted TSE MRCP images were obtained with coronal and
± 15° and 30° oblique coronal angles. Thin-slab T2-weighted
HASTE MRCP images were obtained with 15 sections acquired per breath-hold
(volume of coverage, 60 mm3). Other imaging parameters are
described in Table 2.
|
Three-dimensional MRCP was performed on 24 of the 27 study patients. All 3D MRCP was performed on a Sonata system, and fat saturation was used. For 3D MRCP, the following parameters were used (Table 2): 60 partitions; interpolated partition thickness, 1 mm; typical voxel size, 1 x 1 x 1 mm with coronal orientation; minimum imaging time, 5 minutes. The 3D MRCP images were reconstructed with a maximum-intensity-projection algorithm. Respiratory monitoring was performed with navigator echoes. For the 12 patients who underwent a biliary drainage procedure before MRI, 20 mL of saline solution was infused through the drainage tube for better visualization of the overall ductal structure.
Transverse T2-weighted HASTE images were obtained; the parameters are shown in Table 2. Unenhanced T1-weighted imaging was performed with in-phase and opposed-phase spoiled GRE [14] techniques and a 3D GRE technique (volumetric interpolated breath-hold examination [VIBE]) with fat saturation. All T1-weighted images were acquired in the transverse plane. Two-dimensional dual-echo GRE (in phase, opposed phase) sequences were performed during two suspended breath-holds. Twenty to 30 sections were obtained to cover the entire liver and pancreas for the in-phase and opposed-phase 2D GRE sequences. Other imaging parameters for these two sequences are summarized in Table 2. Dynamic images were obtained with a gadolinium-enhanced 3D fast GRE sequence with intermittent fat-suppression pulse after administration of 0.1 mmol of gadobenate dimeglumine (MultiHance, Bracco) per kilogram of body weight at an injection rate of 2 mL/s. The 36–40 partitions interpolated to 72–80 with a partition thickness of 5 mm. Arterial, venous, and equilibrium phase images were obtained serially 20–35 seconds, 45–60 seconds, and 3 minutes after contrast injection. Other imaging parameters are shown in Table 2.
Image Analysis
All images were reviewed on a PACS workstation monitor (m-view, Marotech)
by two board-certified radiologists. These radiologist had worked mainly as
hepatobiliary radiologists and had interpreted abdominal CT and MR images of
the liver as part of their daily clinical and research practice (11 years and
3 years of experience). They independently reviewed the two image sets in two
steps, that is, the CT with cholangiography imaging set and the MRCP with
dynamic MRI set. To minimize learning bias, there was a 2-week interval
between the two readings, and the images were randomly presented whether they
were part of the CT with combined cholangiography imaging set or the dynamic
MRI with combined MRCP set. That is, there was no order between the two data
sets. Both reviewers were partially blinded to the pathologic results. They
knew, however, that a potentially resectable malignant biliary tumor had been
initially detected on preoperative imaging and confirmed at surgical resection
and pathologic examination. The radiologists reviewed the images using the
cine mode on the PACS viewer.
On the CT images, biliary involvement was considered present if there was irregular ductal wall thickening with asymmetric upstream intrahepatic ductal dilatation, if the ductal wall was hyperattenuating compared with the liver, or if an intraductal soft-tissue mass or thickened ductal wall obliterated the lumen [15]. On direct cholangiographic images, irregular bile duct obliterating the lumen with or without asymmetric upstream dilatation of the intrahepatic duct was considered involved [16]. On MR images, loss of continuity of the bile duct or ductal obstruction, abrupt and irregular narrowing of the distal segment and prestenotic biliary dilatation, irregularly shaped intraluminal filling defects, and segmental enhancing wall thickening of the bile duct were considered findings indicating biliary involvement [17]. The reviewers assessed the longitudinal extent of the tumor in terms of involvement of the secondary confluence of both intrahepatic ducts and of the intrapancreatic CBD.
The diagnostic criteria for vascular involvement on CT and MR images included vessel occlusion, stenosis or contour deformity, and more than half of the perimeter in contact with the tumor [18]. Lymph node metastasis was considered present on CT and MR images when the short-axis diameter was longer than 10 mm or when central necrosis was present with any size or its attenuation or signal intensity was greater than that of liver parenchyma in the portal venous phase [12, 19].
The reviewers determined tumor resectability according to the previously reported revised criteria for hilar cholangiocarcinoma [12]. Criteria for unresectability were as follows: Bismuth type IV lesion and tumor extending farther than 2 cm from the hilum; invasion of the main portal vein or the proper hepatic artery with the involved segment longer than 2 cm; atrophy of one hepatic lobe with contralateral vascular invasion; atrophy of one hepatic lobe with contralateral tumor extension to the second biliary confluence; invasion of the second biliary confluence on one lobe and contralateral vascular invasion; metastasis to celiac, portacaval, or paraaortic lymph nodes; distant metastasis.
Standard of Reference
For all patients, the final diagnosis was established during surgery.
Surgery was performed by two board-certified surgeons, each of whom had worked
primarily as a hepatobiliary specialist for more than 10 years. Exposure of
the biliary confluence and assessment of vascular involvement were
accomplished during surgery, and histologic assessment of the resection margin
was performed by the surgeons. Full exploration was followed by curative or
palliative surgery. Because 11 patients were found to have unresectable tumor
at surgery, a palliative operation was performed. Brief information on each of
the study patients and the surgical findings is presented in
Table 3.
|
Statistical Analysis
Preoperative assessments of biliary, vascular, and nodal involvement and
overall resectability were compared with the surgical findings and the final
pathologic results. For both interpretation sets, we evaluated the individual
performance of each radiologist on CT and MRI with respect to the diagnosis of
tumor involvement of the secondary confluence of both intrahepatic ducts and
the intrapancreatic CBD, vascular involvement, lymph node status, and
resectability.
The sensitivity, specificity, and diagnostic accuracy of each technique were calculated, and the accuracy rates of the two techniques for each reviewer were obtained with a nonparametric method in MedCalc software for Windows (MedCalc Software). We also calculated the perpatient accuracy for interpretation of each image set in terms of longitudinal tumor extent. The McNemar test and the same software were used to compare the accuracy of CT combined with cholangiography with that of MRI combined with MRCP and dynamic axial images. A value of p < 0.05 was considered to indicate a statistically significant difference.
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Prediction of Biliary and Vascular Involvement and Lymph Node Metastasis
The sensitivity, specificity, and overall accuracy for each analysis item
for both reviewers are shown in Figure
1A,
1B,
1C and
Table 4. In cases of bilateral
hilar involvement, for reviewer 1, there was suggestion of a trend favoring
the MRI set rather than the CT set in the assessment of tumor involving
bilateral secondary biliary confluences (90.7% for MRI and 85.1% for CT)
(p > 0.05). For reviewer 2, the accuracy rates of the techniques
were the same (87.0% each) (p > 0.05). In cases of intrapancreatic
CBD involvement, the MRI set and the CT set had comparable accuracy for both
reviewers (p > 0.05). Furthermore, with respect to assessment of
hepatic arterial and portal venous involvement, both CT and MRI sets showed
similar diagnostic performance (p > 0.05). The incidence of lymph
node metastasis was 34.3% (12 of 35 patients), and the two image sets were not
significantly different in terms of performance.
|
Prediction of Tumor Extent and Resectability
With respect to longitudinal tumor extent evaluation, the overall accuracy
of the MRI set (22/27 [81.5% for reviewer 1; 21/27 [77.8%] for reviewer 2) was
roughly equivalent to that of the CT set (21/27 [77.8%] for reviewer 1; 20/27
[74.1%] for reviewer 2) (p = 0.625; p = 0.899). In the tumor
resectability analysis, MRI had the same overall accuracy as CT for reviewer 1
(both techniques, 77.8%), and MRI had slightly higher accuracy (77.8%) than CT
(74%) for reviewer 2. The differences between MRI and CT did not attain
statistical significance in the evaluation of resectability of bile duct
cancer.
Interobserver Agreement
With respect to interobserver agreement, biliary tumor involvement and
vascular involvement had overall good agreement (weighted 
=
0.654–0.784), and lymph node metastasis and tumor resectability had fair
to moderate agreement (weighted = 0.305–0.643). There was a
trend favoring CT rather than MRI in interobserver agreement with respect to
evaluation of the secondary biliary confluence (weighted = 0.784 vs
0.696), portal venous involvement (weighted = 0.675 vs 0.654), lymph
node status (weighted = 0.643 vs 0.305), and tumor resectability
(weighted = 4.64 vs 0.403). However, there was a tendency of better
agreement on MRI than on CT with regard to evaluation of the intrapancreatic
CBD (weighted = 0.767 vs 0.684) and hepatic arterial involvement
(weighted = 0.770 vs 0.715). The differences did not reach statistical
significance (p > 0.05).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
|
|
|
|
|
We used a breath-hold, 3D GRE technique (VIBE) for dynamic MRI. Earlier studies had shown that the VIBE technique allowed better spatial resolution by minimizing the partial volume effect and allowing more efficient postprocessing compared with the 2D GRE sequence. The previously reported accuracy of MRCP in determining the extent of biliary tumors ranged from 71% to 96% [16, 22, 24, 25]. Our results were better than or comparable with those in the previous reports. Addition of contrast-enhanced dynamic images to MRCP allowed better diagnostic performance and interobserver agreement for assessment of the longitudinal tumor extent of bile duct cancer [26]. The extent of bile duct wall thickening, degree of enhancement, and hepatic parenchymal tumor extension also are criteria used to determine the presence of biliary malignancy [17]. We believe that the dynamic MRI–MRCP protocol with improved spatial resolution used in our study contributed to its comparable diagnostic performance with CT. However, although MRI with MRCP can be used instead of CT with direct cholangiography, the latter still has a role in cases in which preoperative biliary drainage is needed.
In the assessment of vascular involvement, our study showed at least the same or slightly greater accuracy of CT than of MRI, although the results were not statistically significant (Fig. 4A, 4B). Dynamic imaging with improved spatial resolution is assumed to contribute to this comparable MRI performance for vascular evaluation. In earlier studies [27, 28], MR angiography was used for the evaluation of vascular invasion in pancreatobiliary malignant disease, and the results showed a slightly higher accuracy than our MRI results. Addition of an MR angiographic sequence to axial MRI may improve the performance and potential to replace CT for vascular evaluation.
|
|
In the tumor resectability assessment, both techniques showed similar diagnostic performance (p > 0.05). Diagnostic inaccuracy in our study was primarily caused by underestimating bile duct involvement; missing tumor infiltration along the hepatoduodenal ligament; and missing paraaortic lymph node metastasis, inferior vena caval invasion, and subtle peritoneal seeding. This inaccuracy caused misinterpretation of unresectable cases as resectable (false-positive), resulting in low specificity for tumor resectability. Underestimation of biliary extension occurred mainly on CT rather than on MRI, whereas subtle peritoneal seeding and vascular invasion were more frequently missed on MRI. A switch from curative to palliative surgery occurred in four cases. Two cases were preoperatively diagnosed as a Bismuth type IIIa lesion, but the resection margin for the left secondary confluence was positive. One case was preoperatively diagnosed as a type IIIb lesion, but the right secondary confluence was determined to be involved at surgery. The stage was underestimated with both imaging techniques in these cases, probably as the result of the low sensitivity of CT and MRI in the detection of microscopic tumor infiltration. In the fourth case, peritoneal seeding was missed on MRI by one of the reviewers and on both MRI and CT by the other reviewer. In our study, overestimates of biliary involvement status occurred mainly in the intrapancreatic CBD rather than in the hilar or intrahepatic bile duct. Although reports [17, 31] have suggested that CT or MRI can be used to differentiate malignant from benign CBD abnormalities, it is still a challenge in our daily practice.
To our knowledge, there has not been a report such as ours that compares the full set of preoperative imaging techniques for bile duct cancer while applying a strict standard of reference and introducing a state-of-the-art MRI technique such as 3D MRCP and 3D GRE. There were, however, several limitations to our study. First, because our study population included only patients who had undergone surgery, many cases of tumors in inoperable stages were excluded that probably would have been more easily diagnosed correctly. Therefore, selection bias was present, and the overall diagnostic performance results might have been emphasized more so than in daily clinical practice. Second, direct cholangiography was performed for therapeutic purposes, and sometimes it did not yield full evaluation of the biliary system. In those cases, there was no real correlation with the very complete MRCP examination. However, direct cholangiography could not be performed on the obstructed segments in those patients owing to risk of sepsis. Third, this study was retrospective, and data from various CT and MRI models with slightly different protocols were used in the analysis. Fourth, the 2-week interval between the two image interpretation sessions might not have been long enough to minimize learning bias. However, in light of the minute nature of the bile duct tumors, the interval between the two image interpretation sessions might not have affected the results of the diagnostic performance on the CT or MRI examinations in regard to determination of resectability. Furthermore, to compensate for this potential problem of learning bias, the images were presented randomly during the interpretation session. Finally, despite the interval between the interpretations of the two image sets, there may have been recall bias in the subsequent analysis.
In our study, in the evaluation of the resectability of bile duct cancer, the diagnostic performance of gadolinium-enhanced MRI combined with MRCP was generally comparable with that of CT combined with direct cholangiography. Therefore, because it yields information comparable with that obtained on CT combined with cholangiography, MRI with MRCP can be used as an alternative imaging technique to establish the diagnosis of bile duct cancer. MRI with MRCP has the additional benefits of having a shorter preoperative diagnostic evaluation period, being a noninvasive procedure, and not exposing patients to radiation.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  Y. E. Chung, M.-J. Kim, Y. N. Park, J.-Y. Choi, J. Y. Pyo, Y. C. Kim, H. J. Cho, K. A. Kim, and S. Y. Choi Varying Appearances of Cholangiocarcinoma: Radiologic-Pathologic Correlation1 RadioGraphics, May 1, 2009; 29(3): 683 - 700. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J.-Y. Choi, M.-J. Kim, J. M. Lee, K. W. Kim, J. Y. Lee, J. K. Han, and B. I. Choi Hilar Cholangiocarcinoma: Role of Preoperative Imaging with Sonography, MDCT, MRI, and Direct Cholangiography Am. J. Roentgenol., November 1, 2008; 191(5): 1448 - 1457. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
