| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1
Department of Radiology, University of Pittsburgh Medical Center, 200 Lothrop
St., Pittsburgh, PA 15213-2582.
2
Department of Medicine, University of Pittsburgh Medical Center, Pittsburgh,
PA 15213-2582.
3
Present address: Real Hospital Português, Setor
de Endoscopie (Endoscopie-Centro de
Diagnóstico e Tratamento), Av. Portugal 163,
Derby-Recife PE, 51010-010 Brazil.
4
Present address: Department of Radiology, University of Brescia, 1 via
Valsabbina, Brescia 25100, Italy.
5
Present address: Department of Radiology, Yamanashi Medical University,
Tamaho-cho, Nakakoma-gun, Yamanashi, Japan 409-3815.
6
Present address: Department of Radiology, Osaka University Medical School, 2-2
Yamadaoka, Suita Osaka, Japan 565-0871.
7
Department of Biostatistics, Graduate School of Public Health, University of
Pittsburgh, Pittsburgh, PA 15261.
Received May 26, 2000;
accepted after revision May 1, 2001.
E. S. Siqueira was supported by a grant from the Fundacao Coordenacao de
Aperfeicoamenta de Pessoal de Nivel Superior (CAPES), Brazil, process number
BEX0752/96-7.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. One hundred thirteen abdominal CT examinations and cholangiograms in 45 patients with primary sclerosing cholangitis, including 18 patients with established biliary tract carcinoma, were analyzed for tumor. Four radiologists who were unaware of the presence or absence of carcinoma rated each study as to the probability of malignancy. Receiver operating characteristic curve analysis was used to assess the diagnostic performance of CT and cholangiography, the value of imaging signs, and the degree of inter-observer variation in interpretation. Sensitivity and specificity values were calculated.
RESULTS. CT outperformed cholangiography in the detection of
carcinoma. The average area under the receiver operating characteristic curve
was 0.82 for CT and 0.57 for cholangiography (p = 0.003). Sensitivity
and specificity for detecting carcinoma using CT were good, with average
values of 82% and 80%, respectively. Average sensitivity and specificity for
cholangiography were 54% and 53%, respectively. The most reliable sign of
tumor on CT was a discrete mass. Progressive biliary dilatation on sequential
studies was the most useful sign on cholangiography. Interobserver agreement
assessed using the Cronbach 
was fair for cholangiography and good for
CT.
CONCLUSION. CT provides good sensitivity and specificity and significantly outperforms cholangiography in detecting biliary tract carcinoma complicating primary sclerosing cholangitis. Despite limitations, CT and cholangiography provide useful information not otherwise available in the treatment of patients with primary sclerosing cholangitis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Cholangiography has traditionally been used to diagnose and follow up patients with primary sclerosing cholangitis for the progression of disease and the development of cholangiocarcinoma [4]. In recent years CT has been more widely used, especially in the setting of end-stage liver disease. CT and cholangiography are the imaging modalities most relied on to detect biliary tract carcinoma complicating primary sclerosing cholangitis. The imaging features of such tumors have been described [4, 5].
Uncertainty as to the presence of tumor can arise because the imaging findings in primary sclerosing cholangitis may mimic those of cholangiocarcinoma. Although such tumors are often visible on imaging studies, to our knowledge no study has examined the efficacy of imaging using methodology in which the interpreters of imaging examinations were unaware of the presence or absence of malignancy. Such an approach should provide a more realistic assessment of the value of imaging and permit a more accurate comparison of the diagnostic capabilities of CT and cholangiography. We therefore performed a study using unaware observers and receiver operating characteristic curve analysis to evaluate the efficacy of CT and cholangiography for detecting biliary tract carcinoma complicating primary sclerosing cholangitis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A pathologic diagnosis of carcinoma was established by analysis of whole liver specimens obtained at liver transplantation in eight patients, percutaneous needle or open biopsy in six patients, endoscopic brush biopsy specimens of bile ducts for cytology examination in two patients, and cholecystectomy specimens in two patients.
The CT scanners used were HiSpeed Advantage, Advantage, or 9800 model units (General Electric Medical Systems, Milwaukee, WI). Because studies were performed over a 9-year period, scanning techniques varied. Contrast-enhanced CT was most commonly done using 150 mL of iothalamate meglumine (Conray-60; Mallinckrodt Medical, St. Louis, MO), iopamidol (Isovue 300; Squibb Diagnostics, Princeton, NJ), or ioversol (Optiray 320; Mallinckrodt Medical) given as a bolus injection with a mechanical power injector (Medrad, Pittsburgh, PA). The rate of injection was 2-5 mL/sec with a dynamic incremental or helical technique. For the helical technique, a pitch of 1.0-1.5 was used. Section thickness ranged from 5 to 10 mm, and in most cases was 5 or 7 mm.
Cholangiography was performed in the standard fashion with the use of endoscopic retrograde and percutaneous transhepatic techniques. Because cholangiography was performed with standard technique by a number of interventional radiologists and gastroenterologists, the cholangiography techniques were not controlled.
Four experienced abdominal radiologists independently interpreted the imaging studies. The interpreters were informed that all patients had primary sclerosing cholangitis and that some had biliary tract carcinoma. Interpreters were unaware of the number of patients with carcinoma versus control patients; other clinical, laboratory, or pathology data; or the results of other imaging examinations. CT scans and cholangiograms of the same patient were interpreted in separate sessions conducted days to weeks apart. Sequential studies done with either modality were interpreted together as one evaluation to allow the assessment of interval change. CT scans were evaluated for the presence of the following features: marked biliary dilatation; progressive biliary dilatation (worsening biliary dilatation on sequential CT scans); bile duct wall thickening greater than 4 mm; presence of mass; presence of multiple masses; delayed enhancement of mass; and location of mass, if present. Cholangiograms were evaluated for the presence of a dominant biliary stricture (single stricture disproportionately severe relative to other strictures); a progressive biliary stricture (worsening stricture on sequential cholangiograms); marked biliary dilatation; progressive biliary dilatation (worsening biliary dilatation on sequential cholangiograms); biliary filling duct, either infiltrating or polypoid; gallbladder mass; and location of filling defect or mass, if present. Confidence in each finding was scored on a five-point scale: 0, not present; 1, probably not present; 2, possibly present; 3, probably present; and 4, definitely present.
One hundred thirteen imaging studies—57 cholangiograms and 56 CT examinations—were reviewed, including 23 cholangiograms and 24 CT studies in patients with carcinoma and 34 cholangiograms and 32 CT studies in control patients. Eleven patients—four patients with carcinoma and seven control subjects—had more than one cholangiogram, 10 had two cholangiograms, and one had three cholangiograms. Eleven patients—six with carcinoma and five control patients—had two CT examinations.
Receiver operating characteristic (ROC) curve analysis was performed.
Estimates of the areas under ROC curves (Az) were calculated using
a nonparametric method [6,
7]. Diagnostic capability was
determined by the area under the ROC curve for each modality, imaging feature,
or interpreter. Results were expressed in the form of the mean plus or minus
one standard deviation. To determine interobserver variability in diagnostic
performance, statistical comparison between the areas under the ROC curves for
each observer was done to determine the Cronbach , a measure of
agreement. Test of the difference between the areas under the ROC curves for
the imaging features using cholangiography and CT was performed using the
paired Student's t test. Values for p of less than 0.05 were
considered statistically significant.
Diagnostic scores for the detection of carcinoma were also used to determine sensitivity and specificity for each observer. Scores of 0 or 1 were considered an interpretation of no carcinoma, whereas scores of 3 or 4 indicated the presence of carcinoma. A score of 2, or possible carcinoma, was considered an invalid interpretation. Sensitivity was calculated as (number of correct diagnoses of carcinoma / number of proven carcinomas) x 100; and specificity, as (number of correct diagnoses of no carcinoma / number of proven patients without carcinoma) x 100.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
The Az values for each individual observer's ROC curves for individual diagnostic criteria for carcinoma using CT are shown in Table 2. The mean area under the ROC curve for carcinoma present was 0.82 ± 0.07; for marked biliary dilatation, 0.63 ± 0.09; for progressive biliary dilatation, 0.63 ± 0.17; and for bile duct wall thickening greater than 4 mm, 0.69 ± 0.09. The direct visualization of a mass was a significantly more reliable criterion than was the indirect sign of marked biliary dilatation (p < 0.05). The CT finding of bile duct wall thickening greater than 4 mm was significantly associated with carcinoma in the ratings of only two of the four observers.
|
The Az values for each individual observer's ROC curves for the individual diagnostic findings of cholangiography are shown in Table 3. Mean area under the ROC curve was 0.82 ± 0.12 for progressive biliary dilatation, 0.74 ± 0.14 for a progressive biliary stricture, 0.59 ± 0.09 for marked biliary dilatation, and 0.57 ± 0.08 for a dominant biliary stricture. Among these criteria, only progressive ductal dilatation (Fig. 1A,1B,1C,1D) was significantly more accurate than the others (p < 0.01).
|
|
|
|
|
Sensitivity and specificity for the detection of carcinoma were calculated for the four observers (Table 4). Using CT, the mean sensitivity of the observers was 82% and mean specificity was 80%. For cholangiography, the scores of the individual observers varied more; mean sensitivity was 54% and mean specificity was 53%.
|
The interobserver agreement in assigning confidence levels for
abnormalities was assessed using the Cronbach . The coefficient alpha
value was 0.86 for CT and 0.75 for cholangiography. Coefficient alpha values
of less than 0.6 are considered poor agreement, 0.6-0.8 are considered fair
agreement, 0.8-0.9 are considered good agreement, and 0.9-1.0 are considered
excellent agreement. Therefore, these results indicate fair interobserver
agreement for cholangiography and good interobserver agreement for CT.
The mean score for the presence of gallbladder carcinoma was greater for CT (2.8) than for cholangiography (0.7), but given the small number of cases, the difference was not statistically significant.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
The limitations of cholangiography for diagnosing biliary tract carcinoma were seen in our patients. These shortcomings stem primarily from the difficulty in distinguishing benign dominant strictures of primary sclerosing cholangitis from malignant strictures caused by cholangiocarcinoma. Only progressive biliary dilatation on serial examinations (Fig. 1A,1B,1C,1D) was significantly better than the other cholangiographic signs of dominant stricture, marked biliary dilatation, or worsening stricture on serial studies (Fig. 3A,3B). Although progressive ductal dilatation is an imperfect sign of complicating carcinoma, its detection should prompt careful correlation with CT and with directed endoscopic brushings cytology.
|
|
Clearly, the diagnosis of biliary tract carcinoma in patients with primary sclerosing cholangitis remains challenging. Although the interobserver variability noted in our study could relate to differences in observer experience, this variability may also reflect the intrinsic difficulty of diagnosing cholangiocarcinoma in the setting of primary sclerosing cholangitis. The diagnosis of biliary tract carcinoma may also be difficult by clinical evaluation, brushings cytology of dominant bile duct strictures, or serial measurements of serum tumor markers. In several studies, brushings cytology was shown to have limited sensitivity, and tumor markers were found to have low specificity [8,9,10]. However, recent experience at our medical center has shown that serial brushings cytology has sensitivity and specificity approaching 50% and 100%, respectively, and that serum tumor markers have diagnostic accuracy of 75-95% [11]. Although such an investigation is beyond the scope of our present study, a prospective study of the combination of serial brushings cytology, serum tumor markers, and CT scan findings may determine a more true accuracy for the noninvasive diagnosis of biliary tract malignancy in primary sclerosing cholangitis.
With regard to imaging, some added improvement in diagnostic accuracy might be achieved if serial high-quality CT using state-of-the-art scanners and optimized contrast material administration, including biphasic (arterial and portal venous) contrast-enhanced CT and delayed imaging, were more widely used. A recent report of the usefulness of multiphasic helical CT in the staging and diagnosis of hilar cholangiocarcinoma showed statistically greater enhancement of infiltrating stenotic lesions on hepatic arterial phase imaging and statistically greater enhancement of exophytic hilar lesions on portal venous phase imaging [12]. Although that study was not limited to primary sclerosing cholangitis patients and did not correlate multiphasic CT findings with delayed phase imaging or cholangiography, some improvement in the CT diagnosis of biliary tract carcinoma, particularly infiltrating stenotic lesions, would likely be expected if primary sclerosing cholangitis patients were imaged with this technique. MR imaging, including MR cholangiopancreatography, is theoretically attractive because it provides both cross-sectional imaging and noninvasive cholangiography. Increased reliance on MR cholangiopancreatography might allow less frequent use of endoscopic retrograde cholangiography. At present, however, endoscopic cholangiography remains necessary to perform brushings cytology and stricture dilatation. A new imaging approach to detecting carcinoma in primary sclerosing cholangitis—positron emission tomography—has been recently described and awaits confirmation regarding its possible role [13].
The discovery of biliary tract carcinoma in a patient with primary sclerosing cholangitis is obviously an important development that mandates a change in treatment. Liver transplantation has emerged as an effective treatment for patients with primary sclerosing cholangitis and end-stage liver disease. However, differences of opinion exist among transplantation surgeons as to whether liver transplantation is appropriate in the presence of a complicating biliary tract carcinoma. Although patients with biliary tract carcinoma that is discovered incidentally in hepatectomy specimens after transplantation can have improved survival [14], posttransplantation survival of patients with cholangiocarcinoma is markedly decreased compared with primary sclerosing cholangitis transplant recipients without cholangiocarcinoma. Some transplantation surgeons will consider transplantation when a cholangiocarcinoma is present if no evidence is seen of extrahepatic tumor. In such cases, transplantation may be done if frozen section biopsies of bile duct margins and hilar lymph nodes at time of surgery have negative findings. Other surgeons believe that carcinoma precludes transplantation because of the risk of tumor recurrence after surgery. Thus, the value of imaging will depend in part on how the results will be used by the referring physicians. CT can help avoid futile surgery for patients with advanced carcinoma and can guide the surgeon to areas suspicious for tumor in less advanced cases. Although in our study interpretations of possible carcinoma (score = 2) were considered to be incorrect diagnoses for calculations of sensitivity and specificity, in clinical practice the finding of a possible mass may be useful. For example, a patient with such a mass might be scheduled for liver transplantation, but with a second potential recipient available to receive the donor liver should tumor be found at surgery.
Our study has several limitations. The use of observers who are specialists in abdominal imaging, some with extensive experience in hepatobiliary imaging, tends to exaggerate the value of CT and cholangiography. Also, observers were necessarily aware that carcinoma was present in a high proportion of the patients, a condition not likely in routine practice. Tending to understate the value of imaging was the lack of opportunity for the observers to correlate CT and cholangiographic findings in the same patient. Finally, brushings cytology was the only proof of carcinoma in two patients.
Another topic of interest that is beyond the scope of our retrospective assessment of CT and cholangiographic signs for the diagnosis of cholangiocarcinoma in primary sclerosing cholangitis is the usefulness of these modalities for screening patients with primary sclerosing cholangitis for the development of biliary tract tumors. Imaging, brushings cytology, and serum tumor markers are now being used prospectively in screening studies of primary sclerosing cholangitis patients who have an estimated 15% incidence of biliary tract carcinoma. Clearly, this important topic of assessing the practical usefulness of these imaging findings in all patients with primary sclerosing cholangitis, both with and without cholangiocarcinoma, merits further research with a controlled prospective study.
In summary, CT provides fairly good sensitivity and is significantly superior to cholangiography for detecting biliary tract carcinoma in primary sclerosing cholangitis. Direct visualization of a liver, bile duct, or gallbladder mass on CT is the most useful imaging sign. A dominant stricture on cholangiography is not a reliable indicator of carcinoma, but progressive marked biliary dilatation is suspicious for complicating malignancy. The potential exists for considerable variation in the interpretation of imaging findings. CT and cholangiography are imperfect methods for detecting complicating biliary tract carcinoma, but they nevertheless provide valuable information not otherwise attainable in patients being monitored for primary sclerosing cholangitis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  C. O. Menias, V. R. Surabhi, S. R. Prasad, H. L. Wang, V. R. Narra, and K. N. Chintapalli Mimics of Cholangiocarcinoma: Spectrum of Disease1 RadioGraphics, July 1, 2008; 28(4): 1115 - 1129. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
