Curved Planar Reformatted Images of MDCT for Differentiation of Biliary Stent Occlusion in Patients With Malignant Biliary Obstruction

Transcription

1 Gastrointestinal Imaging Original Research Bang et al. MDCT of Biliary Stent Occlusion Gastrointestinal Imaging Original Research Byoung Wook Bang 1 Seok Jeong 1 Don Haeng Lee 2 Chul Hyun Kim 1 Soon Gu Cho 3 Yong Sun Jeon 3 Bang BW, Jeong S, Lee DH, Kim CH, Cho SG, Jeon YS Keywords: CT technique, Hounsfield unit, MDCT, self-expandable metallic stent DOI: /AJR Received May 15, 2009; accepted after revision November 16, This study was supported by an Inha University Research Grant. B. W. Bang and S. Jeong contributed equally to this study. 1 Division of Gastroenterology, Department of Internal Medicine, Inha University School of Medicine, Incheon, South Korea. 2 Division of Gastroenterology, Department of Internal Medicine and Center for Advanced Medical Education by Brain Korea 21 Project, Inha University School of Medicine and Utah-Inha DDS & Advanced Therapeutics Research Center, Incheon, South Korea. 3 Department of Radiology, Inha University College of Medicine, 7-206, 3-Ga, Sinheung-Dong, Jung-Gu, Incheon , South Korea. Address correspondence to Y. S. Jeon (radjeon@inha.ac.kr). AJR 2010; 194: X/10/ American Roentgen Ray Society Curved Planar Reformatted Images of MDCT for Differentiation of Biliary Stent Occlusion in Patients With Malignant Biliary Obstruction OBJECTIVE. We prospectively evaluated the usefulness of MDCT using a curved planar reformation technique for the noninvasive assessment of the causes of biliary stent occlusion in patients with malignant biliary obstruction. SUBJECTS AND METHODS. Between December 2004 and January 2009, 173 patients with unresectable malignant biliary obstruction underwent biliary stent insertion. Among them, 26 patients with suspected biliary stent occlusion underwent 29 sessions of ERCP within 2 weeks after MDCT was performed. Curved planar reformation images were obtained along the pathway of the biliary stent. We interpreted tissue growth or stent clogging by comparing attenuation values inside the biliary stent between the unenhanced and contrast-enhanced phases of CT. The cause of biliary stent occlusion was confirmed by using ERCP. RESULTS. The differences in attenuation value inside the biliary stent between the contrast-enhanced and unenhanced phases of CT in the tissue growth group was 27.7 ± 21.7 HU (SD) and 4.2 ± 10.6 HU in the stent-clogging group (p = 0.002). The sensitivity and specificity of MDCT for the diagnosis of tissue growth were 86.7% and 85.7%, respectively. The overall accuracy of curved planar reformation images of MDCT for diagnosing the causes of stent occlusion was 86.2%. CONCLUSION. Curved planar reformation MDCT is a useful noninvasive technique that is relatively accurate for diagnosing the cause of biliary stent occlusion and is helpful for planning the therapeutic management of such patients. T he endoscopic and percutaneous techniques for placement of selfexpandable metallic stents are widely accepted for the palliative treatment of unresectable malignant biliary obstruction [1 3]. However, these stents tend to occlude within the expected stent life for various reasons. In spite of the short life expectancy for patients, the stent occlusion rate until death is approximately 20% [4 6]. The main causes of biliary stent occlusion are biliary stent clogging and tissue growth. Stent clogging is caused by biliary sludge, hemobilia, retained mucus, and, rarely, food materials that are refluxed from the duodenum. Tissue growth includes tumor ingrowth through the cells of the stent wall, tumor overgrowth outside the stent ends, or reactive mucosal hyperplasia [7]. Stent obstruction caused by food or sludge can be relieved endoscopically with a balloon catheter or basket or both, whereas obstruction caused by tissue growth can be treated by additional stent insertion [8]. Therefore, in patients with suspected biliary stent occlusion, it is important to predict the causes of biliary stent occlusion before biliary intervention because of the different therapeutic techniques needed. There is no established noninvasive method to identify the two types of stent occlusions until now, to our knowledge. The aim of this study was to prospectively evaluate the usefulness of curved planar reformation MDCT images for differentiation of the causes of biliary stent occlusion and to determine whether it is helpful to plan for endoscopic intervention in patients with biliary stent occlusion by such a noninvasive method. Subjects and Methods Patients Between December 2004 and January 2009, 173 patients were treated with insertion of metallic stents for palliation of inoperable malignant biliary obstruction at our institution. Among those patients in whom malignancy was confirmed by AJR:194, June

2 Bang et al. endoluminal or percutaneous biopsy, 26 patients who were suspected to have biliary stent occlusion and who underwent 29 sessions of ERCP within 2 weeks after MDCT was performed were enrolled in this study. Three patients had second stents inserted because of stent occlusion by tissue growth, but recurrent occlusion of the biliary stent occurred and the patients underwent further intervention during the follow-up period. The subjects included seven men and 19 women, with an age range of years (mean, 67 years). The causes of the malignant biliary obstruction included bile duct carcinoma (n = 15), pancreatic carcinoma (n = 6), ampulla of Vater carcinoma (n = 2), gallbladder carcinoma (n = 1), and metastatic lymphadenopathy from colon and stomach cancer (n = 2). The stents were inserted endoscopically or percutaneously. We inserted two stents into both intrahepatic ducts in six patients with hilar cholangiocarcinoma. Uncovered and covered self-expandable metal stents were used in 23 and three patients, respectively. Stent occlusion was diagnosed according to clinical, laboratory, and radiologic findings. This study was approved by the institutional review board of our institution, and informed consent for the procedure was obtained from each patient. Imaging Techniques All patients were examined with a 16-MDCT scanner (Sensation 16, Siemens Healthcare). An unenhanced scan of the upper abdomen was obtained to screen the biliary tree and for stent placement. The triple-phase contrast protocol for imaging of the pancreatobiliary system was performed, with a delay of 18 seconds for the arterial phase, 35 seconds for the portal phase, and 70 seconds for the delayed phase, after starting an infusion of ml of nonionic contrast medium (ioversol [Optiray 320, Tyco Healthcare]) at a rate of 3 ml/s through the antecubital vein. For all phases of CT, uniform technical parameters were used, as follows: detector row configuration, mm; collimation, 0.75 mm; slice thickness, 5 mm; table speed, 12 mm/rotation; pitch, 1.0; rotation time, 0.5 second; and 120 kvp. The reconstructed images were then transferred to a dedicated workstation (Wizard, Siemens Healthcare), and the coronal, sagittal, and oblique planes of the multiplanar reformation were created by one radiologist who had sufficient experience with multiplanar reformation techniques. Both the thickness and the interslice spacing of the reconstruction slices for each of the phases of CT were 3 mm. Curved planar reformation images were also obtained interactively by tracing a curved path through the imaging volume along the course of the biliary stent pathway among the unenhanced and portal venous phase images. Region-of-interest (ROI) markers were placed manually in the center of the suspicious lesion, avoiding the inclusion of an air shadow or an adjacent biliary stent. Attenuation values of the suspicious lesions were measured for the unenhanced and portal venous phase images using the same size ROI marker on the same slice. The measurements were performed three times, and the average data were recorded. The contrast enhancement value of the ROI was calculated by subtracting the unenhanced attenuation value from the contrast-enhanced attenuation value. We interpreted an increment of 10 HU in density as tissue growth, whereas an increase < 10 HU in density was considered to be stent clogging. MDCT findings were interpreted by an experienced radiologist before ERCP was performed. The diagnostic efficacy of the MDCT with curved planar reformation images was evaluated and compared with the ERCP findings. After our study, two reviewers who did not know the results of ERCP reviewed curved planar reformation images of MDCT and measured again the HU densities of the ROI. We informed the reviewers which of the precise lesions we measured, because they may have selected different lesions for measurement. We compared the results obtained by the two reviewers with our original results. ERCP Procedure ERCP was performed using a standard duodenoscope (TJF-240, Olympus Optical) for diagnostic or therapeutic purposes by one gastroenterologist. The causes of biliary stent obstruction were confirmed on the basis of the typical ERCP findings. The classification of the cause of stent occlusion was modified by the criteria of Rogart et al. [9], as follows: tissue growth, which includes tumor ingrowth and overgrowth and is seen radiographically or endoscopically as a narrowing within the stent or at the proximal or distal margins of the stent on fluoroscopy; or stent clogging, which is seen as multiple radiographic filling defects that disappeared after extraction with a retrieval balloon catheter along with endoscopic visualization of extracted sludge. Stent occlusion due to stent clogging was cleared by sweeping the lumen with a retrieval balloon catheter. Stent occlusion due to tissue growth was treated by placement of an additional metallic stent through the peroral transpapillary or percutaneous transhepatic route. Statistical Analysis Statistical analysis of the attenuation value from unenhanced and portal phase CT was performed using the Mann-Whitney U test. A p value < 0.05 was considered significant. All data are expressed as the mean (± SD). Statistical analyses were performed using the SPSS computer program (version 12.00, SPSS). Results Initial stent placements were technically successful without complications for all patients. Stent occlusions occurred between 2 and 30 months after initial insertion. Time to stent occlusion was 11.1 ± 9.7 months (mean ± SD) in the tissue growth group and 8.2 ± 6.5 months in the stent clogging group. Stent migration was not observed in either group during the follow-up period. Among three patients who underwent additional stent insertion as a result of tissue growth-related stent occlusion, follow-up ERCP revealed tissue growth in one patient and stent clogging in two patients. Attenuation value was measured for suspicious obstructive lesions inside biliary stents on both unenhanced and contrast-enhanced CT scans (Fig. 1). The contrast enhancement value of each lesion was calculated by subtracting the unenhanced attenuation value from the contrast-enhanced attenuation value (Fig. 2). The interpretation of the MDCT images matched the findings of ERCP in all cases except four. Of the 14 cases that were interpreted as stent clogging by MDCT, 12 cases were also diagnosed as stent clogging by ERCP (Fig. 3). Of 15 cases that were interpreted as tissue growth by MDCT, 13 cases were also diagnosed as tissue growth by ERCP (Fig. 4). There were two cases misdiagnosed by MDCT in each group. The difference in density in ROI between the portal venous phase and the unenhanced phase was 27.7 ± 21.7 HU (SD) in the tissue growth group and 4.2 ± 10.6 HU in the stent clogging group; the difference was statistically significant (p = 0.002). The sensitivity and specificity of MDCT for tissue growth was 86.7% and 85.7%, respectively. The overall accuracy of the curved planar reformation images of MDCT in the determination of the causes of stent obstruction was 86.2%. For the generalizability of our study, attenuation values of all patients were measured by two reviewers retrospectively. The attenuation value, sensitivity, specificity, and diagnostic accuracy from our study and two reviewers results are shown in Table 1. Fourteen cases that were diagnosed with stent clogging were treated successfully by sludge extraction with a retrieval balloon. Fifteen cases that were diagnosed as tissue growth were treated by additional stent insertion. Stents were inserted endoscopically in seven patients and percutaneously in eight patients AJR:194, June 2010

3 MDCT of Biliary Stent Occlusion Density, HU Change in Density, HU Cases Fig. 1 Hounsfield units (HU) measured in regions of interest on unenhanced (white bars) and contrastenhanced (black bars) CT scans. HU for stent clogging groups are shown in cases 1 14, whereas tissue growth groups are shown in cases Cases 20 Fig. 2 Contrast enhancement values for regions of interest. Contrast enhancement values for stent clogging groups are shown in cases 1 14, whereas tissue growth groups are shown in cases White bars represent patients with stent clogging who were misdiagnosed by tissue growth. Gray bars show patients with tissue growth who were misdiagnosed by stent clogging. Black bars represent patients who show concordance between MDCT interpretation and ERCP finding. Discussion Many imaging techniques may be used to evaluate the biliary tree, but until now there have been limitations in evaluating patients with suspected biliary stent obstruction. Abdominal sonography has limited value because of its poor accuracy and visibility inside the metallic stent [10]. MR cholangiopancreatography is a well-established noninvasive diagnostic tool for assessing the biliary system, but it is not useful for patients who have a biliary stent. The stent lumen cannot be evaluated because of susceptibility artifacts [11]. Conventional CT does not provide adequate information inside a biliary stent because the orientation of these ducts is not suitable for axial images [12]. ERCP and percutaneous transhepatic cholangiography have been used as the most sensitive and specific diagnostic techniques. Moreover, these techniques have therapeutic potential as well. However, percutaneous transhepatic cholangiography and ERCP are invasive techniques and cannot evaluate peribiliary surrounding structures. Recently, the MDCT scanner has provided unprecedented image quality. MDCT is a major advance in the field of diagnostic imaging because it allows a fast table speed and, when combined with thin slices, permits data collection that is well suited for workstation analysis [13]. By using a workstation with an advanced postprocessing technique, such as multiplanar reformation, rapid assessment of the bile ducts along different planes without losing information about surrounding structures is possible. MDCT using the multiplanar reformation technique is considered a useful imaging technique in determining the cause of biliary stent obstruction by obtaining a onestep evaluation of the biliary tract and the surrounding structures. However, MDCT with multiplanar reformation images has some limitations because the biliary system does not always course in the orthogonal axial, coronal, or sagittal planes. On the other hand, curved planar reformation images can permit even the most tortuous anatomy to be unwound and can illustrate complex anatomy in a single image along its long axis, which is of particular usefulness in evaluation of the pancreatobiliary system and coronary arteries [14 16]. In many studies, MDCT using the curved planar reformation technique has been studied for the noninvasive assessment of luminal patency and visibility of a stentimplanted coronary artery [17]; however, there are no published reports that have analyzed intraluminal lesions in patients with biliary stent occlusion. We evaluated the curved planar reformation MDCT images to predict the cause of biliary stent occlusion, and the Hounsfield unit intensity of ROI inside the biliary stent was measured on both unenhanced and contrast-enhanced (portal venous) CT for quantitative analysis. The Hounsfield unit scale, which measures the x-ray attenuation of tissue in CT, is linearly related to the mass density of a tissue. Clinically, IV contrast agents cause well-perfused tissues or tumors to increase in brightness on CT [18]. In the current study, we interpreted tissue growth or stent clogging according to whether the increment was 10 HU. The reason is that an attenuation increase of < 10 HU is considered to be within AJR:194, June

4 Bang et al. A C E G B D F H Fig year-old man with cholangiocarcinoma. A and B, MDCT using curved planar reformation images shows no enhancing wall or intraluminal mass when unenhanced image (A) is compared with contrast-enhanced image (B). C and D, Circles represent region-of-interest (ROI) markers used for measuring attenuation of suspicious lesions. Attenuation values of identically sized ROI markers were measured in unenhanced (C) and portal (D) venous CT phase of curved planar reformation images. Measurements were performed three times, and average data were recorded. E, Side view duodenoscopy shows sludge in distal biliary stent. F, Cholangiogram shows multiple filling defects in biliary stent. G, Patient was treated by basket extraction. H, After basket extraction, cholangiogram shows no filling defects in biliary stent. the limits of artifactual pseudoenhancement [19]. With a threshold of 10 HU, the sensitivity, specificity, and accuracy of MDCT using curved planar reformation images for tissue growth were all high (86.7%, 85.7%, and 86.2%, respectively). In addition, there are good interobserver agreements. Even though the retrospective study could have introduced selection bias, because we informed the reviewers of the precise lesions that we measured, the results have good agreement with our original study. Therefore, a curved planar reformation image using the Hounsfield unit scale is a very useful method for evaluating patients with biliary stent obstruction. When we analyzed our four misdiagnosed results, two stent clogging cases that were misdiagnosed as tissue growth by MDCT showed higher attenuation values than adjacent structures on contrast-enhanced images. One case was diagnosed as hemobilia by ERCP. The other case may have been misdiagnosed because of the position of the biliary stent during the creation of curved planar reformation images. On the other hand, two tissue growth cases were misdiagnosed as stent clogging. We may have missed the focal tumor ingrowth because the lesions were too small. Although metallic stents of various designs have recently been developed to improve clinical efficacy, metallic stents have been occluded over time for various causes. The median metallic stent patency is days [20], and the stent occlusion rate until death is approximately 20% [4 6]. The main cause of early occlusion (within 6 months) is bile sludge or a clog, whereas tumor ingrowth and overgrowth are responsible for most late occlusions (after 6 months) [21]. There was no difference in the stent patency period between both groups in this study. Little is 1512 AJR:194, June 2010

5 MDCT of Biliary Stent Occlusion A C E known about the biologic reaction to metallic biliary stents. Some researchers have investigated the tissue reaction of the common duct wall to metallic stents in dogs, and the changes induced by the stent in healthy dogs were focal denudation and proliferation of the mucosa associated with chronic inflammation and fibrosis of the submucosa. These changes were mainly secondary to pressure erosion of the bile duct by metallic wires [22, 23]. Similar results were observed in human studies [7, 24]. The biologic reaction in the early phase (< 4 weeks) was mainly attributable to the trauma of stent insertion. The F B D Fig year-old woman with cholangiocarcinoma. A and B, MDCT using curved planar reformation images shows enhancing intraluminal mass comparing unenhanced (A) with contrast-enhanced (B) image on middle portion of biliary stent (arrow). C and D, Circles represent region-of-interest (ROI) markers used for measuring attenuation of suspicious lesions. Hounsfield unit densities of identically sized ROI markers were measured in unenhanced (C) and portal (D) venous CT phase of curved planar reformation images. Measurements were performed three times, and average data were recorded. E, In spite of basket sweeping, cholangiogram shows stricture in middle portion of biliary stent (arrow). F, Cholangiogram shows that second metallic stent (arrow) was placed. normal epithelium of the bile duct was completely destroyed. In the later stages (2 12 months), the surface of the stent was covered by a layer of fibrous tissue. The main reaction to the stent was connective tissue formation, which never occurred before 3 months, and complete coverage of the entire stent by tissue was first seen only after > 1 year. The current study has some limitations. First, although we differentiated the causes of stent occlusion by sweeping stent lumen using a retrieval balloon catheter and then identifying whether disappearance of the luminal narrowing on fluoroscopy had occurred, we did not confirm the cause of stent occlusion histologically. The reason is that the endoluminal biopsy of the bile duct has been reported to reveal limited and various diagnostic yields for malignant biliary obstruction [25]. In addition, all of the patients in this study had advanced unresectable cancer; therefore, obtaining a tissue diagnosis by ERCP or surgery was useless and did not substantially alter therapeutic management. Second, a bias may have been introduced when measuring attenuation value. Even a small change in ROI can change attenuation value considerably. Also, regardless of careful attention, the defining ROI may include air or biliary stent. In addition, because of the small diameter of the biliary stent, we measured relatively small ROIs, which can magnify the effects of quantum mottle. To overcome the bias, we tried to place the largest size of the ROIs as possible without including air bubble and stent. We also measured three times at the same region. Third, this study was based on a relatively small number of cases. A larger number of subjects is necessary to validate our results. In conclusion, MDCT with the curved planar reformation technique provides useful information for the causes of biliary stent AJR:194, June

6 Bang et al. TABLE 1: Comparison between our and reviewers results Variable Our Study Reviewer 1 Reviewer 2 Change in Hounsfield units Tissue growth 27.7 ± ± ± 22.0 Stent clogging 4.2 ± ± ± 8.1 Sensitivity (%) Specificity (%) Accuracy (%) Note Change in attenuation value was calculated by subtracting the unenhanced attenuation value from the contrast-enhanced attenuation value. obstruction and is well correlated with ERCP findings. Thus, it may be feasible as a noninvasive imaging technique for visualization inside the biliary stent and may be helpful in selecting the appropriate therapeutic option. References 1. Lammer J, Klein GE, Kleinert R, Hausegger K, Einspieler R. Obstructive jaundice: use of expandable metal endoprosthesis for biliary drainage. Radiology 1990; 177: Irving JD, Adam A, Dick R, Dondelinger RF, Lunderquist A, Roche A. Gianturco expandable metallic biliary stents:results of a European clinical trial. Radiology 1989; 172: Kaw M, Singh S, Gagneja H. Clinical outcome of simultaneous self-expandable metal stents for palliation of malignant biliary and duodenal obstruction. Surg Endosc 2003; 17: Mueller PR, Ferrucci JT Jr, Teplick SK, et al. Biliary stent endoprosthesis: analysis of complications in 113 patients. Radiology 1985; 156: Rossi P, Bezzi M, Rossi M, et al. Metallic stents in malignant biliary obstruction: results of a multicenter European study of 240 patients. J Vasc Interv Radiol 1994; 5: Inal M, Akgul E, Aksunger E, Demiryurek H, Yagmur O. Percutaneous self-expandable uncovered metallic stents in malignant biliary obstruction: complications, follow-up and reintervention in 154 patients. Acta Radiol 2003; 44: Boguth L, Tatalovic S, Antonucci F, Heer M, Sulser H, Zollikofer CL. Malignant biliary obstruction: clinical and histopathologic correlation after treatment with self-expanding metal prostheses. Radiology 1994; 192: Lee MJ, Dawson SL, Mueller PR, et al. Percutaneous management of hilar biliary malignancies with metallic endoprostheses: results, technical problems, and causes of failure. RadioGraphics 1993; 13: Rogart JN, Boghos A, Rossi F, et al. Analysis of endoscopic management of occluded metal biliary stents at a single tertiary care center. Gastrointest Endosc 2008; 68: Zemel G, Zajko AB, Skolnick ML, Bron KM, Campbell WL. The role of sonography and transhepatic cholangiography in the diagnosis of biliary complications after liver transplantation. AJR 1988; 151: Merkle EM, Boll DT, Weidenbach H, Brambs HJ, Gabelmann A. Ability of MR cholangiography to reveal stent position and luminal diameter in patients with biliary endoprostheses: in vitro measurements and in vivo results in 30 patients. AJR 2001; 176: Park SJ, Han JK, Kim TK, Choi BI. Three-dimensional spiral CT cholangiography with minimum intensity projection in patients with suspected obstructive biliary disease: comparison with percutaneous transhepatic cholangiography. Abdom Imaging 2001; 26: Hu H, He HD, Foley WD, Fox SH. Four multidetector-row helical CT: image quality and volume coverage speed. Radiology 2000; 215: Desser TS, Sommer FG, Jeffrey RB Jr. Value of curved planar reformations in MDCT of abdominal pathology. AJR 2004; 182: Vargas R, Nino-Murcia M, Trueblood W, Jeffrey RB Jr. MDCT in pancreatic adenocarcinoma: prediction of vascular invasion and respectability using a multiphasic technique with curved planar reformations. AJR 2004; 182: Nino-Murcia M, Jeffrey RB Jr, Beaulieu CF, Li KC, Rubin GD. Multidetector CT of the pancreas and bile duct system: value of curved planar reformations. AJR 2001; 176: Jones CM, Chin KY, Hamady M, Yang GZ, Darzi A, Athanasiou T. Coronary stent assessment with 64-slice multislice computed tomography: a literature review. Semin Ultrasound CT MR 2008; 29: Berber E, Foroutani A, Garland AM, et al. Use of CT Hounsfield unit density to identify ablated tumor after laparoscopic radiofrequency ablation of hepatic tumors. Surg Endosc 2000; 14: Bae KT, Heiken JP, Siegel CL, Bennett HF. Renal cysts: is attenuation artifactually increased on contrast-enhanced CT images? Radiology 2000; 216: Moparty B, Carr-Loche DL. Metal or plastic stent for malignant biliary obstruction: what s got the most bang for your buck? Eur J Gastroenterol Hepatol 2007; 19: Suk KT, Kim JW, Kim HS, et al. Human application of a metallic stent covered with a paclitaxelincorporated membrane for malignant biliary obstruction: mulitcenter pilot study. Gastrointest Endosc 2007; 66: Carrasco CH, Wallace S, Charnsangavej C, et al. Expandable biliary endoprothesis: an experimental study. AJR 1985; 145: Alvarado R, Palmaz JC, Carcia OJ, Tio FO, Rees CR. Evaluation of polymer-coated balloon-expandable stents in bile ducts. Radiology 1989; 170: Hausegger KA, Kleinert R, Lammer J, Klein GE, Flückiger F. Malignant biliary obstruction: histologic findings after treatment with self-expandable stents. Radiology 1992; 185: Ponchon T, Gagnon P, Berger F, et al. Value of endobiliary brush cytology and biopsies for the diagnosis of malignant bile duct stenosis: results of a prospective study. Gastrointest Endosc 1995; 42: AJR:194, June 2010