CT Differentiation of Mucin- Producing Cystic Neoplasms of the Liver From Solitary Bile Duct Cysts

Gastrointestinal Imaging Original Research Kim et al. CT of Hepatic Cystic Neoplasms and Cysts Gastrointestinal Imaging Original Research Hyoung Jung Kim 1 Eun Sil Yu 2 Jae Ho Byun 1 Seung-Mo Hong 2 Kyoung Won Kim 1 Jong Seok Lee 1 So Yeon Kim 1 Kim HJ, Yu ES, Byun JH, et al. Keywords: CT, intraductal papillary neoplasm of bile duct, mucinous cystic neoplasm, solitary bile duct cyst DOI:10.2214/AJR.12.9170 Received April 30, 2012; accepted after revision May 8, 2013. 1 Department of Radiology and Research Institute of Radiology, Asan Medical Center, University of Ulsan College of Medicine, 86 Asanbyeongwon-gil, Songpa-gu, Seoul, 138-736, Korea. Address correspondence to H. J. Kim (hjk@amc.seoul.kr). 2 Department of Pathology, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Korea. AJR 2014; 202:83 91 0361 803X/14/2021 83 American Roentgen Ray Society CT Differentiation of Mucin- Producing Cystic Neoplasms of the Liver From Solitary Bile Duct Cysts OBJECTIVE. The purpose of this study was to identify the CT features required for differentiating mucin-producing cystic neoplasms of the liver (mucinous cystic neoplasms and of the bile duct) from solitary bile duct cysts. MATERIALS AND METHODS. CT images of pathologically confirmed mucinous cystic neoplasms (n = 15), of the bile duct (n = 16), and solitary bile duct cysts (n = 31) were reviewed. Analysis of the CT findings included shape, presence of septa, location of septa (peripheral vs central), thickness of septa (thin vs thick), mosaic pattern, mural nodules, intracystic debris, calcification, upstream bile duct dilatation, downstream bile duct dilatation, and communication between a cystic lesion and the bile duct. The maximum size of a cystic lesion and the maximum size of the largest mural nodule were measured. RESULTS. The presence of septa, central septa, mural nodules, upstream bile duct dilatation, and downstream bile duct dilatation were found to be significant CT findings for differentiating mucinous cystic neoplasms and of the bile duct from solitary bile duct cysts (p < 0.05 for each finding). When two of these five criteria were used in combination, the sensitivity and specificity for diagnosing mucin-producing cystic neoplasms and of the bile duct were 87% (27 of 31) and 87% (27 of 31), respectively. When three of the five criteria were used in combination, the sensitivity and specificity were 58% (18 of 31) and 100%. CONCLUSION. With the use of specific CT criteria, mucin-producing cystic neoplasms of the liver can be differentiated from solitary bile duct cysts with a high degree of accuracy. M ucinous cystic neoplasm and cystforming intraductal papillary neoplasm of the bile duct are mucinproducing cystic neoplasms of the liver that have a propensity toward malignant change, and the two cystic neoplasms have similar macroscopic features [1]. Mucinous cystic neoplasm has previously been referred to as biliary cystadenoma and biliary cystadenocarcinoma [1]. Cases of biliary cystadenocarcinoma have been reported with equal frequency in men and women [2]. Mucinous cystic neoplasm is currently defined as a cystforming epithelial neoplasm composed of mucin-producing epithelium and associated with ovarian-type subepithelial stroma [1]. Intraductal papillary neoplasm of the bile duct includes the previous categories biliary papilloma and papillomatosis. It is characterized by dilated intrahepatic ducts filled with a papillary biliary neoplasm [1]. Although many cases of intraductal papillary neoplasm of the bile duct exhibit tubular or fusiform dilatation of the involved duct, some cases may exhibit cystic dilatation of the affected duct [3]. These of the bile duct lack ovarian-like stroma in the cyst wall and usually exhibit luminal communication with the bile duct, thus differing from mucinous cystic neoplasms [1]. Mucinous cystic neoplasms have often been incorrectly diagnosed at preoperative imaging as solitary bile duct cysts. Because they were then treated with aspiration or fenestration [4, 5], the result was recurrence or persistence of the cyst and its associated symptoms [5]. Therefore, preoperative differentiation of mucin-producing cystic neoplasms from solitary bile duct cysts is important because these conditions have different clinical significance and different management strategies [4]. To our knowledge, no study has compared mucin-producing cystic neoplasms and solitary bile duct cysts. The purpose of this study was AJR:202, January 2014 83

Kim et al. to identify CT findings that differentiate mucin-producing cystic neoplasms of the liver (mucinous cystic neoplasms and cyst-forming intraductal papillary neoplasms of the bile duct) from solitary bile duct cysts. Materials and Methods This retrospective study included patients at a single institution. It was approved by our institutional review board (2012 0220), which waived the requirement for informed consent. Some of the patients with mucinous cystic neoplasms or cyst-forming intraductal papillary neoplasms of the bile duct had been included in a previous study at our institution [6]. However, image analysis and histologic evaluation are wholly unique to the current study. Patients We searched the pathology database from January 2001 to December 2011 using the terms biliary cystadenoma, biliary cystadenocarcinoma, or mucinous cystic neoplasm and liver, papillomatosis or intraductal papillary neoplasm and bile duct or liver, and solitary bile duct cyst or hepatic cyst and liver. A radiologist who did not participate in image analysis used the following exclusion criteria to establish a study group of suitable cases for comparing the CT findings in the three diseases and to eliminate bias after review of the images of the final study group: intraductal papillary neoplasms of the bile duct without cystic dilatation, solitary bile duct cysts treated by fenestration, and absence of both unenhanced CT and contrast-enhanced portal venous phase CT images. In patients with intraductal papillary neoplasms of the bile duct, the presence of cystic dilatation was defined as gross cystic dilatation of tumor-bearing bile ducts and balloonlike dilatation of the bile duct in which there were tumor nodules [3]. Solitary bile duct cysts treated by fenestration were excluded because both a radiologist and a surgeon had preoperatively diagnosed them as solitary bile duct cysts without suspicion of mucinous cystic neoplasm or other cystic neoplasm. Both unenhanced and contrast-enhanced CT acquisitions were essential in the imaging analysis of mural nodule and intracystic debris. Finally, 62 patients with hepatic cystic lesions were included in the study (Fig. 1). Reference Standard and Medical Record Review Among the 62 patients, 42 underwent hepatic resection; three, fenestration followed by resection; 16, enucleation; and one, fenestration. Pathology slides of the 62 patients were reviewed by a pathologist with 20 years of clinical experience in hepatobiliary pathology. The pathology slides were reviewed according to the diagnostic Pathology Database From January 2001 to December 2011 + using the search terms, A = biliary cystadenoma, biliary cystadenocarcinoma, or mucinous cystic neoplasm and liver B = papillomatosis or intraductal papillary neoplasm and bile duct or liver C = solitary bile duct cyst or hepatic cyst and liver A (n = 22) No Yes A (n = 21) Mucinous cystic neoplasm (n = 15) B (n = 63) Without cystic dilatation of intrahepatic bile duct on CT (n = 53) C (n = 77) B (n = 10) C (n = 40) Both unenhanced and contrast-enhanced portal venous phase CT? A (n = 1) B (n = 10) C (n = 31) Review of the pathology slides for 62 patients with hepatic cystic lesion Fig. 1 Flow diagram shows study group. Yes Cyst-forming intraductal papillary neoplasm of bile duct (n = 16) Epithelium-lined hepatic cystic lesions (n = 62) Biliary/mucinous/oncocytic (n = 31) (with or without papillary architecture) With ovarian-like stroma (n = 15) Without epithelial invasion into wall Mucinous cystic neoplasm (n = 15) Epithelial invasion into wall Mucinous cystic neoplasm with invasive carcinoma (n = 0) No Yes Solitary bile duct cyst (n = 31) Without ovarian-like stroma (n = 16) Without epithelial invasion into wall Cyst-forming intraductal papillary neoplasm of bile duct (n = 13) Fenestration (n = 37) C (n = 9) Cuboidal/low columnar (n = 31) Epithelial invasion into wall Cyst-forming intraductal papillary neoplasm of bile duct with invasive carcinoma (n = 3) Mucinous cystic neoplasm with low-, intermediate-, or high-grade intraepithelial neoplasia Cyst-forming intraductal papillary neoplasm of bile duct with low-, intermediate-, or high-grade intraepithelial neoplasia Fig. 2 Flow diagram shows review of pathology slides of 62 patients with hepatic cystic lesion. Solitary bile duct cyst (n = 31) 84 AJR:202, January 2014

CT of Hepatic Cystic Neoplasms and Cysts TABLE 1: Definition of CT Findings in Image Analysis Size Shape Smooth Lobulate Cyst on cyst Pleomorphic Presence of septa Location of septa Peripheral Central CT Finding Thickness of septa Thin Thick Mosaic pattern Mural nodule Size of mural nodule Intracystic debris Calcification Upstream bile duct dilatation Downstream bile duct dilatation Communication TABLE 2: Clinical Data Characteristic Maximum size of hepatic cystic lesion in millimeters Round or ovoid shape with smooth margin Round or ovoid shape with lobulate margin Definition Multilocular cystic lesion consisting of round or ovoid cysts Multilocular cystic lesion consisting of round or ovoid cysts and tubular or bizarre-shaped cysts Visible septum on CT images Septum within one half of the radius from the periphery of the hepatic cystic lesion Septum within one half of the radius from the center of the hepatic cystic lesion Any septum with thickness < 3 mm Any septum with thickness 3 mm Locules of the hepatic cystic lesion with different densities Nodules on the wall or the septum showing enhancement Largest diameter of mural nodule in millimeters Intermediate or high density without enhancement Focal high density, similar to that of cortical bone Diameter of bile duct > 2 mm or 40% of the adjacent portal vein diameter Diameter of common duct > 8 mm Communication between a hepatic cystic lesion and bile duct Mucinous Cystic Neoplasms (n = 15) Cyst-Forming Intraductal Papillary Neoplasms of the Bile Duct (n = 16) Solitary Bile Duct Cysts (n = 31) p a Sex b 0.076 Men 0 6 2 Women 15 10 29 Age (y) c 44 (40 56) 63.5 (50.75 68.5) 62 (51 67) 0.255 Symptoms b 0.670 None 11 (73) 9 (56) 23 (74) Pain 4 (27) 5 (31) 8 (26) Cholangitis 0 1 (6) 0 Weight loss 0 1 (6) 0 Aspartate aminotransferase (IU/L) c 20 (17 27) 30.5 (24 42.5) 22 (18 27) 0.130 Alanine aminotransferase (IU/L) c 18 (11 29) 28 (14.5 58.5) 16 (12 26) 0.157 Alkaline phosphatase (IU/L) c 64 (50 84) 82.5 (66 113.3) 64 (57 81) 0.269 Total bilirubin (mg/dl) c 0.9 (0.6 1.1) 0.8 (0.6 1.1) 0.7 (0.5 0.8) 0.019 CA19-9 (IU/mL) c 34.7 (24.2 107.2) 14.2 (8.3 21.6) 7.4 (2.8 21.9) 0.013 Carcinoembryonic antigen (ng/ml) c 0.9 (0.6 2.1) 1.4 (1.0 2.4) 2.15 (0.7 3.5) 0.327 a Between mucin-producing cystic neoplasms (mucinous cystic neoplasms and of the bile duct) and solitary bile duct cysts calculated with Fisher exact test for comparison of categoric variables, Student t test for comparison of age, and Mann-Whitney test for comparison of the other continuous variables. b Data are numbers of patients with percentages in parentheses. c Values are median with interquartile range in parentheses. AJR:202, January 2014 85

Kim et al. algorithm for cystic lesions of the liver proposed by the World Health Organization [1] (Fig. 2). Among the 21 suspected cases of biliary cystadenoma, biliary cystadenocarcinoma, and mucinous cystic neoplasm, 15 were confirmed as mucinous cystic neoplasm, and six diagnoses were revised to cyst-forming intraductal papillary neoplasm of the bile duct. Ten suspected cases of papillomatosis or intraductal papillary neoplasm were confirmed as cyst-forming intraductal papillary neoplasm of the bile duct. Three suspected cases of invasive carcinoma were cyst-forming intraductal papillary neoplasm of the bile duct (Fig. 2). Thirty-one suspected cases of solitary bile duct cysts were confirmed as solitary bile duct cysts. The final study group consisted of 15 patients with mucinous cystic neoplasm, 16 patients with cyst-forming intraductal papillary neoplasm of the bile duct, and 31 patients with solitary bile duct cysts (Fig. 1). The medical records of the 62 patients with hepatic cystic lesions were reviewed. Patient sex, age, and symptoms at presentation were recorded. Serum aspartate aminotransferase, alanine aminotransferase (ALT), alkaline phosphatase, total bilirubin, carbohydrate antigen 19-9 (CA19-9), and carcinoembryonic antigen levels were recorded. CT CT examinations were performed with a Somatom Plus 4, Sensation 16, or Somatom Definition scanner (Siemens Healthcare) and a Light- Speed QX/I, LightSpeed 16, or LightSpeed VCT scanner (GE Healthcare). For contrast-enhanced CT, 120 150 ml iopromide (Ultravist 370 or Ultravist 300, Bayer Schering Pharma) was administered IV at a rate of 2.5 or 3 ml/s through an automatic power injector. Liver dynamic CT consisting of unenhanced, arterial, portal venous, and delayed phase imaging was performed for 58 patients. Abdominal CT consisting of unenhanced and portal venous phase images was performed for four patients. Portal venous phase images were obtained 70 80 seconds after the start of the IV injection. Axial images were reconstructed with section thicknesses of 3 mm in 13 patients, 5 mm in 47 patients, and 7 mm in two patients. Coronal reformation portal venous phase images were reconstructed with a section thickness of 5 mm in 13 patients. Image Analysis The axial unenhanced and portal venous phase images were evaluated. The coronal reformation portal venous phase images of 13 patients were also included in the image analysis. Two radiologists evaluated the images in consensus. They had 13 and 11 years of experience in abdominal radiology as part of their daily clinical and research practice. The reviewers were blinded to the pathologic diagnosis of hepatic cystic lesions. The items evaluated in our study are summarized in Table 1. Statistical Analysis All comparisons were performed between mucin-producing cystic neoplasms (mucinous cystic neoplasms and cyst-forming intraductal papillary neoplasms of the bile duct) and solitary bile duct cysts. To determine whether there was a significant difference in each demographic data point and laboratory data point between mucin-producing cystic neoplasms and solitary bile duct cysts, the Fisher exact test was used for comparison of sex and symptoms, Student t test for comparison of age, and Mann-Whitney test for comparison of TABLE 3: CT Findings CT Finding Mucinous Cystic Neoplasms (n = 15) Cyst-Forming Intraductal Papillary Neoplasms of the Bile Duct (n = 16) Solitary Bile Duct Cysts (n = 31) p a Size (mm) b 87 (64 136) 78 (44 114) 90 (73 129) 0.223 Shape c 0.145 Smooth 9 (60) 7 (44) 23 (74) Lobulate 6 (40) 3 (19) 7 (23) Cyst on cyst 0 (0) 4 (25) 1 (3) Pleomorphic 0 (0) 2 (13) 0 (0) Presence of septa c 14 (93) 11 (69) 12 (39) 0.001 Location of septa Peripheral 4/14 (29) 1 /11 (9) 10/12 (83) 0.001 Central 10/14 (71) 10/11 (91) 2 /12 (17) Thickness of septa Thin 10/14 (71) 9/11 (82) 12/12 (100) 0.149 Thick 4/14 (29) 2/11 (18) 0/12 (0) Mosaic pattern c 3 (20) 1 (6) 0 (0) 0.113 Mural nodule c 4 (27) 15 (94) 0 (0) < 0.001 Size of mural nodule (mm) b 8 (5 15) 12 (6 27) Intracystic debris c 6 (40) 6 (38) 6 (19) 0.093 Calcification c 7 (47) 1 (6) 5 (16) 0.349 Upstream bile duct dilatation c 4 (27) 11 (69) 3 (10) 0.001 Downstream bile duct dilatation c 0 (0) 7 (44) 0 (0) 0.011 Communication d 0 (0) 5 (31) 0 (0) 0.053 a Between mucin-producing cystic neoplasms (mucinous cystic neoplasms and of the bile duct) and solitary bile duct cysts calculated with chi-square or Fisher exact test for comparison of categoric variables and Mann-Whitney test for comparison of size. b Values are median with interquartile range in parentheses. c Data are numbers of patients with percentages in parentheses. d Communication between hepatic cystic lesion and bile duct. 86 AJR:202, January 2014

CT of Hepatic Cystic Neoplasms and Cysts Fig. 3 64-year-old woman with solitary bile duct cyst. Transverse contrast-enhanced CT image shows large ovoid cystic lesion without lobulation and thin peripheral septum (arrows). aspartate aminotransferase, aspartate aminotransferase, total bilirubin, CA19-9, and carcinoembryonic antigen results. To determine whether there was a significant difference in each CT finding between mucin-producing cystic neoplasms and solitary bile duct cysts, the Student t test was used for comparison of size, and the chi-square or Fisher exact test was used for comparison of shape, presence of septa, location of septa, thickness of septa, mosaic pattern, mural nodules, intracystic debris, calcification, upstream bile duct dilatation, downstream bile duct dilatation, and communication between a hepatic cystic lesion and the bile duct. For comparison of location and thickness of septa between mucin-producing cystic neoplasms and solitary bile duct cysts, only patients with the presence of septa were included. A value of p < 0.05 was considered to indicate a significant difference. Odds ratios with 95% CIs for differentiating mucin-producing cystic neoplasms from solitary bile duct cysts were calculated for each CT criterion. The sensitivity and specificity of each significant CT criterion and a combination of the significant CT criteria were also calculated. Each sensitivity was calculated to consider the number of CT findings as cutoff points. The numbers of CT findings were classified into two types, for example, at least two CT findings and another. For statistical analysis, we used SPSS software (version 18, SPSS). Results The clinical data on the patients with mucinous cystic neoplasms, cyst-forming intraductal papillary neoplasms of the bile duct, and solitary bile duct cysts are summarized in Table 2. There were significant differences between mucin-producing cystic neoplasms Fig. 4 71-year-old woman with solitary bile duct cyst. Transverse contrast-enhanced CT image shows multilocular cystic lesion consisting of round or ovoid cysts with cyst on cyst shape (arrows). Thin septa are present in central portion of cystic lesion. and solitary bile duct cysts with respect to serum total bilirubin and CA19-9 concentrations, but the median values were within a normal range. All 15 patients with mucinous cystic neoplasm were women, but six of the 16 patients with cyst-forming intraductal papillary neoplasm of the bile duct were men. CT Findings in Mucinous Cystic Neoplasms, Cyst-Forming Intraductal Papillary Neoplasms of the Bile Duct, and Solitary Bile Duct Cysts The CT findings of the patients with mucinous cystic neoplasms, cyst-forming intraductal papillary neoplasms of the bile duct, and solitary bile duct cysts are summarized in Table 3. Septa were present in all three groups. However, the presence of septa differed significantly between mucin-producing cystic neoplasms (mucinous cystic neoplasms, 93% [14/15]); intraductal papillary neoplasms of the bile duct, 69% [11/16]); and solitary bile duct cysts, 39% [12/31] (p = 0.001). With regard to the location of septa, peripheral septa and central septa were present in all three groups (Figs. 3 5). However, mucin-producing cystic neoplasms (mucinous cystic neoplasms, 71% [10/14]; cyst-forming intraductal papillary neoplasms of the bile duct, 91% [10/11]) and solitary bile duct cysts (17% [2/12]) differed significantly with respect to central septa (p = 0.001) (Fig. 5). Thick septa were present only in mucinous cystic neoplasms (29% [4/14]) and cyst-forming intraductal papillary neoplasms of the bile duct (18% [2/11]) (p = 0.149) (Fig. 5). Mural nodules were also noted only in mucin-producing cystic neoplasms (mucinous cystic neoplasms, 27% [4/15]; cyst-forming intraductal papillary Fig. 5 57-year-old woman with mucinous cystic neoplasm. Transverse contrast-enhanced CT image shows ovoid cystic lesion with calcification and multiple septa. Thick septum (arrow) is present in central portion. neoplasms of the bile duct, 94% [15/16]) (p < 0.001) (Figs. 6 and 7). Upstream bile duct dilatation was present in all three groups, but it differed significantly between mucin-producing cystic neoplasms (mucinous cystic neoplasms, 27% [4/15]; cyst-forming intraductal papillary neoplasms of the bile duct, 69% [11/16]); and solitary bile duct cysts, 10% [3/31] (p = 0.001) (Figs. 8 and 9). Downstream bile duct dilatation was noted only in of the bile duct (44% [7/16]) (p = 0.011) (Fig. 9). Mosaic pattern and communication between a hepatic cystic lesion and the bile duct were present only in mucin-producing cystic neoplasms (Fig. 6), but statistical significance was not found. Size and shape of the hepatic cystic lesion did not differ significantly between mucin-producing cystic neoplasms and solitary bile duct cysts. Intracystic debris and calcification of the hepatic cystic lesion also were not significantly different between mucin-producing cystic neoplasms and solitary bile duct cysts (Fig. 10). Sensitivity and Specificity Values for the CT Diagnosis Five CT findings, that is, the presence of septa, central septa, mural nodules, upstream bile duct dilatation, and downstream bile duct dilatation, were statistically significant predictors for differentiating mucin-producing cystic neoplasms from solitary bile duct cysts. The sensitivity and specificity of each significant CT criterion for differentiating mucin-producing cystic neoplasms from solitary bile duct cysts are summarized in Table 4. Except for mural nodules and downstream bile duct dila- AJR:202, January 2014 87

Kim et al. tation, the highest odds ratio was achieved for central septa, followed by upstream bile duct dilatation and the presence of septa. When two of these five criteria were used in combination, the sensitivity and specificity for diagnosing mucinous cystic neoplasm and cystforming intraductal papillary neoplasms of the bile duct both were 87% (27/31). When three of these five criteria were used in combination, the sensitivity and specificity for diagnosing mucin-producing cystic neoplasm were 58% (18 of 31) and 100% (Table 5). A C Fig. 6 59-year-old woman with mucinous cystic neoplasm. A and B, Transverse unenhanced CT images show ovoid cystic mass with different densities in each locule, constituting mosaic pattern, and two high-density lesions (arrow, A; arrowhead, B). C and D, Transverse contrast-enhanced CT images show subtle enhancement (arrow) and no enhancement (arrowhead, D) in high-density lesion visible in A and B. Enhancing lesion is mural nodule (arrow), and nonenhancing lesion is intracystic debris (arrowhead, D), such as hemorrhage. B D Discussion In our study, six of 21 patients with former diagnoses of biliary cystadenoma or cystadenocarcinoma had the diagnosis confirmed as cyst-forming intraductal papillary neoplasm of the bile duct. According to the current pathologic requirement for the presence of ovarian-type stroma in the diagnosis of mucinous cystic neoplasm, it is likely that many biliary cystadenocarcinomas would now be classified as intraductal papillary neoplasms of the bile duct with marked cystic changes [1]. Treatment of hepatic cystic lesions can be determined according to the preoperative diagnosis. Symptomatic solitary bile duct cysts can be treated by aspiration followed by sclerotherapy or fenestration [7]. Mucinous cystic neoplasm should be treated by enucleation or hepatic resection [4]. Cyst-forming intraductal papillary neoplasms of the bile duct can be treated by hepatic resection because they have a tendency toward superficial spreading along a dilated bile duct [8]. Laboratory tests may be used in the diagnosis of hepatic cystic lesions because the serum concentration of CA19-9 may increase in some patients with mucinous cystic neoplasms [9]; however, this is not a universal finding [10]. Measurement of the cystic fluid CA19-9 has been proposed as a method for differentiating mucinous cystic neoplasms from solitary bile duct cysts [11]; however, data suggest that levels of cyst fluid tumor marker levels are not useful information in patients with suspected mucinous cystic neoplasms [10]. The cytologic findings in cyst fluid may be used in the diagnosis of hepatic cystic lesions, but there is risk of peritoneal tumor cell spillage in the aspiration procedure, and normal cytologic results give a potentially false sense of security [4]. Therefore, imaging studies should have an important role in the preoperative diagnosis of hepatic cystic lesions. Our study results show that five CT findings the presence of septa, central septa, mural nodules, upstream bile duct dilatation, and downstream bile duct dilatation are statistically significant predictors for differentiating mucinous cystic neoplasms and cyst-forming intraductal papillary neoplasms of the bile duct from solitary bile duct cysts. In addition, when these CT findings were used in combination as the diagnostic criteria, mucinous cystic neoplasms and cyst-forming intraductal papillary neoplasms of the bile duct could be differentiated from solitary bile duct cysts with a high degree of diagnostic accuracy. When the results of our study are used for differentiating mucinous cystic neoplasms and of the bile duct from solitary bile duct cysts at Fig. 7 79-year-old woman with cyst-forming intraductal papillary neoplasm of the bile duct. Transverse contrast-enhanced CT image shows multilocular cystic mass consisting of ovoid cysts (thick arrows) and long tubular cystic lesions (arrowhead) with pleomorphic shape. Mural nodules (thin arrows) are also present. 88 AJR:202, January 2014

CT of Hepatic Cystic Neoplasms and Cysts Fig. 8 66-year-old woman with solitary bile duct cyst. A and B, Transverse contrast-enhanced CT images show round cystic lesion. Upstream bile duct dilatation (arrow) is present in left lateral section, but septum or mural nodule is not evident. CT, they may be helpful in establishing an accurate diagnosis, thereby reducing the chance of incomplete surgery, such as fenestration, for mucin-producing cystic neoplasms. In our study, the presence of septa was one of the statistically significant findings differentiating mucinous cystic neoplasms and A A A B B B Fig. 9 63-year-old woman with cyst-forming intraductal papillary neoplasm of bile duct. A, Transverse contrast-enhanced CT image shows round cystic mass with mural nodule (arrows) and communication (arrowhead) between cystic lesion and bile duct. B, Coronal contrast-enhanced reformatted CT image shows both upstream and downstream bile duct dilatation (arrowhead) and mural nodules (arrows). Fig. 10 56-year-old woman with solitary bile duct cyst. A, Transverse unenhanced CT image shows highdensity nodule in dependent portion of lesion (arrow). B, Transverse contrast-enhanced CT image shows high-density nodular lesion without enhancement (arrow). It was interpreted as intracystic debris. Thin septum (arrowhead) is present in peripheral portion of cyst. of the bile duct from solitary bile duct cysts. The presence of septa had high sensitivity (81%) and relatively low specificity (61%). This result is consistent with those of previous reports describing septa in mucinous cystic neoplasms, cyst-forming intraductal papillary neoplasms of the bile duct, and solitary bile duct cysts [8, 12]. In the previous studies, septa were present in 17 of 24 patients (71%) with mucinous cystic neoplasms or cyst-forming intraductal papillary neoplasms of the bile duct [8] and in 17 of 20 patients (85%) with mucinous cystic neoplasms [12]. Furthermore, we divided the septa according to location and thickness. Central septa had a high sensitivity (80%) and high specificity (83%) for differentiating mucinous cystic neoplasms and of the bile duct from solitary bile duct cysts. Thick septa did not have a statistically significant difference in this respect. However, thick AJR:202, January 2014 89

Kim et al. TABLE 4: Sensitivity and Specificity of CT Findings in Diagnosis of Mucin- Producing Cystic Neoplasms and Solitary Bile Duct Cysts CT Finding Sensitivity (%) Specificity (%) Odds Ratio 95% CI Presence of septa 80.6 (25/31) 61.3 (19/31) 6.6 2.1 20.8 Central septa 80.0 (20/25) 83.3 (10/12) 20.0 3.3 121.8 Mural nodule 61.3 (19/31) 100.0 (31/31) NA Infinite Upstream bile duct dilatation 48.4 (15/31) 90.3 (28/31) 8.8 2.2 34.9 Downstream bile duct dilatation 22.6 (7/31) 100.0 (31/31) NA Infinite Note Data in parentheses are numbers of patients. NA = not applicable (no solitary bile duct cysts exhibited mural nodule or downstream bile duct dilatation). septa were noted only in mucinous cystic neoplasms and cyst-forming intraductal papillary neoplasms of the bile duct. Therefore, thick septa was also an important finding differentiating mucinous cystic neoplasms and cystforming intraductal papillary neoplasms of the bile duct from solitary bile duct cysts. To our knowledge, no previous study has evaluated the location and thickness of septa for differentiating hepatic cystic lesions. Our study showed that the presence of mural nodules was a statistically significant finding for differentiating mucinous cystic neoplasms and cyst-forming intraductal papillary neoplasms of the bile duct from solitary bile duct cysts. The presence of mural nodules had low sensitivity (61%) and high specificity (100%). No solitary bile duct cysts had mural nodules. According to results of a previous study [8], mural nodules were present in 13 of 24 patients (54%) with mucinous cystic neoplasms or cyst-forming intraductal papillary neoplasms of the bile duct. Another study [12] showed that mural nodules were present in 6 of 20 patients (30%) with mucinous cystic neoplasms but was not present in any patient with a solitary bile duct cyst. However, because intracystic debris can mimic mural nodules of hepatic cystic lesions, meticulous comparison of unenhanced and contrast-enhanced CT images is important to ensure correct image interpretation for patients with hepatic cystic lesions. Our study showed that upstream and downstream bile duct dilatation were statistically significant findings differentiating mucinous cystic neoplasms and cyst-forming intraductal papillary neoplasms of the bile duct from solitary bile duct cysts. In our study, upstream bile duct dilatation was present in 3 of the 31 patients (10%) with solitary bile duct cysts and 15 of the 31 patients (48%) with mucinous cystic neoplasms or cyst-forming intraductal papillary neoplasms of the bile duct. Upstream bile duct dilatation caused by solitary bile duct cysts is rare, and such cysts are usually large and centrally located in the liver [13]. According to results of a previous study [8], upstream bile duct dilatation was present in five of seven patients (71%) with cyst-forming intraductal papillary neoplasm of the bile duct and in 5 of 17 patients (29%) with mucinous cystic neoplasms. In our study, downstream bile duct dilatation was present in 7 of 16 patients (44%) with cyst-forming intraductal papillary neoplasms of the bile duct but was not noted in any patient with mucinous cystic neoplasms or solitary bile duct cysts. The previous study [8] showed that downstream bile duct dilatation was present in four of the seven patients (57%) with cyst-forming intraductal papillary neoplasms of the bile duct. Communication between a cystic lesion and the bile duct is an important macroscopic feature of cyst-forming intraductal papillary neoplasm of the bile duct [1, 14]. Our study showed that 5 of the 16 patients (31%) with these lesions had communication between a cystic lesion and the bile duct and that no patient with a mucinous cystic neoplasm or solitary bile duct cyst had such a communication. A previous study [3] showed that communication was present in 6 of 12 patients (50%) with cyst-forming intraductal papillary neoplasms of the bile duct. Our study showed that the mosaic pattern was found in 3 of the 15 patients (20%) with mucinous cystic neoplasms and in 1 of the 16 patients (6%) with cyst-forming intraductal papillary neoplasms of the bile duct and that no patient with a solitary bile duct cyst had a mosaic pattern. Communication and mosaic pattern were not statistically significant in differentiating mucinous cystic neoplasm and cyst-forming intraductal papillary neoplasm of the bile duct from solitary bile duct cyst. However, communication between a cystic lesion and the bile duct was a specific finding for cyst-forming intraductal papillary neoplasms of the bile duct, and mosaic pattern was a specific finding for mucinous cystic neoplasm. Thick septa, mural nodule, and downstream bile duct dilatation are specific findings of mucin-producing cystic neoplasm of the liver. But the presence of septa, central septa, and upstream bile duct dilatation may also be findings of solitary bile duct cysts. The gray zone between mucin-producing cystic neoplasms and solitary bile duct cysts is inevitable. To solve this problem, we presented combined CT findings (Table 5). If one or two CT findings are present, the possibility of solitary bile duct cyst cannot be excluded. However, if more than three CT findings are present, the possibility of solitary bile duct cyst is low. There were several limitations to our study. First, some degree of selection bias could not be avoided because the study was retrospective. We included only solitary bile duct cysts treated by enucleation or hepatic resection. Because most solitary bile duct cysts require no treatment or follow-up, the results of our study should be applied to hepatic cystic lesions that are symptomatic or mimic cystic neoplasms of the liver. Second, we did not use other imaging modalities, such as MRI and direct cholangiography. The incidence of septa and mural nodules may be increased at MRI, and the incidence of communication TABLE 5: Combined CT Findings in Patients With Mucin-Producing Cystic Neoplasms and Intraductal Papillary Neoplasms of the Bile Duct No. of CT Findings Mucin-Producing Cystic Neoplasms + Intraductal Papillary Neoplasms of the Bile Duct (n = 31) Solitary Bile Duct Cysts (n = 31) 1 30 (97) 13 (42) 2 27 (87) 4 (13) 3 18 (58) 0 4 8 (26) 0 5 3 (10) 0 Note Data are numbers of patients with one or more of the following CT findings: presence of septa, central septa, mural nodule, upstream bile duct dilatation, and downstream bile duct dilatation. Data in parentheses are percentages. 90 AJR:202, January 2014

CT of Hepatic Cystic Neoplasms and Cysts between a hepatic cystic lesion and the bile duct may be increased at direct cholangiography. However, CT is currently the primary imaging modality for evaluating patients for surgical resection of the liver. Therefore, the findings in our study are valuable when CT is used for the evaluation of cystic lesions of the liver. Third, coronal reformation images were available for only 13 of 62 patients, and these images might have caused bias in our study. Fourth, we did not use a multiple-reader study design, so we did not evaluate interreader agreement on CT findings. Conclusion The presence of septa, central septa, mural nodules, upstream bile duct dilatation, and downstream bile duct dilatation are characteristic CT findings of mucinous cystic neoplasms and cyst-forming intraductal papillary neoplasms of the bile duct that differ from those of solitary bile duct cysts. When these CT findings are used in combination, mucinous cystic neoplasms and cyst-forming intraductal papillary neoplasms of the bile duct can be differentiated from solitary bile duct cysts with a high degree of diagnostic accuracy, and such a distinction provides valuable information to clinicians for guiding patient treatment. Acknowledgment We thank Seon Ok Kim, Department of Clinical Epidemiology and Biostatistics, Asan Medical Center, for help with statistical analysis in this study. References 1. Bosman FT, Carneiro F, Hruban RH, Theise ND, eds. WHO classification of tumours of the digestive system, 4th ed. Lyon, France: International Agency for Research on Cancer Press, 2010:217 224, 236 238, 254 261 2. Devaney K, Goodman ZD, Ishak KG. Hepatobiliary cystadenoma and cystadenocarcinoma: a light microscopic and immunohistochemical study of 70 patients. Am J Surg Pathol 1994; 18:1078 1091 3. Lim JH, Zen Y, Jang KT, Kim YK, Nakanuma Y. Cyst-forming intraductal papillary neoplasm of the bile ducts: description of imaging and pathologic aspects. AJR 2011; 197:1111 1120 4. Emre A, Serin KR, Ozden I, et al. Intrahepatic biliary cystic neoplasms: surgical results of 9 patients and literature review. World J Gastroenterol 2011; 17:361 365 5. Vogt DP, Henderson JM, Chmielewski E. Cystadenoma and cystadenocarcinoma of the liver: a single center experience. J Am Coll Surg 2005; 200:727 733 6. Oh TH, Kim MH, Lee SK, et al. Thirteen cases of intrahepatic biliary cystadenoma and cystadenocarcinoma: a single center experience. Korean J Gastroenterol 2006; 47:379 385 7. Blonski WC, Campbell MS, Faust T, Metz DC. Successful aspiration and ethanol sclerosis of a large, symptomatic, simple liver cyst: case presentation and review of the literature. World J Gastroenterol 2006; 12:2949 2954 8. Lim JH, Jang KT, Rhim H, Kim YS, Lee KT, Choi SH. Biliary cystic intraductal papillary mucinous tumor and cystadenoma/cystadenocarcinoma: differentiation by CT. Abdom Imaging 2007; 32:644 651 9. Thomas JA, Scriven MW, Puntis MC, et al. Elevated serum CA 19-9 levels in hepatobiliary cystadenoma with mesenchymal stroma: two case reports with immunohistochemical confirmation. Cancer 1992; 70:1841 1846 10. Choi HK, Lee JK, Lee KH, et al. Differential diagnosis for intrahepatic biliary cystadenoma and hepatic simple cyst: significance of cystic fluid analysis and radiologic findings. J Clin Gastroenterol 2010; 44:289 293 11. Koffron A, Rao S, Ferrario M, et al. Intrahepatic biliary cystadenoma: role of cyst fluid analysis and surgical management in the laparoscopic era. Surgery 2004; 136:926 936 12. Seo JK, Kim SH, Lee SH, et al. Appropriate diagnosis of biliary cystic tumors: comparison with atypical hepatic simple cysts. Eur J Gastroenterol Hepatol 2010; 22:989 996 13. Kanai T, Kenmochi T, Takabayashi T, et al. Obstructive jaundice caused by a huge liver cyst riding on the hilum: report of a case. Surg Today 1999; 29:791 794 14. Li T, Ji Y, Zhi XT, et al. A comparison of hepatic mucinous cystic neoplasms with biliary intraductal papillary neoplasms. Clin Gastroenterol Hepatol 2009; 7:586 593 AJR:202, January 2014 91