J Hepatobiliary Pancreat Sci DOI 10.

1007/s00534-011-0419-0

ORIGINAL ARTICLE

Bile aspiration cytology in diagnosis of bile duct carcinoma: factors associated with positive yields
Yasser A. Abdelghani Yoshifumi Arisaka Daisuke Masuda Michiaki Takii Reiko Ashida Madeha M. Makhlouf Yasser M. Fouad Motomu Tsuji Yoshitaka Kurisu Kazuhide Higuchi

Japanese Society of Hepato-Biliary-Pancreatic Surgery and Springer 2011

Abstract Background/Purpose In bile duct carcinoma (BDC) patients, bile aspiration cytology (BAC) is an established method for cytodiagnosis. However, almost all previous reports investigated the biliary strictures caused not only by BDC but also by gallbladder and pancreatic carcinomas. Therefore, BAC in BDC patients only has not yet been investigated sufciently. The aim of this study was to evaluate the actual sensitivity of BAC and to evaluate the factors that affect positive yields of BAC in patients with dened BDC. Methods Data on 47 consecutive patients with denite BDC, who underwent BAC via endoscopic nasobiliary drainage (ENBD) or percutaneous transhepatic cholangiodrainage (PTCD), were retrospectively collected. Fourteen factors were studied for association with positive BAC. Results The number of cytological samplings ranged from 1 to 14 times. The cumulative diagnostic yield was 72.3% (34/47), and 32 positive results were obtained at a maximum of six samplings. Independent factors associated with positive BAC were perihilar location, stricture length C2 cm, and macroscopic papillary type. Conclusion In BDC patients with ENBD or PTCD, repeated BAC is useful, and six times was the optimum
Y. A. Abdelghani Y. Arisaka D. Masuda M. Takii R. Ashida K. Higuchi (&) Second Department of Internal Medicine, Osaka Medical College, 2-7 Daigaku-machi, Takatsuki, Osaka 568-8686, Japan e-mail: higuchi@poh.osaka-med.ac.jp Y. A. Abdelghani M. M. Makhlouf Y. M. Fouad Department of Tropical Medicine and Gastroenterology, Minya University, Minya, Egypt M. Tsuji Y. Kurisu Department of Pathology, Osaka Medical College, 2-7 Daigaku-machi, Takatsuki, Osaka 568-8686, Japan

number of repeat samplings. Although the sensitivity of BAC is not sufcient for the preoperative diagnosis of malignant biliary stricture, the three independent factors noted above predict positive yields and indicate whether or not BAC should be repeated up to six times. Keywords Biliary cytology Bile aspiration cytology Endoscopic nasobiliary drainage Percutaneous transhepatic cholangiodrainage Bile duct carcinoma

Introduction Although there are many reports regarding the preoperative histological diagnosis of malignant biliary strictures, the sensitivities of the reported methods are not sufcient for obtaining a denite preoperative diagnosis. It is often difcult to distinguish malignant from benign biliary strictures by imaging only, so obtaining histological evidence is very important for planning the therapeutic strategies in these patients. Endoscopic transpapillary biopsy or brushing cytology under uoroscopic guidance at the time of endoscopic retrograde cholangiopancreatography (ERCP) is one useful method, but it is not always simple to obtain adequate specimens, from the viewpoint of technical skills. Cholangioscopy-guided targeted biopsy is another possible method to dene the diagnosis; however, this manipulation is more complicated, and the sample sizes are often too small for interpretation by pathologists. Bile aspiration cytology (BAC) is the easiest and oldest method of obtaining a cytological specimen, and the level of technical skill required is lower than that of the abovestated methods. Moreover, as biliary drainage such as endoscopic nasobiliary drainage (ENBD) or percutaneous

123

J Hepatobiliary Pancreat Sci

transhepatic cholangiodrainage (PTCD) must be performed in most patients with biliary stricture, BAC can be done easily and can be repeated any number of times without difculty. Although the sensitivity of BAC for diagnosing malignant biliary stricture is still low, ranging from 33 to 56%, the specicity is very high and few false-positive diagnoses have been reported [1]. Although there are several descriptions of biliary cytology in large numbers of patients in previous reports, most of these reports included patients with not only malignant but also benign strictures, and not only patients with bile duct carcinoma (BDC) but also those with gallbladder carcinomas, pancreatic carcinomas, and other malignant biliary strictures. Essentially, the biliary strictures caused by BDC are different from those caused by other malignant tumors. Certainly BDCs always arise from the columnar epithelium lining the bile duct (BD) wall, and theoretically malignant cells would be present in the mucosal surface of the BD wall. On the other hand, most of the other malignant strictures are caused by extrinsic invasion or compression of the BD wall, and all cases do not always invade to the mucosal layer of the BD, so malignant cells are not always revealed in the mucosal surface. Thus, when we focus on BDC patients only, it is predictable that the sensitivity of BAC would be higher in these patients. But there are very few reports that address biliary cytology in patients with dened BDC only, and the accuracy and sensitivity of biliary cytology in BDC has not yet been investigated. In addition, the effects of the clinical, biochemical, imaging, and pathological characteristics of BDC on the results of BAC have been rarely reported. Recently, advances in imaging methods have provided us with marvelous information, and it is now not so difcult, preoperatively, to differentiate primary biliary lesions, including BDC, from biliary strictures caused by other lesions such as pancreatic tumor. After imaging, the strategy for obtaining histological evidence should be planned. However, in BDC patients, the sensitivity of biliary cytology, and the effects of the clinical, biochemical, imaging, and pathological characteristics of BDC on the results of BAC have been rarely reported. The aim of this retrospective study was to examine the actual sensitivity of BAC in patients with dened BDC only and to evaluate the factors that affect the positive yields of BAC.

1997 to August 2010 and in whom ENBD or PTCD was done. Written informed consents were obtained from all patients. For inclusion in the study, patients had to have a denite nal diagnosis of extrahepatic BDC, based either on denitive surgical histology (n = 29), autopsy histology (n = 2), or other histopathological evidence, except for patients with BAC with a poor convalescent clinical course, such as those with metastases or those who died from cancer (n = 16). Pancreatic carcinoma, gallbladder carcinoma, ampullary carcinoma, intrahepatic cholangiocarcinoma, and strictures possibly caused by non-primary BDC, such as lymph node metastasis, were all excluded. Regarding samples for BAC, we included aspirated bile juice only. Samples obtained with brushing cytology, with cleaning by saline, and other added methods were all excluded from data analysis. Methods Evaluated factors The following 12 clinical, biochemical, imaging, and pathological characteristics were compared in BDC with positive and negative biliary cytology yields: (1) age, (2) gender, (3) serum total bilirubin, (4) serum carbohydrate antigen 19-9 (CA 19-9), (5) stricture location, (6) stricture length, (7) presence or absence of mass on imaging, (8) size of the mass, (9) pancreato-biliary ductal union (PBDU), (10) macroscopic tumor type, (11) histological tumor differentiation, and (12) depth of tumor invasion. In addition, the method of obtaining the bile sample and the number of cytological examinations necessary to obtain a positive yield were assessed. Clinical and biochemical factors The evaluated clinical factors were age and gender. The biochemical factors were serum total bilirubin (normal 0.11.0 mg/dL) and serum CA 19-9 (normal 037 IU/ml). The serum total bilirubin was obtained from the patient just before biliary drainage. The serum CA 19-9 was obtained on admission or just before drainage or after jaundice had been normalized. Imaging factors

Patients and methods Patients The subjects were 47 consecutive patients with accurate diagnoses of BDC obtained during the period from June Stricture location and length, presence or absence of mass on imaging, and size of the mass were assessed by ultrasonography, multidetector row computed tomography (MDCT), magnetic resonance imaging (MRI), magnetic resonance cholangiopancreatography (MRCP), and

123

J Hepatobiliary Pancreat Sci

endoscopic ultrasonography (EUS). The size of the tumor mass was estimated by ultrasound or MDCT. We classied biliary stricture location into two groups: distal and perihilar bile ducts. The location of extrahepatic BDCs have been subdivided by the Japanese Society of Biliary Surgery (JSBS) into inferior bile duct (Bi), middle bile duct (Bm), and superior bile duct (Bs). In our study, distal bile ducts were dened as inferior and middle bile duct. perihilar bile ducts were dened as superior bile duct and the right or left hepatic ducts. In patients with large tumor size or supercial spreading tumor invading more than one location, we considered the main tumor location. Biliary drainage For ERCP, the procedure was performed in a standard fashion with a side-viewing duodenoscope (Olympus JF200, JF-240 or JF 260V; Olympus, Tokyo, Japan). The insertion of an ENBD tube was performed in the standard manner. The procedure was conducted without dilation of the biliary stricture or endoscopic sphincterotomy. For ENBD tubes, we used 5-, 6- or 7-Fr pigtail types (PBD-21Z; Olympus, or Wilson Cook Medical, Bloomington, IN, USA). For insertion of the PTCD tube, puncture was done under ultrasound image guidance using a 23-gauge Chiba needle. After identifying the stricture by direct cholangiography, a 7.5-Fr sheath introducer and 7.2-Fr pigtail drainage catheter (Wilson Cook Medical) were then inserted over the guidewire and positioned above the stricture. PTCD was indicated in the following cases: (1) strong suspicion of lateral spreading carcinoma, needed for further examination by percutaneous transhepatic cholangioscopy (PTCS). (2) Some types of hilar BDC in which it was necessary to drain 2 or 3 bile duct branches. (3) Patients with previous gastric surgery, e.g., Billroth II gastrectomy or Roux-en-Y operation. (4) Patients in whom cannulation or passage through the stricture failed in ERCP. Cholangiography Cholangiography through the ENBD or PTCD tube enabled us to assess stricture location and length, PBDU, and macroscopic tumor type. Bile aspiration cytology The rst cytological sample of bile juice was obtained immediately after insertion of the ENBD or PTCD tube, and as much fresh bile as possible was aspirated, using a 20-mL syringe. Bile from a placed bag was never used. Specimens were brought to the laboratory as soon as possible under cooling with ice. Bile juice was centrifugally

separated (3000 rpm, 3 min), and the deposit was prepared on glass slides with smear preparation. The slides were stained with Papanicoloau stain, Giemsa stain, and periodic acid-Schiff stain. Bile juice sampling was obtained once per day, and sampling was repeated several times (range 114 times). The endpoint of bile juice sampling was dened as when a positive yield was obtained. If we could not get a positive yield while the patients general condition was recovering, we considered another method for cytological and histological diagnoses, and we then added brushing cytology and/or forceps biopsy sampling. Brushing cytology was done under uoroscopic imaging, using a 5-Fr cytoscopic brush catheter (Wilson Cook Medical). Cytological interpretation The cytopathology slides of patients with BDC who met the inclusion criteria of our study were re-evaluated retrospectively by cytopathologists using the diagnostic bile cytology criteria published by the Japanese Society of Clinical Cytology [2]. In these criteria, clusters of cells are classied into (A) large clusters (with more than 50 cells) and (B) small clusters and isolated cells (with fewer than 50 cells) (Fig. 1). Large clusters showed 3 characteristics: (1) irregularly overlapped nuclei, (2) irregularly arranged nuclei, and (3) irregular cluster margins. Small clusters and isolated cells showed 3 characteristics: (1) enlarged nuclei, (2) irregularly shaped nuclei, and (3) abnormal chromatin. Each bile specimen was classied as negative, suspicious, or positive for malignancy. Cells or clusters were diagnosed as suspicious for adenocarcinoma if 2 characteristics were met for A or B, while cells or clusters were diagnosed as positive for adenocarcinoma if 3 characteristics were met for A or B. Macroscopic ndings In 34 patients in whom either operation or autopsy was done, the resected specimens were opened longitudinally and grossly examined for detection of macroscopic tumor type. In this study, we used the macroscopic classication proposed by the Japanese Society of Biliary Surgery (JSBS) [3]. This classication is based on growth characteristics, with tumors being identied as papillary, nodular, or at type. In addition, there is another classication of growth pattern characterized by the intraepithelial spread of the carcinoma cells. Each type is further divided into expanding type and inltrating type (Fig. 2). In the macroscopic papillary type the margin of the elevated lesion is sharp compared to the surrounding at normal mucosa. The height of the elevated lesion is more than 23 mm.

123

J Hepatobiliary Pancreat Sci

Fig. 1 Cytology of a malignant duct epithelium with clusters of cells (Papanicolaou stain, originally 9400). a large clusters (with more than 50 cells): there are irregularly overlapped nuclei and irregular

cluster margins. b Small clusters and isolated cells (with fewer than 50 cells): there are enlarged nuclei and abnormal chromatin

Fig. 2 Photograph of longitudinally opened gross specimens of bile duct carcinoma. a Macroscopic papillary type (after xation): tumor can be seen in the distal bile duct, and it appears as an elevated lesion with a sharp margin compared to the surrounding at mucosa (arrow). b Macroscopic nodular type (before xation): the margin of the tumor

located at the hilum is gradually elevated from the surrounding at mucosa (arrow); asterisk indicates thickening of the bile duct wall. c Macroscopic at type (after xation): a narrow duct segment with minimal irregularities can be seen in the distal bile duct (arrow), but the tumor lesion cannot be clearly detected

The elevated lesion consisted of tumor cells conned to the mucosal layer. Regarding the macroscopic nodular type; the margin of the elevated lesion is gradually elevated from the surrounding at mucosa. The elevated lesion consisted of tumor cells invading to deeper layers. Regarding the macroscopic at type there is no clear elevated lesion, the expanding type is very rare and almost all cases are inltrating type. Sometimes at type is referred to as inltrating type or diffuse inltrating type. The Armed Forces Institute of Pathology (AFIP) [4] has proposed that carcinomas of the extrahepatic BD appear macroscopically as polypoid indurated masses, nodular lesions, or as diffusely inltrating or constricting lesions. Polypoid indurated masses of the AFIP classication match the papillary type according to the JSBS classication used for this study; similarly, the nodular type matches the same type in the JSBS classication, and the diffusely inltrating type matches the at type. In the sixteen patients in whom neither operation nor autopsy was done, the macroscopic tumor type was

evaluated from the uoroscopic image and peroral cholangioscopy (POCS) or PTCS image (Fig. 3). Histological evaluation After the specimens were examined macroscopically, they were xed in 10% buffered formalin. Serial sections, at 36 mm intervals, were prepared from the entire area of the extrahepatic bile duct. Histological specimens were similarly made from entire areas of the hilar and large intrahepatic bile ducts in the patients who underwent concurrent hepatectomy. These sections were embedded in parafn, and more than 20 sections were cut from each block. All of them were processed routinely for histological study. We evaluated histological tumor differentiation as follows; papillary adenocarcinoma, tubular well-differentiated adenocarcinoma, tubular moderately differentiated adenocarcinoma, and tubular poorly differentiated adenocarcinoma. According to the currently used JSBS classication of tumor invasion into the bile duct wall, serosal invasion is

123

J Hepatobiliary Pancreat Sci

Fig. 3 Macroscopic ndings of bile duct carcinoma (BDC) on cholangiography (endoscopic retrograde cholangiopancreatography [ERCP]). a Macroscopic papillary type: there is an irregular lling defect, denoting fungating papillary tumor, with a sharp edge; also

shown is an irregularity of the bile duct wall above and below (arrow). b Macroscopic nodular type: it appears as stricture with a sharp edge (arrow). c Macroscopic at type: it appears as a stricture with a gradually tapering edge as compared to the nodular type (arrow)

histologically divided into 5 grades in anatomical fashion: mucosal (m), bromuscular (fm), subserosal (ss), exposed to serosa (se), and invaded adjacent organs (si). We evaluated depth of tumor invasion, and we dened early BDC as grades m and fm, and advanced BDC as grades ss, se, and si. Statistical analysis Statistical analysis was performed by using the Statistical Package for Social Science SPSS software (Version 16; SPSS, Chicago, IL, USA) on a personal computer. Statistical analysis was performed using the v2 test or independent sample t-test to assess differences between proportions. Differences were considered statistically signicant when the P value was less than 0.05.

percentage of patients with serum bilirubin C3 mg was not associated with positive BAC. Regarding CA 19-9, we obtained data in 41 patients, and in the other 6 patients insufcient data were available. No statistically signicant difference was noted in serum CA 19-9 between the positive and negative biliary cytology groups; a reference value of CA 19-9 greater than 100 U/mL was associated with an elevated odds ratio (OR 2.1). Imaging factors (Table 1) Biliary stricture (distal; n = 20, and perihilar; n = 27) was revealed in all 47 patients. The yield of biliary cytology showed signicant differences concerning the location of the stricture, with the highest positivity for perihilar BD (88.8%), followed by distal BD (50%; P \ 0.005, OR 8). The length of the biliary strictures ranged from 8 to 53 mm (22.3 10.6 vs. 17.5 8.06) (not signicant). A reference value of 2 cm was selected for the stricture length. A stricture length of C2 cm was more likely to be associated with a positive yield obtained on biliary cytology (P \ 0.05, OR 4.1) (Table 1). Patients with a mass on imaging were almost 3 times (OR 3.3) more likely to have a positive cytology result compared to those without a mass. Mass size (15.8 13.2 vs. 11.3 11.3 cm) seemed to be linked to positive yield on biliary cytology, but the difference was not signicant. However, a cutoff point for mass size of C1 cm was associated with an elevated OR estimate for positive cytology (OR 1.7). Normal PBDU was associated with positive BAC (OR 1.3). Bile aspiration cytology (Tables 2, 3) Cytological detection of malignancy was dened as cases in which the cytologist reported clearly malignant cells or cells suspicious of cancer. All patients underwent repeated cytological sampling via an ENBD tube (n = 31) or PTCD tube

Results Clinical and biochemical factors (Table 1) The forty-seven patients with BDC included 31 males and 16 females, and their ages ranged from 24 to 89 years. Among the forty-seven patients with BDC, we got a positive BAC in 34 patients (72.3%). The sensitivity was 74.1% (23/31) in males and 68.7% (11/16) in females. Age of presentation (mean SD, in the positive and negative biliary cytology groups was 67.1 12.8 vs. 59.7 14.2 years, respectively; independent t-test) seemed to be higher in patients with positive biliary cytology, but the difference was not signicant. However, when we selected a cutoff point of C65 years for age, it was associated with signicant differences (P \ 0.05) and elevated odds ratio (OR) estimates for positive cytology (OR 3.8; v2 test). Total serum bilirubin did not inuence the yield of biliary cytology (6.4 6.8 vs. 4.9 4.08 mg/dl), and total

123

J Hepatobiliary Pancreat Sci Table 1 Clinical, biochemical, and imaging features of cholangiocarcinoma in positive and negative biliary cytology groups P valuea

Variable Sex Male Female Age group C65 years \65 years Total bilirubin C3 mg/dL \3 mg/dL CA 19-9b C100 U/mL \100 U/mL Stricture location Distal Perihilar Stricture length C2 cm \2 cm

No.

Positive ratio (No.)

Odds ratio (95% CI)

31 16 29 18 25 22 15 26 20 27 26 21 34 13 30 20 44 3

74.1% (23/31) 68.7% (11/16) 82.7% (24/29) 55% (10/18) 68% (17/25) 77.2% (17/22) 80% (12/15) 65.3% (17/26) 50% (10/20) 88.8% (24/27) 84.6% (22/26) 57.1% (12/21) 79.4% (27/34) 53.8% (7/13) 80% (24/30) 65% (13/20) 72.7% (32/44) 66.6% (2/3)

NS

0.71 (0.2-2.8)

0.04**

3.8 (114.6)

NS

0.6 (0.172.3)

NS

2.1 (0.479.5)

0.004**

8 (1.835.3)

0.03**

4.1 (1.0416.2)

CI condence interval, No. number, P probability value, NS non-signicant, CA carbohydrate antigen, PBM pancreatico-biliary maljunction, PBDU pancreatobiliary ductal union ** P \ 0.05
a b

Mass on imaging Yes No Mass size C1 cm \1 cm PBDU Normal PBM NS 1.3 (0.1116.08) NS 2.1 (0.637.5) NS 3.3 (0.8313.02)

v2 test

CA19-9 was measured in 44 patients

Table 2 Cumulative sensitivity of bile aspiration cytology related to numbers of sequential samples required to establish a diagnosis Examination times Number of positive cases Cumulative sensitivity 1 18 38.2% 2 21 44.6% 3 26 55.3% 4 29 61.7% 5 30 63.8% 6 32 68% 8 33 70.2% 14 34 72.3%

Table 3 Number of examinations in false-negative cases Number of examinations Number of cases 1 0 2 1 3 2 4 2 5 2 6 0 7 4 13 1 14 1 Total 13

(n = 16). In 18 of the 47 patients, the BAC was positive at the rst cytological examination (sensitivity 38.2%). On repeating BAC, we were able to obtain more positive cases, thus improving the cumulative sensitivity. Cytology was cumulatively positive in 34 of the 47 patients (sensitivity 72.3%; Table 2). The sensitivity of ENBD cytology was 77.4%, while the sensitivity of PTCD cytology was 68.7%. Additionally, at our hospital, in patients in whom bile

aspiration was cytology-negative, brushing cytology and/or biopsy sampling was added, and we nally achieved cytological and histological conrmation of malignancy before operation in 42 out of the 47 patients (89.3%). In our study, the number of repeated BACs ranged from 1 to 14 times. In the 34 patients with positive BAC, we correlated the cumulative positive rate and the number of cytological examinations performed. Of the 34 patients

123

J Hepatobiliary Pancreat Sci Table 4 Pathological features of cholangiocarcinoma in positive and negative biliary cytology groups P valuea

Variable Macroscopic type Papillary Nodular Flat Histological type Papillary Tubular (well) Tubular (moderate)

No.

Positive ratio (no.)

Odds ratio (95% CI)

18 25 4 6 13 10 2 5 26

94.4% (17/18) 52% (13/25) 75% (3/4) 83.3% (5/6) 69.2% (9/13) 80% (8/10) 100% (2/2) 60% (3/5) 69.2% (18/26)

Papillary/nodular Papillary/at Nodular/at NS

0.003** NS NS

15.7 (1.8136) 5.6 (0.27117) 0.36 (0.033.9) _

CI condence interval, No. number, P probability value, NS non-signicant ** P \ 0.05

a

Tubular (poor) Depth of invasion Early (m, fm) Advanced (ss, se, si)

Advanced/early

NS

2.2 (0.188.8)

v2 test,

with positive BAC, the result of cytology was positive in 18 (38.2%) patients at the rst cytological sampling, in 21 (44.6%) patients at the second, in 26 (55.3%) patients at the third, in 29 (61.7%) at the fourth, in 30 (63.8%) at the fth, in 32 (68%) at the sixth, in 33 at the eighth (70.2%), and in 34 (72.3%) at the fourteenth. Positive results were not obtained after the sixth cytological sampling, except for 2 patients. Thus, in 32 of 34 (68%) patients, positive results were obtained by or at the sixth examination. Of the 13 cytology-negative patients, 7 underwent cytological sampling less than six times. Cytological samplings in these 7 patients were performed twice in one patient, three times in 2 patients, four times in 2 patients and ve times in 2 patients. Pathological ndings (macroscopic and microscopic ndings) (Table 4) On analysis of the macroscopic tumor type of BDC, there were 18 papillary types, 25 nodular types, and 4 at types. The distribution of different macroscopic patterns of BDC in relation to the biliary stricture location is demonstrated in Fig. 4. There were no statistically signicant differences. The sensitivity of BAC for the macroscopic papillary type was remarkably high (94.4%), and it was signicantly higher than the sensitivity for the other types (papillary/ nodular: P \ 0.01, OR 15.7, papillary/at: P = 0.09, OR 5.6). The histological tumor differentiation in surgical or autopsy specimens was reviewed. Tubular well-differentiated adenocarcinoma showed the lowest BAC sensitivity (69.2%), while tubular poorly differentiated adenocarcinoma showed the highest sensitivity (100%); there were no signicant differences according to tumor differentiation. On analysis of tumor invasion depth, there were ve patients with an early BDC (m = 4, fm = 1) and 26
Fig. 4 Macroscopic ndings of BDC according to stricture locations

patients with an advanced BDC (ss = 16, se and si = 10). Advanced BDC seemed to be linked to positive BAC (OR 2.2). Multivariate analysis for factors associated with positive BAC (Table 5) In a univariate analysis that we performed, there were four factors with a P value of\0.05 for patients in whom all the data were available (age, perihilar stricture location, stricture length C2 cm, and macroscopic ndings of papillary type). These four factors were entered into a step-wise multivariate logistic regression analysis to search for independent factors associated with a positive biliary cytology result. In this analysis, 3 of these four factors (except for age); namely, perihilar stricture location,

123

J Hepatobiliary Pancreat Sci Table 5 Results of multivariate analysis for factors associated with positive biliary cytology Factor Stricture location Stricture length Macroscopic ndings Age Coefcient -2.116 -1.313 -1.513 Standard error 0.915 0.921 1.707 P value 0.02 0.01 0.03 0.3

stricture length C2 cm, and macroscopic ndings of papillary type, were found to have a predictive value for positive BAC. Complications Only two of the patients who had ERCP developed postERCP pancreatitis, and in both patients the acute pancreatitis was mild. None of the patients with PTCD had any associated complications.

Discussion Most BDC presents with obstructive jaundice, and therefore biliary drainage is mandatory in these patients. The biliary drainage is performed either through the transpapillary route or the percutaneous transhepatic route. Transpapillary drainage is roughly divided into two categories: ENBD for external drainage and endoscopic biliary stenting (EBS) for internal drainage. ENBD or PTCD is not always performed in all patients with BDC, and the drainage method should be selected depending on the patients situation. In patients in whom ENBD or PTCD has been done, the access to the bile duct is already achieved, and BAC can be done more simply than brush cytology or biopsy, and we can achieve a cytological diagnosis with no added effort or risk. In patients in whom EBS is selected for biliary drainage, BAC may be done at the time of ERCP, but the sensitivity will be low. In fact, in our study, the sensitivity of BAC at rst cytological examination was 38.2%, and even when we added the result of brushing cytology and biopsy sampling to the result of BAC at the time of ERCP, the overall sensitivity was 68%, which was less than the cumulative sensitivity of BAC (72.3%). In addition, BAC is easier and safer than brushing cytology and biopsy, which are technically difcult and carry some degree of complication. Therefore, BAC is an effective method for cytological diagnosis of BDC. The sensitivity of BAC ranges from 33 to 56% and is approximately 30% in most series [1, 58]. In our study, the sensitivity of BAC was somewhat higher than that in

most reported studies, and this is because our absolute sensitivity of 72.3% was the cumulative sensitivity, and we included only patients with dened BDC. The sensitivity of BAC via an ENBD tube or PTCD tube at rst biliary sampling was 38.2%, and the cumulative sensitivity was improved by repeated cytological examinations to 72.3%; thus, the sensitivity for detection of malignancy was improved without effort when cytological analysis of bile juice was repeated. In addition, to improve the sensitivity, samples were sent to the laboratory as soon as possible. When bile juice was sticky, saline solution was added to the sample before centrifugation. When the deposit was gummy, the material was pinched with two glass slides for preparation, to spread the specimen. Recently, the ThinPrep technique (Hologic, Bedford, MA) was evaluated in some reports, and it was associated with increased sensitivity of diagnosis [9, 10]. The sensitivity of ENBD cytology was comparatively higher than that of PTCD cytology (77.4 vs. 68.7%). This might be explained by the passage of the ENBD tube through the stricture in all patients with ENBD; thus, there was a continuous brushing effect on the tumor lesion, while in most patients with PTCD, the PTCD tube was positioned above the stricture. The results for the sensitivity of ENBD cytology are in accordance with those of Davidson et al. [6]. Therefore, we suggest that the ENBD tube is a better method for the sampling of biliary cytology than PTCD tube. Regarding the sensitivity of the number of times examination was done, biliary cytological examination done after 6 times was associated with very low positivity. Moreover, biliary sampling done more than 6 times may require a longer period of ENBD tube insertion, which often causes discomfort to the patient. Therefore, from the viewpoint of cost and convenience, six times is costeffective and more convenient to the patient. On the other hand, we found that the sensitivity of 68% (32/47), obtained with six cytology samples of bile, included cases in which sampling was repeated fewer than six times. Additionally, 7 of the 13 patients who were cytologynegative underwent cytological sampling less than six times in this retrospective study. In these patients, BAC was stopped a few times before obtaining positive yields because of the next examination such as brushing cytology or biopsy. Accordingly, if the cytological samplings had been repeated six times in all cases, the sensitivity is likely to have been much higher. Based on these results, it is noteworthy that six examination times using bile obtained via an ENBD tube or PTCD tube was the optimum number. And it anticipated that six repetitions of BAC in BDC patients would provide much higher sensitivity than 72.3%. Although a number of ancillary techniques, such as uorescence in situ hybridization, digital image analysis,

123

J Hepatobiliary Pancreat Sci

and the investigation of K-ras and p53 mutations, have been used in attempts to achieve better sensitivity of biliary cytology, these methods are costly, and are quite technically complex [11, 12]. BAC is not costly and is technically simple; therefore, despite the emergence of these new approaches, BAC remains convenient and useful, when it is repeated. Studies that evaluate the factors affecting the positive yields of BAC are very few. Our study found that there was an association between the yield of BAC and the stricture location, with higher rates of positive cytology in BDC with perihilar location than that located in the distal BD. There have been reports that the intraepithelial spread of BDC is signicantly associated with perihilar location compared to the distal BD [13, 14]. The intraepithelial spread of BDC is associated with the exfoliation of many malignant cells in bile juice; therefore, this might explain the signicantly higher yield of BAC in the diagnosis of BDC located in the perihilar BD. The papillary macroscopic type of BDC is usually friable and sometimes sloughs spontaneously, often spreading supercially along the mucosal surface, resulting in multiple tumors in various BD locations [15]. In our study, papillary macroscopic tumor type was linked with the highest sensitivity for predicting positive results obtained by biliary cytology, followed by at type and nodular type. Histological analysis of tumor differentiation showed that samples from well-differentiated tumors had the lowest sensitivity (69.2%) and this is in agreement with Layeld et al. [16], who reported that well-differentiated tumors were difcult to interpret because the cytological abnormalities may be minimal. Regarding depth of tumor invasion, we found that advanced BD cancer seemed to be linked to positive biliary cytology, and this might be explained by the larger tumor size in the advanced stage accompanied by extensive friability and thus abundant cellularity. Because a higher yield on biliary cytology was found to be associated with advanced bile duct cancer, strict preoperative staging of tumor extent is necessary in that situation. Based on the results of our multivariate analysis for the factors associated with positive BAC in patients with suspicion of biliary stricture caused by BDC, if the lesion has the following three features: perihilar location, length C2 cm, and macroscopic type assessed as papillary after cholangiography, a higher yield of BAC is strongly predicted. Additionally, if we repeat the cytological examination up to six times, the sensitivity will be improved. In conclusion, in BDC patients with ENBD or PTCD, repeated BAC is useful, and six times was the optimum number to repeat the BAC. Although the sensitivity of BAC is not sufcient for the preoperative diagnosis of malignant biliary stricture, the three independent factors

noted above predict positive yields and indicate whether or not BAC should be repeated up to six times.

References
1. Ohtsuka M, Ito H, Kimura F, Shimizu H, Togawa A, Yoshidome H, et al. Results of surgical treatment for ICC and clinicopathological factors inuencing survival. Br J Surg. 2002;89:152531. 2. Hirooka Y, Nakaizumi A, Oka T, Naito Y, Arisaka Y, Minamiguchi S et al. Report of the clinical study for methods to improve the diagnostic accuracy of bile cytology. J Jpn Soc Clin Cytol. 2010;49:714 (in Japanese with English abstract). 3. Japanese Society of Biliary Surgery. General rules for surgical and pathological studies on cancer of biliary tract. 1st English ed. Tokyo: Kanehara; 2001. 4. Albores-Saavedra J, Henson DE, Klimstra S. AFIP atlas of tumor pathology. In: Tumors of the gallbladder, extrahepatic bile ducts, and ampulla of vater, Series III. Washington: Armed Forces Institute of Pathology; 2000. p. 198. 5. Foutch PG, Kerr DM, Harlan JR, Kummet TD. A prospective, controlled analysis of endoscopic cytotechniques for diagnosis of malignant biliary strictures. Am J Gastroenterol. 1991;86:57780. 6. Davidson B, Varsamidakis N, Dooley J, Deery A, Dick R, Kurzawinski T, et al. Value of exfoliative cytology for investigating bile duct strictures. Gut. 1992;33:140811. 7. Manseld JC, Grifn SW, Wadebra V, Matthewson K. A prospective evaluation of cytology from biliary strictures. Gut. 1997;49:6717. 8. Sugiyama M, Atomi N, Wada N, Kuroda A, Muto T. Endoscopic transpapillary bile duct biopsy without sphincterotomy for diagnosing biliary stricture: a prospective comparative study with bile and brush cytology. Am J Gastroenterol. 1996;91:4657. 9. Sheehan MM, Fraser A, Ravindran R, McAteer D. Bile duct brushings cytologyimproving sensitivity of diagnosis using the ThinPrep_ technique: a review of 113 cases. Cytopathology. 2007;18:22533. 10. Waugh MS, Guy CD, Maygarden SJ, Livasy CA, Jones CK, Volmar KE. Use of the ThinPrep method in bile duct brushings: analysis of morphologic parameters associated with malignancy and determination of interobserver reliability. Diagn Cytopathol. 2008;36:6516. 11. Athanassiadou P, Grapsa D. Value of endoscopic retrograde cholangiopancreatography- guided brushings in preoperative assessment of pancreaticobiliary strictures: whats new? Acta Cytol. 2008;52:2434. 12. Hema G, Vijaya R, Larry K, Diana T, Razan MW, Suzanne S, et al. Brush cytology of the biliary tract: retrospective study of 278 cases with histopathologic correlation. Diagn Cytopathol. 2002;26:2737. 13. Nakanishi Y, Zen Y, Kawakami H, Kubota K, Itoh T, Hirano S, et al. Extrahepatic bile duct carcinoma with extensive intraepithelial spread: a clinicopathological study of 21 cases. Mod Pathol. 2008;21:80716. 14. Igami T, Nagino M, Oda K, Nishio H, Ebata T, Yukihiro Y, et al. Clinicopathologic study of cholangiocarcinoma with supercial spread. Ann Surg. 2009;249:296302. 15. Lim JH, Yi C, Lim HK, Won JL, Soon JL, Seung HK. Radiological spectrum of intraductal papillary tumors of the bile ducts. Korean J Radiol. 2002;3:5763. 16. Layeld LJ, Wax TD, Lee JG, Cotton PB. Accuracy and morphological aspects of pancreatic and biliary brushings. Acta Cytol. 1995;39:118.

123