Role of biliary tract cytology in the evaluation of extrahepatic cholestatic jaundice.

BACKGROUND: Endoscopic evaluation is critical in assessing the cause of obstructive jaundice. Cytological techniques including bile aspiration and biliary brushings have become the initial diagnostic modality. AIM: The aim of this study is to evaluate the role of endoscopic biliary tract cytology as a diagnostic tool in the evaluation of extrahepatic cholestatic jaundice.
MATERIALS AND METHODS: A total of 56 biliary tract specimens including 34 bile aspirations and 22 biliary brushings from 41 consecutive patients who had presented with obstructive jaundice and underwent endoscopic retrograde cholangiopancreatography (ERCP) were assessed by cytological examination. The smears prepared were analyzed for standard cytological features.
RESULTS: Cytologic diagnosis was adenocarcinoma in 13 (31.7%) cases, atypical in 2 (4.9%), reactive in 3 (7.3%) and benign changes in 19 (46.3%) cases. 4 (9.8%) cases were non-diagnostic. Serum bilirubin was significantly elevated in the malignant group. Biliary stricture was the most common finding on ERCP (68.3%). On cytological examination, presence of solitary, intact atypical cells, enlarged nuclei, irregular nuclear membrane, coarse chromatin and nucleoli were important cytologic criteria for differentiating malignant from benign biliary specimens.
CONCLUSIONS: Regular use of bile cytology and brushings during ERCP evaluation of extrahepatic cholestatic jaundice is invaluable in obtaining a morphologic diagnosis. A systematic approach, use of strict cytomorphologic criteria and inclusion of significant atypia as malignant diagnosis may improve the sensitivity.

Introduction

Cholestasis is known as an etiologically diverse clinical entity which requires a broad differential diagnostic workup.[1] Extrahepatic cholestasis, characterized by dilated bile ducts, is caused by either a bile duct stone or stricture, with stricture most often related to a malignancy.[2]

Suspicion of biliary duct malignancy is often raised by clinical symptoms and signs of cholestasis (painless jaundice, pruritus) with elevated serum hepatic enzyme level, where as pancreatic cancer is more insidious in its growth and presents with less specific signs and symptoms. Pre-operative diagnosis of these malignancies remains a challenge.[3]

The direct visualization by endoscopic retrograde cholangiopancreatography (ERCP) remains the gold standard in the evaluation of the extrahepatic bile duct.[1] In addition to establishing an accurate anatomic diagnosis, it provides an etiological diagnosis with cytologic methods for evaluating biliary ductal epithelium by brushings and direct examination of bile and therapeutic biliary stenting to relieve symptoms of obstruction.[45]

Endoscopic evaluation of biliary tract is critical in assessment of various inflammatory, infectious and neoplastic conditions and it also complements the available radiological techniques, especially in cases of obstructive jaundice.[6]

Cytological techniques have become the initial diagnostic modality in many cases since biliary lesions are not always readily accessible to biopsy. Examination of bile duct fluid is a relatively non-invasive method for establishing a diagnosis of malignancy. Currently, bile duct brushings are the favored method of epithelial cell sampling.[78]

The aim of the present study was to evaluate the role of endoscopic biliary tract cytology as a diagnostic tool in evaluation of the cause of extrahepatic cholestatic jaundice.

Materials and Methods

Over a period of 1½ years, we analyzed 56 biliary tract specimens from 41 consecutive patients who had presented with symptoms of obstructive jaundice and underwent ERCP. Bile aspiration and biliary brushing specimens were obtained from 34 and 22 patients respectively. Both brushings and bile aspiration specimens were obtained from 15 cases. Smears from patients who had not presented with symptoms of obstructive jaundice were excluded from the study.

During ERCP, bile for cytological examination was aspirated through a catheter from bile duct by the Gastroenterologist and was collected in clean plastic specimen container. The specimen was sent to the laboratory immediately and was processed within 1 h of collection. However, in case of delay in transportation the specimen was kept in refrigerator as these specimens degenerate very rapidly due to enzymatic activity and presence of bile salts. The specimen was examined grossly for its volume and appearance. Bile was centrifuged for 5 min at 1500 revolutions/min. The sediment was smeared on 2 slides, immediately fixed in methanol and stained with H and E and Papanicolaou (Pap) technique.

Endoscopic brushings from abnormal appearing mucosa were directly smeared onto the slides in endoscopy room. The slides were fixed in alcohol and stained with Pap technique. Both preparations were analyzed for standard cytologic features.

Criteria for cytomorphologic assessment

The cytologic features analyzed were cellularity, architecture, presence/absence of solitary, intact atypical cells, cell size, nuclear features, necrosis and presence/absence of bile and inflammatory cells in the background. Based on the analysis of the above features the specimens were categorized into four groups: Benign, reactive, atypical and malignant.

Smears diagnosed as benign showed monolayered sheets or palisading strips of cells with low nuclear to cytoplasmic (N/C) ratio, smooth nuclear membranes, fine chromatin pattern and absence of nucleoli. Smears diagnosed as reactive showed slight nuclear crowding and mild nuclear enlargement with hyperchromasia, but with preserved polarity. Smears with features that fell short of a malignant diagnosis, but showed significant architectural and nuclear atypia were categorized as atypical. Smears with characteristics of malignancy displayed marked architectural disarray, presence of 3 D clusters and micropapillae, crowded sheets and solitary, intact atypical cells. There was nuclear enlargement with increased N/C ratio, irregular nuclear membrane, coarse chromatin and prominent nucleoli.[4]

In addition, associations of factors including age, gender, liver function tests and stricture location were also assessed. The threshold values of liver function tests used for comparison between positive and negative for malignancy were as follows-serum bilirubin (75μmol/L = 4.5 mg/dL); alanine transaminase (ALT) (100 U/L); aspartate transaminase (AST) (150U/L) and alkaline phosphatase (ALP) (400 U/L).[5] Statistical analysis was carried out using the SPSS17.0 version and results were compared between groups using χ2 test wherever required.

Results

There were 56 biliary tract specimens comprising of 34 bile aspiration and 22 biliary brushings obtained by ERCP from 41 patients. 19 were males and 22 were females, age ranging from 35 to 80 years with mean age of 57 years.

All patients had presented with features of obstructive Jaundice. The other associated symptoms were abdominal pain in 58.5%, fever in 29.8%, decreased appetite in 26.8% and weight loss in 26.8%.

For statistical purpose, cytological diagnoses were categorized into two broad categories, i.e., positive for malignancy in 13 cases (31.7%) and negative for malignancy in 24 cases (58.5%), the latter including smears with benign cytological features in 19 (46.3%), reactive in 3 (7.3%) and atypical in 2 (4.9%) cases. 4 cases (9.8%) were considered non-diagnostic because of absence of biliary epithelial cells. [Table 1]

Table 1

Results of cytologic findings

Serum bilirubin was significantly elevated in the malignant group (P < 0.05). Mean bilirubin level in malignant group was 12.3 mg/dL and in benign group was 7.2 mg/dL. However, no significant correlation was observed between ALT, AST and ALP levels and a malignant diagnosis.

Biliary stricture was the most common finding observed on ERCP in 28 (68.3%) cases, of which lower common bile duct (CBD), mid CBD and hilar stricture were seen in 14 (50%), 2 (7.1%) and 12 (42.9%) cases respectively. A provisional diagnosis of malignant stricture for evaluation was suggested in 16 (39%) cases, periampullary carcinoma in 4 (9.8%) cases, pancreatic carcinoma in 3 (7.3%) cases and Klatskin tumor in 5 (12.2%) cases in correlation with other radiological findings. Other findings on ERCP were cholangitis (14.6%), cholelithiasis and/or choledocholithiasis (7.4%) and extrinsic compression or CBD obstruction (9.7%). The sensitivity of ERCP to cytological diagnosis was 61.5% and specificity was 84.2%.

Cytologic features

Bile specimens (N = 34) comprised of bile aspiration without manipulation in 19 cases and following biliary brushings in 15 cases. Biliary epithelial cells were absent in 6 cases out of which in 4 cases (66.6%) bile was collected without manipulation of stricture. One case showed poorly preserved morphology. All these cases were considered non-diagnostic. Bile aspiration smears were adequately cellular in 59.3% of cases.

The smears were positive for malignancy in 5 cases and negative in 22 cases, which included 20 cases with benign cytological findings and one case each with reactive and atypical changes.

In the benign group, the epithelial cells were arranged in flat sheets (61.9%) and groups (28.6%) with absence of nuclear overlapping. The cells had low N/C ratio, round to oval nucleus with smooth nuclear membranes and evenly dispersed fine chromatin [Figure 1a]. Reactive cells in addition displayed slight variability in nuclear size and inconspicuous nucleoli [Figure 1b]. Atypical smears showed cells in clusters and groups with moderately increased N/C ratio, nuclear enlargement with irregular nuclear membrane, coarse chromatin and occasional cells with prominent nucleoli and absence of bile pigment [Figure 1c].

Figure 1

Bile cytology (a) benign ductal cells in honeycomb pattern (Pap, ×400). (b) Reactive cells with slight variability in nuclear size and inconspicuous nucleoli, bile pigment is present (Pap, ×400). (c) Atypical cells with moderate nuclear enlargement and prominent nucleoli, absence of bile pigment (Pap, ×400). (d) Malignant cells with overlapping nuclei, pleomorphic nucleus and prominent nucleoli, solitary, intact atypical cells, inflammatory cells in background and absence of bile pigment (Pap, ×100) inset — nuclear molding (Pap, ×400)

The smears reported positive for malignancy displayed 3D clusters and papillaroid fragments with overlapping of nuclei in 80%. Presence of solitary, intact atypical cells, nuclear enlargement and high N/C ratio, vesicular to hyperchromatic chromatin with irregular nuclear membrane were seen in all cases. 60% of the cases showed prominent nucleoli. Necrosis was seen in 40% of the cases with the presence of inflammatory cells and blood. Bile pigment was absent in all cases [Figure 1d] [Table 2].

Table 2

Cytologic features of bile aspiration

Biliary brushings (N = 22) were adequately cellular in 86.6% of cases with good preservation. 11 cases were positive for malignancy and an equal number of cases were negative for malignancy. Negative diagnosis included 8 cases with benign changes, 2 cases with reactive and 1 case with atypical cytologic features. Benign smears were composed of cohesive flat sheets of well-aligned glandular epithelial cells. The cells formed 2 D groups and clusters. The cells had low N/C ratio, round to oval nucleus with smooth nuclear membranes and evenly dispersed fine chromatin [Figure 2a]. Reactive smears showed cells in clusters and displayed cytological features similar to the benign category except slight variability in nuclear size and inconspicuous nucleoli [Figure 2b]. Atypical smears showed cells in clusters and groups with moderately increased N/C ratio, nuclear enlargement with irregular nuclear membrane, fine to coarse chromatin and occasional cells with small nucleoli [Figure 2c].

Figure 2

Biliary brush cytology (a) benign tall columnar bile duct cells with basal nuclei (Pap, ×400). (b) Reactive cell clusters in mosaic with slight nuclear enlargement and inconspicuous nucleoli (Pap, ×400). (c) Atypical cell cluster with overlapping, moderate nuclear enlargement, irregular nuclear membrane and absence of bile pigment (Pap, ×400). (d) Malignant cells with overlapping, hyperchromatic coarse clumped chromatin and prominent nucleoli (Pap, ×400). Inset — solitary, intact atypical cells (Pap, ×400)

Smears positive for malignancy displayed tightly cohesive 3 D clusters and papillaroid fragments (63.6%), small groups (18.2%) and glandular arrangement (18.2%). There was considerable nuclear overlapping (72.7%). The cells had high N/C ratio, enlarged nuclei with irregular nuclear membrane, vesicular to hyperchromatic nucleus with prominent nucleoli. Solitary, intact atypical cells were present in all cases. The background showed presence of inflammatory cells (100%), blood (45.5%) and necrosis (36.4%). Bile pigment was absent in 72.8% of cases [Figure 2d] [Table 3].

Table 3

Cytologic features of biliary brushing

Of the above 41 cases, in 15 both bile aspiration and biliary brushing specimens were obtained. Adenocarcinoma was diagnosed in 3 cases on smears prepared by both procedures. However, in four cases, malignant diagnosis was made on brushings though benign cytological features were reported on bile aspiration. This was due to scant cellularity in 3 cases and poorly preserved morphology in one case. Two cases, which were non-diagnostic on bile aspiration due to absence of biliary epithelial cells were satisfactory for interpretation on biliary brushings. Both these cases were benign on brush cytological examination. The bile aspiration showed a specificity of 100%, sensitivity of 42.9% and over all diagnostic accuracy of 73.3%.

Clinico-pathological correlation confirmed a benign diagnosis in 21 cases. However, false negative diagnosis was observed in 3 cases including 2 cases with atypical cytomorphological features and 1 case displaying reactive changes.

One case with atypical morphology had clinically presented with ascites and enlarged abdominal lymph nodes. Serum amylase, gamma glutamyltranspeptidase(GGTP), carcino embryonic antigen (CEA) and carbohydrate antigen (CA) 19-9 were markedly elevated. Ultrasonography revealed a pancreatic mass with similar lesions in liver. On ERCP lower CBD stricture was seen. Ascitic fluid analysis was positive for adenocarcinoma. Diagnosis of pancreatic carcinoma with metastasis to liver was made.

Other case with atypical cytomorphological features was clinically and radiologically suggestive of periampullary carcinoma. On ERCP ampulla was bulky with dilated CBD and intrahepatic biliary radicals. CA19-9 and GGTP were elevated. However, lesional biopsy carried out subsequently showed only necrotic material and was inconclusive for opinion.

In one case with reactive cytological features, ultrasonography revealed grossly dilated gall bladder, dilated hepatic duct, proximal duct and main pancreatic duct and CA 19-9 was markedly elevated. Whipple's procedure was performed and moderately differentiated periampullary adenocarcinoma was diagnosed on histopathology.

Thus, on clinico-pathological correlation malignant diagnoses were confirmed in 16 cases. 6 cases had bile duct adenocarcinoma, 4 cases cholangiocarcinoma, 3 cases periampullary adenocarcinoma and 3 cases pancreatic adenocarcinoma.

Discussion

Biliary tract cancer is the second most common primary hepatobiliary cancer. They have been divided into cancers of extra hepatic bile ducts, gallbladder and ampulla of vater. Approximately, 2000-3000 new cases of bile duct cancers are diagnosed per year.[9]

Obstructive jaundice is the most common presentation associated with biliary system caused by a stricture or obstruction. Strictures could be due to pancreatitis, ascending cholangitis, gall stones or neoplasms. Obstructions are caused by either extrinsic compression by enlarged abdominal lymph nodes or carcinoma head of pancreas or intrinsic abnormality most likely due to neoplasms within the bile duct system.[10]

Malignancy was more common in older age when compared to the benign group. Furthermore, women were affected more frequently (53.7%) than men. Similar observation were seen by other authors.[710]

Serum bilirubin is an important discriminator between malignant and benign conditions and can be utilized in the initial diagnosis.[5] In the present study, serum bilirubin levels were significantly elevated in malignant group; however, elevation in liver enzymes was statistically not significant. Mansfield et al.[11] observed a mean bilirubin level of 9.4 mg/dL in pancreatic and distal biliary tract malignancies. Al-Mofleh et al.[12] observed serum bilirubin level of 84 mmol/L as the best predictor of malignant biliary stricture with high sensitivity. They also observed a poor sensitivity and specificity for AST, ALT and ALP levels.

ERCP has been recognized as a useful and safe procedure for evaluation of the biliary systems. The two most common cytologic methods employed are examination of bile and brushings obtained at ERCP.[410] In this study, both these methods were evaluated. Tissue biopsies for histologic examination were not performed due to a relatively high rate of complications including direct leakage with peritonitis, hemorrhage and scar formation with stricture.[4]

With advances in fibre optics, the yield of exfoliative cytology of bile samples has improved with reductions in false positives.[13] In the present study for the diagnosis of malignancy, specificity was 100% though sensitivity was 42.9%. Results from literature show sensitivity ranging from 6% to 63% and 100% specificity.[1415]

The limited sensitivity of this method may be due to lower number or absence of cells in the samples, degenerative changes in cells, tumor desmoplasia and submucosal location of neoplasms without significant intraductal exfoliation and extrinsic tumors that compress the duct, leading to the impression of a primary stricture. Further, shortcomings in cytological preparations such as air drying artifacts and cellular obscuring by clumping, necrosis and inflammation contribute to interpretative problems.[1314]

Osnes et al.[16] first reported obtaining brushings from biliary tract. However, it took another decade for it to become the favored method of sampling. Since then biliary tract brushings have taken precedence over the bile exfoliative cytology. This may be because samples obtained are more cellular due to traumatisation of the duct epithelium through contact with abrasive bristles on brush. Results from literature show sensitivity rates ranging from 30% to 80% with nearly 100% specificity.[17]

In an effort to improve the diagnostic accuracy of cytologic examination several investigators have proposed specific diagnostic criteria for malignancy. Jin et al.[13] and Nakajima et al.[18] have proposed the cytologic criteria for bile specimens and determined that loss of honeycombing arrangement, cell in cell arrangement, enlarged nuclei, loss of polarity, flat nuclei and bloody background were associated with malignancy. Cohen et al.[19] used multivariate logistic regression analysis on bile duct brushings to analyze 18 variables and concluded that the criteria of nuclear molding, chromatin clumping and increased N/C ratio resulted in an increased diagnostic accuracy.

Renshaw et al.[8] observed that an overall assessment of malignancy based on degree of atypia was more reproducible than any set of criteria and also resulted in higher sensitivity for malignancy with small decrease in specificity.

Our study shared many similarities in the cytomorphologic analysis with other studies.[3810131718] Small 3D clusters, acinar pattern, papillaroid arrangement and bloody and necrotic background were highly specific for malignancy. Presence of solitary, intact atypical cells showed a high sensitivity and specificity for malignancy. Nuclear features including nuclear enlargement, high N/C ratio, irregular nuclear membrane, prominent nucleoli and vesicular to coarse chromatin were highly sensitive for malignancy.

Diagnostic accuracy that stems from endoscopic brushings is not absolute and false negatives may occur. Application of strict diagnostic criteria is essential to ensure high specificity for carcinoma and also to avoid false positive cytologic results.[20]

A diagnosis of atypia or lack of obvious features of carcinoma on smears should prompt the clinician to perform repeat brushings or additional studies to establish a conclusive diagnosis of malignancy,[21] as observed in the present study. Volmar et al.[14] has also reported a significant number of cases with suspicious and atypical cytological diagnoses, which were found to be malignant on follow-up.

Renshaw et al.[8] observed that false negative cases were due to sampling errors. This was more common with pancreatic malignancy, which is difficult to diagnose on basis of bile duct brushings.

Suggested reasons for high false negative rate on exfoliative cytology of pancreatic carcinoma include enzymatic digestion of cells, pancreatic duct stricture, intramural and extramural tumor growth and the distance between sampling site and the tumor. Sampling directly from pancreatic duct may improve diagnostic sensitivity. However sampling from pancreatic ducts is limited by inaccessibility, tortuosity, small caliber of pancreatic ducts and high possibility of complications.[322]

In conclusion, a regular use of this simple procedure of bile cytology and brushings during ERCP evaluation of extrahepatic cholestasis is invaluable in obtaining a morphologic diagnosis with definitive malignant cytology. A larger series of cases will add to the experience of the cytopathologist enabling to make a confident diagnosis. Inclusion of significant atypia as malignant results will improve the sensitivity of this test.

The challenge for future developments in this field is to develop strategies to improve lesion targeting facilitating retrieval of cytology specimens with higher quality.[7]