Contraindications

Contraindications to brushing of the biliary tract or pancreatic ducts include contraindications to the techniques used to guide the procedures for ERCP and PTC. Absolute contraindications to ERCP include patient refusal to undergo the procedure; unstable cardiopulmonary, neurologic, or cardiovascular status; and existing bowel perforation. Relative contraindications include structural abnormalities of the esophagus, stomach, or small intestine. PTC is contraindicated in patients with bleeding diatheses or significant ascites.

Contraindications to EUS-guided FNAB include the presence of an uncorrectable bleeding disorder and lack of a safe needle access route.⁶ Gastrointestinal obstruction is an absolute contraindication for EUS-FNAB because of the risk of intestinal perforation.⁷

Rapid On-Site Evaluation

The role of rapid on-site evaluation (ROSE) is not clearly established for brushing cytology. However, if smears are prepared on site, then it is preferable that they be prepared by an experienced cytotechnologist or pathologist.

ROSE, performed by a cytotechnologist or a pathologist, is probably the single greatest factor responsible for improvement in the diagnostic accuracy of FNAB for solid masses of the pancreatobiliary tract. ROSE of selected smears ensures adequacy of the specimen and reduces the number of inadequate specimens caused by poor localization of the lesion, scant cellularity, poor sample preparation, poor sample preservation, or obscuring blood. Thus, ROSE produces significant cost savings and may spare the patient from the need for another procedure.^8–12 ROSE also allows the interpreter to determine the need for adjunctive studies such as flow cytometry, immunohistochemistry, or microbiology and prepare the sample accordingly. ROSE is not indicated for purely cystic masses but can be performed on the solid component of a cystic mass.

Prospective and retrospective studies have documented a higher level of sensitivity for biliary brushings compared with exfoliative cytology in the detection of biliary carcinomas.¹³ A review of the literature shows that duct brushing cytology has an overall sensitivity of 26% to 88.9%, a specificity of 80% to 100%, and an overall accuracy of 48.1% to 96%. In PSC, the sensitivity was reported to be 60%, with a specificity of 89%. The sensitivity of bile duct brushings has been shown to increase after repeated attempts.^14,15 In fact, after three negative brushings, the probability that a patient has a carcinoma is less than 6%.¹⁶ Predictors of positive yield include older patient age, mass size greater than 1 cm, and stricture length greater than 1 cm^17,18 or the presence of a true stricture. The presence of stones correlates with benign cytology.¹⁹

Most false-negative results are caused by sampling error,^14,15 which can occur if the tumor does not invade biliary mucosa. In addition, tumors with sclerotic desmoplastic stroma do not exfoliate cells as readily as tumors without desmoplasia and therefore may not be detected with brushings. Poor visualization of the area by the endoscopist may negatively impact the sampling. Interpretation errors (17%) and technical errors (17%) are the second most frequent causes of false-negative results. Most interpretation errors result from underinterpretation of adenocarcinoma caused by the difficulty of distinguishing adenocarcinoma from reactive changes.²⁰ The converse is also true: Reactive changes can mimic adenocarcinoma and lead to false-positive findings. Degeneration of malignant cells is another source of false-negative results.

False positive findings commonly result from overinterpretation of reactive and degenerative changes²¹ or of adenomatous epithelium from villous tumors of the ampulla²² or the bile duct. Intraductal neoplasms and dysplasia within the biliary ductal system may also lead to false-positive results.

The traditional method of specimen preparation is to prepare smears from the brushing—both air-dried smears for on site interpretation and alcohol-fixed smears for Papanicolaou stains. The brush may also be collected in Cytolyt solution (Hologic, Marlborough, Mass.) and the sample prepared as a ThinPrep (Hologic). The latter technique shows greater sensitivity.¹⁹ In another method, the brush is vortexed and then the sheath and guide wire are cut into 5-cm pieces and vortexed separately. The specimen is prepared as a cytospin. This procedure significantly increases the cellularity.²⁹ Another published procedure entails processing the entire brush as an actual surgical pathology specimen.³⁰

Contaminants

Contaminants from the duodenum may occasionally be seen with ERCP-guided brushing cytology of the pancreatobiliary tree, and contaminants from other sites may be obtained with FNAB of either the biliary tract or pancreas (Box 35.1).

Mesothelial Cells.

Mesothelial cells are polygonal cells arranged in flat sheets that show round to oval nuclei and intercytoplasmic “windows” (Fig. 35.1). They may be mistaken for benign ductal cells or possibly squamous cells.

Hepatocytes.

Normal hepatocytes are large polygonal cells with dense, sharply defined cytoplasm and round, central or eccentric nuclei with prominent nucleoli (Fig. 35.2).

Gastrointestinal Epithelium.

Gastrointestinal epithelium is a frequent contaminant in EUS-FNAB samples.

Gastric Epithelium.

The surface mucus cells of the gastric epithelium are columnar cells arranged in large folded sheets, palisaded rows, or single cells, usually associated with background mucin. Typically, a luminal border may be seen along one edge of the cellular aggregates. The sheets are typically monolayered but occasionally may be folded or thick (Fig. 35.3). The gastric pits may appear in some groups of cells as rosettes present in the center of a cellular sheet. The mucus glands of the cardia or pylorus are indistinguishable from the foveolar surface mucus cells. The cells derived from the foveolae and mucus glands display mucinous cytoplasm, often contained in the upper third of the cytoplasmic compartment (Fig. 35.4), although the mucin may extend to the nucleus.

Gastric epithelium may also appear as stripped, bland, slightly elongated nuclei within a background of mucin (Fig. 35.5).^32,33 Chief and parietal cells may be observed if the needle has traversed the fundic or body region of the stomach (Fig. 35.6). Intact fragments may show attachment of surface epithelium to the lamina propria. Gastric epithelium is associated with abundant mucus that may contain degenerated cells if there is an inflammatory process.

Small-Intestinal Epithelium.

Small-intestinal epithelium has an architectural appearance similar to that of the surface mucus epithelium of the stomach, but the epithelial component is composed of absorptive enterocytes with interspersed goblet cells and is also associated with background mucin (Fig. 35.7, A). The brush border is visible when the cells are seen on edge (see Fig. 35.7, B). The epithelium may contain lymphocytes, which appear as darker, small cells within the epithelium.^32,33 Paneth cells are occasionally seen and are identified by the presence of coarse eosinophilic granules in the apical cytoplasm.

Both small-intestinal epithelium and gastric epithelium may produce abundant degenerated material that should not be confused with neoplastic cells. Both may also be associated with lamina propria that may resemble a stromal neoplasm.

Normal Biliary Tract

Normal bile duct epithelial cells are tall and columnar or cuboidal in appearance The cells are typically arranged in flat monolayered sheets (Fig. 35.8, A) or in a picket-fence arrangement with basally located nuclei (see Fig. 35.8, B). The nuclei are round to oval, and the cytoplasmic borders are distinct (Box 35.2).

Reactive Changes

Inflammation of the biliary system results from choledocholithiasis, sclerosing cholangitis, acute cholangitis, infections, calculi, stents, or instrumentation. The changes induced by inflammation include loss of surface structures, vacuolization of the cytoplasm, nuclear enlargement, coarse hyperchromasia, and multinucleolation³⁴ (Fig. 35.9). The epithelium remains monolayered and lacks pseudoacinar structures. In situations of inflammation or injury, the epithelial lining may develop mucinous metaplasia or squamous metaplasia (Box 35.3).

Dysplasia

Premalignant lesions of the bile ducts have historically been referred to as biliary dysplasia or atypical biliary epithelium. A new consensus classification of biliary intraepithelial neoplasia (BilIN) was published in 2007.³⁵ This proposal classified BilIN into three grades, similar to the scheme used for other organs such as the pancreas and prostate. The histopathologic criteria are similar to those for other intraepithelial lesions. The cytopathologic criteria have not been defined but can be assumed to be similar to what has been described as dysplasia in the biliary tract, with grade 1 lesions causing atypia on bile duct brushings previously referred to as low-grade dysplasia⁵ and grade 3 lesions causing atypia previously referred to as high-grade dysplasia.

Lesions with grade 2 dysplasia may fall into either category on brushings. Low-grade dysplasia is characterized by sheets and clusters of cells that show nuclear crowding and overlapping, smooth nuclear membranes, and a moderate nucleus–to-cytoplasm (N : C) ratio (Fig. 35.10, A). The chromatin is characteristically clear in appearance and granular, with mild clumping. Low-grade dysplastic cells may have one or two distinct nucleoli. More pronounced nuclear crowding and overlapping occurs in BilIN grade 3, or high-grade dysplasia. In these cases, nuclear membranes are irregular and the N : C ratio is significantly increased; chromatin is coarse; and nucleoli are distinct and prominent (Box 35.4; see Fig. 35.10, B).

Adenoma of the Biliary Tract and Ampulla

Adenomatous epithelium appears as slender, elongated, columnar cells arranged singly or in small sheets and clusters. The long, thin, basally oriented nuclei occupy approximately one third of the cell. The chromatin is typically fine and granular with one or more nucleoli (Fig. 35.11).³⁶ It may be difficult to separate these lesions from adenocarcinoma or dysplasia based on brushings (Box 35.5).

Intraductal Neoplasms of the Biliary Tract

Intraductal papillary mucinous neoplasms of the bile ducts (IPMN-B) correspond to lesions previously referred to as biliary papillomatosis or as cystic or mucinous lesions of the biliary tract,³⁷ similar to intraductal papillary mucinous neoplasms of the pancreas (IPMN-P).³⁸ The role of cytology in diagnosing these lesions is not established. What have been previously described as atypical papillary or mucinous cells indicative of well-differentiated adenocarcinoma are probably derived from these lesions.¹⁷ Cytologic features of papillary differentiation include groups with crowding and overlapping, papillary formations, and elongated nuclei. Features of mucinous differentiation are groups with mucinous cytoplasm (Fig. 35.12).

Adenocarcinoma

Cytology preparations of adenocarcinoma tend to be cellular and contain both cohesive and single cells. Groups of cells are usually arranged in sheets with nuclear overlapping and crowding and loss of polarity. A very useful finding is the presence of two-cell populations (Fig. 35.13). The nuclei lose their normal round shape and become rectangular, pointed, or angulated with convolutions and notches, particularly in high-grade carcinomas (Fig. 35.14, A). The N : C ratio is altered. The chromatin is typically coarse but may be pale in well-differentiated adenocarcinomas. Carcinomas with more abundant mucin in their cytoplasm show both nuclear and cellular enlargement and appear in an exaggerated honeycomb pattern. Pseudoacinar formations are features of adenocarcinoma (Box 35.6; see Fig. 35.14, A).

Mucinous and papillary differentiation have been described as subtle findings that need to be recognized because they may represent well-differentiated adenocarcinoma.¹⁷ The recommendation was to report any mucinous change as possibly representative of a mucinous neoplasm and any papillary change as suggestive of a papillary neoplasm. It now seems that these morphologic patterns are derived from IPMN-B if identified on a brushing (see earlier discussion). These findings identified on an FNAB of a solid mass are more likely to correlate with an invasive process.

A pitfall occurs with the papillary pattern because normal bile epithelium may form pseudopapillary clusters. Papillary clusters in normal duct cells are two dimensional. The cell clusters in true papillary lesions are crowded, and nuclei are usually arranged haphazardly. The nuclei of true papillary lesions are pointy and elongated, in contrast to their normal round to oval shape. Occasionally, the nuclei of benign epithelium may appear elongated, but again, the architectural arrangement is key. Carcinomas show a greater degree of architectural abnormalities and the presence of hypochromatic, transparent nuclei.

The key problem when assessing pancreatobiliary brushing cytology specimens is differentiating a reactive process from adenocarcinoma. A number of studies have focused on identifying sets of criteria that are most specific for adenocarcinoma. Initial studies assessing cytologic criteria on smears found architectural abnormalities, nuclear enlargement, “cell-in-cell” arrangement, loss of polarity, a bloody background, the presence of flat nuclei, nuclear molding, and chromatin clumping^39–41 to be indicative of the presence of adenocarcinoma, although a cytologist’s overall gestalt assessment appeared to be just as good.⁴¹ Another report described adenocarcinoma as showing architectural features of high-grade dysplasia but characterized by the presence of a greater number of single cells.³⁴

In ThinPrep-prepared specimens, features typical of carcinomas were three-dimensional micropapillae, anisonucleosis, high N : C ratio, nuclear contour irregularity, and prominent nucleoli. Cytomorphologic features that were not helpful in distinguishing positive and negative cases were single naked nuclei, chromatin granularity, and necrosis.⁴²

A prospective study of brush smears obtained during ERCP found tightly cohesive, small, three-dimensional cell clusters that formed cell balls as the most consistent feature of malignancy. The cells in the clusters displayed features of malignancy. The authors concluded that the presence of single cells, cytoplasmic vacuoles, and prominent nucleoli are not essential for a diagnosis of malignancy.⁴³

In summary, cytologic features that are indicative of malignancy include loss of normal honeycomb pattern, loss of polarity, nuclear enlargement, anisonucleosis, nuclear membrane irregularities, and chromatin abnormalities. Identification of a two-cell population is also a very helpful feature. Single cells are not always identified in malignancy, and macronucleoli and necrosis may be seen in inflammatory processes (Table 35.1).

One study that assessed a panel of 12 polymorphic microsatellite markers linked to six tumor suppressor genes and KRAS codon 12 mutation detection showed that this panel was effective.⁴⁴ A subsequent study assessing 20 bile duct brushing specimens for mutations in TP53, KRAS, BRAF, and EGFR found these mutations too infrequent in biliary tract adenocarcinomas to be of added diagnostic benefit.⁴⁵

Normal Pancreas and Contaminants

Table 35.2 summarizes the cytologic features of normal pancreatic epithelia.

Ductal Cells

The lining of the large pancreatic ducts is composed of columnar epithelium arranged in flat, honeycombed, two-dimensional sheets of cells with centrally located nuclei (Fig. 35.15). The ductal cells may also be present in a palisaded, “picket-fence” arrangement in which the nuclei are basally located. In contrast to the epithelium of the large ducts, intralobular duct epithelium is composed of cuboidal-shaped cells with scant basophilic cytoplasm and are usually present as flat sheets, small clusters, or tubular structures.

Acinar Cells

Aspirates of the normal pancreas consist predominantly of acinar-type epithelium, which is typically arranged either singly or in small acinar structures (Fig. 35.16). The cells are pyramidal or triangular in shape and contain abundant, granular cytoplasm with numerous intracytoplasmic zymogen granules. The nuclei are round, have a granular chromatin pattern, contain prominent nucleoli, and are either central or eccentrically located.

Islet Cells

Islet cells are only rarely detected in aspirates of the normal pancreas. More often, they are seen in aspirates from patients with chronic pancreatitis with islet cell hyperplasia. When present, they occur as loose aggregates of cells that contain wispy, ill-defined amphophilic cytoplasm and oval nuclei with a stippled chromatin pattern (Fig. 35.17).

Primary neoplasms of the pancreas manifesting as solid masses have predominantly two distinct morphologic patterns. The first is the typical histopathologic pattern of ductal neoplasms, which shows ductal-type groups with significant pleomorphism associated with desmoplastic stroma. Aspiration may be difficult because of the fibrous stroma, and sometimes the stroma results in samples that are scantly cellular. The differential diagnosis is usually with pancreatitis. The other pattern is that of the solid, cellular neoplasm. PanNET, ACC, SPN, and pancreatoblastoma (PB) fall in this category (see later discussion). Histopathology shows a neoplastic process growing as solid cell sheets, without a desmoplastic stroma. The cytology smears are diffusely cellular with numerous single cells and loose groups. These neoplasms may also be vascular.

Reactive Processes

Reactive processes of the pancreas that may be sampled by FNAB include acute and chronic pancreatitis and autoimmune pancreatitis. The role of FNAB is to rule out adenocarcinoma or other neoplasms. Occasionally, the FNAB may sample a reactive process adjacent to a neoplasm; clinical follow-up and further evaluation is always indicated for patients with clinical and imaging findings suspicious for adenocarcinoma or neoplasia (Box 35.8).

Acute Pancreatitis

Typical smears from patients with acute pancreatitis show dirty, necrotic background cells, cellular debris, and fat necrosis with saponification. Acute inflammation may be prominent. The normal pancreatic elements, when present, may show evidence of necrosis and degeneration (Fig. 35.18). Pancreatic ductal epithelium may show various degrees of reactive atypia as well (see Box 35.8).

Chronic Pancreatitis

Aspirates of chronic pancreatitis are not typically very cellular; they consist of varying amounts of inflammatory cells, ductal cells, acinar cells, calcifications, fibrous tissue fragments (Fig. 35.19), and debris.⁵⁵ Islet cells are sparse or absent. The inflammatory component usually consists of mononuclear cells and histiocytes. Acinar cells may be associated with inflammation and fibrosis. The ductal cells show mild atypia but remain monolayered and organized (see Box 35.8).

Reactive Atypia

Reactive atypia induced by pancreatitis or other processes shows a spectrum of changes (Fig. 35.20). When severe, the atypia may mimic adenocarcinoma. A key feature is that the atypical groups are fewer in number than the number of groups associated with ductal adenocarcinoma. Another feature is that there is usually a range of atypia in the ductal groups, from benign to malignant. The differentiation of benign and reactive processes from adenocarcinoma is discussed later (see Differential Diagnosis and Pitfalls).

Ductal Adenocarcinoma

Diagnosis of pancreatic ductal adenocarcinoma (PDA), not otherwise specified (NOS), based on FNAB samples is usually straightforward. Occasionally, it is difficult to distinguish between benign lesions and adenocarcinoma or between reactive lesions and adenocarcinoma.