CHAPTER 11 Pancreas, biliary tract and intra-abdominal organs
Clinical aspects
In the developed world, the incidence of pancreatic cancer has risen threefold since the 1920s. Prognosis is very poor, this tumor accounting for almost a quarter million deaths annually worldwide. In the USA, cancer of the pancreas is now the fourth leading cause of cancer deaths in male and females. The majority of patients have irresectable disease or metastases at the time of diagnosis. Therapeutic options are limited.1 A diagnostic test such as fine needle biopsy (FNB), which combines accuracy with minimal intervention, is of unchallenged value in these patients.
Ultrasonography (US) and computed tomography (CT) made mass lesions of the pancreas, biliary tree and elsewhere in the abdomen readily accessible to FNB. Teams in Sweden and Denmark were the first to use FNB, guided by angiography or US, to investigate pancreatic masses.2,3 Endoscopic evaluation of pancreatic and biliary disease has long been part of the work-up of patients with obstructive symptoms or mass lesions, with additional retrograde imaging studies and/or the collection of pancreatobiliary fluid or brushing specimens. However, the recent addition of a fine needle biopsy channel and linear ultrasonographic functions to the endoscope, have made this the method of choice in the early assessment of patients with pancreaticobiliary symptomatology. There are several good reviews,4–6 and many recent series attest to the value, efficacy and low complication rate of endoscopically performed, ultrasonographically directed (EUS) procedures.7–14 An alternative method of obtaining a tissue diagnosis is by transduodenal FNB at laparotomy, performed with curative intent, or for surgical bypass of obstructive jaundice.
As confidence in cytological diagnosis has become well established, there is an onus on the cytologist to provide specific information on tumor type as a basis for therapeutic decisions. This has led to use of supplementary techniques such as cell blocks, EM, immunocytochemistry, and analysis of pancreatic cyst fluid.4–6,15,16 Core needle and wedge biopsies are generally discouraged in the pancreas owing to the propensity for fistula formation, spill of lytic enzymes, fat necrosis of peripancreatic tissues, and tumor track spread (see Complications below).
The place of FNAC in the investigative sequence
Fine needle biopsy of a pancreatic mass is performed after imaging has defined site, dimensions, anatomic relationships, depth and solid or cystic nature. This information is important not only for the biopsy procedure and its safety but also for the subsequent interpretation of the smears. The biopsy should be performed by the radiologist and the pathologist in close cooperation. EUS procedures are done by the endoscopist, usually a gastroenterologist. Again, the presence of a cytologist in the endoscopy suite to effect rapid evaluation is highly advisable. Cytological examination may unexpectedly reveal a benign neoplasm, an islet cell tumor, a metastatic malignancy or a lymphoma instead of the presumed pancreatic adenocarcinoma, with its inherently poorer prognosis. If pancreatic cancer is diagnosed microscopically, staging will determine further management. Unresectable or metastatic lesions may require stent insertion for relief of obstructive jaundice. Stent placement may be performed after FNB, at the same endoscopic procedure. If deemed operable, cancer cases can undergo subsequent surgery with the diagnosis and staging already accomplished.17
FNB does not contribute significantly to early diagnosis of pancreatic cancer. Neither is it suited, generally speaking, to the diagnosis of pancreatitis. Inflammatory pancreatic masses, conversely, constitute the chief clinical differential diagnosis of pancreatic neoplasia and must be recognisable to the cytopathologist.18,19 In pancreatic pseudocyst, FNB can be both diagnostic and therapeutic.15,20,21 Decompression of an acutely developed cyst may relieve symptoms and facilitate surgical treatment. Some patients are subjected to laparotomy without prior FNB in anticipation of resectability or for biliary bypass palliation to alleviate symptomatic jaundice. Under these circumstances, the FNB can be undertaken easily and with high precision intraoperatively.22–24A FNB at operation is done through the wall of the opened duodenum for pancreatic head masses. It has the advantage over the traditional wedge biopsy for frozen section of being virtually free of complications such as hemorrhage and fistula formation. It can be repeated to sample many different parts of a large mass, thereby increasing the probability of obtaining representative material. A thin needle can pass through the stomach or duodenum without risk. Rapid on-site staining and interpretation of intraoperative smears is quicker than cutting and staining frozen sections. With experience, interpretation of technically satisfactory smears is often easier than that of frozen section because of better preservation of cell detail.
Mass lesions of the gastrointestinal tract are, as a rule, investigated by radiological imaging and by endoscopy. FNB can be done percutaneously or endoscopically, an approach well-suited to submucosal neoplasms, and to lymph nodes adjacent to the bowel.6,25,26 Endoscopically visible lesions in the upper or lower gastrointestinal tract are better sampled by exfoliative means, such as brushing, or by tissue biopsy. The cytopathology of exfoliative gastrointestinal and pancreatobiliary brushings is beyond the scope of this book.
Accuracy of diagnosis
Diagnostic accuracy is related primarily to the adequacy and representativeness of the biopsy, which in turn depends on the site, size and nature of the lesion, and on the expertise of the operator. The relative merits of available radiological techniques in guiding the biopsy are discussed in Chapter 3. The requirement for technical expertise is greatest for EUS-FNA, the success of which is almost entirely operator dependent. Although technical progress continues to improve the quality of radiological tumor imaging, there is still a minimum size of approximately 1.0 cm below which a lesion cannot be clearly demonstrated by traditional US or by CT, and precise needling is not possible. EUS appears to have the capability of detecting lesions of 0.3–1.0 cm. Intraoperatively, under the control of direct palpation, lesions measuring only a few millimeters in diameter can be biopsied successfully. Very large tumors present the difficulty of identifying viable areas from which to obtain well-preserved cells, in a background of extensive necrosis or hemorrhage.
Diagnostic specificity for malignant pancreatic lesions is 100% in nearly all published series. However, occasional false-positive diagnoses have been reported in cases of chronic pancreatitis or in the presence of pancreatic intraepithelial neoplasia (PanIN).19,27 Diagnostic sensitivity is more variable, particularly in series of US- or CT-directed percutaneous biopsies with reported sensitivity of between 50% and 90%.19,28–33 Intraoperative FNB achieves sensitivities in excess of 90%.22–2434 Recent series of pancreatic EUS-FNA report sensitivies of between 90% and 100%.9,11,12,14,35–39 Factors influencing the sensitivity include pre-analytical issues such as: the size of the lesion, whether the lesion is solid or cystic, the experience of the operator, the number of needle passes, and the availability of on-site assessment by a cytologist or cytopathologist,4 as well as analytical issues. It has been reported that repeat EUS-FNA in indeterminate cases further improves the sensitivity.39,40
Many of the cited studies focus on solid lesions of the pancreas and it is acknowledged that the sensitivity and specificity in diagnosis of cystic lesions is lower.4 Biochemical analysis of fluid from cystic lesions improves the sensitivity and specificity of these lesions over cytologic assessment alone (see Intraductal papillary – mucinous and mucinous cystic neoplasms below).15,41–44
Complications
Significant complications are rare if thin needles of 0.08 mm (21 gauge) or less are used. Such needles pass through stomach or bowel without causing peritonitis, although caution is recommended in cases of bowel obstruction or distension. In a comprehensive review of the literature, and a questionnaire study, Smith documented six deaths after pancreatic FNB.45 Of these, five were due to pancreatitis and one followed sepsis post aspiration of a pancreatic pseudocyst. In an earlier review of 184 procedures, Mueller and colleagues reported a 3% incidence of (nonfatal) severe pancreatitis.46 All of the cases developing pancreatitis in either series had inflammatory mass lesions mimicking carcinoma. Hence, FNB is not recommended for the investigation of classic clinical pancreatitis but should be reserved for cases with a radiologically localized lesion. Major hemorrhage is very rare. The coagulation status should be noted but a full investigation is unnecessary unless the spleen or the liver is involved. A recent study of 1034 pancreatic EUS-FNA identified 10 hemorrhages (0.96%), none of which was fatal, two cases of acute severe pancreatitis (0.19%) and one duodenal perforation (0.09%) which lead to a post-surgical death.47 Two cases of pancreatic ascites due to fistula formation have received literature attention.48,49 Both fistulas closed with regression of the ascites.
Tumor seeding in the needle track has been reported in a small number of cases following FNB of pancreatic carcinoma. In a literature review by Smith, the risk of needle track seeding was 4.45 per 100 000 transabdominal FNB procedures in four questionnaire studies of 156 652 patients.45 Five of 11 instances of needle track spread after FNB of the pancreas used a needle greater than 21 gauge. It is thus apparent that cutaneous needle track seeding is an extremely rare event. More frequently than is clinically realized, however, may be the tracking of cells to the peritoneal surfaces as a result of FNB. Some authors have reported a significant increase in incidence of exfoliated malignant cells within operative peritoneal washings after FNB,50 while others have failed to confirm this.51 More recently, it has been shown that EUS-FNA results in positive peritoneal cytology significantly less often than percutaneous FNB.52 The significance of this positive peritoneal cytology remains uncertain, with some authors finding that it was an indicator of unresectability, advanced disease, early metastasis and short survival,51 others identified a trend to decreased survival,53 while others did not show any survival disadvantage.54 The latter group concluded that cell spillage into the peritoneum at FNB did not render the procedure unsafe, particularly if neoadjuvant chemo-radiation was part of the management protocol. Clinicians wishing to minimise this potential hazard may prefer the transduodenal route of FNB offered by EUS, or by intraoperative sampling.
Cytological findings
The pancreas and biliary tract
Normal structures (Figs 11.1–11.3)
Fig. 11.3 Duodenal surface epithelium
Monolayered sheet of duodenal epithelial cells with interspersed goblet cells (Pap, LP).
Fine needle biopsy smears from normal pancreatic tissue can occasionally be surprisingly cellular. Most of the cells are acinar unless the needle traverses a major duct. Acinar cells have indistinct cell borders; their acinar arrangement suggests an individual triangular shape to the cells, the nuclei being disposed in a circle at the periphery of small, round clusters (Fig. 11.1). Larger fragments are composed of multiple acini held together by sparse fibrovascular stroma, often surrounding an intact duct. The nuclei are uniformly small, round and of similar size and shape to those of small lymphocytes. Many cells are represented by single, naked nuclei. The nuclear chromatin is densely granular and evenly distributed. There may be a single, prominent nucleolus or multiple small nucleoli. The cytoplasm is dense and granular. The ductal epithelial cells form monolayered sheets (Fig. 11.2). Cell borders are usually visible, the cytoplasm is pale and the nuclei are regularly spaced within the sheets. The nuclei are larger than those of the acinar epithelial cells and are more ovoid. They are paler and have finely granular chromatin and small, usually single, nucleoli. Palisading may be seen along the edge of sheets. Ductal epithelial cells from major bile ducts look similar and cannot be clearly distinguished from those of major pancreatic ducts in smears. Columnar cells of duodenal origin, however, although also in monolayers, are distinguished by the presence of goblet cells within the sheets (Fig. 11.3).
Pancreatitis (Fig. 11.4)
Acute diffuse hemorrhagic pancreatitis is unlikely to be subjected to FNB. Percutaneous FNB is not a suitable method to confirm clinically suspected chronic pancreatitis, partly because the inflammatory cells found in smears are often too sparse to be diagnostic, partly because of the risk of the biopsy causing exacerbation of pre-existing pancreatitis.18,19 However, it may be utilized if fibrosis with resultant nodularity leads to a radiological presentation which is suspicious of a neoplastic mass, and, if on surgical exploration, the gland is felt to be increased in size and/or consistency, whether focally or diffusely. Since duct obstruction, edema and inflammation peripheral to a carcinoma can simulate true pancreatitis, multiple biopsies should be taken both proximal to the mass and from different parts of the abnormal area to exclude malignancy. This can be done endoscopically or intraoperatively without significantly increasing the risk of local complications.
In severe chronic pancreatitis much of the exocrine parenchyma may be destroyed and be replaced by fibrous tissue. In such cases, remaining ductal, acinar and endocrine epithelium can show prominent reactive/regenerative atypia which can be difficult to distinguish from well-differentiated adenocarcinoma (Fig. 11.4). There may be crowding of nuclei and microglandular arrangement of cells with nuclear enlargement and variation in nuclear size and shape. The atypia is usually variable in degree between cells.
Infectious pancreatitis is uncommon in immunocompetent hosts, and occurs more frequently in circumstances of immune suppression. It may be caused by a variety of pathogens, including viruses, parasites, bacteria and fungi.55 Organisms gain access to the pancreas either by the hematogenous route in systemic infections, or from the intestinal tract through the pancreaticobiliary system.
Cysts
Cysts in the pancreas can be congenital or post-pancreatitic pseudocysts. Some tumors are inherently and primarily cystic; in addition, any neoplasm, benign or malignant, may undergo cystic degeneration. Evaluation of cysts by radiologic or endoscopic features alone can sometimes be problematic.56 Cytologic evaluation of aspirated cyst fluid is widely used and often fully diagnostic of the nature of the cyst, although sampling and interpretative errors are acknowledged.57,58 Sensitivity of cytodiagnosis of pancreatic cystic lesions was considerably lower than that of solid lesions (64% vs 98%) in one study.57 Biochemical analysis of fluid from cystic lesions improves the sensitivity and specificity over cytologic assessment alone (see Intraductal papillary mucinous and mucinous cystic neoplasms below).
Serous cystadenoma (microcystic/glycogen-rich adenoma) (Fig. 11.6)
These are uncommon benign lesions, often incidental findings in older individuals. Their presentation must be familiar to cytopathologists, as failure to identify the constituent bland monolayered sheets might result in an erroneous ‘nonrepresentative’ report. Histopathologically, these tumors demonstrate classic microcysts, lined by bland, uniform glycogen-rich mucin-negative cells (Fig. 11.6).59 Several reports have appeared in the literature on FNB cytodiagnosis.58,60,61 The aspirate is clear and watery, some are acellular, while others contain a sparse exfoliate of monomorphic round cells with finely vacuolated, glycogen-positive, mucin-negative cytoplasm, and centrally disposed nuclei containing fine chromatin. Occasionally, larger monolayered sheets may be obtained (Fig. 11.6). The cell groups may resemble mesothelial cells or bland acinar cells. Cell features are thus non-specific, but are assessed in conjunction with the watery nature of the aspirate and the very characteristic radiological picture of small microcysts within a well-demarcated round mass lesion arising anywhere in the pancreas. A PAS stain will highlight the glycogen-rich cytoplasm.
Biochemical analysis of cyst fluid generally shows a low viscosity and low CEA and amylase levels.62 A recent study which combined imaging, biochemical and cytological data in reaching a diagnosis concluded that the preoperative diagnosis of serous cystadenomas remains a challenge.63
Solid-psedopapillary neoplasm (Figs 11.7 and 11.8)
Criteria for diagnosis
This relatively uncommon low malignant potential pancreatic neoplasm, with its many pseudonyms, occurs almost exclusively in young women under the age of 20 years.64 They usually behave in a benign manner but 10–15% of cases show spread or metastases. Still considered to be of uncertain histogenesis, this tumor is thought to be hormonally dependent. Occurring anywhere along the length of the pancreas, the mass is usually sizeable at presentation, averaging 8 cm. Grossly, as well as on imaging studies, it is characterized by good circumscription, multiloculation and solid and cystic areas. The histopathologic features, which are specific and diagnostic, are mirrored exactly by the cytologic pattern in FNB smears. Initial tumor growth is solid, but degeneration results in clefts, eventually enlarging to cysts, between vascularized cell fragments, which then appear ‘pseudopapillary’.
Fifty isolated case reports or limited series attesting to the ability of FNB to confidently identify this unusual but widely reported neoplasm were summarized in 2002 by Pettinato and colleagues in a definitive paper.64 Richly cellular smears demonstrate characteristic pseudopapillae and pseudorosettes, composed of small tumor cells adherent to delicate metachromatic fibrovascular stalks. Recognition of this stromal component is essential in distinguishing this tumor of relatively good prognosis from similar small-celled pancreatic neoplasms. The outline of both the pseudopapillae and the pseudorosettes appears irregular, due to exuberant exfoliation of single cells into the smear background. These vascular-based aggregates vary from large grape-like branching structures to single capillaries with leaf-like single-cell linings (Fig. 11.7). These cells are small, round to oval, plasmacytoid or cuboidal and extremely monotonous. Cytoplasm is variably preserved, vesicular to faintly granular, containing characteristic eosinophilic, hyaline intracytoplasmic globules, which are PAS positive. Nuclei are round and even, with grooves but no major irregularities or significant pleomorphism. Chromatin is described as finely granular, without clumping or clearing. Small nucleoli may be appreciated and are occasionally multiple. In the background, foam cells, multinucleated giant cells, debris and laminated psammoma bodies reflect the cystic and papillary nature of the parent tumor. Mucus is absent. Diagnostic accuracy of EUS-FNA has been reported as 75% in a series of 28 cases.65
Immunocytochemistry is extremely variable but most neoplasms studied stain positively for vimentin, α1-antitrypsin, α1-antichymotrypsin and progesterone receptor (Fig. 11.7C). Unlike PET, they are typically negative for chromogranin, cytokeratin, EMA and specific pancreatic hormones. Both tumors may express CD10, synaptophysin and CD56.64,66–69 More recently, nuclear expression of beta-catenin together with loss of normal membrane localization of E-cadherin have been proposed as useful in separating the two entities.66,67,69
