Utilization of ancillary studies in the cytologic diagnosis of biliary and pancreatic lesions: The Papanicolaou Society of Cytopathology Guidelines

Abstract

The Papanicolaou Society of Cytopathology has developed a set of guidelines for pancreatobiliary cytology including indications for endoscopic ultrasound guided fine needle aspiration, terminology and nomenclature of pancreatobiliary disease, ancillary testing and post-biopsy management. All documents are based on the expertise of the authors, a review of the literature, discussion of the draft document at several national and international meetings and synthesis of selected online comments of the draft document. This document presents the results of these discussions regarding the use of ancillary testing in the cytologic diagnosis of biliary and pancreatic lesions. Currently, fluorescence in-situ hybridization (FISH) appears to be the most clinically relevant ancillary technique for cytology of bile duct strictures. The addition of FISH analysis to routine cytologic evaluation appears to yield the highest sensitivity without loss in specificity. Loss of immunohistochemical staining for the protein product of the SMAD4 gene and positive staining for mesothelin support a diagnosis of ductal adenocarcinoma. Immunohistochemical markers for endocrine and exocrine differentiation are sufficient for a diagnosis of endocrine and acinar tumors. Nuclear staining for beta-catenin supports a diagnosis of solid-pseudopapillary neoplasm. Cyst fluid analysis for amylase and carcinoembryonic antigen aids in the pre-operative classification of pancreatic cysts. A number of gene mutations (KRAS, GNAS, von Hippel-Lindau, RNF43 and CTNNB1) may be of aid in the diagnosis of cystic neoplasms. Other ancillary techniques do not appear to improve diagnostic sensitivity sufficiently to justify their increased costs.

INTRODUCTION

Ancillary testing of pancreatobiliary cytology specimens is becoming increasingly important as we learn more about the molecular mutations associated with pancreatobilary disease and develop clinically useful tests that aid in the diagnosis and management of patients. Management algorithms for pancreatic neoplasia are evolving; some are shifting from surgical intervention to conservative surveillance, while others are recommending pre-operative chemoradiation prior to surgery to improve the yield of R0 resections. As an invaluable member of the patient care team, the pathologist is at the forefront of diagnosis, both pre- and post-operatively. Ancillary tests are invaluable tools that assist in refining our diagnoses and include a wide range of biochemical and molecular tests with variable diagnostic and prognostic utility.

These proposed guidelines on ancillary testing of pancreatobiliary cytology specimens stems from the expertise of the authors, review of the literature, discussions with pathologists at several national and international meetings over an 18 month period and synthesis of selected online comments of the draft document on the Papanicolaou Society of Cytopathology web site [www.papsociety.org].

ANCILLARY TESTING OF PANCREATOBILIARY STRICTURES

Cytological evaluation of bile duct brushings is known to have suboptimal sensitivity ranging from 40% to 61%, respectively.[1,2,3,4,5,6,7,8,9,10,11] A number of approaches have been developed to augment the purely cytologic analysis of bile duct brushing specimens. These have included protocols to improve diagnostic sensitivity of brushing specimens designated as inconclusive, indeterminate or negative following on-site cytopathologic examination.[3,12,13,14] Ancillary procedures potentially useful in interpretation of negative and indeterminate cytological results include intraductal ultrasound examination, digital image analysis (DIA), immuno-labeling, fluorescence in-situ hybridization (FISH), genetic analysis for neoplasia-specific mutations and sequential mutational analysis.[1,2,3,4,5,6,7,8,9,10,11,12,13,15]   Table 1 demonstrates markers which have been used in ancillary testing of brushing specimens.

Table 1.

Proposed ancillary testing for pancreatobiliary trictures

DIA of samples from duct strictures

At the level of the biosample, DIA has been used to identify abnormalities of nuclear deoxyribonucleic acid (DNA) content using spectrophotometric techniques.[16,17,18] DIA utilizes Feulgen reactions which hydrolyze DNA into constituent nucleic acids that stoichiometrically bind to the Feulgen dye. Using this technique, DNA ploidy can be assessed by a variety of commercially available image analyzers.[16,19] DNA ploidy status is assessed on the collected cells using a histogram generated by commercially available quantitative DNA analysis programs. Results are characterized as diploid, aneuploid or tetraploid. Aneuploid and tetraploid results are supportive of malignancy.[8]

The sensitivity of DIA does not appear to improve diagnostic accuracy beyond that achievable with routine cytology for patients with primary sclerosing cholangitis. However, in patients without primary sclerosing cholangitis, the technique does appear to improve diagnostic sensitivity. DIA has been reported to have excellent specificity for the diagnosis of carcinoma but only moderate sensitivity.[17,18] Thus, DIA has diagnostic characteristics similar to routine cytology in that a positive test is highly accurate but a negative result is of little clinical value.

Immunocytochemistry

A number of immunocytochemical markers have shown promise distinguishing benign from malignant biliary epithelium. These include S100P, von Hippel-Lindau gene product (pVHL), IMP3, and CD10. Tretiakova et al.[20] have demonstrated loss of CD10 immunopositivity in most examples of high-grade dysplasia and all cases of invasive carcinoma. CD10 positivity was present in the overwhelming majority of benign lesions. In a separate study, Levy et al.[21] have demonstrated strong staining for S100P in 90% of adenocarcinomas with 83% of adenocarcinomas showing diffuse staining for S100P. IMP3 staining was present in 77.5% of adenocarcinomas. Total loss of pVHL staining was seen in 37 of 40 adenocarcinomas. Nearly 70% of adenocarcinomas display the staining pattern S100P+/IMP3+/pVHL−.[21] Benign biopsies were invariably negative for S100P, but 94% were pVHL positive.[21] As discussed in the section on solid pancreatic lesions, adenocarcinomas frequently stain for MUC4 and mesothelin (MSLN) but demonstrate an absence of staining for clusterin-beta.[22]

MOLECULAR ANALYSIS OF DUCT STRICTURE BRUSHING SPECIMENS

KRAS mutational analysis

KRAS gene mutation analysis has been used as an ancillary testing procedure for analysis of pancreatic juice, bile duct brushings and fine-needle aspiration (FNA) of solid and cystic pancreatic masses.[23,24,25] The majority of studies involving KRAS mutations have investigated their relationship with pancreatic adenocarcinoma[26,27,28,29] and genetic progression in pancreatic duct lesions and intraductal papillary mucinous neoplasms (IPMN).[30,31,32] KRAS mutational analysis of pancreatic juice has even been used as a potential method for the early diagnosis of pancreatic carcinoma.[33,34,35] Although KRAS mutations can be found in pre-invasive dysplastic lesions and invasive carcinomas of the pancreas,[23,24,25,26,27,28,29,30] several studies have shown that KRAS mutation analysis is a sensitive test for pancreatic adenocarcinoma.[36,37,38] Sturm et al.[24] studied 312 consecutive patients with extrahepatic biliary stenosis and found that conventional cytology combined with KRAS mutational analysis was more sensitive than conventional cytology alone. Kipp et al.[39] studied 35 brushing cytology samples collected during endoscopic retrograde cholangiopancreatography and demonstrated a combined sensitivity of 86% for KRAS mutation and FISH analyses in the diagnosis of pancreatic adenocarcinoma. Hruban et al.[31] have described a progression model for pancreatic carcinoma in which both KRAS and telomere abnormalities are seen in low grade dysplasias as well as malignancies. KRAS mutations are, therefore, not specific for invasive cancer, and have been described in chronic pancreatitis.[27,28] Reicher et al.[23] have demonstrated some utility for KRAS mutational analysis in the evaluation of endoscopic ultrasound (EUS)-FNA specimens. They described KRAS mutations in approximately 9% of benign specimens and 56% of malignant specimens. Importantly, KRAS mutational analysis was helpful in dividing atypical specimens into benign and malignant categories. Neither of the two cases in Reicher's study considered cytologically atypical but on follow-up demonstrated to be benign, contained KRAS mutations.[23] On the other hand, in cases considered atypical with malignant follow-up, 20% of cases (2/10) demonstrated KRAS mutations. Additional studies will be necessary to determine the value of KRAS mutational analysis in both bile duct stricture cytology samples and FNA of solid pancreatic masses, but current data do not support KRAS testing of solid pancreatic masses and bile duct strictures as a useful ancillary test for diagnosis. The utility of KRAS mutational analyses in the evaluation of cystic lesions of the pancreas is discussed below.

The potential application of analytic techniques for evaluating sequential mutation accumulation in bile duct brushing specimens has been studied. Lapkus et al.[15] have shown that there is considerable overlap in the spectrum of mutational markers in pancreatic duct and biliary brushings, but the temporal profile of accumulation of these mutations differs significantly between pancreatic and biliary neoplasms. These authors studied the time course of mutation accumulation in pancreatic and biliary tract lesions by microdissecting cell clusters on the basis of cytomorphologic features and analyzed the cells for the loss of heterozygosity with a panel of 15 markers (1p, 3p, 5q, 9p, 10q, 17p, 17q, 21q, 22q) as well as point mutations in KRAS using polymerase chain reaction (PCR)/capillary electrophoresis. The prevalence of loss of heterozygsity (LOH) and KRAS mutations in these lesions varied.[15] While distinctive prevalences for sequential mutation accumulation were demonstrable, diagnostic utility of this approach is still to be determined.

FISH

FISH analysis of bile duct brushing specimens for polysomy using a commercially available DNA probe set (UroVysion; Abbot Molecular, USA) has been reported as a useful technique by a number of authors.[3,8,11,12,40,41,42] This commercial kit utilizes probes targeting the pericentromeric regions of chromosomes 3 (CEP3), 7 (CEP7), and 17 (CEP17) as well as chromosomal bands 9p21 (LSI 9p21). The method can be automated using the Bioview Imaging Duet system (Bioview, Ltd., Nes Zionu, Israel). Others[8] have developed probe sets of their own based on known chromosomal alterations in genes associated with pancreatic carcinoma including: TP53, CDKN2A/p16 and epidermal growth factor receptor. In a series of 93 pancreaticobiliary brushings, Barr Fritcher et al.[41] demonstrated a specificity of 100% and a sensitivity of approximately 60% for the identification of carcinoma using FISH probes targeting centromeric regions of chromosomes 3, 7, 17 and the 9p21 band. They considered a specimen as positive for malignancy by FISH when five or more cells showed polysomy (>2 signals in at least 2 of the 4 probes). In that study, FISH analysis outperformed routine cytology and review consensus cytology of the brushing specimens. Boldorini et al.[40] also reported similar findings with FISH of brushing specimens outperforming routine cytology. In that study, the sensitivity of FISH was 90% with 94% specificity while the positive predictive value was 98% and negative predictive value was 75%. Levy et al.[12] have reported similar success with FISH. In that study, the authors considered trisomy for chromosome 7 to be benign. In a series of morphologically negative cytology samples, FISH was able to suggest malignancy in 62% of cases.[12] Barr Fritcher et al.[41] investigated the utility of ancillary studies including DIA and FISH in a series of 498 consecutive patients with pancreaticobiliary strictures. They found that FISH had a sensitivity of 43% and was significantly better than the sensitivity of routine cytology (20%) when equivocal cytology samples were considered as negative. They concluded that FISH had a higher sensitivity than cytology without compromising specificity.[41]

Of all the ancillary techniques currently available for analysis of cytology specimens obtained by brushings from pancreaticobiliary strictures, FISH appears to improve diagnostic sensitivity the most over that achievable by routine cytology.[3,12,14,23,40,41,42] While not directly addressing bile duct brushings, Kubiliun et al.[13] recommended the use of FISH for the diagnosis of pancreatic carcinoma in inconclusive cytologic evaluations. It appears that a similar approach is successful for the evaluation of negative and inconclusive pancreaticobiliary tract brushing specimens.

Additional markers of malignancy in bile duct brushing are under investigation. While experience is limited, ribonucleic acid (RNA)-binding protein-3 (IMP3) shows promise as a marker for adenocarcinoma in bile duct brushings. In one study, IMP3 immunohistochemical expression demonstrated a sensitivity of 64% with a specificity of 100% for adenocarcinoma.[43]

ANCILLARY TESTING FOR PANCREATIC CYSTIC LESIONS

A number of ancillary tests have been proposed to aid in the diagnosis of cystic lesions of the pancreas. These vary from measuring carcinoembryonic antigen (CEA) and amylase levels in cyst fluid, to histochemical stains for mucin on cytologic smears, to mutational analysis (KRAS, GNAS, TP53, VHL, CTNNB1 and RNF43) and measures of polysomy.[32,37,44,45] Table 2 lists markers, which have been used in the diagnosis of pancreatic cystic lesions.

Table 2.

Proposed ancillary testing for cystic pancreatic lesions

Gross cyst fluid evaluation

A gross description such as “thick, white, viscous, sticky fluid” and cyst fluid that is difficult to pull into the needle and express from the needle clearly indicates a mucinous cyst fluid. These descriptions act as a surrogate marker for viscosity, a test that is not readily available in the biopsy suite. Leung et al.[45] examined the role of the “string sign” as a marker of viscosity. By placing the fluid between the thumb and index finger and gently pulling the fingers apart, the fluid would “string” to 3.5 mm if mucinous. Ancillary testing adds little to this simple visual test. Similarly, thin, non-mucoid, serosanguinous or frankly bloody cyst fluid is typical of serous cystadenoma due to the high vascularity of the septae in these cysts, which causes them to bleed internally and during aspiration.

Histochemical stains for mucin

Staining of direct smear preparations for mucin by a variety of techniques including the mucicarmine stain for neutral mucin and alcian blue pH 2.5 for acid mucin can be performed on direct smear preparations, but it is not diagnostically valid when performed on liquid-based preparations. Staining of smear preparations for mucin aids in the establishment of mucinous differentiation of the lining epithelium; however, it cannot separate benign from malignant lesions, and standardization of how much staining constitutes a positive result in EUS-FNA specimens with contamination from the gastrointestinal tract has not been determined. Gastrointestinal contamination of mucin can lead to a false positive interpretation. Thick colloid-like mucin with cellular debris floating in the mucin is consistent with mucinous cyst contents and not gastrointestinal contamination.[47] Documentation of mucinous differentiation in cystic pancreatic lesions is clinically helpful in that it points to either a mucinous cystic neoplasm or an IPMN.

Biochemical tests of cyst fluid

Cyst fluid amylase

Amylase testing quantifies α-amylase using an enzymatic colorimetric assay to measure the formation of degradation products saccharogenically or kinetically with the aid of enzyme-catalyzed subsequent reactions. The color intensity of the degradation product formed is directly proportional to the α-amylase activity, which is determined by measuring the increase in absorbance. Cyst fluid amylase is elevated in pseudocysts (typically in the thousands) and in cysts that communicate with the pancreatic ductal system, such as IPMNs, but not in cystic lesions that do not, such as serous cystadenoma and cystic neuroendocrine tumors.[47] Although mucinous cystic neoplasms (MCNs) do not communicate with the pancreatic ductal system, amylase levels can be quite elevated in MCNs making the specificity of this test for distinguishing IPMN and MCN of little value. High amylase combined with low CEA is consistent with a pseudocyst, low amylase and low CEA is typical of a serous cystadenoma and high amylase and high CEA is consistent with cystic-mucinous neoplasm with rare exceptions. A pancreatic cyst fluid amylase below 250 U/L is associated with a low risk for a pseudocyst.[48]

Cyst fluid CEA

Cyst fluid CEA level is a reliable indicator of mucinous differentiation in a cyst, but unfortunately it does not reliably predict the presence or absence of malignancy.[49] Suggested cut-off values for distinguishing between non-neoplastic and neoplastic cysts vary, with early reports suggesting 192 ng/mL as a useful cut point.[49] More recent data suggests a cut point of 110 ng/mL. Increasing the cut-off point for support of a mucinous cyst increases specificity but at the expense of sensitivity. Early studies suggest that CEA level was helpful in separating benign neoplastic cysts from malignant neoplastic cysts,[44] but subsequent studies have made it clear that an elevated CEA is not a reliable test for malignancy.[48,49] In addition, a low CEA level does not exclude a mucinous cyst in general, nor malignant cyst in particular. However, in a single study, mean CEA levels of cyst fluid demonstrated a striking difference between benign and malignant cystic lesions.[44] CEA greater than 693 ng/mL predicted malignancy with a sensitivity of 80% and a specificity of 90%.[48]

CA 125 in pancreatic cyst fluid

A few studies have investigated cyst fluid CA 125 levels to assess the usefulness of this marker in determining the cyst type and discriminating between benign and malignant neoplasms.[50,51,52,53] These studies have demonstrated that while the CA 125 levels are generally low in a pseudocyst and high in a cystic neoplasm, significant overlap exists between serous cystadenoma and mucinous cystic neoplasm. Furthermore, low levels of CA 125 have been reported in cystic neuroendocrine tumors,[54] and high levels in ciliated enteric duplication cyst[55] and lymphoepithelial cyst.[56] For these reasons, cyst fluid CA 125 assay does not have an important role in the evaluation of pancreatic cysts.

MUC protein expression patterns in pancreatic cyst fluid

Mucin expression patterns demonstrate a correlation with the presence of dysplasia and carcinoma. Several studies have demonstrated that analysis of pancreatic cyst fluid or serum can predict the presence of dysplasia or malignancy in the surrounding pancreatic cyst lining.[22,57,58] Jhala et al.[22] showed that 91% of pancreatic ductal adenocarcinomas expressed MUC4 but no evidence of staining was found in reactive ductal epithelium. In a study of 90 patients, Carrara et al.[57] investigated the expression of MUC1, MUC2, MUC3, MUC4 MUC5A, MUC5B, MUC6 and MUC7 and found that MUC7 expression was a strong marker for adenocarcinoma and borderline for IPMN. However, MUC7 was expressed in 37% of cases of chronic pancreatitis.[57] Maker, et al.[58] showed that high-risk IPMNs demonstrated elevated concentrations of MUC2 and MUC4 in pancreatic cyst fluid. From these data, it appears that evaluation of cyst fluid for MUC2, MUC4 and MUC7 may be helpful in the recognition of dysplasia and malignancy in cystic neoplasms of the pancreas.

Immunohistochemical analysis of pancreatic cyst samples

Most cyst fluids are usually very scant and of insufficient volume and cellularity for cellblock preparation. Exceptions to this are secondarily cystic neoplasms such as pancreatic neuroendocrine tumors (PanNET) and solid-pseudopapillary neoplasms (SPNs) where ancillary testing is often vital to a specific diagnosis.[62] See discussion of immunohistochemical testing of solid pancreatic lesions.

DNA analysis of cyst fluid

DNA analysis may also aid in the separation of non-neoplastic and neoplastic, and benign from malignant neoplastic cystic lesions. The success of DNA analyses depends on the amount of recoverable DNA.[9] Measurements of cyst fluid DNA can be quantitated using a variety of commercially available techniques.[60] The concentration of DNA is correlated with optical density (OD) as measured at a wavelength 260/280. The mean concentration of DNA present within a fluid from a pancreatic cystic lesion documented by OD ranges from a low of 6.5 in benign cysts to 16.5 in malignant cysts.[60]

Molecular testing of cyst fluid and tissue

Molecular analysis for KRAS mutations and for LOH has been reported to be helpful in the separation of benign from malignant cysts.[36,37,49] Khalid et al.[37] studied cyst fluid aspirates from 36 pancreatic cysts with histologically confirmed pathology. KRAS gene mutations were not observed in the benign cysts (a mix of pseudocysts and benign cystic neoplasms such as serous cystadenoma), while 40% of the cysts in the “premalignant” group (neoplastic cysts with low or intermediate-grade dysplasia) had a KRAS mutation, as did the majority of “malignant” cysts (neoplastic cysts with high-grade dysplasia or an associated invasive carcinoma). Similar percentages were reported in an analysis of a larger number of pancreatic cysts in the “pancreatic cyst DNA analysis study.”[60] Of interest, in this latter study, all malignant cysts (those with either high-grade dysplasia or an associated invasive carcinoma) with negative cytologic evaluation could be diagnosed as malignant using DNA analyses.[60] Others[32] have had less success in using mutational analysis for the distinction of benign from malignant cystic lesions. Chadwick et al.[32] have demonstrated that while KRAS point mutations were more common in malignant lesions than in benign lesions, they also could be found in both benign and malignant intraductal papillary mucinous tumors. Others have had a similar experience with solid pancreatic neoplasms.[26,27]

Shen et al.[61] studied the utilization of a commercially available test that combines the detection of KRAS mutation, LOH and DNA quantity/quality in the diagnosis of pancreatic cystic lesions. The concordance between the clinical consensus diagnosis and the commercial test was high with the commercial test showing a sensitivity of 83% and specificity of 100% for a malignant cyst and a sensitivity of 86% and specificity of 93% for a benign mucinous cyst. The authors concluded that the molecular analysis of pancreatic cyst fluids adds diagnostic value to the pre-operative diagnosis.[61]

From the available data, it appears that the analysis of pancreatic cyst fluid for mucin by histochemical stains on direct smears and CEA level are helpful diagnostic adjuncts for the recognition of a cystic lesion showing mucinous differentiation. Cyst fluid amylase levels appear to be of great value for the recognition of pseudocysts. KRAS and other mutational analyses may aid in the characterization of pancreatic mucinous cysts.[39,44,49]

Recent whole exome sequencing of the four most common cystic neoplasms of the pancreas (serous cystadenoma, SPN, mucinous cystic neoplasm and IPMN) has identified a specific mutational profile in each cyst type. VHL mutations are seen in serous cystic neoplasms, CTNNB1 (beta-catenin) in SPNs, RNF43, KRAS TP53 and SMAD4 in MCN, and KRAS, RNF43, GNAS TP53 and SMAD4 in IPMN.[62,63] It has therefore been suggested that mutational analysis for GNAS, KRAS, VHL, CTNNB1, RNF43, TP53 and SMAD4 may aid in the differential diagnosis of cystic lesions of the pancreas.

Mutations in the GNAS gene appear to be an important marker for IPMNs. The GNAS gene encodes for stimulatory G-protein alpha subunit which is a component of many transduction pathways.[64] GNAS gene protein product is defective in McCune-Albright Syndrome and some forms of fibrous dysplasia.[64,65] GNAS mutations appear to be specific for IPMNs,[66,67] while KRAS and RNF43 mutations can also be seen in MCNs.

The VHL is a tumor suppressor gene linked to sporadic hemangioblastomas and clear cell renal cell carcinomas.[68] This gene is somatically mutated in serous cyst adenomas.[62,69] Mutations in VHL gene are not seen in other cystic lesions of the pancreas.

RNF43 (ring finger protein 43) product is a HAP95 binding ubiquitin ligase that promotes cell growth.[70] RNF43 has recently been linked to the beta-catenin pathway. RNF43 mutations have been shown to occur in MCN along with IPMN.[71] Unlike IPMN, MCN do not typically harbor GNAS mutations.[71]

Virtually all SPNs harbor a CTNNB1 gene mutation, and these mutations, in the absence of other mutations, appear to be relatively specific for SPNs in the pancreas. Because the beta-catenin protein abnormally accumulates in the nuclei of cells with CTNNB1 gene mutations, immunolabeling for the beta-catenin protein is also a useful aid in establishing the diagnosis on limited cytology samples.[72]

MicroRNA analysis

The expression of selected microRNA (miRNA) is disregulated in cystic neoplasms of the pancreas, and Matthaei et al.[73] and others[74,75] have shown that the patterns of miRNA expression can help classify cyst type and in some instances even suggest the degree of dysplasia. Some groups have used panels of miRNA, while others have used a more focused approach using selected miRNA such as miRNA 21 and miRNA 221.[74,75]

ANCILLARY TESTING FOR SOLID PANCREATIC NEOPLASMS

A number of ancillary procedures have been investigated for the detection of malignancy in solid pancreatic masses. Table 3 lists markers that have been used in the diagnosis of solid pancreatic lesions.

Table 3.

Proposed ancillary tests for solid pancreatic lesions

Immunohistochemical testing

Immunohistochemical markers for adenocarcinoma

SMAD4 loss is observed in pancreatic intraepithelial neoplasia (PanIN)-3 high grade lesions and over half of pancreatic ductal adenocarcinomas.[76,77] SMAD4 loss also appears to be associated with a poor prognosis for pancreatic ductal adenocarcinoma.[78] Immunolabeling for SMAD4, the protein product of the SMAD4 gene, is a good surrogate for SMAD4 genetic alterations and is useful in cytologic material processed as cellblocks allowing for easy testing of atypical to suspicious cellular groups. Loss of SMAD4 staining supports a malignant diagnosis.

MSLN is a 40-kDa glycosylphosphatidylinositol-linked protein that has been demonstrated to be highly overexpressed in pancreatic adenocarcinoma[72] a finding corroborated in a subsequent study utilizing three platforms.[79] Differential expression of MSLN in pancreatic adenocarcinoma has also been documented in resected samples.[22] Three independent studies have shown that when applied to FNA material, immunolabeling for MSLN can serve as a useful marker for supporting the diagnosis of pancreatic adenocarcinoma.[22,80]

Immunohistochemical testing for PanNET

A variety of immunohistochemical markers are useful in the diagnosis and grading of PanNET. These include general markers of endocrine differentiation (chromogranin, synaptophysin, CD57, CD56 and neuron-specific enolase), cytokeratins CK8 and CK18 and CA 19-9.[81] The proliferation marker Ki-67 is of importance in the histologic classification of PanNET into low and high-grade tumors, but its utility in cytologic preparations including cell block material remains to be determined. Functional pancreatic endocrine neoplasms produce a number of hormones, which can be demonstrated by immunohistochemical techniques including insulin, glucagon, somatostatin, gastrin, vasoactive intestinal protein and pancreatic polypeptide.[82,83,84,85,86]

The separation of PanNET into high and low grade neoplasms based on mitotic count is proposed by the World Health Organization.[87] Some authors have shown that Ki-67 score appears to be a superior predictor of outcome than degree of tumor differentiation.[87,88] Application of Ki-67 staining to cell block material may have value in stratifying tumors into low and high grade forms.[89]

Immunohistochemical testing for acinar cell carcinoma

Immunohistochemical staining for enzymes produced by acinar cells aids in the distinction of these neoplasms from ductal carcinoma and PanNET. Immunolabeling for amylase, trypsin, chymotrypsin and lipase is most helpful in recognition of these neoplasms in cellblock material.[90,91] Expression of CK7 appears to aid in separation of acinar cell neoplasms (positive) from normal acinar cells (negative).[92] Abnormalities of beta-catenin can also be seen focally in acinar cell carcinomas.[93]

Immunohistochemical testing for pancreatoblastoma

Pancreatoblastoma displays diverse directions of differentiation as demonstrated by multiple patterns of antigen expression. The various components of differentiation-ductal, acinar and endocrine-will label with markers for these lines of differentiation as described above. The most common line of differentiation is along acinar cell differentiation.[98,99]

Immunohistochemical testing for SPN

SPNs express some markers seen in PanNET and acinar cell neoplasms including alpha-1-antitrypsin, neuron-specific enolase and CD56.[72,94,95] CD10 is also expressed in SPNs.[95] Beta-catenin abnormalities are also characteristic of SPNs and diffuse nuclear positivity for these markers usually suffices for the confirmation of the diagnosis.[96]

Immunohistochemical testing in pancreatic and biliary tract lymphoma

Both B-cell and T-cell lymphomas arise within the pancreas and biliary tract. The majority of these neoplasms will react with antibodies directed against common leukocyte antigen.

Immunohistochemical testing via either flow cytometry analysis or antibody panels performed on cellblock specimens is necessary for definitive classification of these neoplasms.[97] In depth discussion of these panels is beyond the scope of this document.

Molecular analysis

Molecular testing for pancreatic ductal adenocarcinoma

Mutational analysis

KRAS and p16 point mutation analysis, FISH and DNA ploidy analysis have been used as potential methods for separating benign and malignant masses by FNA.[3,26,27,28,49,61,70,71,100] Invasive pancreatic adenocarcinoma is believed to develop from two non-invasive precursor lesions, IPMN and PanIN by the accumulation of a series of mutations.[31] In this progression model for pancreatic carcinoma, activating point mutations in the KRAS oncogene and telomere shortening appear to be early events.[70,71,72,73,74,75,76,77,100,102,103] The prevalence of KRAS mutations in IPMN and PanIN increases along with increasing degrees of dysplasia.[100] Additional mutations affecting other genes occur later in the course of progression with increasing dysplasia, with p16/CDNK2A loss occurring in intermediate to high-grade lesions, and TP53 and SMAD4 inactivation occurring in high-grade precursors and in invasive cancers.[35,77,102,104] KRAS gene mutations can be found in 95% of invasive ductal adenocarcinomas of the pancreas, and p16/CDKN2A is inactivated in 90%, TP53 in 75%, and SMAD4 in 55%.[32]

A retrospective study by Khalid et al. demonstrated that testing of EUS-FNA material for LOH and KRAS mutations is useful in differentiating autoimmune pancreatitis from pancreatic cancer.[105] KRAS mutations were detected in 10 of 11 (91%) pancreatic cancer cases that yielded DNA amplification and in none of the autoimmune pancreatitis cases suggesting that KRAS mutation in a pancreatic mass FNA is associated with malignancy and may aid in the distinction from benign processes such as autoimmune pancreatitis.[105] However, KRAS mutations have been shown to occur in ductal hyperplasias, ductal metaplasias and chronic pancreatitis.[26,27,28,100]

FISH

FISH performed for evaluation of solid lesions has been shown to significantly improve diagnostic sensitivity without loss of specificity for ductal adenocarcinoma.[13,40] The probes utilized are directed against chromosomes 3, 7 and 17 as well as 9p21. Loss of 9p21 or alterations in copy number for 3, 7 and 17 are diagnostically important. The technique can be applied using commercially available probe sets. Disadvantages include cost, and that the method is time consuming and requires a fluorescence microscope. FISH analysis when combined with routine cytology has a sensitivity of approximately 85%.[13] Because routine cytology has an excellent specificity, FISH is most useful in improving diagnostic accuracy for cases reported as cytologically negative or inconclusive. Kubiliun et al.[13] have recommended the use of FISH for the evaluation of cases assessed on site as inconclusive or negative.

miRNA

Additional techniques for determining malignancy in pancreaticobiliary lesions include miRNA analysis. miRNAS are short 20-25 nucleotide RNA molecules containing regions complementary to various messenger RNAS (mRNA). Specific binding of a miRNA to a mRNA blocks translation. miRNA can be robustly quantified by real-time PCR and localized by in-situ hybridization. Steele et al.[106] have reviewed profiling studies of pancreatic adenocarcinoma. Some miRNA, like miR-21 and miR-155, are highly expressed in early lesions.[107,108,109] Most studies have analyzed resection specimens, one study analyzed FNA specimens,[110] but none have analyzed brushings. The clinical utility remains to be more firmly established, especially in comparison and conjunction with other techniques.

Molecular testing for PanNETs

Loss of heterozygosity

Nodit et al.[111] explored the utility of analysis of microsatellite loss in the diagnosis of PanNET. They concluded that microsatellite loss analysis of EUS-FNA material obtained from PanNET can be performed reliably and that losses of chromosome arms 3p, 6pq and 10pq along with gains of 5q, 12q, 18q and 20q were associated with malignant behavior.[111,112,113,114] LOH studies appear to have some technical limitations when using cytologic techniques, because these techniques require microdissection of small numbers of neoplastic cells followed by amplification of DNA and resultant stochastic effects may alter the validity of test results. In addition, the presence of polysomy or amplification in a cell population being analyzed by LOH may result in allelic imbalance due to other issues than loss of tumor suppressor genes. Polysomy of chromosome 17 might result in one TP53 allele appearing more prominent than the other in some of these assays. Thus, techniques must be used that normalize overall signals and control for neoplastic cellularity to assure validity.

A number of additional techniques have been investigated for their use in determining the presence of malignancy in pancreaticobiliary lesions. These include investigation of miRNAs.

Mutational analysis

A number of chromosomal alterations occur in PanNET but do not appear to play a role in the etiology of these neoplasms.[114,115,116] PanNETS arising in association with multiple endocrine neoplasia type 1 (MEN1) syndrome demonstrate germline mutations in the MEN1 gene characterized by losses at the 11q13 locus.[117] PanNETs do not characteristically harbor mutations in KRAS, TP53, p16 and SMADA4/DPC4.[117]

Molecular testing of acinar cell carcinoma

Acinar cell carcinomas appear to be associated with losses in the chromosome arm 11p.[92,93] Mutations in the APC gene have also been reported.[92,93] The diagnostic significance of these alterations is not yet clear.

Molecular testing of SPNs

The majority (>95%) of SPNs have a somatic point mutation in the beta-catenin gene (exon3).[94,95,96] Mutations in KRAS, p16 and SMAD4/DPC4 genes have not been reported in SPNs.

SUMMARY

Currently, FISH is the most clinically relevant ancillary technique applicable to FNA material from pancreatic lesions, because the addition of FISH analysis to routine cytologic evaluation appears to yield the highest sensitivity without loss in specificity. Loss of immunohistochemical staining for the protein product of SMAD4 and positive staining for MSLN support a diagnosis of ductal adenocarcinoma. Immunohistochemical markers for endocrine and exocrine differentiation are sufficient for a diagnosis of endocrine and acinar tumors, respectively. Nuclear staining for beta-catenin supports a diagnosis of solid-pseudopapillary neoplasm. Cyst fluid analysis for amylase and CEA also appears of diagnostic utility for classification of pancreatic cysts. A number of gene mutations (KRAS, GNAS, VHL, RNF43 and CTNNB1) may be of aid in the identification of specific cystic neoplasms. Other ancillary techniques do not appear to improve diagnostic sensitivity sufficiently to justify their increased cost for the evaluation of EUS-FNA and brushing specimens.

COMPETING INTERESTS STATEMENT BY ALL AUTHORS

The authors declare that they have no competing interests.

AUTHORSHIP STATEMENT BY ALL AUTHORS

All authors of this article declare that we qualify for authorship as defined by ICMJE http://www.icmje.org/#author. Each author has participated sufficiently in the work and take public responsibility for appropriate portions of the content of this article. Each author acknowledges that this final version was read and approved.

EDITORIAL/PEER-REVIEW STATEMENT

To ensure the integrity and highest quality of CytoJournal publications, general manuscripts undergo a double blind peer review process (authors are blinded for reviewers and vice versa) through automatic online system. However, the current article is part of guidelines by Papanicolaou Society of Cytopathology (PSC) submitted by the committee members as authors and is being published after due copy-editing without peer review. Because it is agreed to be published as Open Access article under Creative Commons Legal Code (http://creativecommons.org/licenses/by/2.0/), this intellectual property (IP) would be retained in public domain.