The World Health Organization Reporting System for Pancreaticobiliary Cytopathology

The World Health Organization (WHO), the International Academy of Cytology (IAC), and the International Agency for Research on Cancer (IARC) have joined forces to publish a series of cytopathology reporting systems, which present an international approach to reporting cytopathology of various body sites and mirror the WHO Classification of Tumours series with links between the two series on the website and in the text. Like the WHO Classification of Tumours, the WHO reporting systems provide an evidence-based terminology system with associated risks of malignancy (ROMs) and diagnostic management recommendations for each diagnostic category aimed to facilitate diagnosis and patient management. Each standardized terminology system provides key diagnostic cytopathological features of specific lesions or neoplasms, discusses ancillary studies for diagnostic and prognostic evaluation, and touches on the implications of diagnosis for patient care and management, all of which aim to improve the quality of diagnosis and reporting of cytopathology. Importantly, however, the reporting systems include reporting options that will recognize the variations in the availability of diagnostic and prognostic ancillary testing modalities in low- and middle-income countries, where cytopathology is particularly useful and is increasingly available in the absence of histopathological services. In addition to the printed versions, the cytopathology volumes will be available through the WHO Classification of Tumours online website, which includes all figures in each book as well as additional whole slide images [https://tumourclassification.iarc.who.int/welcome/].

A Standing Committee or “series editors” for these books include co-chairs Ian Cree and Andrew Field, Fernando Schmitt, Martha Pitman, and Ravi Mehrotra. This Standing Committee oversees the organization, development, writing, and editing of the WHO systems. The first four books in the series include reporting systems for pancreaticobiliary, lung, lymph node and spleen, and soft tissue cytopathology. Each specific terminology system has an Expert Editorial Board (EEB), who are also the authors of this pancreaticobiliary system review. The EEB members were chosen based on their expertise in the field and/or diversity of geographical representation using the same methodology used for the 5th Editions of the WHO Tumour Classification books. The content of the books was distributed among the EEB as responsible authors and/or editors, and co-authors were then added throughout the sections. The assignment of writing and editing responsibilities used the same model used for the WHO Classification of Tumours (https://whobluebooks.iarc.fr/about/faq/) [1‒3].

The WHO Reporting System for Pancreaticobiliary Cytopathology (WHO system) [4] revises and updates the Papanicolaou Society of Cytopathology (PSC) System for Reporting Pancreaticobiliary Cytology published in 2015 [5, 6]. The PSC system utilizes a six-tiered system and advocates close correlation with imaging and incorporation of ancillary testing into the final diagnosis, such as biochemical (CEA and amylase) and molecular testing of pancreatic cyst fluid [5]. The use of molecular tests on solid pancreatic neoplasms, biliary brushings, and neuroendocrine tumors is also addressed. The clear benefits of incorporating imaging information and ancillary testing are the decrease in the number of atypical and nondiagnostic reports and an increase in both sensitivity and specificity for the detection of neoplasia, which has helped refine the ROM in the diagnostic categories [5, 7‒15].

This approach is maintained in the WHO system. The principal differences between the WHO system and the PSC system revolve around the classification of neoplasia. In the PSC system, there was a single category for “neoplastic” lesions not diagnostic of adenocarcinoma or other aggressive malignancies that included two groups, one for benign neoplasms (primarily serous cystadenoma [SCA]) and one named “neoplastic: other,” dominated by premalignant intraductal neoplasms [primarily intraductal papillary mucinous neoplasms (IPMNs)] and low-grade malignant neoplasms [pancreatic neuroendocrine tumors (PanNET) and solid pseudopapillary neoplasms (SPN)]. The inclusion of both premalignant and low-grade malignant neoplasms in the “neoplastic: other” category made it difficult to determine a meaningful ROM for the neoplastic category as a whole and especially for the “other” group. In the WHO system, benign neoplasms with virtually no ROM are included in the “benign” category and low-grade malignancies (PanNET and SPN) are included in the malignant category, as per the WHO Classification of Digestive System Tumours [16], thus leaving in the “neoplasm” category primarily those noninvasive premalignant lesions of the ductal system. Because these lesions can be divided by the cytomorphological grade of the epithelium into low and high-grade with distinctly different ROMs [17], there are two distinct categories for these lesions in the WHO system: “pancreatic neoplasm of low risk/low-grade” (PaN-low) and “pancreatic neoplasm of high risk/high-grade” (PaN-high).

There are very few studies that provide data on the associated ROM that translate well into this new categorization system [10, 13, 17‒19]. With the redistribution of certain tumors within the categories from the PSC system to this new WHO system, particularly given that both PanNET and SPN are now categorized as malignant, the associated ROM for the “PaN-low” and “PaN-high” categories likely lies within their estimated ranges. The ROM for bile duct brushings is higher per diagnostic category than for fine-needle aspiration biopsies (FNAB) of pancreatic masses. Because criteria are lacking for the diagnosis of specific premalignant intraductal bile duct neoplasms, most cytopathological interpretations fall in the “benign (negative for malignancy),” “malignant,” or conventional indeterminate categories of “atypical” and “suspicious for malignancy” and not into the “PaN-low” or “PaN-high” categories.

The WHO system also discusses sampling techniques and tissue triage as well as rapid on-site evaluation (ROSE) and ancillary techniques, but the focus of this commentary is on the actual diagnostic categories, their ROM, and the recommended management per category, with an emphasis on the differences with the PSC Reporting System for Pancreaticobiliary Cytology. The categories of the WHO system along with the associated ROM and diagnostic management recommendations are detailed in Table 1 for pancreas and Table 2 for bile duct brushings and are briefly described in the sections below along with selective illustrations.

A specimen categorized as “inadequate/insufficient/nondiagnostic” is one that for qualitative and/or quantitative reasons does not permit a diagnosis of the targeted lesion.

Since precise terminology for this category varies among institutions, three terms are listed as options in the current terminology and any institution or cytopathology service should select one term and use it consistently to optimize communication with clinicians [6]. “Nondiagnostic” rates from published series with endoscopic ultrasound-directed FNABs of the pancreas show an average of 12% [20]. Diagnostic yield largely depends on the operator’s experience, technique employed, and use of ROSE, which has proven to reduce the number of “insufficient/inadequate/nondiagnostic” samples [21‒23]. Use of larger “core-like” needles has also been associated with fewer needle passes and shorter procedure times [24, 25]. The reason for an “insufficient/inadequate/nondiagnostic” sample may also be due to the nature of the lesion, since chronic and autoimmune pancreatitis shows extensive fibrosis and usually produces scant material. Brush cytopathology may also be paucicellular and often obscured by mechanical or preparation artifact, and an FNAB might be recommended when there is a significant thickening of the wall.

In pancreaticobiliary cytopathology, there is no established number of cells or epithelial tissue fragments defining a minimum for adequate cellularity, mainly due to its multidisciplinary approach, taking into consideration not only cytomorphological analysis of the slides but also imaging findings and chemical analysis of cyst fluids. While solid lesions or duct strictures with acellular to very paucicellular samples should be placed in this category, this is not true for cystic lesions, for which the presence of background mucin or high CEA levels may be enough to classify the cyst fluid as mucinous, even in the absence of an epithelial component [26, 27]. In this context, the main reasons for “insufficient/inadequate/nondiagnostic” samples include acellular or very paucicellular samples of a solid mass or duct brushing, technical issues, which obscure the cellular component, or the presence of only normal pancreatic tissue in the context of a well-formed solid or cystic lesion on imaging. Specimens containing only benign pancreatic acinar and/or ductal epithelial cells in the setting of a distinct solid or cystic lesion on imaging are classified as “nondiagnostic” by some cytopathologists, since the FNAB is regarded as most likely representing a sampling error as it does not explain the mass seen on imaging (Fig. 1). But it is recognized within the WHO system that some laboratories choose to classify such cases as “benign,” describing what is present on the slides, and then adding a caveat in the report conclusion, “that the biopsy most likely is not representative of the targeted lesion.” These different uses of the categories “nondiagnostic” and “benign” with a caveat may impact the ROM of these categories and require further study. It is important to note that the presence of any cellular atypia precludes categorization as “insufficient/inadequate/nondiagnostic” and the specimen should be placed in the “atypical” category for further follow-up.

ROM for the “insufficient/inadequate/nondiagnostic” category is based on retrospective and prospective studies and ranges from 5 to 25% [10, 13, 17, 20]. In biliary brushing specimens, the ROM in bile duct brushing specimens in the “insufficient/inadequate/nondiagnostic” category is high at 28–69% due to the sampling bias of targeting duct strictures with an inherently high ROM [28‒34].

Diagnostic management recommendations depend on the cause of an “insufficient/inadequate/nondiagnostic” sample. Operator experience, needle type and technique, lack of available organ-specific cytopathology expertise, and ROSE may all contribute to an insufficient biopsy [21‒23]. A repeat biopsy is recommended in most cases, possibly with different gauges and type of needles, and an added focus on obtaining tissue in formalin since sclerosis from pancreatitis and cancer often contributes to scant specimens. Patients clinically suspected of autoimmune pancreatitis may have a trial of steroids [35]. Brush cytopathology specimens evaluated with direct smears can be repeated using liquid-based cytopathology to improve cellular preservation. For some patients, further clinical management decisions may be based on the clinical scenario and imaging alone, but neoadjuvant therapy requires a definitive malignant diagnosis.

A specimen categorized as “benign (negative for malignancy)” demonstrates unequivocal benign cytopathological features, which may or may not be diagnostic of a specific process or benign neoplasm.

A “benign” categorization is rendered when the prepared slides do not show evidence of malignancy or cellular atypia. In those cases where a specific benign condition is recognized, this should be clearly stated in the diagnostic summary field of the report. When a specific lesion is identified, a microscopic description may not be necessary. The interpretation of a cytopathology sample as “benign” implies that the cellularity of the sample is adequate and that there is no evidence of cytopathological atypia. A benign diagnosis may be made on the basis of diagnosing a specific lesion such as splenule or pseudocyst, but may also include cases where there are only exocrine and endocrine pancreatic elements on the slides. A sample composed of normal pancreatic tissue in the appropriate clinical setting and in the absence of a distinct mass lesion is appropriately placed in the “benign” category. However, if there is a distinctive mass detected on imaging and the cytopathology preparations only show normal pancreatic tissue, some cytopathologists will categorize such a case as “nondiagnostic,” while others will categorize the case as “benign,” with a caveat in the report conclusion that the sample may not be representative. These different uses of the categories are recognized by the WHO system, and such disparate use may impact the category ROM, and needs further research. If only gastrointestinal tract contaminants are present, the categorization is “insufficient/inadequate/nondiagnostic” [8, 20].

The benign category includes both non-neoplastic and neoplastic entities. This is in contrast to the PSC system that places benign neoplasms, mainly SCAs, in the “neoplastic:benign” category. The benign entities include acute pancreatitis, chronic pancreatitis, autoimmune pancreatitis, cholangitis, pseudocyst, lymphoepithelial cyst, accessory spleen (splenule), SCA (Fig. 2), and other rare benign neoplasms, such as lymphangioma and schwannoma.

The ROM for a benign pancreatic FNAB ranges from 0 to 15% [10, 13, 17, 19]. However, the ROM for benign lesions may be overestimated since it is usually calculated using histopathological confirmation, and most cases with high clinical and imaging suspicion are surgically excised after a “benign” diagnosis [20]. The ROM for a “benign” bile duct brushing is as high as 55% based on retrospective studies [28‒34], and this high ROM should not be surprising given the high threshold for a malignant diagnosis in bile duct brushing cytopathology.

Surgical resection is not required for benign lesions. However, surgery may be selected to alleviate symptoms. Pseudocysts may be drained, while treating the underlying etiology of acute pancreatitis is the focus of the patient management. Management of the causes of chronic pancreatitis and improvement of nutrition are essential in improving patient health. Corticosteroids are used in the treatment of autoimmune pancreatitis, which is why it is essential for cytopathologists to consider the possibility of the disease based on the FNAB. After a diagnosis of lymphoepithelial cyst and splenule, the patient can be discharged.

A specimen categorized as “atypical” demonstrates features predominantly seen in benign lesions and minimal features that may raise the possibility of a malignant lesion, but with insufficient features either in number or quality to diagnose a benign, PaN-low, PaN-high, or malignant process or lesion.

As in the PSC system, the WHO system retains the two indeterminate categories, “atypical” and “suspicious for malignancy.” The major benefit of implementing the PSC system in 2015 with the integration of imaging findings and ancillary tests into the final diagnosis was the decreased number of “atypical” diagnoses and the increase in definitive diagnoses of a “neoplasm,” particularly for FNAB of cystic lesions [6]. The reasons for using the “atypical” category are multiple. The inherent characteristics of pancreaticobiliary lesion, low tumor cellularity, and anatomically challenging tumor sites may all preclude the rendering of a definitive diagnosis. Technical factors, such as the endosonographer’s skill in procuring adequately cellular and representative samples, specimen preparation artifact, and the availability of ROSE will all also influence the quality of the specimen [36, 37]. In addition, different levels of experience and expertise, training and institutional case volume can result in significant inter- and intra-observer variability in the use of the “atypical” category [38]. The histopathological outcome of this category ranges from benign to premalignant and malignant entities [10‒12, 17, 39‒41]. FNAB with no supportive evidence of a specific neoplasm remains in the “atypical” category.

The reported frequency of the “atypical” category for FNAB of pancreas ranges from 0 to 14% with an average of 5.5% [20, 42]. The frequency in bile duct brushings is higher, ranging from 11 to 39.8%, which may be partially due to the challenge in recognizing reactive atypia inherent to primary sclerosing cholangitis, stents, and biliary stones [29, 43‒45].

In the PSC system, indeterminate FNAB not diagnostic of a well-differentiated PanNET or SPN, which are tumors that were in the “neoplastic: other” rather than “malignant” category, were classified as “atypical” rather than “suspicious for malignancy.” Now that these tumors are classified as “malignant,” if an FNAB of either of these tumors is not diagnostic, but is suspected, then the interpretation should be “suspicious for malignancy” and not “atypical.” The “atypical” category is appropriate when the differential diagnosis of finding a few endocrine cells includes islet cell hyperplasia in the clinical setting of chronic pancreatitis [46, 47]. It is important to understand that “atypical” should be used judiciously and as seldom as possible, because an “atypical” report may lead to unnecessary additional investigations and procedures.

For bile duct brushings, the “atypical” category is applied to cases in which the atypia observed is beyond that seen for reactive and inflammatory changes while quantitatively and qualitatively insufficient for categorization as “suspicious for malignancy” (Fig. 3). Although low-grade biliary intraepithelial neoplasia and intraductal papillary neoplasm of the bile duct (IPN-B) are entities included in the “PaN-low” category in this WHO system, the most appropriate category is “atypical” because these entities lack well-defined criteria and the degree of atypia is mild [48, 49]. The known overlap in cytomorphology between reactive and reparative changes in bile duct epithelium from stents, stones, and inflammation, and well-differentiated adenocarcinoma leads to a high use of indeterminate interpretations [50].

The ROM associated with the diagnostic categories of the PSC system has been refined [11, 40, 43, 51]. ROM of the “atypical” category is different for bile duct brushings and EUS-FNAB of pancreatic lesions. There is a remarkable range in the ROM for bile duct brushings and pancreas FNAB depending on the type of study, date of the study (e.g., pre-PSC system), and experience of the institution reporting the study. Using studies following the establishment of the PSC system where the category “neoplastic: other” provided a home for nonmalignant neoplasia that otherwise would have been classified as “atypical” or “suspicious for malignancy,” the ROM of the “atypical” category for pancreas FNAB is 30–40% and for bile duct brushings 25–77% [10, 13, 17, 28‒31].

The management of an “atypical” cytopathological diagnosis should include multidisciplinary discussion, consensus review, expert consultation, the use of ancillary tests, and repeat sampling with ROSE. In recent years, evolving technologies such as multiprobe fluorescence in situ hybridization (FISH) [52‒55] and molecular tests with genetic sequencing have been reported to improve the classification of pancreatic cysts [56, 57] and the sensitivity of detection of malignancy in bile duct brushings [30, 58‒61]. Molecular testing using either otherwise discarded supernatant fluid of cytocentrifuged bile duct brushing specimens [58] or an aliquot of a brushing rinsed in preservative has significantly increased the sensitivity of detecting malignancy to 93%, while maintaining a specificity of 100% and also identifying markers potentially predictive of a response to targeted therapy [59]. Therefore, molecular tests increase the information available for clinical decision-making. Precisely defining the correct categorization and specific diagnosis of pancreaticobiliary cytopathology is challenging, and thus, consensus review or second opinion from experienced pancreaticobiliary cytopathologists may be cost-effective and is recommended [62]. If ancillary testing is not available and multidisciplinary discussion is inconclusive, further workup may include repeat sampling with ROSE.

A specimen categorized as “PaN-low ” has features of an intraductal and/or cystic neoplasm with low-grade epithelial atypia.

The grading of intraductal neoplasia is now two-tiered as low and high-grade, with low-grade epithelial atypia representing low- to intermediate-grade dysplasia [63, 64]. The cytopathological ROM of PaN-low lesions is directly related to the epithelial atypia in the cyst fluid or brushing cytopathology. In one study on cases categorized in the PSC system as “neoplastic: other,” which included all the lesions placed in that category, low-grade atypia had a ROM of 4.3% compared to 90% for high-grade atypia [65]. IPN-B is classified into two types, one of which is similar to pancreatic IPMN and occurs mostly in the intrahepatic ducts, and the other which is unlike pancreatic IPMN and occurs in the extrahepatic ducts [66‒68]. IPMN and mucinous cystic neoplasm (MCN) are well-known cystic lesions for which cytopathological distinction between low and high-grades is essential for patient care. Cyst fluid CEA analysis is a valuable adjunct to the cytopathological diagnosis of these mucinous lesions. When its level is greater than 192 ng/mL, it is approximately 80% accurate for a mucin-producing neoplasm [69, 70]. Accurate grading of the epithelial atypia as low risk/low-grade, which includes intermediate-grade dysplasia, and distinguishing it from high-risk/high-grade atypia, is challenging requiring well-preserved epithelium and diagnostic experience. Undoubtedly, there will be interobserver variability that will need close cytohistologic correlation and quality assurance review in a prospective manner.

Low-grade IPMN (Fig. 4) is a mucin-producing epithelial neoplasm of the main and/or branch ducts of the pancreas characterized by an intraductal proliferation of columnar mucinous epithelium with mild to moderate cytoarchitectural atypia, lining cystically dilated ducts with or without intraductal papillae. Low-grade neoplasms typically show gastric and/or intestinal type epithelium [71]. The primary differential diagnosis for low-grade IPMN is gastrointestinal contamination [72]. Low-grade MCN is a cyst-forming and mucin-producing epithelial neoplasm with distinctive ovarian-like subepithelial stroma and low-grade dysplasia. Low-grade IPMN is not readily distinguished from low-grade MCN on cytopathology because subepithelial ovarian-type stroma is usually not sampled with aspiration of cyst fluid. The use of micro-forceps biopsy, which obtains a bite of the cyst wall, can help make a specific diagnosis of MCN by obtaining a tissue sample demonstrating the ovarian-type stroma [73]. Identification of a GNAS mutation supports the diagnosis of IPMN [56, 74, 75]. IPMN harbors KRAS mutations, but KRASmutations are identified only in a small subset of low-grade MCN [76, 77]. Other gene alterations such as TP53 and SMAD4 mutations are absent in low-grade cysts [56, 75].

Stratifying epithelial atypia into high-grade or low-grade leads to better patient management, as pancreatic cyst fluid FNAB is rarely interpreted as suspicious for or diagnostic of malignancy due to their scant cellularity. In the absence of cytopathological high-grade atypia, and in the presence of low-grade atypia, the patient has the option of surveillance depending on the clinical and imaging features [78‒80]. Patients with MCN and low-grade atypia on cytopathology can be referred for surveillance in the absence of other high-risk factors [81, 82].

A specimen categorized as “PaN-low” has features of an intraductal and/or cystic neoplasm with high-grade epithelial atypia.

With the removal of PanNET and SPN from the PSC “neoplastic” category and the inclusion of SCA in the WHO system “benign” category, the “pancreatic neoplasm” category is left almost exclusively containing cystic mucinous and intraductal neoplasms. Given the marked difference in the ROM between low-grade and high-grade epithelial atypia [17], the neoplastic category was renamed “pancreatic neoplasm” and subdivided into low-grade (PaN-low) and high-grade (PaN-high) categories. PaN-high lesions include all intraductal and cystic pancreaticobiliary lesions with high-grade epithelial atypia [HGA]. In cytopathology of the pancreas, HGA encompasses lesions with high-grade dysplasia and invasive carcinoma since these usually cannot be distinguished in aspirates of cyst fluid. PaN-high is predictive of an increased risk for high-grade dysplasia or carcinoma and carries an estimated ROM of 60–95% [10, 13, 17, 19]. The PaN-high category provides a more flexible and less anxious patient management paradigm than the “suspicious for malignancy” category, particularly when conservative patient management is recommended. HGA has an 89% sensitivity and 98% specificity for detecting a high-risk cyst [65].

HGA is defined as a cell smaller than a 12-µ duodenal enterocyte with high nuclear-to-cytoplasmic ratio and abnormal chromatin, which can be hypochromatic or hyperchromatic and with or without background necrosis. When these criteria are used, there is overall good interobserver agreement in distinguishing low-risk/grade from high-risk/grade cysts [63, 80, 83‒85]. However, if an intermediate-grade dysplasia is included in the grading scheme, it is virtually impossible to accurately stratify cysts with intermediate-grade dysplasia into low- and high-risk groups, making the addition of genetic testing very important in accurately identifying high-risk cysts [56, 84].

Lesions included in the PaN-high category include flat high-grade pancreatic/biliary intraepithelial neoplasia, intraductal mucinous and nonmucinous lesions with HGA (IPMN, IPN-B, intraductal oncocytic papillary neoplasm, and intraductal tubulopapillary neoplasm) and MCN with HGA. The PaN-high category is dominated by IPMN-high-grade, which is a grossly visible cystic lesion involving the main and/or branch pancreatic ducts. Histopathologically IPMN-high-grade forms macro- and micropapillae lined by cells with severe cytopathological atypia. Key cytopathological features include hypercellularity relative to low-grade IPMN, small cell size that is smaller than a 12-µm duodenal enterocyte, increased nuclear-to-cytoplasmic ratio, nuclear irregularity, abnormal chromatin, which can be hypochromasia or hyperchromasia, prominent nucleoli, variable cytoplasmic mucin, background necrosis, and inflammation (Fig. 5). The primary differential diagnosis is with MCN with HGA, which is a cystic mucin-producing epithelial neoplasm with distinctive subepithelial ovarian-type stroma, high-grade dysplasia with or without associated invasive carcinoma (15%), and no connection to the pancreatic ducts [86‒89]. Almost all MCN with HGA occurs in middle-aged women, and more than 90% arise in the pancreatic body/tail. On FNAB, MCN with HGA shows variable cellularity with thick, colloid-like mucin, necrosis and crowded sheets, papillae, and singly dispersed pleomorphic cells with high nuclear-to-cytoplasmic ratio, nuclear irregularities, hypo- or hyperchromasia, mitotic figures, and scant mucin. The entity-defining ovarian-type stroma is often not sampled on FNAB [82].

IPMN and MCN with HGA may show mutations known to be associated with adenocarcinoma and high-grade dysplasia. Late mutations in the adenoma-carcinoma progression include TP53 [56, 57, 90, 91], SMAD4 [56, 90, 91], CDKN2A (p16) [56, 91], PTEN [57, 92], PIK3CA [57], and aneuploidy/LoH in certain regions [93]. Mutant allelic frequency mutations in PIK3CA, PTEN, and TP53 genes equivalent to KRAS or GNAS are sensitive and specific for IPMN with either high-grade dysplasia or invasive carcinoma [57]. Low-level mutations of PTEN, PIK3CA, and TP53 in low-grade IPMN suggest risk for transformation to high-grade [57]. Mutational allelic frequencies for GNAS greater than 55% correlate with high-grade IPMN [57]. Immunohistochemical stains on cell blocks include p53 mutant expression demonstrated by strong nuclear staining or no expression [null pattern] [94, 95], and loss of nuclear SMAD4 [96, 97] or p16 [98]. These stains should be interpreted with caution on scant specimens.

Surgically fit patients with PaN-high should be considered for surgical resection. However, conservative observation is a reasonable option where the risk of surgery is high and especially if the imaging features do not appear high risk and the patient is a poor surgical candidate.

A specimen characterized as “suspicious for malignancy” demonstrates some cytopathologic features suggestive of malignancy but with insufficient features in either number or quality to make an unequivocal diagnosis of malignancy.

The cytomorphological features of specimens obtained by FNAB or bile duct brushings show a spectrum of features ranging from those clearly representative of benign epithelium to those definitively diagnostic of malignancy. This spectrum is nearly continuous without obvious cut points for subcategorization. The two indeterminate categories of “atypical” and “suspicious for malignancy” help optimize the accuracy and predictive values of the categories “benign/negative for malignancy” and “malignant.” Assignment of specimens to the “suspicious for malignancy” category varies among cytopathologists, and this variance is reflected in the published range for ROM [10, 11, 17, 19, 99] and reported interobserver agreement statistics [100, 101]. The reported ranges in ROM and interobserver reproducibility are associated with the overlapping clinical and imaging features of benign and malignant entities, difficulty in sampling and varying experience levels of both endoscopists and cytopathologists [17, 102‒105]. All of these factors result in the need for intermediate categories which group the range of cytomorphological features into clinically useful and reproducible categories.

The use of these indeterminate categories should be limited to retain the diagnostic utility of the categorization system. The “suspicious for malignancy” category accounts for approximately 4.7–16% of all cytopathological diagnoses reported in the literature for the pancreaticobiliary tract [29, 51, 106]. The use of the “suspicious for malignancy” category indicates that the cytopathological findings are highly concerning for but not diagnostic of a malignancy. Factors contributing to the “suspicious for malignancy” category include scant cellularity, technical limitations of specimen staining or preparation, cytomorphological characteristics of the specimen along the spectrum of features, and caution on the part of the cytopathologist. These limiting factors may be especially constraining when insufficient tissue is present for ancillary testing. Additional factors resulting in a “suspicious for malignancy” rather than a “malignant” diagnosis include concurrent pancreatitis, stent placement, stones, inflammatory conditions such as sclerosing cholangitis, and sampling of subclinical premalignant lesions of the pancreas (high-grade pancreatic intraepithelial neoplasia/high-grade dysplasia/carcinoma in situ) [17, 107‒109].

Specimens assigned to the “suspicious for malignancy” category demonstrate significant alterations in both architecture and the cytomorphology of single cells. Assignment to the “atypical,” “suspicious for malignancy,” or “malignant” categories depends on the combination of features and the degree to which they are expressed [110‒112]. When the architectural and cytopathological features of malignancy are not there in quality or quantity to make a diagnosis of malignancy but are present to a significant enough degree to suspect that a malignancy is present, then the “suspicious for malignancy” category is used (Fig. 6). The cytopathologist should include details of the features that are suspicious and that have led to the categorization of a specimen as “suspicious for malignancy,” to minimize the tendency to use this category as a waste basket. A discussion of any features limiting or compromising the evaluation should be noted, and the need for correlation with clinical and imaging findings should be emphasized. Finally, the use and results of any ancillary testing performed including immunohistochemical and molecular analysis should be included and evaluated.

The ROM varies depending on the solid or cystic nature of the lesion. Reported ROM for the “suspicious for malignancy” category for pancreatic FNAB has varied from 80 to 100% [10, 11, 17, 19, 99]. Malignancy risks for bile duct brushing specimens range from 74 to 100% [28‒31, 33, 44, 113, 114]. These ranges in ROM are relatively wide and reflect the degrees of interobserver reproducibility [100, 101].

Categorization of a specimen as “suspicious for malignancy” is not equivalent to a “malignant” diagnosis and should not by itself result in neoadjuvant therapy or radical surgery. In all cases, further patient management and clinical decisions require correlation of the “suspicious for malignancy” cytopathological categorization with clinical and imaging findings which are best correlated in a treatment planning multidisciplinary conference. When clinical and imaging findings strongly support the diagnosis of malignancy, a categorization of a specimen as “suspicious for malignancy” may in some restrictive settings allow for definitive therapy without further histopathological or cytopathological procedures. Ancillary testing also may contribute significantly to the clinical decision-making process.

A specimen categorized as “malignant” demonstrates unequivocal cytopathological features of malignancy.

Tumors in this category include both primary and secondary malignancies with pancreatic ductal adenocarcinoma (PDAC) and cholangiocarcinoma accounting for most primary malignancies [115]. As in the PSC system [5], pancreatic acinar cell carcinoma, neuroendocrine carcinomas, pancreatoblastoma, primary and secondary hematopoietic malignancies, and sarcomas are also included in this category. A change from the original PSC system is the inclusion of well-differentiated PanNET and SPN in this category to make this system consistent with the classification of pancreatic tumors in the WHO 5th edition [115]. In the PSC system, these were included in the “neoplastic other” category.

The ROM of a positive pancreatic FNAB ranges between 97 and 100% [10, 11, 17, 48, 116], and the ROM of a positive bile duct brushing ranges between 96 and 100% [10, 11, 17, 48, 116]. The cytopathological features of PDAC on FNAB smears include hypercellularity, loss of the normal honeycomb pattern in cell groups, nuclear enlargement, chromatin clearing, nuclear membrane irregularities, anisonucleosis greater than 4: 1 in tissue fragments, cytoplasmic mucin, atypical mitotic figures, and dispersed single malignant cells. Coagulative background necrosis is present in higher grade carcinomas, but well-differentiated PDAC may have a clean background (Fig. 7) [110, 117‒120]. The less common cytomorphological subtypes of PDAC include adenosquamous carcinoma characterized by the presence of keratinized tumor cells mixed with glandular elements, colloid adenocarcinoma with abundant background mucin, undifferentiated carcinoma which can be anaplastic or sarcomatoid, and undifferentiated carcinoma with osteoclast giant cells, which represent benign multinucleated histiocytes. Each of these subtypes has unique cytopathological features [121‒123].

Nonductal neoplasms of the pancreas include PanNET (Fig. 8), SPN (Fig. 9), neuroendocrine carcinoma, acinar cell carcinoma (Fig. 10), and pancreatoblastoma. These share some common cytopathological features including hypercellular smears composed of loosely cohesive, monotonous appearing neoplastic cells, often with a vascular stroma, and stripped nuclei. However, they do have distinguishing cytopathological and immunophenotypical features (Table 3).

Surgical management is usually the first-line treatment for malignancies of the pancreaticobiliary tract [124]. An exception is patients with a PanNET smaller than 2 cm and a Ki-67 less than 3%, who may be managed with surveillance [125]. Patients presenting with borderline or locally advanced PDAC may be treated with neoadjuvant therapies in an attempt to convert the PDAC to resectable disease [126]. Patients with unresectable disease are typically treated with a combination of chemotherapeutic agents and radiation therapy. The type of systemic therapy administered is dependent on the tumor type. Metastases may be treated with resection or systemic therapy, depending on the tumor type, e.g., metastatic renal cell carcinoma occurring many years after initial presentation may be resected.

As with all reporting systems involving categorization of cytopathology specimens, the new WHO Reporting System for Pancreaticobiliary Cytopathology is designed to improve communication between clinicians and cytopathologists about their patient’s biopsy results. Each category has a calculated ROM that aims to assist the clinical care team in patient management. The authors recognize that the performance indicators for each category of the system are derived from reviews of relatively recent literature, so we hope that the WHO system will encourage research into the utility of the system and its management recommendations and provide an ever more precise ROM for each category.

The WHO system also defines through an international consensus the key diagnostic cytopathological criteria and differential diagnosis for each lesion or tumor, which is essential to improving the quality of diagnostic assessment and reporting of pancreaticobiliary cytopathology. In addition, the WHO system provides the current best practice application of ancillary testing, including immunocytochemistry and molecular testing and, importantly, provides detailed descriptions of sampling and processing techniques to optimize the handling and preparation of the cytopathology sample. The authors of the WHO system recognize that local medical and pathology resources and infrastructure will vary, particularly in low- and middle-income countries. To make the WHO system applicable worldwide, the system is based on cytomorphology and provides options for further diagnostic workup of the specimen.

The WHO Reporting System for Pancreaticobiliary Cytopathology provides a direct and dynamic link to the WHO Classification for Gastrointestinal Tumours, 5th edition, and raises the profile and use of cytopathology by increasing awareness of its current role in diagnosis and management of patients with pancreaticobiliary disease.

The content of this article represents the personal views of the authors and does not represent the views of the authors’ employers and associated institutions. Where authors are identified as personnel of the International Agency for Research on Cancer/World Health Organization, the authors alone are responsible for the views expressed in this article and they do not necessarily represent the decisions, policy, or views of the International Agency for Research on Cancer/World Health Organization.

There are no funding sources.

Martha B. Pitman, Barbara A. Centeno, Michelle D. Reid, Momin T. Siddiqu, Lester J. Layfield, Miguel Perez-Machado, Birgit Weynand, Edward B. Stelow, Maria D. Lozano, Noriyoshi Fukushima, Ian A. Cree, Ravi Mehrotra, Fernando C. Schmitt, and Andrew S. Field: conceptualization. Martha B. Pitman, with collaboration of all authors: writing original draft. All authors read and approved the final manuscript: final review.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.