Diagnostic Pearls and Pitfalls in the Evaluation of Biopsies of the Pancreas

Pancreatic ductal adenocarcinoma (PDAC) is the most common malignancy of the pancreas.^1–9  The diagnosis of this entity based on small biopsies can be challenging for pathologists. However, given the critical implications of such a diagnosis, the diagnosis should be based on robust and undebatable criteria. The following features can be diagnostic of invasive adenocarcinoma, above all if they coexist on the same diagnostic material: (1) presence of ruptured glands/incomplete lumina with necrotic debris, (2) cytologic atypia with nuclear size variation greater than 4:1, (3) presence of mitoses (and above all atypical mitoses) in glandular elements, (4) loss of nuclear polarity, (5) infiltrative growth around muscular vessels, (6) perineural invasion, and (7) glands present within adipose tissue, and/or thick smooth/skeletal muscle (Figure 1, A through C; Table 1).^5,7,8  The normal pancreatic parenchyma present in small biopsies can be used as a reference point for comparison with neoplastic components. It can also allow evaluation of the infiltrative nature of the lesion. Globally, these criteria are useful in diagnosing PDAC.

However, the documentation of such findings can be difficult for small biopsies. Indeed, tissue cores from endoscopic ultrasound–guided procedures are often thin and fragmented, sometimes harboring hemorrhagic foci, which pose substantial challenges for diagnostic pathologists.^10–12  Moreover, the phenomenon of core fragmentation can produce displacement of the epithelium, even if nonneoplastic, into stromal tissue, mimicking true invasion.⁷  Thus, in cases of marked tissue fragmentation in biopsies, pathologists should be aware of the possibility of displacement. Another important pitfall is the distinction between reactive and atrophic ducts and neoplastic glands. In addition to the aforementioned 7 criteria classically associated with infiltrative glands, pathologists should also evaluate the tissue architecture because nonneoplastic elements usually preserve an organoid structure, maintaining the lobulated architecture of the pancreatic parenchyma. Other important pitfalls include: (1) the evaluation of fibrosis, because fibrosis in chronic pancreatitis can mimic the desmoplastic reaction of PDAC; and (2) the potential misinterpretation of the presence of glands within the muscle. This phenomenon can be observed in cases where the biopsy could have traversed the ampulla, also including normal ducts intermingled with smooth muscle bundles of Oddi sphincter.^5,7,8 

Of note, a nonnegligible number of biopsies may harbor small ducts and glands that have an angulated shape and infiltrative appearance but at the same time lack cytologic atypia or show only 1 or very few PDAC diagnostic criteria. In such cases, morphology alone is unconvincing, leading to a descriptive diagnosis without definitive conclusions. To overcome these difficulties, specific immunohistochemistry (IHC) for p53 and Smad4 (Dpc4) can be used (Figure 1, D and E; Table 1).^5,7,8  These proteins are encoded by homonymous genes that act as tumor suppressors. If present, their mutational status represents a late event in pancreatic carcinogenesis; thus, an altered pattern of protein staining on IHC can be used as a marker of adenocarcinoma.^13–17  Normally, Smad4 is universally intact, whereas p53 shows weak nuclear staining in up to 10% to 15% of cells. Mutations in SMAD4 are seen on IHC as a loss of labeling, whereas mutations in TP53 are seen as diffuse strong nuclear overexpression or a complete loss of nuclear expression.^13–17  It should be noted that about 40% to 45% of PDACs do not harbor SMAD4 mutations, and a similar situation is observed for TP53. Thus, although the altered Smad4/p53 staining pattern in atypical glands is highly specific to PDAC, the intact expression of Smad4 and normal expression of p53 are inconclusive for the diagnosis of PDAC. Along this line, an additional helpful immunohistochemical marker may be ARID1A, a protein encoded by a homonymous gene that acts as a chromatin remodeler.^18,19  Its mutational status has been reported in up to 15% of PDAC cases and is usually seen as a loss of labeling by IHC.¹⁹  Although histomorphology remains the most important tool for diagnosis, its integration with specific IHC may be decisive in providing a correct interpretation of challenging cases, avoiding potential misinterpretation. Along this line, other potentially useful immunohistochemical markers for supporting the diagnosis of PDAC are S100P, insulin-like growth factor 2 messenger RNA–binding protein 3 (IMP-3), and Maspin, which are usually positive in PDAC, and von Hippel-Lindau tumor suppressor gene protein (pVHL), which is usually negative in PDAC but positive in normal ducts and acini.^20,21  Notably, such markers can also give reliable results on material from fine-needle aspiration/biopsy.^20,21 

It is also important to remember that autoimmune pancreatitis (AIP), a rare but nonnegligible entity, mimics PDAC both radiologically and histologically.^22,23  The classic histologic features of AIP, including inflammatory changes, fibrosis with a storiform pattern obscuring the underlying pancreatic architecture, inflammation with a duct-centric pattern, granulocytic-epithelial lesions, obliterative phlebitis, and immunoglobulin G (IgG)/IgG4 immunostaining, are often poorly represented and patchy on small biopsies.^5,7,8,22  In this context, the most important task for diagnostic pathologists remains excluding the presence of a PDAC, also with the support of IHC (p53, Smad4, ARID1A). Given their specific expression pattern, S100P, IMP-3, Maspin, and pVHL can also be used in the differential diagnosis of PDAC versus AIP.²¹  Pathologists suggest the presence of an autoimmune disorder only in cases with classical AIP-associated histologic features; however, the final diagnosis should be made by integrating the pathology report with clinical parameters and specific data from imaging and clinical laboratories (always considering IgG4 serum levels).

Regarding PDAC variants, it is important to note that a significant fraction of PDACs may show peculiar cytologic/histologic features, such as adenosquamous, medullary, or micropapillary components, or signet ring, osteoclast-like, and sarcomatoid cells.^6,8,9,24  However, the diagnosis of such variants must satisfy specific thresholds belonging to each component according to the World Health Organization (WHO) classification (eg, adenosquamous: at least 30% of squamous cells; signet ring: at least 80% of signet ring cells).⁶  Thus, the final diagnosis of PDAC variants is possible only on surgical specimens. However, pathologists should report any carcinoma components that are not conventional PDAC on small biopsies because these findings may be useful for the differential diagnosis and clinical management of patients.²⁵ 

Pancreatic solid cellular neoplasms comprise different entities with striking morphologic and immunohistochemical similarities, making their distinction challenging in small biopsies. They are: neuroendocrine neoplasms (NENs), acinar cell carcinomas (ACCs), pancreatoblastomas (PBs), and solid pseudopapillary neoplasms (SPNs).⁶  The integration of morphologic criteria with specific IHC represents the best approach for this complex scenario.

NENs represent an important topic for diagnostic pathologists. The first task involves the identification of NEN-related material within the biopsy specimen. Subsequently, it becomes imperative to provide an estimation of the proliferative index via Ki-67 IHC and to distinguish between well-differentiated G1, G2, and G3 pancreatic neuroendocrine tumors (PanNETs) versus poorly differentiated G3 pancreatic neuroendocrine carcinoma (PanNEC).⁶ 

Well-differentiated PanNETs usually show classic histologic architecture, such as solid-trabecular, nested, or ribbonlike. G1 and G2 tumors have a Ki-67 proliferative index less than 20%, whereas in G3 PanNETs, it is greater than 20%.⁶  Neoplastic cells are usually monomorphic, showing chromatin typically organized with a “salt and pepper” appearance.^6,26–30  PanNECs are usually very different from PanNETs in that they show marked cytologic atypia, diffuse solid architecture, tumor necrosis, and high Ki-67 levels (usually >60%).^6,26–31  With the exception of PanNET G3, where G1 and G2 areas are typically intermingled with G3 foci, the estimation of the Ki-67 proliferative index is often reliable for small biopsies using both manual count and digital pathology–based devices.³¹  Although the morphologies of PanNETs and PanNECs are strikingly different, the differential between G3 PanNETs and PanNECs represents a substantial potential pitfall in small biopsies, where the recognition of classic neuroendocrine versus poorly differentiated features may be challenging. Notably, the distinction between PanNETs and PanNECs has important clinical consequences, as the prognosis and therapeutic strategies are very different. Table 2 and Figure 2 summarize the most significant differences between these 2 entities.

To support this distinction, recent discoveries in molecular studies have provided important opportunities for neuroendocrine diagnostic workflows. Indeed, well-differentiated PanNETs express the classical driver gene MEN1 and 2 chromatin remodelers, death domain–associated protein (DAXX) and α-thalassemia/mental retardation X-linked chromatin remodeler (ATRX).^24,32–36  In IHC, DAXX/ATRX mutations, which are mutually exclusive, can be detected as a complete loss of nuclear expression of homonymous proteins.^24,32–36  Because DAXX/ATRX mutations are typical of PanNETs, their immunohistochemical loss of expression represents important support for a PanNET diagnosis. Moreover, they are important PanNET prognostic factors and can be reliably determined from small biopsies.³⁷  Conversely, TP53 and RB1 are typical drivers of PanNECs.^24,27,38,39  In IHC, if TP53 mutations are seen as diffuse strong nuclear overexpression or complete loss of nuclear expression of p53, as already mentioned for PDAC, RB1 alterations in PanNECs usually result in a complete loss of nuclear staining (Figure 2, D).^24,29,39  In a neuroendocrine setting, this alteration is important for the diagnosis of PanNECs. Along this line, the immunohistochemical use of p16 may also be useful for distinguishing PanNECs (diffuse strong staining in up to 70% of cases) from G3 PanNETs.³⁹ 

In addition to the differences between PanNETs and PanNECs, there are other critical pitfalls in the neuroendocrine topic that should be acknowledged. These intricacies manifest as histologic variants of PanNETs, which have the potential to give rise to misinterpretations and diagnostic challenges. These include the oncocytic variant of PanNETs, which can mimic an intraductal oncocytic neoplasm and an acinar cell carcinoma; the clear cell variant, which can mimic metastasis from a clear cell renal cell carcinoma; the hepatoid variant, which can mimic a hepatoid carcinoma or a hepatoid SPNs; and the SPN-like variant, which can closely mimic an SPN.^26,40 

For all the aforementioned reasons, the diagnosis of PanNENs should be made based on robust histologic criteria, but also on standardized immunohistochemical markers, including positive immunostain for synaptophysin and chromogranin-A (diffuse and intense cytoplasmic expression) and INSM1 (nuclear positivity), as well as negative immunostain for BCL10 and β-catenin.^{6,24,26,41,42}  Another potentially useful marker for supporting the neuroendocrine nature of a neoplasm is NKX6.1, which is expressed in about 70% of PanNETs and is also expressed in PanNECs.⁴³  Specific IHC for selected cases, as discussed, includes DAXX, ATRX, p53, and Rb1. One of the most important considerations regarding the use of IHC in the diagnostic workflow of PanNENs is to prefer the adoption of a panel of antibodies, rather than only 1 marker. Indeed, for example, synaptophysin can be expressed by SPNs (IHC positive in up to 40% of cases, albeit with a patchy pattern), and chromogranin-A can be positive in ACCs (in up to 10%–20% of cases).

A last consideration should be made on those PanNETs treated by neoadjuvant therapy–based approaches. Although existing data are currently insufficient to draw reliable conclusions, special attention should be paid to this scenario. The potential impact of neoadjuvant therapy on the mitotic count and Ki-67 of resected specimens could pose significant challenges to tumor grading.⁴⁴  Further studies are needed to improve our understanding of this important topic.

Pancreatic neoplasms with acinar cell differentiation include ACC, PB, and mixed acinar-neuroendocrine carcinoma.^5,6 

ACC is a rare, malignant pancreatic neoplasm that is usually highly cellular; typically, it lacks desmoplastic stroma, a classical morphologic aspect of PDAC. It can display different morphologic features, ranging from bland cytologic atypia, with neoplastic elements resembling benign acinar cells and maintaining a nested/lobular architecture, to marked atypia, mimicking PDAC cytologic features (Figure 3, A through C).^45,46  In all cases, the hallmarks of neoplastic cells in the ACC are granular eosinophilic cytoplasm and prominent central nucleoli (Figure 3, C). Thus, the distinction between ACC and other entities, and above all, PDAC and PanNET/NEC, is based on histologic and cytologic features. However, there are also some immunohistochemical markers that can be very useful for demonstrating acinar differentiation, such as BCL10 (Figure 3, D) and trypsin.^45–48  If used simultaneously, these markers can reach very high values of sensitivity and specificity (>95%).⁴⁸  Other potentially useful markers for demonstrating the acinar cell differentiation are carboxypeptidase A1 (CPA1), carboxypeptidase A2 (CPA2), and glycoprotein 2 (GP2), with the first 2 markers being the most specific.⁴⁹  In some cases, diagnostic pathologists are required to distinguish ACC from benign acinar cells, which may represent a critical pitfall in the diagnosis of ACC using small biopsies. Under these circumstances, BCL10 and trypsin cannot rule out normal acinar cells, because both (neoplastic and nonneoplastic) are positive for these markers. This distinction should be based on cytohistologic features, but IHC for Ki-67 can be of great help in the differential diagnosis between normal acinar cells and ACC. Indeed, it is usually very low in normal acinar elements (<1%), whereas it is significantly higher in ACC (usually >20%, and even >80% in some cases). Along these lines, another entity that can enter the differential diagnosis of ACC, because of the presence of cells with acinar differentiation, is acinar cystic transformation (ACT) of the pancreas. It is a cystic formation usually lined by normal acinar epithelium. Because ACT is considered a benign entity, its diagnosis in the preoperative setting can spare patients from unnecessary surgical resections in cases of asymptomatic lesions.^50–53 

PBs are rare pancreatic tumors with acinar differentiation.⁶  In small biopsies, neoplastic elements of PBs usually appear as monotonous, blastlike cells with a high nuclear-to-cytoplasmic ratio, fine chromatin, small, distinct nucleoli, and a resemblance to neuroendocrine elements.⁵³  The cells are positive for BCL10 and trypsin expression. The diagnostic hallmark of PBs is the presence of squamoid nests within the tumor mass.^6,54  However, this feature is not always present in small biopsies, sometimes rendering the identification of this tumor type very challenging or even impossible on this type of material. When present, neoplastic cells of squamous nests usually show an enlarged and pale cytoplasm with nuclear clearing; abnormal nuclear positivity for β-catenin is usually detected.⁵⁴  Moreover, considering its rarity, integration of histology with IHC remains the most important diagnostic tool for PB.

SPNs usually exhibit peculiar and distinctive morphologic features that are useful for the differential diagnosis of other solid cellular neoplasms of the pancreas (Figure 4, A through C). These features include the presence of discohesive cells, oval nuclei with nuclear grooves, hyaline globules, and degenerative modifications, such as cholesterol clefts, foamy macrophages, and multinucleated giant cells.^5–7,55,56  Myxoid changes may also occur in small biopsy specimens. All these aspects are suggestive and indicative of SPNs, but they should be further supported by nuclear immunohistochemical positivity for β-catenin (Figure 4, D), which is a diagnostic hallmark of this tumor type.^6,57,58  Other useful immunohistochemical markers include LEF1 (Figure 4, E), vimentin, CD10, and progesterone receptor.^6,59  Potential immunohistochemical pitfalls can be caused by the distinction between SPNs and PanNETs because synaptophysin and CD56 may be positive in SPNs (up to 40%–50% of cases) and CD200 is usually positive in both entities, with a similar prevalence.^6,55,60,61  Notably, it should be acknowledged that the neuroendocrine marker chromogranin-A is negative in SPNs, thus being very helpful in this differential diagnosis.^5,6,61  Along these lines, the lack of CAM5.2 is a useful support for excluding a PanNEN from the differential diagnosis, but caution should be given with this biomarker because it can be positive in at least 10% of SPNs. A last consideration regards β-catenin, which can be positive in up to 5% to 7% of ACCs, thus the differential diagnosis with SPN cannot be based only on such a marker.^47,62 

Neoplastic and nonneoplastic cystic lesions of the pancreas represent a challenging target not only for endoscopists but also for diagnostic pathologists dealing with small biopsies. Tissue fragments obtained from cystic lesions are often more fragmented than those obtained from solid nodules.⁷  Furthermore, in a small proportion of cases, the material used for diagnosis may be composed only of the cyst contents, including acellular debris and macrophages. Gastrointestinal tract contamination is another potential health concern. This can be particularly challenging because fragments of benign gastric or duodenal mucosa can strongly mimic fragments of a neoplastic mucinous cyst, such as low-grade intraductal papillary mucinous neoplasms (IPMNs) or mucinous cystic neoplasms (MCNs).^7,63  Based on the location, gastric contamination is a very important pitfall for lesions from the body and tail because the needle is most likely to traverse the stomach; for the same reason, duodenal contamination is more likely for lesions of the pancreatic head. Notably, if the biopsy shows the diffuse presence of mucinous epithelial fragments with complex architecture and high-grade dysplasia, the exclusion of contamination is easier. At the same time, such findings introduce additional issues for distinguishing high-grade dysplasia versus invasive adenocarcinoma. In these cases, a diagnosis of “neoplastic mucinous cyst with high-grade dysplasia or atypia” can be made, with a specific note suggesting invasive adenocarcinoma if abundant background necrosis is present.⁶⁴  Regarding mucinous cysts, the distinction of MCNs from IPMNs can be made based on 2 main histologic features: (1) papillary structures are typical of IPMN, whereas MCNs rarely show pseudopapillae; (2) ovarian-like stroma is a diagnostic hallmark of MCNs, although it is a difficult target for biopsy-based approaches.

Additional cystic lesions that can occasionally be sampled using small biopsies are the following: (1) serous cystadenomas: on small biopsies, epithelial cells with bland atypia and roundish centrally located nuclei in a background of clear cytoplasm are important diagnostic features, and the diagnosis can be supported with positive immunohistochemical staining for cytokeratin, MUC6, inhibin, CAIX, HIF-1α, and GLUT-1, among others⁶ ; (2) lymphoepithelial cysts: the combination of squamous epithelial lining with organized lymphoid tissue is an important diagnostic aspect, but there are no definitive immunohistochemical markers available to unequivocally support this diagnosis; (3) pseudocysts: they have no true epithelial lining, thus this is an exclusion-based diagnosis, that cannot be based on specific histologic findings.^7,65  Also based on such considerations, the management of patients affected by pseudocysts should be based on a multidisciplinary approach⁶⁶ ; and (4) pancreatic ACT: benign cyst lined by normal acinar epithelium—as already stated in the paragraph on ACC, the diagnosis of ACT in the preoperative setting is very important, potentially sparing patients from unnecessary surgical resections, at least in the case of asymptomatic lesions.^50–53 

For some cystic lesions, including MCNs, serous cystadenomas, and rare ACTs of the pancreas, it is important to note that a biopsy approach based on the use of Moray microforceps can provide a unique opportunity to obtain adequate tissue for diagnosis.^7,50,67,68  This further highlights the importance of adequate sampling of pancreatic lesions for diagnosis and the presence of a multidisciplinary team to tailor the best approach for each patient.

Regarding topical pearls for the diagnosis of cystic lesions, recent evidence has highlighted that molecular analysis of cyst fluids can be of great help in routine diagnostic workflow.^69–72  Specific genetic biomarkers truly represent the next level of diagnosis in this field, including PRKACA/B fusions for intraductal oncocytic papillary neoplasms, FGFR2 fusions for intraductal tubulopapillary neoplasms, VHL mutations for serous cystadenomas, and GNAS mutations for intestinal IPMNs.^70–72  Moreover, the combination of mitogen-activated protein kinase/GNAS and TP53/SMAD4/CTNNB1/mammalian target of rapamycin alterations showed high sensitivity (approximately 90%) and specificity (almost 100%) for the detection of advanced neoplasia.^70–72  The introduction of molecular analysis into the routine workflow and its integration with the morphology and IHC of biopsy/cytology samples represents a decisive step in the diagnosis of pancreatic cystic lesions.

A recent survey by the College of American Pathologists investigated the pancreaticobiliary cytology/fine-needle biopsy practice in domestic and international laboratories in 2021.⁷³  Its findings revealed significant differences among institution types and between domestic and international laboratories, above all regarding specimen volume/type, slide preparation, and rapid on-site evaluation (so-called ROSE) practice. It is evident that there is not a singular correct approach to be used. Critical issues in the realm of pancreatobiliary cytology/fine-needle biopsy include: (1) the importance of standardized procedures/guidelines to be followed in each laboratory; (2) to consider the possibility of preparing cell blocks, not only for the possibilities to perform ancillary analyses, such as IHC, but in general for maximizing the diagnostic potential of pathologists; (3) to have highly specialized pathologists, ideally with the knowledge for simultaneously approaching material from fine-needle aspiration and biopsy; and (4) to maximize the quantity of material, with the priority given to biopsy/cell blocks, also considering an aspect with growing importance in this context: the preservation of sufficient material for molecular investigations/next-generation sequencing.

Along those lines, it is important to acknowledge that cell block processing may alter the overall morphology, at least in some circumstances. An example is given by the so-called Azzopardi phenomenon, which can be observed more often in small biopsies and cell blocks than in fine-needle aspiration, in small-cell neuroendocrine carcinomas.⁷⁴  That said, in the pancreas but also in other organs, cell block preparations can give more advantages than limitations.⁷⁵  Interestingly, a recent investigation regarding cell block preparations pointed out that the most robust modality along these lines, also considering the preservation of morphology, is the thrombin clot method cell block.⁷⁶  In pancreatic pathology, an in-depth knowledge of the advantages and limitations of cell blocks represents an important point for diagnostic pathologists. Overall, choosing the best diagnostic strategy/workflow should be based on multiple factors, including a multidisciplinary approach to each individual patient in the context of personalized medicine.

Another consideration regarding real-world scenarios should be made regarding IHC. Indeed, it is important to highlight that the most important tool for the diagnostic activity is represented by morphology. Thus, IHC should be used only in particular/specific circumstances, and it should be integrated with morphology for improving the diagnostic reliability. Some of the most important situations that can require IHC are: (1) PDAC: difficulties in the differential diagnosis between infiltrative versus reactive glands (p53 and/or SMAD4 should be requested in all these cases); (2) solid cellular neoplasms: difficulties in (or support for) the differential diagnosis among NENs versus SPNs versus PBs versus ACCs; and (3) tumor origin: primary neoplasm versus metastasis.

The pancreatic gland is the metastatic site for other malignant primary tumors. Therefore, diagnostic pathologists should be aware of previous and/or simultaneous neoplastic diseases, including hereditary cancer syndromes, in all patients. The most common primary tumors that potentially metastasize to the pancreas include clear cell renal cell carcinoma and lung and breast cancers, although other tumors can also involve the pancreatic gland.^77,78  To overcome the potential pitfalls of metastasis versus pancreatic primaries, it is important to collect an adequate medical history and perform specific immunohistochemical staining to explore all potential extrapancreatic origins.

The pancreatic gland can also produce rare mesenchymal lesions. In such cases, the most convenient approach is based on the integration of clinical history, histologic examination, and a specific immunohistochemical panel, with a descriptive diagnosis ideally including the most common differential diagnoses. In the differential diagnosis of sarcomas, sarcomatoid carcinoma of the pancreas should be considered.^79,80  In a significant proportion of these cases, however, the final diagnosis can be made only on surgical specimens.