Pancreatic endocrine neoplasm (PEN) and solid pseudopapillary neoplasm of the pancreas (SPN) frequently pose diagnostic challenges. We sought to determine which markers could provide the best immunophenotypic characterization of PEN and SPN, allowing separation on limited cytology samples. We retrieved 22 resected PEN (n = 12) and SPN (n = 10) tumors to serve as a training set for the performance of extensive immunohistochemical staining. Based on these results, we selected a subset of antibodies for application to 25 fine-needle aspiration (FNA) samples from PEN (n = 16) and SPN (n = 9).
Chromogranin A, synaptophysin, CD56, and progesterone receptor (PR) highlighted PEN cases in the training set; E-cadherin was noted in a membranous pattern. SPN cases were most immunoreactive for α1-antitrypsin, vimentin, CD10, and PR, with nuclear staining for β-catenin; E-cadherin did not show a membranous pattern. Among all FNA samples tested, the immunohistochemical staining of E-cadherin (P = .0003), β-catenin (P = .00004), and CD10 (P = .00006) demonstrated the greatest difference between PEN and SPN. The pattern of E-cadherin/β-catenin expression was highly specific for distinguishing PEN from SPN. On limited FNA samples, the characteristic expression of E-cadherin/β-catenin and the expression of CD10 can be used to distinguish PEN from SPN.
The origin of solid pseudopapillary neoplasm of the pancreas (SPN) remains largely unknown. This tumor has a predilection for young females and is usually located in the head or tail of the pancreas.1–4 Abdominal pain is the most common presenting clinical symptom,1 with other signs and symptoms including nausea, vomiting, fever, weight loss, and jaundice. This tumor is often found incidentally via imaging studies. Radiologically, this tumor exhibits solid and cystic areas, often with areas of hemorrhage.5 SPN can have a radiologic appearance and clinical features that overlap with pancreatic pseudocyst and radiologic/cytologic appearances that overlap with other pancreatic neoplasms, including acinar cell carcinoma, mucinous neoplasms, and pancreatic endocrine neoplasm (PEN).6–9 It is, therefore, important to establish a preoperative diagnosis of SPN because these tumors exhibit a different biologic behavior, and their management protocols may differ accordingly.
Fine-needle aspiration (FNA) has been proven as a useful method for diagnosing pancreatic lesions.10,11 The cytomorphologic features of SPN and PEN have been described.8,12–16 However, their distinction in cytologic samples can be challenging owing to overlapping diagnostic features.12 In such circumstances, ancillary stains can provide much needed supportive evidence. Many studies have used a large number of stains to document an immunohistochemical profile associated with SPN and PEN.4,17–19 Review of these immunophenotypic patterns shows that SPN and PEN can have overlapping immunohistochemical staining patterns. Specifically, SPN has been noted to demonstrate neuroendocrine differentiation on immunohistochemical stains and on electron microscopic studies.20–23 Thus, relying only on evidence of neuroendocrine differentiation on cytologic samples can lead to diagnostic errors. To add to this dilemma, often FNA samples do not have enough cellularity to perform a large battery of immunohistochemical stains. Therefore, antibodies need to be chosen judiciously to best use limited cytologic material to distinguish among these morphologic mimickers.
This study was undertaken with an objective to identify the minimum number of immunohistochemical “stains” that can best distinguish SPN from PEN on FNA samples.
Materials and Methods
In an effort to determine the ideal number of antibodies best suited to distinguish SPN from PEN with a limited amount of clinical material, we used a 2-step approach. We used resected sections to perform immunohistochemical analyses based on the published literature and our own experience. Following that, we used a smaller panel of reagents on cytologic samples to determine those that could best distinguish SPN from PEN.
The training set included 22 resected pancreatic tumors (10 SPN and 12 PEN) from our surgical pathology files. The demographic profile, tumor location, and tumor size for each were recorded. A panel of antibodies was chosen based on literature review and applied to these cases Table 1. The immunostaining results were considered positive when greater than 10% of tumor cells were immunoreactive in the resected tissues. The immunophenotype was recorded for each tumor type, and from this profile, a limited panel of antibodies was selected to be applied to the test cases (endoscopic ultrasound–guided [EUS]-FNA samples).
We retrieved 25 EUS-FNA samples from patients with SPN (n = 9) and patients with PEN (n = 16). These tumors were diagnosed based on morphologic characteristics and immunohistochemical staining pattern, as has been previously reported.8,11–13 A demographic profile and the tumor location and size were recorded for each case. On-site assessment was performed on air-dried rapid Romanowsky stains, and additional sample was collected for preparing paraffin-embedded cell blocks. In these cases, a limited panel of immunohistochemical stains (Table 1) was performed. This panel was chosen based on the profiles obtained on resected tissue samples. The cases were categorized as positive or negative.
A 2-tailed Fisher exact probability test was used to determine if significant differences existed in immunohistochemical expression patterns of PEN and SPN. Furthermore, operating characteristics were calculated to determine performance of each antibody to arrive at a diagnosis of SPN or PEN based on their staining patterns.
Demographics and Tumor Characteristics of the Training Set
Patients with resected SPN were predominantly young females (mean age, 34 years) with most tumors (6/10 [60%]) located in the body/tail of the pancreas Table 2. These demographics are similar to those for patients in the test set (Table 2). In comparison, patients with resected PEN tumors were slightly older (mean age, 49 years), with no specific sex bias. PEN lesions were also noted more frequently in the body/tail of the pancreas (Table 2). The mean tumor size of SPN was larger in resected tissues than in those aspirated by EUS-guided FNA (Table 2).
Immunophenotyping of Specimens in Training Set
Extensive immunophenotyping performed on the resected tissue sections demonstrated that PENs were immunoreactive for chromogranin A (16/16 [100%]), synaptophysin (16/16 [100%]), and CD56 (16/16 [100%]). These tumors also demonstrated immunoreactivity for E-cadherin in a membranous pattern of immunoreactivity (16/16 [100%]). Progesterone receptor (PR) demonstrated positive immunoreactivity in PEN (14/16 [88%]). In comparison, immunoreactivity for CD10 was noted in only 31% (5/16) of cases with PEN. In contrast, all SPN tumors (100%) were immunoreactive for CD10 and CD56. Simultaneously, none of the tumors demonstrated membranous immunoreactivity for E-cadherin in the resected SPN tumors. All 4 SPN tumors that were stained for α1-antitrypsin and vimentin were positive. Four SPN cases were positive for synaptophysin (4/9 [44%]). None of the SPN tumors were positive for chromogranin, HMB-45, CD117 (c-kit), or inhibin. Pancytokeratin stain performed on a subset of these tumors demonstrated positive staining in only 2 (33%) of 6 cases. Review of the literature shows that pancytokeratin may be noted in up to 52% of SPNs.17
Test Set Results
Of 9 SPN FNA cases included in the test set, 7 were initially classified as SPN at the time of diagnosis. Supportive immunohistochemical stains were not required in only 1 case. Of these 7 cases, 1 was reclassified as SPN (from the original impression of PEN) after immunohistochemical staining was performed. Two test set SPN cases were originally diagnosed as PEN at FNA, both with positive chromogranin and synaptophysin immunostains (one with only focal synaptophysin staining). Subsequent histologic examination of the tumor resections provided adequate material and features to amend the diagnosis. In addition, all 16 PEN FNA cases in the test set were originally diagnosed correctly at the time of FNA, with the majority having supporting immunohistochemical data.
Training set results were used in constructing an immunohistochemical panel for the test set. Lack of adequate material in the cytologic sample prevented the performance of immunophenotyping on 1 SPN case. In addition, E-cadherin immunohistochemical analysis could not be performed on an additional 2 SPN cases, and material for β-catenin staining was available in 5 cases.
Our results on FNA samples demonstrate that chromogranin A, CD56, and synaptophysin were consistently expressed in PEN cases Table 3. E-cadherin was noted in 14 of 16 PEN cases in a membranous pattern. E-cadherin expression in SPN cases was noted in the cytoplasm and nucleus of the neoplastic cells but never in a membranous pattern. Chromogranin was expressed in 15 (94%) of 16 PEN cases, but was also expressed in 3 (38%) of 8 SPN cases. All PEN cases stained with β-catenin showed cytoplasmic or membranous staining. In comparison, all SPN cases showed a strong nuclear staining pattern, with 3 cases having concurrent cytoplasmic staining, although less pronounced than the nuclear reactivity. The SPN cases that were positive for PR demonstrated a strong nuclear pattern. However, faint dot-like nuclear immunoreactivity for PR was also noted in the FNA samples of a few PEN cases. Our results show that expression of E-cadherin/β-catenin, CD10, and PR is significantly different between SPN and PEN cases (Table 3).
Similar to previous reports, our study shows that SPN is almost always noted in young women and in the body/tail of the pancreas. However, as highlighted in this series, we also show that this tumor may be noted at both extremes of age and in males and is increasingly noted in the head of the pancreas (Table 2). In a recent study by Audard et al,24 11 of 28 cases were located in the head of the pancreas.
Most reports highlight that cellular aspirates with papillary groups, metachromatic matrix material, cytoplasmic globules, and characteristic nuclear features such as round to oval eccentrically placed nuclei with fine granular chromatin, when identified, are virtually diagnostic of SPN on FNA samples Image 1 and Image 2.7,12,25 SPN shares some cytologic characteristics with PEN, which include many single cells and occasional cells with a plasmacytoid appearance. Overall, PEN exhibits bland, uniform cells with finely stippled (“salt-and-pepper”) chromatin Image 3. PEN may show cytoplasmic neurosecretory granules and a pseudorosette pattern of growth Image 4. It is recognized that the diagnosis of SPN remains challenging, especially on cytologic samples. In an earlier study, it was noted that 3 of 5 such cases were not initially recognized as SPN.12 Jhala et al26,27 suggested that when overlapping features are noted between PEN and SPN on FNA samples, cytoplasmic vacuoles can serve as a clue to diagnosis of SPN. It therefore appears that immunophenotyping may aid in distinguishing SPN from PEN on challenging cytologic cases.
More recently, E-cadherin and β-catenin, members of the Wnt signaling pathway, have been searched for in SPN and its mimickers. Literature shows that the expected β-catenin staining in SPN tumors is a nuclear pattern.18,19,28 Similar to these reports, our results also show nuclear expression of β-catenin in all 5 of the FNA SPN samples tested.
In our resected samples, 12 (100%) of 12 PEN cases demonstrated a membranous expression pattern of immunoreactivity for E-cadherin Image 5. Similar to our results, a recent report from El-Bahrawy et al29 showed membranous expression of E-cadherin in 14 (93%) of 15 cases of neuroendocrine tumors. It has also been suggested that a lack of membranous expression by E-cadherin antibodies has a role in the pathogenesis of SPN.19 This lack of membranous expression pattern has been independently validated in subsequent reports of SPN.28–30 In keeping with these reports, our results also show in all cases of SPN that the tumor cells did not demonstrate a membranous expression pattern but instead demonstrated cytoplasmic, nuclear, or, more typically, no expression Image 6.
E-cadherin is known to have 3 cellular domains: extracellular, transmembranous, and intracytoplasmic. As a result, varying patterns of expression of E-cadherin have been recorded for SPN cases in the literature. E-cadherin expression may be noted in the nucleus or as a granular pattern in the cytoplasm or may not be highlighted at all.31,32 In their report of 20 SPN cases, Chetty and Serra31 demonstrated that E-cadherin could be identified in the nucleus or was not present at all. These reported differences in expression patterns in SPN cases were attributed to differences in the commercially available antibody clone. Regardless, SPN cases, unlike PEN, do not show a membranous staining pattern.
Although these results were noted on resected samples, our report demonstrates for the first time that E-cadherin may be differentially expressed on PEN and SPN cytologic specimens. Here, E-cadherin was consistently noted in a membranous expression pattern in PEN cases, and these results were highly specific. In contrast, a lack of membranous staining was noted in FNA samples derived from SPNs. Furthermore, E-cadherin expression was not seen in membranous expression in 2 of the 3 cases that also demonstrated chromogranin A expression in the SPN FNA samples. In a third case, subsequent resection confirmed the diagnosis of SPN. Therefore, E-cadherin seems to be used as a “stain of choice,” especially on limited FNA samples in which the morphologic features show overlapping features between SPN and PEN.
Earlier studies had consistently demonstrated that chromogranin was a reliable marker to support the diagnosis of PEN Image 7 because SPN was usually not immunoreactive for chromogranin.17 Our training set cases also demonstrated this difference, with only 2 (20%) of 10 cases of SPN showing focal expression of chromogranin. However, chromogranin positivity was noted in 3 of 9 FNA samples from SPN Image 8. In these 3 cases, E-cadherin staining did not reveal membranous expression, similar to what would be expected in PEN. This lack of a membranous pattern further reaffirms our contention that E-cadherin may serve as a valuable marker to differentiate PEN from SPN, especially on FNA samples. Earlier it was documented that chromogranin could be expressed in some SPNs.22 In their recent report, Chetty and Serra31 documented in their case selection that all 20 cases of SPN were immunoreactive for chromogranin. In this same report, they also documented lack of membranous expression of E-cadherin in SPN cases. These data, taken together, suggest that chromogranin alone should not be used as a sole marker to support or rule out PEN or SPN on FNA samples.
We used 2 additional neuroendocrine markers on FNA samples to determine if they could help support the diagnosis of PEN and rule out SPN on limited FNA samples. Similar to what was noted on our resected samples and what has been reported in the literature, synaptophysin did not serve as a reliable marker to distinguish PEN from SPN Table 4. CD56, a neural cell adhesion molecule, is also known as a marker of neuroendocrine differentiation. In our study, CD56 was expressed in 7 (44%) of 16 cases of PEN on FNA samples. This marker was also expressed in all cases of SPN, making it an unreliable marker to distinguish PEN from SPN. This finding is in keeping with an earlier report by Notohara et al,23 who also demonstrated expression of CD56 in PEN (18/20 [90%]) and SPN (19/19 [100%]) cases. In this study, however, the pattern of expression was different. The majority of PEN cases were weakly immunoreactive for CD56, and they stained in a cytoplasmic pattern. In comparison, CD56 was more strongly expressed in SPN and was predominantly noted in a membranous pattern. CD56 expression in the absence of concomitant lymphoid markers has been noted in poorly differentiated neuroendocrine carcinomas, including small cell carcinoma, Merkel cell carcinoma, and Ewing sarcoma on cytologic samples.33,34 In this study, we show that CD56 can be expressed in PEN and in SPN.
It has been shown that 80% of SPN cases demonstrate PR immunoreactivity.17 In this expanded study, we achieved similar results Image 9. It is important to note that we also demonstrated PR expression in our resected specimens of PEN. When extended to FNA samples, we could not demonstrate a similar result. Instead, there was a significant difference in the expression of PR between SPN and PEN cases. It was noted that the PR stain was overwhelmingly “positive” in the SPN cases. In our study, faint dot-like nuclear immunoreactivity was noted in the FNA samples of a few PEN cases Image 10. Thus, although our results suggest that PR expression supports a diagnosis of SPN, it should not be used in isolation.
Our findings further show that CD10 is expressed in 100% of SPN cases but in only 30% of PEN cases and was focal in its expression pattern. Other investigators have also detected CD10 expression in resected SPN tumors.23,30 Kim et al30 found this marker to be expressed in 83% of their SPN tumors. In comparison, its expression was noted in only 10% of pancreatic carcinomas and 11% of resected PEN cases. In the present study, FNA samples from all cases of SPN demonstrated CD10 expression Image 11 and, similar to what was noted by Kim et al,30 was seen to be expressed in 2 (13%) of 16 cases of PEN Image 12. This difference was statistically significant.
We show that the use of ancillary studies could be extremely helpful to differentiate SPN and PEN in FNA specimens. While chromogranin has been used for a long time to support the diagnosis of PEN, it should be recognized that SPN may occasionally demonstrate immunoreactivity for chromogranin. Our results further suggest that the lack of membranous staining with E-cadherin antibodies and positive nuclear staining with β-catenin by immunohistochemical analysis further support the diagnosis of SPN (specificity, 100%), whereas strong membranous expression supports the diagnosis of PEN (specificity, 100%). Thus, pattern recognition of a single stain can serve as one powerful marker on limited cytologic samples to help distinguish PEN from SPN. In cytologic cases in which additional stains can be performed, use of CD10, chromogranin A, and PR will further help establish the diagnosis on FNA samples.
Upon completion of this activity you will be able to:
describe the characteristic cytologic features of solid pseudopapillary neoplasm of the pancreas (SPN) and its differential diagnosis.
define the demographic profile in which SPN is most often identified.
list possible methods for obtaining tissue samples from a pancreatic mass.
The ASCP is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The ASCP designates this educational activity for a maximum of 1 AMA PRA Category 1 Credit™ per article. This activity qualifies as an American Board of Pathology Maintenance of Certification Part II Self-Assessment Module.
The authors of this article and the planning committee members and staff have no relevant financial relationships with commercial interests to disclose.
Questions appear on p 972. Exam is located at www.ascp.org/ajcpcme.
- fine needle biopsy
- tissue membrane
- needle device
- neural cell adhesion molecules
- receptors, progesterone
- islet cell tumor
- chromogranin a
- solid pseudopapillary tumor of the pancreas
- endoscopic ultrasound