SEARCH

SEARCH BY CITATION

Keywords:

  • serous cystadenoma;
  • pancreas;
  • α-inhibin;
  • pancreatic cystic neoplasm;
  • inhibin;
  • pancreatic fine-needle aspiration

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

The diagnosis of serous cystadenoma (SCA), a rare benign pancreatic neoplasm, can alter the management of patients with pancreatic masses. Although characteristic imaging findings and fluid chemical analysis have been described, SCAs are not always recognized preoperatively. Furthermore, scant cellular yield on fine-needle aspiration (FNA) often leads to a nondiagnostic or nonspecific benign diagnosis. α-Inhibin (AI), a sensitive marker for SCA, is infrequently required for diagnosis in surgical specimens due to their characteristic histologic appearance. The objective of the current study was to determine whether AI staining can improve SCA diagnosis on FNA specimens.

METHODS

Fifteen confirmed cases of SCA with prior FNA specimens were selected for this study. FNAs were evaluated for cellularity, cellular arrangement, and cytomorphology. Resection specimens were reviewed.

RESULTS

Of the 15 FNA cases, approximately 75% demonstrated scant cellularity (11 of 15 cases). On smears, the cells were arranged as flat sheets, corresponding to strips of cells on cell block sections. The cells were small and round to cuboidal, with clear cytoplasm; occasional plasmacytoid cells and oncocytic cells were identified. Flattened cells, corresponding to attenuated epithelial cells lining macrocysts on the resections, were also noted. Stromal fragments were present in 5 FNAs and correlated with the hyalinized stroma in the resection specimens. AI immunostaining was positive in 88% of cases (7 of 8 of cases), thereby supporting the diagnosis of SCA.

CONCLUSIONS

The results of the current study indicate that low cellularity and bland cytology are inherent to SCAs. Performing cell blocks and AI staining on FNA specimens is useful for establishing the diagnosis of SCA. An immunohistochemical panel including AI, chromogranin, and synaptophysin may enhance the diagnostic accuracy of pancreatic FNA specimens. Cancer (Cancer Cytopathol) 2014;122:33–39. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Serous cystadenomas (SCAs) are rare benign neoplasms that account for approximately 1% of exocrine pancreatic neoplasms.[1] Due to their generally indolent clinical course, small and asymptomatic SCAs can be treated expectantly with radiological and clinical follow-up. Confirming the diagnosis prospectively with any or all of several available modalities, however, is challenging.

Although characteristic imaging findings and chemical analyses of cyst fluid can help to narrow the differential diagnoses, they are not entirely sensitive or specific, leaving a percentage of cases unrecognized. Similarly, fine-needle aspiration (FNA) not infrequently results in either a nondiagnostic or nonspecific benign diagnosis due to limited cellularity and/or an inability to appreciate and confirm that the bland cells aspirated are of “serous” origin. The numbers of such cases are likely to grow as the frequency of enhanced imaging modalities continues, resulting in the greater detection of pancreatic cysts.

Staining for α-inhibin (AI), a highly sensitive immunohistochemical marker for SCA,[2, 3] has been described on surgical pathology (SP) resection specimens, but it is infrequently applied in routine practice due to the characteristic histology of SCA. More importantly, to the best of our knowledge its applicability in cytology specimens has not been tested to date. The objective of the current study was to determine whether AI immunohistochemical staining performed on cell blocks aids in improving the diagnostic yield of FNA diagnosis by confirming that the cells are representative of SCA.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

A retrospective computerized search for SP resections diagnosed as SCAs and endoscopic ultrasound (EUS)-guided pancreatic FNAs with diagnoses of (or suggestive of) SCA was performed. Cases that had FNAs and corresponding SPs or a follow-up FNA were selected for the study. This study was approved by the Institutional Review Board of the study institution.

The FNAs were evaluated by 2 study pathologists for cellularity, architectural arrangement, cell cytomorphology (shape), and stroma. Available SP specimens were also reviewed for diagnosis, cytomorphology, and stromal tissue.

AI staining with appropriate controls was conducted both prospectively and retrospectively on cases with available cell blocks. After antigen retrieval in ULTRA Cell Conditioning Solution (Ventana Medical Systems, Tucson, Ariz), slides from all cases were stained on an autostainer (Ventana Benchmark Ultra; Ventana Medical Systems) for AI (clone R1, prediluted mouse monoclonal antibody; Cell Marque, Rocklin, Calif). A case was considered positive with strong staining demonstrated in > 95% of cells, negative in the absence of any immunoreactivity, and equivocal when positive and negative criteria were not fulfilled.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

We identified a total of 15 FNA specimens from 15 patients (9 women and 6 men with a median age of 79 years [range, 40 years-79 years]) with SP and/or clinical/FNA follow-up over 109 months. Thirteen patients underwent SP resections, 1 had a concurrent EUS-guided SP biopsy, and 1 underwent an FNA (13 weeks later) and had clinical follow-up of 54 weeks demonstrating no significant increase in the size of the lesion.

The original cytology diagnoses were “nondiagnostic” (7 cases); “negative for malignant cells, suggestive of SCA” (3 cases); “atypical” (2 cases); “atypical, cannot exclude endocrine neoplasm” (1 case); “suspicious for neoplasia, suggestive of SCA” (1 case); and “positive for neoplastic cells, suggestive of endocrine neoplasm” (1 case) (Table 1).

Table 1. Summary of Cases
Case No.Tumor Size, cmFNA DiagnosisSurgical DiagnosisCell Morphologies on FNAα-Inhibin Staining
  1. Abbreviations: -, negative; +, positive; endocrine, endocrine neoplasm; FNA, fine-needle aspiration; I, insufficient material; ; LC, round/ low cuboidal cells with clear/vacuolated cytoplasm typically characteristic of serous cystadenomas; NA, not available; O, oncocytic; Ov, ovoid; P, plasmacytoid; SCA, serous cystadenoma; St, stroma.

  2. a

    Size obtained from imaging studies.

  3. b

    Tumor not resected.

  4. c

    α-inhibin immunohistochemistry performed prospectively.

  5. d

    Biopsy specimen.

12.5NondiagnosticSCALC, ONA
21.5NondiagnosticSCA, oligocysticLC, P, OvNA
33NondiagnosticSCALC, O, Ov
49Positive for neoplasia, endocrineSCA, papillary featuresLC, Ov, St+
53.3AtypicalSCALC, StNA
61.5aNegative, suggestive of SCANAbLC, P, ONA
79.5Negative, suggestive of SCASCALC, P, O+
89.8NondiagnosticSCALC, O, St+
91AtypicalSCALC, Ov, St+
101.8aSuspicious, suggestive of SCAcSuggestive of SCAb, dLC, P+
111.1Atypical, cannot exclude endocrineSCALC, Ov, St+
123NondiagnosticSCALC, O, Ov+
132.2NondiagnosticSCA NA
142.2Negative, suggestive of SCASCALC, INA
153NondiagnosticSCA, papillary featuresLC, OvNA

Thirteen cases, including the cases diagnosed as “atypical, cannot exclude endocrine neoplasm” and “positive, suggestive of endocrine neoplasm,” had resection specimens that demonstrated SCA. Original FNA results of 2 cases without SP resections were “negative” and “suspicious for SCA,” respectively. The former had a negative FNA and clinical follow-up and the latter had a concurrent biopsy that was diagnosed as having features suggestive of SCA after negative immunohistochemical staining for endocrine markers.

Overall, the aspirates were not very cellular. One had no diagnostic cells, 10 had rare/scant cells, 3 had moderate numbers of cells, and 1 had relatively abundant cells. When present, the cells were arranged as 2-dimensional sheets, but 2 cases (those that were suspected to be endocrine neoplasms) had cellular overlapping. In addition to the round/low cuboidal cells with clear/vacuolated cytoplasm typically characteristic of SCAs, 3 additional cell types were noted (Table 1). Four of the 15 FNA specimens had plasmacytoid cells, including the case designated as“atypical, cannot exclude endocrine neoplasm.”. Oncocytic-appearing cells (6 cases; only rare cells were identified in 2 of these 6 cases) were noted; there were many such cells in the FNA reported as “atypical, cannot exclude endocrine neoplasm” (Table 2). Ovoid cells were evident in 7 cases.

Table 2. Features of FNA Specimens With Diagnoses Suggestive of Endocrine Neoplasms
FeaturesAtypicalPositive
  1. Abbreviations: Atypical, atypical, cannot exclude endocrine neoplasm; FNA, fine-needle aspiration; Positive, positive for neoplastic cells, suggestive of endocrine neoplasm.

CellularityAbundantScant
Plasmacytoid cellsRareAbsent
Oncocytic cellsManyAbsent
OverlappingPresentPresent
Nuclear groovesPresentPresent

Cell blocks were prepared in 13 of the 15 cases. Epithelial cells were arranged as strips/single-file cords on 7 cell blocks (Fig. 1A) and hyalinized eosinophilic stromal tissue was present in 5 cases (Fig. 2). Of the 13 cases, 8 had cells and were available for the performance of AI immunohistochemical staining prospectively (1 case) or retrospectively (7 cases); 7 cases (88%) were immunoreactive and 1 was negative (Figs. 1B and 2C). Synaptophysin and chromogranin stains were performed in 4 cases and were positive in 1 case. A review of the corresponding SP in this positive case demonstrated that the lining serous cells were negative for these stains. Moreover, we identified numerous islets of Langerhans entrapped between and around the serous cysts, suggesting that the islet cells might have been sampled in the FNA. Indeed, a review of the cell block indicated that the scant chromogranin- and synaptophysin-positive cells did not overlap with the AI-positive cells.

image

Figure 1. Cytologic features of pancreatic serous cystadenomas are shown. (A) A cell block preparation of a typical pancreatic serous cystadenoma is shown demonstrating benign-appearing, low cuboidal epithelial cells arranged in strips (H & E, × 400). (B) Inhibin immunohistochemical staining was found to be strongly positive in the cytoplasm of these cells (× 400).

Download figure to PowerPoint

image

Figure 2. Cell block preparation of a pancreatic serous cystadenoma is shown demonstrating (A) strips of bland-appearing epithelial cells and (B) hyalinized eosinophilic stromal tissue (arrow) (H & E, × 400). (C) Inhibin immunohistochemical staining highlighted the epithelial cells (× 400). (D) The surgical resection specimen confirmed prominent hyalinized stroma in this serous cystadenoma specimen (H & E, × 100).

Download figure to PowerPoint

Diagnoses of the 13 SP resections were microcystic SCA in 10 cases, oligocystic SCA in 1 case, and microcystic SCA with papillary features in 2 cases. The FNA diagnosis on one of the latter cases was “positive for neoplastic cells, suggestive of endocrine neoplasm”; the aspirate contained few clusters of overlapping nuclei with grooves. Oncocytic cells were readily observed in only this SP resection. The SP also demonstrated cysts and papillae lined by clear cuboidal cells with uniform round nuclei (Fig. 3). In addition, small ovoid cells identified on FNA specimens corresponded to the attenuated epithelium lining the cysts on the resection specimens (Fig. 4). Of the 7 cases found to have ovoid cells on FNA, 5 had similar cells on SP resections, 1 SP specimen was unavailable for rereview, and such cells were not readily identified in 1 specimen. Hyalinized tissue between the cysts corresponded to stroma that was evident on the cell blocks.

image

Figure 3. This example of pancreatic serous cystadenoma demonstrated (A) numerous overlapping cells and nuclear grooves in the aspirate specimen, originally interpreted as being suggestive of an endocrine neoplasm (Romanowsky stain, × 600). The resection specimen shows (B) cysts and papillary structures lined by (C) cuboidal cells with uniform round nuclei and focal intracellular globules (arrows) (B: H & E, × 100; C: H & E, × 200).

Download figure to PowerPoint

image

Figure 4. The fine-needle aspiration smear preparation of this pancreatic serous cystadenoma demonstrated (A) flattened cells (Romanowsky stain, × 400). (B) The corresponding surgical resection specimen demonstrated cysts lined by cuboidal and attenuated epithelial cells (arrowhead) (H & E, × 200).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

With greater and improved imaging techniques, increasing numbers of pancreatic cysts are being encountered, often incidentally during evaluation for unrelated symptoms.[4] Pancreatic cysts, including SCAs, mucinous cystic neoplasms, and intraductal papillary mucinous neoplasms, have been reported to have been identified in 2.6% of patients who have undergone a computed tomography scan.[5] SCAs account for approximately 30% of pancreatic cystic neoplasms[6] and 1% of all nonendocrine pancreatic neoplasms.[1] Generally, neoplasms of serous origin are benign and approximately one-half are asymptomatic[7]; to the best of our knowledge, only rare cases of serous cystadenocarcinomas have been reported to date.[8-12]

Therefore, conservative management, with a few exceptions, has been suggested for patients with SCAs[6] because surgical intervention can result in morbidity, mortality, and loss of pancreatic tissue.[13] Proposed indications for the surgical intervention of SCAs include a symptomatic mass measuring > 4 cm in size, rapid growth,[14] uncertainty about the SCA diagnosis, and lesions in the pancreatic body/tail.[6]

Although SCAs can be followed conservatively, a prospective diagnosis is not always attainable.[7] Accurate radiographic recognition of SCAs ranges from 27%[15] to 78%,[7, 16, 17] and the rate for achieving a diagnosis with EUS is also low.[6] Evaluation of cyst fluid is helpful but not entirely sensitive or specific.

FNAs offer the opportunity to assess the cells but, similar to the above-mentioned modalities, the diagnostic yield is low.[7, 18, 19] In the current series, a prospective diagnosis of SCA was suggested in 4 of 15 cases (27%), but the majority were deemed nondiagnostic. Similar (7 of 28 cases; 25%),[19] lower (1 of 21 cases; 1%),[7] and higher (10 of 11 cases; 91%)[20] rates have been reported.

The low cellular yield of SCA FNA specimens appears to be intrinsic to this entity.[19] It is a reflection on their histological appearance: the area of the SCA is predominantly cystic with a low percentage of epithelial cells relative to the size of the mass, a finding that is accentuated further in oligocystic SCAs. In addition to their scant cellularity, a lack of experience with recognizing this lesion on FNA may contribute to nondiagnostic and false-negative diagnoses.[6] Differences in sampling techniques (EUS-guided vs transabdominal computed tomography-guided), operators, and thoroughness of correlation with the imaging and clinical findings may also account for the variability noted.[6]

When present, aspirates are reported to consist mostly of monolayered sheets of round-to-low cuboidal cells with clear/vacuolated glycogenated nonmucinous cytoplasm, typically described for SCAs.[19] Focally, overlapping was noted, a finding that has been described in the literature.[19] In addition, plasmacytoid cells, cells with granular/oncocytic features, and ovoid cells also were noted. Features not present in the current study but described by others include papillary clusters (corroborated on histology),[8, 19, 21] mild nuclear atypia[8] consisting of nuclear enlargement, pleomorphism, hyperchromasia, and columnar cells suggestive of mucin-producing cells.[19]

We noted stromal tissue, mostly on cell blocks, in 5 of the 15 cases in the current study (33%). Collagenous stromal fragments with attached neoplastic cells were noted in 50% of SCA aspirates (14 of 28 cases) evaluated by Huang et al.[19] These correlate with the stroma separating the cysts and sometimes resulting in a central stellate scar noted on SP resections.

The results of the current study highlight the importance of obtaining sufficient material for a cell block preparation. Cell blocks were prepared in 13 cases, and 8 had material available for further immunohistochemical staining. Although the usefulness of cell blocks has been recognized previously,[6] experience with cell block processing for diluted specimens, such as those from SCAs, is variable; some studies have described insufficient material and others have reported cell blocks in a subset of cases (10 of 28 cases; 36%), with the majority having rare clusters of tumor cells.[19, 22] Typically, cell blocks are used to confirm the presence of glycogen with periodic acid-Schiff and periodic acid-Schiff with diastase. The majority of cell blocks in the current study had scant cells, but staining for AI was readily interpretable and positive in 88% of cases (7 of 8 cases), a figure similar to that reported by Kosmahl et al,[2] who noted staining in 82% of SP SCA resections (31 of 38 resections). They postulated that the rate may be even higher because paraffin blocks aged > 10 years did not stain. An even greater number of SCAs with AI staining was reported by Marsh et al[3] (26 of 27 cases; 96.3%).

Distinguishing SCAs from other relatively bland-appearing entities including solid pseudopapillary neoplasms, low-grade mucinous cystic neoplasms (MCNs), acinar cell carcinomas, incidental gastrointestinal contamination, and neuroendocrine tumors can be challenging in scant FNA specimens, but AI can be useful in differentiating them. Unlike SCAs, solid pseudopapillary neoplasms were reportedly nonimmunoreactive for AI in all 59 cases stained in the study by Kosmahl et al,[23] but they demonstrated β-catenin reactivity in 95% to 100% of cases in the study by Tanaka et al.[24] Differentiating SCA from a low-grade MCN can be difficult in the presence of atypical SCA cellular features (eg, columnar cells, abundant cytoplasm). In addition to chemical analysis of cystic fluid, AI staining may prove helpful. When present in MCNs, AI stains the ovarian stromal cells, rather than the epithelial cells as observed in SCAs. In addition, the MCN stromal cells are positive for progesterone and estrogen receptor immunohistochemistry and negative in SCA.[25] A subset of acinar cell carcinomas can reportedly stain with AI.[2] However, cells of acinar cell carcinomas have a characteristic large, central nucleolus, which is absent in SCAs. Gastrointestinal epithelium contamination by bland-appearing epithelial cells and mucin also may be a source of diagnostic pitfalls,[19] but intestinal cells can be recognized by goblet cells and gastric epithelial cells appear to be negative for AI.[26]

Because 2 of the cases in the current study were suggestive of endocrine neoplasms, we performed AI staining on tissue microarrays of pancreatic endocrine neoplasms. In our experience with these tissue microarrays, 11.9% of neuroendocrine tumors stained positive for AI (unpublished data). Marsh et al described a comparable rate of staining on a tissue microarray of endocrine tumors (2 of 49 cases; 4.1%).[3] In the current series, 2 SCAs were misinterpreted and suspected to represent an endocrine neoplasm. The relatively abundant cellularity and reactivity for synaptophysin and chromogranin in islet cells may have resulted in this misinterpretation in one case. The possible presence of entrapped benign islet cells should be considered when interpreting neuroendocrine markers. It is interesting to note that none of SCAs reported in 2 large case series stained with chromogranin.[2, 3] In the second case, monomorphism, loose cohesiveness, overlapping, and grooves may have prompted the suggestion of an endocrine neoplasm. Intriguingly, the SP resection demonstrated SCA with a papillary pattern. This histological pattern has been previously reported as a diagnostic pitfall on FNA.[21] Similar to our case, it also demonstrated crowding and rare grooves, but unlike ours, it showed papillary fronds and was diagnosed as a solid pseudopapillary neoplasm.

In scant FNA specimens, the presence of unusual cytomorphology may raise suspicion of a neuroendocrine tumor. Therefore, even in cases with negative immunohistochemistry for chromogranin and synaptophysin, the pathologist may be inclined to render a “nondiagnostic” result in view of scant cellularity.

We strongly believe that an immunohistochemical panel including AI, chromogranin, and synaptophysin may enhance the diagnostic accuracy of pancreatic FNA specimens. Recognition of possible pitfalls is vital when interpreting these problematic cases. In these cases and those with scant cells, AI staining can be used to identify and at least suggest a diagnosis of SCA in the appropriate clinical and radiological setting. Because the staining is strong and clean, AI can be used even in sparsely cellular specimens. Although it was not performed in the current study, another immunohistochemical stain, calponin, has also been found to be sensitive and specific in differentiating SCAs from pancreatic adenocarcinomas and pancreatic endocrine neoplasms.[2] The expected scarcity of SCAs and the difficulties in obtaining a cell block sample limited the number of cases in the current study. Despite the sample size limitation, we believe this study achieved its aims. Moreover, our recent experience has demonstrated that several techniques can facilitate cell block preparation. For example, gentle suction may be applied to create a bloody aspirate, which can be allowed to clot before placement in 10% formalin.

In conclusion, a diagnosis of SCA can significantly alter patient management. The results of the current study indicate that low cellularity and bland cytology are inherent to SCAs, and a diagnosis of SCA on FNA can be challenging. In other instances, the presence of atypia or other cell types can lead to erroneous diagnoses suggestive of a pancreatic endocrine neoplasm. In these cases, we recommend performing AI immunohistochemistry on cell blocks to help establish the diagnosis of SCA. AI positivity on cytology specimens, even those with sparse cells, is useful and highly indicative of SCA, particularly when neuroendocrine markers are negative.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Dr. Poneros acts as a consultant for Boston Scientific. Dr. Sethi acts as a consultant for Boston Scientific and Xlumera and has received travel expenses for unrelated meetings from Beacon Endoscope and Mauna Kea Technologies. Dr. Saqi has provisional patents filed and a patent pending.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES