Fax: (617) 726-7474
Cystic pancreatic endocrine tumor
A variant commonly confused with cystic adenocarcinoma
Version of Record online: 15 DEC 2006
Copyright © 2006 American Cancer Society
Volume 111, Issue 1, pages 47–53, 25 February 2007
How to Cite
Deshpande, V. and Lauwers, G. Y. (2007), Cystic pancreatic endocrine tumor. Cancer, 111: 47–53. doi: 10.1002/cncr.22422
- Issue online: 15 FEB 2007
- Version of Record online: 15 DEC 2006
- Manuscript Accepted: 25 SEP 2006
- Manuscript Revised: 17 SEP 2006
- Manuscript Received: 5 JUL 2006
- pancreatic endocrine tumor;
- cystic pancreatic neoplasm;
- fine-needle aspiration biopsy;
- endoscopic ultrasound;
Most cystic pancreatic neoplasms are currently evaluated by an endoscopic ultrasound-guided fine-needle aspiration biopsy (FNAB). In the authors' experience, FNAB of cystic pancreatic endocrine tumors (PETs) frequently causes diagnostic difficulties, partly because of unexpected overlapping features with cystic ductal adenocarcinomas.
The authors identified 5 histologically confirmed cystic PETs that were evaluated by FNAB and compared their cytomorphologic features to cystic ductal adenocarcinomas (n = 5) and solid PETs (n = 39) of the pancreas.
Cytologically, 2 of the aspirates of cystic PETs were devoid of tumor cells whereas the other 3 were variably cellular and composed of cohesive aggregates of monomorphic cellular elements with variably coarse chromatin. Tumor necrosis and nuclear membrane irregularities were not identified in cystic PETs. Alternatively, in contrast to PETs, cystic ductal adenocarcinomas were characterized by nuclear pleomorphism, nuclear membrane irregularities, and tumor necrosis.
Given the clinical implications, awareness of cystic PETs and their deceptive cytologic features would assist in distinguishing these lesions from cystic ductal adenocarcinomas. Cancer (Cancer Cytopathol) 2007 © 2006 American Cancer Society.
Common cystic pancreatic lesions encountered in clinical practice include inflammatory pseudocysts, intraductal papillary mucinous neoplasms, mucinous cystic neoplasms, and serous cystadenoma.1
The recognition of cystic pancreatic neoplasms has increased in the last decade, owing in part to improved abdominal imaging.2, 3 Concurrently, the improvement in endoscopic techniques has allowed nonsurgical sampling and evaluation by endoscopic, ultrasound-guided fine-needle aspiration biopsy (EUS-guided FNAB).
In addition to cystic pancreatic neoplasms, primarily solid pancreatic tumors such as pancreatic endocrine tumors (PETs), acinar cell carcinoma, and pancreatic ductal adenocarcinoma can occasionally undergo cystification and, thus, enter the differential diagnosis of cystic pancreatic neoplasms.4 The fraction of PETs with cystification is limited to <2%.5 Naturally, although the cytomorphologic features of solid PETs have been well characterized, those of their cystic counterparts have received little attention. Although, in our experience, cystic PETs may cause diagnostic challenges, only 1 report of FNAB findings of this particular subtype had been published.6 We review herein our experience with 5 cases of cystic PETs and contrast the cytomorphologic features of these cases with those of solid PETs. Subsequently, we highlight diagnostic pitfalls and differential diagnosis of these cystic PETs.
MATERIALS AND METHODS
We identified 5 EUS-guided FNABs from pancreatic cystic neoplasms that were eventually diagnosed as cystic PET. In each case, gross and histologic examination of the resection specimens confirmed the cystification of the tumor. In addition, we compared these cases with 5 cystic ductal adenocarcinomas on which a liquid-based preparation was available.
Endoscopic ultrasound (EUS), and computed tomography (CT) findings were recorded when available. Results of cyst fluid analysis, when available, were also reviewed. At this institution, cyst fluid analysis, including concentrations of carcinoembryonic antigen and amylase, is performed on all pancreatic cystic lesions that yield sufficient fluid.1 Cyst fluid concentrations of carcinoembryonic antigen and amylase were performed, as previously reported.1
Papanicolaou-stained slides were available on all cases. ThinPrep and cytospin preparation techniques were used on 2 cases each. One case consisted entirely of direct smears. Cellularity, presence of single cells and cell clusters, plasmacytoid and monomorphic cell populations, chromatin pattern, and presence of nucleoli, as well as necrosis were systematically evaluated. The presence and absence of gastrointestinal contamination and background inflammatory cells were also recorded. A rapid onsite interpretation was not performed on any of these cases. As a comparative measure, we also reviewed 39 solid PETs that underwent a EUS-guided FNAB and a cohort of morphologically confirmed cystic ductal adenocarcinomas7 on which a cytospin and/or ThinPrep preparation was available. This study has been approved by the Massachusetts General Hospital institutional review board.
Demographic, Clinical, and Radiologic Information (Table 1)
The patients were all men and showed no evidence of multiple endocrine neoplasia, type 1. The mean age was 62.8 years (range, 47 to 74 years). All 5 PETs were clinically nonfunctional. Based on EUS and CT scans, the pancreatic lesions were uniloculated (n = 2) or multiseptated (n = 2) and ranged from 2 cm to 5 cm in greatest diameter (Fig. 1d and 1e).
|1||71||M||50 mm cyst pancreatic head, wall thickness <2 mm, septated, septae thin, mass lesion absent, 12 cc fluid aspirated||50 mm cyst, mass absent, septated, thin septae||1.2||46|
|2||47||M||20×15 mm multiseptated cyst pancreatic tail, thin walled, 2 cc fluid aspirated||NA||NA||NA|
|3||58||M||20 mm unilocular cyst, pancreatic tail, 3 cc aspirated||16 mm cystic lesion, well circumscribed, low density||2||NA|
|4||64||M||NA||40 mm cyst pancreatic head||NA||NA|
|5||74||M||25 mm unilocular cyst, tail of pancreas||NA||NA||NA|
Cyst fluid analysis
Cyst fluid analysis was performed on 2 cases, and in both the carcinoembryonic antigen level was <5 ng/mL.
Cytomorphologic features (Table 2)
Cases 1, 2, and 3
These aspirates showed a monomorphic population of tumor cells (Fig. 1a,b,c). The cellularity varied from low (case 1) to highly cellular (cases 2 and 3). The first 2 cases were interpreted as “atypical glandular cells”, and a diagnosis of adenocarcinoma was rendered in case 3.
|Case no.||Cellularity||Type of preparation||Cytology diagnosis||Cluster||Single cells||Monotonous tumor cells||Plasmacytoid cells||Chromatin||Nucleoli|
|1||Hypo||TP||Atypical glandular cells||Present||Absent||Y||N||Fine & evenly distributed||Present|
|2||Hyper||CS||Atypical glandular cells||Present||Rare||Y||Y||Fine & evenly distributed||Present|
|3||Hyper||CS||Adenocarcinoma||Present||Rare||Y||Y||“Salt and Pepper”||Present|
|5||Hypo||TP||Acute & chronic inflammation||No tumor||NA||NA||NA||NA||NA|
The cells were arranged predominantly as tightly cohesive aggregates, many in a 3-dimensional pattern. The cells were monomorphic with round and regular nuclear contours. Rare single tumor cells as well as plasmacytoid cells were identified in 2 cases. Although nuclear moulding was observed, it was focal and never reached the degree seen in small-cell carcinoma. Coarse and evenly distributed chromatin was noted in only 1 case. Small nucleoli were identified in all cases. Case 2, in addition, showed a diffusely vacuolated cytoplasm. Tumor necrosis and nuclear membrane irregularities were not identified.
Cases 4 and 5
These aspirates were composed of neutrophils and occasional histiocytes, and tumor cells were not identified.
Comparative Analysis of Solid PETs
Thirty-four of the 39 (87.2%) cases were accurately diagnosed as PETs. These aspirates showed dyshesive and frequently plasmacytoid cells. The monomorphic population of tumor cells frequently showed a “salt and pepper” pattern of nuclear chromatin. Two (5.1%) cases showed rare neoplastic cells and were diagnosed as suggestive for, but not diagnostic of, a PET. Three cases were either acellular or showed only gastrointestinal contamination and, thus, were deemed unsatisfactory.
Cytomorphologic features of cystic ductal adenocarcinoma (Table 3)
The cellularity of specimens was variable, and most showed both clusters and single tumor cells (Fig. 2). Two of the 5 cases showed tumor necrosis, and all demonstrated fine, evenly distributed chromatin. Significantly, the characteristic monomorphism of pancreatic endocrine tumors was only focally present (Fig. 3). Nuclear pleomorphism was further enhanced by the presence of irregular nuclear membranes (Figs. 4 and 5). Notably, 1 case, in addition to showing scant amounts of extracellular mucin, also demonstrated a vacuolated cytoplasm, similar to that seen with the cystic PET (case 5) (Fig. 2c). The remaining 4 cases were negative for both intracellular and extracellular mucin.
|Case no.||Cellularity||Type of preparation||Original cytology diagnosis||Clusters||Single cells||Monotonous cells||Plasmacytoid cells||Chromatin||Nucleoli||GI contamination||Necrosis|
|6||Hypo||TP||Adenocarcinoma||Present||Rare||No angulated nuclei||No||Fine and evenly distributed||None||Absent||Absent|
|7||Hyper||TP||Adenocarcinoma||Present||Present||No irregular nuclear membrane||Present||Fine and evenly distributed||Present||Absent||Present|
|8||Hypo||TP||Adenocarcinoma||Rare||Present||No irregular nuclear membrane||No||Fine and evenly distributed||None||Absent||Present|
|9||Hypo||TP||Atypical glandular cells||Present||Absent||No irregular nuclear membrane||No||Fine and evenly distributed||Present||Absent||Absent|
|10||Hypo||CS||Adenocarcinoma||Absent||Present||No irregular nuclear membrane||No||Fine and evenly distributed||None||Absent||Absent|
All 5 tumors showed evidence of cystic degeneration. Monomorphic tumor cells, arranged in both an insular and trabecular pattern, rimmed the cystic space(s). As expected, the tumors were positive for both chromogranin and synaptophysin on immunohistochemistry.
Owing to recent improvements in abdominal imaging, an increasing number of pancreatic cystic lesions are identified in patients who present with jaundice, pancreatitis, and/or abdominal pain as well as asymptomatic patients who are tested for unrelated causes.8 EUS-guided FNAB is considered an integral part of the evaluation of these cystic pancreatic lesions.1
The distinction of pancreatic adenocarcinoma from a PET has obvious clinical implications, whether they are resectable or not. Metastatic and nonresectable ductal adenocarcinoma can be treated by 5-fluorouracil (5-FU)-based chemotherapy and gemcitabine, whereas metastatic PETs may still be surgically managed and amenable to hormonal manipulation.9 Ultimately, metastatic PETs have a significantly longer survival time than those with pancreatic ductal adenocarcinomas.
Cystic PET, an uncommon macroscopic variant, has surprisingly been reported to be seen in 13% of PETs, according to a recent series.10 Interestingly, PETs arising in the setting of multiple endocrine neoplasia, type I syndrome, appear to be more frequently cystic than nonsyndromic PETs.10 In the current series, PETs were both nonfunctional and nonsyndromic.
Radiological evaluation with observation of a “cyst into cyst” and hypervascularity at the periphery of the tumor10 may suggest the diagnosis, because these findings are uncommon for a pancreatic mucinous neoplasm. However, none of our cases were recognized on imaging studies.
Chemical analysis of cyst fluid contents significantly contributes to preoperative diagnosis of cystic pancreatic lesions.1 Mucinous lesions of the pancreas, including mucinous cystic neoplasms and intraductal papillary mucinous neoplasms, are associated with elevated levels of carcinoembryonic antigen. Conversely, cystic PETs show low carcinoembryonic antigen levels, as in 2 of these cases. However, a low carcinoembryonic antigen level does not exclude a neoplastic mucinous cystic lesion.1
The cytomorphologic features of conventional (solid) PETs have been extensively characterized.11–15 On FNAB, these neoplasms are highly cellular and composed predominantly of either single cells or loosely cohesive cell clusters. The cells are typically plasmacytoid and occasionally binucleated. The presence of coarse ‘salt and pepper’ chromatin is also suggestive of neuroendocrine differentiation, a feature readily confirmed by chromogranin and synaptophysin immunostains. These cytomorphologic features allow an accurate and precise diagnosis of a PET. Indeed, 87.2% of our series of 39 solid PETs were correctly interpreted on FNAB.
In sharp contrast, none of the 5 cystic PETs were correctly interpreted preoperatively. FNAB of cystic PETs demonstrates subtle but significant cytomorphologic differences. The aspirates show monotonous but tightly cohesive tumor cells, whereas single cells are poorly represented. Consequently, 3 of our cases closely mimicked a ductal adenocarcinoma. In the only other report of a cystic PET, the tumor cells were also described to be predominantly cohesive.6
Cystic change within a ductal adenocarcinoma is not an infrequent event.7 A cursory examination of smears from cystic ductal carcinomas confirmed their resemblance to cystic PETs. We identified 5 biopsy-proven, EUS-guided, FNABs of cystic pancreatic ductal adenocarcinomas. To ensure an appropriate comparison, we restricted our analysis to cases where either a cytospin or ThinPrep preparation was available. Of note, the presence of irregular nuclear outlines and nuclear pleomorphism assisted in distinguishing ductal adenocarcinomas from PETs. Furthermore, extracellular and intracellular mucin as well as tumoral necrosis, when present, favor a diagnosis of ductal adenocarcinoma. These characteristics mirror the morphology of these tumors because cystic transformation of ductal adenocarcinoma is frequently caused by intratumoral necrosis,7 a finding typically absent in cystic PETs.10 Interestingly, 1 case each of cystic PET and cystic ductal adenocarcinoma showed a diffusely vacuolated cytoplasm, an uncommon change, rarely documented in both PETs and ductal adenocarcinomas.16, 17
Two of the cases in this series lacked cellularity, and they were nondiagnostic. This relatively low sensitivity of EUS-guided FNAB for cystic PET is also of concern because the clinical, radiological, and cyst-fluid findings of serous cystadenoma overlap with those of cystic PETs. Given the recent tilt toward conservative management of cystic pancreatic lesions, a failure to recognize cystic PETs would delay recognition of this potentially malignant neoplasm.18 On site assessment of the adequacy of the specimen (either in the radiology or endoscopy suite) has been reported to increase the diagnostic yield of pancreatic FNABs.11, 19 However, its effectiveness in the assessment of cystic lesions has yet to be demonstrated. On site rapid assessment of cystic pancreatic neoplasms is not performed at this institution.
Immunohistochemistry for chromogranin and synaptophysin would have provided an accurate diagnosis. These were not performed because of the cytomorphologic features reported above, as well as the lack of clinical suspicion to guide cytologists.
Conventional wisdom would suggest that the cytomorphologic features of cystic PET would overlap with other cystic pancreatic neoplasms with monomorphic cells, including solid-pseudopapillary tumor and acinar-cell carcinoma.20 The cytomorphologic features of solid-pseudopapillary tumors include papillary fragments with vascular cores composed of myxoid tissue21 and uniform, round to oval, bland-appearing tumor cells. Additionally, solid-pseudopapillary tumors display fine chromatin and lack prominent nucleoli. Furthermore, immunohistochemistry would provide additional ancillary proof in equivocal cases. Solid-pseudopapillary tumors are keratin negative and beta-catenin positive, the exact opposite of PETs. However, it is important to recognize that solid-pseudopapillary tumors can be focally positive for synaptophysin. The presence of cherry red macronucleoli in conjunction with immunohistochemical markers for acinar cells such as trypsin would help distinguish acinar cell carcinomas from endocrine neoplasms.
Neoplastic mucinous lesions intraductal papillary mucinous neoplasms and mucinous cystic neoplasms) represent the most common pancreatic cystic neoplasms. Aspirates from these neoplasms frequently show a background composed of abundant and dense mucinous material.22 In contrast, cystic PETs lack mucin. However, mucin from the gastric or duodenal surface may present diagnostic difficulties.
On imaging studies, oligocystic serous cystadenomas can mimic cystic PETs. In our experience, the lesional cells of serous cystadenomas, although identified on CT-guided biopsies, are only rarely noted on a EUS-guided biopsy.23 When present, the tumor cells are monotonous and are either single or in flat sheets. In distinction to PETs, the cells show a moderate amount of glycogen-rich and clear cytoplasm and finely and evenly distributed chromatin.
In conclusion, cystic PET is an uncommon neoplasm, which can present cytologists with unique diagnostic challenges. Apart from their resemblance to solid-pseudopapillary tumors, these tumors could be mistaken for cystic ductal adenocarcinoma, particularly when direct smears are not available. Awareness of these cytomorphologic features and appropriate use of immunohistochemical markers would avoid this pitfall.
- 6Cystic islet cell neoplasm: Differential diangnosis with other pancreatic cystic masses. Pathology Case Reviews. 2001; 6: 127–134.,