Fax: (011) 04 91 22 35 44
Cystic pancreatic endocrine tumors
An Endoscopic Ultrasound-Guided Fine-Needle Aspiration Biopsy Study With Histologic Correlation
Article first published online: 13 APR 2009
Copyright © 2009 American Cancer Society
Volume 117, Issue 3, pages 203–210, 25 June 2009
How to Cite
Charfi, S., Marcy, M., Bories, E., Pesanti, C., Caillol, F., Giovannini, M., Viret, F., Robert Delpero, J., Xerri, L. and Monges, G. (2009), Cystic pancreatic endocrine tumors. Cancer Cytopathology, 117: 203–210. doi: 10.1002/cncy.20024
- Issue published online: 11 JUN 2009
- Article first published online: 13 APR 2009
- Manuscript Accepted: 29 JAN 2009
- Manuscript Revised: 6 JAN 2009
- Manuscript Received: 11 NOV 2008
- pancreatic endocrine tumor;
- cystic pancreatic tumor;
- endoscopic ultrasound;
- fine-needle aspiration biopsy;
Cystic pancreatic endocrine tumors (PETs) are rare neoplasms with a preoperative diagnostic challenge. The aim of this study was to evaluate the preoperative diagnostic strategy for these tumors and to assess the clinical and pathologic characteristics.
Six cases of cystic PET were retrospectively enrolled. Endoscopic ultrasound-guided fine-needle aspiration biopsy (EUS-FNAB) was performed in 4 cases. All cytomorphologic data from conventional smears, ThinPrep preparations, and cell block preparations were reported in detail.
There were 3 male and 3 female patients with a mean age of 52.3 years. Tumor size ranged from 10 mm to 60 mm (mean, 29.8 mm). EUS-FNAB contributed to an accurate diagnosis in all cases. Cytologically, loosely cohesive aggregates and single cells were predominant. Cells were small and typically plasmacytoid, with occasional cytoplasmic vacuolization. Nuclei were round or oval, uniform, with finely and evenly distributed chromatin. Immunocytochemistry confirmed the endocrine differentiation. Histologic findings were typical for endocrine proliferation. All tumors were well differentiated.
Cystic PET is an unusual finding that presents diagnostic challenges for both endoscopists and cytologists. EUS-FNAB with the Thinprep preparation technique and cell block material were found to be helpful in improving diagnostic accuracy. Immunocytochemical staining with endocrine markers confirmed the diagnosis. Cancer (Cancer Cytopathol) 2009. © 2009 American Cancer Society.
Owing to recent improvements in pancreatic imaging, an increasing number of cystic lesions have been identified in asymptomatic as well as symptomatic patients. Concurrently, the application of endoscopic ultrasound-guided fine-needle aspiration biopsy (EUS-FNAB) has expanded the utility of EUS by allowing cytologic material to be obtained for a preoperative diagnosis. The latter is essential for appropriate tumor management.1, 2 Pseudocyst or serous cystadenoma can be conservatively managed with clinical and radiologic follow-up; however, mucinous neoplasms and solid pseudopapillary neoplasms require surgery.3, 4 In addition, primary solid pancreatic tumors such as ductal adenocarcinoma, acinar cell carcinoma, and pancreatic endocrine tumors (PETs) can be observed with cystic features.5 Cystic PETs represent <5% of all cystic pancreatic tumors.5 Cystic changes occur in approximately 2% to 10% of PET.1, 4, 6-8 To the authors' knowledge, the physiopathologic mechanisms involved in this setting are unknown.9
EUS-FNAB findings have been well‒described for solid PETs,10, 11 but have rarely been described in cases with cystic presentation.2, 12 The current study analyzed the clinical, radiologic, and histologic features of 6 cystic PET cases. Cytomorphologic and immunocytochemical EUS-FNAB findings are described in 4 cases.
MATERIALS AND METHODS
During 4 years, from 2003 to 2007, 6 patients who underwent surgical exploration at the Department of Surgical Oncology at the Paoli-Calmettes Institute with a pathologic diagnosis of cystic PET were retrospectively studied. Four patients had a preoperative EUS-FNAB procedure with available material. In each case, clinical, radiologic, and follow-up data were recorded. Biopsy procedures were performed in the endoscopy unit using a linear EUS. FNAB was performed under direct EUS guidance with a 22-gauge needle (Wilson-Cook Corporation, Winston Salem, NC). Direct smears were prepared by the endoscopist in 3 cases and were air-dried for May-Grünwald-Giemsa staining. The ThinPrep preparation was used in all cases. Cell block material, fixed in 10% neutral buffered formalin, was collected from aspirated material. Cyst fluid analysis was performed in 2 cases for CA 19.9 antigen and carcinoembryonic antigen (CEA). Surgical specimens were available from all patients. Specimens were received immediately after resection and fixed in 10% neutral buffered formalin. Thin sections (5 μm) were cut from paraffin-embedded cell blocks and stained with hematoxylin, eosin, and safran. Immunohistochemical analysis was performed according to the streptavidin-biotin technique. The list of antibodies is given in Table 1. Tumors were classified according to both World Health Organization classification of endocrine and digestive tumors and TNM staging as recently suggested.13, 14
Clinical and Radiologic Findings
The patients were 3 men and 3 women. The mean age was 52.3 years (range, 43-69 years). One patient had a history of colon carcinoma. Patients were symptomatic for abdominal discomfort and pain. No patient had multiple endocrine neoplasia type 1 (MEN1) (Table 2).
|Case No.||Age, Years/Sex||Clinical Presentation||Imaging Impression||EUS-FNAB*||Site||Size, mm||Surgical Procedure|
|1||57/Woman||Abdominal pain||Cystic lesion consisting in mucinous tumor||No||Tail||60||Left panceatecomy with Splenectomy|
|2||51/Woman||Incidental||Cystic lesion||No||Head-body||NA||Isthmic pancreatectomy|
|3||51/Man||Abdominal pain||Cystic lesion with hyperechoic wall||Yes||Tail||28||Left panceatecomy with splenectomy|
|4||43/Man||Abdominal pain||Multiseptated lesion consisting in mucinous tumor||Yes||Uncinate process||46||Duodenopancreatectomy|
|5||69/Man||Abdominal pain||Cystic lesion with thick wall consisting with malignant IPMT||Yes||Body-tail||15||Left and body pancreatectomy|
|6||43/Woman||NA||Cystic lesion with thick wall||Yes||Body||30||Tumor enucleation|
All cystic lesions were detected on imaging studies, in the majority of cases by both EUS and computed tomography. The pancreatic lesions were uniloculated in 5 cases and multiseptated in 1 case. Tumor size ranged from 15 mm to 60 mm in greatest dimension (mean, 35.8 mm). EUS findings were quite similar in the 4 cases in which EUS-FNAB was performed. Cystic PET appeared as a cystic lesion with a thick wall (>2 mm) and a hypoechogenic center. During EUS-FNAB, 1 cc of fluid or less was aspirated in all cases (Fig. 1).
Tumors were located in the tail in 2 cases. Other tumors were located in the body, in the head-body junction, in the body-tail junction, and in the uncinate process in 1 case each, respectively. Radiologically, mucinous cystadenoma was suggested in 2 cases and malignant intraductal papillary mucinous neoplasm in 1 case. No diagnosis was considered for the 3 other cases.
Surgical treatment included left pancreatectomy with splenectomy in 2 cases, a duodenopancreatectomy in 1 case, a left and body pancreatectomy in 1 case, and an isthmic pancreatectomy in 1 case. Simple enucleation of the tumor was performed in 1 case (Case 6).
All detailed cytologic findings are presented in Table 3. Cellularity was low in most cases, especially in conventional cytologic smear preparations. The background was bloody in 1 case. The cellular pattern included loosely cohesive small clusters and single cells. Tumor cells were mostly round. Cytoplasm was moderately abundant with an amphophilic or finely granular appearance. Cell borders were indistinct in 3 cases. Nuclear membranes had smooth contours in all cases. No nuclear pleomorphism was obvious. Cells with eccentric nuclei were present in 2 cases, giving a plasmacytoid appearance. No binucleated or multinucleated cells were noticed. The chromatin was evenly distributed with a salt‒and‒pepper appearance in all cases. Fine nucleoli were present in all cases. No mitotic figures were encountered (Fig. 2).
|Feature||Case 3||Case 4||Case 5||Case 6|
|Smears||NP||No tumor cells||Low||Low|
|Type of preparation||ThinPrep||ThinPrep||ThinPrep||ThinPrep|
|Cell shape||Round||Round||Round and oval||Oval|
|Nucleus||Round, eccentric||Round||Round, eccentric||Round, eccentric|
|Chromatin||Fine and even||Fine and even||Fine and even||Fine and even|
|Immunohistochemistry on cell block preparation|
|Ki-67||No tumor cells||−||−||NP|
|p53||No tumor cells||No tumor cells||−||NP|
The immunocytochemical study demonstrated positivity for chromogranin A and synaptophysin in all cases. No cells reacted with Ki-67. No reactivity with cytokeratin 7 and MUC1 was detected.
Macroscopic Features and Histologic Characteristics
Table 4 shows the macroscopic and histologic characteristics of the tumors. Cystic tumors were solitary in all cases. The greatest tumor dimension ranged from 10 mm to 60 mm (mean, 29.8 mm). Cystic content was bloody in 3 cases. One tumor had a multiseptated appearance. Internal vegetations were noticed in 1 case (Fig. 3). Histologically, typical nests of endocrine tissue were present along the cyst wall in all cases. Cellularity was marked in 2 cases, moderate in 3 cases, and low in 1 case. Superficial layers of endocrine cells were in direct contact with the cyst content. The endocrine proliferation demonstrated trabecular, insular, or cord arrangements. Angiomatoid change was noted in 1 case. Cells were monomorphous, and small to medium-sized. Mitoses were observed in 1 case (1 mitosis per 10 high‒power fields). The stroma was abundant, and usually hyalinized. The peritumoral capsule was inconspicuous. Numerous large arterial vessels were present within the capsule or at its contact. Vascular invasion was observed in 1 case. There was no necrosis. Foci of hemosiderin and cholesterol deposits were observed. One tumor was associated with a well-differentiated solid PET measuring 4 mm (Fig. 4). A summary of the pathologic diagnoses with pTNM classification is presented in Table 5.
|Case No.||Size, mm||Cystic Content||Septa||Cellularity||Mitotic Index(/10 HPF)||Vegetation*||Vascular Invasion||Necrosis||Immunohistochemistry||Other Findings|
|Ki-67, %||p53, %||Hormonal Expression|
|5||12||Clear||No||+++||0||No||No||No||1||0||PP||Solid PET (4 mm)|
|Case No.||Cytologic Diagnosis||Histologic Diagnosis||pTNM Classification|
|1||NP||Well-differentiated endocrine carcinoma||pT3NxM0|
|2||NP||Well-differentiated endocrine tumor with benign behavior||pT1NxM0|
|3||Endocrine tumor||Well-differentiated endocrine tumor with uncertain behavior||pT2NxM0|
|4||Consistent with endocrine tumor||Well-differentiated endocrine tumor with uncertain behavior||pT3NxM0|
|5||Consistent with endocrine tumor||Well-differentiated endocrine tumor with benign behavior||pT1mNxM0|
|6||Endocrine tumor||Well-differentiated endocrine with uncertain behavior||pT2NxM0|
Cyst Fluid Analysis
This analysis was performed on 2 cases, and in both instances the CEA and CA 19.9 levels were normal (Case 3 and 4).
The cystic presentation of PET is rare, reportedly occurring in <2% to 10% of all PETs.1, 4-6, 8 To our knowledge, the mechanisms involved in this change are unknown.9 Most authors have suggested that intratumoral hemorrhage and necrosis account for the cystic appearance.8, 9, 15-20 In 2 reports,9, 16 a direct correlation between tumor size and cystic change was found. In these reports, cystic PETs had a mean size of 6.9 mm and 48 mm versus 2.9 mm and 19 mm for their solid counterparts, respectively.9, 16 The cases presented herein demonstrated an occasionally blood content and angiomatoid features, thereby supporting the role of hemorrhage phenomenon in cystic change of these tumors. However, according to the results of the current series, tumor size does not appear to be correlated with cystic presentation. Tumors in the current study had a mean dimension of <30 mm, and 2 cases measured <15 mm. Therefore, as suggested by Geisinger et al,12 we believe that it is important to distinguish large solid PETs with cystic degeneration and necrosis from “true” cystic PETs.
Clinically, cystic PETs can occur in 2 different settings: as sporadic tumors and in the context of MEN1. Sporadic lesions are reported in patients who are mostly in their sixth or seventh decade of life with no evident gender predilection.2, 9 In MEN1 syndrome, cystic PETs appear to be more prevalent, accounting for up to 13% of PETs. They occur in younger patients and are usually multiple.7, 8, 21 Secreting tumors are mostly diagnosed in a MEN1 context. Glucagon is the most frequent peptide detected in these tumors. In the current series, no patient presented with MEN1 syndrome and only 1 tumor demonstrated hormonal secretion on immunohistochemical tests. In the literature, cystic PETs occur predominately in the tail, in accordance with the results of the current series.2, 8, 9, 22
Radiologically, 2 features were described as being suggestive of diagnosis: hypervascularity of the tumor periphery (related to the presence of numerous large vessels surrounding the tumor capsule) and the image of cyst into cyst.8, 23, 24 The latter sign is related to cystic change arising in a hypodense tumor. However, the diagnosis of cystic PET is rarely suggested by radiologists2, 8, 9, 25 for cystic pancreatic tumors. Therefore, preoperative diagnosis constitutes a challenge and the distinction between cystic PETs and other cystic lesions has obvious implications because PETs may be managed, even if metastatic, by either surgery or hormonal manipulation. Generally, the value of cytologic examination is limited due to the difficulty in obtaining sufficient material for analysis. Cyst fluid analysis may contribute to preoperative diagnosis of cystic pancreatic lesions. Mucinous lesions of the pancreas are associated with elevated levels of CEA and CA 125 antigen. Conversely, cystic PETs and serous tumors demonstrate a low CEA level.5, 26, 27 In 2 of the cases in the current study, the cystic level of CEA and CA 19.9 antigen were low.
The cytomorphologic features of solid PETs have been extensively characterized to date.9-11, 28-31 Jhala et al31 demonstrated the superiority of EUS-FNAB on computed tomography-FNAB in obtaining cells for the diagnosis of PET. On EUS-FNAB, these neoplasms are highly cellular, and are comprised of either single cells or loosely cohesive cell clusters. Cells are typically plasmacytoid and occasionally binucleated. Chromatin has a salt‒and‒pepper appearance.9-11, 30, 31 Coupled with immunocytochemical study (chromogranin and synaptophysin immunostaining), EUB-FNAB allowed in up to 87.2% a correct diagnosis.10, 11, 30, 31 In sharp contrast to this finding, the diagnosis of cystic PETs appears to be more difficult at the preoperative stage. None of the 5 cystic PETs reported in the series of Deshpande et al,2 was correctly diagnosed. Furthermore, only 3 cases had sufficient material for an appropriate examination.2 In that report, aspirated material demonstrated monomorphous, tightly cohesive tumor cells interpreted as ductal adenocarcinoma.2 In contrast to the cases in the current study, single cells were poorly represented. To the best of our knowledge, only 2 cases of cystic pancreatic lesion were preoperatively diagnosed by FNA cytology using US and computed tomography guidance12, 22; the report of Geisinger et al12 demonstrated small cohesive aggregates of tumor cells with an occasionally plasmacytoid appearance and cytoplasmic vacuolization.12 According to our results, EUS-FNAB successfully contributed to an accurate diagnosis in all cases. Cytomorphologic data were similar to those reported for their solid counterparts, including round monomorphous and loosely cohesive cells, and regular nuclei with fine chromatin. Plasmacytoid features and cytoplasm vacuolization were helpful findings in making a diagnosis. ThinPrep preparation and cell blocks contributed more to making a diagnosis than smears, which yielded only rare neoplastic cells. Immunocytochemical staining on cell blocks confirmed the diagnosis in all cases. In the report of Deshpande et al, neither immunocytochemical study nor cell block technique was performed.2
The presence of irregular nuclear outlines and nuclear pleomorphism is helpful in distinguishing ductal adenocarcinoma from PET. Furthermore, extracellular and intracellular mucin as well as tumoral necrosis favor a diagnosis of ductal adenocarcinoma.2 We noticed that the cytomorphologic features of PETs overlap with other cystic pancreatic neoplasms with monomorphous cells, including solid and pseudopapillary tumor and acinar cell carcinoma. The cytologic features of solid and pseudopapillary tumors include papillary fragments with vascular cores comprised of myxoid tissue and uniform, round to oval, bland‒appearing tumor cells with fine chromatin without prominent nucleoli. Immunocytochemical stains demonstrate keratin negativity and β‒catenin positivity. These tumors can be focally positive for synaptophysin.32 Acinar cell carcinomas demonstrate loosely cohesive cells with uniform nuclei and prominent nucleoli with finely granular and eosinophilic cytoplasm. Trypsin and chymotrypsin immunostaining confirm the diagnosis.33 Neoplastic mucinous lesions, intraductal papillary mucinous neoplasms, and mucinous cystic neoplasms represent the most common pancreatic cystic tumors, accounting for approximately 30% of all cystic tumors.5 Cytologically, they demonstrate a background comprised of abundant and dense mucinous material. Mucin from the gastric or duodenal surface may represent a diagnostic pitfall. Cytologic material is not highly cellular, and contains characteristically small aggregates of columnar cells with a low nuclear-cytoplasmic ratio. The bulk of the cytoplasm is occupied by mucin, which may appear as multiple droplets, a solitary vacuole, or as very fine granules. When frankly carcinomatous areas of the tumor are aspirated, nuclear features of malignancy can be recognized.34, 35 Oligocystic serous neoplasms can mimic cystic PETs. In these lesions, most aspiration samples were hypocellular and cells were arranged in loose clusters with a vaguely acinar pattern. Tumor cells were uniform and cuboidal. Immunocytologic stains and the detection of intracytoplasmic glycogen are helpful in this setting.3 In view of the nature of the EUS-FNAB procedure, normal cellular components are commonly observed in specimens in addition to neoplastic cells. The presence of acinar cells when they are numerous may present a diagnostic challenge. These cells are pyramidal or triangular in shape with round, eccentric, or central nuclei that demonstrates granular chromatin and small nucleoli.10 As observed in the current series and in the report of Deshpande et al, EUS-FNAB yielded neoplastic cells only rarely; this fact should be considered when developing a diagnostic strategy of cystic PET. The histologic diagnosis of these tumors is relatively easy, because most cases present with a well-differentiated pattern.8, 9 Changes such as hemorrhage, hemosiderin deposits, and calcification may be observed.8, 9 Cystic PETs are classified in the same manner as their solid counterparts: based on tumor size, tumor extension, vascular invasion, mitotic index, and Ki-67 index.13, 14
In summary, cystic PET is an unusual tumor that presents diagnostic challenges for both endoscopists and cytologists. EUS-FNAB with ThinPrep preparation techniques and cell block material are helpful in maintaining diagnostic accuracy. Cytomorphologic features are similar to those of solid tumors. Immunocytochemical staining with endocrine markers confirms the diagnosis.
Conflict of Interest Disclosures
The authors made no disclosures.
- 12Cystic islet cell neoplasm: differential diagnosis with other pancreatic cystic masses. Pathol Case Rev. 2001; 6: 127-134., .
- 14Tumours of the endocrine pancreas. In: DeLellisRA, LloydRV, HeitzPU, EngC, eds. Tumours of the Endocrine Organs. Lyon, France: IARC Press, Inc; 2004: 175-207., , , et al.