The Role of Endoscopic Ultrasound in Cystic Pancreatic Tumor Treatment

Abstract

Pancreatic cysts are being detected with increasing frequency. These lesions may be pseudocysts, serous cystadenomas, mucinous neoplasms, or other rare pancreatic malignancies. It is important to differentiate among the various cystic lesions in order to further guide management. We review the role of endoscopic ultrasound, fine needle aspiration including cytology, and tumor markers in the evaluation of these cystic tumors. We also discuss emerging concepts in the management of cystic pancreatic tumors.

Recent advances in radiologic imaging techniques have allowed for the more frequent detection of pancreatic cysts, including increased incidental detection of pancreatic cysts in asymptomatic patients. These cysts may have evolved from benign, premalignant, or malignant processes. Difficulty arises in appropriately differentiating these lesions in order to guide management, particularly in asymptomatic patients. This review focuses on pancreatic cystic lesions and the role of endoscopic ultrasound (EUS) and fine needle aspiration (FNA) in the difficult problem of differential diagnosis of pancreatic cystic tumors. The role of EUS in assisting endoscopic drainage of classic pancreatic pseudocysts will not be discussed.^1,2

Pancreatic cystic lesions include pseudocysts, which account for 90% of all cysts. The remainder comprises mucinous cystic neoplasms, serous cystadenomas, intraductal papillary mucinous tumors (IPMT), and rarer pancreatic tumors such as neuroendocrine tumors and metastases.

Types of Pancreatic Cysts

Serous Cystadenomas

In the past, serous cystadenomas have been referred to as microcystic adenomas. They are more commonly seen in women and most frequently are located in the pancreatic body or tail. Grossly, serous cystadenomas consist of many tiny cysts containing clear fluid separated by septa, giving them a characteristic honeycomb-like appearance.³ On radiologic imaging they may have a well-defined capsule with an enhancing cyst wall and septa.⁴ A characteristic central stellate scar is present in only about 30% of cases.^5-7 These lesions are considered benign for the most part, but there have been a few case reports of serous cystadenocarcinomas.⁸

Mucinous Cystic Neoplasms

These are the most common cystic tumors and are classically macrocystic. They include cystadenocarcinomas and mucinous tumors, which are further divided into benign (cystadenoma), borderline, and malignant classifications.⁹ These mucinous tumors, even when classified as benign, are still considered premalignant, occur more frequently in women, and are found mostly in the pancreatic body or tail.¹⁰ On radiologic imaging, these cysts can be multiloculated and usually do not communicate with the duct. Calcifications, thick walls, and septations can be suggestive of malignancy.¹¹

Intraductal Papillary Mucinous Tumor

Intraductal papillary mucinous tumors are premalignant lesions that occur more frequently in males between the ages of 60 and 70. They present more often in the head of the pancreas and may involve the main pancreatic duct or pancreatic side branches. These tumors may present as cystic-lesion, branch-duct IPMTs, which generally have a better prognosis than main duct IPMT. The ampulla may be patulous and may have what has been described as a “fish-eye” or “fish-mouth” appearance, with mucin draining from the opening. On radiologic imaging, IPMT can present as diffuse or segmental dilation of the pancreatic duct.

Pseudocysts

Pseudocysts are benign, nonneoplastic lesions occurring as the result of acute or chronic pancreatitis or trauma. These cysts are generally sterile and contain a large amount of enzymes due to communication with the pancreatic duct. On radiologic imaging, pseudocysts generally appear unilocular with uniform enhancing walls.

Other Pancreatic Cysts

Congenital cysts, lymphangiomas, solid papillary tumors, islet cell tumors, and pancreatic metastasis can also present as pancreatic cystic lesions.

Endoscopic Ultrasound of Pancreatic Cystic Lesions

Endoscopic ultrasound allows for high resolution imaging of the pancreas to better define pancreatic cysts. EUS may display the characteristic honeycomb pattern of serous cystadenomas and may also reveal numerous cysts with thin-walled septae. Mucinous lesions tend to be macrocystic and may be unilocular or multilocular (Figures 1 and 2). The presence of an adjacent mass and septations may increase the chance of malignancy.¹² Small cyst size does not necessarily exclude malignancy.¹³ A cyst without an associated mass or mural nodules in a patient with a prior history of acute pancreatitis is more likely to be a pseudocyst.¹⁴

Figure 1.

A small unilocular cystic lesion in the pancreatic head incidentally found with endoscopic ultrasound-guided fine needle aspiration and subsequent surgery confirming the presence of mucinous neoplasm.

Figure 2.

Endoscopic ultrasound image of a complex cystic mass with septation.

Several studies have examined the role of EUS in pancreatic cystic lesions. Koito and associates¹⁵ performed retrospective correlation to differentiate nonneoplastic from neoplastic cysts by classifying 52 pancreatic cysts, based on EUS and pathology, into six types. They included thick-wall, tumor-protruding, thick-septal, microcystic, thin-septal, and simple types. All neoplastic cysts were categorized in the first four types and all nonneoplastic cysts in the last two types. The estimated accuracy between of EUS in making the above differentiation with two observers was 96% and 92%.

Gress and colleagues¹⁶ correlated EUS findings with surgical pathology in 35 patients. Their study included 14 mucinous cystadenocarcinomas that were more likely characterized by hypoechoic-cystic/solid-mass or complex cyst and were more likely to be associated with a dilated main pancreatic duct. Benign mucinous duct ectasia lesions in 6 patients showed a dilated main pancreatic duct in conjunction with hyperechoic thickening of the pancreatic duct wall. Two cases of intraductal mucinous hyperplasia additionally showed a hypoechoic mass. Intraductal papillary carcinoma in 11 patients had features similar to mucinous cystadenocarcinomas but also had echogenic foci in the mass and intraductal hyperechoic lesions. Two cases of microcystic adenoma showed either a mixed hypoechoic solid or cystic mass or a complex cyst without the additional features seen in mucinous cystadenocarcinoma.

Sedlack and coworkers,¹⁷ in an analysis of 34 patients, found that the following features on EUS imaging predict a malignant or potentially malignant cystic lesion (with a sensitivity of 91%, specificity of 60%, and accuracy of 82%): wall thickness of 3 mm or more, macroseptation with cystic compartments greater than 10 mm, presence of mass or intramural growth, or cystic dilation of the main pancreatic duct.

O’Toole and associates¹⁸ compared EUS findings in macrocystic serous and mucinous lesions. The presence of microcysts was only seen in serous lesions (P<.0008). All mucinous lesions had a thickened cyst wall, compared to 56% of macrocystic serous lesions (P<.0001).

The echo features on EUS of pancreatic cysts in the above study and by others discussed above may not provide an absolute differentiation between cystic pancreatic lesions. Ahmad and colleagues^19,20 attempted to determine whether or not EUS alone can differentiate between malignant and benign cystic lesions of the pancreas. Original EUS images were reviewed by two endosonographers, who were blinded to each other’s interpretations and to the surgical pathology findings. They concluded that EUS features cannot reliably differentiate benign from malignant lesions. Another study by the same group evaluated interobserver agreement for distinguishing cystic lesions. Accuracy of EUS for the diagnosis of neoplastic versus nonneoplastic lesions ranged from 40% to 93%. In this study, less than half of the patients underwent surgery. Information regarding the negative predictive value of EUS is not clearly explained and also what clinical (eg, history of acute pancreatitis) or EUS (eg, no septations, absence of solid component, no lymphadenopathy) findings led to clinical observation as opposed to surgery.²¹

Endoscopic Ultrasound Fine Needle Aspiration

Linear EUS allows real-time ultrasound-guided FNA of pancreatic cystic lesions in a minimally invasive manner. FNA is generally performed using a 19- or 22-gauge needle. Fluid from the aspirate can be analyzed for cytologic and morphologic features and tumor markers (Figure 3). On cytology, serous cystadenomas contain cuboidal epithelium. The presence of glycogen on Periodic Acid Schiff (PAS) staining without diastase, along with negative mucin staining, is suggestive of a serous cystadenoma. Mucinous lesions have neoplastic epithelium with columnar cells. Mucinous lesions stain positive by PAS with diastase and mucicarmine staining.

Figure 3.

Endoscopic ultrasound imaging of an incidental unilocular cystic lesion in pancreatic body (A). Endoscopic ultrasound-guided fine needle aspiration (B, C) of the cystic lesion shown in part A. Note the tip of the needle in the cyst (arrowhead) with decreasing size due to aspiration. Cytopathology revealed a mucinous cystadenoma.

The success rate and accuracy of EUS-guided FNA cytology, however, is somewhat variable among centers. In the study by Brandwein and associates,¹² EUS-guided FNA was only 50% sensitive for evidence of malignancy in cystic pancreatic tumors. In a study by Sedlack and coworkers,¹⁷ the accuracy of EUS-guided FNA cytology for diagnosing malignant or potentially malignant pancreatic cystic lesion was 55%. Although the specificity of EUS-guided FNA was 100%, the sensitivity was only 27%. Frossard and colleagues²² prospectively studied 127 consecutive patients with pancreatic cystic lesions. The results of EUS and EUS-guided FNA were compared with the final diagnosis in 67 cases. EUS-guided FNA correctly identified 65 cases (97%) and EUS correctly identified 73% of cases.

The Frossard study revealed that a cyst fluid CA19-9 measure greater than 50,000 U/mL had a 15% sensitivity and 85% specificity in identifying mucinous lesions.²² The multicenter Cooperative Pancreatic Cyst (CPC) Study²³ investigated the role of various tumor markers in cyst fluid in correctly classifying cystic lesions. The study included 341 patients who underwent EUS with FNA, including 112 with available surgical correlation. Receiver operating curves revealed that a pancreatic cyst fluid carcinoembryonic antigen (CEA) cut-off value of 192 ng/mL would differentiate mucinous from nonmucinous lesions with 79% accuracy, which was significantly greater than EUS morphology (51%) or cytology (59%). Recently, a small study by Raty and coworkers²⁴ showed that tumor-associated trypsin inhibitor was significantly higher in patients with mucinous neoplasms and endocrine islet tumors, when compared to pseudocysts and serous cystadenomas. Sperti and colleagues²⁵ evaluated the presence of mucin-like, carcinoma-associated antigen in pancreatic cyst fluid obtained during surgery or percutaneously. The antigen was elevated in 7 of 8 patients with mucinous lesions and higher levels noted in mucinous cystadenocarcinoma than cystadenomas.²⁵

A recent pooled analysis²⁶ reviewed published data on pancreatic cyst fluid obtained by radiologic, surgical, or endoscopic means from 12 studies including 450 patients. Analysis revealed that CEA levels less than 5 ng/mL or CA 19-9 levels less than 37 U/mL suggested a serous cystadenoma or pseudocyst (sensitivity 50%, specificity 95%). A CEA level greater than 800 ng/mL strongly suggested a mucinous lesion (sensitivity 48%, specificity 98%). Also, cyst fluid amylase levels less than 250 U/L essentially excluded the possibility of pseudocyst. Cytology revealed malignancy in 48% of mucinous cystadenocarcinomas.

Complications from FNA include pancreatitis (2–3%), within-cyst hemorrhage (<1%), and infection (<1%).²⁷ A recent retrospective review²⁸ reported complications in 2.2% (13 of 603) patients, who had undergone EUS-guided FNA of pancreatic cysts. Six patients had pancreatitis, 4 patients had abdominal pain, 1 patient had a retroperitoneal bleed, 1 patient had an infection, and 1 patient had bradycardia. Most experts recommend antibiotics for 3–5 days postprocedure to decrease infection risk.²⁷ It has been our practice and that of others to completely aspirate a unilocular cyst or the locule that is entered in order to decrease the risk of infection.¹⁴ In the event that a second pass is needed during EUS-guided FNA for technical reasons, a new needle is recommended (by some physicians, including us) to decrease the risk of infection.

Emerging Concepts

Molecular markers have been recently analyzed in pancreatic cyst fluid. Khalid and associates²⁹ evaluated pancreatic cyst fluid for DNA mutations to detect malignancy in mucinous neoplasms. Thirty-six cysts were analyzed. Malignant cysts were differentiated from premalignancies based on CEA level, DNA quality, number of mutations, and the sequence of mutations acquired. The most predictive was early k-ras mutation followed by allelic loss, with a sensitivity of 91% and specificity of 93%. Anderson and colleagues³⁰ analyzed FNA specimens for gene expression biomarkers. They were able to differentiate pancreatic neoplasm from benign pancreatic conditions. Mutational analysis of cyst fluid may help in differentiating the various mucinous lesions and lead to better clinical management of the patient.

EUS-guided Trucut biopsy (TCB) may be performed to obtain a biopsy specimen of various organs. A TCB needle (Quick-Core; Wilson-Cook Medical) can be used with linear endoscopes.³¹ Recently, Levy and associates³² examined the use of EUS-guided TCB of the pancreatic cyst wall to differentiate cystic lesions. EUS-guided TCB was performed in 10 patients and was nondiagnostic in 3 of 10. Of the remaining 7 patients, the histology from the TCB tissue core established the diagnosis in 6. Varadarajulu and coworkers³³ evaluated the role of the EUS-guided TCB device in 18 patients. Two complications were reported: bleeding of a gastric stromal tumor that was successfully managed by endoscopic injection of epinephrine and mediastinitis after EUS-guided TCB sampling of a bronchogenic cyst requiring surgical drainage. Trucut biopsy may provide more tissue to help differentiate pancreatic cystic lesions, but larger studies must be performed to further assess the safety and clinical utility of this procedure.

A recent pilot study examined ethanol lavage of pancreatic cystic lesion as a nonoperative method of ablation of pancreatic cystic lesions. Asymptomatic patients (N=25) were enrolled in the study. The cyst was collapsed and lavaged for 3–5 minutes with ethanol. Mean CEA levels in the cyst fluid of patients in this study was 5,916 ng/mL. Eight patients had complete resolution of their cyst at 6–12 months follow-up. Five patients underwent surgical resection of the cyst, which revealed various degree of histologic evidence of epithelial ablation of the cyst lining.³⁴ Further studies and long-term follow-up are needed to assess the clinical application of this technique.

Conclusion

Mucinous lesions can be premalignant or malignant tumors. The natural history of progression to malignancy has not been well studied. There is also a lack of long-term follow-up in patients with no symptoms, who have apparently benign cysts on imaging and/or FNA, who do not undergo surgery. EUS, cyst fluid analysis, and cross-sectional imaging aid in the differentiation of the various pancreatic cystic lesions. Management should be individualized based on symptoms, age of the patient, location of the cyst, and comorbid conditions, which would increase surgical risk. Surgery is recommended if a patient is symptomatic from a cystic lesion or if a mucinous lesion is confirmed or strongly suspected. Conservative management with radiologic or EUS follow-up imaging can be offered to patients who decline surgery or those in whom benefits of immediate surgical intervention are lower than with watchful waiting and periodic imaging follow-up. Recent advances in molecular markers, imaging, sampling techniques, and ethanol cyst ablation may guide us in the future to better differentiate and treat pancreatic cystic lesions.