The evaluation of pancreatic lesions, from solid pancreatic masses to pancreatic cysts, remains a clinical challenge. Although cross-sectional imaging remains the cornerstone of the initial evaluation of an indeterminate pancreatic lesion, advances in imaging with the advent of endoscopic ultrasound scan, elastography, contrast-enhanced endoscopic ultrasound scan, and probe-based confocal laser endomicroscopy have allowed us to visualize the pancreas in even higher resolution and diagnose premalignant and malignant lesions of the pancreas with improved accuracy. This report reviews the range of imaging tools currently available to evaluate pancreatic lesions, from solid tumors to pancreatic cysts.
Key points
- •
Preoperative imaging of the pancreas has become a critical part of the evaluation, diagnosis, and staging of pancreatic cancer and other pancreatic neoplasms.
- •
Endoscopic ultrasound scan is associated with higher sensitivities for diagnosing pancreatic solid lesions and cystic lesions compared with cross-section imaging with MRI or multidetector computed tomography.
- •
New advances in endoscopic ultrasound scan enhancement with elastography and contrast enhancement may provide improvement in distinguishing malignant from benign inflammatory pancreatic diseases.
Introduction
Pancreatic mass lesions are highly concerning for pancreatic cancer, which remains highly fatal cancer and the fourth leading cause of cancer-related deaths. The American Cancer Society estimates 48,960 new cases of pancreatic cancer diagnosed in 2015 with an almost equal number of cancer-related deaths. Preoperative imaging of the pancreas has become a critical part of the evaluation, diagnosis, and staging of pancreatic cancer and other pancreatic neoplasms. Early and accurate detection and staging of pancreatic neoplasms allows for curative resection in select patients, and avoidance of surgery in those who would not benefit and in whom further workup is indicated.
With the increased use of cross-sectional imaging, incidentally discovered pancreatic cysts have also become a common clinical problem. An estimated 15% of patients undergoing MRI of the abdomen are found to have a pancreatic cyst. Although cystic neoplasms of the pancreas only account for 1% to 5% of all malignant pancreatic neoplasms, they generate a substantial volume of subsequent imaging and procedures to differentiate between benign and malignant lesions. This imaging, in turn, has led to advances in radiologic and endoscopic-based imaging studies to better characterize the nature of these cysts. This report reviews the range of imaging tools currently available to evaluate pancreatic lesions, from solid tumors to pancreatic cysts. This article reviews noninvasive radiologic imaging and moves to the evolving role of endoscopic ultrasound scan (EUS) and the newer techniques of elastography and contrast-enhanced EUS. Finally, recent device innovations moving the field toward in vivo endoscopic microscopy of the pancreas are discussed.
Introduction
Pancreatic mass lesions are highly concerning for pancreatic cancer, which remains highly fatal cancer and the fourth leading cause of cancer-related deaths. The American Cancer Society estimates 48,960 new cases of pancreatic cancer diagnosed in 2015 with an almost equal number of cancer-related deaths. Preoperative imaging of the pancreas has become a critical part of the evaluation, diagnosis, and staging of pancreatic cancer and other pancreatic neoplasms. Early and accurate detection and staging of pancreatic neoplasms allows for curative resection in select patients, and avoidance of surgery in those who would not benefit and in whom further workup is indicated.
With the increased use of cross-sectional imaging, incidentally discovered pancreatic cysts have also become a common clinical problem. An estimated 15% of patients undergoing MRI of the abdomen are found to have a pancreatic cyst. Although cystic neoplasms of the pancreas only account for 1% to 5% of all malignant pancreatic neoplasms, they generate a substantial volume of subsequent imaging and procedures to differentiate between benign and malignant lesions. This imaging, in turn, has led to advances in radiologic and endoscopic-based imaging studies to better characterize the nature of these cysts. This report reviews the range of imaging tools currently available to evaluate pancreatic lesions, from solid tumors to pancreatic cysts. This article reviews noninvasive radiologic imaging and moves to the evolving role of endoscopic ultrasound scan (EUS) and the newer techniques of elastography and contrast-enhanced EUS. Finally, recent device innovations moving the field toward in vivo endoscopic microscopy of the pancreas are discussed.
Radiologic imaging of the pancreas
Solid Pancreatic Lesions
The most common etiology of a solid pancreatic tumor is adenocarcinoma, which accounts for 85% to 95% of all pancreatic tumors. A few other malignant pancreatic neoplasms in the differential diagnosis of a solid pancreatic tumor have a generally more favorable prognosis compared with adenocarcinoma ( Box 1 ). Noninvasive cross-sectional imaging of the pancreas remains the first-line imaging modality of choice in the evaluation of a pancreatic mass ( Table 1 ). Cross-sectional studies can provide a general assessment of malignancy potential, resectability, presence of lymphadenopathy, and distant metastases.
Pancreatic adenocarcinoma
Pancreatic neuroendocrine tumor
Pancreatoblastoma
Pancreatic lymphoma
Solid pseudopapillary tumor
| Imaging Modality | Sensitivity for Tumor Detection (%) | Specificity (%) | Accuracy of Tumor Nodal Staging (%) |
|---|---|---|---|
| CT | 68–86 | 50–64 | 47–68 |
| MRI | 83–96 | 100 | 56–60 |
| EUS | 87–100 | 75–100 | 44–84 |
Dual-phase multidetector computed tomography (MDCT) is an excellent initial imaging choice in the evaluation of pancreatic masses and is often referred to as pancreatic protocol computed tomography (CT). It provides 1-mm thick cross-sectional images of the pancreas with volume acquisition and allows for 3-dimensional reconstruction and vascular mapping in multiplanar views. The dual contrast phase obtains images of the pancreatic parenchyma during the arterial phase and peripancreatic vasculature during the portal venous phase. Most pancreatic masses are hypoattenuating and best visualized during the portal venous phase, although certain neuroendocrine tumors and metastatic deposits to the pancreas can be hypervascular. Signs used on MDCT to detect a small pancreatic mass even when no overt lesion is visible include concurrent biliary and pancreatic ductal dilation or “double duct” sign for pancreatic head tumors, subtle changes in the contour of the pancreas, or loss of perivascular fat planes. In a head-to-head prospective comparison of the diagnostic yield of MDCT to EUS, DeWitt and colleagues found MDCT to have lower sensitivity of overall cancer detection (86% vs 98% for EUS, P = .01), but MDCT was equivalent to EUS for tumor nodal staging. Although CT imaging overall has a high positive predictive value of more than 90% for tumor detection, it is limited by poor detection of small tumors, critically those tumors in the early resectable stage, small hepatic metastases, and peritoneal implants.
On MRI, pancreatic tumors appear as a hypoenhancing lesion on T1-weighted images compared with the surrounding pancreatic parenchyma. Overall, MRI sensitivity for tumor detection is similar to that of MDCT with reported ranges up to 90%. However MRI may have an advantage of superior detection of smaller tumors and metastases, especially with the addition of newer techniques of diffusion-weighted MRI. One distinct advantage of pancreatic MRI is the ability to obtain magnetic resonance cholangiopancreatography sequences to evaluate for pancreatic ductal dilation and biliary tract pathology, which has comparable diagnostic value for biliary obstruction to endoscopic retrograde cholangiopancreatography.
Pancreatic Cysts
Cystic pancreatic neoplasms account for 1% to 5% of malignant pancreatic neoplasms, but they are increasingly recognized as an incidental finding during abdominal imaging studies obtained for other indications. Because the etiologies of pancreatic cysts encompass a wide range—from benign, inflammatory to malignant—determining the underlying etiology of the cyst is critical for prognostication and planning for either surgical resection or continued surveillance of lesions with low malignant potential ( Box 2 ). Of the etiologies of cystic lesions, pseudocyst and serous cystadenoma are benign lesions with little to no malignant potential. Mucinous cystic neoplasm (MCN) is considered a premalignant lesion with a reported rate of malignant transformation of between 6% and 36%. Intraductal papillary mucinous neoplasms (IPMN) have variable malignant potential depending on the subtype of IPMN with main-duct IPMN (MD-IPMN) having the highest malignant potential of between 57% and 92% in different series.
Serous cystadenoma
Mucinous cystic neoplasm
Intraductal papillary mucinous neoplasm
Pseudocyst
Cystic neuroendocrine tumor
Polycystic disease
Parasitic infection
Solid pseudopapillary neoplasm
Cystic degeneration of necrotic adenocarcinoma or metastases
In general, when an incidental cystic lesion is discovered, the radiologic imaging study of choice to further delineate cyst characteristics is MRI with magnetic resonance cholangiopancreatography because of MRI’s superior soft tissue resolution compared with MDCT and the ability to evaluate the pancreatic duct and biliary tree. MRI can identify pathognomonic features of the cyst subtypes, detect nodules, and main pancreatic duct involvement for risk stratification ( Table 2 ). Retrospective studies show MRI accuracy in correctly classifying cysts as mucinous versus nonmucinous as 79% to 84%, but MRI performs poorly in diagnosing the specific subtype with an accuracy of only 44%. However, MRI accuracy in predicting malignancy in a cystic lesion is reported to be 75%, comparable to EUS in one retrospective series.
| Cyst Type | Clinical Features | Imaging Characteristics |
|---|---|---|
| Pseudocyst | Sequela of acute necrotizing pancreatitis, no sex predominance | Unilocular, encapsulated cyst with a well-defined wall |
| Serous cystadenoma | Older age, female predominance | Multilocular, microcystic appearance, central scar (seen in 30%) is pathognomonic |
| Mucinous cystic neoplasm | Sixth decade of life, almost all female | Multilocular, well encapsulated, smooth margins, preferentially in body or tail of pancreas, internal septations, peripheral calcification (16%) rare but specific for malignancy |
| SPEN | Young age, female predominant | Unilocular, encapsulated, can have solid and hemorrhagic components |
| IPMN | Older age, mild male predominance, often incidental finding | Any region of pancreas, MD-IPMN associated with pancreatic duct dilation >5 mm, dilated side branches seen in BD-IPMN |
Endoscopic ultrasound scan
Solid Pancreatic Mass
EUS is widely considered the most reliable and accurate test in the detection and diagnosis of pancreatic masses, including pancreatic cancer ( Fig. 1 ). Reported sensitivity of EUS in detection of pancreatic cancer is between 94% and 100%. Compared with MDCT, EUS can detect between 13% and 14% of pancreatic cancers that were missed on MDCT. In particular, EUS performs better with detection of tumors smaller than 20 mm where MRI and CT have higher miss rates. For lymph node staging, EUS has an accuracy of 44% to 84% with the lower accuracy attributed to variable appearance of metastatic lymph nodes and adenopathy that are obscured by the primary tumor and its associated inflammatory changes. Current ultrasound criteria for malignant adenopathy include round, hypoechoic, lymph nodes with diameter greater than 1 cm with distinct margins. However, these features can be seen even in benign lymph nodes, and malignant nodes do not uniformly have all these features. The last component of cancer staging is an evaluation of vascular invasion. EUS sensitivity and specificity for vascular involvement is reported to be 56% to 87% and 89% to 97%, respectively. EUS evaluation of the superior mesenteric vessels can be particularly difficult, as the course of these vessels through the uncinate and inferior head can be challenging to see. Nevertheless, a retrospective analysis of Surveillance, Epidemiology, and End Results data found that patients who underwent EUS for pancreatic cancer were more likely to have earlier-stage disease, receive curative-intent surgery, and undergo chemotherapy and radiation therapy. Overall, EUS was an independent predictor of improved survival in those who had it as part of their diagnosis or staging.
Despite the overall reliability and utility of EUS in pancreatic imaging, it does have notable limitations. EUS can miss true pancreatic masses in the setting of chronic pancreatitis, at the site of the ventral/dorsal split of the pancreas, and when there are indwelling biliary or pancreatic stents causing acoustic shadowing. EUS can also misdiagnose a pancreatic tumor in autoimmune pancreatitis, a well-known mimicker of pancreatic cancer that often presents with obstructive jaundice, abdominal pain, and a focal pancreatic head mass. Recently, 2 additional EUS-based imaging modalities, elastography and contrast-enhanced EUS, were studied with variable success in providing additional yield in the evaluation of pancreatic tumors, and some of the limitations of conventional EUS were addressed.
Elastography
Elastography refers to the measurement of relative tissue stiffness to differentiate benign and malignant pancreatic masses. Malignant tissue is thought to have a harder consistency and thus has a different degree of compression compared with benign tissue. During EUS, real-time calculation of tissue elasticity can be performed using the natural arterial pulsations and respiratory movements as the source of pressure; the results are displayed in color superimposed onto conventional B-mode gray-scale scans. Initial reports of the diagnostic accuracy of elastography showed sensitivity of near 100% in both malignant pancreatic mass and lymph nodes but low specificity caused by false-positive results of 67% for malignant lesions and 50% for lymph nodes. Follow-up studies found inconsistent results, with one prospective study of 70 patients with undifferentiated pancreatic masses reporting much lower overall sensitivity of elastography of 41% for malignancy, specificity of 53%, and accuracy of only 45%. Recent efforts to improve the reproducibility, accuracy, and clinical utility of elastography in pancreatic imaging have moved toward developing quantitative scoring systems for elastography to better delineate the relative differences in the elasticity of solid pancreatic masses. Although performing elastography does not add significant time to an EUS examination, there are some limitations to adaptation of this technology. Elastography is not universally available as a built-in feature in all processors. Expensive software is required to supplement this feature into processors such as the Olympus ProSound F75. Furthermore, all endosonographers are not trained in elastography further limiting its availability. Special training in interpretation of elastography should be obtained, as varied interobserver agreement exists and can improve with experience ( Fig. 2 ).
Contrast-enhanced endoscopic ultrasound scan
Use of contrast-enhanced endosonography was initially introduced for ultrasound examinations of hepatic lesions in which microbubbles of inert gas were injected intravenously to enhance visualization of mass vascularity. Similarly, contrast-enhanced EUS (CH-EUS) depends on the differential perfusion or enhancement pattern of pancreatic tumors to identify them, namely the hypovascular appearance of adenocarcinoma compared with the hypervascular appearance of pancreatic neuroendocrine tumor (NET) and isovascular appearance of the pancreatic parenchyma in chronic pancreatitis. A meta-analysis of available data on CH-EUS with 1139 patients shows it has a pooled sensitivity of 94% for the diagnosis of pancreatic adenocarcinoma with specificity of 89%. The meta-analysis included patients with chronic pancreatitis in whom conventional EUS is known to have poorer detection rates. Despite the promising data, the interpretation of CH-EUS images remains subjective, and there is a lack of data on the reproducibility of this novel technique outside of expert hands, thus limiting its widespread adoption. Furthermore, contrast agents are not currently approved or available in countries such as the United States ( Fig. 3 ).
Endoscopic ultrasound scan fine-needle aspiration
The ability of EUS to obtain tissue or core biopsy specimens of pancreatic lesions is one of the most attractive features of endosonography. Before the advent of EUS fine-needle aspiration (EUS-FNA), percutaneous needle aspiration or intraoperative tissue acquisition was needed, but each had its own drawbacks such as malignant needle tract seeding and the invasive morbidity of diagnostic laparoscopy. Overall EUS-FNA has a sensitivity of 85% to 94% and specificity of 100% for pancreatic cancer. EUS-FNA sensitivity is lower in patients with chronic pancreatitis and a focal mass decreasing to 54% in some reports. Much research has focused on improving the cellular yield of FNA sampling based on needle size selection (FNA needles vary from 19 to 25 gauge), increasing the number of needle passes, “fanning” technique, use of suction, and pull-back methods of removing the stylet. What stands out clearly among these variations of technique is that the availability of on-site cytopathology improves the diagnostic yield of FNA, and when this is unavailable, at least 7 FNA passes are needed to achieve a high sensitivity (83%) for diagnosing malignant pancreatic lesions.
Another variation of EUS tissue sampling technique is the development of fine-needle biopsy, in which core biopsy samples are obtained rather than cellular aspirates from FNA. This technique is particularly suited when there is clinical suspicion for lymphoma or stromal tumors where the tissue architecture needs to be preserved and full examination of the tissue histopathology is essential for accurate diagnosis. Additionally, core biopsy samples allow immunohistochemistry staining and other molecular studies to be performed on the sample. One large, multicentered retrospective cohort study of 109 patients with both intraintestinal and extraintestinal lesions sampled using a 19-gauge fine-needle biopsy device, EchoTip ProCore designed by Cook Endoscopy (Limerick, Ireland) had a technical feasibility rate of 98%, adequate tissue sampling for full histologic evaluation in 89% of cases, and an overall diagnostic accuracy for malignancy of 92.9% with no reported complications.
Pancreatic Cysts
Although the combination of clinical and radiologic data can provide an initial diagnosis in the evaluation of pancreatic cysts, EUS can add additional diagnostic value in indeterminate cysts based on their endosonographic features, cyst fluid analysis for amylase, carcinoembryonic antigen (CEA), and cytology. Generally, all symptomatic or obstructing cysts, even when there is low suspicion for malignancy, should be referred for surgical resection or endoscopic drainage in the case of symptomatic pseudocysts. For asymptomatic pancreatic cysts, initial clinical decision making depends on differentiating between mucinous and nonmucinous cysts because of the malignant potential of mucinous cysts. Pseudocysts on EUS appear as unilocular, anechoic lesions arising within pancreatic parenchyma that has changes consistent with acute or chronic pancreatitis ( Fig. 4 ). Fluid aspirate from a pseudocyst has elevated amylase (>5000 U/mL) and lipase (>2000 U/mL) levels. Serous cystadenomas classically have a microcystic or honeycomb appearance with multiple small cysts that together form the lesion and on fluid analysis have a CEA level less than 5 ng/mL ( Fig. 5 ). Mucinous cystadenomas are macrocystic, unilocular lesions usually found in the body and tail with viscous, thick mucin on FNA, which may have markedly elevated CEA level on fluid analysis, although a low CEA level does not exclude the possibility of a mucinous lesion. IPMNs are classified as MD-IPMN, which have a mean risk of malignancy of 61%, and side-branch or branch duct IPMN (BD-IPMN), which have much lower rates of malignancy ( Fig. 6 ). The International Consensus Guidelines of 2012 for management of IPMN and MCN recommend surgical resection of all MD-IPMN in surgically fit patients to prevent development of malignancy. The management of BD-IPMNs is more controversial. EUS examinations of BD-IPMN should focus on the evaluation of high risk features or predictors of malignancy: solid component or mural nodule, main pancreatic duct dilation ≥1 cm, and cyst size ≥3 cm 2 . The 2015 American Gastroenterological Association guidelines on asymptomatic pancreatic cysts recommend EUS examination only when 2 such high-risk features are present and noninvasive surveillance with MRI in all other cases. Prospective validation of these recommendations is lacking, and clinical practice patterns can vary widely in BD-IPMN surveillance.
