Diagnosis of Pancreatic Neuroendocrine Tumors and the Role of Endoscopic Ultrasound

Pancreatic neuroendocrine tumors (PETs) such as the one described in the interesting case report by Adebajo and associates¹ are exceedingly rare but generate much clinical interest due to their protean and often dramatic manifestations. They comprise 1–2% of all pancreatic neoplasms. The annual incidence of clinically symptomatic PETs is less than 1 in 100,000, although the prevalence of asymptomatic PETs in autopsy studies is higher, ranging from 0.8% to 10%.^2,3 Nonfunctioning PETs are most common, accounting for 70–90% of all PETs,³ followed by insulinomas and gastrinomas, whereas vasoactive intestinal peptide-producing tumors, glucagonomas, somatostatinomas, and others occur very rarely, in decreasing order of incidence. Among the other hormones secreted by PETs extremely rarely are growth hormone–releasing factor, parathyroid hormone–related protein, and adrenocorticotropic hormone (ACTH). Although PETs do occur sporadically, they have a much higher incidence (30–75%) in patients with the hereditary syndrome multiple endocrine neoplasia-type 1 (MEN-1). Other inherited syndromes less commonly associated with PETs include von Hippel-Lindau disease, von Reckling-hausen disease, and tuberous sclerosis.

Clinically, PETs are classified as nonfunctional or functional, depending upon the absence or presence of clinical symptoms related to hormone release. Therefore, nonfunctional PETs may not secrete hormones, or the secreted hormones may not produce symptoms. Accordingly, nonfunctional PETs are larger and more often malignant at presentation than functioning PETs.⁴ Malignancy is defined by the invasion of adjacent organs, spread to lymph nodes, or presence of distant metastases, and is not strictly based upon histology. The World Health Organization classifies PETs into the following categories: well-differentiated endocrine tumors with benign or unknown behavior; well-differentiated endocrine carcinomas with low-grade malignant potential; or poorly differentiated endocrine carcinomas with high-grade malignant potential.⁵ A tumor lymph node metastasis (TNM) classification analogous to the TNM system for other solid tumors has also been developed.⁶ Both appear to accurately predict long-term survival for PETs.⁷ Recent data have also suggested the ability to prognosticate from endoscopic ultrasound–fine needle aspirate (EUS-FNA) specimens of PETs. Malignant PETs contained significantly greater DNA microsatellite losses than benign lesions, and more microsatellite loss was associated with higher 2-year recurrence and lower 5-year survival.⁸

Other factors influencing prognosis include the type of PET, primary tumor size and location, extent or rate of growth of liver metastases, presence of bone metastases, histologic features, high proliferative indices, flow cytometric features, and development of ectopic Cushing syndrome. Although insulinomas have low malignant potential (5–15%), the other PETs have a much higher malignant potential, ranging from 50% to 90%, with a greater-than-90% malignancy rate in ACTH-producing PETs. ACTH-producing PETs metastasize early (occasionally even before clinical manifestation of Cushing syndrome) and frequently to the liver, leading to a poor 5-year survival rate of 16%.⁹ Therefore, it is certainly unusual that the patient in this case study did not present with metastatic disease.

Diagnosis of functional PETs is guided by clinical symptomatology. For suspected insulinomas, serum glucose, insulin, proinsulin, and C peptide levels should be checked, with a prolonged 72-hour fast as the gold standard for diagnosis. In potential gastrinomas, fasting serum gastrin, basal acid output, and secretin tests may be used for diagnosis. Serum vasoactive intestinal peptide, glucagon, and somatostatin levels are diagnostic for their respective PETs. Diagnosis of ectopic ACTH-PET, as outlined in this case, involves documentation of hypercortisolism, typically with late-night serum cortisol, followed by demonstration of ACTH dependence with elevated ACTH, and finally unresponsiveness to glucocorticoid feedback using the high-dose dexa-methasone suppression test and corticotropin-releasing hormone test, which implies an ectopic, nonpituitary source of ACTH. Chromogranin A is widely used to diagnose and follow especially nonfunctional PETs, with a sensitivity of 60–100% in metastatic disease but only 50% for local disease.¹⁰

With a biochemical diagnosis, radiologic studies are critical to identify the location of the tumor as well as metastases in order to guide appropriate management. This approach can be challenging, particularly with functional tumors, which are often small. Traditional radiologic imaging with computed tomography (CT) scan is 64–82% sensitive for detecting the primary tumor, whereas magnetic resonance imaging (MRI) has an equivalent or superior sensitivity of 74–100%.¹¹ Classically, PETs demonstrate hypervascular enhancement during the arterial phase of CT due to their vascular nature. Because PETs often contain somatostatin receptors, functional imaging with somatostatin receptor scintigraphy (SRS) identifies 50–70% of primary tumors, with the exception of insulinomas that express somatostatin receptors in only approximately half of cases.¹⁰ Other limitations of SRS include reduced accuracy in localizing the tumor within the pancreas, differentiating between an intrapancreatic lesion and a peripancreatic lymph node, and detecting small PETs less than 1 cm in size. Positron-emission tomographic scanning with standard substrates such as ¹⁸F-deoxyglucose is ineffective due to the low metabolic activity of most PETs; however, use with ¹¹C-5-hydroxy-tryptophan (¹¹C-5-HTP) or ⁶⁸Ga labeled somatostatin analogues appears very promising.¹² With the advent of functional imaging and ever-improving CT and MRI, invasive angiographic techniques involving arterial stimulation with secretagogues and subsequent selective hepatic venous sampling are infrequently utilized.

Given the limitations of the current radiologic studies, EUS has become an integral part of the diagnosis of PETs because of its high sensitivity for detecting, localizing, and diagnosing pancreatic PETs. In fact, when a lesion is not visualized on CT scan in patients with PETs, sensitivity of EUS-FNA for diagnosing PETs is 70%.¹³ In an older series of 82 patients with suspected PETs due to clinical, biochemical, or radiologic evidence, the sensitivity, specificity, and accuracy of EUS imaging for localizing PETs was 93%, 95%, and 93%, respectively.¹⁴ Other studies have reported sensitivity rates of EUS ranging from 83% to 94% for detecting PETs.^15,16

Most commonly, PETs appear hypoechoic, round, homogeneous, and well defined on EUS, though they may be isoechoic and, on rare occasions, hyperechoic with irregular margins. Malignant PETs are larger, with irregular margins, compared to benign PETs. Cystic lesions are the least common presentation, accounting for 8–17% of PETs, and may be unilocular, septated, microcystic, or mixed solid-cystic.^17,18 Compared to solid PETs, cystic PETs are twice as large, more often symptomatic, 3.5 times more likely to be associated with MEN-1, and more likely nonfunctional.¹⁸ Approximately 81% of cystic PETs are nonfunctional, and diagnosis with radiologic or EUS imaging alone is unlikely due to the lack of distinguishing characteristics of these cystic lesions. Therefore, cytologic and immunohistochemical evaluation of EUS-FNA specimens is essential and appears to have similar sensitivity as EUS-FNA of solid PETs.¹³

The addition of FNA to EUS using a 22- or 25-gauge needle enables tissue diagnosis, which allows differentiation from pancreatic adenocarcinoma and is more relevant for diagnosis of nonfunctioning or cystic PETs. One could argue that FNA was not essential in the current case, given a biochemical diagnosis of ectopic ACTH syndrome with a mass seen on EUS. Two recent large studies of EUS-FNA in PETs reported overall 87–90% sensitivity, with one study finding diminished sensitivity (66%) for tumors that were smaller than 15 mm or benign, though the other publication failed to replicate these results.^13,19 Although cytomorphology alone was adequate for diagnosis in one of these studies, the use of immunohistochemistry on cytology specimens, as in this case report, may aid in cytologic diagnosis of PETs.

To improve FNA yield, ideally onsite cytopathology examination should be performed. This examination significantly reduces the rate of unsatisfactory cytology specimens from 20% to 9%.²⁰ If a cytopathologist is unavailable, 5–7 passes should be performed for pancreatic masses (as in this case), 2–3 for liver metastases, and 2–5 for lymph nodes.^21,22 Use of a 22- versus a 25-gauge needle does not affect diagnostic yield, though comparative experience specifically with PETs is limited.²³ The trucut needle biopsy uses a 19-gauge needle to obtain core biopsies. Despite interest in this technique, studies have not consistently demonstrated superior diagnostic yield, and technical failures occur, particularly with the duodenal approach.²⁴ Therefore, lesions in the pancreatic head and uncinate process are difficult to access, and, again, there are inadequate data with PETs. EUS-guided brush of pancreatic cysts appears to have similar or possibly superior diagnostic yield to FNA; however, it may carry an increased bleeding risk.²⁵

Despite EUS and improved radiologic imaging, small PETs may be difficult to localize in the operating room. Intraoperative palpation combined with intraoperative ultrasound is over 95% sensitive; however, they prolong operative time, have rarely been associated with splenic vessel rupture from manipulation of the pancreas, and are not practical with laparoscopic resections.²⁶ Tattooing the lesion during EUS appears to be safe, according to small case series. The agents injected into the pancreas have included presterilized, diluted, and filtered India ink, indocyanine green, methylene blue, and GI Spot.^27,28 EUS-fine needle injection with GI Spot lasted up to 83 days after injection, and operative time was significantly reduced, with no repeat surgery necessary in tattooed patients.²⁸

In conclusion, PETs are unusual entities offering an extraordinary glimpse into the workings of various hormones, including, on rare occasions, ACTH. Diagnosis of functional PETs usually relies upon biochemical and imaging studies, particularly EUS, given the smaller size of functional tumors. Nonfunctional PETs are more readily detected with radiology, though they will typically require EUS-FNA for definitive diagnosis. It is clear that endoscopy with EUS and EUS-FNA has become a cornerstone in the diagnosis of these fascinomas.