RadioGraphics
The radiologic appearance of islet cell tumors varies according to the biologic behavior and natural history of the lesion. Larger masses, such as non-functioning islet cell tumors, non-insulin-producing tumors, and non-gastrin-producing functioning tumors, are usually easily identified and characterized with cross-sectional imaging studies. Small lesions (insulinomas and gastrinomas) are more elusive and may require more invasive or multiple localization techniques.
The radiologist's role in the work-up of patients with insulinomas and gastrinomas is different from what is usually expected. With other patients, the radiologist first detects a previously unknown lesion and then attempts to make a specific diagnosis based on its appearance at various imaging studies. In patients with small, functioning islet cell tumors, the endocrinologist has already established the diagnosis, and the presence of a tumor is implied by the diagnosis. It is the radiologist's job to locate the lesion by using appropriate imaging modalities.
In this regard, insulinoma and gastrinoma must be considered separately. In addition to the differences in epidemiology and biologic behavior described, insulinoma and gastrinoma differ in other ways as well. Over 90% of insulinomas are palpable at surgery (4). They are much more easily localized than are gastrinomas. Because gastrinomas are so often extrapancreatic or multiple, they are much more difficult to detect than are insulinomas. Localization of small, extrapancreatic lesions is so difficult that in the past, despite exhaustive attempts, nearly 40% of patients with gastrinomas have gone to surgery without preoperative identification of the primary lesion (4).
Herein, we do not attempt to explore which localization method or imaging modality is best suited for identifying islet cell tumors. The choice usually depends on the preference, skill, and expertise of the radiologist at each institution. A discussion of the various techniques currently used for localizing and evaluating islet cell tumors follows.
Sonographic Appearance
Transabdominal ultrasound (US) consistently enables identification of larger islet cell tumors, but it is less effective for locating smaller (less than 2 cm in diameter) islet cell tumors (24). The appearance of larger tumors on US scans is variable and depends on the degree of hyalinized stroma, hemorrhage, and cystic degeneration (Figure 2). Cystic areas appear as anechoic or hypoechoic areas with through transmission. Calcification may be seen as isolated hyperechoic areas with shadowing.
Smaller lesions are not readily identified with transabdominal US. The sensitivity of US is only 20%-30% in the detection of gastrinoma and is 20%-75% for insulinoma (25,26,27,28). Most of these lesions appear as round or oval, smoothly marginated masses that are hypoechoic compared with the surrounding pancreatic parenchyma.
Transabdominal US is helpful in detecting metastatic disease to the liver, which may appear hypoechoic to hyperechoic. Hyperechogenicity (Figure 3), when present, is helpful in characterizing and differentiating islet cell metastases from the metastases of primary adenocarcinoma of the pancreas, which usually appear as hypoechoic lesions.
Endoscopic US is performed with a high-frequency (10-mHz) transducer incorporated into an endoscope and placed into the stomach or duodenum, adjacent to the pancreas. With the exception of the pancreatic tail, which is not closely apposed to the stomach, the entire pancreas is well visualized with this technique. Sensitivities of 77%-94% (29,30,31,32,33,34,35,36) are reported with the use of this technique. Lesions seen at endoscopic US have a round or oval contour and decreased echogenicity (Figure 4). Endoscopic US is particularly useful for the detection of smaller lesions.
In intraoperative US, a high-frequency transducer is placed directly on the pancreas. This technique provides greater spatial resolution than does transabdominal US. Islet cell tumors typically appear as hypoechoic, sharply marginated masses (Figure 5). Small cystic areas may be seen (Figure 6). Intraoperative US is a valuable technique for localizing small lesions. It has a greater sensitivity for intrapancreatic lesions than for extrapancreatic lesions and is therefore more helpful for localization of insulinoma than for gastrinoma. The overall sensitivity of intraoperative US ranges between 75% and 100%, and the technique is more sensitive than all other preoperative imaging methods (37,38,39,40).
CT Appearance
CT accurately demonstrates larger islet cell tumors (Figure 5, Figure 6), but it is less reliable for detection of smaller tumors. The detection rate for gastrinoma and insulinoma varies between 30% and 75% (4). CT accurately depicts areas of calcification, necrosis, and cystic degeneration (Figure 2, Figure 7). CT scans obtained during the intravenous administration of contrast material may demonstrate islet cell tumors that are of lower, equal, or higher attenuation than that of the surrounding pancreas. Similar variability in findings is noted on helical CT scans obtained during the arterial and portal venous phases (Figure 8) (4,24,41,42). Use of CT during the intraarterial infusion of contrast material may improve the conspicuity of the lesion, but this technique has not been shown to improve sensitivity (43). Larger lesions demonstrate hypervascular solid components, usually in the periphery, and central nonenhancing areas, which may represent necrosis, fibrosis, or cystic degeneration (Figure 7).
CT is an effective method for detecting liver metastases and vascular invasion (Figure 9, Figure 10, Figure 11). Hypervascular hepatic metastases from islet cell tumors may become isoattenuating relative to the liver after administration of contrast material (44). However, recent studies in which more rapid CT scanning was used have demonstrated that there is no significant difference in the detection of hypervascular metastases with the use of contrast-enhanced and unenhanced images (45,46).
MR Imaging Appearance
MR imaging has rapidly improved as a method for imaging the pancreas. New imaging sequences have overcome the previous drawbacks of motion artifacts caused by respiration and peristalsis in the surrounding bowel. MR imaging also offers additional information that may help characterize islet cell tumors. Pancreatic islet cell tumors typically demonstrate very high signal intensity on T2-weighted images and fat-suppressed inversion-recovery images (Figure 2, Figure 6, Figure 7, Figure 12). Furthermore, MR imaging performed with gadolinium is more sensitive to tumor vascularity than is CT performed with intravenous contrast material and therefore permits the detection and characterization of lesions better than CT (Figure 2, Figure 12) (29,47,48).
MR imaging is also useful in the detection of liver metastases and may be superior to CT and US for this purpose (Figure 11) (28,47).
Scintigraphic Appearance
Somatostatin receptor imaging with octreotide, a somatostatin analogue (Sandostatin; Sandoz, East Hanover, NJ), permits scintigraphic detection of many lesions that contain somatostatin receptors (Figure 13). However, lesion detection is not predictable for a number of reasons. Although most pancreatic endocrine tumors express somatostatin receptors, only about 60%-70% of insulinomas are somatostatin receptor positive (30,49,50,51). Somatostatin receptor imaging is generally not helpful for detecting insulinomas. In addition, uptake of the radiotracer is theoretically based on the number of somatostatin receptors, rather than on the absolute size of the tumor. Therefore, large tumors with few somatostatin receptors may go undetected, but small tumors with numerous somatostatin receptors can be imaged clearly.
Octreotide scintigraphy may also be used to screen for metastatic disease, especially extrahepatic metastases (26,32,51). In a prospective study of 80 patients with gastrinomas, somatostatin receptor imaging was shown to be the most sensitive of the localization techniques evaluated; it was superior to CT, MR imaging, US, and angiography in the detection of both primary gastrinoma and metastases (52).
Angiographic Appearance
With the advent of cross-sectional imaging techniques, angiography is used less frequently for the localization and work-up of islet cell tumors. Virtually all islet cell tumors demonstrate an intense stain in the late arterial and capillary phases (Figure 5, Figure 6) (4). Larger tumors, which may contain central areas of necrosis or cystic degeneration, demonstrate hypervascularity in the viable peripheral portions of the tumor. This feature is helpful for distinguishing islet cell tumors from other pancreatic neoplasms, particularly adenocarcinoma, which are hypovascular. Smaller, solid lesions demonstrate an intense homogeneous blush, and even lesions as small as 5 mm may be visualized (24). Hepatic arteriography is helpful in demonstrating liver metastases as small as 5 mm in diameter (Figure 3) (53).
Portal Venous Sampling
Portal venous sampling is a technique that requires percutaneous transhepatic catheterization of the portal vein and selective venous sampling along the pancreatic venous arcade and the small pancreatic veins that drain into the splenic vein. In this technique, the hormone concentration of each sample collected is plotted on a diagram of the portal venous system (4,54). The location of the lesion is indirectly localized by identifying the vein in which the highest concentration of hormone is detected (Figure 14). Because of variations in venous drainage and because the procedure is difficult and time-consuming, it has been replaced with selective arterial stimulation testing (SAST).
Selective Arterial Stimulation Testing
SAST provides information similar to that obtained through portal venous sampling but does not require selective catheterization of the portal venous system. In SAST, a catheter is placed in the right hepatic vein from a femoral vein approach, and a second catheter is used to selectively catheterize the arteries supplying the pancreas. Secretagogues (drugs that stimulate hormone production by the specific islet cell tumor) are injected selectively into arterial branches that supply the pancreas or the liver. Blood samples are drawn from the hepatic vein at intervals of 1-2 minutes after the arterial injection (Figure 15). Hepatic venous blood reflects the hormonal composition of the pancreatic venous effluent. A twofold increase in the hormone concentration in the hepatic vein helps identify the artery supplying the tumor and, in turn, the approximate location of the lesion.
Different secretagogues are used for different tumors. Secretin is used to stimulate gastrinomas, and calcium is used to stimulate insulinomas. This technique is used to find small tumors that are not detected with other modalities (55,56,57,58,59).