- Email: [email protected]
Somatostatin receptor scintigraphy: A first–line imaging modality for gastroenteropancreatic neuroendocrine tumors
October 1998
of patients or more. The pathogenesis of pancreatitis in the setting of a endoscopic retrograde cholangiopancreatography (ERCP) (with or without sphincterotomy) is unclear. A leading hypothesis suggests that local trauma to the papilla results in swelling and edema and this in turn leads to an obstruction to pancreatic flow with subsequent pancreatitis. Elta et al. randomized 170 patients undergoing endoscopic sphincterotomy to either a ‘‘blended’’ group (blended ‘‘three’’ setting on a Valleylab electrosurgical unit (Valleylab, Boulder, CO) with power settings of 30 W for pure and coagulation current) or ‘‘pure current’’ group (30 W) in a blinded fashion. Overall there were fewer complications in the pure current group than in the blended group (5% vs. 14%; P , 0.05). Only 3 cases of mild pancreatitis occurred in the pure group (3%) compared with a total of 10 cases (7 mild, 2 moderate, and 1 severe) in the blended group. Although only 1 case of bleeding requiring transfusion occurred in each group, patients in the pure group were observed to have more (mild) bleeding at the time of the procedure. The investigators concluded that the use of pure cut current is associated with a lower incidence of pancreatitis. Comment. The two most common serious complications of endosocopic sphincterotomy are pancreatitis and bleeding (N Engl J Med 1996;335:909–918). Various patient-dependent risk factors have been implicated in the development of these complications (Gastroenterology 1991;101:1068–1075, Pancreas 1991;6:350–367), but the role of current-induced injury has not been well defined. Broadly speaking, two types of electrical currents are available for use in endoscopic procedures. Pure cutting current, which results in much more rapid cutting and less edema, has traditionally not been used for sphincterotomy because of a fear of inducing significant bleeding from uncauterized vessels. Therefore, coagulation current, either by itself or blended with cutting current, has been preferred. The theoretical advantage of better hemostasis with this method may be offset by greater injury to local tissue from coagulation and edema, which in turn could lead to clinical significant complications such as pancreatitis. However, this hypothesis has not been well tested, and most standard teaching on this subject advocates the use of coagulation current in some form. Investigators at Johns Hopkins were the first to describe the potential benefits of sphincterotomy using a pure cutting current (Gastrointest Endosc 1994;40:66). In a randomized blinded trial they showed the following: (1) recannulation of the bile duct was easier after sphincterotomy with pure cutting current than after sphincterotomy with blended current; (2) sphincterotomy with either pure cutting or blended currents results in a similar mild decrease in hemoglobin level; (3) the use of blended but not pure cutting current for sphincterotomy results in a significant increase in postprocedural serum amylase level and; (4) resolution of abnormal liver function test results and bilirubin levels is swifter after sphincterotomy with pure cutting current. Although the number of patients in that study was inadequate to assess the impact of current settings on the risk of clinical pancreatitis, it was clear that the use of a pure cutting current for sphincterotomy had the potential for reducing local tissue damage without increasing the risk for bleeding. The investigators of the present study have now taken this issue further and shown that the use of pure current may indeed decrease the risk of pancreatitis. In a relatively simple randomized study, they showed an impressive fourfold reduction in the incidence of pancreatitis in the pure current group. This beneficial effect was observed
SELECTED SUMMARIES
1025
without a concurrent increase in bleeding complications, although, as the authors point out, this may have been caused by inadequate statistical power. Unfortunately, we are not told if they studied additional parameters such as resolution of obstructive-pattern liver function tests or ease of recannulation immediately after sphincterotomy. This would not only have added to the clinical value of this study but would also have strengthened the hypothesis that the risk of pancreatitis is related to tissue injury induced by the current (and not due to some other effect such as stimulation of locoregional neuroendocrine reflexes). Despite more than two decades of ERCP and sphincterotomy, pancreatitis remains an important and poorly understood complication of this procedure. Nevertheless, glimmers of progress in this area have been observed recently. Thus, prophylactic gabexate, a potent protease inhibitor, has been shown to reduce the risk of pancreatitis after ERCP. Although this approach may not be practical or costeffective, adjusting current settings at the time of sphincterotomy is a simple enough matter. We have been using a pure current for most cases of sphincterotomy for several years without any significant increase in bleeding complications. It should be borne in mind, however, that although a pure current setting during sphincterotomy may diminish the frequency of pancreatitis, it will clearly not eliminate the risk that remains significant even for a diagnostic ERCP (N Engl J Med 1996;335:909–918). Elta et al. are to be congratulated on successfully carrying out an important and provocative study that should stimulate other much needed research on this subject. PANKAJ PASRICHA, M.D.
Reply. Although we did not formally study and therefore not report in the study, we did find that pure cut electrocautery for sphincterotomy simplifies recannulation due to a significant decrease in local edema. Occasionally the increased minor bleeding at the sphincterotomy from pure cut current causes a temporary loss of the endoscopic view, although this disadvantage is outweighed by the lower incidence of postprocedural pancreatitis and the improved recannulation rate. We did not assess the resolution of obstructive-pattern liver function tests. We are currently comparing pure cut electrocautery current for sphincterotomy with a combination of pure cut current for the first half of the sphincterotomy and a blended current for the remainder, because most bleeding occurs at the end of the cut. GRACE H. ELTA, M.D.
SOMATOSTATIN RECEPTOR SCINTIGRAPHY: A FIRST–LINE IMAGING MODALITY FOR GASTROENTEROPANCREATIC NEUROENDOCRINE TUMORS Lebtahi R, Cadiot G, Sarda L, Daou D, Faraggi M, Petegnief Y, Mignon M, Le Guludec D (Departments of Medicine and Nuclear Medicine, Hoˆpital Bichat, Paris, France). Clinical impact of somatostatin receptor scintigraphy in the management of patients with neuroendocrine gastroenteropancreatic tumors. J Nucl Med 1997:38:853–858. Gastroenteropancreatic neuroendocrine tumors (e.g., gastrinoma and carcinoid) are slow growing and despite the associated peptide hypersecretion often remain undetected for years. Tumor localization is often difficult because the primary
1026
SELECTED SUMMARIES
tumors are frequently small, but is of paramount importance because surgical resection remains the preferred treatment in most patients without metastases. The discovery of highaffinity somatostatin receptors on neuroendocrine tumors has led to the use of radiolabeled somatostatin analogues for the detection of these tumors. The present study was designed to prospectively evaluate the clinical impact of somatostatin receptor scintigraphy (SRS) in the detection and management of patients with gastroenteropancreatic neuroendocrine tumors. The impact of SRS was analyzed in 160 consecutive patients (72 women and 88 men) with biologically and/or histologically proven gastroenteropancreatic neuroendocrine tumors seen from November 1992 to September 1995. The study population consisted of 78 patients with gastrinoma, 38 patients with carcinoid, and 44 patients with other types of neuroendocrine tumors. Diagnosis of gastrinoma was based on histopathology and specific biological syndrome (n 5 60) or only specific biological syndrome (n 5 18). The diagnosis of all other tumors was histologically confirmed. Thus, histological confirmation of tumors was obtained in 142 of 160 patients (89%). Chest radiography, contrast-enhanced abdominal computerized tomography (CT), and abdominal ultrasonography were performed in all patients. Endoscopic ultrasonography (EUS) of the duodenopancreatic area was performed in 59 patients, magnetic resonance imaging of the abdomen in 17 patients, and abdominal angiography in 9 patients. To evaluate the additional value of SRS in patient management, patients were classified into three groups according to the presence or absence of liver and/or extrahepatic metastases on the basis of conventional imaging results, before SRS. Group 1 consisted of 90 patients without detected metastases, group 2 consisted of 59 patients with metastases limited to the liver, and group 3 consisted of 11 patients with known extrahepatic metastases. SRS was performed using 111In–diethylenetriaminopentaacetic acid (DTPA) D-phenylalanine-1-octreotide (135 MBq). Abdominal images were obtained at 4 hours after injection, in the anterior and posterior views. At 24 hours, the acquisition included anterior and posterior views for the head, chest, and pelvis, and anterior, posterior, lateral, and oblique views of the abdomen. At 30–48 hours after injection, delayed images of the abdomen were obtained in the anterior and posterior views. The images were visually analyzed by two blinded independent observers with disagreements resolved by consensus. Conventional imaging including EUS was positive in 114 of the 160 patients (71%), and SRS was positive in 125 patients (78%). Although the global detection rate was similar, SRS was positive in 28 of 46 patients (61%) with previously undetected tumors on the basis of conventional imaging, and was negative in only 17 of the 114 patients (15%) with known tumor sites. In the overall population of 160 patients, 111 new sites were discovered by SRS in 80 patients; 37 of these sites (12 celiac, 3 liver, 9 chest, 10 bone, and 3 abdomen) were confirmed during follow-up either by conventional imaging and/or surgical findings.
GASTROENTEROLOGY Vol. 115, No. 4
In the duodenogastric area of the 21 patients with surgical confirmation of tumor, CT was positive in 1 patient, EUS was positive in 11 patients, SRS was positive in 11 patients, and a combination of EUS and SRS was positive in 19 patients (90%). In group 1 patients (90 patients without known metastases), SRS detected 22 primary tumors in 17 of 46 patients (37%) without detectable tumors by conventional imaging and discovered 29 metastatic sites in 25 patients (28%). In group 2 patients (59 patients with hepatic metastases only by conventional imaging), SRS confirmed these metastases in 56 patients (95%), 18 extrahepatic metastases in 13 patients (22%), and new liver sites in 5 patients (8%). In group 3 patients (11 patients with known extrahepatic metastases), SRS missed 5 tumor sites in 3 patients (27%; 2 in the lung and 3 in the abdomen) and discovered 7 new sites in the duodenopancreatic area in 4 patients and 4 new metastatic sites in 3 patients. Overall, SRS discovered 11 new sites in 7 patients (64%). The investigators conclude that SRS improves tumor detection, alters clinical therapeutic decision making, and should be performed in patients with gastroenteropancreatic tumors. Comment. Gastroenteropancreatic neuroendocrine tumors are rare, slow-growing tumors that often present late, with metastases, most often in the liver, and with symptoms more related to endocrine hypersecretion than neoplastic growth. The majority of these tumors arise in the pancreas; others originate from the mucosa of the stomach, duodenum, intestine, or retroperitoneum. These tumors have been called ‘‘APUDomas’’ because they arise from cells that share APUD characteristics, i.e., they are capable of amine precursor uptake and then decarboxylation into amines and/or peptides. The release of these bioactive substances is responsible for many of the presenting signs and symptoms of these tumors. The most frequently encountered gastroenteropancreatic neuroendocrine tumors are gastrinomas, carcinoids, VIPomas, glucagonomas, insulinomas, and somatostatinomas. Only 10% of insulinomas are malignant, but more than 60% of gastrinomas, VIPomas, and glucagonomas, and nearly 100% of somatostatinomas are malignant (Digestion 1993;54[suppl 1]:68– 71). The treatment of choice for most of these tumors is surgical excision; however, a curative resection is frequently not possible because of multicentricity, particularly in the multiple endocrine neoplasia type 1 syndrome, and metastases. Somatostatin is a peptide existing in two different forms, a 14–amino acid and a 28–amino acid molecule. It is abundantly present in the gastrointestinal tract, in endocrine cells and neurons, with highest concentrations in the stomach, upper small intestine, and pancreas. Five different somatostatin receptor subtypes (sst1, sst2, sst3, sst4, and sst5), belonging to the G protein–coupled receptor superfamily, have been cloned (Trends Pharmacol Sci 1995;16:86–88). All five human subtypes bind somatostatin 14 and somatostatin 28 with high affinity. The short-chain somatostatin analogues such as octreotide bind sst2 receptors with high affinity, sst3 and sst5 receptors with intermediate affinity, and sst1 and sst4 receptors with little or nonexistent affinity (EMBO J 1995;14:727–735). Most gastroenteropancreatic neuroendocrine tumors, with the exception of insulinoma, contain an abundance of sst2 messenger RNA and express a high density of sst2 receptors (Cancer Res 1994;54:3455–3459, Nucl Med Annu 1995:1–50, Metabolism 1996;45[suppl 1]:39–41); consequently, they avidly bind radiolabeled octreotide.
October 1998
SELECTED SUMMARIES
In 1989, Lamberts et al. (Lancet 1989;1:242–244, N Engl J Med 1990;323:1246–1249) developed a technique that allowed in vivo visualization of tumors in humans after intravenous administration of 123I coupled to octreotide and was able to detect primary tumors or metastases in 12 of 13 patients with carcinoids and 7 of 9 patients with pancreatic neuroendocrine tumors. Although promising, this technique was hampered by the difficulty in preparing pure 123I, by the short half-life of the radionuclide, and by hepatobiliary clearance of the radiolabeled peptide, which created a high background of abdominal radioactivity and made localization of small tumors difficult (Gastroenterology 1993;105:1909–1914). These limitations have been overcome by the development of 111In-DTPA–labeled somatostatin analogues, which are easier to prepare, have longer half-lives, and are excreted primarily by the kidneys (Life Sci 1991;49:1593–1601, Semin Oncol 1994;21[Suppl 13]:6–14). Using 111In-DTPA-D-PHE1-octreotide in 40 patients with immunohistologically proven gastroenteropancreatic neuroendocrine tumors, mostly carcinoids, Scheru¨bl et al. (Gastroenterology 1993;105:1705–1709) reported successful visualization of the tumors in 80% of patients; in 16 of 40 patients (40%) additional tumors were localized that were not detected by ultrasonography, CT scanning, or magnetic resonance imaging. In the European Multicenter Trial (Nucl Med Annu 1995:1–50), SRS visualized 297 of 388 (77%) known sites and revealed another 166 unsuspected lesions in 308 patients (54%); 40% of these unsuspected lesions were subsequently confirmed as true positives based on the results of additional imaging procedures or histology obtained during the follow-up period. In the present study by Lebtahi et al., which included 160 patients, almost half with gastrinoma, SRS was confirmed to be accurate, visualizing 84% of 196 tumor sites detected by conventional imaging and discovering 111 unsuspected new tumor sites in 50% of the patients. Although the demonstration of additional tumor sites is unlikely to alter the management of patients with known metastases, it carries critical clinical relevance for the management of patients with a single known lesion or without any known lesion because it may obviate the need for surgery or alter the type of surgery performed. In the present study, therapeutic management was altered in 25% of the patients due to the discovery of unsuspected liver tumors in 7 patients, contralateral liver tumors before hepatectomy in 2 patients, and extrahepatic metastases in 31 patients. SRS with 111In-DTPA-PHE1-octreotide, also known as 111Inpenetreotide and commercially available as Octreoscan (Mallinckrodt Medical, Petten, The Netherlands, and St. Louis, MO), has come of age as a safe and accurate first-line imaging modality for localizing and determining the extent of spread of gastroenteropancreatic neuroendocrine tumors. The cost and radiation exposure are similar to that of a CT scan of the chest, yet SRS images the entire body, is relatively easy to interpret, has a high power of resolution for tumors as small as 1 cm, and detects tumors unsuspected by CT in 40%–60% of patients. Accurate staging by SRS impacts on the choice of therapy (surgery with curative or palliative intent, medical treatment with somatostatin analogues, hepatic artery embolization, or systemic chemotherapy) offered to the patient. For tumors in the duodenopancreatic area, EUS may be a valuable adjunct; a combination of EUS and SRS detected 90% of primary neuroendocrine tumors in the present study. MITCHELL L. SCHUBERT, M.D.
Reply. Several studies have now shown that Octreoscan scintigraphy is the first-line imaging modality for the detection of endocrine tumors. It evaluates the spread of the disease, and its results alter management in many patients (J Nucl Med 1995;36:542–549, Gut
1027
1997;41:107–114, Gastroenterology 1997;112:335–347, J Nucl Med 1997;38:853–858). The ability of Octreoscan to alter management varies according to the clinical setting. In patients without metastases, in whom the goals of imaging are to detect unknown liver metastases and to find the primary tumor, Octreoscan is the first-line technique. It is the most sensitive imaging modality to detect liver metastases (Ann Intern Med 1996;125:26–34), and its negative predictive value is very high (Radiology 1997;202:151–158). Although its sensitivity for the detection of primary tumor is not as high as for the detection of liver metastases, Octreoscan is often the only imaging technique that visualizes midgut carcinoids (Surgery 1994;116:1112–1122). Ninety percent of duodenopancreatic tumors, especially gastrinoma, are detected by the combination of Octreoscan and EUS (Gastroenterology 1996;111:845–854). In patients with advanced disease, Octreoscan can detect metastases that would contraindicate curative liver surgery or local treatment such as liver embolization. Finally, Octreoscan can identify multiple endocrine neoplasia type 1–related tumors, especialy in the chest (Gut 1997;41:107–114). R. LEBTAHI G. CADIOT M. MIGNON D. LE GULUDEC
INTERFERON IN WONDERLAND Bennett WG, Inoue Y, Beck JR, Wong JB, Pauker SG, Davis GL (University of Florida College of Medicine, Gainesville, Florida; Baylor College of Medicine, Houston, Texas; Yamaguchi University School of Medicine, Ube, Yamaguchi, Japan; and New England Medical Center and Tufts University School of Medicine, Boston, Massachusetts). Estimates of the costeffectiveness of a single course of interferon-a-2b in patients with histologically mild chronic hepatitis C. Ann Intern Med 1997;127:855–865. Noting that there are 3.9 million people chronically infected with the hepatitis C virus (HCV) and that 8300 deaths and 730 additional liver transplantations occur annually as a consequence, Bennett et al. performed a decision analysis to attempt to assess the impact that interferon (IFN) treatment might have on the long-term course and costs of this infection in the subgroup of patients having histologically mild chronic disease. Using data from the literature, they created a model predicting the long-term course of the disease. Employing data from five prospective randomized controlled trials of IFN (and the individual patient databases provided by Schering-Plough, the sponsor of those trials), they calculated that 27% of patients with mild or moderate chronic hepatitis would have a desirable long-term response (i.e., become seronegative for markers of active HCV infection). As is the case in all such analyses, the authors had to make assumptions when actual data were not available. They assumed that patients who lose HCV would not develop progressive liver disease. They also apparently assumed that the course of the disease in nonresponders was the same as in those who had never been treated. To make some of their cost comparisons, they asked a panel of hepatologists to make quality-of-life adjustments for various states of disease. (These
of patients or more. The pathogenesis of pancreatitis in the setting of a endoscopic retrograde cholangiopancreatography (ERCP) (with or without sphincterotomy) is unclear. A leading hypothesis suggests that local trauma to the papilla results in swelling and edema and this in turn leads to an obstruction to pancreatic flow with subsequent pancreatitis. Elta et al. randomized 170 patients undergoing endoscopic sphincterotomy to either a ‘‘blended’’ group (blended ‘‘three’’ setting on a Valleylab electrosurgical unit (Valleylab, Boulder, CO) with power settings of 30 W for pure and coagulation current) or ‘‘pure current’’ group (30 W) in a blinded fashion. Overall there were fewer complications in the pure current group than in the blended group (5% vs. 14%; P , 0.05). Only 3 cases of mild pancreatitis occurred in the pure group (3%) compared with a total of 10 cases (7 mild, 2 moderate, and 1 severe) in the blended group. Although only 1 case of bleeding requiring transfusion occurred in each group, patients in the pure group were observed to have more (mild) bleeding at the time of the procedure. The investigators concluded that the use of pure cut current is associated with a lower incidence of pancreatitis. Comment. The two most common serious complications of endosocopic sphincterotomy are pancreatitis and bleeding (N Engl J Med 1996;335:909–918). Various patient-dependent risk factors have been implicated in the development of these complications (Gastroenterology 1991;101:1068–1075, Pancreas 1991;6:350–367), but the role of current-induced injury has not been well defined. Broadly speaking, two types of electrical currents are available for use in endoscopic procedures. Pure cutting current, which results in much more rapid cutting and less edema, has traditionally not been used for sphincterotomy because of a fear of inducing significant bleeding from uncauterized vessels. Therefore, coagulation current, either by itself or blended with cutting current, has been preferred. The theoretical advantage of better hemostasis with this method may be offset by greater injury to local tissue from coagulation and edema, which in turn could lead to clinical significant complications such as pancreatitis. However, this hypothesis has not been well tested, and most standard teaching on this subject advocates the use of coagulation current in some form. Investigators at Johns Hopkins were the first to describe the potential benefits of sphincterotomy using a pure cutting current (Gastrointest Endosc 1994;40:66). In a randomized blinded trial they showed the following: (1) recannulation of the bile duct was easier after sphincterotomy with pure cutting current than after sphincterotomy with blended current; (2) sphincterotomy with either pure cutting or blended currents results in a similar mild decrease in hemoglobin level; (3) the use of blended but not pure cutting current for sphincterotomy results in a significant increase in postprocedural serum amylase level and; (4) resolution of abnormal liver function test results and bilirubin levels is swifter after sphincterotomy with pure cutting current. Although the number of patients in that study was inadequate to assess the impact of current settings on the risk of clinical pancreatitis, it was clear that the use of a pure cutting current for sphincterotomy had the potential for reducing local tissue damage without increasing the risk for bleeding. The investigators of the present study have now taken this issue further and shown that the use of pure current may indeed decrease the risk of pancreatitis. In a relatively simple randomized study, they showed an impressive fourfold reduction in the incidence of pancreatitis in the pure current group. This beneficial effect was observed
SELECTED SUMMARIES
1025
without a concurrent increase in bleeding complications, although, as the authors point out, this may have been caused by inadequate statistical power. Unfortunately, we are not told if they studied additional parameters such as resolution of obstructive-pattern liver function tests or ease of recannulation immediately after sphincterotomy. This would not only have added to the clinical value of this study but would also have strengthened the hypothesis that the risk of pancreatitis is related to tissue injury induced by the current (and not due to some other effect such as stimulation of locoregional neuroendocrine reflexes). Despite more than two decades of ERCP and sphincterotomy, pancreatitis remains an important and poorly understood complication of this procedure. Nevertheless, glimmers of progress in this area have been observed recently. Thus, prophylactic gabexate, a potent protease inhibitor, has been shown to reduce the risk of pancreatitis after ERCP. Although this approach may not be practical or costeffective, adjusting current settings at the time of sphincterotomy is a simple enough matter. We have been using a pure current for most cases of sphincterotomy for several years without any significant increase in bleeding complications. It should be borne in mind, however, that although a pure current setting during sphincterotomy may diminish the frequency of pancreatitis, it will clearly not eliminate the risk that remains significant even for a diagnostic ERCP (N Engl J Med 1996;335:909–918). Elta et al. are to be congratulated on successfully carrying out an important and provocative study that should stimulate other much needed research on this subject. PANKAJ PASRICHA, M.D.
Reply. Although we did not formally study and therefore not report in the study, we did find that pure cut electrocautery for sphincterotomy simplifies recannulation due to a significant decrease in local edema. Occasionally the increased minor bleeding at the sphincterotomy from pure cut current causes a temporary loss of the endoscopic view, although this disadvantage is outweighed by the lower incidence of postprocedural pancreatitis and the improved recannulation rate. We did not assess the resolution of obstructive-pattern liver function tests. We are currently comparing pure cut electrocautery current for sphincterotomy with a combination of pure cut current for the first half of the sphincterotomy and a blended current for the remainder, because most bleeding occurs at the end of the cut. GRACE H. ELTA, M.D.
SOMATOSTATIN RECEPTOR SCINTIGRAPHY: A FIRST–LINE IMAGING MODALITY FOR GASTROENTEROPANCREATIC NEUROENDOCRINE TUMORS Lebtahi R, Cadiot G, Sarda L, Daou D, Faraggi M, Petegnief Y, Mignon M, Le Guludec D (Departments of Medicine and Nuclear Medicine, Hoˆpital Bichat, Paris, France). Clinical impact of somatostatin receptor scintigraphy in the management of patients with neuroendocrine gastroenteropancreatic tumors. J Nucl Med 1997:38:853–858. Gastroenteropancreatic neuroendocrine tumors (e.g., gastrinoma and carcinoid) are slow growing and despite the associated peptide hypersecretion often remain undetected for years. Tumor localization is often difficult because the primary
1026
SELECTED SUMMARIES
tumors are frequently small, but is of paramount importance because surgical resection remains the preferred treatment in most patients without metastases. The discovery of highaffinity somatostatin receptors on neuroendocrine tumors has led to the use of radiolabeled somatostatin analogues for the detection of these tumors. The present study was designed to prospectively evaluate the clinical impact of somatostatin receptor scintigraphy (SRS) in the detection and management of patients with gastroenteropancreatic neuroendocrine tumors. The impact of SRS was analyzed in 160 consecutive patients (72 women and 88 men) with biologically and/or histologically proven gastroenteropancreatic neuroendocrine tumors seen from November 1992 to September 1995. The study population consisted of 78 patients with gastrinoma, 38 patients with carcinoid, and 44 patients with other types of neuroendocrine tumors. Diagnosis of gastrinoma was based on histopathology and specific biological syndrome (n 5 60) or only specific biological syndrome (n 5 18). The diagnosis of all other tumors was histologically confirmed. Thus, histological confirmation of tumors was obtained in 142 of 160 patients (89%). Chest radiography, contrast-enhanced abdominal computerized tomography (CT), and abdominal ultrasonography were performed in all patients. Endoscopic ultrasonography (EUS) of the duodenopancreatic area was performed in 59 patients, magnetic resonance imaging of the abdomen in 17 patients, and abdominal angiography in 9 patients. To evaluate the additional value of SRS in patient management, patients were classified into three groups according to the presence or absence of liver and/or extrahepatic metastases on the basis of conventional imaging results, before SRS. Group 1 consisted of 90 patients without detected metastases, group 2 consisted of 59 patients with metastases limited to the liver, and group 3 consisted of 11 patients with known extrahepatic metastases. SRS was performed using 111In–diethylenetriaminopentaacetic acid (DTPA) D-phenylalanine-1-octreotide (135 MBq). Abdominal images were obtained at 4 hours after injection, in the anterior and posterior views. At 24 hours, the acquisition included anterior and posterior views for the head, chest, and pelvis, and anterior, posterior, lateral, and oblique views of the abdomen. At 30–48 hours after injection, delayed images of the abdomen were obtained in the anterior and posterior views. The images were visually analyzed by two blinded independent observers with disagreements resolved by consensus. Conventional imaging including EUS was positive in 114 of the 160 patients (71%), and SRS was positive in 125 patients (78%). Although the global detection rate was similar, SRS was positive in 28 of 46 patients (61%) with previously undetected tumors on the basis of conventional imaging, and was negative in only 17 of the 114 patients (15%) with known tumor sites. In the overall population of 160 patients, 111 new sites were discovered by SRS in 80 patients; 37 of these sites (12 celiac, 3 liver, 9 chest, 10 bone, and 3 abdomen) were confirmed during follow-up either by conventional imaging and/or surgical findings.
GASTROENTEROLOGY Vol. 115, No. 4
In the duodenogastric area of the 21 patients with surgical confirmation of tumor, CT was positive in 1 patient, EUS was positive in 11 patients, SRS was positive in 11 patients, and a combination of EUS and SRS was positive in 19 patients (90%). In group 1 patients (90 patients without known metastases), SRS detected 22 primary tumors in 17 of 46 patients (37%) without detectable tumors by conventional imaging and discovered 29 metastatic sites in 25 patients (28%). In group 2 patients (59 patients with hepatic metastases only by conventional imaging), SRS confirmed these metastases in 56 patients (95%), 18 extrahepatic metastases in 13 patients (22%), and new liver sites in 5 patients (8%). In group 3 patients (11 patients with known extrahepatic metastases), SRS missed 5 tumor sites in 3 patients (27%; 2 in the lung and 3 in the abdomen) and discovered 7 new sites in the duodenopancreatic area in 4 patients and 4 new metastatic sites in 3 patients. Overall, SRS discovered 11 new sites in 7 patients (64%). The investigators conclude that SRS improves tumor detection, alters clinical therapeutic decision making, and should be performed in patients with gastroenteropancreatic tumors. Comment. Gastroenteropancreatic neuroendocrine tumors are rare, slow-growing tumors that often present late, with metastases, most often in the liver, and with symptoms more related to endocrine hypersecretion than neoplastic growth. The majority of these tumors arise in the pancreas; others originate from the mucosa of the stomach, duodenum, intestine, or retroperitoneum. These tumors have been called ‘‘APUDomas’’ because they arise from cells that share APUD characteristics, i.e., they are capable of amine precursor uptake and then decarboxylation into amines and/or peptides. The release of these bioactive substances is responsible for many of the presenting signs and symptoms of these tumors. The most frequently encountered gastroenteropancreatic neuroendocrine tumors are gastrinomas, carcinoids, VIPomas, glucagonomas, insulinomas, and somatostatinomas. Only 10% of insulinomas are malignant, but more than 60% of gastrinomas, VIPomas, and glucagonomas, and nearly 100% of somatostatinomas are malignant (Digestion 1993;54[suppl 1]:68– 71). The treatment of choice for most of these tumors is surgical excision; however, a curative resection is frequently not possible because of multicentricity, particularly in the multiple endocrine neoplasia type 1 syndrome, and metastases. Somatostatin is a peptide existing in two different forms, a 14–amino acid and a 28–amino acid molecule. It is abundantly present in the gastrointestinal tract, in endocrine cells and neurons, with highest concentrations in the stomach, upper small intestine, and pancreas. Five different somatostatin receptor subtypes (sst1, sst2, sst3, sst4, and sst5), belonging to the G protein–coupled receptor superfamily, have been cloned (Trends Pharmacol Sci 1995;16:86–88). All five human subtypes bind somatostatin 14 and somatostatin 28 with high affinity. The short-chain somatostatin analogues such as octreotide bind sst2 receptors with high affinity, sst3 and sst5 receptors with intermediate affinity, and sst1 and sst4 receptors with little or nonexistent affinity (EMBO J 1995;14:727–735). Most gastroenteropancreatic neuroendocrine tumors, with the exception of insulinoma, contain an abundance of sst2 messenger RNA and express a high density of sst2 receptors (Cancer Res 1994;54:3455–3459, Nucl Med Annu 1995:1–50, Metabolism 1996;45[suppl 1]:39–41); consequently, they avidly bind radiolabeled octreotide.
October 1998
SELECTED SUMMARIES
In 1989, Lamberts et al. (Lancet 1989;1:242–244, N Engl J Med 1990;323:1246–1249) developed a technique that allowed in vivo visualization of tumors in humans after intravenous administration of 123I coupled to octreotide and was able to detect primary tumors or metastases in 12 of 13 patients with carcinoids and 7 of 9 patients with pancreatic neuroendocrine tumors. Although promising, this technique was hampered by the difficulty in preparing pure 123I, by the short half-life of the radionuclide, and by hepatobiliary clearance of the radiolabeled peptide, which created a high background of abdominal radioactivity and made localization of small tumors difficult (Gastroenterology 1993;105:1909–1914). These limitations have been overcome by the development of 111In-DTPA–labeled somatostatin analogues, which are easier to prepare, have longer half-lives, and are excreted primarily by the kidneys (Life Sci 1991;49:1593–1601, Semin Oncol 1994;21[Suppl 13]:6–14). Using 111In-DTPA-D-PHE1-octreotide in 40 patients with immunohistologically proven gastroenteropancreatic neuroendocrine tumors, mostly carcinoids, Scheru¨bl et al. (Gastroenterology 1993;105:1705–1709) reported successful visualization of the tumors in 80% of patients; in 16 of 40 patients (40%) additional tumors were localized that were not detected by ultrasonography, CT scanning, or magnetic resonance imaging. In the European Multicenter Trial (Nucl Med Annu 1995:1–50), SRS visualized 297 of 388 (77%) known sites and revealed another 166 unsuspected lesions in 308 patients (54%); 40% of these unsuspected lesions were subsequently confirmed as true positives based on the results of additional imaging procedures or histology obtained during the follow-up period. In the present study by Lebtahi et al., which included 160 patients, almost half with gastrinoma, SRS was confirmed to be accurate, visualizing 84% of 196 tumor sites detected by conventional imaging and discovering 111 unsuspected new tumor sites in 50% of the patients. Although the demonstration of additional tumor sites is unlikely to alter the management of patients with known metastases, it carries critical clinical relevance for the management of patients with a single known lesion or without any known lesion because it may obviate the need for surgery or alter the type of surgery performed. In the present study, therapeutic management was altered in 25% of the patients due to the discovery of unsuspected liver tumors in 7 patients, contralateral liver tumors before hepatectomy in 2 patients, and extrahepatic metastases in 31 patients. SRS with 111In-DTPA-PHE1-octreotide, also known as 111Inpenetreotide and commercially available as Octreoscan (Mallinckrodt Medical, Petten, The Netherlands, and St. Louis, MO), has come of age as a safe and accurate first-line imaging modality for localizing and determining the extent of spread of gastroenteropancreatic neuroendocrine tumors. The cost and radiation exposure are similar to that of a CT scan of the chest, yet SRS images the entire body, is relatively easy to interpret, has a high power of resolution for tumors as small as 1 cm, and detects tumors unsuspected by CT in 40%–60% of patients. Accurate staging by SRS impacts on the choice of therapy (surgery with curative or palliative intent, medical treatment with somatostatin analogues, hepatic artery embolization, or systemic chemotherapy) offered to the patient. For tumors in the duodenopancreatic area, EUS may be a valuable adjunct; a combination of EUS and SRS detected 90% of primary neuroendocrine tumors in the present study. MITCHELL L. SCHUBERT, M.D.
Reply. Several studies have now shown that Octreoscan scintigraphy is the first-line imaging modality for the detection of endocrine tumors. It evaluates the spread of the disease, and its results alter management in many patients (J Nucl Med 1995;36:542–549, Gut
1027
1997;41:107–114, Gastroenterology 1997;112:335–347, J Nucl Med 1997;38:853–858). The ability of Octreoscan to alter management varies according to the clinical setting. In patients without metastases, in whom the goals of imaging are to detect unknown liver metastases and to find the primary tumor, Octreoscan is the first-line technique. It is the most sensitive imaging modality to detect liver metastases (Ann Intern Med 1996;125:26–34), and its negative predictive value is very high (Radiology 1997;202:151–158). Although its sensitivity for the detection of primary tumor is not as high as for the detection of liver metastases, Octreoscan is often the only imaging technique that visualizes midgut carcinoids (Surgery 1994;116:1112–1122). Ninety percent of duodenopancreatic tumors, especially gastrinoma, are detected by the combination of Octreoscan and EUS (Gastroenterology 1996;111:845–854). In patients with advanced disease, Octreoscan can detect metastases that would contraindicate curative liver surgery or local treatment such as liver embolization. Finally, Octreoscan can identify multiple endocrine neoplasia type 1–related tumors, especialy in the chest (Gut 1997;41:107–114). R. LEBTAHI G. CADIOT M. MIGNON D. LE GULUDEC
INTERFERON IN WONDERLAND Bennett WG, Inoue Y, Beck JR, Wong JB, Pauker SG, Davis GL (University of Florida College of Medicine, Gainesville, Florida; Baylor College of Medicine, Houston, Texas; Yamaguchi University School of Medicine, Ube, Yamaguchi, Japan; and New England Medical Center and Tufts University School of Medicine, Boston, Massachusetts). Estimates of the costeffectiveness of a single course of interferon-a-2b in patients with histologically mild chronic hepatitis C. Ann Intern Med 1997;127:855–865. Noting that there are 3.9 million people chronically infected with the hepatitis C virus (HCV) and that 8300 deaths and 730 additional liver transplantations occur annually as a consequence, Bennett et al. performed a decision analysis to attempt to assess the impact that interferon (IFN) treatment might have on the long-term course and costs of this infection in the subgroup of patients having histologically mild chronic disease. Using data from the literature, they created a model predicting the long-term course of the disease. Employing data from five prospective randomized controlled trials of IFN (and the individual patient databases provided by Schering-Plough, the sponsor of those trials), they calculated that 27% of patients with mild or moderate chronic hepatitis would have a desirable long-term response (i.e., become seronegative for markers of active HCV infection). As is the case in all such analyses, the authors had to make assumptions when actual data were not available. They assumed that patients who lose HCV would not develop progressive liver disease. They also apparently assumed that the course of the disease in nonresponders was the same as in those who had never been treated. To make some of their cost comparisons, they asked a panel of hepatologists to make quality-of-life adjustments for various states of disease. (These