Prospective Study of the Ability of Computed Axial Tomography to Localize Gastrinomas in Patients With Zollinger-Ellison Syndrome

Transcription

1 GASTROENTEROLOGY 1987;92; Prospective Study of the Ability of Computed Axial Tomography to Localize Gastrinomas in Patients With Zollinger-Ellison Syndrome S. A. WANK, J. L. DOPPMAN, D. L. MILLER, M. J. COLLEN, P. N. MATON, R. VINAYEK, J. I. SLAFF, J. A. NORTON, J. D. GARDNER, and R. T. JENSEN Digestive Diseases Branch, National Institute of Arthritis, Diabetes, and Digestive and Kidney Diseases; Department of Radiology, Clinical Center; and Surgery Branch, National Cancer Institute, National Institutes of Health, Bethesda, Maryland The ability of routine computed tomography (CT) performed with oral and intravenous contrast to localize gastrinomas in 61 consecutive patients with Zollinger-Ellison syndrome was evaluated prospectively. The results of CT scanning were subsequently evaluated in all patients by either surgery, autopsy, or percutaneous biopsy. Thirteen of 14 patients with CT scans positive for hepatic metastases and 5 of 13 patients with CT scans negative for hepatic metastases were found to have gastrinoma in the liver. For gastrinoma metastatic to the liver, CT scanning had a specificity of 98%, a sensitivity of 72%, a positive predictive value of 93%, and a negative predictive value of 90%. Twenty-two of 23 patients with positive extrahepatic CT scans and 15 of 33 patients with negative extrahepatic CT scans were found to have extrahepatic gastrinomas. For extrahepatic gastrinoma, CT scanning had a specificity of 95%, a sensitivity of 59%, a positive 'predictive value of 96%, and a negative predictive value of 54%. The ability of CT scan to detect gastrinomas both in the liver and extrahepatically was directly related to tumor size, detecting 0% of tumors <1 cm and 83%-95% of tumors >3 cm. The location of the extrahepatic gastrinoma was also an important determinant in that ~80% of pancreatic gastrinomas but only 35% of extrapancreatic gastrinomas were detected. The present results indicate that because of its convenience and accuracy, CT scanning with Received March 26, Accepted October 16, Address requests for reprints to: Dr. R. T. Jensen, Building 10, Room 9C-l03, National Institutes of Health, Bethesda, Maryland The authors thank Mary O'Shaughnessy for preparing this manuscript for publication /87/$3.50 oral and intravenous contrast material should be the initial procedure to evaluate the extent of gastrinoma. A positive CT scan is almost always correct; therefore, a CT scan detecting metastatic gastrinoma to the liver would avoid unnecessary surgery and, if positive for extrahepatic gastrinoma, would assist the surgeon in finding the gastrinoma. A negative CT is less reliable; therefore, patients should undergo other localizing studies before exploratory laparotomy. In early studies most patients with Zollinger-EilisOIi syndrome died from complications related to gastric acid hypersecretion (1). With the general availability of gastrin radioimmunoassays leading to earlier diagnosis (2), and the development of both medical and surgical means to effectively control gastric acid hypersecretion, complications due to the gastrinoma itself are becoming increasingly important determinants of long-term survival (2). A number of factors indicate that it is important to determine preoperatively the location and extent of the gastrinoma if appropriate therapy is to be undertaken. First, in most studies only 10%-20% of patients are cured surgically (3-6) because the gastrinomas are frequently multiple or extrapancreatic and iri about 50% of patients without metastatic disease (2,3,7,8) gastrinomas are not found. Second, in most studies 10%-44% of patients are found to have metastatic disease, usually to the liver, at the time of surgery (1,5,6). Because total gastrectomy is now rarely required to control gastric acid hypersecretion (2,7) if metastatic disease can be identified preoperatively, Abbreviation used in this paper: CT, computed axial tomography.

2 906 WANK ET AL. GASTROENTEROLOGY Vol. 92, No.4 an unnecessary laparotomy can be prevented. Third, in early studies patients with metastatic disease were reported to have a 5-yr survival of 42% and a 10-yr survival of 30% (9). However, a recent study reports the importance of determining the extent of the gastrinoma prior to therapy, because patients with extensive disease had a 5-yr survival of only 20% and thus chemotherapy should be considered in these patients (10). This study also identified a small group of patients with extensive disease who might be helped by surgical resection (10). Several imaging studies such as selective abdominal angiography, computed axial tomography (CT), and ultrasound have been used to investigate the location and extent of gastrinomas with variable results in different series (2,3,8). Of these imaging procedures, CT scanning has the advantage of convenience, widespread availability, and good resolution for both pancreatic and extrapancreatic gastrinomas, but in previous studies specificity and sensitivity have ranged from 0% to 100% (3,6, 7,11-15). The results ofthese studies were limited in different cases because (a) only small numbers of patients were included; (b) consecutive cases were not evaluated in a prospective fashion; (c) other islet cell tumors, which are rarely extrapancreatic in location, were included with gastrinomas, which are frequently extrapancreatic in location; (d) patients with Zollinger-Ellison syndrome with multiple endocrine neoplasia-type 1, who frequently have multiple tumors, were not differentiated from patients without multiple endocrine neoplasia-type 1; (e) all patients did not undergo surgery; (f) the CT technique was not standardized (use of oral and intravenous contrast material); and (g) in many cases older generation CT scanners were used (2-4,6,13,15,16). The purpose of the present study was to examine prospectively the ability of CT scanning to localize gastrinbmas preoperatively in-. patients with Zollinger-Ellison syndrome using a standardized protocol. We report here the results from 61 consecutive patients with Zollinger-Ellison syndrome who underwent CT scanning and subsequent tissue evaluation by either surgery, autopsy, or percutaneous biopsy. Materials and Methods Patients Seventy-five consecutive patients referred to the National Institutes of Health with a diagnosis of Zollinger-Ellison syndrome were considered for the present study. Ten patients were excluded from the present study as they were not considered surgical candidates because of age or other concomitant medical illness. Before 1983 all patients (n = 5) with multiple endocrine neoplasia-type 1 and Zollinger-Ellison syndrome were included for the full protocol. Since 1983 such patients with negative imaging studies (CT scan, ultrasound, or selective angiography) were not considered surgical candidates and 4 such patients were excluded. The 61 remaining patients (38 men, 23 women) were included in the study. Each patient satisfied the following criteria for the diagnosis of Zollinger-Ellison syndrome: basal acid output 2:15 meq/h in patients without previous gastric surgery or 2:5 meq/h in patients who had previous acid-reducing surgery, elevated fasting serum gastrin concentration (>100 pg/ml), and a diagnostic rise in the serum gastrin concentration following provocative testing with secretin or calcium as outlined previously (7,17). Each patient was part of an ongoing study approved by the Clinical Research Committee of the National Institute of Arthritis, Diabetes, and Digestive and Kidney Diseases, National Institutes of Health. Protocol and Investigation The diagnostic evaluation of all patients included the following studies: measurement of fasting serum gastrin concentration by Bioscience Laboratories (New York, N.Y.) using Walsh gastrin antibody No (lot 4A) which recognizes gastrin 17 and gastrin 34 equally, secretin and calcium provocative testing performed as outlined previously (7,17), and basal and Histalogstimulated (Histalog, Eli Lilly and Company, Indianapolis, Ind.) maximal acid output in the absence of anti secretory medication as described previously (18). The duration of disease was calculated from the onset of symptoms until the date of CT scan and is expressed in months. The antisecretory drug dosage requirement was that amount of anti secretory medication necessary to suppress gastric acid secretion to <10 meq/h in the final hour before the next scheduled dose of medication determined as described pl'eviously (18). This quantity is expressed in terms of milligrams per 24 h, although the actual dose schedule was a function of the total prescribed at intervals varying from 4 to 24 h. Each patient underwent a CT scan of the upper abdomen before the definitive evaluation for the presence of gastrinoma (surgery, autopsy, or liver biopsy). Nine patients between 1977 and 1980 were evaluated using the EMI 5005 whole body scanner. The remaining 52 patients were evaluated using the GE 8800 and GE 9800 whole body scariners. Patients were routinely scanned after oral water-soluble iodinated contrast material (Gastrografin) Intravenous contrast, either 60% Conray or Angiovist, was injected as a bolus of 100 ml in the antecubital vein over a 3-5-min period. Computed axial tomography scans of the upper abdomen from the dome of the liver to the inferior margin of the liver were performed using l-cm sections. Computed axial tomography scans over the head, body, and tail of the pancreas were performed using l-cm sections with a 5-mm overlap between adjacent sections. All CT scans were evaluated by a single radiologist O.L.D.) within 3 mo prior to surgery, autopsy, or liver biopsy in all patients.

3 April 1987 CT SCAN IN ZOLLINGER-ELLISON SYNDROME 907 All patients whose preoperative evaluation did not suggest metastatic gastrinoma (n = 51) underwent exploratory laparotomy, regardless of whether a gastrinoma was localized. At laparotomy the pancreas, stomach, duodenum, liver, and mesenteric and retroperitoneal regions in the upper abdomen were carefully explored as reported previously (19). Patients with suspected metastatic gastrinoma in the liver who were not candidates for attempted cure or debulking surgery (10) underwent either percutaneous or laparoscopic liver biopsy (n = 5). Five patients died with metastatic diseaes and were evaluated by autopsy. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated for the results from CT scanning (20). Results from surgery, autopsy, or liver biopsy served as the standard of reference. Results The pertinent clinical and laboratory characteristics of the 61 patients studied are given in Table 1. These patients closely resemble the clinical and laboratory characteristics for all patients with Zollinger-Ellison syndrome reported in the literature (2,7) in terms of age, sex distribution, disease duration, mean antisecretory drug dose, mean basal and maximal acid output, and both fasting and provoked serum gastrin concentration. Unlike previous reports (2,7), this series has a low percentage (10%) of patients with multiple endocrine neoplasiatype 1, because only those patients with multiple endocrine neoplasia-type 1 with a positive preoperative localization study for gastrinoma underwent surgery after In view of the clinical implications of gastrinoma metastatic to the liver, the results of CT evaluation for gastrinoma metastatic to the liver and gastrinoma occurring outside the liver (extrahepatic) will be evaluated separately. The results of CT identification of gastrinoma metastatic to the liver are shown in Table 2 where the results of CT are compared with the subsequent presence or absence of gastrinoma found at surgery, autopsy, or liver biopsy. Fourteen patients had a positive CT scan for gastrinoma in the liver. In each patient there was more than one area of involvement with a typical CT appearance of gastrinoma metastatic to the liver (shown in Figure 1). Thirteen of the 14 patients with positive CT scans were found to have gastrinoma in the liver, 9 by surgery, 3 by autopsy, and 1 by liver biopsy (Table 2). Thus, the CT had a positive predictive value of 93% (Table 3). Forty-seven patients had a negative CT scan of the liver (Table 2). Five of these patients, however, were found to have gastrinoma in the liver, 4 by surgery and 1 by autopsy (Table 2). Thus, the CT had a negative predictive value of 90% (Table 3) in terms Table 1. Clinical and Laboratory Characteristics of Patients With Zollinger-Ellison Syndrome Patients Number (n) Sex Male Female Age Mean (yr) Range (yr) MEN-Ib Present (n) Disease duration c Mean (mo) Range (mo) Basal acid output (meq/h) Mean Range Maximal acid output (meq/h) Mean Range Serum gastrin (pg/ml) Mean Range Positive secretin/calcium provocative test (%) Antisecretory drug dosed Histamine H 2 -receptor antagonist daily dose (mg/24 h) Cimetidine (n = 28) Ranitidine (n = 25) Anticholinergic (% required) H+, K+ -ATPase inhibitor daily dose (mg/24 h) Omeprazole (n = 3) ± ± ± ± ,094 ± 26, , ,032 ± 1,837 2,761 ± 1, ± 23 CT, computed axial tomography; H+, K+ -ATPase, hydrogen-potassium-stimulated adenosine triphosphatase. Mean, plus or minus one standard deviation from the mean. b MEN-I = multiple endocrine neoplasia, type 1. C Disease duration = time in months from onset of symptoms until CT scan. d Drug doses are listed in dose/24 h. Except for omeprazole, which was taken once daily, all drugs were taken every 4-8 h. Five patients underwent total gastrectomy and took no medication. of reliably predicting that a gastrinoma would not be found. The CT had a sensitivity of 72% for localizing gastrinoma in the liver (Table 3), having identified 13 of the 18 patients with gastrinoma found in the liver at surgery, autopsy, or liver biopsy. The CT had a specificity of 98%, having falsely identified a gastrinoma in the liver in only 1 of 43 patients without evidence of gastrinoma in the liver at surgery, autopsy, or liver biopsy. In most patients with a positive CT scan for extrahepatic gastrinoma a single gastrinoma was identified. A typical CT appearance of extrahepatic gastrinoma is illustrated in Figure 2. The results of CT identification of extrahepatic gastrinomas are summarized in Table 4 where the results of CT are compared with the subsequent presence or absence of gastrinoma found at surgery or autopsy. Twenty-

4 908 W ANK ET AL. GASTROENTEROLOGY Vol. 92, No.4 Table 2. Evaluation of Results of Computed Axial Tomography for Gastrinoma Metastatic to the Liver CT results Positive Negative CT results Positive Negative Surgery (n = 13)0 9 b 4 Surgery (n = 38)0 Gastrinoma found at Liver Autopsy biopsy (n = 4) (n = 1) Gastrinoma not found at Liver Autopsy biopsy (n = 1) (n = 4) Total (n = 18) 13 5 Total (n = 43) CT, computed axial tomography. 0 n refers to the number of patients in whom the CT result was evaluated by the indicated methods. b Numbers in the table refer to the number of patients evaluated by the indicated method with the indicated CT result. three patients had a positive CT scan for extrahepatic gastrinoma, and gastrinoma was found in 17 at surgery and 5 at autopsy. Thus, the CT has a positive predictive value of 96% in terms of reliably predicting that a gastrinoma would be found (Table 3). Thirty-three patients had a negative CT scan for extrahepatic gastrinoma (Table 4). Fifteen of these patients, however, were found to have an extrahepatic gastrinoma at surgery. Thus, the CT had a negative predictive value of 54% in terms of reliably predicting that a gastrinoma would not be found (Table 3). The CT had a sensitivity of 59% for localizing extrahepatic gastrinoma (Table 3), having identified 22 of 37 patients with gastrinoma found at 1 42 Table 3. Sensitivity, Specificity, and Predictive Value of Computed Axial Tomography CT results for gastrinoma In the Outside liver the liverb Parameter O (%) (%) Specificity Sensitivity Positive predictive value Negative predictive value CT, computed axial tomography. 0 Specificity, sensitivity, and positive and negative predictive values were calculated as described in Methods on the basis of results from gastrinoma found at surgery, autopsy, or liver biopsy. b Gastrinomas found either in the pancreas, gastrointestinal tract, or lymph glands. surgery or autopsy. The CT had a specificity of 95%, having falsely identified an extrahepatic gastrinoma in 1 of 19 patients with no evidence of extrahepatic gastrinoma at surgery or autopsy. To evaluate the extent to which CT results were dependent on the size of the gastrinoma, the ability of CT to identify gastrinomas of different size (greatest dimension determined at surgery or autopsy) was examined in Table 5. For gastrinoma metastatic to the liver no tumors (0/2) <1 em in greatest dimension were detected, whereas 75% (3/4) of tumors 1-3 em and 83% (10/12) of tumors >3 em in size were detected. Similarly, in the case of extrahepatic gastrinoma, no tumors (0/5) <1 em in diameter were detected. Unlike gastrin om a metastatic to the liver, gastrinomas of 1-3 em in size were detected in only 30% (3/10) of the cases, whereas gastrinomas >3 em were detected in 95% (18/19) of the cases. To determine whether CT scanning localized all extrahepatic Figure 1. Computed axial tomography scan from a patient with gastrinoma metastatic to the liver. The location of the metastatic gastrinoma is indicated by the arrows. Figure 2. Computed axial tomography scan demonstrating the appearance of a gastrinoma in the pancreatic head. The location of the gastrinoma is labeled "tumor" and indicated by the arrow.

5 April 1987 CT SCAN IN ZOLLINGER-ELLISON SYNDROME 909 Table 4. Evaluation of Results of Computed Axial Tomography for Extrahepatic Gastrinoma Gastrinoma found at Surgery Autopsy Total CT results (n = 32)0 (n = 5) (n = 37) Positive 17 b 5 22 Negative Gastrinoma not found at Surgery Autopsy Total CT results (n = 19)0 (n = 0) (n = 19) Positive 1b 0 1 Negative CT, computed axial tomography. 0 n refers to the number of patients in whom the CT result was evaluated by the indicated method. b Numbers in the table refer to the number of patients evaluated by the indicated method with the indicated CT result. gastrinomas equally well, the ability of CT scanning to localize pancreatic and extrapancreatic gastrinomas was examined (Table 6). Thirty-nine patients were found to have extrahepatic gastrinomas at surgery or autopsy. Twenty-two of these 39 gastrinomas were in the pancreas and 17 were outside the pancreas (stomach, duodenum, lymph nodes, or renal capsule). The CT scan localized 77% (17/22) of pancreatic gastrinotnas and 35% (6/17) of extrapancreatic gastrinomas (Table 6). However, only 1 of the 22 pancreatic gastrinomas was $1 cm, whereas 6 of the 17 extrapancreatic gastrinomas were ::.;1 cm. All 5 of these small ($1 cm) extrahepatic gastrinomas were not detected on CT scan. For extrahepatic gastrinomas, the CT scan correctly localized the gastrinoma to a specific organ in 18 of the 22 gastrinomas detected. However, in 4 patients the CT scan was unable to localize the gastrinoina to one of several adjacent structures such as pancreatic head, duodenal wall, or adjacent lymph node. The CT scan of 1 of these patients is shown in Figure 3. In this patient a gastrinoma was localized by the CT scan to the pancreatic head. However, at surgery the gastrinoma was found in a lymph node in the pancreatic head area, but not in the pancreas. To assess the possibility that patients with a truepositive CT scan differed in clinical and laboratory characteristics from patients with a false-negative CT scan, the two groups were compared. Patients with true-positive and false-negative CT results for either metastatic gastrinoma to the liver or extrahepatic gastrinoma did not differ significantly in terms of disease duration, fasting serum gastrin concentrations, basal or maximal acid output, or anti secretory drug requirements (data not shown). Table 5. Evaluation of the Ability of Computed Axial Tomography to Detect Gastrinoma of Different Sizes CT results Positive Negative % Positive" CT results Positive Negative % Positive c <1 Ob 2 0 <1 Ob 5 0 Gastrinoma in the liver Tumor size (cm)o 1-3 >3 No data Gastrinoma outside the liver Tumor size (cm)o 1-3 >3 No data CT, computed axial tomography. a Size of the tumor in its longest dimension. b Number of patients with a gastrinoma of the indicated size and CT results. C Percent of the total CT scans positive for a gastrinoma of the indicated size. Discussion The results of the present prospective study in 61 consecutive patients with Zollinger-Ellison syndrome demonstrate that routine CT scanning is much more useful in localizing and determining the extent of gastrinoma than reported in most previous studies. In earlier studies routine CT scanning gave a relatively low yield, localizing gastrinomas in only 0%-43% of patients, with frequent false negatives, as well as occasional false-positive results (6,11,12). In the present study, CT scanning had a sensitivity of 72% for identifying metastatic gastrinoma to the liver and a sensitivity of 59% for identifying extrahepatic gastrinoma that was subsequently found at surgery or autopsy. Furthermore, false-positive results were very uncommon in that CT scanning had a specificity of 98% for determining tht=l absence of metastatic gastrinoma to the liver and 96% for extrahepatic gastrinoma subsequently determined at sur- Table 6. Correlation of Gastrinoma Location and Results of Computed Axial Tomography Location of gastrinoma o CT results Pancreatic Extrapancreatic b Positive 17 6 Negative 5 11 % Positive o Based on findings at surgery or autopsy. b Gastrinomas outside both the liver and pancreas were in the wall of the stomach or duodenum (n=8), lymph nodes (n=7), renal capsule (n=1), or ovary (n=1).

6 910 WANK ET AL. GASTROENTEROLOGY Vol. 92, No.4 Figure 3. Computed axial tomography scan from a patient with Zollinger-Ellison syndrome found to have tumor in a lymph node in the pancreatic head. The location of the gastrinoma is labeled "tumor" and indicated by the arrow. The gastrinoma on the basis of the CT scan was localized to the pancreatic head but subsequently at surgery was found to be in a l ymph node adjacent to the pancreatic head. gery or autopsy. A number of factors probably contribute to the improved results in the present study compared to earlier studies using routine CT. In earlier studies older generation CT scanners were frequently used which give poorer resolution and longer scanning times (11-13). Recent studies have demonstrated that the ability to localize gastrinomas can be enhanced by holus infusions of contrast, with up to 80% demonstrating enhancement (15). In earlier studies, unlike the present study, bolus contrast infusions were not routinely performed (11). Because in previous studies poor results were reported with routine CT scanning, a number of more elaborate, time-consuming CT scanning techniques were described to improve localization of islet cell tumors (3,13,15). Dynamic CT scanning, utilizing rapid bolus contrast injection and rapid sequence scanning, was reported to have' a sensitivity of 43%-78% for functioning islet cell tumors, and a specificity of 100% (3,13,15). These results are almost identical to the present results utilizing routine CT scanning with bolus contrast injection. Because of the relative rarity of Zollinger-Ellison syndrome, most previous studies of the ability of CT scanhihg to localize and determine the extent of gastrinoma have involved small numbers of selected cases (3,11-14). A number of recent findings suggest that, in contrast to the present study, which is a prospective evaluation of consecutive cases, previous studies may not be applicable to most patients currently diagnosed with Zollinger-Ellison syndrome. First, because of the present increased awareness of this syndrome and because of availability of gastrin radioimmunoassay, Zollinger-Ellison syndrome is being diagnosed earlier (21,22), gastrinomas are smaller at the time of diagnosis, and in some series up to 50% of patients have no localizable tumor at the time of diagnosis (2,7,8,19). In contrast, in previous studies (5,23-25) 80%-90% of patients had tumors found at surgery and thus, if selected patients were examined by CT scanning and not all patients underwent surgical evaluation, the results would not be currently generally applicable. Second, in previous studies the results of CT scanning for all islet cell tumors included only small numbers of gastrinomas (3,6,11,14,15). This type of study assumes that the factors determining the ability to localize all islet cell tumors are similar, but this is probably not the case. Gastrinomas are much more likely to be extrapancreatic (5,26) than insulinomas (27). This increases the difficulty of localizing the gastrinoma because a wider area must be considered, confusion with fluid-filled opacified bowel is more likely, and the likelihood of the gastrinoina occurring in an area overlying other areas that enhance with contrast is increased (7). Third, recent studies also suggest that gastrinomas and insulinomas vary markedly in the percentage that enhance with contrast and can be demonstrated by angiography. A relatively high percentage of insulinomas and only a small percentage of gastrinomas are identified by angiography, suggesting differences in vascularity and enhancement with contrast (7,14,28,29). Each of the above points is avoided in the present study because all patients underwent definitive procedures to confirm the results of the CT scanning, only patients with Zollinger-Ellison syndrome were iricluded, and the study was prospective and involved consecutive cases. The present results provide data that allow the potential usefulness of the CT to be assessed in each of the clinical decisions that must be made in currently managing a patient with Zollinger-Ellison syndrome. Initially, the extent of the gastrinoma must be determined to decide whether surgical exploration, chemotherapy, or only medical treatment should be pursued. A CT scan for metastatic gastrinoma to the liver has a negative predictive value of 90%, a specificity of 98%, a positive predictive value of 93%, and a sensitivity of 72%. These results demonstrate that false-positive CT scans of the liver are very uncommori and thus, almost all patients without metastatic gastrinoma to the liver will have a negative CT of the liver. Conversely, almost all patients with a positive hepatic CT scan for gastrinoma will be found to have metastatic tumor at surgery. In contrast to positive CT scans, false-negative results are riot rare and -25% of patients with metastatic disease will not be identi-

7 April 1987 CT SCAN IN ZOLLINGER-ELLISON SYNDROME 911 fied by the CT scan. These results suggest that, in the patient who is considered a surgical candidate, if the CT scan is negative, additional imaging studies may be helpful. One recent study reported that angiography had a greater sensitivity than CT scanning in identifying metastatic gastrinoma to the liver (30). To evaluate this finding and determine the role of angiography after first performing a CT scan, we have analyzed our experience with selective angiography in the accompanying paper. In determining the extent and location of extrahepatic gastrinoma, the CT scan resulted in only one false-positive result and thus, had a high positive predictive value (95%) and a high specificity (95%). In contrast, CT scanning had frequent false negatives with the result that the sensitivity was 59% and the negative predictive value of localizing extrahepatic gastrinoma was 54%. These results indicate that a positive extrahepatic CT scan is a reliable indicator of the location of the gastrinoma. However, because of lower sensitivity and lower negative predictive value, a negative CT scan should generally not be used as a basis to exclude surgical exploration. In ~40% of patients with a negative CT scan, a gastrinoma will be found at surgery or autopsy. These results suggest that if the CT scan does not demonstrate extrahepatic gastrinoma, additional localization studies should be done. Selective angiography is reported to have a sensitivity of 13%-35% (B,30) and ultrasound a sensitivity of 2B%-60% (13,14,30). Selective venous sampling for gastrin has been reported by some to be useful in localizing extrahepatic gastrinoma (B,31) and to have a sensitivity up to 93% (B), whereas our recent studies (32) indicate that this procedure adds little to careful imaging studies and thus is not routinely indicated. Although the results vary from study to study and some investigators report frequent false negatives as well as false positives (6,7,B) in some studies selective angiography detected gastrfnomas not detected by CT scanning (30,33). In another recent study ultrasound was superior to CT scanning in detecting small gastrinomas «7 mm) in the pancreas (13). The results of the latter two studies suggest that careful angiography and ultrasound by an experienced radiologist may identify gastrinomas not identified by CT scanning. Analysis of the CT scan results demonstrates a number of factors that contribute to the lower sensitivity and negative predictive values in identifying extrahepatic versus hepatic gastrinomas, and may be important in the clinical management of an individual patient. As reported previously, the size of the gastrinoma is an important determinant for localizing hepatic and extrahepatic gastrinomas (3,11-13). However, size was more important for localizing extrahepatic than hepatic gastrinomas. For both hepatic and extrahepatic gastrinoma BO%-90% of all tumors >3 cm and 0% of tumors <1 cm were localized. However, 75% of gastrinomas metastatic to the liver and only 30% of primary gastrinomas 1-3 cm in size were localized. The location of the extrahepatic gastrinoma was also a critical factor in localization. Computed axial tomography scanning localized ~BO% of gastrinomas within the pancreas and only 35% of extrapancreatic gastrinoma. This is particularly important clinically because it demonstrates that CT scanning has the least sensitivity in localizing gastrinomas outside the pancreas and yet this is the group that recent studies suggest may have a much higher cure rate (up to 50%) (5,26). Because patients with false-negative CT scans cannot be distinguished clinically, the present results indicate that surgical exploration should still be considered in patients with a negative extrahepatic CT scan. In conclusion, the results of this prospective study suggest that because of its convenience, high specificity, and high positive predictive value, CT scan with bolus infusion of contrast should be the initial procedure for localizing and assessing the extent of gastrinoma in patients with Zollinger-Ellison syndrome. In patients with negative CT localization, additional imaging studies should be considered and the negative CT result alone should not be used to determine the need for surgical exploration. References 1. Ellison EH, Wilson SD. The Zollinger-Ellison syndrome: reappraisal and evaluation of 260 registered cases. Ann Surg 1964;160: Jensen RT, Doppman JL, Gardner JD. Gastrinoma. In: Brooks F, DiMagno E, Gardner JD, Go VLW, Lebenthal E, Scheele G, eds. The exocrine pancreas: biology, pathobiology and diseases. New York: Raven, 1986: Deveney CW, Deveney KE, Stark D, Moss A, Stein S, Way LW. Resection of gastrinomas. Ann Surg 1983;198: Malagelada J-R, Edia AJ, Adson MA, Von Heerden JA, Go VLW. Medical and surgical options in the management of patients with gastrinoma. Gastroenterology 1983;84: Stabile BE, Morrow DJ, Passaro E Jr. The gastrinoma triangle: operative implications. Am J Surg 1984;147: Thompson JC, Lewis BG, Wiener I, Townsend CM Jr. The role of surgery in the Zollinger-Ellison syndrome. Ann Surg 1983;197: Jensen RT, Gardner JD, Raufamn J-p, Pandol SJ, Doppman JL, Collen MJ. Zollinger-Ellison syndrome. NIH combined clinical staff conference Uensen RT, moderator). Ann Intern Med 1983;98: Roche A, Raisonnier A, Gillon-Savouret Me. Pancreatic venous sampling and arteriography in localizing insulinomas and gastrinomas: procedure and results in 55 cases. Radiology 1982;145: Zollinger RM, Ellison EC, Fabri PJ, Johnson J, Sparks J, Carey Le. Primary peptic ulcerations of the jejunum associated with islet cell tumors: twenty-five-year appraisal. Ann Surg 1980;191 :

8 912 WANK ET AL. GASTROENTEROLOGY Vol. 92, No Norton IN, Sugarbaker PN, Doppman SL, et al. Aggressive resection of metastatic disease in selected patients with malignant gastrinoma. Ann Surg 1986;203: Damgaard-Petersen K, Stage JG. CT scanning in patients with Zollinger-Ellison syndrome and carcinoid syndrome. Scan J Gastroenterol 1979;14(Suppl 53): Dunnick NR, Doppman JL, Mills SR, McCarthy DM. Computed tomographic detection of nonbeta pancreatic islet cell tumors. Radiology 1980;135: Gunther RW, Klose KJ, Ruckert K, et al. Islet-cell tumors: detection of small lesions with computed tomography and ultrasound. Radiology 1983;148: Krudy AG, Doppman JL, Jensen RT, et al. Localization of islet cell tumors by dynamic CT. Am J RoentgenoI1984;143: Stark D, Moss A, Goldberg H, et al. CT of pancreatic islet cell tumors. Radiology 1984;150: Thompson NW, Lloyd RV, Nishiyama RH, et al. MEN-I pancreas: a histological and immunohistochemical study. World J Surg 1984;8: Deveney CW, Deveney KS, Jaffe BM, Janes RS, Way LW. Use of calcium and secretin in the diagnosis of gastrinoma (Zollinger-Ellison syndrome). Ann Intern Med 1977;87: Collen MJ, Howard JM, McArthur KE, et al. Comparison of ranitidine and cimetidine in the treatment of gastric hypersecretion. Ann Intern Med 1984;100: Harmon JW, Norton JA, Collen MJ, et al. Removal of gastrinomas for control of Zollinger-Ellison syndrome. Ann Surg 1984;200: Weinstein MC, Fineberg HV. Clinical decision analysis. Philadelphia: WB Saunders, 1980: McCarthy DM. Zollinger-Ellison syndrome. Ann Rev Med 1982;33: Deveney CW, Deveney KS, Way LW. The Zollinger-Ellison syndrome--23 years later. Ann Surg 1978;188: Creutzfeldt W, Arnold R, Creutzfeldt C, Tract NS. Pathomorphological, biochemical, and diagnostic aspects of gastrinomas (Zollinger-Ellison syndrome). Hum Pathol 1975; 6: Fox PS, Hofmann JW, Wilson SD, DeCosse JJ. Surgical management of the Zollinger-Ellison syndrome. Surg Clin North Am 1974;54: Ellison EH, Wilson SD. The Zollinger-Ellison syndrome: re-appraisal and evaluation of 2600 registered cases. Ann Surg 1964;160: Wolfe MM, Alexander RW, McGuigan JE. Extrapancreatic, extraintestinal gastrinoma. N Engl J Med 1982;306: Giercksky KE, Hahne J, McHisen W, et al. Endocrine tumors of the pancreas. Scan J Gastroenterol 1980;15: Mills SR, Doppman JL, Dunnick NR, McCarthy DM. Evaluation of angiography in Zollinger-Ellison syndrome. Radiology 1979;131: Ofstad E, Kluge T, Fretheim B. Non beta cell tumors of the pancreas. Acta Chim Scand 1972;138: Ruszniewski PH, Mignon M, Rene E, Bonfils S. Localization of tumoral process in Zollinger-Ellison syndrome (ZES): a retrospective study in 76 patients (abstr). Gastroenterology 1986;90: Glowniak JV, Shapiro B, Vinnik AI, Glaser B, Thompson NW, Cho KJ. Percutaneous trans hepatic venous sampling of gastrin. N Engl J Med 1982;307: Cherner JA, Doppman JL, Norton JA, et al. Selective venous sampling for gastrinomas: a prospective assessment. Ann Intern Med 1986;105: Mee AS, Imail S, Bornman PC, Marks IN. Changing concepts in the presentation, diagnosis and management of the Zollinger-Ellison syndrome. Q J Med 1983;206: