Patients with gastroenteropancreatic tumors are assessed for location of the primary tumor (to assist in possible tumor resection [7, 11-13]), for metastasis to the liver (for possible resectability [8, 14, 19, 20]), for the need for tumoricidal therapy [21], and for distant metastases (for possible specific tumoricidal therapies, such as local radiation to bone metastases [22]). Different localization methods are better for certain clinical situations [8, 9, 17], and somatostatin receptor scintigraphy needs to be compared with other methods in each of these clinical circumstances.

Sixty percent to 90% of patients with the Zollinger-Ellison syndrome have malignant tumors, and their tumors thus resemble all pancreatic endocrine tumors with the exception of insulinomas [8, 23, 24]. The Zollinger-Ellison syndrome occurs more frequently than other malignant pancreatic endocrine tumors. Therefore, several groups, including ours, have a sufficient number of patients with this syndrome to be able to systematically address questions about localization in different clinical situations [8, 25, 26]. Gastrinomas resemble other, less common pancreatic endocrine tumors in that they are composed of amine precursor uptake and decarboxylation (APUD) cells and have similar growth patterns, locations, imaging properties, metastatic rates, immunohistochemistry, and rates of occurrence of somatostatin receptors [24, 27]. Gastrinomas are therefore an excellent model from which to obtain information that is also pertinent to the less common pancreatic endocrine tumors [25].

We prospectively compared the ability of somatostatin receptor scintigraphy with that of other conventional localization methods—ultrasonography, CT, MRI, bone scanning, and selective angiography—in the localization of primary and metastatic gastrinoma in patients with the Zollinger-Ellison syndrome.

Methods

Top Methods Results Discussion Author & Article Info References

We prospectively studied 80 consecutive patients with the Zollinger-Ellison syndrome who were admitted to the National Institutes of Health (NIH) between June 1994 and May 1995. Our study is part of a prospective study of patients with the Zollinger-Ellison syndrome that has been ongoing at the NIH since 1974, as approved by the clinical research committee of the National Institute of Diabetes and Digestive and Kidney Diseases. Thirty-one of the patients had no previous gastrinoma resection, and 49 had had noncurative resections of gastrinomas 0.25 years to 13 years before the study.

The Zollinger-Ellison syndrome was diagnosed as described elsewhere [28]. Serum gastrin levels were determined by Bioscience Laboratories (New York, New York). The diagnostic criteria for the presence of multiple endocrine neoplasia type I in a patient with the Zollinger-Ellison syndrome have been described elsewhere [29].

Basal acid output and maximal acid output were determined for each patient by using methods described previously [30]. Doses of oral antisecretory drug were determined by establishing the dose required to reduce gastric acid output to less than 10 mEq per hour before the next dose of medication and to less than 5 mEq per hour for patients who had had gastric acid-reducing surgery or who had advanced esophageal disease [31].

Specific Protocol

The localization and the extent of gastrinomas were evaluated in all patients as described elsewhere [17, 32] by using upper gastrointestinal endoscopy, CT, MRI, transabdominal ultrasonography [33], and bone scanning. With MRI, T1-weighted spin-echo sequences and STIR sequences were obtained with a repetition time of 400 to 600 ms and an echo time of 10 ms as described [16]. We did CT as described [16, 17] at 10-mm thickness with an oral contrast agent (diatrizoate sodium, Winthrop-Breon, Rensselaer, New York) with and without the rapid (2 mL/s) intravenous injection of the contrast agent iopamidol 300 (Winthrop-Breon). If surgical exploration was being considered or the extent of disease remained unclear, selective abdominal angiography was done with injection of the splenic, superior mesenteric, gastroduodenal, and hepatic arteries as described elsewhere [17, 34]. One radiologist evaluated the results of all conventional imaging studies.

For somatostatin receptor scintigraphy studies, patients were hydrated before and after the intravenous injection of [¹¹¹In-DTPA-DPhe¹]octreotide and were given a laxative on the night of administration to avoid artifacts from radioactive accumulation in the intestines. Each patient received 6 mCi of [¹¹¹In-DTPA-DPhe¹]octreotide intravenously as recommended by the manufacturer for single-photon emission computed tomographic (SPECT) imaging (Mallinckrodt Diagnostic Imaging Service Radiopharmacy, Beltsville, Maryland). Images were obtained 4 and 24 hours after injection using a TRIONIX (Twinburg, Ohio) or ADAC (Milipitas, California) dual-headed camera with 20% windows at 173 and 247 keV. At 4 hours, a 30-minute whole-body scan was obtained; 10-minute planer spot views of the abdomen and other regions were obtained as needed. At 35 minutes and at 24 hours, SPECT images of the abdomen were obtained. Forty-second, 128 x 128 matrix SPECT images were acquired at 4-degree intervals over 360 degrees. The images were reconstructed with the manufacturer's software using a standard filter back projection algorithm. A Hamming filter was used. Images were displayed as orthogonal (transverse, coronal, sagittal) sections and as reprojected views. Results of somatostatin receptor scintigraphy were obtained while the investigator was blinded to the results of the conventional imaging studies.

Surgical exploration was done in all new patients and all patients who had a positive extrahepatic gastrinoma localized, had no liver metastases, had multiple endocrine neoplasia type I or a recent exploration (< 2 years), and had had exploratory laparotomy for possible gastrinoma resection (n = 15). All patients with possible liver metastases had percutaneous CT- or ultrasonography-guided biopsies (n = 24).

Statistical Analysis

The Fisher exact test was used to compare independent percentages, and the McNemar test was used to compare sensitivities in the same patients. Two-tailed P values and 95% CIs from the StatXact statistical package (Version 3.0, Cytel Software Corp., Cambridge, Massachusetts) are reported. P values less than 0.05 were considered statistically significant.

Results

Top Methods Results Discussion Author & Article Info References

The clinical and laboratory characteristics of the 80 study patients are shown in Table 1. These patients resemble patients in other large series [8, 35] with regard to sex, age, multiple endocrine neoplasia type I, basal acid output, maximal acid output, fasting serum gastrin concentration, and disease duration.

Somatostatin receptor scintigraphy detected tumors either inside or outside of the liver in 56 of the 80 patients (70%); conventional imaging studies were significantly less sensitive (P < 0.001) (Figure 1). Angiography, CT, and MRI each detected tumors in 38% to 45% of patients; ultrasonography detected tumors in 19% of patients (n = 15). The sensitivity of somatostatin receptor scintigraphy alone (70% [95% CI, 59% to 80%]) was equal to that of all the conventional imaging studies combined (59% [CI, 47% to 70%]) (Figure 1). In 13 patients (16%), a tumor was seen only with somatostatin receptor scintigraphy; in 4 patients (5%), a tumor was seen only with conventional imaging studies. When somatostatin receptor scintigraphy was combined with all other conventional imaging methods, tumors were localized in 60 patients (75% [CI, 64% to 84%]) (Figure 1). This result was nearly the same as the result with somatostatin receptor scintigraphy alone (70% [CI, 59% to 81%]).

View larger version (22K): [in this window] [in a new window]   Figure 1. Results of tumor localization studies for the identification of hepatic or extrahepatic gastrinoma in patients with the Zollinger-Ellison syndrome. Results are expressed as the percentage of the 80 patients in whom a tumor was located extrahepatic or metastatic to the liver. Each patient was counted only once. All patients but one (who did not have selective angiography because recent liver transplantation resulted in contraindication) had all of the studies done. Angio equals angiography; CT equals computed tomography; MRI equals magnetic resonance imaging; SRS equals somatostatin receptor scintigraphy; US equals ultrasonography. ***P < 0.001 for the method compared with somatostatin receptor scintigraphy alone.

Extrahepatic tumors were detected in 46 patients (58%) with somatostatin receptor scintigraphy only; somatostatin receptor scintigraphy was significantly more sensitive (P < 0.001) Figure 2 than any of the conventional imaging methods. Angiography, CT, and MRI identified tumors in 28% to 31% of patients, and ultrasonography identified extrahepatic tumors in 7 patients (9%) (Figure 2). The sensitivity of somatostatin receptor scintigraphy alone (58% [CI, 46% to 68%]) was equal to that of all the conventional imaging methods combined (48% [CI, 36% to 59%]). Extrahepatic tumors were identified in 16 patients (20%) who had only somatostatin receptor scintigraphy and in 8 patients (10%) who had only conventional imaging studies. Figure 3 shows a case in which the results of CT, MRI, and the other conventional imaging studies were negative but somatostatin receptor scintigraphy localized a gastrinoma in the pancreatic head-duodenal area that was subsequently found in the duodenum at laparotomy. When somatostatin receptor scintigraphy was combined with the conventional imaging methods, extrahepatic tumors were identified in 68% (CI, 56% to 78%) of patients (Figure 2). This result was similar to the result obtained with somatostatin receptor scintigraphy alone (58% [CI, 46% to 68%]).

View larger version (21K): [in this window] [in a new window]   Figure 2. Results of tumor localization studies for identification of an extrahepatic tumor in patients with the Zollinger-Ellison syndrome. Results are expressed as the percentage of the 80 patients in whom any extrahepatic tumor was localized. Each patient was counted only once. Angio equals angiography; CT equals computed tomography; MRI equals magnetic resonance imaging; SRS equals somatostatin receptor scintigraphy; US equals ultrasonography. ***P < 0.001 for the method compared with somatostatin receptor scintigraphy alone.

View larger version (55K): [in this window] [in a new window]   Figure 3. Localization of a duodenal gastrinoma by somatostatin receptor scintigraphy alone. Top. Negative magnetic resonance imaging STIR (short-time inversion-inversion recovery) result. Middle. Negative computed tomography scan. Bottom. Somatostatin receptor scintigraphy image showing a gastrinoma in the pancreatic head-duodenal area. At surgery, a 0.4-cm duodenal wall gastrinoma was removed.

In the 24 patients with proven metastatic liver disease, somatostatin receptor scintigraphy had the highest sensitivity of any localization method: It detected liver metastases in 22 patients (92%) (Figure 4). Other imaging studies (with the exception of MRI [P = 0.12]) were less sensitive than somatostatin receptor scintigraphy alone. Angiography (P = 0.016) Figure 4 identified metastatic disease in 62% of these patients, and ultrasonography (P = 0.001) or CT (P < 0.001) Figure 4 detected tumors in 42% to 46% of these patients. In three patients (12%), metastatic disease to the liver was detected with somatostatin receptor scintigraphy only; in 1 patient (4%), metastatic disease to the liver was detected with conventional imaging studies only. Figure 5 shows an example of the greater sensitivity of somatostatin receptor scintigraphy compared with conventional imaging studies. The patient in whom the image was taken had left-lobe liver metastases detected with somatostatin receptor scintigraphy only, whereas right-lobe liver metastases were detected with both MRI and somatostatin receptor scintigraphy. When somatostatin receptor scintigraphy and all of the conventional imaging methods were combined, metastatic liver disease was identified in 23 of the 24 of patients with proven liver metastases (96% [CI, 79% to 99.9%]) (Figure 4); this result overlapped considerably with the results of somatostatin receptor scintigraphy alone (92% [CI, 73% to 99%]). In the remaining 4% of patients (1 patient), liver metastases were detected only by exploratory laparotomy. One patient had localization to the left lobe of the thyroid. This patient had Hashimoto thyroiditis, which was confirmed by aspiration and cytologic examination.

View larger version (24K): [in this window] [in a new window]   Figure 4. Results of tumor localization studies for the identification of liver metastases in patients with proven metastatic gastrinomas to the liver. Results are expressed as the percentage of the 24 patients with proven metastatic liver disease in whom metastasis was identified. Each patient was counted only once. Angio equals angiography; CT equals computed tomography; MRI equals magnetic resonance imaging; SRS equals somatostatin receptor scintigraphy; US equals ultrasonography. *P equals 0.016, **P equals 0.001, and ***P < 0.001 for the method compared with somatostatin receptor scintigraphy alone.

View larger version (50K): [in this window] [in a new window]   Figure 5. Localization of liver metastases by somatostatin receptor scintigraphy alone in a patient with the Zollinger-Ellison syndrome. Top. Magnetic resonance imaging (MRI) STIR (short-time inversion-inversion recovery) showing metastasis in the right lobe of the liver only (arrow). Bottom. Somatostatin receptor scintigraphy (SRS) image showing metastases in both the right and left liver lobes (arrows). This patient was treated with interferon.

Fifteen patients Table 2 had exploratory laparotomy for possible removal of the primary gastrinoma after somatostatin receptor scintigraphy and other imaging studies were done. Gastrinomas were found in all 15 patients at surgery; a total of 29 gastrinomas were found. Ten patients each had a duodenal gastrinoma; 1 patient had a pancreatic tail gastrinoma; and 18 lymph nodes positive for gastrinomas were resected in 10 patients (Table 2). Twenty-four of these 29 gastrinomas were identified by somatostatin receptor scintigraphy before surgery, significantly more than were identified by CT, MRI, ultrasonography, and angiography combined (n = 15) (P = 0.004). Somatostatin receptor scintigraphy missed 17% of the tumors that were identified at surgery, including one duodenal tumor and four small tumors that had metastasized to the lymph nodes (Table 2).

Discussion

Top Methods Results Discussion Author & Article Info References

Numerous studies have recently reported that somatostatin receptor scintigraphy can localize carcinoid tumors and various pancreatic endocrine tumors, including gastrinomas [1, 2, 18, 36-45]. However, many of these studies are limited in various ways, making it difficult to completely evaluate the possible role of somatostatin receptor scintigraphy in comparison with existing methods. First, most of the studies did not differentiate between localization of metastatic lesions in the liver and localization of extrahepatic lesions. This distinction is clinically important because the reasons for doing localization studies and thus the usefulness of existing localization methods differ in these two situations [8, 9, 17]. Second, the number of patients with pancreatic endocrine tumors was small in every study except the recent study by de Kerviler and coworkers [18]. Third, somatostatin receptor scintigraphy was systematically compared with the most sensitive conventional imaging methods in only a few patients in these studies. Specifically, few patients had selective abdominal angiography, which is considered to be the most sensitive conventional imaging technique for localization of primary lesions and, in many studies, has been shown to be the most sensitive technique for localization of metastatic disease in the liver [9, 17, 34]. Also, although MRI was assessed in some studies [40, 44, 45], it is unclear whether MRI STIR sequences were routinely done. These sequences have been shown to more clearly image liver metastases than standard T1-weighted imaging in patients with metastatic pancreatic endocrine tumors [16, 17].

Previous studies involving fewer patients have reported that somatostatin receptor scintigraphy detects either a primary tumor or liver metastases in 80% to 100% of patients with various pancreatic endocrine tumor syndromes, in as many as 100% of patients with gastrinomas, and in only 61% of patients with insulinomas [2, 38, 45, 46]. This difference is supported by in vitro somatostatin receptor autoradiography, with which 72% of insulinomas and 100% of gastrinomas, glucagonomas, and nonfunctioning pancreatic endocrine tumors were found to have somatostatin receptors [2, 47]. Despite reports that 100% of gastrinomas have somatostatin receptors [2, 47], we found that somatostatin receptor scintigraphy detected a primary gastrinoma or metastatic disease in 70% of patients. In previous studies [41], it has been suggested that somatostatin receptor scintigraphy failed to image various pancreatic endocrine tumors because SPECT imaging was not used, because low doses of [¹¹¹In-DTPA-DPhe¹]octreotide were used, because a single-headed camera was used, or because the patients studied might not actually have had pancreatic endocrine tumors. Our study has none of these shortcomings. We used SPECT imaging in all patients at 4 and at 24 hours, we used the recommended [¹¹¹In-DTPA-DPhe¹]octreotide dose in all patients [6], and we used a dual-head camera in all patients. Furthermore, all patients met strict criteria for diagnosis of active Zollinger-Ellison syndrome [7, 8, 35, 36]. At present, it remains unclear whether decreased density of somatostatin receptors in the gastrinomas or some other factors account for the finding that at least one third of patients with the Zollinger-Ellison syndrome do not have gastrinomas localized by somatostatin receptor scintigraphy.

In previous studies of patients with gastroenteropancreatic tumors, somatostatin receptor scintigraphy reportedly identified liver metastases in 80% to 100% of patients in whom such disease had been identified by other imaging methods, such as CT or ultrasonography [38, 42, 48]. In most studies, the true sensitivity of somatostatin receptor scintigraphy was not established because more sensitive imaging studies—such as angiography or STIR MRI sequences that, when combined with CT, can identify liver metastases in almost all patients—were not used [16, 17, 34]. In our study, somatostatin receptor scintigraphy identified metastatic gastrinoma to the liver in 92% of patients with proven liver metastases. This shows that somatostatin receptor scintigraphy does not identify liver metastases in all patients. Similarly, in previous series of patients with gastroenteropancreatic tumors (excluding insulinomas), the sensitivity of somatostatin receptor scintigraphy varied from 14% to 90% in the localization of a possible primary tumor [18, 36-3840, 44, 48]. In our study, somatostatin receptor scintigraphy identified a primary gastrinoma in 58% of patients. This result shows that somatostatin receptor scintigraphy does not identify primary tumors in 40% of patients. These results, in showing that somatostatin receptor scintigraphy does not localize a substantial percentage of primary pancreatic endocrine tumors and that it does not identify some patients with metastatic disease to the liver, raise the following question: Should additional imaging studies be done to exclude metastatic liver disease and to localize the primary tumor in patients with pancreatic endocrine tumor syndromes in whom the results of somatostatin receptor scintigraphy are negative, especially before surgery?

Numerous studies have reported that somatostatin receptor scintigraphy [18, 36, 38-4045, 49] is superior to other conventional localization methods, but this conclusion is often not proven because investigators do not routinely use the most sensitive conventional imaging tests, such as angiography and MRI. Furthermore, few studies have assessed the ability of somatostatin receptor scintigraphy combined with other imaging methods to identify more lesions. For the localization of primary pancreatic endocrine tumors, somatostatin receptor scintigraphy has been reported to be equal to CT [42], superior to CT or ultrasonography [39, 40, 44, 45], and equal [44] or superior [40] to MRI. In our study, somatostatin receptor scintigraphy was more sensitive than any other imaging method for the identification of primary gastrinomas, and the sensitivity of somatostatin receptor scintigraphy alone was equal to that of all conventional methods combined. Only one study [18] has examined somatostatin receptor scintigraphy in combination with other methods. In that study, using CT in addition to somatostatin receptor scintigraphy identified lesions in an additional 12% of patients. In our study, combining other imaging methods with somatostatin receptor scintigraphy identified a possible primary tumor in 68% of patients, 10% more than were identified with somatostatin receptor scintigraphy alone. This suggests that additional imaging methods should be used in patients in whom somatostatin receptor scintigraphy fails to localize a possible primary tumor.

For the identification of patients with liver metastases, somatostatin receptor scintigraphy has been reported to be equally sensitive to ultrasonography, CT, or MRI [18, 45] or superior to ultrasonography, CT, MRI, or angiography [18, 36, 40]. Our study shows that in patients with proven liver metastases, the ultrasonography results were positive in 46% of patients, the CT results were positive in 42% of patients, the MRI results were positive in 71% of patients, the angiography results were positive in 62% of patients, and the results of a combination of all conventional studies were positive in 83% of patients. Somatostatin receptor scintigraphy alone was the most sensitive method; it had a sensitivity of 92% and was equal to all of the conventional methods combined. When somatostatin receptor scintigraphy was added to all conventional studies, an additional 4% of patients had liver metastases localized. These results suggest that if somatostatin receptor scintigraphy does not show metastatic liver disease in a patient with a pancreatic endocrine tumor, the results of additional localization studies (even those done using sensitive methods, such as MRI or selective angiography) will almost always be negative for metastatic liver disease. Thus, they are not routinely indicated.

Our study has some potential limitations. First, it was done at a tertiary referral center; thus, our patients may have been at a different stage of disease than patients seen by other physicians in primary care settings. However, our patients are similar to those described in other large studies of patients with the Zollinger-Ellison syndrome, and this possibility is therefore unlikely to be an important limitation. Second, the sensitivity achieved in our study may not be attainable in centers that do not do many of these procedures. To achieve the sensitivity we report, it is particularly important that somatostatin receptor scintigraphy be done only at institutions that have the facilities to do SPECT imaging during somatostatin receptor scintigraphy. If SPECT imaging cannot be done, the sensitivity of somatostatin receptor scintigraphy is probably considerably less than that reported here [50]. If facilities similar to those used for our patients are available in an institution, then our conclusions on the role of somatostatin receptor scintigraphy in the localization of pancreatic endocrine tumors are probably applicable to that institution.

Because we found somatostatin receptor scintigraphy to have the highest sensitivity of any imaging method for localization of the primary tumor and identification of the presence of liver metastases in our study, it should be the initial tumor localization study done in patients with the Zollinger-Ellison syndrome. This conclusion probably also pertains to other pancreatic endocrine tumor syndromes, excluding insulinomas, and to carcinoid tumors. Insulinomas are excluded because only 60% to 70% of them have somatostatin receptors [2, 41]. Carcinoid tumors and other pancreatic endocrine tumors are included because they have the same percentage of somatostatin receptors as gastrinomas and have been reported to be imaged by somatostatin receptor scintigraphy [2, 41, 47].

Whereas all of the conventional imaging studies combined (ultrasonography, CT, MRI, selective angiography, and bone scanning) had a sensitivity equal to that of somatostatin receptor scintigraphy, these combined studies cost at least three times as much as somatostatin receptor scintigraphy and involve considerably more time, radiation exposure, and inconvenience for the patient. However, if the results of somatostatin receptor scintigraphy are positive for metastatic liver diseases and tumoricidal treatment is considered, then an additional imaging technique, preferably MRI, should be considered, because it is easier to assess changes in metastatic tumor size with such a method. If the somatostatin receptor scintigraphy results are negative for either metastatic liver disease, the primary tumor, or both in a patient with a pancreatic endocrine tumor syndrome (excluding insulinoma), our data allow a decision to be made about whether additional tumor localization studies should be done. Additional imaging studies to identify additional patients with metastatic liver disease do not seem justified.

Somatostatin receptor scintigraphy identifies approximately 90% of patients with metastatic liver disease, and the addition of ultrasonography, CT, MRI, and angiography identifies only 4% more patients with metastatic liver disease. In a patient with a positive somatostatin receptor scintigraphy localization of a possible primary tumor, additional imaging studies are not indicated because they infrequently localize additional extrahepatic tumors. However, if somatostatin receptor scintigraphy results are negative for a primary tumor and metastatic liver disease, additional localization studies should only be done if surgery is considered. Somatostatin receptor scintigraphy alone will localize only 60% of the primary tumors; using other imaging studies in conjunction will identify a possible primary tumor in an additional 10% of patients and thus should be done. At present, because of recent advances that have increased its sensitivity and because of its wide availability, MRI [17] should be the next procedure done to identify the primary tumor. Furthermore, because somatostatin receptor scintigraphy misses 20% of the extrahepatic gastrinomas found at surgery and because a recent study has shown that tumor resection substantially decreases the percentage of patients who develop liver metastases [51], all patients with the Zollinger-Ellison syndrome who do not have liver metastases, multiple endocrine neoplasia type I, or a medical condition limiting life expectancy should have exploratory laparotomy even if the results of somatostatin receptor scintigraphy are negative [51-53].

In conclusion, we show that somatostatin receptor scintigraphy is the single most sensitive method for localizing primary gastrinomas and assessing tumor extent in patients with the Zollinger-Ellison syndrome, and we provide data that allow for the development of guidelines for use of somatostatin receptor scintigraphy in the localization of pancreatic endocrine tumors. In the future, it will be important to extend these studies to evaluate the ability of the various imaging methods to alter management: A difference in sensitivity does not necessarily reflect changes in management. With the current emphasis on cost constraints, an assessment of the effect of an imaging study on management is ultimately the most important measure of the usefulness of that study.

Drs. Reynolds and Chen: National Institutes of Health, Nuclear Medicine, Building 10, Room 1C-401, Bethesda, MD 20892-1804.

Dr. Doppman: National Institutes of Health, Diagnostic Radiology, Building 10, Room 660, Bethesda, MD 20892.

Dr. Venzon: National Institutes of Health/National Cancer Institute, Biostatistics and Data Management Section, 8601 Old Georgetown Road, Bethesda, MD 20892.