The computed tomographic (CT) scan (A) and magnetic resonance image (B) show a 5-cm recurrent tumor in liver segment VI 11 years after the initial central hepatectomy.
A, Computed tomographic (CT) scan of the patient in Figure 1 shows a new recurrence in liver segments VI and VII 4 years after the first recurrent tumor excision and 15 years after the initial central hepatectomy. B, Gallium 68–labeled DOTATOC positron emission tomographic (PET) scan shows that the recurrent tumor takes up the somatostatin analogue and is consistent with a gastrinoma.
The liver primary tumors are in segments II, IV, and V. The primary bile duct tumors involve the right, common, and left hepatic bile ducts. The size of the circles indicates the relative size of the tumor. IVC indicates inferior vena cava.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Norton JA, Foster DS, Blumgart LH, et al. Incidence and Prognosis of Primary Gastrinomas in the Hepatobiliary Tract. JAMA Surg. 2018;153(3):e175083. doi:10.1001/jamasurg.2017.5083
Do primary gastrinomas of the liver exist?
In this series of 223 patients with Zollinger-Ellison syndrome, primary gastrinomas of the liver were the second most common extrapancreatic, extraintestinal site for a primary gastrinoma and may metastasize to regional lymph nodes. Aggressive resection is indicated because cure rates and survival are similar to those for primary gastrinomas in other locations.
Liver primary gastrinomas occur and warrant aggressive resection including regional lymph nodes.
Zollinger-Ellison syndrome (ZES) is a life-threatening disease caused by a malignant tumor that secretes gastrin (gastrinoma). Gastrinomas typically occur in the pancreas or the duodenum.
To describe the incidence and prognosis of very unusual gastrinomas originating in the hepatobiliary tract.
Design, Setting, and Participants
This study included 223 consecutive patients at the National Institutes of Health and Stanford University Hospital who were enrolled in a prospective protocol to treat ZES using proton pump inhibitors to control acid hypersecretion and surgical resection to ameliorate the tumoral process. Data were collected from June 1982 to August 2017.
Main Outcomes and Measures
Incidence, location, surgical results, and cure rate and overall survival among patients with gastrinomas that originate in the liver or bile ducts. Cure was defined as serum gastrin levels within the reference range, negative results of a secretin test, and no tumor found on imaging.
Of the 223 patients who underwent surgery to remove gastrinomas, 7 (3.1%) had liver or biliary tract primary tumors, including 5 men and 2 women (mean age at diagnosis, 43 years; range 27-54 years). The mean serum gastrin level was 817 pg/mL (range, 289-2700 pg/mL). Each patient had positive results of a secretin test. None had evidence of multiple endocrine neoplasia 1. Four patients had primary tumors in the liver (1 in segment II, 2 in segment IV, and 1 in segment V); 3, in the bile duct (1 in the right hepatic duct, 1 in the left hepatic duct, and 1 in the common hepatic duct). Surgical resection required 1 right hepatic lobectomy, 1 left lateral segmentectomy, 2 left hepatic lobectomies, 1 central hepatectomy, and 2 bile duct resections. Four patients had nodal metastases, and no patient had distant metastases. No operative deaths occurred, but 3 patients had complications, including bile duct stricture, portal vein stricture, and biliary fistula. Each patient was disease free in the immediate postoperative period, and 3 had recurrences in the liver and portal lymph nodes (at 3, 11, and 15 years). Three patients (43%) remained free of disease at follow-up ranging from 24 months to 26 years.
Conclusions and Relevance
Primary gastrinomas of the hepatobiliary tract are uncommon (3%), but the hepatobiliary system is the second most frequent extraduodenopancreatic primary location (after the lymph nodes). These tumors can occur outside the gastrinoma triangle and must be specifically considered. Furthermore, their discovery changes the operative approach because aggressive liver or bile duct resection is indicated, with high rates of long-term cure and survival and acceptable rates of complications. In addition, their discovery dictates that lymph nodes in the porta hepatis should be routinely excised because nearly 50% of patients will have lymph node metastases.
Zollinger-Ellison syndrome (ZES) is the second most common functional neuroendocrine tumor (NET).1 Zollinger-Ellison syndrome is one of the most difficult NETs to manage because the symptoms of peptic ulcer disease and diarrhea may be life threatening, and the tumor may be malignant and also life threatening.2-4 Because ZES is rare, physicians usually require years to make the correct diagnosis.5 Once the symptoms of ZES are controlled medically with proton pump inhibitors (PPIs),6,7 the workup continues with imaging studies to identify and localize the extent of tumor.8Quiz Ref ID Most gastrinomas in patients with ZES are localized to the duodenum (70%-90%) or the pancreas (2%-30%), but ectopic sites, including the ovary,9,10 stomach, lymph nodes,11 heart,12,13 liver or bile ducts,14-38 and rarely other sites,39 have been reported.
Primary gastrinomas of the hepatobiliary tract are controversial22,40 because duodenopancreatic gastrinomas characteristically metastasize to the liver1,37; therefore, distinguishing whether an NET containing gastrin in the liver is a primary or secondary tumor can be difficult. Furthermore, the biliary tract is not always well explored because lymph node metastases within the pancreatic head area are frequently found in patients with gastrinoma, often without a primary localized tumor, presumably because a small duodenal primary tumor is missed.37,40-44 Furthermore, because fasting serum gastrin levels can normalize transiently after a noncurative resection,45 careful follow-up with functional (fasting serum gastrin levels and secretin testing) and detailed imaging studies are required to establish that the resected gastrinoma was actually a primary tumor.45-48 Thus, limited information, primarily from case reports on hepatobiliary gastrinomas, is available. Liver metastases have been associated with a greater chance of subsequent death from gastrinoma,40,49 whereas a primary tumor of the liver may be curable. Therefore, a better understanding of possible primary hepatobiliary gastrinomas could markedly affect clinical management in some patients with ZES.
Since 1982 at the National Institutes of Health and 2003 at Stanford University Hospital, 223 patients with ZES were included in a prospective strategy to perform surgical exploration for cure, as described previously.2,11 The main outcomes measured were overall survival, disease-related survival, disease-free status, the need for reoperation, and time to recurrence. After controlling the gastric acid hypersecretion with histamine H2 receptor antagonists or PPIs,7 confirming the diagnosis, and establishing the possible presence of multiple endocrine neoplasia 1 (MEN1),8 all patients underwent detailed imaging studies, including a computed tomographic (CT) scan with intravenous contrast, magnetic resonance imaging with gadolinium, ultrasonography, and somatostatin receptor scintigraphy (since 1994 using indium In 111–labeled DTPA-D-Phe1–octreotide, 6 mCi, with whole-body, planar, and single-photon emission CT views or more recently using gallium 68–labeled DOTATOC positron emission tomography and CT).1,37,50-56 Patients were entered into a prospective study approved by the institutional review board of the National Institutes of Health and invited to undergo surgery to remove the tumor if they had no comorbid medical conditions that markedly limited life expectancy, had an apparently operable tumor, and if MEN1 was present, had a tumor of at least 2.0 to 2.5 cm in diameter.8,57 All patients provided written informed consent.
Data were collected from June 1982 to August 2017. The diagnosis of ZES was based on results of acid secretory studies, measurement of fasting serum levels of gastrin, and the results of secretin and/or calcium provocative tests.8,40 Basal and maximal acid output were determined for each patient using methods described previously.6 Doses of oral gastric antisecretory drug were determined as described previously.6,46 All patients referred with a diagnosis of possible ZES underwent an evaluation to establish the diagnosis of ZES and to determine whether MEN1 was present6,8,58 and studies to determine the suitability of surgical exploration for cure.57,58 These latter studies included tumor localization studies, studies to determine the presence or absence of MEN1,8 and studies to determine the presence of another disease that might contraindicate surgery. The presence of MEN1 was established by assessing levels of plasma hormones (intact and midmolecule parathyroid hormone, prolactin, insulin, proinsulin, and glucagon), serum calcium (ionized and total), and glucose as well as personal and family history.8 The operative techniques have been described previously.43,47,57,59 Patients underwent evaluation for disease-free status immediately after surgery (ie, 2 weeks after resection), within 3 to 6 months after resection, and then yearly.43,47 Disease-free status was defined as no evidence of tumor on conventional imaging studies, fasting gastrin levels within reference range, and negative results of secretin provocative testing as described previously.47 Persistent postoperative disease was defined as any persistent elevation of fasting serum gastrin level and/or a positive secretin test result. A postresection recurrence was defined as occurring in a patient who was initially disease free after resection but then lost disease-free status at follow-up evaluation by developing positive findings on imaging studies that were compatible with a recurrent pancreatic NET and/or recurrent elevated fasting serum gastrin levels or positive secretin provocative test results.43,47 The Fisher exact test was used for 2-group comparisons. All continuous variables were reported as means (SEM). The probabilities of survival were calculated and plotted according to the Kaplan-Meier method.60
The following criteria were used to identify patients with likely hepatobiliary primary gastrinomas: the patient had to undergo a detailed exploratory laparotomy as described above; a possible primary gastrinoma was found only in the hepatobiliary area on surgical exploration or imaging studies; during detailed follow-up with imaging studies as described above, no other primary gastrinoma was found; and after resection of the hepatobiliary primary gastrinoma, the patient was disease free for at least 1 year. Seven patients met these criteria. The short-term follow-up of 3 of these patients was included in a previous report of liver surgery in patients with gastrinomas with potentially resectable disease in the liver.36,37
Quiz Ref IDOf the 223 patients in our series with ZES, 7 (3%) had primary gastrinomas of the hepatobiliary tract (5 men and 2 women; mean age, 43 years; range, 27-54 years). None of these patients had MEN1. Each patient had typical signs and symptoms of ZES with peptic ulcer disease and diarrhea due to acid hypersecretion. Each had an elevated fasting serum gastrin level. Mean preoperative serum gastrin level was 817 pg/mL (reference level, <100 pg/mL), with a range of 289 to 2700 pg/mL (to convert to picomoles per liter, multiply by 0.481). Each patient achieved control of all symptoms of peptic ulcer disease and diarrhea with a PPI. The usual PPI treatment consisted of oral omeprazole, 40 mg/d. Each patient gained weight during PPI treatment. Each was found to have a primary gastrinoma of the liver (n = 4) or the bile duct (n = 3).
Each patient underwent advanced imaging studies that did not show any tumors outside the hepatobiliary area. Six patients underwent a single hepatobiliary gastrinoma resection. One patient underwent 3 resections. Initially, the patient underwent a central hepatectomy for a 5.5-cm tumor in liver segments V and VIII and was rendered disease free. The patient developed a recurrence of ZES at 11 years postoperatively and was found to have a large 5-cm recurrence in the right posterior sector (Figure 1) that was subsequently excised surgically. In addition, the patient had 5 positive lymph node metastases in the porta hepatis that were surgically removed. Postoperatively, the patient was found to be free of disease, but 4 years later, hypergastrinemia and diarrhea developed, and the patient had a liver gastrinoma in segments VI and VII (Figure 2) and 1 positive celiac lymph node that was removed at a third operation. The postoperative fasting serum gastrin level was within reference limits, but the secretin test result was positive. The patient is included as alive with disease.
Quiz Ref IDOf the 7 hepatic or bile duct primary gastrinomas, the final pathologic evaluation demonstrated that 4 were in the liver and 3 were within a major bile duct. The locations of the 4 primary tumors in the liver included segment II (n = 1), segment IV (n = 2), and segment V (n = 1) (Figure 3). The location of the 3 primary bile duct tumors were at the bifurcation of the common hepatic duct (n = 1), in the left hepatic duct (n = 1), and in the right hepatic duct (n = 1) (Figure 3). The mean primary liver and bile duct gastrinoma tumor size was 2.6 cm, with a range of 1.4 to 5.5 cm. The surgical procedures included 1 right hepatic lobectomy, 1 left lateral segmentectomy, 2 left hepatic lobectomies, 1 central hepatectomy followed by a tumor excision and a subsequent right posterior sectorectomy, 4 portal lymph node dissections (3 at the initial operation and 1 at a second operation), 1 celiac lymph node dissection at a third operation, 2 bile duct resections, and 3 Roux-en-Y hepaticojejunostomies.
Lymph node metastases were found in each of the 4 patients who underwent a portal and/or celiac lymph node dissection. Three dissections were performed at the initial operation; 1, at a subsequent operation; and 1, at a third operation. None of the patients had distant metastases. Operative complications included 2 bile duct leaks that healed without further intervention. One patient developed narrowing of the portal vein with mild portal hypertension. Quiz Ref IDAll 7 patients were disease free for at least 1 year postoperatively, with 6 of 7 patients (86%) having no evidence of ZES at the second yearly postoperative evaluation. Three patients had recurrences at 3, 11, and 15 years after resection. One of these patients with recurrence subsequently became disease free after operations at 11 and 15 years postoperatively. Quiz Ref IDHowever, at long-term follow-up (median, 13 years; range, 24 months to 26 years), 3 (43%) remained disease free, 1 had died of progressive metastatic tumor at 7 years, and 3 were alive with disease. During follow-up in all patients, imaging studies showed no evidence of a gastrinoma outside the liver and adjacent portal or celiac lymph nodes.
Gastrinomas were the second functional pancreatic NET syndrome described (in 1955)3 after insulinomas, and like insulinomas, they were originally thought to occur in the pancreas. However, it has become apparent that gastrinomas differ markedly from insulinomas in their primary tumor location. Whereas extrapancreatic insulinomas are extremely rare (<0.1%), currently only 10% to 30% of gastrinomas are pancreatic in location, with 70% to 90% occurring in the duodenum.38,61 At present, even with extensive imaging and detailed surgical exploration, no primary tumor has been found in the duodenum or the pancreas in a proportion of patients (10%-30%).11,43,47,62 Primary gastrinomas have been reported in numerous other locations,62 including the lymph nodes,11 stomach, mesentery, ovaries,9,63 heart,12,13 lung, hepatobiliary tract, jejunum, and, rarely, in other locations.38,61 Except for ovarian gastrinomas, which are well established but rare, and lymph node primary tumors, for which existence is controversial but which made up 10% of primary gastrinomas in a previous series,11 descriptions of the other locations are uncommon (<1%), usually in case reports, and not clearly established. In these uncommon locations, establishing that the resected NET is the primary gastrinoma can be difficult.22,39 This difficulty occurs because numerous gastrointestinal tract NETs can have gastrin found immunocytochemically, but the gastrin is not secreted and the NETs are not causing ZES.14,38,61 Furthermore, gastrinomas can metastasize widely or invade other tissues, and when the metastases are resected, the fasting serum gastrin level can decrease transiently often to reference range, although the true primary gastrinoma is still present. The patient is assumed to be cured unless prolonged follow-up is performed with secretin testing and imaging.37,45,64 Furthermore, duodenal gastrinomas can be small (1-5 mm) and easily missed42,43,59; thus, another site could be thought to be a primary site.
Primary hepatobiliary gastrinomas have been reported in a small number of case reports (<3 cases/report) in patients with sporadic ZES14,15,17-21,23-39 and with MEN1 with ZES.16,22 However, for the aforementioned reasons, their existence is questioned and their frequency and natural history, including presentation, resectability, possible cure, and pattern of recurrence, are unknown.22,39 We should address these factors because important changes might result in the surgical approach, prognosis, follow-up, and subsequent treatment of the patient. In one case, a patient with a surgically resected hepatobiliary gastrinoma could be treated as having a primary resection and, in another case, as having metastatic disease to the liver, which carries a poor prognosis and more aggressive treatment.36,40
Our results support the conclusion that primary gastrinomas of the hepatobiliary tract in patients with sporadic ZES (ie, without MEN1) do occur. One could dispute this conclusion and suggest that a small duodenal primary tumor or a primary tumor in another location was missed and that the gastrinomas are not hepatobiliary primary tumors but hepatic metastases. Our study supports the existence of hepatobiliary primary tumors. First, each patient had only a hepatobiliary NET identified on the best possible imaging study that was available at the time. All patients had magnetic resonance imaging and CT examinations with contrast infusions, and most had somatostatin receptor scintigraphy,1 with one having a subsequent DOTATOC positron emission tomographic somatostatin receptor scintigraphy scan, which is the most sensitive imaging modality available at present.1 Each of these examinations identified a hepatobiliary location as the only possible primary site of disease. Second, after surgical excision of the hepatobiliary gastrinoma, all patients were biochemically cured of ZES for a minimum of 1 year.11,45,47 A previous study45 has demonstrated that if a recurrence occurs after gastrinoma resection, in more than 95% of such patients, the recurrence occurs within 6 to 12 months; if the patient is disease free at 1 year, the positive predictive value for long-term cure is 95%. Third, the long-term cure rate with a median follow-up of 13 years was 43%, which is similar to the reported long-term cure rate for gastrinomas excised from the duodenum or pancreas and better than results reported for patients after liver resection for metastatic gastrinoma.37,43,47,57
Our results demonstrate that primary hepatobiliary gastrinomas occur in 3% of patients with ZES, which is more than 10 times the frequency of primary gastrinomas in the ovaries (ie, <0.3% in our series) and of gastrinomas in other rare ectopic sites, such as the stomach, mesentery, heart, or jejunum or gastrinomas due to lung cancer (<0.2%). Primary lymph node gastrinomas are reported in 0 to 10% of patients with ZES undergoing surgical exploration11,38,61 and occurred in a previous series11 in 10% of patients who had surgical exploration. Therefore, the hepatobiliary system is the second most frequent extraduodenopancreatic site for a primary gastrinoma to occur and thus needs to be considered when a duodenal or pancreatic gastrinoma is not found at exploration. This latter point is important and can affect the surgical approach.
Our results also provide support for the proposal that primary gastrinomas can occur in the liver and the biliary tract, as has been reported in previous case reports.15-39 In our series, 4 of 7 primary hepatobiliary gastrinomas (57%) were found in the liver and the remaining 3 (43%) in the biliary tract, suggesting that approximately an equal proportion occur in each location. This distribution is at odds with the frequency of possible hepatic or biliary tract gastrinomas causing ZES reported in the literature. Previous reviews of literature on hepatobiliary NETs report that 20 to 25 patients with sporadic ZES and proposed hepatic gastrinomas have been described15-39 compared with only 5 patients with sporadic ZES and possible biliary tract gastrinomas,16-18,23,36 which suggests that hepatic gastrinomas are more frequent than biliary tract gastrinomas. Coupled with the fact that biliary tract gastrinomas can be easily missed because they can occur within the bile duct or can be attached to the duct and mistaken for possible lymph node metastases, our data raise the possibility that gastrinomas at this site occur more frequently than previously reported.
One feature of the present study that was previously not generally recognized is that hepatobiliary primary tumors often metastasize to regional lymph nodes and lymph nodes within the porta hepatis and celiac axis, and they can recur locally after resection. In various reports involving 27 cases of possible hepatobiliary gastrinomas,15-20,23-36 only 3 patients (11%) were reported to have lymph node metastases21,31,34 and only 3 (11%) had recurrent disease after resection,22,29,32 suggesting that hepatobiliary gastrinomas might be less invasive than duodenopancreatic primary gastrinomas. In our study, 4 of the 7 patients with long-term follow-up (57%) had recurrent disease. Furthermore, 3 patients (43%) had lymph node metastases during the original surgery that were identified and found by portal lymph node dissection. Nonimaged portal and celiac lymph node metastases were also identified in 1 patient with liver recurrence at 11 years after resection (Figures 1 and 2). Therefore, 4 patients (57%) had regional lymph node metastases, similar to rates seen for duodenal or pancreatic primary gastrinomas.43,65 This finding suggests that regardless of the results of the imaging studies, if a hepatobiliary primary gastrinoma is apparent and considered as a working diagnosis, a portal lymph node dissection should be performed at the time of surgery. This procedure has a high probability of demonstrating and removing lymph node metastases.
The exact origin of hepatobiliary tract gastrinomas is unclear. Three theories have been proposed, including differentiation of ectopic pancreatic or adrenal tissue in the liver, transformation of malignant stem cells in the liver, and transformation of neuroendocrine cells in the intrahepatic or biliary hepatobiliary ducts.16-18,21,66 The bile duct cells contain NET argentaffin cells, some of which show gastrin immunoreactivity.16,17,21 In our series, hepatobiliary tract gastrinomas were identified in 3 of the 7 tumors as originating within the bile duct. Although bile ducts are scattered throughout the liver and neuroendocrine cells can be seen within them, we have no direct proof that the other 4 hepatic gastrinomas originated from them, as proposed in other studies.16
The limitations of this study include the impossibility of being certain that a primary gastronome outside the liver and bile ducts was not missed. DOTATOC scans, which are the most sensitive test for gastrinoma detection, were not available for most of the study period. When the earlier patients underwent evaluation, an extrahepatic gastrinoma may have been missed, although we do not believe this result to be the case.
Primary hepatobiliary tract gastrinomas were found in 3% of patients with sporadic ZES, and the hepatobiliary system is the second most frequent extraduodenopancreatic primary location of gastrinomas (after the lymph nodes). Furthermore, the discovery of these gastrinomas changes the operative approach because aggressive resection is indicated; the long-term cure rate and survival rates are high and comparable with those observed among patients with duodenal or pancreatic primary gastrinomas. In addition, our data suggest that discovery of liver or biliary gastrinomas dictates removal of regional lymph nodes (portal and celiac) because nearly half of these patients will have lymph node metastases.
Corresponding Author: Jeffrey A. Norton, MD, Department of Surgery, Stanford University, 300 Pasteur Dr, Room H3591, Stanford, CA 94305 (email@example.com).
Accepted for Publication: October 4, 2017.
Published Online: January 24, 2018. doi:10.1001/jamasurg.2017.5083
Author Contributions: Dr Norton had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Norton, Blumgart, Poultsides, Alexander, Jensen.
Acquisition, analysis, or interpretation of data: Norton, Foster, Blumgart, Poultsides, Visser, Fraker, Jensen.
Drafting of the manuscript: Norton, Foster, Blumgart, Jensen.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Norton, Jensen.
Obtained funding: Jensen.
Administrative, technical, or material support: Foster, Blumgart, Poultsides, Visser, Alexander, Jensen.
Study supervision: Norton, Jensen.
Conflict of Interest Disclosures: None reported.