2293 Liver Embolizations of Patients with Malignant Neuroendocrine Gastrointestinal Tumors B. K. Eriksson, M.D., Ph.D.1 E. G. Larsson, R.N.1 B. M. Skogseid, M.D., Ph.D.1 A. M. Löfberg, M.D.2 L. E. Lörelius, M.D., Ph.D.2* K. E. Öberg, M.D., Ph.D.2 1 Department of Internal Medicine, Uppsala Hospital, Uppsala, Sweden. 2 Department of Diagnostic Radiology, University Hospital, Uppsala, Sweden. * Deceased. BACKGROUND. Patients with neuroendocrine gastrointestinal tumors usually present with inoperable metastatic disease and severe hormonal symptoms. Specific chemotherapy, interferon-␣ (IFN), and somatostatin analogs are established therapies for these patients, but all of them eventually fail. Hepatic arterial embolization can provide reduction of both hormonal symptoms and tumor burden in these patients. METHODS. Between 1981 and 1995, a total of 55 liver embolizations with gel foam powder were performed on 41 patients with histopathologically verified neuroendocrine tumors; 29 had carcinoid tumors and 12 had endocrine pancreatic tumors (EPTs). All patients had received medical treatment, including chemotherapy (n ⫽ 18), IFN (n ⫽ 31), and octreotide (n ⫽ 19), and were experiencing treatment failure when liver embolization was performed at a median of 37 months after diagnosis of liver metastases. Medical treatment was continued after embolization. RESULTS. An overall objective response was noted in 15 of 29 patients with carcinoid tumors (52%). The median duration of effect was 12 months in patients with midgut carcinoid tumors. An overall objective response was observed in 6 of 12 patients with EPTs (50%), with a median duration of effect of 10 months. Adverse events were observed, and, in agreement with earlier reports, the rate of serious complications was 10%. Survival analyses showed a median survival of 80 months and a 5-year survival rate of 60% from the performance of embolization on patients with midgut carcinoid tumors, whereas for patients with EPTs the median survival from embolization was only 20 months. CONCLUSIONS. Liver embolizations performed relatively late in the clinical course in our series appeared to be as effective as “early” embolizations in other series of patients with carcinoid tumors. The results for those with EPTs were poorer, and earlier embolizations may result in better outcomes for these patients. Considering the morbidity associated with the procedure, it is imperative to select patients according to extent of liver involvement, severity of carcinoid heart disease, and somatostatin receptor status. Cancer 1998;83:2293–301. © 1998 American Cancer Society. KEYWORDS: carcinoid tumor, endocrine pancreatic tumor, gel foam powder, liver embolization, p-chromogranin, U-5-hydroxyindoleacetic acid. M Supported by the Swedish Cancer Foundation. Address for reprints: Barbro Eriksson, M.D., Ph.D., Department of Internal Medicine, University Hospital, S-751 85 Uppsala, Sweden. Received December 24, 1997; revision received March 30, 1998; accepted April 23, 1998. © 1998 American Cancer Society ost patients with neuroendocrine gastrointestinal tumors, i.e., carcinoid tumors and endocrine pancreatic tumors (EPTs), present with disseminated disease and involvement of the liver at the time of diagnosis.1,2 Debilitating symptoms associated with hormonal excess are usually present. Treatment of localized hepatic metastases is surgical, if feasible. In the more frequent cases of multifocal lesions in the liver, a multidisciplinary approach, including systemic chemotherapy3,4 and biotherapy, i.e., interferon-␣ (IFN)5,6 and somatostatin analogs,7,8 should be used in the management of these patients. 2294 CANCER December 1, 1998 / Volume 83 / Number 11 Although such medical therapies can control both hormonal symptoms and tumor growth for extended time periods, they are rarely curative, and eventually most tumors become resistant to them. Hepatic metastases from neuroendocrine gastrointestinal tumors are usually hypervascular, and this provides a good rationale for interrupting the arterial blood supply as a therapeutic measure.9,10 Palliative treatment with arterial embolization of hepatic tumors was initially described for hepatocarcinomas.11 Since then, several reports have demonstrated that hepatic arterial embolization can provide both reduction of hormonal symptoms and reduction of tumor burden in patients with neuroendocrine gastrointestinal tumors.12–14 However, the exact role and timing of the procedure (i.e., on which patients it should be performed and when in the clinical course of the disease) is unclear. In this study, we report the results of embolizations with gel foam powder in a large cohort of patients with neuroendocrine gastrointestinal tumors. The embolizations were performed relatively late in the clinical course, i.e., in most cases when the patients were experiencing failure with medical treatment. MATERIALS AND METHODS Patients Over a 14-year period (1981–1995), a total of 55 liver embolizations with gel foam powder were performed on 41 patients (23 males, 18 females). All patients had histopathologically verified diagnoses of neuroendocrine tumors of the gastrointestinal tract. Twenty-nine patients had liver metastases from carcinoid tumors (21 midgut, 4 foregut, and 4 hindgut), and they were diagnosed at a median age of 53 years (range, 36 –73 years) and embolized at a median age of 56 years (range, 40 –73 years)( median age for patients with midgut carcinoid tumors, 59 years). The median duration of disease from the time of diagnosis was 37 months for patients with midgut metastases and 28 months for those with foregut or hindgut metastases. Twelve patients had liver metastases from endocrine pancreatic tumors (1 insulinoma, 2 gastrinomas, 5 nonfunctioning tumors, 3 vipomas, and 1 somatostatinoma); they were diagnosed at a median age of 46 years (range, 34 – 69 years) and embolized at a median age of 52.5 years (range, 35–71 years), and the median duration of their disease from the time of diagnosis was 37 months. Two patients with EPT (one with vipoma and one with a nonfunctioning tumor) had multiple endocrine neoplasia (MEN) type 1. Previous Treatment All patients had received medical treatment before embolization. Among the carcinoid patients, 24 had received IFN, 15 octreotide, and 6 chemotherapy; and among EPT patients, 12 had received chemotherapy, 7 IFN, and 4 octreotide. Responses were of varying duration. All patients were showing signs of disease progression when the decision to embolize the liver was made. Informed consent to perform the procedure was obtained from all patients. The medical treatment was continued after embolization. Eleven patients underwent more than one procedure, and the frequency at which the embolizations were performed was primarily dictated by the patients’ symptoms and/or biochemical or tumor status, i.e., when signs of progression occurred. Two patients had another embolization of the same lobe within 2– 4 months (without signs of progression) to enhance the effect. Embolization Procedure Arteriography of the superior mesenteric artery and celiac axis was done routinely before the embolization to assess arterial anatomy and define the patency of the portal vein. Depending on which hepatic lobe contained more tumor, or where progression of metastases had been demonstrated, very selective catheterization of either the left or the right hepatic artery was performed; because of the risk for severe postembolic complications, embolization of the proper hepatic artery was avoided whenever possible (it was performed, however, on 10 patients in whom selective catheterization could not be achieved). Gelfoam powder (Spongostan, A/S Ferrosan, Denmark) with a particle size of 50 –150 m mixed with iodinated contrast was used as embolic material. The amount of gel foam powder was individually determined for each patient and administered until arterial blood flow ceased completely as assessed by angiogram. The mixture was injected slowly to avoid backflow and spillover to the gastroduodenal, splenic, dorsal pancreatic, or gastric arteries. After embolization, the patients were carefully monitored (up to 1990, in the intensive care unit for the first 24 hours) with regard to vital signs, including pulse, blood pressure, body temperature, weight, hepatic functions (transaminases, alkaline phosphatase, and lactate dehydrogenase), renal functions (creatinine), and routine biochemistry (hemoglobin, leukocyte count, and electrolytes). With a few exceptions early in the study, all patients were carefully hydrated (a diuresis of ⬎3 liters per day was required) and, if necessary, were also given diuretics for the first 4 –5 days after embolization to ensure protection of the Liver Embolizations for Neuroendocrine Gastrointestinal Tumors/Eriksson et al. kidneys. From 1985, all patients with hormonal symptoms were given a continuous intravenous infusion of octreotide 50 –100 g/hour, started immediately after the embolization procedure was finished and then continued for 24 – 48 hours, followed by subcutaneous injections 2–3 times daily to avoid hormonal crisis. Analgesic treatment (pethidine hydrochloride 50 –100 mg intravenously) was given during the embolization procedure and then when required during the days that followed. No prophylactic antibiotics were given. Evaluation of Biochemical and Radiologic Responses The therapeutic effect was evaluated by both measurements of biochemical tumor markers (including urinary 5-hydroxyindoleacetic acid [U-5-HIAA] in carcinoid tumor patients and serum gastrin, insulin, pancreatic polypeptide, plasma glucagon, and vasoactive intestinal polypeptide in patients with EPTs) and radiology (computed tomography [CT] or ultrasonography) every 3 months. Since 1989, plasma chromogranin has routinely been used as a tumor marker in all patients.15 In the CT studies, the size of the metastases before and after embolization was determined by multiplying the longest diameter by the greatest perpendicular diameter of the hepatic tumor mass. Response Criteria An objective biochemical response was defined as a ⬎50% reduction of hormone levels. An objective radiologic response was defined as a ⬎50% reduction in tumor size at CT. An overall objective response required an objective biochemical and/or radiologic response. Stable disease (SD) was defined as a ⬍50% reduction of hormone levels and tumor size and progressive disease (PD) as a ⬎25% increase in tumor markers and/or tumor size. Statistics Survival analyses, with death from any cause as the outcome (related or unrelated to neuroendocrine tumor disease), were performed by the Kaplan–Meier method. An unpaired Student’s t test was used to test for differences between diagnostic groups, and analysis of variance was used to test for differences between response groups. RESULTS The therapeutic effect of all the 55 embolizations are given in Tables 1 and 2. Carcinoids and EPTs were evaluated separately. To avoid excessive detail, results per patient are given in the following text, whereas results per embolization are given in the tables. 2295 Carcinoid Tumors Response Results for patients with carcinoid tumors are presented in Table 1. Eleven of 21 patients with midgut carcinoid tumors (52%) showed an overall objective response (OR), with a median duration of effect of 12 months, whereas 2 of 4 patients with foregut carcinoid tumors and 2 of 4 with hindgut carcinoid tumors responded objectively, with a median duration of effect of 11 months. In 8 of 21 midgut carcinoid patients (38%), a biochemical response (⬎50%) was noted, and a radiologic response (⬎50%) was also demonstrated in 8 of 21 patients (38%). One midgut carcinoid patient who achieved both a biochemical and a radiologic response showed progression of metastases in the untreated lobe (not indicated in Table 1). Two of four patients with foregut and one of four with hindgut carcinoid tumors responded biochemically, whereas a radiologic response was noted in one foregut and two hindgut carcinoid patients. An overall stabilization (SD) was achieved in 6 of 21 midgut carcinoid patients (38%) with a median duration of 7 months, whereas 1 foregut carcinoid and 2 hindgut carcinoid patients showed SD with a median duration of 8 months. All together, eight patients were reembolized. Five patients with midgut carcinoid tumors who had achieved OR (n ⫽ 3) or SD (n ⫽ 2) at first embolization, 2 foregut carcinoid patients who had both achieved OR, and 1 hindgut carcinoid patient who had achieved SD were reembolized. In 2 of these patients, reembolizations were performed within 2– 4 months, without signs of progression, to enhance the effect. In the other 6, reembolizations were performed at signs of progression, with an OR achieved in 2 patients (1 midgut and 1 hindgut) and SD (n ⫽ 3) and PD (n ⫽ 1) achieved in the remaining 4 patients. Progressive disease was seen in 4 of 21 patients with midgut carcinoid tumors (19%). When pretreatment levels of U-5-HIAA and chromogranin in patients with midgut carcinoid tumors were compared, there was no significant difference between responding (OR) and nonresponding (SD, PD) patients. Median duration of disease from diagnosis in the two groups did not differ either. For midgut carcinoid patients, survival analyses showed a median survival of 120 months and an 83% rate of survival 5 years after diagnosis. In the same patient category, the median survival and rate of survival 5 years after the first embolization were 80 months and 60%, respectively. For the smaller group of foregut-hindgut carcinoid patients, the median survival from diagnosis and median survival from first 2296 CANCER December 1, 1998 / Volume 83 / Number 11 TABLE 1 Effects of Hepatic Arterial Embolization on Patients with Carcinoid Tumors Biochemical response Tumor type Midgut Midgut 1 2 3 Midgut 1 2 Midgut Midgut Midgut 1 2a Midgut Midgut Midgut Midgut Midgut Midgut Midgut Midgut Midgut Midgut 1 2 Midgut 1 2 Midgut Midgut Midgut Midgut Foregut 1 2 Foregut 1 2a Foregut Foregut Hindgut Hindgut 1 2 Hindgut Hindgut Overall response duration (mos) % Duration (mos) Radiologic response duration (mos) Posttreatment OR 7 55 7 SD 20 IFN OR SD SD 16 5 3 75 49 0 24 5 — OR SD SD 16 2 10 IFN, Oct IFN IFN OR OR OR OR 10 20 7 12 50 20 36 — 10 20 36 — OR OR OR OR 24 96 7 12 IFN IFN IFN IFN OR — OR OR OR OR OR SD SD SD SD 60 — 7 17 3 3 5 5 12 3 4 0 0 59 60 77 90 55 37 0 23 8 — — 7 20 3 3 5 5 — 3 4 OR — OR OR SD SD OR SD SD SD SD 60 — 19 17 12 3 5 10 12 24 4 None STZ, 5-FU IFN IFN Oct IFN, Oct IFN, Oct IFN IFN IFN, Oct IFN, Oct SD SD 5 3 38 20 5 3 SD SD 11 9 Oct Oct SD PD PD PD PD un 8 — — — — — 43 PD PD 85 20 un 8 — — 12 18 — SD SD SD PD PD un 11 8 7 — — — IFN IFN, Oct Oct IFN, Oct IFN; Oct IFN OR SD 11 10 60 47 11 10 SD SD 11 10 IFN, Oct IFN, Oct OR OR SD un OR 15 — 4 — 11 81 0 44 un — 15 — 4 — — OR — SD un OR 24 — 6 — 11 IFN, Oct IFN, Oct STZ, 5-FU, IFN Oct IFN SD OR SD SD 1 3 24 3 7 51 — 30 4 3 — 3 SD OR SD SD 1 9 24 3 IFN-␣, -␥ IFN-␣, -␥ IFN none OR: objective response; SD: stable disease; PD: progressive disease; IFN: interferon-␣; Oct: octreotide; STZ: streptozocin; 5-FU: 5-fluorouracil; un: undetermined. a Reembolization was performed within 2–4 mos without any signs of progression but rather regression, to enhance the effect. embolization were 85 and 40 months, respectively, and the 5-year survival rates were 67% and 40%, respectively. Causes of Death Among the 21 midgut carcinoid patients, 10 have died; tumor progression was the cause of death for 9 of Liver Embolizations for Neuroendocrine Gastrointestinal Tumors/Eriksson et al. 2297 TABLE 2 Effects of Hepatic Arterial Embolization on Patients with Endocrine Pancreatic Tumors Biochemical response Tumor type Nonfunc Nonfunc 1 2 Nonfunc Nonfunc 1 2 Nonfunc WDHA WDHA WDHA 1 2 3 4 Gastrinoma Gastrinoma Insulinoma Somatostatinoma Overall response duration (mos) % Duration (mos) Radiologic response duration (mos) Posttreatment OR 6 77 6 SD 11 None SD PD PD 6 — — 32 PD — 6 — — SD PD PD 12 — — IFN, Oct IFN, Oct None PD PD — — — — — — PD PD — — STZ, DOX STZ, DOX OR OR 6 12 77 87 8 12 SD SD 6 12 Oct STZ, DOX, IFN-␣, -␥ OR OR PD PD OR PD PD OR 5 5 — — un — — 16 94 57 PD PD 61 PD 50 48 5 5 — — un — 1 16 OR SD SD SD SD PD PD OR 7 — — — un — — 22 IFN-␣, -␥ IFN-␣, -␥ IFN-␣, -␥ Cispl, VP16 IFN-␣ STZ, DOX, IFN None IFN a a a OR: objective response; SD: stable disease; PD: progressive disease; IFN: interferon-␣; Oct: octreotide; STZ: streptozocin; DOX: doxorubicin; un: undetermined. a The patient died within 3 days after embolization due to liver necrosis. these patients, and the 10th died during surgery for carcinoid heart disease. Five of eight patients with foregut-hindgut carcinoid tumors died of tumor progression. Endocrine Pancreatic Tumors Response The therapeutic effect of liver embolization in EPTs is shown in Table 2. An overall objective response was achieved in 6 of 12 patients (50%), with a median duration of effect of 10 months. A biochemical response was demonstrated in 5 of 12 patients (42%), and a radiologic response was noted in 2 (17%). One patient (the case of somatostatinoma), who achieved both a biochemical and a radiologic response, showed progression in the untreated liver lobe 3 months postembolization (not indicated in Table 2). Three patients were later reembolized at signs of progression, with no effect achieved in two patients and a biochemical response achieved in one patient who then failed to respond to two further embolizations. One of 12 patients (8%) achieved SD with a duration of 6 months, whereas 2 patients (16%) had PD and 3 died of postembolic complications (2 hepatorenal syndrome, 1 liver necrosis). Survival analyses in EPTs showed a median survival of 60 months from diagnosis and a median sur- vival of 20 months from first embolization. The 5-year survival rate from diagnosis was 50%. There was no patient with EPT alive 5 years after embolization. Eleven of 12 patients have died, 3 of whom had complications from embolization. Other causes of death were tumor progression for five patients, myocardial infarction for one, and disseminated intravascular coagulation for one. Adverse Events All patients had some degree of postembolization syndrome (Table 3). Pain requiring injections of morphine analgesics, a febrile reaction, and leukocytosis were noted in the majority (⬎80%) of patients, reaching a maximum after 2–3 days. Nausea was experienced by 33% of patients. Persistent fever for longer than 1 week was noted in 7 patients, and in 5 a urinary infection probably caused by the urinary catheter (to ensure diuresis) was found. The other two patients had septicemia requiring systemic antibiotics. Elevations in liver enzymes occurred in ⬎90% of patients, indicating damage to normal liver parenchyma (transaminases) and possibly also tumor necrosis (lactate dehydrogenase) with a maximum on Day 2–3 after embolization. However, we found no significant difference in the increase of lactate dehydrogenase between responding and nonresponding 2298 CANCER December 1, 1998 / Volume 83 / Number 11 TABLE 3 Summary of Complications of 55 Hepatic Arterial Embolizations in 41 Patients with Neuroendocrine G-I Tumors Complications Frequency Comments Pain Pyrexia (⬎37.5°C) 48/54, 1 un 51/54, 1 un Infection 7/23, 32 un Leukocytosis 40/49, 6 un No. of days analgesics required: md 4 (range, 0–27) No. of days: md 7 (0–18) No prophylactic antibiotics 4 patients with persistent pyrexia received cortison 5 patients with persistent pyrexia had urinary infection, possibly due to catheter 2 patients with persistent pyrexia had septicemia Maximum value: md 13.65 (4.4–25.5) Day of maximum value: md 4.6 Nausea Hepatic enzyme derangement GOT 16/49, 6 un GPT 53/54, 1 un 52/53, 2 un Alkaline phosphatase 48/51, 4 un Bilirubin 33/49, 6 un LD 52/54, 1 un Renal insufficiency 2/55 Liver necrosis 1/55 Extrahepatic embolization 1/55 Carcinoid crisis 0/29 Maximum value: md 5.9 (0.55–59) Day of maximum value: md 2 Maximum value: md 2.6 (0.5–27) Day of maximum value: md 2 Maximum value: md 9.9 (2.9–63) Day of maximum value: md 7 Maximum value: md 25 (7.9–114.3) Day of maximum value: md 3 Maximum value: md 33 (4.7–347) Day of maximum value: md 2 Two early patients; thereafter avoided by hydration and forced diuresis Occurred 36 hours after embolization with fatal outcome. Septicemia with Clostridium perfringens and previous surgery (Whipple’s operation) could have contributed Small intestinal iscemia due to spillover of embolic material: the patient had an arterial anomaly Avoided by continuous infusion of octreotide the first 24 hours No. of days of hospitalization: md 12 (range, 6–52). GOT: glutamic–oxaloacetic transaminase; GPT: glutamic pyruvic transaminase; LD: lactate dehydrogenase; un: undetermined; md: median. patients. Moderate elevations of alkaline phosphatase and bilirubin levels were also noted in the majority of patients. Our first two patients in the early 1980s, with very advanced hepatic metastases of EPT, developed hepatorenal failure with fatal outcome. Since then, all the patients have been kept on a strict hydration scheme, and no renal complications have occurred. Serious adverse events were noted in two more patients. One patient, who had undergone Whipple’s operation because of a nonfunctioning EPT several years earlier, died of liver infarction on the third day after unilobar embolization. Retrospectively, there appeared to be nothing exceptional about the procedure, and the patency of the portal vein had been established. Blood cultures showed growth of Clostridium perfringens in this patient. The other patient, who was embolized despite an anomaly of the arterial tree (the right hepatic artery originated from the mesenteric artery), developed signs of intestinal ischemia with dilated small intestine and diffuse abdominal pain 5 days postembolization. Exploratory laparotomy revealed gangrene of a portion (1 meter) of the small intestine, which was resected. The patient survived several other postoperative complications and showed a dramatic biochemical response (90%). The median number of days the patients needed to recover from the postembolic complications, i.e., hospitalization days, was 12 days (range, 6 –52 days). DISCUSSION Neuroendocrine gastrointestinal tumors commonly metastasize to the liver, and the clinical consequences of these tumors are mainly seen as disabling symptoms related to the secretion of hormonal substances: the carcinoid syndrome with flush and diarrhea, or symptoms associated with EPTs, such as hypoglycemia, multiple ulcers, diarrhea, skin rash, and diabetes. Liver Embolizations for Neuroendocrine Gastrointestinal Tumors/Eriksson et al. Chemotherapy with streptozocin plus 5-fluorouracil or doxorubicin produces biochemical and/or tumor responses in about 50% of EPTs, which can last for approximately 2–3 years,3 whereas midgut carcinoid tumors are resistant to this treatment.4 In both carcinoid tumors and EPTs, IFN treatment is relatively effective, with responses in approximately 40 –50% of patients that have a duration of 2–3 years.5,6 The role for octreotide and other somatostatin analogs has been established in several series,7,8 and recently the importance of the presence of somatostatin receptors in the tumors for the therapeutic outcome has been demonstrated.16 The hepatic metastases of neuroendocrine gastrointestinal tumors derive almost all of their blood supply from the hepatic artery. The normal liver parenchyma receives only 20 –25% of its blood supply from this artery, whereas 75– 80% is supplied by the portal vein.17 Occlusion of hepatic arterial branches will cause varying degrees of necrosis of hepatic metastases, whereas the normal liver parenchyma is protected from infarction by its dual blood supply. Therefore, occlusion of the portal vein is a contraindication to arterial embolization. Because endocrine tumors and metastases are usually hypervascular, they should, at least theoretically, be sensitive to devascularization. Various particulate embolization materials have been used, e.g., gel foam powder, polyvinyl alcohol particles, to achieve peripheral arterial embolization, which is preferred to the proximal embolization used earlier (ligation of hepatic artery or the use of metal coils), because the collateral circulation will be less effective and the tumor necrosis therefore greater. Peripheral arterial embolization also allows repeated embolizations, because the vessels undergo recanalization. Our overall response rates of approximately 50% for both carcinoids and EPTs were somewhat lower than the 80 –90% that others have reported,12–14 but the duration of effect (median, 12 months for midgut carcinoids and 10 months for EPTs) was similar to the 11–15 months in other series.13,14 It is important to remember that all of our patients had advanced stage disease, having had their neuroendocrine tumor with liver metastases diagnosed for a median of 37 months, and having received systemic medical treatment during that time. The lower response rate could be explained by the advanced stage; but among our midgut carcinoid patients, which was the largest group, we could not find a significant difference in the duration of disease before embolization or pretreatment hormone levels between responders and nonresponders. It could also be that others have recommended repeated embolizations (up to 8),14 whereas we per- 2299 formed only one embolization on the majority of patients (11 of 41 patients were reembolized). When should patients be embolized? At least theoretically, the approach of reducing tumor burden first and then giving medical treatment appears to be preferrable. In a relatively recent randomized study,18 carcinoid patients were treated at the time of diagnosis with IFN with or without prior liver embolization, and the objective response rate after 1 year was higher in the embolized group (86%) than in the group that received IFN only (42%). However, embolization could not be shown to have any significant effect on survival. In this series, the 5-year survival rate was 38%. Several reports from another group have advocated embolization early in the clinical course of midgut carcinoids.19,20 Aggressive surgery, followed by octreotide treatment and unilobar or bilobar liver embolizations in 40 patients with midgut carcinoids, produced a 55% reduction of U-5-HIAA, which had still been maintained after 71 ⫾ 11 months, and the 5-year survival rate was 56%. Survival was calculated from the time of diagnosis of the carcinoid syndrome. In our current study, the median survival after diagnosis for patients with midgut carcinoids and liver metastases was 120 months, and the 5-year survival rate from diagnosis 83%. Liver embolizations were undertaken at a median of 37 months after diagnosis, but still the patient’s life expectancy after embolization was a median of 80 months and the 5-year survival rate 60%, indicating that “late” embolization was also very effective. Our data on malignant EPT embolized at a median of 37 months after diagnosis indicated that the median survival postembolization was only 20 months (from diagnosis, 60 months), suggesting that in this group earlier embolization may improve the results. The rate of occurrence of complications in patients with postembolic syndrome, including pain, fever, nausea, leukocytosis, and liver enzyme derangements, in our study was about the same as in other series.12–14 Severe complications, i.e., renal failure, liver necrosis, and intestinal ischemia, occurred in a total of 4 patients (10%), and 3 of them (7%) died. However, two of these patients were embolized in the early 1980s, when the risk for renal damage was not recognized. Since then, all our patients have been vigorously hydrated after the procedure, and we have had no further problems with renal insufficiency. One patient died of acute liver infarction 36 hours after embolization. At autopsy, this patient had a fresh portal thrombosis (the patency of the portal vein had been established before the embolization). Blood cultures also showed septicemia with Clostridium perfringens. It was unclear whether this had any signifi- 2300 CANCER December 1, 1998 / Volume 83 / Number 11 cance and whether prophylactic antibiotics may have helped the patient. It is also unclear whether the Whipple’s operation she had undergone made embolization more risky. The fourth patient, who had an extrahepatic spillover and accidental embolization of the small intestine, which was manageable with surgery, had an anomaly of the superior mesenteric artery that may have contributed to the occurrence of this complication. However, this type of anomaly is quite common, and other patients with the same anomaly have been safely embolized without complications. No hormonal crises occurred in our patients, possibly because of the octreotide infusion administered to patients with hormonal symptoms. To avoid reducing the blood flow to the liver before the embolization, which, at least theoretically, may have reduced the effect of embolization, we usually started the infusion of octreotide when the embolization had just been finished. This study began in the early 1980s, and both the embolization technique and the postembolization care has improved during this time, but the procedure is still associated with considerable toxicity and morbidity, which usually subside within 1–2 weeks. In our study, the median number of days of hospitalization was 12. Thus, hepatic artery embolizations are costly procedures not only for society, but also for the patients, in that their quality of life is impaired. However, this impairment is temporary, and if the embolization is successful it usually improves hormonal symptoms and hormone levels much more quickly than other therapies, i.e., biotherapy and chemotherapy. Which patients should be embolized? This question also remains to be answered. With the knowledge that we have acquired, that the procedure is associated with significant morbidity even in experienced hands, we think patient selection is imperative. We do not recommend embolization for patients with inoperable but few small lesions in the liver. The risk for complications also disqualifies patients with poor performance status (e.g., due to severe carcinoid heart disease) and patients with massive involvement of the liver who have very little normal liver parenchyma left. One group of patients that should definitely be considered for embolization is the relatively small group (10 –20%) who are somatostatin receptor negative according to octreotide scintigraphy. These patients would not benefit from standard treatment with currently available somatostatin analogs. In conclusion, it is very difficult to compare data regarding treatment results and survival between different groups of patients and also between treatment centers with different therapeutic traditions. In our study, with patients embolized “late” in the clinical course of the disease, the median survival from embolization for patients with midgut carcinoids was as long as 80 months, and the 5-year survival rate was 60%, similar to what others have achieved with “early” embolizations. We cannot see an obvious advantage to early embolization for patients in this category, and we suggest that both the individually chosen medical treatment (IFN and somatostatin analogs) and the embolization have contributed to the relatively long survival (median, 120 months after diagnosis) and the high 5-year survival rate of 83% for this patient category. Although the response rate was similar for patients with advanced EPTs who had received chemotherapy, IFN, and somatostatin analogs before embolization, the median survival after embolization was only 20 months. Even though the median survival after diagnosis of 60 months and the 5-year survival rate of 50% was better than in other series of malignant EPTs, earlier liver embolization (possibly before biotherapy) may improve the results. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. Norheim I, Öberg K, Theodorsson-Norheim E, Lindgren PG, Lundqvist G, Magnusson A et al. Malignant carcinoid tumors. An analysis of 103 patients with regard to tumor localization, hormone production, and survival. Ann Surg 1987;206:115–25. Eriksson B, Arnberg H, Lindgren PG, Lörelius LE, Magnusson A, Lundqvist G, et al. Neuroendocrine pancreatic tumors: clinical presentation, biochemical and histopathological findings in 84 patients. J Intern Med 1990;228:103–13. Öberg K, Norheim I, Lundqvist G, Wide L. Cytotoxic treatment in patients with malignant carcinoid tumors: response to streptozocin alone or in combination with 5-FU. Acta Oncol 1987;26:429 –32. Moertel CG, Lefkoponto M, Lipsity M. Streptozocin-doxorubicin, streptozocin-5-fluorouracil or chlorozotocin in the treatment of advanced islet-cell carcinoma. N Engl J Med 1992;326:519 –23. Öberg K, Norheim I, Lind E, Alm G, Lundqvist G, Wide L, et al. Treatment of malignant carcinoid tumors with human leukocyte interferon: long-term results. Cancer Treat Rep 1986;70:1297–304. Eriksson B, Öberg K. An update of the medical treatment of malignant endocrine pancreatic tumors. Acta Oncol 1993; 32:203– 8. Kvols LK, Moertel CG, O’Connell MJ, Schutt AJ, Rubin J, Halen R, et al. Treatment of the malignant carcinoid syndrome: evaluation of a long-acting somatostatin analogue. N Engl J Med 1986;315:663– 6. Maton PN. Use of octreotide acetate for control of symptoms in patients with islet cell tumors. World J Surg 1993; 17:504 –10. Breedis C, Young G. The blood supply of neoplasms in the liver. Ann J Pathol 1954;30:969 – 85. Liver Embolizations for Neuroendocrine Gastrointestinal Tumors/Eriksson et al. 10. Bierman HR, Byron RL, Kelly KH, Grady A. Studies on the blood supply of tumors in man: vascular patterns of the liver by hepatic arteriography in vivo. J Natl Cancer Inst 1951;12: 107–31. 11. Yamada K, Sato M, Kawabata M. Hepatic artery embolization of 120 patients with unresectable hepatoma. Radiology 1983;148:397– 401. 12. Mitty H, Warner R, Newman L, Train J, Parnes I. Control of carcinoid syndrome with hepatic artery embolization. Radiology 1983;149:79 – 83. 13. Carrasco CH, Chuang V, Wallace S. Apudoma metastatic to the liver: treatment by hepatic artery embolization. Radiology 1983;149:79 – 83. 14. Carrasco CH, Charnsangavej C, Ajani J, Samaan N, Richli W, Wallace S. The carcinoid syndrome: palliation by hepatic artery embolization. AJR Am J Roentgenol 1986;147:149 –54. 15. Eriksson B, Arnberg H, Öberg K, Hellman U, Wernstedt C, Wilander E. Chromogranins: new sensitive markers for neuroendocrine tumors. Acta Oncol 1989;28:325–9. 16. Tiensuu Janson E, Westlin JE, Eriksson B, Ahlström H, Nil- 17. 18. 19. 20. 2301 sson S, Öberg K, et al. [111In-DTPA-D-Phe1]-octreotide scintigraphy in patients with carcinoid tumors: the predictive value for somatostatin analog treatment. Eur J Endocrinol 1994;131:577– 81. Ackerman NB, Sien WM, Silverman NA. The blood supply of experimental liver metastases: the effects of acute ligation of hepatic artery or portal vein. Surgery 1972;71:636 – 41. Jacobsen M, Hanssen LE, Kolmanskog, Schrumpf E, Vatn MH. Interferon-␣ 2b, with or without prior hepatic artery embolization: clinical response and survival in midgut carcinoid patients. Scand J Gastroenterol 1995;30:31– 8. Geternd K, Tylén U, Jansson D, Stenqvist D, Tisell LE, Ahlman H. Hepatic arterial embolization in the treatment of midgut carcinoid syndrome and other advanced endocrine tumors metastatic to the liver. J Intervent Radiol 1990;5:69 – 76. Wängberg B, Westberg G, Tylen U, Tisell LE, Jansson S, Nilsson O, et al. Survival of patients with disseminated midgut carcinoid tumors after aggressive tumor reduction. World J Surg 1996;20:892–9.