When physicians analyze therapeutic options for patients with metastatic carcinoid tumors or islet cell carcinomas, they should consider the course of the patient's malignant disease, the distribution of metastasis, the severity of clinical manifestations, and the relative contributions of the hormonal syndromes and tumor bulk to symptoms and disability. The oncologist must be conservative, because these diseases frequently run an indolent course, allowing years of comfortable life with no treatment whatsoever. However, aggressive approaches may be justified when the patient develops disabling symptoms because tumor debulking or correction of mechanical problems may sometimes provide the patient with additional years of comfortable life.

If the patient has a clinically dominant hormonal syndrome and has no imminent threat from tumor bulk, it is frequently possible to produce substantial palliation with minimal therapeutic risk. In earlier years, only patients with gastrinoma could be helped in this manner by the use of histamine-2 blockers (for example, adequate doses of cimetidine or ranitidine or, more recently, omeprazole [Prilosec, Merck and Company, West Point, Pennsylvania]).

The development and clinical application of an analog of somatostatin (octreotide; Sandostatin, Sandoz Pharmaceuticals, East Hanover, New Jersey) has provided a novel and frequently highly effective tool for control of almost all the endocrine syndromes. Striking decreases in hormonal levels and substantial or complete relief from symptoms can be achieved in approximately 80% of patients [1, 2]. Although octreotide usually produces only minimal immediate side effects, gallstones are a frequent long-term complication. However, octreotide therapy is cumbersome to the patient and is costly. Further, most patients become resistant to this therapy; for patients with carcinoid tumor, the median interval is about 1 year and for patients with islet cell carcinomas, only a few months. For these reasons, octreotide therapy for the carcinoid syndrome should be reserved for patients with severe flushing or diarrhea. In those with islet cell carcinomas, it should be used primarily to restore the patient to a better general and nutritional status so that more aggressive and definitive therapy can be undertaken with an acceptable risk.

Regression rates after systemic chemotherapy or immunotherapy (using either single drugs or drug combinations) for metastatic carcinoid tumors and islet cell carcinomas have been variable, as is typical for solid tumors [3]. For carcinoid tumors, regression rates (even with our most effective regimens) seldom exceed 20%, and the discouragingly short duration of these responses hardly justifies the risks and side effects of treatment. For this reason, we urge that systemic therapy for carcinoid tumors be carried out in a research setting. For islet cell carcinoma, however, treatment may be of true clinical value. Streptozocin (Zanosar, The Upjohn Company, Kalamazoo, Michigan) seems to have specific cytotoxicity for the islet cell carcinoma and produces credible tumor regressions in about 30% of patients. Because of its minimal hematologic toxicity, it is useful in drug combinations with chemotherapeutic agents that are dose limited by leukopenia and thrombocytopenia. In a randomized trial, fluorouracil plus streptozocin was found to produce a 45% regression rate and seemed to improve survival when compared with streptozocin alone [4]. In a recent randomized trial, the combination of doxorubicin and streptozocin increased the regression rate to 69% and produced a statistically significant increase in survival [5]. However, fewer than one third of patients have a substantive and prolonged response to chemotherapy, and it produces considerable toxicity.

A special consideration in managing these patients is the usually dominant involvement of the liver in the metastatic process. For selected patients, the surgeon can make a major and usually long-lasting contribution by resection of large solitary tumor masses or well-localized clusters of metastatic lesions [6]. With modern surgical technology, operative morbidity and mortality have been reduced to acceptable levels. Among 40 such patients reported by McEntee and colleagues [6], only a single postoperative death occurred. Usually, however, the diffuse nature of hepatic metastasis precludes effective surgical debulking. Therefore, hepatic artery infusion of various drugs has been attempted but with little documented evidence of success. Occlusion of hepatic arterial flow is attractive, because metastatic tumor neovascularity and oxygenation derive almost entirely from the hepatic artery. In contrast, hepatocytes are resilient, and viability may be maintained through the portal vein until rearterialization occurs. Although many researchers have shown that hepatic artery occlusion effectively shrinks primary or metastatic cancer in the liver, for the more common and more aggressive malignant diseases, tumor regrowth occurs so quickly that any net gain is difficult to discern. Because carcinoid tumors and islet cell carcinomas usually grow slowly, lasting benefit may be possible using hepatic artery occlusion.

This review documents our experience with hepatic arterial occlusion, by either surgical ligation or embolization, in selected patients with neuroendocrine tumors and hepatic-dominant metastatic disease. Although frequent and substantial tumor regressions were produced, the duration of these regressions, even with these indolent neoplasms, was discouragingly short. We therefore developed and tested a regimen of sequential hepatic arterial occlusion and chemotherapy with the hope that regressions could be enhanced and prolonged. For chemotherapy, we elected to give each of the drugs that have been most active for these tumors when used alone (that is, fluorouracil, streptozocin, doxorubicin, and dacarbazine [DTIC]).

Methods

Top Methods Results Discussion Author & Article Info References

Patient Selection

All patients selected for study had histologically confirmed carcinoid tumor or islet cell carcinoma with metastasis clinically limited to or dominant in the liver. They were ambulatory and maintaining a reasonable state of oral nutrition (at least 1200 calories daily). One or more measurable parameters of malignant disease were required to serve as objective indicators of response to therapy. For liver metastasis, this required clearly demarcated lesions on liver imaging that measured at least 5 cm in greatest diameter using radioisotope scanning or at least 3 cm in diameter using computed tomography or magnetic resonance imaging. Alternatively, if the patient had malignant hepatomegaly with a distinct liver edge extended at least 5 cm below the xiphoid or costal margins on quiet respiration, this was also used as a marker for response to therapy. In the absence of measurable liver involvement, hormonal assays could be used alone as markers for response to therapy, if they were greater than twice the upper limit of normal.

Contraindications to study entry were a total serum bilirubin level more than 51.3 µmol/L (3 mg/dL), previous radiation therapy to the liver or hepatic arterial chemotherapy, and the standard contraindications for each of the involved cytotoxic drugs. In practice, protocol entry was limited to those patients who, in the judgment of the physician, had clinically significant symptoms related to tumor bulk or to the endocrine syndrome or had extensive metastasis associated with abnormal test results for liver function.

Treatment

Hepatic artery occlusion was done by surgical ligation in those patients who had other specific indications for abdominal surgery (for example, an obstructing carcinoid tumor of the small bowel) or for those patients in whom catheterization and embolization were not considered to be technically feasible. For all other patients, occlusion was done by catheterization and embolization. Before hepatic artery occlusion by either method, the anatomy of the hepatic vasculature was determined by angiography. If any evidence of portal vein occlusion or compromise of portal vein flow was noted, occlusion was not done.

In patients having operative tumor devascularization, the hepatic artery was approached for ligation through the gastrohepatic ligament of the lesser omentum. The primary right and left hepatic arteries were ligated and divided individually distal to the gastroduodenal artery. Dearterialization was completed by division of the entire gastrohepatic omentum from the diaphragm to the bile duct. Any accessory or replaced hepatic arteries were also identified, ligated, and divided; lymphoareolar tissue in the hepatoduodenal ligament was also divided. Cholecystectomy was done to avoid gallbladder ischemia. For occlusion by embolization, the hepatic artery was selectively catheterized with injection of embolic material, usually gel foam or polyvinyl alcohol or both. Because of the technical difficulty encountered in the procedure, the pretreatment impairment of liver function, and the total mass and distribution of hepatic metastasis, eight patients had embolization done at two or three sittings separated by 3 to 4 work-week intervals.

After hepatic artery occlusion, the patient was observed in the hospital for a minimum of 5 days to monitor complications. In some instances, prophylactic antibiotics were used. The patient was released from hospital when there was no clinical evidence of complications, definite recovery of abnormal test results for liver function, and reduction of fever.

Sequential chemotherapy was initiated 3 to 4 weeks after the occlusion procedure, first with doxorubicin at a dose of 60 mg/m² on day 1 and dacarbazine at a daily dose of 250 mg/m² on days 1 to 5, both by intravenous injection. Four weeks later, streptozocin was given at 500 mg/m² and fluorouracil at 400 mg/m², both by intravenous injection daily for 5 days. At 8 weeks, this cycle was repeated, but the interval between initiation of individual courses was increased to 5 weeks. Earlier in our study, this cyclic chemotherapy was continued until tumor progression. Because responding patients became intolerant to such long-term chemotherapy, we changed this plan to one of continuing chemotherapy until the patient had stabilized at a maximum tumor regression (a median of 6 months of treatment). Doxorubicin was limited to a total dose of 400 mg/m (²). Streptozocin was discontinued if the serum creatinine level reached 177 µmol/L (2.0 mg/dL) or greater.

Objective tumor regression required a greater than 50% decrease in the product of longest perpendicular diameters of lesions seen using liver imaging (magnetic resonance or computed tomography or radioisotope scanning) or, in the case of hepatomegaly, a 30% decrease in the sum of the distances below the xiphoid and costal margins at the midclavicular lines. Hormonal variables were required to be decreased by at least 50%. Most patients had measurable tumor and hormonal variables, and both of the above criteria were required for a regression to be declared. The timing of events related to therapy was measured from the day of the first hepatic artery occlusion. The duration of regression was measured to the last day on which the patient met criteria for regression. Survival and interval-to-progression distributions were estimated by the Kaplan-Meier method [8].

Results

Top Methods Results Discussion Author & Article Info References

Between September 1982 and June 1991, 123 patients gave their consent to participate in our studies. In three patients, hepatic artery occlusion could not be attempted because angiography showed either occlusion or substantial impairment of the portal vein flow. In nine other patients, embolization was not technically possible. Among the remaining 111 patients, 48 had hepatic artery occlusion done by surgical ligation and 63 had catheterization and embolization. Forty patients were selected for treatment with hepatic artery occlusion only, and 71 received sequential occlusion and chemotherapy. Sixty-five patients had carcinoid tumors, and 46 had islet cell carcinomas. Nine patients selected for sequential therapy did not receive subsequent chemotherapy—4 because of death due to complications of hepatic artery occlusion, 3 because of clinical deterioration, and 2 because of patient refusal. These patients, however, are included as having received sequential therapy in all analyses below.

The characteristics of our patients are shown in Table 1 according to tumor type and treatment regimen. Probably because more attractive chemotherapy alternatives are available for islet cell carcinoma, 13 of 17 of those patients who were managed with hepatic artery occlusion alone had previous chemotherapy compared with only 8 of 23 patients with carcinoid tumor. Sixteen patients, 11 with islet cell carcinomas and 5 with carcinoid tumors, had previous octreotide therapy. Patients with islet cell carcinoma who were treated with occlusion alone had a much longer duration of known hepatic metastasis than did patients with carcinoid tumor or patients with islet cell carcinoma who were treated with sequential occlusion and chemotherapy. They also more frequently had substantial impairment of performance status. Certainly, all of these factors could have a substantive influence on response to therapy and survival. All patients had advanced hepatic metastasis, and measurable hepatic tumor was present in 80%. Among our patients with carcinoid tumor, the median 5-hydroxyindoleacetic acid level (5-HIAA) was more than 1046 µmol/d (200 mg/24 h); normal levels are 10 to 31 µmol/d (2 to 6 mg/24 h). Ninety-two percent of patients had clinical manifestations of the carcinoid syndrome. Only two patients, one with a gastric and one with a rectal primary tumor, had no increase in 5-HIAA levels. Hormonal excess was documented in 83% of patients with islet cell carcinoma, and 28% had multiple endocrine abnormalities. The most common was the Zollinger-Ellison syndrome; the median gastrin level in patients with this syndrome was 5000 ng/L and ranged to 460 000 ng/L (normal, 0 to 200 ng/L). Only three patients, all with carcinoid tumor, had evidence of extra-abdominal metastasis (one each with bone, pulmonary, and supraclavicular nodal).

Therapeutic Results

Among the 40 patients treated with hepatic artery occlusion alone, our overall regression rate was 60% (95% CI, 43% to 75%) and among the 71 patients who had sequential occlusion and chemotherapy, it was 80% (CI, 69% to 89%) (Table 2). Tumor regressions were unequivocal, and those shown in Figure 1, Figure 2, and Figure 3 are typical. Other examples are shown in an earlier publication [3]. The median duration of regression with occlusion alone was only 4 months (CI, 2 to 8 months) compared with 18 months (CI, 14 to 23 months) with the combined therapy. These differences were comparable for islet cell carcinoma and for carcinoid tumor, and similar differences were seen in duration of time to tumor progression.

View larger version (90K): [in this window] [in a new window]   Figure 1. Metastatic gastrinoma. Before (left) and after (right) sequential hepatic artery occlusion and chemotherapy. Gastrin levels are given in ng/L.

View larger version (122K): [in this window] [in a new window]   Figure 2. Metastatic polyfunctional islet cell carcinoma. Before (left) and after (right) sequential hepatic artery occlusion and chemotherapy. For SI units: gastrin and glucagon (pg/mL = ng/L); multiply 5-HIAA values by 5.23 to yield micromole/d; and multiply insulin values by 7.175 to yield pmol/L. 5-HIAA = 5-hydroxyindoleactic acid.

View larger version (88K): [in this window] [in a new window]   Figure 3. Metastatic nonfunctioning islet cell carcinoma. Before (left) and after (right) sequential hepatic artery occlusion and chemotherapy.

For both tumor types, hormonal responses were somewhat more frequent than were tumor regressions. Complete tumor regressions were relatively uncommon, as were complete hormonal regressions in patients with islet cell carcinoma who were treated with hepatic artery occlusion alone. However, the combined regimen produced complete hormonal regressions in more than half of patients with islet cell carcinoma compared with only 12% of patients with carcinoid tumors, perhaps reflecting a greater responsiveness to chemotherapy. This difference did not reflect a greater resistance of serotonin-producing neoplasms, because five of nine patients with serotonin-producing islet cell carcinomas who were treated with the combined regimen had a complete hormonal response (52%).

Although numbers for comparison were small, no suggestion was evident that any specific hormonal production rendered islet cell carcinomas either preferentially sensitive or resistant to therapy. All responding patients with islet cell carcinomas had complete or substantial relief from endocrine syndromes. Hormonal responses were also substantive in patients with carcinoid tumors. The median overall decrease in 5-HIAA levels among responding patients was 75% with hepatic artery occlusion alone and was 89% when chemotherapy was added. Among responding patients, 98% had substantial improvement of pretreatment flushing and 88% had improvement of diarrhea. More than half had complete relief from these symptoms. Figure 4 and Figure 5 show survival curves from the onset of therapy. In patients with islet cell carcinomas, the median survival times were 35 months (CI, 25 to 51 months) with the sequential regimen and 9 months (CI, 14 to 19 months) with hepatic artery occlusion alone. In patients with carcinoid tumors, the median survival times were 49 months (CI, 35 to 69 months) and 27 months (CI, 10 to 61 months), respectively.

View larger version (11K): [in this window] [in a new window]   Figure 4. Islet cell carcinoma.Top. Survival from onset of therapy after hepatic artery occlusion alone. Bottom. Survival from onset of therapy after hepatic artery occlusion plus chemotherapy. Pt = patients.

View larger version (7K): [in this window] [in a new window]   Figure 5. Carcinoid tumor. Left. Survival from onset of therapy after hepatic artery occlusion alone. Right. Survival from onset of therapy after hepatic artery occlusion plus chemotherapy. Pt = patients.

In a comparison of the relative effectiveness of hepatic artery occlusion done by surgical ligation or by embolization in patients with metastatic carcinoid tumor, ligation produced a more favorable regression rate (80% compared with 68%) and duration of regression (16 months compared with 12 months). Patients were not randomly assigned to these two approaches, and differences could be related to patient selection.

Side Effects and Toxicity

After embolization, nausea, vomiting, and liver pain were common side effects, usually subsiding in a few days, although some patients continued to have pleuritic-type discomfort for 3 to 4 weeks. These side effects could not be adequately discerned in surgical patients, because they were usually intubated and were having incisional pain. Fever was almost universal, with highest temperatures observed 2 to 3 days after the procedure and a return to normal within 5 to 6 days. Occasionally, however, patients had low-grade fever continuing for 2 to 3 weeks. If patients had persistent high spikes of fever, the possibility of complicating abscess or sepsis was considered. Almost all patients had rapid increases in serum aminotransferase levels that usually increased to more than 8.34 µkatal/L (500 U/L) (normal, to 0.52 µkatal/L [31 U/L]) 2 to 3 days after the procedure and then decreased rapidly. Less frequently, mild increases were noted in alkaline phosphatase and serum bilirubin levels that generally peaked on day 4 or 5. Continued mild abnormalities in liver function test results were sometimes seen as long as 4 months after occlusion. Usually mild increases in serum creatinine levels were also seen 4 to 5 days after the procedure. Distinctly different severities of side effects were noted when comparing the two methods of hepatic artery occlusion. Temperature increases tended to be higher after embolization, whereas extreme increases in serum aminotransferase levels and increases in serum bilirubin levels were more frequent with surgical ligation. Increases in serum creatinine levels were more frequent with embolization, presumably because of the contrast material used for visualization.

Major complications of hepatic arterial occlusion occurred in 13 patients (12%). Four patients had a severe hepatorenal syndrome with both jaundice and renal failure; 3 of these patients died. One patient died from acute hepatic necrosis without renal failure, and 1 developed prolonged but nonfatal renal failure without liver failure. Two patients, both treated with embolization, had gallbladder perforation with abscess, and 1 of these patients died. One patient with the Zollinger-Ellison syndrome developed a perforated duodenal ulcer after embolization, and a second had gastrointestinal hemorrhage from duodenal ulceration. One patient developed a nonfatal carcinoid crisis during surgery, 1 had a prolonged period of adynamic ileus after embolization, and 1 had an abscess in a large tumor mass requiring surgical drainage. Ten of these 13 major complications occurred after embolization, as did 4 of the 5 deaths.

Toxic reactions to chemotherapy were dominated by nausea and vomiting that occurred in about 90% of patients treated with each regimen. A few patients had mild stomatitis and diarrhea. Alopecia was a universal side effect. About half of the patients had some degree of leukopenia, usually mild, and thrombocytopenia was infrequent. Thirty-two percent of patients had small increases in creatinine levels (all to less than 221 µmol/L [2.5 mg/dL]), which first occurred after combined streptozocin and fluorouracil therapy. A single drug-related death occurred that was caused by profound leukopenia, thrombocytopenia, and sepsis after the first course of combined dacarbazine and doxorubicin.

Discussion

Top Methods Results Discussion Author & Article Info References

Our study shows that most symptomatic patients with neuroendocrine carcinomas and advanced hepatic metastasis have substantial tumor regressions and relief from endocrine syndromes after occlusion of the hepatic artery. When this treatment was used alone, however, the duration of regression was only a few months. Considering the side effects, risk to life, necessity for hospitalizations, and substantial cost, our data would not justify hepatic artery occlusion alone as standard therapy. It may be used for patients who have failed chemotherapy, but expectations must be limited. Most embolized patients had only one embolization, and previous patients were embolized with gel foam alone. Currently, we are embolizing with smaller particulate materials and most commonly in two stages. The M.D. Anderson Hospital group has used up to 12 embolizations (average, 3 or 4 per patient) [9, 10]. In patients with carcinoid tumor, this group obtained objective regressions (by the same criteria that we used) in 10 of 23 patients (43%), with a median duration of regression in excess of 21 months [9]. Among 21 patients with islet cell carcinoma, they observed objective regressions in 12 (57%); the median duration was not stated [10].

Adding planned sequential chemotherapy to hepatic arterial occlusion has yielded a regression rate of 80% for a median duration of 18 months (one patient as long as 7 years). This suggests that for many patients the therapeutic gain probably merited the price in dollars and discomfort. Our most recent experiences have suggested that the major problems of nausea and vomiting during chemotherapy can be substantially relieved by ondansetron (Zofran; Glaxo Pharmaceuticals, Research Triangle Park, North Carolina). We also feel that certain precautions can minimize the occurrence of major complications of hepatic artery occlusion. All patients should be well hydrated. Those with the Zollinger-Ellison syndrome should be brought to optimal pharmacologic control. In patients with pancreatic cholera (the vasoactive intestinal peptide syndrome), severe hypoglycemia, or severe carcinoid syndrome, we have used a few days of octreotide therapy before and after occlusion, until the patient has stabilized. Even for patients with only mild or moderate carcinoid syndromes, octreotide should be immediately available in the operating room or catheterization suite.

Although the different study populations cannot be strictly compared, we believe our combined modality therapy for carcinoid tumor is more favorable than chemotherapy alone. Our regression rates with chemotherapy alone, regardless of regimen, have been low, usually in the 20% to 30% range [3], and have usually been transient. Experience with chemotherapy alone in islet cell carcinoma, however, has been more favorable, with regression rates up to 69% [5], but rates are still lower than those for our combined modality regimen. Our data suggest that our combined modality therapy has altered the natural course of these diseases. All patients who received chemotherapy subsequent to hepatic artery occlusion had a large hepatic tumor burden as reflected by their excessive hormonal levels. Thus, in patients with islet cell carcinoma, the median gastrin level was 5000 ng/L (range, 400 to 460 000 ng/L); among our patients with carcinoid tumor, the median 5-HIAA level was 1203 µmol/d (230 mg/24 h) (range, 68 to 3326 µmol/d [13 to 636 mg/24 h]). Despite these markers of late-stage disease, the median interval to tumor progression was 22 months for patients with islet cell carcinoma and 24 months for patients with carcinoid tumor. Median survivals from onset of therapy were 35 months and 49 months, respectively. These numbers compare favorably with our own previous experience as well as that reported by others. These differences could have resulted from patient selection, however, because we did not do a randomized study. Our assumption that sequential chemotherapy adds to the effectiveness of hepatic artery occlusion might also not be correct for the same reason. Here, any comparative differences could be influenced by differences in pretreatment characteristics (for example, duration of disease and previous chemotherapy). Unfortunately, the rarity of these diseases, the long-term observations required to assess end results, and the complexity of the treatment procedures make it unlikely that these questions can be addressed with an adequate randomized, controlled trial.

In Table 3, we show a schema for the current management of metastatic carcinoid tumors or islet cell carcinomas. These recommendations must, of course, be modified by clinical judgment based on such factors as general and nutritional status of the patient, aggressiveness of malignant disease, organ function impairment, and comorbid disease states. Physicians should consider offering the combined treatment approach that we have described for patients with hepatic-dominant metastasis of well-differentiated neuroendocrine tumors who have substantial symptoms related to tumor bulk or who have life-threatening endocrine syndromes that cannot be controlled by simpler means. We feel this recommendation is appropriate for islet cell carcinoma where chemotherapy is clearly effective and where long-term disease control can be achieved with our combined regimen. In the case of carcinoid tumor where chemotherapy alone has only a marginal effect, the evidence is less persuasive. Disease control with hepatic artery occlusion alone was discouragingly short in our nonrandomized comparison with the combined regimen. We recommend that patients with carcinoid tumor be treated in a research setting using protocols designed to improve the effectiveness of the systemic therapy.