Hepatic resection for metastatic neuroendocrine carcinomas

Hepatic Resection Neuroendocrine

for Metastatic Carcinomas

Florencia G. Que, MD, David M. Nagorney, MD, Kenneth P. Batts, MD, Rochester, Minnesota, Laurie J. Linz, MD. Rochester, Minnesota and St. Louis. Missouri, Larry K. Kvols, MD, St. Louis, Missouri BACKGROUND: Metastatic neuroendocrine malignancies frequently cause incapacitating endocrinopathies, and metastases predominant in the liver. Hepatic resection of metastases from such tumors is attractive because the natural history of neuroendocrine tumors is protracted, clinical severity of the endocriuopathy correlates with tumor volume, and local and intrahepatic growth characteristics often allow complete resection. EATIENTS AND METHODS: TO define the role of hepatic resection for me&static neuroendocrine tumors, the records of 74 patients who underwent hepatic resection for such tumors between 1984 and 1992 were reviewed. Neuroendoerine tumors were classified by site of origin and clinical endocrinopathy. Survival, and type and duration of symptomatic response, were assessed as the major outcomes of this study. HEWLTS: There were SO carcinoid, 23 islet-cell, and 1 atypical neuroendocrine tumors. Resections included 36 hemihepatectomies or extended hepatectomies and 38 nonanatomic resections. Thirty-eight primary tumors were resected concomitantly. Perioperative mortality was 2.7% and morbidity was 24%. Four-year survival was 73%. Overall postoperative symptomatic response rate was 90% with a mean duration of response of 19.3 months. CONCL~IYIONS: Hepatic resection for metastatir neuroendocrine malignancies is safe, provides effective palliation, and probably prolongs survival,

lthough authors have advocated hepatic resection in the management of metastatic neuroendocrine A malignancies, l-7 clinical experience is limited because the some

incidence of these tumors is low. The attractiveness of hepatic resection for functioning neuroendocrine malignancies has been based on several observations: the protracted natural history of most neuroendocrine malignancies. the clinicopathologic growth characteristics of these tumors, and the clinical impression that the severity of endocrinopathy correlates with tumor volume. Clinical experience during the past 2 decades has shown clearly that selected patients with functioning neuroendocrine malignancies can obtain From the Department of Surgery (FGQ. DMN), Department of Lahoratot?, Medicine and Pathology (KPB, LJL) and Division of Medical Oncology (LKK). Mayo Clinic & Mayo Foundation. Rockwater Minnesota. Requests for reprints should be addressed to David M. Nagwney. MD. Department of Surgery, Mayo Clinic. 200 First Street SW. Rochester. Minnesota 55905. Presented at the 35th Annual Meeting of The Society tar Surgery r)j the Alimentary Tract, New Orleans. Loukiana. May 16--1X. 1994.

36

THE AMERICAN

JOURNAL

OF SURGERY=

VOLUME

169

resolution of endocrinopathies and resumption of a normal performance status after resection of all gross disease.4,7-” In fact, some patients will be cured of their disease. Other patients, however, despite apparent complete resection, experience progression of residual metastatic disease. recrudescence of clinical endocrinopathies, and eventually cancer-related death. Whether the degree and duration of clinical benefit from hepatic resection for neuroendocrine melastases outweighs operative risk or the benefits from other therapeutic alternatives is unresolved. Our initial report showed that cytoreductive hepatic surgery for metastatic neuroendocrine malignancy was both safe and temporarily efficacious. and thus supported broader clinical application.4 The potential for cure and the durability of palliation remained unclear. We have subsequently pursued hepatic resection as a primary treatment for symptomatic patients with hepatic metastasesfrom neuroendocrine tumors when resection of nearly all gross disease was considered technically feasible. Here we review the clinical outcomes of our updated experience with 74 patients and define further guidelines for patient selection. PATIENTS

AND

METHODS

We reviewed the medical records of all patients who underwent hepatic resection for metastatic neuroendocrine malignancies from March 1984 through December 1992. Demographics, clinical presentation, laboratory and hormonal findings, operative management, and hospital morbidity and mortality were recorded for all patients. Each neuroendocrine malignancy was classified by site of origin and type of hormone production. An endocrinopathy was considered present if elevated serum hormone levels were documented or clinical signs and symptoms were consistent with a well-established endocrinopathy. Endocrinopathies could not be quantitated or qualified further because of the retrospective study design. Resections were considered curative in intent only if all gross primary, regional. and hepatic extent of the neuroendocrine malignancy was resected-ie, there was no gross residual disease. Resections were considered palliative in intent if any gross residual disease was present either intra-abdominally at operation or extra-abdominally by imaging. Patients with metastatic neuroendocrine malignancies were selected for hepatic resection if the primary tumor had been resected, or was potentially completely resectable, and metastases were predominantly hepatic. Exploration for hepatic resection was pursued if preoperative imaging studies suggested that the primary and regional tumor and 90% or more of the hepatic tumor volume were resectable. Preoperative evaluation included complete history and physical examination, laboratory studies including complete blood cell count (CBC); serum chemistry, hematology, and coagulation profiles; and serum protein electrophoresis and hormonal marker levJANUARY

1995

NEUROENDOCRINE

els. Patients were examined for evidence of carcinoid heart disease. When appropriate, echocardiography was performed to evaluate cardiac function. If indicated, cardiac valve replacement was performed before hepatic resection to improve right heart function. Gastrointestinal barium contrast studies with or without enteroclysis were used to define the site and extent of intestinal carcinoids. The size, number, location, and extent of hepatic metastaseswere defined routinely by computed tomography (CT) and ultrasonography (US). Magnetic resonance imaging (MRI) was used selectively to define the relationship of the hepatic metastases to the major intrahepatic vasculature. Abdominal CT was used to estimate resectability of islet-cell carcinomas. Patients with carcinoid syndrome were injected subcutaneously with 150 to 500 ug of long-acting somatostatin analogue (octreotide) as preoperative prophylaxis for carcinoid crisis. Additional intravenous doses were injected intraoperatively as deemed necessary by the anesthesiologist for maintenance of hemodynamic stability. The type of hepatic resection employed was dictated by the extent of hepatic metastases. Types of resection included: right hepatectomy (polysegmentectomy V-VIII), left hepatectomy (polysegmentectomy II-IV), extended hepatectomy (right or left hepatectomy with either contiguous or noncontiguous anatomic or nonanatomic resection of a part of the contralateral lobe), and nonanatomic resections defined as nonsegmental resections whether intra- or intersegmental. hrtestinal resections for carcinoids included the involved stomach, small intestine. colon, and mesentery. Intestinal continuity was restored by gastroenterostomy, enteroenterostomy, or enterocolostomy. Isletcell carcinomas encountered in this series arose in the distal gland and were resected by distal pancreatectomy and splenectomy. Tumors were classified into the following types: carcinoid, insulinoma, glucagonoma, polyfunctioning islet-cell carcinoma, gastrinoma, or nonfunctioning neuroendocrine. The classification was based on the tumor hormonal products, clinical presentation, and, in some cases, immunohistochemistry performed on resected tissue. The histopathology of each tumor was reviewed independently by two pathologists. Data on gross tumor size and distribution were obtained from original pathology reports. Tumor type and grade were confirmed microscopically. Each neoplasm was graded 1 to 4 using a modification of existing criteria for neuroendocrine neoplasms of the lung.” No attempt was made to distinguish islet-cell tumors from nonislet-cell tumors by growth pattern. Grade 1 neuroendocrine carcinomas were defined as neoplasms that had a “classic” carcinoid tumor growth pattern and showed minimal pleomorphism, scant mitotic activity (53 per 10 high power fields [HPF]), and negligible necrosis. Grade 2 carcinomas were roughly analogous to pulmonary “atypical carcinoid” tumors. demonstrating a preserved organoid growth pattern but containing a greater degree of pleomorphism and either focal tumor necrosis or moderate mitotic activity (4 to 10 per 10 HPF). Tumors of grade 1 or 2 were regarded as “low grade,” while grade 3 and 4 carcinomas were designated as “high grade.” Grade 3 lesions consisted of non-small-cell neuroendocrine neoTHE AMERICAN

CARCINOMAlQUE

ET AL

plasms with ample (>I0 per 10 HPF) mitotic activity or abundant necrosis, with or without pleomorphism. Grade 4 carcinomas were analogous to small-cell carcinoma of the lung, consisting of small cells with scanty cytoplasm, nuclear molding, and ample mitotic activity with or without necrosis. Follow-up data were obtained through patient interviews and periodic examinations with imaging and laboratory evaluations. An effort was made to determine whether symptomatic responses were due to reduced hormone production or alleviation of mechanical tumor-related factors. Changes in serum tumor markers were documented when available to assess objective response. Clinical and hormonal responses were classified as complete, partial, or negative. Duration of clinical response was defined as the time interval between hepatic resection and the first documented recrudescence of symptoms. Actuarial survival curves were calculated by the Kaplan-Meier method.‘* Comparisons of survival curves for discrete variables were made with the log-rank test.Ii Differences in survival were considered significant if P 50.05.

RESULTS Seventy-four patients underwent hepatic resection for metastases from neuroendocrine malignancies. Forty-six were women and 28 men. Their mean age was 55 years (range 25 to 77). The hepatic metastases arose from the following neuroendocrine malignancies: carcinoid tumor (n = 50), glucagonoma (n = 8), multihormonal islet-cell carcinoma (n = 7), nonfunctioning islet-cell carcinoma (n = 5), gastrinoma (n = 2), and insulinoma (n = 1). One patient’s tumor was designated simply as a neuroendocrine malignancy because it could not be subclassified further. Carcinoids arise from the midgut (pancreas, jejunum, ileum, appendix) in 41 patients, the foregut (bronchus, stomach, duodenum) in 7, and the hindgut (colon, rectum) in 2. Both gastrinomas originated in the pancreas. No patient had a multiple endocrine neoplasia syndrome, though 1 patient with a gastrinoma was later found to have a pheochromocytoma. Forty-five patients had histologic grade 1 hepatic metastases, 27 grade 2, and 2 grade 3. Clinical differentiation between symptoms due to functioning hormonal excess and symptoms of mechanical origin related to the primary tumor was difficult. Presenting symptoms at evaluation for hepatic resection were considered to have origins that were primarily hormonal in 5 1 patients, mechanical in 11, and combined in 6. One patient had anemia due to chronic gastrointestinal hemorrhage but denied symptoms. Six patients were asymptomatic. Among the 50 patients with carcinoid tumors, diarrhea (n = 29) and flushing (n = 22) were the most common symptoms. Other clinical features included abdominal pain, abdominal distention, wheezing, syncopal episodes, nausea, vomiting, and audible borborygmi. Clinical findings associated with islet-cell carcinomas were variable. In the 8 patients with glucagonomas. anorexia, weight loss, and necrolytic migratory erythema were the most common findings. Both patients with gastrinomas had diarrhea. Other clinical manifestations associated with islet-cell carcinomas included diarrhea, epigastric pain, fatigue, weight loss, and lightheadedness. JOURNAL OF SURGERY”

VOLUME

I69 JANUARY

1995

37

NEUROENDOCRINE --.___

01 0

wenlsnrbk 74

0-l 0 PUiOllti&rirk

FZC

CARCINOMA/QUE

_

ET AL

2

1

Years following resection 33 es

2

THE AMERICAN

19

9

I

Years following resection E :: 1:

JOURNAL

OF SURGERY”

VOLUME

169

Figure 1. Overallsurvival for patients after hepatic resection of metastatic neuroendocrine carcinoma.

4

3

The hepatic resections performed in this study included right or left hepatectomy (n = 22), right or left hepatectomy with anatomic extension (n = 7), right or left hepatectomy with nonanatomic extension (n = 7), and nonanatomic resection (n = 38). Nonanatomic resections were multiple in 26 of 38 patients. Diagnosis of the primary tumor and hepatic metastases were synchronous and resection of both was concurrent in 37 patients. Nonhepatobiliary procedures included partial small-bowel resection (n = 22), partial ileocolectomy (n = 4), partial pancreatectomy with or without splenectomy (n = 9) and partial gastrectomy (n = 2). Two deaths occurred within 30 days postoperatively or during hospitalization, for a 2.7% postoperative mortality rate. One patient died as a result of intracranial hemorrhage and the other, in whom palliative resection was ineffective, of a refractory carcinoid crisis. Eighteen patients experienced postoperative complications: intra-abdominal hemorrhage requiring transfusion (n = 1) or reoperation (n = 2), infections (urinary tract [n = 21, line sepsis ]n = 21, empyema [n = 1J, unknown origin [n = 1]), cardiac ar38

4

I

I

I

1

3

83

Figure 2. Survival of patients after hep atic resection of metastatic neuroendocrine carcinoma: curative versus pallia tive intent.

rhythmias (n = 3), biliary leaks (n = 2), viral hepatitis (n = l), portal vein thrombosis (n = I), brachial plexopathy (n = l), and anorexia (n = 1). The overall survival at 4 years was 73% (Figure 1). Median survival duration has not been reached. Mean follow-up has been 2.2 years (range 3 months to 5.8 years). Sixty patients were alive at the time of their most recent follow-up examination. Tumor progression was evident in each of the 12 patients who died after initial hospitalization. Deaths were attributed to bile duct obstruction (n = 2), fulminant carcinoid syndrome (n = I), intracranial hemorrhage from brain metastases (n = 1), perforated viscus (n = l), liver failure (n = 1), hypercalcemia (n = 1), and unknown (n = 5). Survival among patients who had resection with curative intent versus those with palliative intent did not differ significantly (Figure 2). Twenty-eight patients were classified as resected with curative intent after having all gross disease removed. Concurrent nonhepatic resections in these patients included partial small-bowel resection (n = 7) partial pancreatectomy with or without splenectomy (n JANUARY

1995

NEUROENDOCRINE

100

so

CARCINOWQUE

ET AL

j---u,:;s ------------------------------L

% 60 Log rank P = 0.72

i.2

Years following resection 20 13

13 6

6 3 WWIIU

0

I

I

I

I

1

0

1

2

3

4

5

4

Years following resection P8lt6nt66lrkk

57

31

13

= 5), and partial gastrectomy (n = 1). No postoperative deaths occurred in this group. Twenty-three were initially symptomatic with endocrinopathies (n = 14), mechanical causes (n = 7), combined hormonal and mechanical causes (n = l), or anemia (n = 1). Five were asymptomatic. All those who were symptomatic experienced a complete clinical response. Twenty-three of the 28 patients who received resection with curative intent had abnormal hormonal markers prior to surgery. Tumors from 16 were assayed for hormonal tumor markers to assess biochemical response. Response was complete in 10 and partial in 6. Two patients with both complete symptomatic and hormonal responses have had tumor progression. All patients with a partial hormonal response have had tumor progression. Survival of patients with a complete biochemical response was not compared to survival of patients with a partial response because of small sample size. Fourteen of me 23 patients who had symptoms initially and underwent potentially curative resection have remained asymptomaticof them disease free and 6 with disease progression-after a mean follow-up of 24 months THE AMERICAN

Figure 3. Survival of patients after hep atic resection of metastatic neuroen docrine carcinoma: grade 1 versus grade 2-3.

Figure 4. Probability of recurrence of symptoms in preoperatively symptomatic patients after hepatic resection of metastatic neuroendocrinecarcinomas. Excluded from this analysis: 6 patients asymptomatic preoperatively (1 with ane mia), 6 patients with no response, and 5 patients with less than 30 days of followup.

(range 0.2 to 70.1). SIeven of the 23 have developed recurrent symptoms after a mean interval of 20.4 months (range 8.3 to 58.2) and have disease progression. Two patients have died with disease at 40 and 61 months postoperatively. Of the 5 patients who were asymptomatic preoperatively and classified as potentially curable, 2 are alive and disease free, 2 are alive with disease progression, and 1 died at 13.3 months postoperatively. Forty-six patients (62%) underwent resections with palliative intent. Concurrent nonhepatic resections included partial small-bowel resection (n = 15), partial pancreatectomy with or without splenectomy (n = 5), ileocolostomy (n = 4), and partial gastrectomy (n = 1). Both postoperative deaths occurred in this group. All 46 were symptomatic with endocrinopathies (n = 38), mechanical causes (n = 4), or combined mechanical and hormonal causes (n = 4). Excluding the 2 patients who died postoperatively, symptoms resolved completely in 38 and remained unchanged in 6. Twenty-six have developed recurrent symptoms after a mean interval of 11.3 months (range 2.2 to 27.6). Forty-one patients treated palliatively had biochemical tumor markers available for analysis of hormonal response. JOURNAL

OF SURGERY@

VOLUME

169

JANUARY

1995

39

NEUROENDOCRINE

CARCINOWQUE .._ .-...-.

ET AL

___.. .__.._

in patients who are unresponsive to nonsurgical treatment. Although few studies have focused on the fate of patients with metastatic gastrointestinal neuroendocrine tumors, survival of patients with such tumors is prolonged compared to many other intestinal or pancreatic carcinomas. An overall 5-year survival of 30% to 40% and a median survival of 3 to 4 years has been reported for untreated or for unresponsive patients with hepatic metastasesfrom either carcinoid or islet-cell carcinoma.‘9,“0 Despite the limitations of comparing survival to such historical controls, the near doubling of survival associated with hepatic resection for similar disease stage is at least noteworthy. Although differences in survival could be related to various pretreatment characteristics, stage of tumor is an unlikely factor because all patients had metastatic liver disease. COMMENTS The symptomatic response of our patients from hepatic Our findings provide further evidence that hepatic resection of metastatic neuroendocrine tumors in selected pa- resection of neuroendocrine metastaseshas been gratify90%--complete. tients is safe and can provide temporary relief of symptoms. ing. It has been prompt and-in We have found that initial symptomatic response is simi- Resection has resulted in discontinuation of long-acting solar in patients with metastatic disease resected either for cu- matostatin analogue injections. continuous glucose infurative or palliative intent. As expected, recrudescence of sions, and nutritional supplements. The patients themsymptoms occurs earlier in patients with resections of pal- selves have attested to an improvement in the quality of liative intent, but survival of patients resected either for cu- life. Prompt and complete resolution of clinical enrative or palliative intent has not differed. Our data support docrinopathies, regardless of type, has been reported preother reports that have shown an improved quality of life viously. ‘-I l However, duration of clinical response has varthrough relief of endocrinopathies.“-x Clinical criteria for ied. Our data suggest the durability of the clinical response selection of patients--especially of patients for resection is related directly to the extent of resected tumor and inwith palliative intent-and comparison of hepatic resection versely to the extent of residual disease. Patients with complete resection of all gross disease and normalization of to other therapeutic alternatives. warrants further study. The operative risk of hepatic resection for metastatic neu- hormonal markers have fared better than those with gross roendocrine tumors has not been prohibitive. Our periop- residual disease or reduction (but not normalization) of erative mortality and morbidity were not excessive, espe- hormonal markers. Although complete response was more cially given the fact that more than half of our patients had prolonged when hepatic resection was performed for cure major or extended hepatic resections and concurrent in- rather than palliation, survival to date has been similar retestinal and pancreatic resections. This perioperative risk gardless of the intent of resect. Consequently, if preoperdoes not differ from that of hepatic resection for other ma- ative imaging suggests that resection of all gross disease lignancies’4m’7or after resection of primary intestinal car- is feasible, it should be undertaken. cinoids or islet-cell carcinomas.?-” We fully acknowledge The major clinical challenge to surgeons managing pathat patient selection bias may account for our relatively tients with hepatic neuroendocrine metastasesis selection low morbidity and mortality after hepatic resection of of patients for palliative resection. Given the brief durametastatic neuroendocrine tumors. However, selection bias tion of response in some of our patients, more accurate alone cannot account for the overall low operative risk of imaging of the extent of metastatic disease is needed to surgery in these patients. Indeed, operative risk would be avoid resections which have limited value. We admit that increased in these patients because of the extent of their our failure to palliate 7 of our patients was due to inademetastatic disease and their reduced clinical performance quate resection because preoperative imaging underestistatus due to the hemodynamic and metabolic effects of mated disease extent. Finally, the extent of resection that carcinoid and islet-cell carcinomas. Carcinoid patients in is planned preoperatively must be achieved intraoperaparticular pose added operative risk from carcinoid heart tively. Technically simple resections of the primary tumor disease and the potential for an intraoperative carcinoid and peripheral hepatic metastasesshould be accomplished crisis. Although we acknowledge the critical importance first so that the operation can be promptly terminated if of intraoperative anesthetic management in treatment of major hepatic resection or debulking should result in a these patients, analysis of its impact on morbidity and mor- compromised patient. We recommend liberal use of heptality cannot be determined by this study. atic vascular isolation techniques to reduce intraoperative The overall survival of our patients was 73% at 4 years, hemorrhage from the enlarged, hypervascular livers with and median survival has not been reached. These findings neuroendocrine metastases. are similar to those of others who have shown a mean surAlternatives to resection for patients with dominant hepvival of nearly 6 years after hepatic resection of metastatic atic metastasesfrom neuroendocrine malignancies have inneuroendocrine carcinomas.l-‘O.‘x These findings support cluded systemic chemotherapy, hepatic artery occlusion or the premise that hepatic resection of neuroendocrine metas- embolization (alone or in combination with systemic tases prolongs survival beyond what is expected in un- chemotherapy), and orthotopic liver transplantation. Retreated patients with hepatic neuroendocrine metastasesor gression rates after systemic chemotherapy for metastatic Two had normal markers preoperatively. Of the remainder, hormonal response was complete in 6, partial in 27, and unchanged in 6. At the time of last follow-up, 35 patients are alive with disease progression, 24 of whom have recurrent symptoms. Eleven patients have died with disease-including the two postoperative deaths-after a median interval of 23.6 months postoperatively. The relationship of intent of resection, tumor grade, and tumor type to survival was assessedby univariate analysis. None of these factors was associated significantly with survival (Figures 2 and 3). The overall probability of symptomatic recurrence among preoperatively symptomatic patients is shown in Figure 4.

40

THE AMERICAN

JOURNAL OF SURGERY”

VOLUME

169 JANUARY

1995

~--_

carcinoid tumors seldom exceed 33%, and duration of response is generally brief. I9 In contrast, in a recent randomized trial, combined doxorubicin and streptozocin increased regression rates to 69% and significantly improved survival for patients with metastatic islet-cell carcinoma.2’ Although the median duration of tumor regression and improvement in endocrine symptoms were 18 months in that study, toxicity was considerable. Octreotide (Sandostatin, Sandoz Pharmaceuticals, East Hanover, New Jersey) has markedly decreased hormonal levels and substantially or completely relieved symptoms in nearly 80% of patients with these tumors.22*2”Long-term somatostatin analogue treatment is very expensive, however, and objective tumor regression is infrequent. Most patients become partially resistant to somatostatin analogue therapy after a median interval of 12 months for those with carcinoids and 3 months for those with islet-cell carcinomas. The limitations of systemic chemotherapy have prompted our evaluation of hepatic ischemia alone or in combination with systemic chemotherapy to slow tumor growth rate and arterial dependency of hepatic metastases in 123 patients with predominantly hepatic neuroendocrine metastases.24We observed objective regression in 60% of patients treated with hepatic arterial occlusion alone and 80% of patients with arterial occlusion and chemotherapy. The median duration of substantial or complete relief from endocrinopathies was 4 months for occlusion alone and 18 months for combined therapy. Mortality associated with hepatic ischemic therapy was 4%. The median survival after hepatic arterial occlusion and sequential chemotherapy was 49 months for patients with carcinoid and 35 months for those with islet-cell carcinomas. Repeated temporary hepatic ischemia using an implantable and inflatable arterial occluder has provided a similar immediate response and this approach may improve the durability of the ischemic response.” Finally, orthotopic liver transplantation has been employed for a small subgroup of symptomatic patients who had completely controlled primary and regional disease and no extrahepatic metastases. Of 15 reported patients, IO had endocrinopathies. 26-31 A complete clinical response followed hepatic transplantation in all symptomatic patients. There were 1 operative and 5 late deaths, with a survival range of 5 to 38 months. Two clinical recurrences and 2 biochemical recurrences without clinical disease have been reported in a limited follow-up interval of 29 months. These data suggest that orthotopic liver transplantation will benefit few patients clinically. Our approach to the management of patients with hepatic metastases from neuroendocrine tumors has evolved over the last decade. Application of our recommendations to the individual patient must make allowance for clinical performance status, nutritional impairment, and comorbid conditions (especially cardiac dysfunction in patients with metastatic carcinoid disease). Regardless of tumor origin, if all gross primary and metastatic neuroendocrine tumors are resectable, the treatment of choice should be resection. Conversely, if neither the primary, regional, nor metastatic disease is resectable, systemic therapy offers the best chance of palliation and resection is contraindicated. Treatment strategy is more complex for those patients

NEUROENDOCRPIE

CARCINOMAIQUE

ET AL

with resected or resectable primary and regional disease and metastases confined to the liver. In such patients we recommend cytoreductive hepatic surgery if careful preoperative imaging confirms that the primary and regional diseases are controlled and 90% or more of all hepatic metastases can probably be removed. Hepatic resection should not be undertaken in patients who have severe heart failure from carcinoid heart disease or massive hepatomegaly from diffuse replacement. We believe that resection should precede combined hepatic arterial occlusion and systemic chemotherapy. Resection has been demonstrated to provide a substantial response rate, and combined arterial occlusion and systemic chemotherapy can be reserved for the expected recrudescence of symptoms with disease progression. Whether or not an initial approach using hepatic arterial occlusion and systemic chemotherapy would increase resectability by reducing tumor size or alternatively reduce resectability by impairing liver function and regenerative capacity is unknown. If resection is the initial approach, the potential associated morbidity and mortality of hepatic arterial occlusion and systemic chemotherapy are deferred. Orthotopic liver transplantation should be considered only for individuals in whom less than 90% of the hepatic metastasesare resectable by partial hepatectomy, who have complete gross control of their primary and regional tumors and no other distant metastases,and who are otherwise candidates for transplantation. Individuals whose cancer would qualify them for transplantation but who are not candidates for other reasons should undergo resection of the primary and regional disease and have subsequent arterial occlusion and systemic chemotherapy. Finally, for patients with unresectable primary and regional disease and resectable hepatic me&stases, systemic chemotherapy probably offers the best chance of palliation. In this group, surgical therapy is relegated to palliation of intestinal obstruction, hemorrhage, or secondary infection of the primary tumor if such problems develop.

ACKNOWLEDGMENT The authors wish to thank Duane M. Ilstrup for statistical support and Charles G. Moertel, MD, for valuable guidance.

REFERENCES I. Stephen JL, Grahame-Smith DG. Treatment of the carcinoid syndrome by local removal of hepatic metastases.Proc K SOC Med. 1972;65:444-445. 2. Strode1WE, %lpOS G, Eckhauser E. Thompson N. Surgical tierapy for small-bowel carcinoid tumors. Arch Swg 1983;118:391-397. 3. Norton JA, Sugarbaker PH, Doppman JL, et al. Aggressive resection of metastatic disease in selectedpatients with malignant gastrinoma. Ann Surg. 1986;203:352-359. 4. McEntee GP, Nagomey DM, Kvols LK, et al. Cytoreductive hepatic surgery for neuroendocrine tumors. Surgery. 1990:I OS: IOYI-1096. 5. Xdakridis C, ijberg K, Juhlin C, et al. Surgical treatment of midgut carcinoid tumors. World J Surg. 1990;14:377-385. 6. Soreide 0, Berstad T, Bakka A, et al. Surgical treatment as a principle in patients with advanced abdominal carcinoid tumors. Surgery. 1992;1I 1~48-54.

THE AMERICAN JOURNAL OF SURGERY” VOLUME 169 JANUARY 1995 41

NEUROENDOCRINE

CARCINOWQUEET AL

7. Grama D, Eriksson B, Martensson H, et al. Clinical characteristics, treatment and survival in patients with pancreatic tumors causing hormonal syndromes. World J Surg. 1992;16:632-639. 8. Ahlman H, Wangberg B, Jansson S, et al. Management of disseminated midgut carcinoid tumours. Digestion. 1991;49:78-96. 9. Goto H, Yamaji Y, Konno T, et al. A glucagon-secreting alphacell carcinoma of the pancreas. WorZdJ Surg. 1982;6:107-109. 10. Nagomey DM, Bloom SR, Polak JM, Blumgart LH. Resolution of recurrent Vemer-Morrison syndrome by resection of metastatic vipoma. Surgery. 1983;93:348-353. 11. Travis WD, Linnoila RI, Tsokos MG, et al. Neuroendocrine tumors of the lung with proposed criteria for large-cell neuroendocrine carcinoma. Am J Surg Putfwl. 1991;15:529-553. 12. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. /Am Stat Assoc. 1958;53:457-481. 13. Peto R, Peto J. Asymptotically efficient rank invariant test procedures. /Roy S&t Sot. 1972;135:185-220. 14. Foster JH. Survival after liver resection for cancer. Cancer. 1970;26:493-502. 15. Iwatsuki S, Shaw BW, Starzl TE. Experience with 150 liver resections.Ann Surg. 1983;197:247-253. 16. Thompson HH, Tompkins RK, Longmire WP, Jr. Major hepatic resection. A 25-year experience. Ann Surg. 1983;197:375-387. 17. Wolf RF, Goodnight JE, Krag DE, Schneider PD. Results of resection and proposed guidelines for patient selection in instances of noncolorectal hepatic metastases.Surg Gynecol Obstef. 1991;173: 454-460. 18. Nagomey DM, Que, FG. Cytoreductive hepatic surgery for metastatic gastrointestinal neuroendocrine tumors: alternative review. Adv Gastrointest Res. In press. 19. Moertel CG. An odyssey in the land of small tumors. J Clin Onrol. 1987;5:1503-1522. 20. Thompson GB, van Heerden JA, Grant CS. et al. Islet cell carcinomas of the pancreas: a twenty year experience. Surgery. 1988; 104:1011-1017. 21. Moertel CC, Lefkopoulo M, Lipsitz S, et al. Streptozocin-doxorubicin, streptozocin-fluorouracil, or chlorozotocin in the treatment of advanced islet-cell carcinoma. NEJM. 1992;326:519-523, 22. Kvols LK, Moertel CG, O’Connell MJ, et al. Treatment of the malignant carcinoid syndrome: evaluation of a long-acting somatostatin. NEJM. 1986;315:663-666. 23. Kvols LK, Buck M, Moertel CG, et al. Treatment of metastatic islet cell carcinomas with a somatostatin analogue (SMS 201-995). Ann Intern Med. 1987;107:162-168. 24. Moertel CG, Johnson CM, McKusick MA, et al. The management of patients with advanced carcinoid tumors and islet cell carcinomas. Ann Intern Med. 1994;120:302-309. 25. PerssonBG. Nobin A, Ahren B, et al. Repeated hepatic ischemia as a treatment for carcinoid liver metastases. World I Surg. 1989;13:307-312. 26. Makowka L, Tzakis AG, Mazzaferro V, et al. Transplantation of the liver for metastatic endocrine tumors of the intestine and pancreas. Surg Gynecol O&et. 1989;168:107-l 11. 27. Arnold JC, O’Grady JG, Bird GL, et al. Liver transplantation for primary and secondary hepatic apudomas. Br J Surg. 1989;76: 248-249. 28. Alsina AE, Bartus S, Hull D, et al. Liver transplant for metastatic neuroendocrine tumor. J Clin Gastroenterol. 1990;12:533-537. 29. Lobe TE, Vera SR, Bowman LC, et al. Hepaticopancreaticogastroduodenectomywith transplantation for metastatic islet cell carcinoma in childhood. J Pediatr Surg. 1992;27:227-229. 30. Gulanikar AC, Kotylak G, Bitter-Suermann H. Does immuncsuppression alter the growth of me&static liver carcinoid after orthotopic liver transplantation? Transplant Proc. 1991;23:2197-2198. 31. Farmer DG, Shaked A, Colonna JO, et al, Radical resection combined with liver transplantation for foregut tumors. Am J &rg. 1993;59:80~812. 42

DISCUSSJON B. Langer, MD (Toronto, Ontario, Canada): This is a very large experiencewith an uncommondisease,and the early operative results certainly are quite outstanding.I’d just take the editorial prerogative to disagree with your comment that liver resection is a safeprocedure.I would say that liver resection is a dangerousprocedurethat can be carried out safely in highly skilled and experienced hands such as Dr. Nagomey and his colleagues. But I would not want any of us to go away thinking that liver resectionis simple. W.C. Meyers, MD (Durham, North Carolina): Do you have any evidenceat all to suggestthat thesepatientslived longer than they would have?For example,did severity of the syndromeaffect the outcome? If you cannot cure patientswho have m&static disease of the liver, then the goal is palliation. Do you have any quality-of-life data, such as time-tradeoff analysis to justify doing the operation? What were your criteria for operating on thesepatients? Was it the presenceof the syndrome?Was it refractorinessto treatment?What is your routine in terms of trying to treat thesepatients medically before operation? We do a numberof plucking operations,wherewe reach deep within the liver with our index finger and pluck out the lesion. Is that the sort of operationyou’re talking about in many of thesepatients?Could you commenton the technical aspects? Dr. Que: Dr. Moe-l’s early study showedthat the natural history of the diseaseresultedin a median duration of survival of 8.1 yearsfollowing the onsetof symptoms.No long-term data are available to determinethe median duration of survival with metastaticdisease.We do not have any quality-of-life data. The procedureof enucleation of the tumors dependson the patient and where the tumors are located. Whenever possiblewe try to do anatomicresections.Many of the patients had multiple tumors in various parts of the liver and anatomicresectionsweren’t possiblewithout resectingtoo much liver. In such caseswe pluck out the tumors. P.B. Lansing, MD (New Orleans, Louisiana): Five patients who were asymptomaticwere resected.Since these are generally slow-growing tumors, at what stageof the diseasewould you recommendresection for an asymptomatic patient? Dr. Que: If the patient has localized diseasein the liver and appearscurable, we feel comfortable resecting these tumors. In general,hepatic resectionis to be reservedfor patients who are symptomatic. R.A. Prim, MD (Chicago, Illinois): It seemsthat your patientswith metastaticneuroendocrinecarcinomafall into two groups.About half had a concomitantresection,while the remainderhad a subsequentor stagedresectionof their disease.Wasthe extent of diseasethe samein both groups or, in other words, was the type and magnitudeof the operative proceduresremoving the metastaticdiseasesimilar in each group? Finally, was there a difference in survival betweenconcomitant and stagedresection? Dr. Que: The patientswho had synchronoustumors underwentresectionof their primary tumors at the sametime as their liver tumors. I did not stratify the data to see if

THE AMERICAN JOURNAL OF SURGERY” VOLUME 169 JANUARY 1995

NEUROENDOCRINE

therewasa differencein survival rateswith respectto those patients as comparedto the overall group. Dr. Langer: Since Dr. Moertel has a median survival of 8.1 years with nonresectedtherapy, why is he now referring patients for resection? Dr. Que: Many of the patients developed debilitating symptoms,and it is the debilitating symptomswe aretreating. We are currently investigating a protocol for hepatic artery occlusion for patients who are not candidatesfor liver resection. Sandostatinisn’t really effectivefor long periodsof time in thesepatientsbecauseof the ameliorationrate.We will often resector treat with hepaticartery chemotherapyinfusion. Dr. Langer: Is there a role for transplantation in this disease? Dr. Que: Transplantationhasbeenusedby other centers for the treatmentof patientswith this disease.At this time,

THE AMERICAN

CARCINOMAIQUE

ET AL

we feel that it’s indicated only for patients in whom the primary tumor hasbeencompletely resected,the local and regional diseaseis under completecontrol, and surgicalresection would not be applicable. A. Sicular, MD (New York, New York): Have you used cryotherapy in any of thesepatients? Dr. Que: No, we have not. D. Pertsemlidis, MD (New York, New York): In your palliative resections,do you have any neuroendocrinepeptides to show that there was a decline in the plasma levels and that your palliative resectionwas indeedeffective? Dr. Que: Yes. Dr. Pertsemlidis: I am talking about the pancreatic polypeptide and other peptides. Dr. Que: Yes, but I did not presentthat data.With a decreasein the symptoms or a complete symptomatic response,there was a decreasein the hormonal levels.

JOURNAL OF SURGERY@

VOLUME

169 JANUARY

1995

43