First experience and technical aspects of isolated liver perfusion for extensive liver metastasis

First experience and technical aspects of isolated liver perfusion for extensive liver metastasis Karl J. Oldhafer, MD, Hauke Lang, MD, Markus Frerker, MS, Laura Moreno, MD, Ajay Chavan, MD, Peer Flemming, MD, Silvio Nadalin, MD, Ekkehard Schmoll, MD, and Rudolf Pichlmayr, MD, FRCS (Hon), FACS (Hon), Hannover, Germany

Background. New drugs and modalities for locoregional tumor treatment in recent years may offer new potential for isolated liver perfusion in patients with nonresectable liver tumors. The purpose of this study was to prove the feasibility of arterial isolated liver perfusion and to assess the tolerance of perfusion with high-dose tumor necrosis factor (TNF). Methods. Twelve patients with extensive liver metastases previously treated unsuccessfully with systemic chemotherapy underwent isolated hyperthermic liver perfusion using a heart-lung machine. High doses of mitomycin were administered in the first six and a combination of TNF and melphalan in the last six patients. Results. No operative death occurred and no direct postoperative liver failure was observed in any patient. In cases of variations of the arterial hepatic blood supply, the perfusion was done through the splenic artery or an angiography catheter. Histologic analysis of tumor biopsy specimens obtained on the first postoperative day revealed major tumor necrosis in 8 of 12 patients. Conclusions. Isolated arterial perfusion of the liver is a complex surgical procedure that is feasible in patients with anatomic variations of the hepatic artery. The remarkable histologic response to perfusion in several pretreated patients, especially after application of high-dose TNF and melphalan, suggests that this modality is very effective in tumor killing. (Surgery 1998;123:622-31.) From the Departments of Abdominal and Transplantation Surgery, Radiology, and Hematology and Oncology, and the Institute of Pathology, Hannover Medical School, Hannover, Germany

NEOPLASMS OF THE LIVER CAUSE a serious health problem worldwide.1 Because surgical resection offers a significant potential for cure, it has remained the treatment of choice.2,3 However, resection is often not feasible because of intrahepatic and extrahepatic tumor spread. Patients excluded from surgery may be treated with systemic chemotherapy, but tumor response is often poor and systemic toxicity severe.4,5 Thus, because of little long-term benefit and only small degrees of palliation, alternative therapies are required. Various regional approaches have been developed.6-11 Their aim is to restrict chemotherapy to the tumor-bearing organ (i.e., liver). Such restricted applications allow dosages higher than those achieved by systemic application. Response rate Accepted for publication Nov. 5, 1997. Reprint requests: Karl J. Oldhafer, MD, Klinik für Abdominalund Transplantationschirurgie, Carl-Neuberg-Strasse 1, 30625 Hannover, Germany. Copyright © 1998 by Mosby, Inc. 0039-6060/98/$5.00 + 0 11/56/87441

622 SURGERY

after regional chemoperfusion for metastatic liver tumors is increased compared with that after systemic administration.12,13 However, there is no convincing evidence that survival is prolonged.14 This may be related to the antitumor agent chosen and its concentration. Isolated organ perfusion systems are based on a concept of regional cancer chemotherapy first described by Klopp et al.15 and further developed by Creech et al.16 Perfusion techniques have been developed for limbs, mid-gut, and liver effectively separating the organ perfusion from systemic circulation.16 Limb perfusions in particular have already been successfully established in clinical use for melanoma17,18 and locally advanced soft tissue sarcomas.19 These good results in limb perfusions indicate that a similar procedure may be applicable for the liver. The liver can be perfused through three routes: hepatic artery, portal vein, or both. Tumor cells seem to enter the liver mainly through the portal vein, particularly in primary tumors of the gastrointestinal tract. Therefore the portal vein

Oldhafer et al. 623

Surgery Volume 123, Number 6

Fig. 1. Schematic diagram of the liver perfusion system: (1) venous reservoir, (2) roller pump, (3) oxygenator, (4) heater, and (5) centrifugal pump.

Table I. Characteristics of patients, histologic and radiologic tumor response, hospital stay, and outcome

Patient

Age/ Gender

Primary tumor

Tumor necrosis 1 postoperative day (%)

Hepatic tumor response*

Hospital stay (days)

1

67/M

Uveal melanoma

10

NC

30

2

41/F

Colon cancer

70

PD

12

3 4 5 6 7 8 9 10 11 12

49/M 38/F 47/F 38/F 53/F 38/F 40/M 48/F 39/M 35/M

Colon cancer 50 Breast cancer 5 Colon cancer No material Breast cancer 50 Breast cancer 50 Colon cancer 80 Colon cancer 90 Colon cancer 10 Small bowel carcinoid 80 Uveal melanoma 50

PR PD PD PR† PR NC PR No data CR NC

14 21 15 60 20 13 25 22 99 27

Outcome Dead (30 days), myocardial infarction, ascites Dead (20 mo), liver metastasis, lung metastasis, ascites Alive (27 mo) Dead (4 mo), VOD, ascites Dead (14 mo), liver metastases Dead (60 days), VOD, ascites Alive (13 mo) Alive (12 mo) Alive (12 mo) Alive (11 mo) Alive (7 mo) Alive (6 mo)

* Assessment of tumor response by computed tomography: CR, complete response: disappearance of all measurable disease in the liver for > 4 weeks; PR, partial remission: regression of the tumor size by > 50% for > 4 weeks; NC, no change: regression < 50% of the tumor in the liver or progression < 25% for > 4 weeks; PD, progressive disease: progression > 25% of the tumor. †Postmortem

examination revealed complete necrosis of the multiple hepatic metastases.

should be selected for perfusion. However, several groups have shown that hepatic neoplasms are mainly supplied by the hepatic artery.20,21 Therefore, liver perfusion through the hepatic artery might be a better route to reach more tumor cells. Typically, the gastroduodenal artery is cannulated for isolated arterial perfusion, whereas the proximal common hepatic artery is occluded. However, anatomic variations in the hepatic arterial blood supply are common and might complicate the perfusion or even exclude patients from this approach.22 Although the first clinical liver perfu-

sions were performed more than 30 years ago, they have been performed only in specialized centers.23 Isolated liver perfusion has been described as a technically difficult procedure that requires an extracorporeal perfusion technique including use of a heart-lung machine, which greatly increases the cost for a single treatment. Consequently, widespread use of this regional antitumor therapy has not been possible. New drugs (e.g., TNF) and new strategies for locoregional tumor treatment (e.g., gene transfer technology) may offer new potential for this technique.

624 Oldhafer et al.

Surgery June 1998

Table II. Operating times, intraoperative transfusion requirements, and intraoperative chemotherapy Patient 1 2 3 4 5 6 7 8 9 10 11 12

Operating time (min) 450 410 425 440 440 395 350 410 420 435 413 375

Time of portal vein occlusion (min)* 102 114 89 110 107 120 120 116 140 136 115 114

Fresh frozen plasma (ml)

Packed red cells (ml)†

1750 1500 1250 1000 1000 1500 0 1500 1500 1750 1500 2000

1500 2250 500 1500 1750 1750 500 750 500 1750 1500 1250

Intraoperative chemotherapy 50 mg mitomycin C 30 mg mitomycin C 20 mg mitomycin C 30 mg mitomycin C 35 mg mitomycin C 35 mg mitomycin C 200 µg TNF + 80 mg melphalan 200 µg TNF + 60 mg melphalan 200 µg TNF + 70 mg melphalan 300 µg TNF + 75 mg melphalan 300 µg TNF + 120 mg melphalan 300 µg TNF + 140 mg melphalan

*Time between occlusion of portal vein for bypass cannulation and portal venous reperfusion after reconstruction. †Including

the packed red cells for preloading of the heart-lung machine.

In 1995 a program of isolated liver perfusion was initiated at Hannover Medical School Hospital. This article describes our first clinical experiences with isolated liver perfusions. PATIENTS AND METHODS Patients. From February 1995 to March 1997, 12 patients with unresectable bilateral hepatic metastases of different primary tumors were selected for isolated liver perfusion. Patient characteristics are listed in Table I. In all patients, the primary tumor had been previously resected. Most patients showed progressive metastatic disease despite various chemotherapy protocols. Extrahepatic tumor growth and recurrence of primary tumor were exclusion criteria for isolated liver perfusion. Angiography of hepatic arteries was performed to assess the technical feasibility of arterial cannulation for liver perfusion. This study was approved by the Medical Ethics Committee of the Hannover Medical School. Surgical technique. First extrahepatic tumor spread of the liver was excluded by surgical exploration of intraabdominal organs. Next the liver was completely mobilized similar to the hepatectomy procedure in liver transplantation or in ex situ liver resection. The arterial system of the liver was dissected according to each patient´s anatomy. The portal vein and infrahepatic vena cava were exposed. Suprarenal veins were identified, ligated, and divided. The retrohepatic vena cava was isolated from the posterior wall, and complete control of the vena cava was obtained. Phrenic veins were ligated when necessary for clamping of the vena cava without compromising hepatic venous drainage. In all patients a cholecystectomy was performed to avoid the possible complication of therapy-related

cholecystitis. Femoral and axillary veins were prepared for veno-venous bypass. Pediatric aortic cannulas (DLP, INC., Grand Rapids, Mich.) were used for arterial cannulation. Thereafter, the venovenous bypass connecting the portal and femoral veins with the axillary vein was established using a centrifugal pump (Bio pump; Medtronic-biomedicus Inc., Eden Prairie, Minn.). This allowed venous decompression during liver perfusion (Fig. 1). Finally, the retrohepatic caval vein was cannulated and connected with the heart-lung machine. After clamping of the infrahepatic and suprahepatic caval veins, the extracorporeal circuit was started (Fig. 1). The extracorporeal circuit consisted of a hollow fiber oxygenator with integrated heat exchanger (Minimax; Medtronic Cardiopulmonary, Anaheim, Calif.) and a Computer-Aided-PerfusionSystem with Roller Pump and Heater (CAPS; Stöckert Instrumente GmbH, München, Germany) and was preloaded with 500 ml saline and 250 ml of packed red cells. Perfusion flows were adjusted between 500 and 700 ml/min to maintain perfusion pressure below 160 mm Hg. Temperature monitoring probes were inserted into the liver. Intrahepatic temperature was elevated to 40.0° to 41.0° C by hyperthermic perfusion. Perfusion time was limited to 60 minutes followed by rinsing with 1000 ml of saline through the artery and 500 ml of saline through the portal vein. Cannulas and clamps were removed. Portal and caval vein incisions were closed with running sutures. The abdominal cavity was temporarily closed and a second look operation was performed the next day. Chemotherapy. Mitomycin (Mitomycin medac; Medac mbH, Hamburg, Germany) was used in the first six patients to use a familiar chemotherapeutic drug while the surgical technique was verified.

Surgery Volume 123, Number 6

Oldhafer et al. 625

Fig. 2. Anatomy of hepatic blood supply and technique of arterial cannulation in patient 2. CT, celiac trunk; GDA, gastroduodenal artery; LGA, left gastric artery; LHA, left hepatic artery; PV, portal vein; RHA, right hepatic artery; SA, splenic artery.

Fig. 3. Anatomy of hepatic blood supply and technique of arterial cannulation in patient 3. CT, celiac trunk; CHA, common hepatic artery; GDA, gastroduodenal artery; LGA, left gastric artery; LHA, left hepatic artery; PV, portal vein; RHA, right hepatic artery; SMA, superior mesenteric artery.

Fig. 4. Anatomy of hepatic blood supply and technique of arterial cannulation in patient 4. CT, celiac trunk; GDA, gastroduodenal artery; LGA, left gastric artery; LHA, left hepatic artery; PV, portal vein; RHA, right hepatic artery; SA, splenic artery.

Mitomycin was administered as bolus into the perfusion fluid at the beginning of the perfusion. Dosages are shown in Table I. After verification of the surgical technique recombinant human tumor

necrosis factor–α (r-hTNF-α) was given in combination with melphalan (Alkeran; Glaxo Wellcome GmbH, Hamburg, Germany) in the last six patients. Whereas r-hTNF-α was administered at the beginning of the perfusion, melphalan was

626 Oldhafer et al.

Fig. 5. Angiogram of the celiac trunk with the catheter tip positioned in the common hepatic artery just proximal to the origin of the gastroduodenal artery (arrow) in which the stump only is visualized because of retrograde flow.

added after 20 minutes. The r-hTNF-α concentrations were determined in the perfusion and in the systemic circulation by means of a radioimmunoassay. Postoperative care. Postoperative liver function was monitored by liver enzymes, bilirubin, and clotting factors. Intravenous heparin was given for the first 7 postoperative days to keep the partial thromboplastin time between 50 and 60 seconds. Liver blood flow was monitored by Doppler sonography during the first postoperative days. Because the majority of patients were referred to us from other institutions, standardized postoperative imaging studies were not possible in all cases. Assessment of tumor response. Clinical responses were defined by computed tomography. Radiologic findings were differentiated as follows: complete response as disappearance of all measurable disease in the liver for more than 4 weeks, partial remission as regression of the tumor size by >50% for more than 4 weeks, no change as regression < 50% of the tumor in the liver or progression < 25% for more than 4 weeks, progressive disease as progression > 25% of the tumor. Histology. Tumor biopsy specimens were taken during the second look operation, fixed with formalin, and embedded in paraffin. The percentage of tumor necrosis was assessed microscopically. RESULTS Vascular isolation of the liver was possible in all patients without technical problems. Complexity of the surgical procedure and extracorporeal perfusion technique required operation times ranging from 350 to 450 minutes, with a mean of 413 ± 29

Surgery June 1998

minutes (Table II). Intraoperative transfusion requirements and other operation parameters are summarized in Table II. In four patients, standard procedure with cannulation of the gastroduodenal artery was feasible. The diameter of the aortic cannulas varied between 2.7 and 4.0 mm and allowed a sufficient flow rate between 500 and 700 ml/min. In all, 8 of 12 patients showed variations in their arterial hepatic blood supply. Therefore, the surgical technique of arterial perfusion was adapted to the individual situation in the following manner. Patient 1: The left hepatic artery arose from the left gastric artery, which was occluded during the perfusion by a vascular clamp. Because this was the only left hepatic artery present, a liver resection of segments II and III was performed after perfusion to remove the nonperfused liver segments. Patient 2: Left and right hepatic arteries had a common origin directly from the celiac trunk together with the splenic artery (Fig. 2). The liver was perfused through a cannula in the splenic artery, and the celiac trunk and the gastroduodenal artery were occluded. Because an end-to-end reconstruction was not possible, the splenic artery was ligated after perfusion. This did not lead to splenic infarction. Patient 3: The common hepatic artery was significantly stenosed. Hepatic blood supply was mainly established by the gastroduodenal artery through the superior mesenteric artery (Fig. 3). The common hepatic artery was dilated intraoperatively and cannulated. After liver perfusion the common hepatic artery was ligated and the hepatic blood supply was then exclusively through the gastroduodenal artery. Patient 4: The patient had a massive tumor in the left liver lobe in addition to metastases in the right lobe. The left hepatic artery arose from the left gastric artery. The tumor had caused a thrombosis of the left portal vein. Because the tumor mass was located mainly in the left liver lobe, a left hepatectomy was performed before perfusion. The remaining right liver lobe was perfused 3 weeks later through a cannula in the gastroduodenal artery. Patient 5: The liver had two variations: the left hepatic artery originated from the left gastric artery, and the gastroduodenal artery arose from the celiac trunk (Fig. 4). The liver was perfused through the gastroduodenal artery, whereas the celiac trunk, splenic, and left gastric arteries were occluded. After liver perfusion the liver segments II and III were resected in a manner similar to that performed in patient 1.

Surgery Volume 123, Number 6

Patient 7: A left hepatic artery originated from the left gastric artery. In contrast to patients 1 and 5, this represented an accessory left hepatic artery. The lateral segments were perfused during the therapeutic liver perfusion through intrahepatic collaterals from the common hepatic artery as presumed by the fact that those segments did not change in appearance after clamping the left gastric artery. Thus, a left lateral liver resection was not performed. Patient 9: This patient had received regional chemotherapy before liver perfusion through a port catheter that had been placed into the gastroduodenal artery. After several intraarterial treatments, the gastroduodenal artery was thrombosed and the port catheter was removed. Therefore the splenic artery was used for liver perfusion. Celiac trunk and left gastric artery were occluded during perfusion. The splenic artery was reconstructed after perfusion. Patient 11: The preoperative angiogram showed no antegrade opacification of the gastroduodenal artery from the celiac trunk or from hepatic arteries. Retrograde filling of the gastroduodenal artery was seen through the superior mesenteric artery. The splenic artery originated separately from the aorta. Because of the unavailability of the gastroduodenal artery for cannulation and the separate origin of the splenic artery, it was decided to place an angiographic catheter into the common hepatic artery through the femoral artery before operation (Fig. 5). This was performed in the radiology suite using a Simmons II catheter (2.7 mm external diameter, 2.2 mm inner diameter, 100 cm long; Cordis Corporation, Miami, Fla.). During operation the catheter was identified in the proper hepatic artery by palpation and controlled by a tourniquet. Perfusion was performed through the angiography catheter, which allowed a maximal flow rate of only 250 to 300 ml/min. This low flow rate allowed only an intrahepatic temperature of 36.0° C, although the perfusion fluid was heated to 42.0° C. The washout procedure was unsatisfactory because of perfusion conditions. Thus, severe systemic side effects of TNF with hypotension and pulmonary insufficiency were observed. TNF concentration in the peripheral blood increased to 1466 pg/ml. Despite low perfusion flow the antitumor effect was remarkable. Two weeks after perfusion, an almost complete necrosis of the metastases was found radiologically (Fig. 6). Complete control of the infrahepatic, retrohepatic, and suprahepatic caval veins allowed isolated perfusion with no appreciable leakage as deter-

Oldhafer et al. 627

Fig. 6. Contrast enhanced CT sections before (A), 1 week after (B), and 4 weeks after liver perfusion (C). The hypodense lesions in A (arrows) show marked necrosis 1 week after liver perfusion (B). The corresponding CT section after 4 weeks shows disappearance of the smaller metastases.

mined by measurement of r-hTNF-α in the isolated and systemic circulation. During the isolated liver perfusion no r-hTNF-α (< 0.02 ng/ml) was detectable in the systemic circulation (Fig. 7, A). Fig. 7, B shows r-hTNF-α levels in the isolated circulation. A rapid increase in systemic r-hTNF-α concentrations was observed only in the early

628 Oldhafer et al.

Surgery June 1998

Fig. 7. Concentrations of TNF in the systemic perfusion (A) and in the perfusion medium (B). Diamond, patient 8 (200 µg TNF); triangle, patient 9 (200 µg TNF); box, patient 10 (300 µg TNF); open circle, patient 12 (300 µg TNF).

reperfusion period as a result of incomplete washout of the liver (Fig. 7, A). No operative death occurred and no postoperative liver failure was observed in any patients. After an early increase, serum aspartate aminotransferase levels dropped. On postoperative day 5, concentrations reached levels below 100 U/L. However, after liver perfusion with mitomycin, ascites was seen in four of six patients. Two had a venoocclusive disease (VOD) and died 2 months and 4 months after liver perfusion as a result of VOD. Another patient died within the postoperative period on the thirtieth day as a result of a myocardial infarction. Patients receiving 300 µg rhTNF-α showed signs of hypotension and capillary leak during the first 2 postoperative days. Patient outcome and duration of hospital stay are summarized in Table I. Temporary abdominal closures and second look operations after 24 hours caused no prob-

lems in surgical and intensive care management. In patients with left lateral liver resections, the second look operation even allowed a local packing of the hepatic transection surface. During the second look operation, gross tumor response and the condition of the normal liver could be verified directly. Histologic analysis of the tumor biopsy specimens obtained during the second look operation revealed major tumor necrosis (≥ 50%) in 8 of 12 patients. Histologic results of all patients are summarized in Table 1 and an example is shown in Fig. 8. Postmortem examination of the sixth patient revealed complete necrosis of the multiple hepatic metastases. Computed tomography showed complete remission in one patient after isolated liver perfusion (Table I). Partial remissions were achieved in four patients, with survival up to 24 months. Three patients showed no change and three others had progressive disease. In one patient, radiologic follow-up was not possible

Surgery Volume 123, Number 6

Oldhafer et al. 629

Fig. 8. Liver metastasis of colon carcinoma (patient 9) (A) before perfusion. Adenocarcinoma with fibrous septae; no necrosis. (Original magnification ×140; hematoxylin-eosin stain.) (B) One day after perfusion with 200 µg r-hTNF-α and 70 mg melphalan. Overall rated tumor necrosis, 90%. Vital tumor cells in the periphery (arrow). (Original magnification ×140; hematoxylin-eosin stain.)

because the patient did not return from the referring peripheral hospital. DISCUSSION Hyperthermic perfusion with TNF of the surgically isolated liver represents a novel therapeutic approach to unresectable liver tumors. Progress in liver surgery and transplantation over the years has improved the surgical safety of liver vascular isolation. Standard procedure for isolated arterial liver perfusion was cannulation of the gastroduodenal artery. Anatomic variations of the arterial blood supply, however, required individualized approaches in 8 of 12 patients. Several variations of the celiac trunk and hepatic arteries have been described.22 In cases in which a relevant left hepatic artery originates from the left gastric artery, two approaches are possible: resection of segments II and III after chemoperfusion (patient 1 and 5 of this series) or additional liver perfusion through the portal vein without therapeutic arterial chemoperfusion of the left lateral segments. In cases in which there is an accessory left hepatic artery, the left lateral segments are also perfused through intrahepatic collaterals from the common hepatic artery; therefore these segments can be preserved (patient 7). When there are abnormalities within the origin of the gastroduodenal or common hepatic artery, perfusion through the splenic artery with clamping of the celiac trunk is an alternative24 and was used in two patients in the present series. Another approach to this problem could be the radiologic insertion of a catheter through the femoral artery into the hepatic artery. This combined radiologic and surgical approach could be appropriate for patients with a right hepatic artery originating from the superior

mesenteric artery. Such patients would otherwise be excluded from isolated arterial liver perfusion. The general feasibility of this combined concept has been proved in patient 11 of this study. However, the length and diameter of the catheter caused perfusion problems; maximal flow rate was only 250 to 300 ml/min, maximal intrahepatic temperature was 36° C, and the washout was insufficient. Shorter and larger diameter catheters may help to avoid these problems. However, because the antitumor effect was remarkable in this case, the low perfusion flow might be enough or even helpful. To isolate the liver the portal vein needs to be clamped, thereby compromising venous return of intestinal blood. This state is tolerated in liver resections for up to 1 hour. Whether a veno-venous bypass is necessary in this procedure is unclear. For safety reasons, we chose the bypass. We cannot be sure whether a patient undergoing isolated hyperthermic chemoperfusion will show reduced tolerance to venous congestion and reperfusion damage; therefore the portal vein should be drained during the clamping period. Several internal shunts have been developed for this purpose,25-27 but we found them difficult to handle. The extracorporeal veno-venous bypass between the portal/femoral vein and the axillary vein regularly used in liver transplantation seemed to be superior.28 However, occlusion time of the portal vein was extended because of its cannulation and subsequent reconstruction. It is not yet possible to evaluate the overall therapeutic results of liver perfusions on the basis of length of survival or relief of symptoms in this study because of the small number of patients and followup period of no more than 2 years. On the other

630 Oldhafer et al.

hand, if tumor necrosis, as determined by the histologic effect of perfusion on the tumor, is used as a criterion, then treatment was successful in 8 of 12 cases. In those patients microscopic evidence of significant tumor necrosis (≥50%) was found 24 hours after treatment. In addition, postmortem examinations of patients who died early showed clear hepatic tumor response to perfusion. The short application period and the technical limitation to only one or two perfusion procedures are disadvantages of this approach, particularly because it requires a major surgical procedure. Therefore, efficiency of the treatment must be substantial during the brief exposure to justify such an extensive operative approach. High-dose melphalan and mitomycin are the most frequently used chemotherapeutic drugs in isolated organ perfusion. Because TNF can cause severe side effects and because there was only little experience with TNF at the beginning of our program, we started with mitomycin. Mitomycin represented a well-established chemotherapeutic drug and had proved effective in the treatment of various gastrointestinal malignancies including liver tumors.29 However, our results with mitomycin in liver perfusions were disappointing. Venoocclusive disease and ascites formation were observed in two of six patients. Marinnelli et al.30 have published similar results. Thus mitomycin should not be used in isolated liver perfusions. In contrast the use of TNF should be advantageous because its antitumor effect begins rapidly after treatment.31 High-dose TNF therapies have been used in isolated limb perfusion with good results. Systemic clinical trials of TNF, however, have been disappointing. Clinical experiences with TNF in liver perfusion are rare. Final results regarding tumor response and survival are not yet available in the literature. The first report about TNF use in liver perfusion cited systemic side effects.32 Hypotension and capillary leak were present in this series, especially after 300 µg r-hTNF-α but could be managed by standard intensive care management. Prophylactic application of dexamethasone or indomethacin may reduce these side effects.33,34 Although our TNF trial is of short duration, results appear promising. For example, the first patient treated with TNF after prior unsuccessful systemic chemotherapy showed continued partial remission after 12 months. Hyperthermia is another important aspect of isolated liver perfusion. Hyperthermia is selectively toxic to malignant cells and synergistic with cytotoxic drugs in vitro.35 In vivo, hyperthermia alone or in combination with high-dose regional chemotherapy has caused regression of large primary tumors and

Surgery June 1998 metastases.36,37 In isolated liver perfusion, hyperthermia can be achieved easily by use of a heat exchanger connected to the extracorporeal circuit. In the present series, portal venous chemoperfusion was not performed, but it has been used by others.38,39 Artificial portal venous perfusion must be handled extremely carefully.40 As part of a low pressure system, the vascular architecture of the portal veins could be severely damaged by high perfusion pressures. However, tumor margins are often supplied with portal venous blood. Thus, combined perfusions through the hepatic artery and the portal vein might improve the efficacy from the oncologic point of view. In summary, the technique of isolated arterial liver perfusion used in this series proved to be feasible without technical problems. Individualized approaches allowed perfusion in patients with anatomic variations of hepatic blood supply. Leakage of TNF into the systemic circulation was not observed. Histologic and radiologic tumor response after perfusion with TNF and melphalan in several unsuccessfully pretreated cases suggests that this modality could be effective in tumor killing. Its place within an oncologic model, however, cannot yet be determined. Apart from its single use as palliation, further applications within a combined therapy are conceivable. Its application as induction therapy with subsequent regional chemotherapy through a therapeutic catheter within the hepatic artery may be a new option to improve results of locoregional chemotherapy. Tumor necrosis factor was provided by Hans KierulffNielsen, The Interferon Resesarch Institute, Hjörring, Denmark. REFERENCES 1. Niederhuber JE. Tumors of the liver. In: Murphy GP, Lawrence W, Lenhard RE, editors. Textbook of clinical oncology, ed 2. Atlanta: The American Cancer Society; 1995. 2. Adson MA. Resection of liver metastases: when is it worthwhile? World J Surg 1987;11:511-20. 3. Steele G, Bleday R, Mayer RL, Lindblad A, Petrelli N, Weaver D. A prospective evaluation of hepatic resection for colorectal carcinoma metastases to the liver: gastrointestinal tumor study group protocol 6584. J Clin Oncol 1991;9:1105-12. 4. Lokich JJ, Ahlgren JD, Gullo JJ, Philips JA, Fryer JG. A prospective randomized comparison of continuous infusion fluorouracil with a conventional bolus schedule in metastatic colorectal carcinoma: a Mid-Atlantic Oncology Program study. J Clin Oncol 1989;7:425-32. 5. Moertel CG. Chemotherapy for colorectal cancer. N Engl J Med 1994;330:1136-42. 6. Ravikumar TS, Kane R, Cady B, Jenkins R, Clouse M, Steele

Oldhafer et al. 631

Surgery Volume 123, Number 6

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17. 18.

19.

20.

21.

22.

G Jr. A 5-year study of cryosurgery in the treatment of liver tumors. Arch Surg 1991;126:1520-4. Curley SA, Stone DL, Fuhrman GM, Hohn DC, Siddik ZH, Newman RA. Increased doxorubicin levels in hepatic tumors with reduced systemic drug exposure achieved with complete hepatic venous isolation and extracorporeal chemofiltration. Cancer Chemother Pharmacol 1993;33:251-7. Stuart K, Stokes K, Jenkins R, Trey C, Clouse M. Treatment of hepatocellular carcinoma using doxorubicin/ethiodized oil/gelatin powder chemoembolization. Cancer 1993;72:3202-9. Ishikawa O, Ohigashi H, Sasaki Y, Furukawa H, Kabuto T, Kameyama M, et al. Liver perfusion chemotherapy via both the hepatic artery and portal vein to prevent hepatic metastasis after extended pancreatectomy for adenocarcinoma of the pancreas. Am J Surg 1994;168:361-4. Taniguchi K, Nakata K, Kato Y, Sato Y, Hamasaki K, Tsuruta S, et al. Treatment of hepatocellular carcinoma with transcatheter arterial embolization. Cancer 1994;73:1341-5. Ku Y, Fukumoto T, Iwasaki T, Tominaga M, Samizo M, Nishida T, et al. Clinical pilot study on high-dose intraarterial chemotherapy with direct hemoperfusion in patients with advanced hepatocellular carcinoma. Surgery 1995;117:510-9. Kemeny N, Daly J, Reichman B, Geller N, Botet J, Oderman P. Intrahepatic or systemic infusion of fluorodeoxyuridine in patients with liver metastases from colorectal carcinoma: a randomized trial. Ann Intern Med 1987;107:459-65. Martin JK, O’Connell MJ, Wieand HS, Fitzgibbons RJ, Mailliard JA, Rubin J, et al. Intra-arterial floxuridine vs systemic fluorouracil for hepatic metastases from colorectal cancer: a randomized trial. Arch Surg 1990;125:1022-7. Venook AP, Warren RS. Regional chemotherapy approaches for primary and metastatic liver tumors. Surg Oncol Clin North Am 1996;5:411-27. Klopp CT, Alford TC, Bateman J, Berry GN, Winship T. Fractionated intra-arterial cancer chemotherapy with methyl bis amine hydrochloride. Ann Surg 1950;132:81132. Creech O, Krementz ET, Ryan RF, Winblad JN. Chemotherapy of cancer: regional perfusion utilizing an extracorporeal circuit. Ann Surg 1958;148:616-32. Hohenberger W, Meyer T, Göhl J. Extremitätenperfusion beim malignen Melanom. Chirurg 1994;65:175-85. Lejeune FJ, Liénard D, Eggermont AMN, et al. Clinical experience with high-dose tumor necrosis factor alpha in regional therapy of advanced melanoma. Circ Shock 1994;43:191-7. Eggermont AMM, Koops HS, Klausner JM, Kroon BBR, Schlag PM, Liénard D, et al. Isolated limb perfusion with tumor necrosis factor and melphalan for limb salvage in 186 patients with locally advanced soft tissue extremity sarcomas. Ann Surg 1996;224:756-5. Ackerman NB. The blood supply of experimental liver metastases. IV. Changes in vascularity with increasing tumor growth. Surgery 1974;71:589-96. Lin G, Lunderquist A, Hägerstrand I, Boijsen E. Postmortem examination of the blood supply and vascular pattern of small liver metastases in man. Surgery 1984;96:517-26. Michels NA. Newer anatomy of the liver and its variant

23.

24. 25. 26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37. 38.

39.

40.

blood supply and collateral circulation. Am J Surg 1966;112:337-47. Fortner JG, Penneman R, Krakoff IH. Actinomycin-D perfusion of the isolated liver for cancer. Bull Soc Int Chir 1975;5:399-403. Aust JB, Ausman RK. The technique of liver perfusion. Cancer Chemother Rep 1960;10:23-33. Ausman RK. Development of a technic for isolated perfusion of the liver. N Y State J Med 1961;61:3993-7. Boddie AW, Booker L, Mullins JD, Buckley CJ, McBride CM. Hepatic hyperthermia by total isolation and regional perfusion in vivo. J Surg Res 1979;26:447-57. Aigner K, Walther H, Tonn JC, Krahl M, Wenzl A, Merker G, et al. Die isolierte Leberperfusion mit 5-Fluorouracil (5FU) beim Menschen. Chirurg 1982;53:571-3. Shaw BW, Martin DJ, Marquez JM, Kang YG, Bugbee AC, Iwatsuki S, et al. Venous bypass in clinical liver transplantation. Ann Surg 1984;200:524-34. Atiq OT, Kemeny N, Niedzwiecki D, Botet J. Treatment of unresectable primary liver cancer with intrahepatic fluorodeoxyuridine and mitomycin C through an implantable pump. Cancer 1992;69:920-4. Marinelli A, de Brauw LM, Beerman H, Keizer HJ, van Bockel JH, Tjaden UR, et al. Isolated liver perfusion with mitomycin C in the treatment of colorectal cancer metastases confined to the liver. Jpn J Clin Oncol 1996;26:341-50. Asher A, Mulé JJ, Reichert CM, Shiloni A, Rosenberg SA. Studies on the anti-tumor efficacy of systemically administered recombinant tumor necrosis factor against several murine tumors in vivo. J Immunol 1987;138:963-74. Fraker DL, Alexander HR, Thom AK. Use of tumor necrosis factor in isolated hepatic perfusion. Circ Shock 1994;44:45-50. Kettelhut IC, Fiers W, Goldberg AL. The toxic effects of tumor necrosis factor in vivo and their prevention by cyclooxygenase inhibitors. Proc Natl Acad Sci U S A 1987;84:4273-7. Tsujimoto M, Okamura N, Adachi H. Dexamethasone inhibits the cytotoxic activity of tumor necrosis factor. Biochem Biophys Res Commun 1988;153:109-15. d´Oleire F, Kutz ME, Bird A, Robins HI. Step down heating and melphalan: cytotoxic interactions and clinical implications. Melanoma Res 1994;4:303-5. Robins HI, Cohen JD, Schmitt CL, Tutsch KD, Feierabend C, Arzoomanian RZ, et al. Phase I clinical trial of carboplatin and 41.8 degrees C whole-body hyperthermia in cancer patients. J Clin Oncol 1993;11:1787-94. Robins HI. Systemic hyperthermia oncological working group. Oncology 1995;52:260-3. Hafström L, Rudenstam CM, Holmberg SB, Schersten T, Ehrsson H. The pharmaco-kinetics of melphalan in regional hyperthermic liver perfusion. Reg Cancer Treat 1990;3:23-5. Naredi P, Holmberg SB, Hafström L, Heath DD, Shalinsky DR, Howell SB. Pharmako-kinetics of cisplatin in an isolated perfusion system in humans. Reg Cancer Treat 1992;4:254-7. Horikawa M, Nakajima Y, Kido K, Ko S, Ohashi K, Nakano H. Simple method of hyperthermo-chemo-hypoxic isolated liver perfusion for hepatic metastases. World J Surg 1994;18:845-51.