228 Heptobiliary Surgery 17 When Should HAIP Therapy Be Considered? At present, HAIP therapy is limited to clinical trials. Various drug regimens and combinations are under investigation; for the purpose of the present discussion, only issues pertaining directly to hepatic artery pump placement are reviewed. Clinical scenarios in which an HAIP is considered include: 1. Liver resection and colectomy in combination with HAIP placement. 2. Liver resection in combination with HAIP placement. 3. HAIP placement alone, without liver resection or colectomy. 4. Nonsurgical ablative therapy in combination with HAIP placement. The rationale for the use of HAIP chemotherapy is to provide regional treat- ment for patients considered to be at high-risk for the local recurrence, progression, or persistence of disease. Patients at high-risk may include those presenting with synchronous colon and hepatic disease, hepatic metastases within 24 months of the primary tumor, or patients undergoing hepatic resection with minimal free margins or demonstrated persistent disease. Nonresectional hepatic ablative therapies (e.g., radiofrequency ablation, cryoablation), for patients not otherwise candidates for surgical resection, have been popularized by the early promising results demonstrated in small series. In these patients, HAIP therapy may be used as a component of their treatment. However, the long-term outcome for this approach remains to be established. Preoperative Patient Evaluation This section will be necessarily brief, as the topic of metastatic work-up for patients with colorectal cancer is covered elsewhere. Patient eligibility for enrollment into clinical trial is specifically defined by the study protocol. In general, the following criteria are common to most HAIP clinical trials: •History of histologically confirmed colorectal adenocarcinoma metastatic to the liver with no clinical or radiographic evidence of extrahepatic disease. • Liver metastases must comprise < 70% of the liver parenchyma. • WBC >3,500 cells/mm 3 and platelet count > 150,000 cells/mm. • Albumin >2.0 gm% •Total serum bilirubin < 2.0 mg/dl. •No active concurrent malignancies. •No active infection, ascites, or hepatic encephalopathy. •Prothrombin time within 1.5 seconds of normal. Furthermore, patients meeting these criteria must have favorable vascular anatomy for HAIP placement. HAIP placement requires that a catheter (Fig. 17.1) be introduced into the gas- troduodenal artery (see following section, technique for placement of HAIP), so that pump flow is directed necessarily towards the liver via the hepatic artery(s). Traditionally, demonstration of the arterial anatomy has been done by celiac (visceral) angiography. This approach is invasive, has some (albeit low) associated morbidity, and is time consuming for the patient. More recently, magnetic resonance angiogra- phy (MRA), which is noninvasive and devoid of the angiography-associated mor- bidities, is being developed. At the present time, the radiographic images obtainable 229 Technique for Placement of the Hepatic Arterial Infusion Pump 17 by MRA provide at least equivalent anatomic information (Fig. 17.2), although their routine use awaits validation. Our preferred operative technique for HAIP placement in patients with stan- dard vascular anatomy is presented. Surgical Technique for the HAIP Placement Patients should receive prophylactic preoperative gram-positive antibiotic cover- age and this should be continued for 24 hours following the operation. Operative draping of the patient for this procedure should extend cranially to the nipple line and inferiorly to the symphysis. The skin is prepared with a bactericidal such as a Betadine or Hibiclens solution. The skin is dried with an absorbent sterile towel and a providone-iodine adhesive drape is placed on the operative field. We prefer the use of a right subcostal incision extended slightly beyond the midline. This approach provides ample exposure to the right upper quadrant and spares the left side of the abdomen for the creation of the pump pocket (Fig. 17.3). Intra-abdominal adhesions will be present in patients who have previously undergone colectomy. Some of these adhesions will need to be cleared to prevent inadvertent injury during retraction, dissection, or when passing the pump catheter across the abdomen from within the peritoneum. Cholecystectomy is routinely performed to prevent chemical cholecystitis. The hepatic artery is easily identified in the porta hepatis, both by location and palpable tell-tale arterial thrill. Favorable anatomy for pump placement has been verified prior to coming to the operating room by angiogram or MRA. Fig. 17.1. Visceral angiogram: (gastroduodenal artery, left hepatic artery, right hepatic artery). 230 Heptobiliary Surgery 17 Fig. 17.2. Visceral MRA: (gastroduodenal artery, left hepatic artery, right hepatic artery. Fig. 17.3. Planned incision for laparotomy (right sub-costal) and pump placement (left sub-costal). 231 Technique for Placement of the Hepatic Arterial Infusion Pump 17 The gastroduodenal artery (GDA) is located immediately inferior to the proper hepatic artery coursing inferiorly towards the proximal duodenum (Figs. 17.1 and 17.2). Sharp dissection to isolate the GDA is performed, taking care to ligate or clip the surrounding lymphatic vessels. The common bile duct (CBD) lies lateral to the GDA, and injury to the CBD must be avoided during the dissection. Medially and deep to the GDA lies the portal vein, and again, great care is needed while dissecting in this area. After circumferentially dissecting the GDA, a right angle instrument is used to facilitate the passing of a suture ligature around this vessel. This suture will be used for traction as arterial branches are cleared off for approximately 1 cm towards the duodenum and head of the pancreas, 1 cm toward the proximal common hepatic artery, and 1 cm towards the bifurcation of the hepatic artery (but proximal to the right and left divisions). Up to 15% of patients may have an accessory left hepatic artery arising from the left gastric artery, which should be (but is not always) identified by the preoperative imaging. In order to maintain drug-only arterial blood flow to the liver, this vessel must be identified and ligated. Elevation of the left lobe of the liver to assess for the presence of a nonvisualized left hepatic artery is mandatory. This is best accomplished by dividing the lesser omentum with ligation of the accessory vessel if it is present. Once this is accomplished, arterial flow from the GDA should be only towards the liver via the common hepatic artery and the liver should receive only drugs containing blood flow. Considerations for Patients with Variant Anatomy The technique previously described is applicable only for patients with standard arterial anatomy. As discussed, it is important to identify any aberrant arterial archi- tecture. By definition, accessory vessels are present, in addition to the normal ana- tomic structures, such that ligation and division of an accessory right or left hepatic artery can be performed without ischemic consequences. In contrast, replaced arter- ies may represent the sole arterial supply to a given lobe of the liver. Prior to ligation and division, replaced arteries should be clamped with a noncrushing vascular clamp to assess for potential liver ischemia. If no obvious ischemia is noted, the vessel can be ligated. However, at least one artery to either lobe of the liver arising from the hepatic artery after the take-off of the GDA must be preserved. Crosshepatic arterial perfusion to the contralateral lobe of the liver develops rapidly after ligation of the replaced lobar artery in the patient with a nontoxic liver. In contrast, if ischemic changes are observed during “trial-clamping”, then that vessel should be preserved and cannulated separately to maintain adequate hepatic perfusion. Replaced Right Hepatic Artery (RRHA) The most common variant arterial anatomy is a replaced RHA with a standard left hepatic artery (LHA) and GDA. A RRHA is most commonly identified by palpation as it exits from behind the duodenum, courses lateral to the common bile duct, and posteriorly within the hepatoduodenal ligament. The RRHA has no side branches that can be used to insert the catheter; however, arteriotomy or direct puncture under vascular control can be used for placement of a polyethylene or 22 232 Heptobiliary Surgery 17 gauge “angiocatheter” with subsequent connection to the HAIP catheter. Placing the catheter directly in the RRHA, and ligating the LHA, results in the develop- ment of crossover perfusion of the left lobe of the liver. Another approach is the placement of a double-catheter pump where one catheter is placed into the LHA from the GDA and the second catheter is inserted directly into the RRHA to assure bilateral perfusion. The latter is the approach we prefer when possible. Replaced Left Hepatic Artery (RLHA) A RLHA can be identified coursing within the gastrohepatic ligament along the inferior edge of the liver. The RLHA arises from the left gastric artery (LGA) and there is a normal RHA. In this setting, the HAIP catheter can be inserted into the GDA to perfuse the RHA with ligation of the RLHA. Another technique is the placement of a double-catheter pump, where one catheter is inserted into the LGA to perfuse the RLHA and the other catheter is inserted into the GDA to perfuse the RHA. Trifurcation of the Common Hepatic Artery (CHA) into RHA, LHA and GDA Trifurcation of the CHA into a RHA, LHA and GDA is another described ana- tomic variant that can be managed by inserting the HAIP catheter into the GDA 1 cm proximal to the takeoff of the GDA from the CHA to allow for adequate drug- blood mixing. The GDA is then ligated. Although feasible, this approach is associ- ated with an increased risk for CHA thrombosis. Alternatively, insertion of the HAIP catheter into the splenic artery, advancing it into the CHA with ligation of the GDA and left and right gastric arteries, will obtain bilateral perfusion. A more complex approach to deal with this anatomy is to construct a conduit using a short segment vein graft for catheter insertion into a selected artery with subsequent ligation of the other vessels. We do not recommend this latter approach. These variant arterial anatomies underscore the importance of preoperative assessment of vascular anatomy in order to be prepared at the time of operation. Failure to appropriately identify and ligate aberrant arteries and arterial branches may result in malperfusion. Creation of the Subcutaneous Pump Pocket The pump pocket is usually created on the left side of the abdominal wall unless a colostomy is present. The pump is placed well below the left costal margin to prevent patient discomfort from the hard metal pump abutting against the ribs. The skin incision is transverse and should measure no more than 1 cm wider than the width of the actual pump to be placed (Fig. 17.4). The subcutaneous tissues are divided using electrocautery up to, but not into, the abdominal wall fascia. Continuing inferiorly along the adipose-fascial interface, the pocket is extended for about 8-10 cm (actual pocket size should be planned based on the size of the actual model of pump to be implanted). Strict attention to hemostasis must be maintained during the dissection of the subcutaneous pocket to minimize the risk of hematoma and lessen the risk of subsequent complications. The automatic mechanism that drives catheter flow is initiated (“primed”) by filling the pump reservoir with a standard volume of heparinized saline (volume is specific to the pump model), which is placed in a warm water bath 233 Technique for Placement of the Hepatic Arterial Infusion Pump 17 at body temperature to activate it. Flow through the catheter is driven by hydro- static pressure secondary to internal expansion from thermal activation. The deliv- ery mechanism is graduated to deliver a set rate of flow, and adjustment of the solution concentration regulates the amount of drug delivery. The pocket is tested for fit of the pump body and catheter while maintaining a “no-contact” technique with the skin. Pump architecture and specifications will vary according to manufacturer and model; however, all designs will have a catheter with an internal component that exits through the pump body (Fig. 17.5). This catheter must be inserted into the abdominal cavity through the anterior abdominal wall and subsequently placed into the GDA. In performing this maneuver, care should be taken to avoid injury to vessels coursing through the posterior surface of the rectus abdominus muscle. Pumps have anchoring “loops” (Fig. 17.5) fused to the pump body housing to prevent “flipping” or axial rotation of the pump, within the subcutaneous pocket. Using a skin retractor, the inferior edge of the pump pocket is elevated towards the ceiling; interrupted nonabsorbable sutures are placed into the fascia of the anterior abdominal wall and passed through each anchor loop. The two distal anchors are sutured first and not tied until the two proximal sutures are placed. Until the sutures are tied, the ends of the cut suture are held in place by individual clamps. The inferior pump pocket edge is elevated and the distal anchor sutures are tied, pulling the pump down into the pocket as it is fastened down. Care should be taken to insure that the pump catheter does not become twisted or damaged during this maneuver. With the pump in place, the catheter is free within the abdomen. Fig. 17.4. Abdominal CT with HAIP in the left lower quadrant. Note that the pump body housing rests directly on the fascia. 234 Heptobiliary Surgery 17 Fig. 17.5. Hepatic artery infusion pump; the catheter exits the pump body housing at the 6 o’clock position. Note the four anchor loops located around the body housing. The catheter is measured and cut sharply such that the tip has an oblique angle and only one of the flanged catheter hubs remains. The catheter length should be long enough to reach the junction of the GDA with the hepatic artery but not extend beyond it. In this manner, flow-out of the catheter will necessarily go only towards the liver. The GDA is ligated in continuity as it enters the duodenum. The position of this suture will determine the length of artery available for catheter insertion. Once the catheter is prepared as described above, and the branches off of the GDA and other arteries have been cleared, proximal and distal control of the hepatic artery is obtained using pediatric vascular clamps. A #11 scalpel blade is used to incise the anterior wall of the GDA in a transverse fashion. An arteriotomy introducer is in- serted into the GDA lumen to facilitate passing of the catheter tip into the vessel lumen. The catheter is advanced using nontoothed forceps and the previously placed circumferential traction suture ligature is tied down onto the catheter behind the flanged hub. Care is needed to insure that the knot is not too tight as to impede or occlude flow through the catheter. We advocate the use of an additional separate suture to prevent inadvertent arterial dislodgement. We place a 4-0 prolene suture through the adventitia of the GDA behind the flanged hub of the catheter and circumferentially encompassing the vessel, again not encroaching on the catheter lumen. The distal (outflow) vascular clamp is removed, followed by the proximal (inflow) clamp. 235 Technique for Placement of the Hepatic Arterial Infusion Pump 17 Assessment of Hepatic Perfusion Intraoperative fluorescein test is generally performed after pump placement to assure liver-only perfusion. Using the side port of the pump, approximately 3-5 cc of fluorescein (nondiluted) are injected, followed by illumination with a Wood’s lamp. If all the arterial branches have been properly ligated or clipped, only the liver should demonstrate fluorescence. Do not aspirate into the pump when injecting it as this will result in small clot formation and subsequent pump malfunction. Subse- quently, the pump is flushed with 10 cc of heparinized saline. The pump pocket is closed in two layers using absorbable suture material. The abdominal incision is closed in standard fashion. Postoperative hepatic perfusion and extravasation (i.e., stomach, duodenum, and spleen) via HAIP is assessed three days postoperatively by a Technetium-labeled macroaggregated albumin hepatic scan. This will confirm liver-only pump perfu- sion and is necessary prior to the initiation of HAIP therapy (Fig. 17.6). Brief Comments on Technical Complications Analysis of the technical complications associated with hepatic artery infusion pump placement, reveals that HAIP complications can be divided into those occur- ring either early (< 30 days postpump placement), or late (>30 days) (Table 17.1). Furthermore, complications are divided into arterial, catheter, perfusion or pump related. Arterial complications include intimal dissection and thrombosis. Arterial throm- bosis that occurs early can be salvaged by reoperation and revision of the catheter; thrombosis in the early postoperative period is often salvaged by thrombolytic therapy. Late arterial thrombosis occurs more frequently and is less likely salvaged. Catheter complications included dislodgement or migration of the GDA cath- eter from its original site or out of the artery with or without hemorrhage. Catheter dislodgement is a rare occurrence in the early postoperative period; most dislodgement occurs late. Catheter dislodgement may be accompanied by hemorrhage that can be life-threatening. Catheter occlusion is a late occurrence and can be salvaged by lytic therapy. Malperfusion is defined as the identification of organ perfusion other than the liver (i.e., stomach, duodenum or spleen). Extrahepatic perfusion can occur with catheter misplacement or failure to identify and ligate the appropriate arterial branches off the GDA. If malperfusion is apparent, a patent branch of the common hepatic arterial system that continues to perfuse the distal stomach or duodenum is often identified. Once localized, this complication is corrected by laparotomy with liga- tion of the patent branches of the GDA or CHA. More recently, patent vessel branches have been embolized at arteriography, obviating the need for re-laparotomy. Pump infections are more frequent in the late period (> 30 days postoperatively) and are the pump related complication that most often results in the premature termination of therapy. However, pump salvage rates of approximately 25% are obtainable for both early and late pump infections. Complications related to local and regional toxicity of the therapeutic agents are well characterized elsewhere. 1,5 236 Heptobiliary Surgery 17 Fig. 17.6. Macro-aggregated albumin scan demonstrating pump to liver-only per- fusion. In the top row images the liver is demonstrated immediately. The catheter is evident and although the actual pump is not well-visualized, it is located where the catheter originates. Table 17.1. Early and late technical complications of HAIP ( n=391) 6 Complication Early Terminated Late Terminated Arterial thrombosis 10 (1%) 6 (2%) 15 (4%) 14 (4%) Catheter dislodgement 3 (1%) 2 (1%) 22 (6%) 12 (3%) Catheter occlusion 0 — 9 (2%) 4 (1%) Pump infection 4 (1%) 3 (1%) 8 (2%) 6 (2%) Malperfusion 10 (3%) 0 3 (1%) 0 Total 27 (7%) 11 (3%) 57 (15%) 36 (9%) The technical complications following HAIP placement between 1986 and 1995 (N=391). Complications were defined as early (<30 days ) and late (>30 days) following pump placement. These data represent a median follow-up interval of 16 months (range 1-110). A total 3% early and 9% late technical failure rates were observed. The prevention of late arterial thromboses and catheter dislodgement would prevent the majority of premature treatment terminations related to pump complications. 237 Technique for Placement of the Hepatic Arterial Infusion Pump 17 Summary In the absence of efficacious systemic chemotherapy regimens to treat hepatic colorectal metastases, regional therapies will continue to be investigated. HAIP place- ment has a low associated mortality; however, the challenge for the surgeon in the management of these patients is to provide a reliable system for the administration of regional chemotherapy. The efficacy of hepatic artery infusion pump (HAIP) chemotherapy for metastatic colorectal cancer confined to the liver is currently the subject of several national multi-center randomized trials. Selected Reading 1. Allen Mersh, TG, Earlam S, Fordy C, Abrams K, Houghton J. Quality of life and survival with continuous hepatic-artery floxuridine infusion for colorectal liver metastases [see comments]. Lancet 1994; 344:1255-6000. 2. Chang AE, Schneider PD, Sugarbaker PH et al. A prospective randomized trial of regional versus systemic continuous 5-fluorodeoxyuridine chemotherapy in the treatment of colorectal liver metastases. Ann Surg 1987; 206:685-933. 3. Ensminger WD, Rosowsky A, Raso V et al. A clinical-pharmacological evaluation of hepatic arterial infusions of 5-fluoro-2'-deoxyuridine and 5-fluorouracil. Can- cer Res 1978; 38:3784-3922. 4. Grage TB, Vassilopoulos PP, Shingleton WW et al. Results of a prospective ran- domized study of hepatic artery infusion with 5-fluorouracil versus intravenous 5-fluorouracil in patients with hepatic metastases from colorectal cancer: A Cen- tral Oncology Group study. Surgery 1979; 86:550-555. 5. Kemeny N, Daly J, Reichman B et al. Intrahepatic or systemic infusion of fluorodeoxyuridine in patients with liver metastases from colorectal carcinoma. A randomized trial. Ann Intern Med 1987; 107:459-655. 6. Dudrick PS, Picon A, Paty PB et al. Technical Complications Associated with Hepatic Arterial Infusion Pumps for Metastatic Colorectal Cancer. (In Prepara- tion) 7. Sigurdson ER, Ridge JA, Kemeny N et al. Tumor and liver drug uptake following hepatic artery and portal vein infusion. J Clin Oncol 1987; 5:1836-400. 8. Wagner J S, Adson MA, Van Heerden JA et al. The natural history of hepatic metastases from colorectal cancer. A comparison with resective treatment. Ann Surg 1984; 199:502-588. 9. Weiss GR, Garnick MB, Osteen R et al. Long-term hepatic arterial infusion of 5-fluorodeoxyuridine for liver metastases using an implantable infusion pump. J Clin Oncol 1983; 1:337-444. 10. Wingo PA, Ries LA, Rosenberg HM et al. Cancer incidence and mortality, 1973- 1995: a report card for the U.S. Cancer 1998; 82:1197-2077. [...]... Mets Allgaier ( 199 9) HCC 23 14 12 Lencioni ( 199 8) HCC 80 85% PEI Rhim ( 199 9) HCC Mets 25 17 Siperstein ( 199 9) NET Mets 66 Bauer ( 199 9) Mets 20 Recurrence rates RFA-2% PEI-13% 69% complete ablation on CT 13% recurrence Four complications 2 IP bleeds 1 needle track seeding 1 diaphargm injury 88% demonstrated shrinkage after ablation 12% recurrence (N = 22) 92 % experience > 50% increase in pre-op CEA No complications... metastases: treatment and follow-up in 16 patients Radiology 199 7;202: 19 5-2 03 Solbiati L, et al Hepatic metastases: percutaneous radio-frequency ablation with cooled-tip electrodes Radiology 199 7;205:36 7-7 3 Steiner P, et al Radio-frequency-induced thermoablation: monitoring with T1weighted and proton-frequency-shift MR imaging in an interventional 0.5-T environment Radiology 199 8;206:80 3-1 0 Trubenbach J, et al... survival Shafir ( 199 6) CRM HCC Other 39 Cryo alone 65% 3-year actuarial survival Crews ( 199 7) CRM HCC Other 40 Cryo alone 30% 5-year actuarial survival Weaver ( 199 5) CRM Other 140 Combined series 62% 2-year actuarial survival 22 mo median survival Wrens ( 199 7) HCC 12 Cryo alone 19 mo median survival Ravikumar ( 199 1) CRM HCC 32 Combined series 63% overall actuarial survival Onik ( 199 1) CRM 18 Combined... Ther Oncol 199 6;1:31 2-6 Patterson EJ, et al Radiofrequency ablation of porcine liver in vivo: effects of blood flow and treatment time on lesion size Ann Surg 199 8;227:55 9- 6 5 Rhim H, et al Radiofrequency thermal ablation of liver tumors J Clin Ultrasound 199 9;27:22 1 -9 Rossi S, et al Percutaneous RF interstitial thermal ablation in the treatment of hepatic cancer AJR Am J Roentgenol 199 6;167:75 9- 6 8 Rossi... electrosurgery Acad Radiol 199 6;3:21 9- 2 4 Marone G, et al Echo-guided radiofrequency percutaneous ablation of hepatocellular carcinoma in cirrhosis using a cooled needle Radiol Med (Torino) 199 8 ;95 :62 4 -9 Mazziotti A, et al An appraisal of percutaneous treatment of liver metastases Liver Transpl Surg 199 8;4:27 1-5 McGahan JP, et al Hepatic ablation using bipolar radiofrequency electrocautery Acad Radiol 199 6;3:41 8-2 2... neuroendocrine tumor metastases Surgery 199 7;122:114 7-5 5 Solbiati L, et al Radio-frequency ablation of hepatic metastases: postprocedural assessment with a US microbubble contrast agent—early experience Radiology 199 9;211:64 3 -9 Solbiati L New applications of ultrasonography: interventional ultrasound Eur J Radiol 199 8;27:S20 0-6 Solbiati L, et al Percutaneous US-guided radio-frequency tissue ablation of... treatment 32% 5-year actuarial survival Haddad ( 199 8) CRM PLC Other 31 Combined series 7 mo median disease free survival Seifert & Morris ( 199 8) CRM 116 Yeh ( 199 7) CRM 24 Combined series 13% 5-year survival 26 mo median survival Combined series 85% 3-year actuarial survival 31 mo median survival Adam ( 199 7) HCC CRM 34 Combined series 52% 5-year survival Korpan ( 199 7) CRM 63 Combined series 4% 5-year actuarial... 199 6;3:41 8-2 2 McGahan JP, et al Hepatic ablation with use of radio-frequency electrocautery in the animal model J Vasc Interv Radiol 199 2;3: 29 1-7 McGahan JP, et al Hepatic ablation using radiofrequency electrocautery Invest Radiol 199 0;25:26 7-7 0 Miao Y, et al Ex vivo experiment on radiofrequency liver ablation with saline infusion through a screw-tip cannulated electrode J Surg Res 199 7;71:1 9- 2 4 Nativ... Radiol 199 6;3:21 2-8 Goldberg SN, et al Radiofrequency tissue ablation: increased lesion diameter with a perfusion electrode Acad Radiol 199 6;3:63 6-4 4 Goldberg SN, et al Tissue ablation with radiofrequency using multiprobe arrays Acad Radiol 199 5;2:67 0-4 Goldberg SN, et al Tissue ablation with radiofrequency: effect of probe size, gauge, duration, and temperature on lesion volume Acad Radiol 199 5;2: 39 9- 4 04... Am 199 5;1:73 Rossi S, et al Thermal lesions induced by 480 KHz localized current field in guinea pig and pig liver Tumori 199 0;76:5 4-7 Scudamore CH, et al Radiofrequency ablation followed by resection of malignant liver tumors Am J Surg 199 9;177:41 1-7 Sinha S, et al Phase imaging on a 2-T MR scanner: application to temperature monitoring during ablation procedures J Magn Reson Imaging 199 7;7 :91 8-2 8 . mortality, 197 3- 199 5: a report card for the U.S. Cancer 199 8; 82:1 19 7-2 077. CHAPTER 18 Hepatobiliary Surgery, edited by Ronald S. Chamberlain and Leslie H. Blumgart. ©2003 Landes Bioscience. Non-Resectional. Surg 198 7; 206:68 5 -9 33. 3. Ensminger WD, Rosowsky A, Raso V et al. A clinical-pharmacological evaluation of hepatic arterial infusions of 5-fluoro-2'-deoxyuridine and 5-fluorouracil. Can- cer. 199 :50 2-5 88. 9. Weiss GR, Garnick MB, Osteen R et al. Long-term hepatic arterial infusion of 5-fluorodeoxyuridine for liver metastases using an implantable infusion pump. J Clin Oncol 198 3; 1:33 7-4 44. 10.