Hepatic resection with vascular isolation and routine supraceliac aortic clamping

Hepatic Resection With Routine Supraceliac

and

Vascular Isolation Aortic Clamping

Michael S. Stephen, FRACS, P. James Gallagher, FRACS, A.G. Ross Sheil, FRACS, Donald M. Sheldon, FRACS, David W. Storey, FRACS, Sydney, Au.stra/ia

BACKGROUND: Hepatic resection with total vascular isolation has been rBpOrted to reduce hemorrhage. Addition of supraceliac aortic clamping putatively avoids hemodynamic instability, but may increase morbidity. MEWODS: This technique was used in 99 major liver resections utilizing scalpel division and suture hemostasis. RESULTS: LiVerS were IIOrIIIBl in 86 patients, cirrhotic with no portal hypertension in 5, and cirrhotic with portal hypertension in 8. There was 1 death in 91 patients with no portal hypertension compared with 5 in 8 patients with portal hypertension due to hepatic failure or bleeding esophageal varices. There were 59 hemihepatectomies and 40 segmentectomies. Median operating time was 145 and 110 minutes, respectively, and mean transfused blood was 4 and 0 units, respectively, with minimal morbidity. CONCLUSIONS: Use of total hepatic vascular isolation with routine supraceliac aortic clamping is a safe and expedient method of hepatic resection that limits blood loss and maintains hemodynamic stability, but does not increase morbidity. However, the presence of portal hypertension precludes safe resection. Am J Surg. 1996;171:351355.

D

escriptions of hepatic anatomy, based on liver vasculature and functional units by Couinaud’ and Ton,* have been popularized by Bismuth et al.3 Adoption of this nomenclature as a guide to hepatic resection, together with advances in preoperative and perioperative hepatic imaging, has enabled resection to be linked to intrahepatic anatomy and has eliminated the confusion that compromised earlier accounts. In the last decade, there have been numerous advances in techniques and technology that have been directed at reducing morbidity and mortality from hepatic resection. However, what is to be considered as a standard technique is yet to be clarified.

From the Departments of Upper Gastrointestinal Surgery (MSS, PJG, DMS, DWS), and Liver Transplantation (AGRS), Royal Prince Alfred Hospital, Camperdown, Sydney, Australia. Requests for reprints should be addressed to Dr. P.J. Gallagher, Suite 11, Missenden Medical Centre, Brfggs Street, Camperdown 2050, New South Wales, Australia. Manuscript submitted October 27, 1994 and accepted in revised form June 14, 1995.

THE AMERICAN

Clamping of the portal triad (the Pringle maneuvre)+ decreases hemorrhage from a cut liver edge. Additi.on of suprahepatic and infrahepatic inferior vena cava (WC) clamps establishes total hepatic isolation (THI), and further limits blood loss from the hepatic venous system.5s Moreover, the risk of air embolism is also decreased.9 The reported 29% mortality hampered initial acceptance of this technique.‘O Heaney and coworkers5 first examined the use of THI in a canine model with routine supraceliac aortic clamping. This putatively avoided the subsequent criticism that total vascular isolation may cause hemodynamic instability in up to 40% of patients.8 Nevertheless, aortic clamping has been avoided completely,6s7 employed seldomly or only in urgent situations,sJ1J2 and studied with and without its application.13 In 1990, we reported an original preliminary study favorably comparing total hepatic vascular isolation and routine supraceliac aortic clamping (THIS) with a historical control group. l4 The present study reports routine application of this technique in a series of patients requiring hepatic resection.

PATIENTS AND METHODS Between February 1988 and September 1993,, 99 patients underwent hepatic resection using THIS. Fifty-nine patients had classically described hemihepatectomies and 40 patients had segmentectomies (Table I). In this latter group, the patients were unable to have more minor techniques (eg, Lin clamp, simple sling, subsegmentectomy, or wedge resection). All patients were assessed by baseline full blood count, coagulation studies, liver function tests, and routine electrolytes. All had a preliminary ultrasound and computed tomography (CT) of the abdomen. The indications for hepatic resection for benign disease were prophylaxis against bleeding in adenomata, a multiloculated Streptococcus ntilkri abscess destroying the right lobe, and a benign cyst in segment VI thought initially to be a malignant extension of a renal cell carcinoma. In patients with hepatic malignancy, extrahepatic malignancy and, if indicated, primary recurrence were excluded. Following this, a CT scan during portal venography was performed. Resection proceeded provided adequate clearance was considered possible. In 3 patients with colorectal cancer secondaries (CRCS) and 2 patients with hepatocellular carcinoma (HCC), resection of a single lesion in an anatomically discrete lobe was performed in addition to the ma:jor resection. Four of these were in addition to a left hemihepatectomy and the other with a bisegmentectomy of VI and VII. The postoperative histologic diagnoses are listed in Table II. Following an upper abdominal transverse incision with a superior midline extension, bilateral subcostal retraction was JOURNAL

OF SURGERYa

VOLUME

171

MAFtCH

1996

351

HEPATIC RESECTION/STEPHEN

ET AL 1

TABLE I Extent of Hepatic Segments Resected in 99 Patients According to the Couinaud Nomenclature1 Total

classical resections Right hemihepatectomy V-VIII Left hemihepatectomy II-IV f I Extended right V-VIII + IV Extended left II-V + VIII f I Total segmentectomies II-111 v + VI VI + VII VII + VIII IV VII VIII

TABLE

II Liver Pathology

Necessitating

59 33 16 6 2 40 13

3 6 12 2 1

3

Hepatic Resection

Total

benign disease Cholangiohepatitis Adenoma Focal nodular hyperplasia Hydatid Simple cyst Abscesses in right lobe Total primary malignancy Hepatocellular carcinoma (11 with cirrhosis) Cholangiocarcinoma Hemangiosarcoma Total secondary malignancy Colorectal carcinoma (2 with cirrhosis) Leiomyosarcoma Melanoma Squamous cell carcinoma Ovarian carcinoma Stomach carcinoma Pancreatic carcinoma Carcinoid

18 6 4 4 2 1 1 25 17 7 1 56 44 3 3 2 1 1 1 1

applied, and a hepatic ultrasoundusingthe mapping technique developed at our hospital15was performedto verify the lesions,particularly their proximity to major vascular structures. All resectionswere performedusingTHIS. The liver was completely mobilized by division of the falciform, left and right triangular, and anterior and posterior coronary ligaments. An aberrant left hepatic artery wassought prior to division of the lesseromentum. The suprahepaticIVC was encircled between the liver and the diaphragm,asopposed to the work of Heaney et al5 when isolation wasobtained via the pericardium.The key to this dissectionwas elevation of a fold of diaphragmabovethe right hepatic vein and IVC. This allowed a plane to develop close to the right atrium, which avoided the posteriortributaries of the WC. The infrahepatic IVC wasencircled superiorto the right renal vein. The supraceliacaorta wasexposed,without being encircled, via the crus of the diaphragmafter lateral retraction of the esophagus. The portal triad, and the aberrant 352

THE AMERICAN

JOURNAL

OF SURGERYB

VOLUME

171

left hepatic artery if present,were encircled.The portions of the liver to be resectedaccordingto Couinaud’ssegments’ weremarkedsuperficiallywith diathermy. Clampswerethen appliedconsecutivelyto the aorta, the infrabepaticIVC, the porta hepatis(and the aberrant left hepatic artery), and the suprahepaticIVC. That order of clamp application permitted hepatic decompression of blood asit betcameavascular. Resection was accomplishedby sharp dissection with a scalpel.The infrahepatic IVC clamp is angledsuperiorlyto occlude the adrenal vein, which has a variable entry into the IVC and can causesignificant bleeding during the resection. The vascularand biliary structureson the cut liver surface were then oversewn with 3-O vascular sutures. Following this, the infrahepatic WC, suprahepaticIVC, and portahepatis clamps were serially released,and bleeding points were securedaseach wasremoved. Finally, the aorta was gradually releasedwith application of bilateral groin pressureto compressthe femoral arteriesand avoid any hypotension due to suddenreperfusionof bol:h lower limbs. Bolusesof fluid were usedto maintain blood pressureasarterial circulation wasrestored.Any residualarterial or portal bleedingpoints wereoversewnwith or without brief reapplication of the portahepatis clamp. The oversewn cut surfaceof the liver wassealedwith Argon BeamCoagulation (Birtcher Le Monte, Irvine, California) andwasthen bathed with a fibrin glue. The abdomenwasclosedwith two suction drains left in situ. In somepatients, due to en bloc resectionof tumor, there wasneed for reconstruction of vascularor biliary structures. Hepaticojejunostomywasusedfor the 7 cholangiocarcinoma resections,and the left hepatic duct was resecteden bloc with tumor and reconstructedover a T-tube in 1 CRCS patient. Similarly, 2 patients required arterial reconstruction with reversedlong saphenousvein. The WC waspatched with autologous material in 4 patients (2 CRCS, 1 leiomyosarcomasecondary,and 1 HCC).Two CRCS patients had en bloc resectionand primary repair of a segment of diaphragm.The abdomenwasclosedwithl suction drains placed along the resectionmargin. Data are presentedasa median with rangesin parentheses,excluding the blood results,which arepresentedasmean k standarderror of the mean (SEM).

RESULTS Forty-two women and 57 men had hepatic resectionsusing THIS. The median agewas 57 years(range 25 to 78). Resectiondata fall into two groupsaccordingto whether a major resectionor segmentectomies were done (Table III). Only 1 patient had a hepatic resection(segmentsV to VII) for CRCS at the sametime asthe primary surgery.This was the only death in the normal liver group, and wasdue to a mixture of high intraoperative blood lossfrom the colon resection, a urinary fistula, and acute Budd-Chiari syndrome from rotation of the hepatic remnant on the IVC. Comparisonof postoperative and preoperative blood results showedno difference except for serumtransaminases (Table IV). Thesereturned to normal by the first postoperative visit. Morbidity included 3 patients requiring return to the operating room for bleeding. All were transfused,but only 2 MARCH

1996

1 HEPATIC RESECTION/STEPHEN had arterial bleeding sites found on the capsule of the liver that were oversewn. All survived the stay and were discharged. The patient with the biliary fistula was drained via operative drains and was hospitalized for a total of 21 postoperative days until the fistula closed. There were 2 patients with deep venous thrombosis and no pulmonary emboli who were heparinized and discharged at day 15. Subphrenic collections occurred in 5 patients; 4 were drained percutaneously and 1 required operative intervention and was hospitalized for 20 postoperative days. One patient developed portal venous thrombosis after resection and repair and was discharged on day 12 after thrombectomy. Mortality during the hospital stay varied according to whether the patient had a normal liver, cirrhosis with no portal hypertension, or cirrhosis with portal hypertension (Table V). The deaths in the latter group were due to esophageal varices hemorrhaging in 3 patients and acute liver failure in 2 patients.

TABLE

III Perioperative Data Classical Resections

Operative time (min) Range Clamp time (min)

Range Blood loss (L) Range Blood transfused (units) Range Postoperative hemoglobin (g/dL) Range Intensive care unit stay (d) Range Postoperative stay (d) Range Data are presented mean * standard

COMMENTS Total hepatic isolation allows parenchymal dissection according to the segmental anatomy of the liver in a bloodless field, which facilitates precise identification and resection of intrahepatic vessels and ducts without risk of air embolism. The amount of liver removed is specifically related to the tumor bulk and position and avoids ligation of vessels providing inflow and oufflow for the portion of the liver to be preserved. As with the intrahepatic resection technique popularized by Launois et al, 16~17there is no need for formal extrahepatic identification of vessels. Removal of more extensive disease involving major vascular structures is possible with an en bloc resection. Furthermore, simpler removal of additional segments in anatomically discrete areas of the liver is facilitated, without adding significantly to the operative time or morbidity and mortality. Total hepatic isolation has been criticized because of hemodynamic instability incurred with application of the IVC clamps.s Routine use of supraceliac aortic clamping5s14 avoids this and also the need for large boluses of fluid. Hemodynamic irregularities can be easily corrected prior to resection by release and reapplication of the aortic and venous clamps. Reported complications of supraceliac aortic clamping, such as ventricular arrhythmias, aortic injury, embolic phenomena, acute renal failure, and spinal cord ischemia,’ were not seen. In addition, there was no increase in venous complications. Meticulous positioning of clamps during THlS is required or hemorrhage continues during the resection from adrenal and posterior IVC tributaries. Application of the aortic clamp reduces blood flow through the lower half of the body and further reduces venous bleeding. An unavoidable 200to X0-mL blood loss occurs during scalpel resection with THIS, as residual blood in the liver drains. Blood transfusion is preferably avoided owing to limited resources, expense, transfer of infection, and potential immunosuppression. This latter contention is supported by improved renal allograft function’s and earlier cancer recurrence”J’ following blood transfusion. In this series, transfusion was required for 38% of segmentectomies and 78% of formal resections. The postoperative hemoglobin in this series was not signifTHE AMERICAN

TABLE

ET AL 1

145 90-260 18 13-38 1.25 0.5-6.22 4 O-18 12.4

‘I 10 70-l 66 13 9-25 0..675 0.32-2.1 0 o-1 1 11.4

7.7-15.1 1 l-32 11 7-38

9-l 4.6 1 ‘I-6 9 7-19

as median error

Segmentectomies

values, of the mean.

except

blood

results

which

are

IV Perioperatlve Parameters Reflecting Renal and Hepatic Function Preoperative Day 1

International numerical ratio (n=32) Prothrombin index (“I) (n = 67) Creatinine (pmol/L) Alkaline phosphatase (U/L) Bilirubin (pmoi/L) Alanine transferase (U/L)

Day 3

1.2 f 0.3

1.4 f 0.5

1.2 t 0.4

99 f 1.5

91 * 2.3

97 zt 1.8

76 zt 6 102 *32 32i 15 465 f 82’

78 i 9 142 * 19 24i 16 301 * 96’

78 f 140 f 18* 62i

8 41 11 15

‘P <0.001.

TABLE

V Postoperative

Mortality

Patient Groups

Death Rate

Normal liver Cirrhosis and no portal hypertension Cirrhosis and portal hypertension

1 of86 0 of 5 5 of 8

icantly different from preoperative levels. This, may reflect that transfusions were used too frequently, especially as familiarization with the technique evolved. Hemoglobin levels after hepatic resection are seldomly reported, but would be safely lower than in this study allowing the transfusion requirements to be reduced. Numerous techniques have been described to enable parenchymal division since the advent of finger fracture. With THIS, scalpel dissection, with oversewing of hepatic venous and portal triad structures, is original. It is essential that the scalpel dissection obtain smooth cut :sutfaces with no crevices in the hepatic parenchyma, or else hemostatic suturing becomes imprecise and troublesome. The technique allows accurate and swift en bloc resection and simplifies removal of anatomical separate segments or vascular structures JOURNAL

OF SURGERY@

VOLUME

171

MARCH

1996

353

that have been described for cure of malignant disease.21J2 Precise control of intrahepatic portal triad and hepatic venous structures is obtained during THIS and is completed as each vascular clamp is released. It is essential that the “fig ure 8” vascular sutures are not tied tightly or they will sever liver parenchyma. Sutures are just snug, and tighten coincident with the engorgement of the liver with blood as revascularization occurs. Few postoperative subphrenic collections, biliary fist&s, or primary arterial hemorrhages support using this technique. However, with THIS, particular care should be taken to oversew capsular arteries prior to coagulation because the two postoperative arterial bleeding sites that were found were from this source and both had been coagulated and were not bleeding at the completion of the resection. The precise anatomical resection eliminates residual devascularized liver parenchyma, which may contribute to these complications. Median postoperative stays of 1 day in the intensive care unit and 10 days of hospitalization reflect the safety of THIS. The liver is exposed to normothermic ischemia with THIS, but there have been numerous reports of the liver tolerating normothermic ischemia7J3.23beyond the classic limit of 15 to 20 minutes.4 Furthermore, it has been shown that morbidity and mortality do not increase if normothermic ischemia is maintained without revascularization for over 1 hour.12*24This length of normothennic ischemia was never necessary in this study, even for complex resections. Hepatic inflow exclusion techniques have reported the necessity for intermittent release of clamps17 to avoid the postoperative hepatic cytolysis reported to occur in animals25 and humans’ in direct correlation with the duration of hepatic ischemia. Hypothermic perfusion of the asanguinous liver has been described during THI.6 Similar to other studies,ss23the hepatic transaminases were raised for a short period of time, but returned to normal by the first postoperative visit, and renal function remained unchanged. Neither intermittent revascularization nor hypothermia were required with THIS and, asshown by Bismuth et al,26 the risk of postoperative liver failure is directly proportional to the severity of preexisting liver disease and the amount of liver parenchyma requiring excision. There is no doubt that hepatic resection is possible in cirrhotic livers, with the major thrust being towards lesser resections. There are conflicting reports about the influence of cirrhosis on the prognosis of hepatic resection, especially in patients with HCC. Methods used to classify cirrhotic patients suitable for resection have difficulty in predicting outcome. Neither Child’s classification27-29 nor mathematical models of metabolic function30 and hepatic blood flow studies29J1have demystified the issue. In this series, the presence of portal hypertension was the main predictor of mortality, irrespective of Child’s classification. No mortality was seen in the group with cirrhosis and no portal hypertension (PHT) compared with 63% once PHT was present. The reason for mortality was either lack of hepatic reserve or variceal bleeding due to the reported striking increase in portal pressure.32 Four of the 5 deaths in the cirrhosis with PHT group required formal resections to obtain adequate tumor clearance. There has been a reported 50% mortality with THI in cirrhotic patients, lo but THIS is not contraindicated in cirrhosis with no coexistent PHT. Lesser hepatic resections are 354

THE AMERICAN

JOURNAL

OF SURGERY@

VOLUME

171

indicated wherever possible ,26but in our series the usual high incidence of advanced stages of primary malignant tumors at presentation precluded application of this philosophy. Hepatic-only secondary malignant tumors were the most common reasons for hepatic resection, and CRCS was the most prevalent diagnosis. From single-institution33*34 and multi-institution studies, 35it is clear that a certain subset of patients with liver-only CRCS can be cured by surgical resection. Whether biological selection or surgery determines long-term survival remains unresolved.36 In. our series, the mortality for resection without PHT is l.l%, and the 5-year survival to date is 41.6% on a standard life table, although the follow-up remains short. Only further multi-institution studies will answer the question of the absolute indications for hepatic-only CRCS resection. Benign disease should only be resected if it is complicated, cannot be treated by conservative measures, or cannot be distinguished from a potentially malignant lesion. The majority of this group required segmentectomies, the morbidity was low and the mortality zero. This confirrm that rigid conservatism should no longer be practiced and that hepatic resection is justified in this gro~p.~’ Precise delineation of anatomy, together wlith use of THIS and scalpel division of hepatic parenchyma, permits segmental and formal hepatic resections to be undertaken with minimal morbidity, mortality, or transfusion requirement in patients without portal hypertension. Furthermore, swift resection and en bloc resection of vascular and biliary structures are facilitated. The authors describe their experience using total vascdar occlusion and supruceliac au& cLrmping in liver resection. Their resuh confirm that this is a very safe methodof hepati resection that muintains kmodynumic stability without an apparent increase in liver-relatedmorbidity. Tk growing confidence of tk surgicalcornmu& in dealingwith complexliver lesions hasbeen a significant advance fursuchpatients.

REFERENCES 1. Couinaud C. Le Masson; 1957.

Foie: Et&es Anatomiques

2. Ton That Tung.Chirurgk

et Chkrgicaks.

Paris:

d’Exerese du Foil. Park: Masson; 1962. 3. Bismuth H, Houssin D, Castaing D. Major and minor segmentectomies ‘Reglees” in liver surgery. World J Surg. 1982;6: 10-24. 4. Pringle JH. Notes on the arrest of hepatic haemorrhage due to trauma. Ann Swg. 1908;48:541-549. 5. Heaney JP, Stanton WK, Halbert DS, et al. An improved technique for vascular isolation of the liver: an experimental study and case reports. Ann Surg. 1963;163:237-241. 6. Former JG, Shia MH, Kinne DW, et al. Major hepatic resection using vascular isolation and hypothermic perfkion. Ann Surg. 1974;180:644-652. 7. Huguet C, Nordlinger B, Galopin JJ, et al. Normothennic hepatic vascular exclusion for extensive hepatectomy. Surg Gynecol Obstet. 1978;147:689-693. 8. Bismuth H, Castaing D, Garden DJ. Major hepatic resection under total vascular exclusion. Am J Surg. 1989;210:13-19. 9. Emre S, Schwartz MD, Katz E, Miller MD. Liver resection under total vmular isolation. Variations on a theme. .Ann Surg. 1992; 217:11-19. 10. Huguet C, Vacher B, Delva E, et al. L’Hepatectomie pour tumeur sous exclusion vasculaire. Evolution des idees sur une decade. A propos d’une experience de 41 cas. Ckwgie. 1983;109:146-151.

MARCH

1996

( HEPATIC

11. Ryan JA, Faulkner DJ. Liver resection without blood transfusion. Am 1 Surg. 1989;157:472-475. 12. Delva E, Camus Y, Nordlinger B, et al. Vascular occlusions for liver resections. Ann Surg. 1989;209:211-218. 13. Delva E, Barberousse JP, Nordlinger B, et al. Haemodynamic and biochemical monitoring during major liver resection with use of hepatic vascular exclusion. Surgery. 1984;95:309-318. 14. Stephen MS, Sheil AGR, Thompson JF, et al. Aortic occlusion and vascular isolation allowing avascular hepatic resection. Arch Surg. 1990;125:1482-1485. 15. Solomon MJ, Stephen MS, White GH, Eyers AA. A new classlfication of hepatic territories using intraoperative ultrasound. Am J Surg. 1992;163:336-338. 16. Launois B, Jamieson GG. The importance of Glisson’s capsule and its sheaths in the intrahepatic approach to resection of the liver. Surg Gynecol Obsret. 1992;174:155-158. 17. Launois B, Jamieson GG, Corbel L, Campion J-P. The intraheptic Glissonian approach to liver resection. Surg Gynecol Obstet. 1992;174:7-10. 18. Opelz G, Terasaki PI. Dominant effect of transfusions on kidney graft survival. Transplantation. 1980;29:153-158. 19. Blumberg N, Heal JM, Murphy P, et al. Association between transfusion of whole blood and recurrence of cancer. BMJ. 1986;293: 530-533. 20. Stephenson KR, Steinberg SM, Hughes KS, et al. Perioperative blood transfusions are associated with decreased time to recurrence and decreased survival after resection of colorectal liver metastases. Ann Surg. 1988;208:679-687. 21. Miller CM, Schwartz ME, Nishizaki T. Combined hepatic and vena caval resection with autogenous caval graft replacement. Arch Surg. 1991;126:106-108. 22. Nakamura S, Sakaguchi S, Kitazawa T, et al. Hepatic vein reconstruction for preserving remnant liver function. Arch Surg. 1990;125:1455-1459. 23. Huguet C, Nordlinger B, Block P, Conrad J. Tolerance of the human liver to prolonged normothermic ischaemia. Arch Surg. 1978;113: 144&1451.

THE AMERICAN

RESECTION/STEPHEN

ET Al

24. Huguet C, Gavelli A, Chicco A, et al. Liver isch,aemia for liver resection. Where is the limit? Surgey. 1992;111:251-259. 25. Kahn D, Hickman R, Dent DM, Terblanche J. For how Long can the liver tolerate ischaemia? Eur Surg Res. 1986;18:277-282. 26. Bismuth H, Houssin D, Mazmanian G. Postoperative liver insufficiency: prevention and management. World J Surg. 1983;7:505-510. 27. Child CC. The Liver and Portal Hypertension. Philadelphia: WB Saunders; 1964. 28. Franc0 D, Capussotti L, Smadji C, et $1. Resection of hepatocellular carcinomas: results in 72 European patients with cirrhosis. Gusaoenteroiogy. 1990;98:733-738. 29. Nagao T, Inoue S, Goto S, et al. Hepatic resection for hepatocellular carcinoma: clinical features and long term prognosis. Ann Surg. 1987;205:33-40. 30. Mori K, Ozawa K, Yamamoto Y, et al. Response of hepatic mitochondrial redox tolerance test as a new prediction of surgical risk in hepatectomy. Ann Surg. 1990;211:43%446. 3 1. Mimura H, Takakura N, Ohno Y, et al. Determination of the extent of feasible hepatic resection from hepatic blood flow. WuddJ Surg. 1986;10:302-310. 32. Kanematsu T, Takenaka K, Furata T, et al. Acute portal hypertension associated with liver resection: analysis of early post operative death. Arch Surg. 1985;120:1303-1305. 33. Adson MA, Van Heerden JA, Adson MH, et al. Resection of hepatic metastases from colorectal cancer. Arch Surg. 1984;119:647-651. 34. Former JG, Silva JS, Golbey RB, et al. Multivariate analysis of a personal series of 247 consecutive patients with liver nnetastases from colorectal cancer. Ann Surg. 1984;199:306-316. 35. Steele G, Bleday R, Mayer RJ, et al. A prospective evaluation of hepatic resection for colorectal carcinoma metastases to the liver: gastrointestinal tumor study group protocol 6584. J Cfin Oncol. 1991;9: 1105-1112. 36. Cady B, Stone MD, McDermott WV, et al. Technical and biological factors in disease-free survival after hepatic resection for colorectal cancer metastases. Arch Surg. 1992;127:561-569. 37. Iwatsuki S, Starzl TE. Personal experience with 4.11 hepatic resections. Ann Surg. 1988;208:421-434.

JOURNAL

OF SIJRGERYm

VOLUME

171

MAR’CH

1996

355