LIVER RESECTION FOR COLORECTAL METASTASES

Purpose: More than 50,000 patients in the United States will present e ach year with liver metastases from colorectal cancers. The current st udy was performed to determine if liver resection for colorectal metas tases is safe and effective and to evaluate predictors of outcome. Mat erials and Methods: Data for 456 consecutive resections July 1985 and December 1991 in a tertiary referral center were analyzed. Results: Th e perioperative mortality rate was 2.8%, with ct mortality rate of 4.6 % for resections that involved a lobectomy or more. The median hospita l stay was 12 days and only 9% of patients were admitted to the intens ive care unit. The 5-year survival rate is 38%, with a median survival duration of 46 months. By univariate analysis, nodal status of the pr imary lesion, short disease-free interval before detection of liver me tastases, carcinoembryonic antigen (CEA) level greater than 200 ng/mL, multiple liver tumors, extrahepatic disease, large tumors, or positiv e resection margin was predictive of poorer outcome. Sex, age greater than 70 years, site of primary tumor, or perioperative transfusion was not predictive of outcome. By multivariate analysis, positive margin, size greater than 10 cm, disease-free interval less than 12 months, m ultiple tumors, and extrahepatic disease were independent predictors o f poorer outcome. Short disease-free interval or multiple tumors were nevertheless associated with a 5-year survival rate greater than 24%. Conclusion: Liver resection for colorectal metastases is safe and effe ctive therapy and currently represents the only potentially curative t herapy for metastatic colorectal cancer. The only absolute contraindic ation to resection is extrahepatic disease. A randomized trial to exam ine efficacy of surgical resection cannot ethically be performed. Live r resection should be considered standard therapy for all fit patients with colorectal metastases isolated to the liver. (C) 1997 by America n Society of Clinical Oncology.