Randomized trial of surgery versus surgery followed by adjuvant hepatic arterial infusion with 5-fluorouracil and folinic acid for liver metastases of colorectal cancer

Objective To determine the impact of adjuvant hepatic arterial infusion (HAI) on survival relative to resection alone in patients with radical resection of colorectal liver metastases. Summary Background Data Nearly 40% to 50% of all patients with colorectal carcinoma develop liver metastases. Curative resection results in a 5-year survival rate of 25% to 30%. Intrahepatic recurrence occurs after a median of 9 to 12 months in up to 60% of patients. The authors hypothesized that adjuvant intraarterial infusion of 5-fluorouracil (5-FU) might decrease the rate of intrahepatic recurrence and improve survival in patients with radical resection of colorectal liver metastases. Methods Between April 5, 1991, and December 31, 1996, patients with colorectal liver metastases from 26 hospitals were stratified by the number of metastases and the site of the primary tumor and randomized to resection of the liver metastases followed by adjuvant HAI of 5-FU (1000 mg/m(2) per day for 5 days as a continuous 24-hour infusion) plus folinic acid (200 mg/m(2) per day for 5 days as a short infusion), or liver resection only. Results The first planned intention-to-treat interim analysis after inclusion of 226 patients and 91 events (deaths) showed a median survival of 34.5 months for patients with adjuvant therapy versus 40.8 months for control patients. The median time to progression was 14.2 months for the chemotherapy group versus 13.7 months for the central group. Grade 3 and 4 toxicities (World Health Organization), mainly stomatitis (57.6%) and nausea (55.4%), occurred in 25.6% of cycles and 62.9% of patients. Conclusion According to this planned interim analysis, adjuvant HAI, when used in this dose and schedule in patients with resection of colorectal liver metastases, reduced the risk of death at best by 15%, but at worst the risk of death was doubled. Thus, the chance of detecting an expected 50% improvement in survival by the use of HAI was only 5%. Patient accrual was therefore terminated.