Abstract
Background
Postoperative portal vein infusion (PVI) of 5-fluorouracil (5-FU) is a well tolerated and widely applicable treatment for colorectal cancer that might have an enormous public health impact, even if it produced survival benefits of just a few per cent. Very large trials are required to detect such differences, and the Adjuvant X-ray and 5-FU Infusion Study (AXIS) is the largest such trial yet reported.
Methods
Consenting patients with presumed colorectal cancer were randomized to surgery with or without 7 days of PVI (1 g 5-FU plus 5000 units heparin in 1 litre 5 per cent dextrose infused over each 24-h period). In addition, patients with rectal cancer could be randomized to radiotherapy or no radiotherapy to be given either before or after surgery.
Results
Between November 1989 and December 1997, 3583 patients were randomized with respect to PVI. The survival hazard ratios (95 per cent confidence interval (c.i.)) in all patients randomized and in the curatively resected subgroup (71Β·2 per cent of patients) were 1Β·00 (0Β·92 to 1Β·11) and 0Β·94 (0Β·83 to 1Β·06) respectively. Tests for heterogeneity suggested a greater treatment benefit for patients with colonic cancer than for patients with rectal cancer with respect to disease-free survival (hazard ratio 0Β·79 versus 1Β·03; P = 0Β·07), and there was a non-significant trend with respect to overall survival (hazard ratio 0Β·87 versus 1Β·03; P = 0Β·17). No survival benefit was seen in the 761 patients randomized with respect to radiotherapy; although not statistically significant, the impact on local recurrence rates was similar to that reported in the literature.
Conclusion
No overall benefit of PVI was established in AXIS when colonic and rectal cancers were considered together, but the evidence suggesting a differential treatment effect according to site of cancer in AXIS was strongly supported by a meta-analysis incorporating the previous trials. Combining the data gave hazard ratios of 0Β·82 and 1Β·00 for colonic and rectal tumours respectively (test for interaction, P = 0Β·024), equating to an absolute survival benefit for patients with colonic cancer of 5Β·8 (95 per cent c.i. 2Β·8 to 8Β·5) per cent, a level close to that seen for prolonged systemic therapy.