Survival impact of chemotherapy in patients with colorectal metastases confined to the liver: A re-analysis of 1458 non-operable patients randomised in 22 trials and 4 meta-analyses

Background: Metastases confined to the liver is a frequent situation in patients with advanced colorectal cancer. For non-operable patients, 5-FU-based chemotherapy is often proposed but the importance of the choice of first line 5-FU regimen remains debatable. Design: In four previously performed meta-analyses, our group had compared bolus intravenous fluoropyrimidines (bolus FU group) with experimental fluoropyrimidines (experimental FU group), consisting of 5-FU plus leucovorin, 5-FU plus methotrexate, continuous infusion 5-FU, or hepatic-artery infusion FUDR. We re-analysed this data set to focus on 1458 patients with non-operable colorectal metastases confined to the liver, randomised in 22 trials. All analyses were stratified by trial and used individual patient data. Results: Median survival times were 11.3 months in the bolus FU group (95% CI: 10.5-12.0 months) compared to 12.7 months in the experimental FU group (95% CI: 12.0-13.1 months). This difference, although clinically small, was statistically significant, with an overall survival hazard ratio of 0.88 (95% CI: 0.79-0.99, P = 0.037). In a multivariate analysis, performance status was the only significant predictor of survival (P < 10(-4)), whereas the statistical significance of allocated treatment was borderline (P = 0.058). Conclusions: The outcome of patient with non-operable colorectal metastases confined to the liver is poor, and mainly driven by their initial performance status. Experimental chemotherapy schedules yield a small improvement in their overall survival, indicating the importance of the choice of first-line chemotherapy.