Taking into account successive treatment lines in the analysis of a colorectal cancer randomised trial.

The FFCD 2000-05 randomised trial included 410 patients with advanced colorectal cancer and compared a sequential arm S treated with 5-fluorouracil and leucovorin (LV5FU2) followed by FOLFOX (LV5FU2+oxaliplatin) and then FOLFIRI (LV5FU2+irinotecan) and a combination arm C that begins directly with FOLFOX followed by FOLFIRI. The first aim of this study was to analyse the prognostic effects on overall survival of disease progression, and of toxicities under first-line therapy. We also studied the benefit of introducing irinotecan in each arm. Finally, we compared the effect of treatment on repeated progression and toxicities. For this purpose, we used Cox regression models with time-dependent variables and shared gamma frailty regression models. We found that early on during follow-up, the prognostic effect on survival of progression under first-line therapy was greater in C (hazard ratio [HR]=18.0 [7.9-41.2]) than in S (HR=7.7 [3.9-17.4]). This difference was significant, but it decreased over time. The prognostic effect of severe toxicities was greater in S (HR=2.0 [1.4-2.9]) than in C (HR=1.3 [0.9-1.9]). Introducing irinotecan was significantly more beneficial in S (HR=0.2 [0.1-0.4]) than in C (HR=0.3 [0.2-1.5]). The risk of repeated progression was not significantly different between the two groups (HR=0.9 [0.8-1.1]) whereas the risk of toxicities was greater in C (HR=1.7 [1.4-2.1]). Overall, this study suggests that starting with less toxic first-line treatment is a valid option since it does not exert a deleterious effect on the risk of overall progression or death.