Overall survival (the primary end point) and radiologically confirmed progression-free survival were analyzed in the intention-to-treat population with the use of a two-sided, stratified log-rank test, with the hazard ratio and two-sided 95% confidence intervals based on a stratified Cox model and the associated Kaplan–Meier survival
estimates. The median follow-up time for survival was calculated by means of the reverse Kaplan–Meier method. Rates of objective response and disease control were compared with the use of Fisher’s exact test in the subgroup of the intention-to-treat population that had measurable disease at baseline. Adverse events and laboratory abnormalities were summarized for all patients who received at least one dose of study drug. Time to worsening of ECOG performance status was analyzed with the same methods used to assess overall survival. All subgroup analyses, as well as the time to worsening ECOG performance status, were prespecified in the protocol or statistical analysis plan before the data were unblinded. Multivariate Cox regression analysis was performed to examine the effect of all prespecified factors (prognostic and predictive) on the overall survival effect of TAS-102.