Prediction of progression-free survival (PFS) adjusted by continuous platinum-free interval (PFI) at fixed timepoints in patients with recurrent ovarian cancer (ROC): Results from OVA-301.

Abstract

5067 Background: OVA-301, a randomized, open-label phase III study comparing trabectedin plus pegylated liposomal doxorubicin (PLD) vs. PLD alone in 672 patients with ROC, showed significant and clinically relevant longer PFS in patients treated with the combination. In ROC the efficacy of re-treatment is broadly accepted to be highly correlated with the extension of PFI. A PFI ≥ 6 months (n=430) predicts platinum sensitivity, but within this group, a PFI of 6-12 months (n=214) is considered to indicate a partially platinum-sensitive (PPS) disease. Kaplan-Meier unadjusted survival estimates are widely used for plotting time-to-event curves. Allmer and Sargent (SUGI-2003) proposed an alternative method to plotting survival estimates at a specified point in time, adjusted by a continuous explanatory variable that may be of interest. Methods: In this study adjusted estimates of PFS by continuous PFI at fixed timepoints (3, 6 and 12 months) according to Allmer and Sargent method have been considered. Results: Trabectedin and PLD provided a 21% risk reduction for disease progression or death compared with PLD (HR = 0.79; p = 0.0190), with median PFS of 7.3 months for the combination and 5.8 months for PLD (Monk et al., 2010). With the adjustment by PFI, and PFI by treatment interaction, plots representing the PFS rate prediction at 3, 6 and 12 months showed a significant difference along PFI axis favoring the combination of trabectedin and PLD. In the PPS subgroup, no PFI imbalance was observed, and the interaction term was not significant. Median adjusted estimates for main effects in PFS were 7.7 months vs. 5.5 months (adjusted HR = 0.63; p = 0.0115), again favoring the combination. Conclusions: ROC relapse outcome of subsequent re-treatment is largely influenced by the PFI extension. Therefore, it is important to adjust this variable in the analysis of PFS. In accordance with previously published estimates for PFS, the new estimation approach for presenting predicted survival data based on the continuous PFI showed an improved PFS in patients treated with the combination of trabectedin and PLD.