Two prognostic populations of ovarian cancer patients defined by CA125 modeled kinetic parameter KELIM (AGO-OVAR 7 & 9; ICON 7 AGO/GINECO/ MRC CTU/GCIG trials).

Abstract

5554 Background: Longitudinal CA125 kinetics during treatment can be assessed by mathematical models. The modeled CA125 elimination parameter KELIM (with 3-4 timepoints) was an early prognostic factor in CALYPSO (Gynecol Oncol 2013). Validating the prognostic value of KELIM in phase 3 trial datasets with different 1st line treatments was warranted. Methods: Data from AGO-OVAR 9 (training set: carboplatin-paclitaxel (CP) +/- gemcitabine; n=1742); AGO-OVAR 7 (validation set: CP +/- topotecan; n=1308); and ICON 7 trials (validation set: CP +/- bevacizumab; n=1528) were analyzed. CA125 profiles were fit to nonlinear mixed effect equations, and KELIM estimated in all patients at 100 days. KELIM prognostic value was tested regarding PFS and OS against other prognostic factors (stage; pathology; grade; arms; GCIG CA125 criteria; Oza groups in ICON 7) using univariate/multivariate tests. Results: KELIM (≤ or > median 0.0598) had reproducible independent prognostic value for PFS (AGO 9: HR = 0.60 [0.53-0.69]; AGO 7: HR = 0.58 [0.40-0.83]; ICON-7: HR=0.65 [0.44-0.96]) and for OS (AGO 9: HR = 0.55 [0.47-0.64]; AGO 7: HR = 0.55 [0.35-0.86]; ICON-7: HR=0.49 [0.41-0.57]) by multivariate tests. Other significant factors for PFS & OS: stage IV in AGO7 & 9 (HR=6.0 to 8.3) and ICON 7 poor progn group (PFS HR=2.24 PFS; OS HR=2.22 [1.9-2.6]). KELIM prognostic value was independent on regimen arms (Table), maintained within ICON7 progn groups (best OS gain with bevac if poor progn & unfav KELIM), and better than GCIG CA125 (inconsistent progn value). Conclusions: The reproducible & independent early prognostic value of KELIM regarding PFS and OS is validated. Easily calculable online, it early discriminates 2 ovarian cancer populations for PFS & OS whatever treatments, and is a new reference prognostic factor. [Table: see text]