Search

Background: Although CA125 kinetic profiles may be related with relapse risk in ovarian cancer patients treated with chemotherapy, no reliable kinetic parameters have been reported. Mathematical modeling may help describe CA125 decline dynamically and determine parameters predictive of relapse. Methods: Data from CALYPSO phase III trial data comparing 2 carboplatin-based regimens in ROC patients were analyzed. Based on population kinetic approach (Monolix software), a semi-mechanistic model was used to fit serum log (CA125) concentration-time profiles with following parameters: tumor growth rate constant (BETA); CA 125 tumor production (KIN); tumor decay rate constant (KOUT) and treatment indirect effect (Emax relationships with A and A50) “d[CA125]/dt=(KIN* exp [BETA*t]) * (1 - [A/{A+A50}]) – KOUT * (CA125)” where t is time. The predictive values of KIN; KOUT; BETA and A50 estimated during the first 50 treatment days were tested regarding progression free survival (PFS) against other reported prognostic factors using Cox-models: treatment arm; platinum-free interval (PFI), metastatic site number, largest tumor size, elevated WBC and measurable disease. Results: The CA125 kinetics from 898 patients were analyzed. Individual CA125 profiles were well fit by the model, as validated by a visual predictive check. Two modeled kinetic parameters had strong predictive values using univariate analyses: BETA (HR=1.38, p