Aurore Carrot (1) ; Olivier Colomban (1) ; Michele Lamuraglia (2) ; Christophe Sajous (3) ; Pauline Corbaux (3) ; Aymeric Favre (3) ; Benoit You (1)(3) ; Gilles Freyer (3)
(1) EA3738 CICLY, UCBL - HCL Faculté de Médecine Lyon-Sud, Université Claude Bernard Lyon 1, univ Lyon, Oullins, France (2) Oncologie médicale, Institut de Cancérologie du CHU de St Etienne, France (3) Institut de cancérologie des Hospices Civils de Lyon (IC-HCL), Oncologie médicale, CITOHL, Lyon, France
Objectives:
In patients with recurrent advanced ovarian cancer (AOC), the development of combined immunotherapy + anti-angiogenic + PARP inhibitor agents offers the opportunity to assess chemotherapy-free regimens. The combination of durvalumab, olaparib and bevacizumab was associated with promising efficacy and safety antitumor effects in MEDIOLA and BOLD trials [1].
The modelling of the kinetics of the serum tumor marker CA-125 during platinum-based chemotherapy using a nonlinear mixed-effects Kinetic-Pharmacodynamics (K-PD) model was shown to be relevant. Indeed, the modeled ELIMination rate contant K (KELIM) is a reproducible indicator of the tumor chemosensitivity [2]. The investigation of the CA-125 kinetics in BOLD trial showed that KELIM adjusted to durvalumab + olaparib + bevacizumab was associated with higher overall survival (manuscript submitted to JCO).
Previous studies showed that CA-125 kinetics and tumor size changes were related [3].
The objective of the present study, performed with the BOLD trial data, was to describe the tumor size changes and the kinetics of CA-125 in a joint K-PD model, in order to determine if adding the radiological assessment to the CA-125 kinetics model enabled to increase the model accuracy.
Methods:
Data: The data came from the phase II clinical trial BOLD (NCT04015739) [1], evaluating the safety and efficacy of the combination of olaparib, bevacizumab and durvalumab in 74 patients with recurrent high-grade AOC Patients with at least three CA-125 titters and two tumor size measurements (sum of the longest diameters according to RECIST criteria) were selected.
Models: Three K-PD models were built to fit 1) CA-125 kinetics; 2) tumor growth inhibition; and 3) the relationships between tumor growth inhibition and CA-125 kinetics during the first 100 days of treatment. CA-125 and/or tumor size were inhibited by the treatment through an indirect effect model. Models validation criteria were the Objective function value, Relative Standard Errors (RSE), shrinkage values, classical Goodness-Of-Fits (GOF) plots (plots of observations vs predictions, NPDE (Normalized Prediction Distribution Errors)) and Visual Predictive Checks (VPC) using NONMEM 7.5.0.
Results:
Out of 74 enrolled patients, 50 were selected. The three models were composed by two compartment models: a central compartment C1 receiving treatment dosing and a transit compartment C2. Two parameters were estimated to define the treatment effects: treatment kinetic rate constant (K) and treatment concentration to obtain 50% of maximal effect (EC50). The tumor growth inhibition model contained a third compartment for tumor size variation, defined by a tumor proliferation rate (KPROL) and a tumor reduction constant rate (KREDUC). The CA-125 kinetic model had a third compartment for CA-125 concentrations, with a CA-125 production rate (KPROD) and a CA-125 elimination constant rate (KELIM). The combined model of tumor growth inhibition and CA-125 concentrations was previously described [3]. The platinum-sensitivity status (platinum–sensitive relapse (PSR) or platinum-resistant relapse (PRR)) was used as a baseline covariate impacting the CA-125 kinetics. CA-125 and tumor size values were log-transformed. Lesions diameters below quantification limit (LOQ = 10mm) were set to LOQ/2. Initial Tumor size (TS0E) and CA-125 value (CA1250E) were also estimated. Two parameters were used to describe the CA-125 production in the joint model, with a baseline production (KPROD1) and a production induced by the tumor (KPROD2).
Most of RSE were < 50%, except those for K and EC50 in the three models (< 100%) and for the baseline CA-125 production (KPROD1) (< 130%) in the joint model. High RSE value was explained by the lack of treatment concentrations information, the low number of patients and the high inter-individual variability. GOF plots and VPC suggested good predictions of CA-125 kinetics and tumor growth inhibition in all models, and enabled to explain these variations with a unique model.
Conclusions:
These models were effective to characterize individually the CA-125 kinetics and tumor growth inhibition in AOC patients during the first hundred days of treatment administration. Combination of tumor size and CA-125 kinetics in a unique model links CA-125 kinetics and tumor size variation, with a more complex and precisely modelling. The prognostic value regarding PFS and OS of parameters extracted from these models will be investigated.
References:
[1] Freyer, G., et al., 733P Bevacizumab (Bev), olaparib (Ola) and durvalumab (Durva) in patients with recurrent advanced ovarian cancer (AOC): The GINECO BOLD study. Annals of Oncology, 2021. 32: p. S734-S735.
[2] You, B., et al., The role of the tumor primary chemosensitivity relative to the success of the medical-surgical management in patients with advanced ovarian carcinomas. Cancer Treat Rev, 2021. 100: p. 102294.
[3] Wilbaux, M., et al., Prediction of tumour response induced by chemotherapy using modelling of CA-125 kinetics in recurrent ovarian cancer patients. Br J Cancer, 2014. 110(6): p. 1517-24.
Reference: PAGE 30 (2022) Abstr 10120 [www.page-meeting.org/?abstract=10120]
Poster: Drug/Disease Modelling - Oncology