A quantitative modelling framework to inform dose selection of Xentuzumab, a dual insulin-like growth factor-I/II neutralizing antibody in cancer patients
Zinnia P Parra-Guillen (1,2), Alvaro Janda (1,2,#), Ulrike Schmid (3), Matthias Freiwald (3), Iñaki F. Troconiz (1,2)
(1) Pharmacometrics & Systems Pharmacology Research Unit, Department of Pharmacy and Pharmaceutical Technology, School of Pharmacy and Nutrition, University of Navarra, Pamplona, Spain. (2) IdiSNA, Navarra Institute for Health Research, Pamplona, Spain. (3) Translational Medicine and Clinical Pharmacology, Boehringer Ingelheim Pharma, Germany. # Current affiliation: School of Engineering, University of Edinburgh, United Kingdom
Objectives: Over the past decade, the insulin-like growth factor (IGF)-signalling pathway has gained substantial interest as a potential new therapeutic target in oncology [1]. Xentuzumab is a humanised IgG1 monoclonal antibody (mAb) binding both IGF-I and IGF-II and thereby inhibiting downstream signalling essential for survival and tumour growth. This pathway is further regulated via binding of IGFs to circulating binding proteins (BPs). The aim of the current work was to develop a mechanistic model characterising the dynamics and interactions of IGF-I, IGF-II, BPs and Xentuzumab to guide dose selection during clinical development.
Methods: To build and validate the mathematical model, in house in vitro studies, literature and clinical data were used. Time courses of IGF-I (total and free) and their main binding protein BP-3 (total and bound) in plasma were obtained from literature [2-4]. Clinical data came from two phase I studies, where Xentuzumab doses of 10-1800 mg weekly or 10-3600 mg every 3 weeks were administered to advanced solid cancer patients (n=125). Within these trials, total IGF-I, IGF-II, BP-3 and Xentuzumab plasma concentrations were measured over time. Part of the clinical data from the higher doses was saved for model validation (n=50 for pharmacokinetics and n = 59 for pharmacodynamics). Model development was performed in three steps: (i) population pharmacokinetic analysis of Xentuzumab, (ii) modelling of the biological system representing endogenous IGF-I, IGF-II and BPs in the absence of drug accounting for synthesis, degradation and binding processes of the free and bound entities, and (iii) integration of the two previous models to characterise IGFs and BPs dynamics during Xentuzumab treatment. Quasi steady-state equilibrium was assumed to compute baseline conditions. A molar excess of BPs over measured BP-3 (characterised by the parameter FACBP) was assumed and optimised during model building. Model calibration was undertaken by comparing model predictions with real data. The final model was used to predict neutralization of free IGF levels at trough steady-state (t=6 weeks) for different Xentuzumab doses and dosing schedules to support Phase II dose selection. A sensitivity analysis evaluated the impact of final model parameters on the predicted neutralization of free IGFs for the selected dosing regimen. Analyses were performed in NONMEM 7.3 and MATLAB.
Results: An adequate description of all evaluated scenarios from literature and clinical data (including training and validation set) was obtained with the final model. Simulations showed that in order to achieve > 90% neutralization of free IGF-I at trough steady-state, a 1000 mg weekly dose was needed. Simultaneously, >64% neutralization of free IGF-II was estimated (higher uncertainty due to limited knowledge on free IGF-II from literature). Interestingly, equivalent total doses administered less frequently (i.e. 2000 mg every 2 weeks or 3000 mg every 3 weeks) resulted in lower IGF neutralization compared to weekly administration, suggesting that over-proportional doses would be needed to obtain the same effect. During the sensitivity analysis, FACBP appeared to be the most influential parameter, as a drop of only 10% translated into more than double levels of free IGF at steady state. However, when looking at the % of IGF neutralization, FACBP values ranging from 1.1 to 1.5 (plausible interval) translated into a minor change in the predicted inhibition of free IGF-I and IGF-II at steady state of 88-91% and 56-66%, respectively, suggesting model robustness.
Conclusions: A mechanistic model solving multiple protein interactions has been developed to characterise the disposition of IGF-I, IGF-II and BPs in the absence and presence of Xentuzumab. The quantitative framework allowed us to predict the time-course of unmeasured markers in Xentuzumab treated cancer patients, particularly plasma concentrations of free IGF-I and IGF-II as ultimate drug targets. The Xentuzumab dose of 1000 mg/week was finally selected as recommended Phase II dose supported by the model predictions of % free IGF neutralization, illustrating the utility of systems pharmacology type models beyond target identification.
References:
[1] Pollak M. Nat Rev Can (2012) 12:159-169.
[2] Laron et al. Acta Endocrinologica (1990)123:378-382.
[3] Boroujerdi et al. Am J Physiol (1997) 273:E438-E447
[4] Mizuno et al. Pharm Res (2001) 18:1203-1209.