A quantitative systems pharmacology model to explore combination efficacy of immuno-oncology compounds: Effects of CXCR2 and PD-1 inhibitions
Oleg Stepanov, Veronika Voronova, Yuri Kosinsky, Kirill Peskov, Eric Masson, Gabriel Helmlinger
M&S Decisions
Objectives: Targeting of the tumor microenvironment and of immune cell-dependent mechanisms is a promising emerging approach for the treatment of different cancers [1]. As monotherapies in a preclinical setting, neither anti-CXCR2 induced decrease in myeloid-derived suppressive cells (MDSC) influx into tumor tissue, nor T-cell checkpoint inhibition via an anti- PD1 antibody significantly affect tumor growth; however, their combination results in impressive tumor growth inhibition [2]. The main aim of this study was to provide mechanistic hypotheses relative to the synergistic effects of combined anti-CXCR2/anti-PD1 treatment, using quantitative systems pharmacology (QSP) modeling. A related goal was to identify system parameters driving inter-animal variability in the observed tumor dynamics responses.
Methods: A QSP model was built upon a set of ordinary differential equations. Tumor dynamics was described by a logistic equation and was directly driven by the time-dependent number of activated cytotoxic lymphocytes (CTL). In the model, tumor growth itself promoted MDSC migration into tumor tissue, a mechanism which may directly inhibit CTL proliferation. Activated CTLs would also negatively affect their own activation via a PD1-dependent negative feedback mechanism. The dynamics of CXCR2 and PD1 antibodies were described using, respectively, one-compartment and two-compartment PK models. The effects of CXCR2 blockade on MDSC migration and of PD1 blockade on free PD1 levels were described using Imax equations. Qualification of the QSP model was based on various data published in the literature [2].
Results: The proposed model adequately described tumor dynamics and CTL levels in control experiments, in CXCR2 and PD1 monotherapies, and in their combination. A local sensitivity analysis of the model demonstrated that inter-animal variability in response to treatment depended mainly on parameters describing the dynamics of active CTLs and their infiltration into tumor tissue. It was also shown that, even for modest changes in these specific parameter values, the output response (tumor dynamics) could be switched from unaffected tumor growth to full tumor growth inhibition or regression. Such quantitative insights may form the basis for the elucidation of patient response vs. no response, as observed in recent immuno-oncology clinical trials [REF].
Conclusions: A preclinical QSP model of the immuno-oncology cycle, which includes mechanisms of action for CXCR2 and PD-1 inhibitions, was developed, based on experimental data published in the literature. It was shown that simultaneous targeting of these two immuno-suppressive pathways would lead to sufficient immune cell activation as well as inhibition of immuno-suppressive MDSCs in the microenvironment, with a net result of tumor growth inhibition or regression. A set of model parameters, critical for a positive treatment outcome and possibly explaining the broad heterogeneity observed in inter-individual responses, was also identified via a model-based sensitivity analysis.
References:
[1] Chen DS and Mellman I. Oncology Meets Immunology:The Cancer-Immunity. Immunity, 2013, 39, 1-10.
[2] Highfill SL, Cui Y, Giles AJ. Disruption of CXCR2-mediated MDSC tumor trafficking enhancesefficacy. Sci Transl Med. 2014 May 21;6(237):237ra67
