Quantitative Systems Pharmacology (QSP) tools to aid in model development and communication: Vantage QSP Modeling Tools (VQM-Tools)
Madhav Channavazzala, Dinesh Bedathuru, Priyamvada Modak, Rukmini Kumar
Introduction: Quantitative Systems Pharmacology (QSP) models connect physiological mechanisms at the cellular and organ level, to responses at the patient and population level. Tool development in QSP is vital to 1) improve efficiencies in model development, 2) provide a framework to capture model design aspects and decisions, and 3) provide a way to communicate the multiple constraints that are incorporated as part of robust QSP model development.
- Develop tools to aid and accelerate the QSP model development
- Develop tools that explicitly assimilate model design features by visualizing modelling constraints and enable effective communication with all non-modeling stakeholders
- Show application of VQM-Tools in the development of the Vantage Rheumatoid Arthritis QSP model
Methods: "Reference Virtual Subjects" are a key milestone in QSP model development (Stage 4,) that capture baseline characteristics of patients and show an “average” (or a range of) response to perturbations (such as treatment). QSP models develop Virtual Subjects that are simultaneously consistent with 1) understanding of physiology and the interactions amongst species (summarized in model equations), 2) model rate constants being within ranges identified from basic science literature (‘bottom-up’ constraints) and 3) model dynamics spanning ranges identified from clinical literature for various perturbations (‘top-down’ constraints). In addition, modelers may need to keep track of constraints such as correlations among parameters, relationships across perturbations, maintenance of steady-state dynamics etc.
Need: Implementation of Virtual Subjects in the model requires exploring plausible parameter ranges, performing parameter sensitivity analysis and defining ranges in multi-dimensional parameter spaces that are consistent with the constraints of physiological feasibility and observed clinical behaviours.
Solution: Vantage QSP Modelling tools (VQM-Tools) have been developed (using Matlab) to:
- Define and visualize response range for Virtual Subjects: VQM-Tools provide a framework that lets the user define and visualize physiological, clinical and modeling constraints and tested repeatedly in the iterative process of creating Virtual Subjects.
- Develop Virtual Cohort: Modelers carry out systematic sensitivity analysis in VQM-Tools that helps in understanding the most and least sensitive parameters. Based on model design requirements, physiological uncertainties and inferences from sensitivity analysis, VQM-Tools are used to vary key subsets of parameters to create feasible Virtual Cohorts.
- Develop Virtual Population: Feasible parametrizations from Virtual Cohort are carried over for development into Virtual Population. Multiple approaches are used in VQM-Tools to develop VPops including those reported in previous literature [2,3]
Results: We illustrate the use of VQM-Tools for Vantage RA QSP model development. This model connects the pathophysiology of Rheumatoid Arthritis with the observed population behaviour in clinical trials using standard therapies (RA-BEAM  and RA-PREMIER ).
VQM-Tools was used to consolidate RA constraints in a single framework to visualize the effect of model parameterization on alignment with constraints. Further the results of selected Parameter Sensitivity analysis and estimation of free/under-constrained parameters for calibrating reference Virtual Subjects are shown. A Virtual Cohort that spans the range of baseline disease characteristics (in terms of cell densities, cytokine concentrations and disease scores) and response to treatment (methotrexate and anti-TNF-alpha) have been created and visualized using these tools.
Conclusions: VQM-Tools helped accelerate development of the Vantage RA model, and provided accessible visualizations of the model constraints for communication to external stakeholders from various non-modeling disciplines. Once tool development is complete, VQM-Tools will be made available for public use.
 Gadkar, K., Kirouac, D.C., Mager, D.E., van der Graaf, P.H. & Ramanujan, S. A six stage workflow for robust application of systems pharmacology. CPT Pharmacometrics Syst. Pharmacol. 5, 235–249 (2016).
 Allen RJ, Rieger TR, Musante CJ. Efficient Generation and Selection of Virtual Populations in quantitative Systems Pharmacology Models. CPT Pharmacometrics Syst Pharmacol. 2016;5(3):140-6.
 Schmidt BJ, Casey FP, Paterson T, Chan JR. Alternate virtual populations elucidate the type I interferon signature predictive of the response to rituximab in rheumatoid arthritis. BMC Bioinformatics. 2013;14:221. Published 2013 Jul 10. doi:10.1186/1471-2105-14-221
 Edward C Keystone et al, Peter C Taylor, Yoshiya Tanaka, Carol Gaich, Amy M DeLozier, Anna Dudek, Jorge Velasco Zamora, Jose Arturo Covarrubias Cobos, Terence Rooney, Stephanie de Bono, Vipin Arora, Bruno Linetzky, Michael E Weinblatt. (2017) Patient-reported outcomes from a phase 3 study of baricitinib versus placebo or adalimumab in rheumatoid arthritis: secondary analyses from the RA-BEAM study. Annals of the Rheumatic Diseases 76:11, 1853-1861.
 Breedveld FC, Weisman MH, Kavanaugh AF, Cohen SB, Pavelka K, van Vollenhoven R, Sharp J, Perez JL, Spencer-Green GT. The PREMIER study: A multicenter, randomized, double-blind clinical trial of combination therapy with adalimumab plus methotrexate versus methotrexate alone or adalimumab alone in patients with early, aggressive rheumatoid arthritis who had not had previous methotrexate treatment. Arthritis Rheum. 2006 Jan;54(1):26-37. PubMed PMID: 16385520.