A multiscale, mechanistic model of rheumatoid arthritis (RA) has been developed to simulate late-stage clinical trials and aid in decision-making for pharmaceutical research and development. The model captures the dynamics of cells and mediators on a physiological scale and translates these dynamics to clinical disease activity scores over a few months. This tool aims to provide a quantitative framework for understanding RA pathophysiology and predicting the impact of therapies.
Model Design and Scope
The model is designed to simulate Phase 2 and Phase 3 clinical trials, focusing on the induction and maintenance phases in patients with established RA. It incorporates a simplified representation of the disease, limited to capturing established steady-state disease of varying severity. The model does not account for the complexities of disease onset, flares, or long-term disease progression.
The primary site of inflammation in RA, the joint, is represented by a well-mixed volume of 1 mL synovium. This volume includes structural cells (fibroblast-like synoviocytes (FLS) and endothelial cells) and immune cells (macrophages, CD8 T cells, Th1 and Th17 cells, B cells, plasma cells, and CD4 Tregs). The model assumes that cell-cell interactions occur via cytokines, with disease severity approximated by cell density in the synovial volume.
Key Components and Interactions
The model includes multiple pro-inflammatory cytokines (TNF-α, IL-6, IL-17, IL-12, and IL-23) and anti-inflammatory cytokines (IL-10 and TGF-β). Chemokines and growth factors, such as RANTES, MIP-3α, MCP-1, and VEGF, are also incorporated to regulate immune cell migration to the joint. The model also includes a generalized plasma compartment that serves as a source of immune cells and a site for modeling the pharmacokinetics of therapies.
The dynamics of cells and cytokines in the synovial compartment are tracked using ordinary differential equations (ODEs). These equations capture cell proliferation, apoptosis, and migration, as well as cytokine secretion and clearance. Cytokines regulate cell life-cycle processes, and cells modulate cytokine secretion rates.
Model Parameterization and Calibration
Model parameters are derived from both "bottom-up" data (in vitro experiments) and "top-down" data (clinical, synovial RA samples). Rate constants for cellular processes (apoptosis, migration) and mediator-driven effects are determined from in vitro experiments, while steady-state cell densities and cytokine concentrations are obtained from literature data on RA patient samples. The model is calibrated to match clinical scores, such as DAS28-CRP and ACR, using empirical relationships between cell densities and disease severity.
Virtual Population (Vpop) Generation
A virtual population (Vpop) is generated by varying model parameters, including cell proliferation and migration rates, cytokine secretion rates, and cytokine effects on cell processes. This Vpop is then calibrated against clinical trial data for methotrexate (MTX), adalimumab (ADA), and tocilizumab (TCZ) to ensure that the model accurately reflects the range of patient responses observed in clinical trials. The Vpop consists of approximately 300 virtual patients (VPs) with diverse disease characteristics and treatment responses.
Sensitivity Analysis
Local and global sensitivity analyses were performed to identify key model parameters that influence the disease state. Local sensitivity analysis revealed that macrophage-related parameters are highly influential, while global sensitivity analysis highlighted the importance of FLS, macrophage, and B cell-related parameters. These analyses provide insights into the critical biological factors driving RA pathogenesis.
Therapeutic Modeling
The model incorporates the pharmacokinetics (PK) and pharmacodynamics (PD) of MTX, ADA, and TCZ. The PK of these drugs are modeled using two- or three-compartment models, while the PD is captured using binding kinetics models. MTX is modeled to affect cytokine secretion, migration, and Treg function, while ADA and TCZ are modeled to reduce free TNF-α and IL-6, respectively.
Clinical Implications
This mechanistic model of RA provides a valuable tool for simulating clinical trials and predicting the impact of therapies. By integrating physiological and clinical data, the model can help researchers and clinicians better understand RA pathogenesis and identify potential therapeutic targets. The model's ability to capture patient heterogeneity through the Vpop approach also allows for the exploration of personalized treatment strategies.
Software and Availability
The model was created and simulated in MATLAB Simbiology, with virtual cohort generation and simulations run using gQSPsim. The Vpop selection was done using a genetic algorithm developed by Vantage Research. The model is available for further research and development.
