A modular framework for multiscale multicellular spatial modeling of viral infection, immune response and drug therapy timing and efficacy in epithelial tissues: A multiscale model of viral infection in epithelial tissues.

Development of predictive quantitative models of all aspects of COVID-19 is essential for rapidly understanding the causes of differing disease outcomes and vulnerabilities, suggesting drug and therapeutic targets, and designing optimized personalized interventions. Easy to implement, predictive multiscale modeling frameworks to integrate the wide variety of clinical and research datasets into actionable insights, which could inform therapeutic regime strategies are lacking. We present a multiscale, multicellular, spatiotemporal model of the infection of epithelial tissue by a generic virus, a simplified cellular immune response and viral and immune-induced tissue damage. Our initial model is built of modular components to allow it to be easily extended and adapted in a collaborative fashion to describe specific viral infections, tissue types and immune responses. The model allows us to define three parameter regimes: where viral infection coincides with a massive cytopathic effect, where the immune System rapidly controls the virus and where the immune System controls the virus but extensive tissue damage occurs. We use the model in a proof-of-concept application to evaluate a number of drug therapy concepts. Inhibition of viral internalization and faster immune-cell recruitment lead to containment of infection. Fast viral internalization and slower immune response lead to uncontrolled spread of infection. Simulation of a drug, whose mode of action is to reduce production of viral RNAs, shows that a relatively limited reduction of viral replication at the beginning of infection greatly decreases the total area of tissue damage and maximal viral load, while even a treatment that greatly reduces the rate of genomic replication rapidly loses efficacy as the infection progresses. A number of simulation conditions lead to stochastically variable outcomes, with some replicas clearing or controlling the virus, while others see virus-induced damage sweep the simulated lung patch. The model is open-source and modular, allowing rapid development and extension of its components by groups working in parallel.