(699d) Data-Driven Modeling Enhances Oncolytic Adenovirus Therapy

Oncolytic adenoviruses exploit the lytic property of viral replication to specifically target and kill cancer cells. The success of such treatment is fundamentally based on the ability to deliver virus to the targeted cells. Many cancer cell lines, however, down-regulate the Coxsackievirus and Adenovirus Receptor (CAR), rendering this population less vulnerable to viral attack. Experimental data demonstrates that disruption of signaling through the RAF-MEK-ERK pathway by inhibition of MEK can up-regulate CAR expression, offering enhanced adenovirus entry into cells. This pharmacological restoration of CAR expression, however, interferes with the replication of ONYX-015 since inhibition of MEK results in G1 cell cycle arrest. While drug mediated up-regulation of CAR might improve virus entry, the consequent cell cycle arrest inhibits viral replication. Enhancement of this combinatorial approach is difficult since the effects of MEK inhibitors, as well as the interaction of adenoviruses with target cells, are highly complex, dynamic, and non-linear processes. We address these complexities through development of a data-driven, mechanistic model that characterizes and predicts the impact of MEK inhibition on cancer cell proliferation, infection, and adenovirus-induced cell death. An ordinary differential equation model is fit to training data and analyzed to predict therapeutic efficiency for a variety of treatment conditions. Simulations suggest that it is not therapeutically optimal to treat cell prior to infection; simultaneous- or post-treatment with MEK-inhibitor is most effective. These predictions were experimentally validated, confirming the utility of computational modeling toward the advancement of combinatorial therapies. Conclusions drawn from this research have the potential to dramatically improve the utility and efficacy of non-surgical cancer treatment, especially in locally advanced or metastatic cases where treatment options remain limited.