Examining the controllability of sepsis using genetic algorithms on an agent-based model of systemic inflammation

The developmental pipeline for novel therapeutics to treat sepsis has diminished to a trickle compared to previous years of sepsis research. While enormous strides have been made in understanding the basic molecular mechanisms that underlie the pathophysiology of sepsis, a long list of novel agents have now been tested in clinical trials without a single immunomodulating therapy showing consistent benefit. The only antisepsis agent to successfully complete a phase III clinical trial was human recumbent activated protein C. This drug was taken off the market after a follow-up placebo-controlled trial (human recombinant activated Protein C Worldwide Evaluation of Severe Sepsis and septic Shock [PROWESS SHOCK]) failed to replicate the favorable results of the initial registration trial performed ten years earlier. We must critically reevaluate our basic approach to the preclinical and clinical evaluation of new sepsis therapies.

To introduce a form of mathematical modeling, agent-based modeling (ABM), and demonstrate its potential uses in the evaluation of the dynamics of the innate immune response (IIR) and the development of possible treatments for systemic inflammatory response syndrome (SIRS)/multiple organ failure (MOF). The IIR can be categorized as a complex system that responds to interventions in a nonintuitive fashion, leading to difficulty in translating basic science knowledge into effective treatments for SIRS/MOF. It is proposed that ABM is particularly well suited to examining the complex interactions of the IIR and its disordered states of SIRS/MOF. Computer simulation and mathematical modeling. Review articles on components and mechanisms involved in the IIR. Published results from phase III anticytokine/mediator trials. Published results from smaller clinical trials and animal studies. An abstract ABM of the IIR was created. The model reproduces the general behavior of the IIR with respect to outcome and cause of system "death." Patterns of levels of individual cytokines matched patterns of measured cytokines reported in the existing literature. Clinical trials of anticytokine therapy were simulated and produced outcomes qualitatively similar to those reported in the literature. A series of hypothetical treatment regimes (variation of dose and length of treatment [anti-tumor necrosis factor and anti-interleukin-1], anti-CD-18, and multiple-drug regimes [combination of anti-tumor necrosis factor, anti-interleukin-1, and anti-CD-18]) were formulated and implemented in the ABM. None of the simulated therapies showed a statistically significant improvement in system mortality. Presented herein is an abstracted ABM of the IIR. This model is intended primarily as an introduction to and demonstration of this technique. However, even this relatively simple model demonstrates counterintuitive system responses and the difficulty of effectively manipulating a complex system like the IIR. ABM may provide a synthetic, analytical platform to integrate basic science data on the IIR, thus eventually aiding in formulating and testing future mediator-directed therapies for SIRS/MOF before clinical trials, and it may provide insights into directions of future research.