PUBLICALLY ACCESSIBLE SIMULATIONS TO EXPLORE THE COST-EFFECTIVENESS OF PHARMACOGENOMICS IMPLEMENTATIONS

Purpose: Genomic medicine programs, such as those implementing pharmacogenomics, need to be sensitive to local conditions, such as minor allele frequencies of targeted gene variants, attributes of the genotyped population, and prevalence of drug exposures or disease states.  A publically accessible simulation, with user defined inputs, could provide the flexibility to evaluate such programs operating within different environments, and with different racial and ethnic admixtures.  

Method: We developed an open source discrete event simulation framework for modeling the cost-effectiveness of genotype tailored therapy based upon a academic medical center based pharmacogenomic program featuring both preemptive and reactive multiplexed testing.   The framework was based on existing discrete event simulations created within a commercial software package. To demonstrate utility for researchers and implementers of precision medicine, we applied the framework to SLC01B1 tailored simvastatin therapy with the purpose of creating a public web application allowing user manipulation of simulation assumptions.

Result: Open source statistical software (R software) was used to replicate the simulation structure generated by a commercial discrete event simulation software package.   The R package, “simmer”, provided much of the core functionality, allowing linkages to powerful simulation routines in C++ and a user-driven web application package (R Shiny).  The application was successfully hosted on an internal web site and will be released publically when model validation is complete.  On-screen manipulation of key parameters allows users to re-run the simulation with local assumptions and facilitates robust, on-demand sensitivity and value of information analyses.  Performance on a basic web server was acceptable, but may be require more robust computational support for complex simulations.  Similar work to host public simulations of genotype guided warfarin and clopidogrel therapy are underway.

Conclusion: Publically accessible simulations of pharmacogenomic scenarios address several challenges that arise in health economic evaluations of genomic medicine and can support user adaptation of the simulation to local needs.