Comprehensive Virtual Patient Platform Implemented For Anti-Angiogentic Drugs Development
R. Ben-Av, V. Vainstein, E. Massuri, M. Shtilerman, Z. Agur
OPTIMATA Ltd. Abba Hillel 7st. Ramat Gan, Israel
Background: Over the past decades biomathematical model have evolved and their usage in actual clinical research is growing. Together with this growth there are also a growing number of computerized tools whose aim is to support a specific research area. There is a variety of tools with various user-levels controls and flexibility that exist today, including NONMEM Pharmacokinetics with extension for population analysis, MATLAB bioSIM for model building and analysis and more. For a comprehensive research in the field of drug development there is frequently a need to combine few tools. Moreover there is no standard computational tools to find an optimum between conflicting requirements, e.g. high efficacy and low toxicity of a drug.
Objectives: Our objective is to provide the researcher with a tool that will be able to suggest best treatment considering computed efficacy and toxicity. Using the concept of Virtual Patient we establish a unified platform that covers different aspects of drug development during clinical trial phases.
Methods: We have developed a unified platform with integrated features that covers all the aspects for virtual drug development. The platform integrates a simulation engine, a database engine, a parameter estimation tool and a treatment optimization tool. The simulation engine includes several Biological-SubSystems (BSS), representing pharmacokinetics as well as pathological and physiological processes affected by a drug (e.g., hematopoesis, tumor growth etc.). In this engine pharmacodynamics is realized as interaction between the PK module and the relevant BSS's. The database module contains the relevant parameters for different individuals, or for groups of individuals. The parameter estimation module performs the task of best fitting the BSS parameters to a given experimental data set. The treatment optimization module automatically searches in the space of all feasible treatment schedules for the best option - given a specified goal function. All the modules can operate in a modular way with GUI. The user can have high degree of flexibility to choose the model to be run, the optimization algorithms, the goal function etc. The output contains graphical representations for relevant dynamics and internal model parameters.
Results: We will describe our technology as well as demonstrate several applications of it in drug development, e.g. for deciphering unknown drug action mechanisms, for predicting Phase I results (eg., DLT, Dose escalation process for establishing therapeutic window) and Phase II optimal treatment regimens. The system can operate on a regular PC and can also be configured to take advantage of a heterogeneous array of computing nodes locally or over the network. Using this platform we have been able to analyze the results of unsuccessful clinical drug development case and suggest a different less toxic highly efficacious treatment schedule, which currently is in a clinical trial. Some results from the clinical trial are expected in few months.
Conclusions: A Virtual Patient system that encompasses a large spectrum of activities in drug development was built and is being used by skilled biologists. The system can significantly improve the drug development process.