Reducing Attrition In Silico

Allan Haberman


March 16, 2010 | Insights Outlook | In silico modeling and simulation (M&S) is used throughout the drug development process, from discovery through late-stage clinical trials. Pharmacodynamic/pharmacokinetic (PD/PK) models are the most widespread and important computer-based mathematical models used in drug development today. Traditionally, using animal model data to determine the starting dose or dose range for use in first-in-human clinical trials was mainly empirical. However, with the industry’s experience of late-stage failures due to suboptimal dosing, companies have increasingly moved to the use of PK/PD modeling aimed at predicting how a drug will perform in patients with respect to dose/concentration relationships and how these may affect safety and efficacy.

Several companies provide software for developing PK/PD models in the pharmaceutical industry. Pharsight’s WinNonlin is the industry standard for PK/PD analysis. In addition to its extensive library of built-in PK, PD, and PK/PD models, WinNonlin supports custom models developed by the users. In June 2009, Pharsight announced the launch of its next-generation version, Phoenix WinNonlin. GastroPlus, an ADME simulation program developed by Simulations Plus, is also used by pharmaceutical researchers in preclinical and clinical PK/PD modeling.

Modeling and Simulation at Novartis

Several large pharmaceutical companies, including Pfizer, GlaxoSmithKline, Lilly, and Novartis, have been implementing PD/PK M&S. In particular, Novartis has established a dedicated M&S department, which works with the company’s drug development therapeutic franchises. Novartis’ M&S department has a very broad scope, ranging from modeling of signal transduction pathways and safety modeling, to clinical trials, economic modeling, and decision analysis. One aspect of the department’s activities involves development of models to help move drug candidates from preclinical into first-in-man studies.

For example, Novartis researchers completed preclinical studies of a monoclonal antibody therapeutic designed to treat spinal cord injuries. However, the human spinal column has a very different geometry and fluid dynamics than spinal columns of animal models. The Novartis M&S team therefore used a biophysical modeling platform to model the human spinal cord and its surrounding tissues. The goal of this modeling effort was to describe the geometry of the spinal canal and the relevant transport properties that are applicable within that geometry, when the monoclonal antibody therapeutic is injected into the intrathecal space at a certain point in the spinal canal. (See, “Novartis Savors Early Modeling SuccessBio•IT World. May 2009.)

Entelos’ “Virtual Patients”

Entelos focuses on building dynamic, large-scale computer models of human physiology and disease. Based on its Physio-Lab technology, Entelos has developed models called “virtual patients.” These models of human disease represent the range of patients affected by a particular complex disease, including the effects of unknown genetic factors and behavioral/environmental factors that are involved in each disease. Disease-specific PhysioLab platforms include systems for diabetes, obesity, and immune/inflammatory diseases. Within each of these platforms, Entelos can create a potentially unlimited number of virtual patients, each of which represents a subpopulation of actual human patients. A virtual patient may represent a known or hypothesized cause of disease. Using simulation experiments, researchers can test marketed and experimental therapies to predict a patient’s likely response to treatment.

Entelos’ virtual patient models are being widely applied across the pharmaceutical industry. Among Entelos’ partners are Pfizer, Merck, Novartis, AstraZeneca, Bayer, Bristol-Myers Squibb, Lilly, Johnson & Johnson, and Roche.

The use of computer models can help researchers more effectively move from preclinical animal studies to human clinical trials. PK/PD models, as well as biophysical models such as those developed by Novartis and physiological models such as those developed by Entelos, can help researchers more effectively use animal model data in the design of clinical trials. In particular, they can help researchers reduce drug attrition in clinical trials due to suboptimal dosing.

However, a computer simulation is only as good as the data used to program it. There are levels of cellular and organismic regulation that have only been discovered within the last few years, and noncoding DNA (which is mainly of unknown function) makes major contributions to the genetics and biology of disease. Computer simulations are, of course, unable to deal with these unknown factors, since no one can program them into a computer model. Creating a simulation of a physiological/pathophysiological process requires knowledge of the process itself, and such data must still come from biological studies.

Further Reading: Animal Models for Therapeutic Strategies, by Allan B. Haberman, PhD, is available from Insight Pharma Reports. For more information, visit

Allan Haberman can be reached at

This article also appeared in the March-April 2010 issue of Bio-IT World Magazine.
Subscriptions are free for qualifying individuals. Apply today.
View Next Related Story
Click here to login and leave a comment.  


Add Comment

Text Only 2000 character limit

Page 1 of 1