New mathematical models predict success rate of drug trials

A large part of drug trials fail when it turns out that the results for laboratory animals do not transfer to humans (1). New mathematical models developed at Centrum Wiskunde & Informatica (CWI) are able to predict the transferability of results between different organisms.

A large part of drug trials fail when it turns out that the results for laboratory animals do not transfer to humans (1). New mathematical models developed at Centrum Wiskunde & Informatica (CWI) are able to predict the transferability of results between different organisms. This could improve the efficiency of drug research and reduce the number of laboratory animals needed in trials. The results were published this week in the journal Bioinformatics.

In a drug trial, medical researchers investigate the effect of potential new drugs by administering these to model organisms, such as mice. If an effect is found, the drug is tested on a small set of humans in phase II of the trials. Only 25% of the experiments survive this second phase, mostly because the results found in the lab animals turn out to be non-transferable to humans. Early prediction of transferability can help prevent the waste of valuable resources and reduce the number of lab animals needed in drug trials.

Conserved cell mechanisms

The mathematical model developed by the researchers at CWI identifies patterns in large networks of genes to discover cellular mechanisms that are conserved throughout the course of evolution. Similar gene activity patterns in two organisms under various circumstances indicate that a particular mechanism is active in both organisms. Such a mechanism will therefore probably be affected by drugs in the same way.  “Existing approaches are of limited use, as they only find conserved mechanisms when all active genes are conserved”, says Gunnar Klau, one of the leading researchers in this project. “Our models are more flexible. We look at similar activity patterns in a network of genes, and can find similar behaviour even if not all involved genes are known to have a common ancestor.”

By determining conserved activity patterns between a lab animal and humans, the transferability of experimental results between species can be predicted. This increases the success rate of phase II drug trials, and allows for fewer, more goal-directed experiments, thus saving lab animals.

Case study

In a case study, the researchers were able to show that the differentiation process of the recently discovered Th17 cell type, which plays an important role for the immune system, is conserved between mouse and human. These insights may be valuable for the development of novel therapeutics for various inflammatory and autoimmune diseases, such as rheumatism, psoriasis and MS.

The project is led by CWI with cooperating partners the Netherlands Cancer Institute (NKI), the University of Bordeaux and the University of Würzburg. The algorithms are open and available online.

 

More information
Full paper (pdf)


Image credit: Marlies van der Wees
(1): Source: Csermely, Péter, Tamás Korcsmáros, Huba J M Kiss, Gábor London, and Ruth Nussinov. 2013. “Structure and Dynamics of Molecular Networks: a Novel Paradigm of Drug Discovery: a Comprehensive Review..” Pharmacology & Therapeutics 138 (3): 333–408. doi:10.1016/j.pharmthera.2013.01.016.