Discovering drugs better: New paper by Farr scientists

The Challenge

Drug development has a low probability of success with only 4% of drug development programmes yielding licensed drugs. This can largely be attributed to unresolved system flaws whereby preclinical drug target identification and validation which occurs in cells, tissues, and animal models of disease are poor predictors of human efficacy. This means definitive evidence on the validity of a new target for the treatment of human disease comes late in drug development, during clinical (phase II or III) randomised controlled trials. Because the drug target hypothesis advanced by preclinical studies is all too frequently false, expensive, late-stage failure due to lack of efficacy is an increasing problem, affecting many therapeutic areas, posing a major threat to the sustainability of the current model of drug development. A seemingly unattainable solution to this problem would be to obtain large-scale randomised human evidence on a target and disease state earlier in a drug development programme, without recourse to developing a medicinal compound that might be destined to fail.

The Research and Results

Our work extended the concept of Mendelian randomisation studies for drug development from individual targets to the whole genome. The study (1) defined a set of genes that not only encode actual (or potential) drug targets but which are also likely to be responsible for genetic associations with complex diseases from prior genome wide association studies (GWAS); (2) led to the design a genotyping array with enriched single nucleotide polymorphism (SNP) coverage of the druggable gene and; (3) linked the proteins encoded by this gene set to licensed drugs or to compounds with bioactivities against these targets. A variety of bioinformatics resources and other in silico tools were used to achieve these aims. The integrity of the analysis was evaluated through a comparison of the consistency between licensed drug indications and GSWAS associations through manual curation and blinded clinical expert review. This analysis showed that GWAS have already ‘rediscovered’ around 70 or so of the approximately 600 targets of licensed drugs through associations with disease indications, disease related biomarkers or mechanism-based adverse effects. The findings suggest the approach has promise as a tool to systematically identify target-human disease indication pairings and to identify drug repositioning opportunities for licensed drugs. Large scale electronic health record resources with and without genetic information are likely to play an important role.

The Impact

The work has highlighted the important part genomics can play in drug target identification and validation, addressing the major reason for late-stage drug development failure. The findings have been influential in directing pharmaceutical industry attention towards the use of genomics for drug target identification (not just drug response prediction), leading to a growing number of pharma-healthcare (particularly those with large scale electronic health records) and pharma-academic genomics partnerships focusing on drug development.

Reference

Finan C, Gaulton A, Kruger F, Lumbers T, Shah T, Engmann J, Galver L, Kelly R, Karlsson A, Santos R, Overington J, Hingorani A, Casas JP. The druggable genome and support for target identification and validation in drug development. Science and Translational Medicine 2017, 9, eaag1166, 29 March 2017.