NEW YORK – New drugs are more likely to be approved if they target a gene linked to disease, according to a new analysis from AbbVie researchers that confirms earlier findings.
Only between 5 percent and 10 percent of new drug candidates that enter early-stage clinical trials are eventually approved. In 2015, researchers led by GlaxoSmithKline's Philippe Sanseau reported that drugs with supporting genetic evidence were twice as likely to be approved.
AbbVie's Emily King and her colleagues have now expanded that analysis to include additional data. As they reported in PLOS Genetics this week, they validated the previous study's findings, but also noted that the association between genetic evidence and drug approval was stronger for genetic evidence derived from severe genetic disorders and from genetic variants that affect protein sequences.
"The findings from this study demonstrate that human genetics evidence is predictive of historical drug development success," King, a postdoctoral fellow at AbbVie, said in a statement. "Moving forward, we believe that a more detailed understanding of the links between genetic variants and diseases will only continue to help scientists design more successful drug development programs."
In the previous Nature Genetics study, GlaxoSmithKline's Sanseau and his colleagues compiled a set of gene-trait combinations by mapping common variant genetic associations from GWASdb and rare Mendelian traits from the Online Mendelian Inheritance in Man (OMIM) database to Medical Subject Heading (MeSH) terms. They likewise used the Informa Pharmaprojects database to generate a set of target-indication pairs. Through this, they found that the target genes for approved drugs were enriched among genes associated with human trait variation.
For their replication study, King and her colleagues similarly gathered data from the GWAS Catalog, OMIM, and Pharmaprojects were not included in that initial analysis. They again found a strong association between genetic evidence and progression through clinical development or drug approval — but that association was stronger for OMIM-derived genetic evidence than for GWAS-derived genetic evidence.
While GWAS-supported target-indication pairs are more likely to be approved than those without a GWAS-linked gene target, the researchers noted a smaller magnitude of association for them than for the OMIM-derived evidence.
This, the researchers said, could be because recently reported GWAS variants typically have smaller effect sizes than ones linked through Mendelian disease and often do not affect protein-coding genes. However, they did not find that GWAS evidence effect size influenced approval, though they noted that when they examined GWAS variants in high linkage disequilibrium with a missense variants or other variant predicted to be moderately or highly deleterious, the estimated effect of the GWAS genetic evidence rivalled that of OMIM.
King and her colleagues developed multivariate logistical regression models to predict target-indication pair approvals, which they built into an app called Shiny that can be used to evaluate target-indication pairs of interest.
"Human genetics has the potential to help us to focus our investment on drug programs that are most likely to have an impact on patients," Howard Jacob, vice president and head of genomic research at AbbVie, said in a statement. "Studies like this one further demonstrate the importance of human disease genetics in drug development."