M4 drug discovery: human drug predictions from integrated pre-clinical insights exemplified with a GLP1-R agonist

Drug discovery utilises cell- and animal-systems to predict human responses. These different systems provide different drug discovery insights. However, there is currently no methodology to integrate all these insights into single quantitative framework. This study presents a new methodology -- M4 drug discovery -- which achieves such an integration. The method integrates: a) Multi-timescale data (short- to long-term responses); b) Multi-level data (intracellular to whole-organism responses); c) Mechanistic models (describing mechanistic underpinning); and d) Multi-species data (for example rodent and monkey responses). We exemplify this new method using 16 new human cell culture studies, 6 pre- existing human cell culture studies, and 6 pre-existing in vivo studies of the glucagon-like peptide-1 receptor agonist exenatide. All data can be simultaneously explained, and we successfully predict new human cell culture studies ({chi}2=64<83, p=0.05), human pharmacokinetics ({chi}2=64<97, p=0.05), and human drug responses ({chi}2=36<45, p=0.05). We found that different systems contribute complementary insights by predicting a 30-week human exenatide treatment study: multi-species data provides pharmacokinetics; human cell data provides human potency readings; and animal data provides drug effects, such as appetite changes, which could not be observed in our human cell culture studies. Our method opens the door to more cost-effective, ethical, and knowledge-based drug discovery.