Approved research

Using UK Biobank data to support the Novartis drug discovery and development pipeline

Lay summary

Novartis' goal is to develop innovative medicines across the entire range of human disease. Drug development is very challenging: it takes a long time and has a high failure rate. This failure rate can be reduced by utilizing genetic information to make better choices about which patients and which diseases are targeted by a drug. We wish to use the combined genetic data and patient phenotype data available in the UK Biobank to: inform our drug development strategy, identify new drug targets and better characterize patient populations so that the drug programs we pursue are targeted at the patients who will benefit most. The initial projects will focus on several heart related diseases i.e. hypercholesterolaemia and hyperlipoproteinaemia. Novartis has active drug programs targeting specific genetic targets of these conditions i.e. a siRNA therapy to lower LDL-C (inclisiran) and an Lp(a) antisense inhibitor (pelacarsen). There is a need to understand the epidemiology and phenotypic characteristics of patients who are at high risk of suffering from cardiovascular events but who could benefit from these classes of therapies. The initial project will aim to determine predictors of cardiovascular risk and mortality in patients with elevated LDL-C and/or elevated Lp(a) levels. The project will involve characterizing the participants who have evidence of atherosclerotic cardiovascular diseases including their current treatment patterns and health outcomes. We will look at the subjects who have high risk genotypes and a history of prior cardiovascular (CV) events and determine if there are demographic, environmental or additional genetic variables that discriminate between this population and those participants who have no equivalent history of CV events. The results of these projects will be used, as input into clinical trials and presented as talks or posters at the relevant international cardiovascular conferences and/or published in peer reviewed cardiovascular journals. This research will help to determine which patients will benefit most from these targeted therapies. This information can then be used to inform the population level strategy for reducing heart disease in the UK (where it is currently the second biggest cause of death).

Scope extension:

Evaluate associations between drug-target genes and health related outcomes (including disease occurrence, prevalence, disease severity, disease progression) by application of systematic phenome-wide association approach (PheWAS). In particular we are interested in using the patient record level data (HES and GP level) and taking advantage of the ICD10 disease hierarchy and ATC drug categorization hierarchies to evaluate PheWAS associations at different levels of the disease and drug hierarchy.

Fine grained description of disease natural history, epidemiology and patient populations. Construction of high precision patient endotypes particularly in chronic diseases where we have constructed and validated clinical phenotypes.

Identification of responders/non-responders to specific medicines based on patient genotype.

Develop learning representations for different diseases using features derived from imaging data (including OCT/retinal scans imagery) in conjunction with other datatypes to identify patient subgroups (e.g. fast vs slow progressors for age related macular degeneration ).

Conduct genomic association analyses and machine learning modelling using NMR metabolomics, imaging, and whole exome sequence data to assess molecular signatures for target validation related to metabolic and other diseases.

Multimodal analyses using statistical, machine learning and AI techniques utilising the diverse data types available in UKBB (genetics, NMR, proteomics, imaging, health outcomes, phenotypes, patient questionnaire responses etc.) to characterise diseases, patient subgroups and patient disease trajectories. These may be used to develop clinical prediction models; to identify and validate drug targets and molecular signatures; development of patient screening tools for clinical trials.