Abstract 776: Utilising genetic susceptibility and big data to inform novel cancer therapies

Abstract

Abstract Despite a move towards personalised medicine, attrition rates for new cancer drugs remain unacceptably high. The pharmaceutical industry has also shown a preference for well-studied targets and pathways, as evidenced by ‘me too' drugs. Together with the challenge of inadequate pre-clinical models, this indicates a need for novel, evidence-based therapeutic targets. Genome-wide association studies (GWAS) have identified over 450 robust genetic variants associated with increased cancer risk. Genes implicated through GWAS are often mutated somatically and therefore represent attractive therapeutic targets. Examples include the target of venetoclax in chronic lymphocytic leukaemia, BCL2. We exploit this principle more generally by integrating genetic associations for common cancers with drug target data and druggability using the canSAR drug discovery knowledgebase (https://cansar.icr.ac.uk). By harnessing the power of Big Data we aim to both identify opportunities for repurposing of existing drugs, and prioritise novel targets for cancer drug discovery. We mined the NHGRI-EBI Catalog of published GWAS for all cancer risk SNPs. We annotated candidate target genes through overlapping topologically associating domains (TADs), a more sensitive technique than previously published methods using linkage disequilibrium. We used canSAR to identify target genes for which there is no FDA-approved small molecule drug, and the resource Probe Miner to identify targets for which high-quality chemical probes exist. We also utilised canSAR's machine learning algorithms to assess the druggability of target genes by structure-, ligand-, precedence- and network-based approaches. We additionally analysed results from cancer drug databases to ascertain whether there is an enrichment of ‘drug target-indication pairs' at successive stages of the drug development pathway for which supporting evidence from GWAS exists: this indicates potential ‘stumbling blocks' that may present a risk for future drug development projects. 7 257 protein-coding genes mapped within TADs overlapping cancer risk SNPs. Of these, 98 were pre-existing targets for which there is an FDA-approved small molecule drug. For the remaining 7 159 genes we performed multi-faceted druggability analyses incorporating assessments of the 3D structure of the target and any protein complexes it exists in, chemical properties of known ligands of the target, and the target's position and role within the human interactome. We comprehensively rank our target-indication pairings by criteria including novelty relative to existing targets and predicted attrition risk. Mapping approved drug targets back to cancer GWAS signals enables identification of both novel drug targets and patient populations. Collectively our findings show the value of investigating germline cancer genetics as part of interdisciplinary, data-driven approaches to inform drug discovery. Citation Format: Elizabeth A. Coker, Ben Kinnersley, Amit Sud, Patrizio Di Micco, Bissan Al-Lazikani, Richard Houlston. Utilising genetic susceptibility and big data to inform novel cancer therapies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 776.