Computational repurposing of oncology drugs through off-target drug binding interactions from pharmacological databases

Abstract

PURPOSESystematic repurposing of approved medicine for another indication represents an attractive strategy to accelerating drug development in oncology. Herein we present a strategy of combining biomarker testing with drug repurposing to identify new treatments for patients with advanced cancer. METHODSTumours were sequenced with Illumina TruSight Oncology 500 (TSO-500) platform or the FoundationOne(R) CDx panel. Mutations were manually screened by two medical oncology clinicians and pathogenic mutations were categorised with reference to the literature. Variants of unknown significance were classified as potentially pathogenic if a plausible mechanism and computational prediction of pathogenicity existed. Gain of function mutations were evaluated through the repurposing databases Probe Miner, the Broad Institute Drug Repurposing Hub (Broad Institute DRH) and TOPOGRAPH. Gain of function mutations were classified as repurposing events if they were identified in Probe Miner, were not indexed in TOPOGRAPH which captures active clinical trial biomarkers and excluding mutations for which a known FDA-approved biomarker label exists. The performance of the computational repurposing approach was validated by evaluating its ability to identify known FDA-approved biomarkers. Exploratory functional analyses were performed with gene expression data and CRISPR-dependency data sourced from the DepMap portal. The total repurposable genome was identified by evaluating all possible gene-FDA drug approved combinations in the Probe Miner dataset. RESULTSThe computational repurposing approach was highly accurate at identifying FDA therapies with known biomarkers (94%). Using a real-world dataset of next-generation sequencing molecular reports (n = 94) and excluding the identification of mutations that would render patients eligible for FDA-licensed therapies or local clinical trials, it was found that a meaningful percentage of patients (14%) would have an off-label therapeutic identified through this approach. Exploratory analyses were performed, including the identification of drug-target interactions that have been previously described in the medicinal chemistry literature but are not well known, and the evaluation of the frequency of theoretical drug repurposing events in the TCGA pan-cancer dataset (73% of samples in the cohort). CONCLUSIONOverall, a computational drug repurposing approach may assist in identifying novel repurposing events in cancer patients with advanced tumours and no access to standard therapies. Further validation is needed to confirm the utility of a precision oncology approach using drug repurposing.