Abstract
Purpose :
Proliferative vitreoretinopathy (PVR) is the dreaded cause of failure following retinal detachment repair; however, to this date, no cures or preventative therapies exists. The purpose of this study is to use advanced bioinformatics tools to study drug-gene associations in order to identify drugs/compounds that interact with genes affected in PVR and that could be candidates for further testing as novel management strategies of PVR.
Methods :
We queried PubMed Gene to assemble a list of genes associated with PVR to date. Gene enrichment analysis for the compiled PVR gene list was executed via ToppGene against the drug databases. This analysis identifies overepresented compounds and predicts their statistical significance from multiple drug-gene interaction databases to formulate a Pharmacome. Compounds with adjusted p-value <0.05 were chosen. Compounds with no clinical indications (e.g., ozone, ethanol) were filtered out from the resulting drug lists.
Results :
Our query identified 34 unique genes associated with PVR. Out of 77,146 candidate drugs/compounds in the drug databases, our analysis revealed multiple drugs/compounds with significant interactions with genes involved in PVR. The drugs predicted amongst the top 100 most significant compounds include anti-proliferatives, corticosteroids, anti-diabetics, antioxidants, statins, and flavonoids. Prednisone (P=2.08x10-7), a corticosteroid that affects 4 of the 34 identified PVR genes, and methotrexate (P=1.70x10-4), an anti-neoplastic agent that also affects 4 of the 34 genes, have shown promising results in animal studies and ongoing clinical trials for PVR. Other predicted top compounds including metformin (P=8.30x10-12), statins (P=9.76x10-19), and curcumin (P=2.03x10-17), which affect 11, 17, and 18 of the 34 PVR genes, respectively, have well-established safety profiles and could be readily repurposed for PVR.
Conclusions :
This bioinformatics approach of studying drug-gene interactions can speculate drugs which may affect genes and pathways implicated in PVR. We use our growing understanding of systems biology to introduce a computational network medicine process for finding novel therapeutic targets for PVR. While further validation from preclinical/clinical studies is required for bioinformatic predictions, this unbiased approach could identify possible candidates out of existing drugs/compounds that may be repurposed for PVR and guide future investigations.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.