A44-33 Applying Causal Discovery Methods to Drug Repurposing in COPD

Rationale Chronic Obstructive Pulmonary Disease (COPD) is a progressive condition characterized by the gradual decline in lung function, resulting from airway wall narrowing and alveolar destruction. While there is no cure for COPD, current treatments such as long- and short-acting bronchodilators as well as inhaled corticosteroids (ICS) can alleviate symptoms and improve quality of life. Beyond these medications there are limited options for COPD patients as recent clinical trials for potential new drugs have failed to show efficacy. In this study, we propose to explore the repurposing of pre-approved drugs for COPD treatment by leveraging causal discovery methods. Specifically, we aim to identify genes with known drug targets that could be modulated to improve COPD outcomes. Methods Using transcriptomic data at five year follow-up from the COPDGene cohort, we first filter for genes with established drug interactions using the DGIdb database. We then identify a Markov Blanket for key spirometric variables like FEV1 (% predicted) and FEV1/FVC using the Incremental Association Markov Blanket (IAMB) algorithm. The Markov Blanket will contain genes that render the target variable conditionally independent of all other genes and serve as a candidate set of genes to further investigate known drug interactions. The IAMB algorithm relies on conditional independence tests which require a significance threshold to be specified. Here we test multiple thresholds to quantify the stability of genes within the Markov Blanket. Results Preliminary results from our analysis identify 74 genes that appear within the Markov Blankets of both FEV1 (% predicted) and FEV1/FVC. Notably, we observe genes like CSF2 and IL1B which have known interactions with Fluticasone (a commonly prescribed corticosteroid for COPD) as well as PDE4B which is directly inhibited by Roflumilast—a medication used to prevent exacerbations in COPD. A total of five genes appeared across all Markov Blankets and thresholds, with two being G-protein coupled receptors (GPR15 and GPR183 ). Conclusion The goal of this study is to provide a more targeted approach to drug repurposing to give better treatment options for COPD patients. Our method was able to identify previously established COPD specific gene-drug interactions which gives some support to the newly discovered interactions. Our future work will incorporate other cohorts as well as include current medication use to better inform Markov Blanket detection. This abstract is funded by: This work was supported by NHLBI R01 HL151421 and R01HL157879. The COPDGene study (NCT00608764) is supported by grants from the NHLBI (U01HL089897 and U01HL089856), by NIH contract 75N92023D00011, and by the COPD Foundation through contributions made to an Industry Advisory Committee that has included AstraZeneca, Bayer Pharmaceuticals, Boehringer- Ingelheim, Genentech, GlaxoSmithKline, Novartis, Pfizer and Sunovion.