Identification of repurposing opportunities for uric acid-lowering drugs using integrative multi-omics and machine learning analyses

Abstract

INTRODUCTION: Previous epidemiological studies have identified that elevated uric acid (UA) levels were significantly associated with increased risk of disease outcomes other than gout, such as cardiovascular diseases (CVDs). However, whether uric acid-lowering therapies could be repurposed for the treatment of CVDs remains unclear. In addition, further investigation is needed to understand possible mechanisms through which uric acid-lowering drugs affect the progression of CVDs.

AIMS: This project aims to estimate the effects of genetically proxied uric acid-lowering treatments on the risk of CVDs by using a candidate drug repurposing strategy. Furthermore, a drug-wide repurposing strategy is employed to provide more details regarding the underlying biological mechanisms by integrating multi-class data sources.

METHODS: First, a systematic review on the methodologies in drug repurposing using human genomic data was conducted. Then, in the candidate drug repurposing strategy, I conducted both observational phenome wide association study (Obs-PheWAS) and polygenic risk score PheWAS (PRS-PheWAS) to examine the links between uric acid levels and a broad spectrum of disease outcomes. For those disease outcomes identified as significant through both approaches, I conducted trajectory analysis to investigate patterns of disease progression following elevated uric acid levels. Meanwhile, with a multivariable logistic regression model, I performed drug repurposing analysis to assess potential therapeutic benefits of uric acid-lowering medications for the associated cardiovascular outcomes. Factorial Mendelian randomization (MR) was also carried out to detect whether there exist potential interactions between uric acid-lowering and antihypertensive therapies. Finally, in the drug-wide repurposing strategy, meta-analysis of genome wide association studies (GWASs) on CVDs was conducted to identify associated genetic variants with greater statistical power and accuracy. Based on the combined summary statistics, transcriptome wide association study (TWAS) was performed to obtain differential expressed genes, which were subsequently included in the following Connectivity Map (CMap) analysis to identify potential repurposing candidates. Six machine learning approaches including linear discriminant analysis(LDA), classification and regression trees (CART), k-nearest neighbors (kNN), support vector machines (SVM), random forest (RF) and gradient boosted method (GBM), were applied to evaluate the validity of the identified repurposing opportunities. Additionally, proteome wide association study (PWAS) and MR analysis were carried out to prioritize compounds with high evidence. Drug and disease information for these candidates was also summarized to provide clinical validation.

RESULTS: We searched MEDLINE and EMBASE databases to identify eligible studies up until 1 May 2023, with a total of 102 studies finally included after two-step parallel screening. Three drug repurposing strategies, including Mendelian randomization, multi-omic-based drug repurposing and network-based drug repurposing were summarized based on approaches and data sources used. Then, we integrated and categorized these methodologies into candidate drug repurposing and drug-wide repurposing strategies based on the objectives of the project. First, the candidate drug repurposing strategy observed a total of 41 overlapping disease outcomes with consistent effect directions across both Obs-PheWAS and PRS-PheWAS, including 17 circulatory diseases, 7 endocrine/metabolic diseases, 7 genitourinary diseases, 2 musculoskeletal diseases, 2 digestive diseases, 2 infectious diseases, 1 respiratory disease, 1 hematopoietic disease and 1 neoplasm. Then, trajectory analysis established 595 possible disease pairs for 35 unique health outcomes that were at an increased risk with elevated uric acid levels. As a result, one cluster mainly including diseases of cardiometabolic system was identified, where the disease tree thrived after the diagnoses of obesity, type 2 diabetes, hypercholesterolemia, essential hypertension, coronary atherosclerosis and myocardial infarction, followed by anemia, pneumonia, heart failure, renal failure, and finally ended up on death. Finally, drug repurposing analysis found that uric acid-lowering drugs exerted a protective role in reducing the risk of coronary atherosclerosis (OR=0.96, 95%CI: 0.93, 1.00, P=0.049), congestive heart failure (OR=0.64, 95%CI: 0.42, 0.99, P=0.043), occlusion of cerebral arteries (OR=0.93, 95%CI: 0.87, 1.00, P=0.044) and peripheral vascular disease (OR=0.60, 95%CI: 0.38, 0.94, P=0.025). Furthermore, the combination of uric acid-lowering therapy with antihypertensive treatment exerted additive effects and was associated with a 6%, 8%, 8% and 10% reduction in risk of coronary atherosclerosis, heart failure, occlusion of cerebral arteries and peripheral vascular disease, respectively.

The drug-wide repurposing strategy identified 270, 25, 33, 166, 30, 13 and 29 genomic loci for CVD, angina, coronary atherosclerosis, myocardial infarction (MI), valvular heart disease (VHD), venous thromboembolism (VTE) and heart failure (HF), respectively. Correspondingly, by integrating the joint effects of multiple variants within specific genes, TWAS detected 269, 32, 54, 215, 29, 17 and 40 signals representing strong evidence of shared causal variants influencing both gene expression and the cardiovascular outcomes. Based on the transcriptome files from both diseases and compounds, CMap analysis revealed 578, 78, 36 and 3 repurposing candidates for CVD, coronary atherosclerosis, MI and VHD, respectively. Machine learning methods exhibit high performance, with a receiver operating characteristic (ROC) value around 0.9, in distinguishing repurposed compounds from non-repurposed ones.

Finally, 16 compounds were prioritized based on their protein levels being significantly related to cardiovascular risk, and 7 of them have completed phase 4 clinical trials for CVDs.

CONCLUSIONS: Our findings from candidate drug repurposing strategy support a role of elevated uric acid levels in advancing cardiovascular dysfunction and identify potential repurposing opportunities for uric acid-lowering drugs in cardiovascular treatment. Furthermore, drug-wide repurposing strategy also validated that uric acid-lowering drugs, particularly xanthine oxidase inhibitors (XOIs), are promising repurposing candidates for CVD treatment.