Drug target gene-based analyses of drug repositionability in rare and intractable diseases

Abstract

Drug development for rare and intractable diseases has been challenging for decades due to the low prevalence and insufficient information on these diseases. Drug repositioning is increasingly being used as a promising option in drug development. We aimed to analyze the trend of drug repositioning and inter-disease drug repositionability among rare and intractable diseases. We created a list of rare and intractable diseases based on the designated diseases in Japan. Drug information extracted from clinical trial data were integrated with information of drug target genes, which represent the mechanism of drug action. We obtained 753 drugs and 551 drug target genes from 8307 clinical trials for 189 diseases or disease groups. Trend analysis of drug sharing between a disease pair revealed that 1676 drug repositioning events occurred in 4401 disease pairs. A score, R_gene, was invented to investigate the proportion of drug target genes shared between a disease pair. Annual changes of R_gene corresponded to the trend of drug repositioning and predicted drug repositioning events occurring within a year or two. Drug target gene-based analyses well visualized the drug repositioning landscape. This approach facilitates drug development for rare and intractable diseases.

Figure 1

Analysis of drug repositioning in rare and intractable diseases based on clinical trial data. (A) Pipeline of extracting information from clinical trials. First, from 412,555 clinical trials in the four registries (JPRN, ClinicalTrials.gov, EU-CTR, and ChiCTR), 15,194 trials which are for rare and intractable diseases and contain drug information were selected. Second, by limiting the drugs to those that have target gene information, 8307 trials containing those drugs were selected. (B) Schema of drug repositioning where a drug in disease A is repositioned to disease B. A drug repositioning occurs when a drug which formerly tested for a disease newly becomes shared. (C) Accumulated number of clinical trials identified as drug repositioning trials enrolled annually in recent 20 years. In total, 1676 clinical trials were identified during the period. (D) Accumulated number of disease pairs combined for drug repositioning first appeared during the period. In total 4401 disease pairs were counted during the period.

Figure 2

R_gene: A score of drug repositionability based on sharedness of drug target genes between a disease pair. (A) Schema of drug repositioning where a drug in disease A is repositioned to disease B, with information of drug target genes. Drugs and their target genes are differently positioned across the three regions (disease A only, disease B only, and disease A and B). (B) Calculation of R_gene as a score of drug target gene sharedness between disease A and disease B. (C) Log distribution of R_gene of 6808 disease pairs. R_gene of 2466 disease pairs (36.2%) was less than 1. Note: disease pairs which had no repositioned drugs are included. (D) Inter-disease network by R_gene. A circular network consists of 33 diseases from the top 100 R_gene disease pairs were depicted. The number of disease pairs (connected lines) are different among the diseases. Multiple sclerosis (ID:13) and Crohn’s disease (ID:96) are paired to many other diseases.

Figure 3

Annual changes of R_gene and drug repositioning between disease pairs in recent 20 years. (A) Annual changes of R_gene and drug repositioning between multiple sclerosis (MS) and Crohn’s disease (CD). R_gene increased gradually correlating to that of the number of repositioned drugs during the whole period. Three characteristic surges were seen for both R_gene and drug repositioning in 2007–2008, 2010–2011, and 2014–2016. (B) For MS and CD pair, annual changes of R_gene and drug repositioning were standardized by SD. The three surges of R_gene (ΔR_gene) appeared to precede corresponding surges of drug repositioning (ΔDrug) by about a year. (C) Odds ratios of R_gene and drug repositioning as an explanatory variable and an objective variable, respectively, by logistic regression analyses. The top 100 repositioning-active disease pairs were divided into 10 groups consecutively and examined the predictability by R_gene. This analysis validated the ubiquitous predictability by R_gene. Note: In fact, 93 disease pairs were used since 7 pairs lacked drug repositioning events.