Uncover the Underlying Mechanism of Drug-Induced Myopathy by Using Systems Biology Approaches

Abstract

Drug-induced myopathy (DIM) is a rare side effect; however, the consequence could be fatal. There are few reports to systematically assess the underlying mechanism of DIM. In this study, we curated the comprehensive DIM drug list based on structured labeling products (SPLs) and carried out the analysis based on chemical structure space, drug protein interaction, side effect space, and transcriptomic profiling space. Some key features are enriched from each of analysis. Specifically, the similarity of DIM drugs is more significant than random chance, which shows that the chemical structure could distinguish the DIM-positive drugs from negatives. The cytochrome P450 (CYP) was identified to be shared by DIM drugs, which indicated the important role of metabolism in DIM. Three pathways including pathways in cancer, MAPK signaling pathway, and GnRH signaling pathway enriched based on transcriptomic analysis may explain the underlying mechanism of DIM. Although the DIM is the current focus of the study, the proposed approaches could be applied to other toxicity assessments and facilitate the safety evaluation.

Figure 1

Flowchart of the study: (1) similarity between DIM drugs based on chemical similarity and protein distance in PPI network; (2) myopathy-related side effects based on Fisher's exact test and network analysis; (3) the DIM transcriptomic analysis based on CMap data; (4) structure alert of DIM.

Figure 2

The therapeutic categories distribution of drug-induced myopathy (DIM) drugs. The DIM drugs were mapped to the second level of the WHO Anatomical Therapeutic Chemical (ATC) Classification System. Then, for each therapeutic category, the number of DIM drugs were counted.

Figure 3

Permutation test for similarity of DIM drug pairs. The distribution of 75 DIM-positive drugs were drawn based on their chemical similarity. Then, the random test was carried out based on chemical similarity of negative DIM drugs, which are randomly picked up for 100,000 times. Then, the p value could be calculated for assessing whether the DIM-positive drugs are more similar.

Figure 4

The protein space of DIM drugs. (a) The top 10 protein targets for DIM; the DIM drug and target relationship was extracted from DrugBank. (b) The STRING PPI for the top 10 protein targets; the top 10 proteins corresponding to more DIM drugs were inputted to the STRING PPI database to exact the subnetwork among the 10 proteins.

Figure 5

Network myopathy-related side effects. The myopathy and side effects were extracted based on SIDER database.