A Brief Introduction to Drug Repositioning
SummaryGenerally speaking, the drug repositioning strategy includes three steps: candidate molecules for target indications; evaluation of preclinical drug action mechanism; evaluation of second-phase clinical drug efficacy (assuming that the first phase of the original indication has been completed and has sufficient security data).
- Author Name: Lisa George
The exploitation of new use of old drugs is also called the repositioning, reanalysis or redevelopment of drugs. It is a strategy to develop new uses in addition to the existing indications of drugs that have been approved or studied. This strategy has multiple advantages over development of new drugs. First, it can reduce the risk of failure compared with new drug development, because the reused drugs have been fully verified by animal models and humans in early clinical trials. Secondly, the drug development cycle is shortened, because most of the preclinical trials, safety evaluation, and even formulation development have been completed. Thirdly, the costs in the drug candidate development stage and process are greatly reduced. Statistics show that the average cost of reusing old drugs from clinical to marketing is about US$300 million, and that of a new drug is about US$2 billion to US$3 billion. In addition, the repositioning of old drugs may reveal new targets and signaling pathways, which can be used for further development and utilization.
Generally speaking, the drug repositioning strategy includes three steps: candidate molecules for target indications; evaluation of preclinical drug action mechanism; evaluation of second-phase clinical drug efficacy (assuming that the first phase of the original indication has been completed and has sufficient security data).
- Drug matching: analysis of drug-disease relationships based on gene expression profiles
The characteristics of drugs generally come from three types of data: transcriptome (RNA), proteomics or metabolomics data, chemical structure, and adverse event spectrum. The drug-drug similarity method aims to determine the common mechanism of action of different drugs. This principle is called correlation inference, and it can help to identify new uses of existing drugs and discover potential off-target effects in clinical applications. Therefore, the transcriptomic characteristics shared between the two drugs may indicate that they also share therapeutic indications, regardless that whether their chemical structures are similar. This principle has proven to be effective, and it is the same when comparing transcription characteristics.
Molecular docking is a structure-based computational strategy used to predict the complementarity of binding sites between ligands (e.g. drugs) and targets (e.g. receptors). If you have a priori knowledge of the receptor target involved in the disease, you can simulate the design of multiple drugs for that specific target (conventional docking: one target and multiple ligands). Instead, drug libraries can be explored against a range of target receptors (reverse docking: several targets and one ligand) to identify new interactions that can be used by old drugs.
- Whole genome association research
The number of genome-wide association studies (GWAS) has increased significantly. Thanks to advances in genotyping technology, a significant reduction in costs has been made possible over the past 10 years. GWAS aims to determine the genetic variation associated with common diseases, and the data obtained may also help to identify new targets, some of which can be compared between drug-treated diseases and disease phenotypes studied by GWAS, thereby repositioning drugs.
- Signaling pathway
The signaling pathway has been widely used to identify drugs or drug targets that may have potential for reuse. GWAS or other means may find some potential drug target genes, but in general, these genes may not be ideal drug targets. In this case, pathway network research can provide information on genes upstream or downstream of GWAS-related targets to see whether these genes contain opportunities for new drugs and new uses. Network analysis involves building a drug or disease network based on gene expression patterns, disease pathology, protein interactions, or GWAS data to help identify older drugs that can be repurposed. Some of the drug matching studies also use pathway network methods.
- Retrospective clinical analysis using electronic health records
Retrospective clinical data can be obtained from various sources, including electronic health records (EHR), post-marketing monitoring data, and clinical trial data. The EHR contains a large amount of structured and unstructured patient outcome data. Structured data is mainly diagnostic and pathophysiological data, including laboratory test results and drug prescription data. Unstructured information such as clinical descriptions of patient symptoms and signs (very important for defining disease phenotypes) and imaging data. A large amount of data existing in EHR can be used as an information source to identify new drugs for old drugs. In addition, a large amount of EHR data also provides a statistical basis. Post-marketing surveillance data and clinical trial data are two other important big data sources, but for commercial or confidential reasons, access to these data will be restricted to a certain extent. However, it is increasingly recognized that open access to such kind of information will help to further develop drug research.
At present, researchers have begun specific research on repositioning drugs in conjunction with university affiliated hospitals and pharmaceutical companies, and have obtained some prediction results. Moreover, an AI system has been developed to analyze long-term accumulated medical data so as to predict whether existing drugs for a certain disease are effective for other diseases.
The researchers noticed that there are different types of protein mutations between different diseases, so AI was used to compare various types of protein mutations, and then the corresponding alternative drugs were matched according to the similarity of the protein mutation types. The specific method of this system is: first, analyze the protein variation in about 1300 diseases and the various components of about 8000 drugs to find out a group of diseases with similar protein variations, and correlate them to predict the interchangeability of drugs. In addition, AI also detects about 20,000 different types of genetic changes produced when the drug is administered to cells as well as changes in physical conditions after taking the drug.