In:
Cancer Research, American Association for Cancer Research (AACR), Vol. 79, No. 13_Supplement ( 2019-07-01), p. 682-682
Abstract:
Despite recent advances in life sciences and technology, bringing a single drug to market has stayed drastically expensive, leading to a decline in the number of new drugs being clinically approved. Drug repurposing, identifying novel indications for approved drugs, presents an opportunity to avoid many common pitfalls, such as toxicity, and bring treatments to market faster. Additionally, drug repurposing, which aims to identify indications for unapproved, shelved drugs, can also be a short-cut to the clinic due to the previously completed preclinical work. Current computational efforts for indication prediction focus on leveraging the similarities between drugs’ approved indications, however these approaches are unsuitable for drug repurposing. Here we introduce a novel big data approach that integrates only compound similarity metrics to identify novel indications for drugs, regardless of the existence of a primary indication. We introduce a computational approach that integrates multiple data types to create a drug-indication network, which can be used to recommend indications a drug may treat. Using this network, we can accurately elucidate novel indications for drugs currently being used as treatments in clinical trials. Additionally, we present multiple promising candidates for future clinical work. This drug-indication network is built using a confidence score that two drugs share an indication, calculated using a gradient boosting classifier. Utilizing only publicly available data, we built this classifier based on drug similarity features and identified previously undocumented associations between approved drugs which share indications such as significantly high side effect (D = 0.26, p & lt; 0.001) and target similarity (D = 0.21, p & lt; 0.001). When trained on FDA approved drugs, this classifier achieved significant predictive performance, with an area under the ROC curve of 0.82.We then used our method to determine potential repurposing opportunities. We predicted numerous statins, including Lovastatin and Cerivastatin, known cholesterol-lowering agents, as potential treatments for nasopharyngeal cancer. Recent work has shown statins to be promising anti-tumor agents, specifically in nasopharyngeal cancer, highlighting the specificity of our approach. Furthermore, we predicted Cinnarizine, a calcium channel blocker with potential antipsychotic effects, to treat schizophrenia. In addition to identifying these and a variety of other novel indications with our drug-indication network, our complete method incorporates large-scale biological/chemical data to help understand the underlying mechanisms behind new indication predictions. Altogether, our method provides a distinctive strategy that can identify novel and diverse indications, allowing for an expedited and more efficient method for future drug development and repurposing efforts. Citation Format: Jamal A. Elkhader, Coryandar M. Gilvary, Neel S. Madhukar, Olivier Elemento, David Solit. Drug repurposing engine fueled by diverse drug similarity data [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 682.
Type of Medium:
Online Resource
ISSN:
0008-5472
,
1538-7445
DOI:
10.1158/1538-7445.AM2019-682
Language:
English
Publisher:
American Association for Cancer Research (AACR)
Publication Date:
2019
detail.hit.zdb_id:
2036785-5
detail.hit.zdb_id:
1432-1
detail.hit.zdb_id:
410466-3
Bookmarklink
