Abstract :
[en] Preclinical models in gliomas —the most prevalent adult brain cancer that originate from glial cells —are either inaccurate in capturing metabolic dependencies in humans or hard to culture. The 2021 World Health Organization classification of the central nervous system tumors stratifies adult gliomas into three subtypes: IDH-mutant astrocytoma, oligodendroglioma and IDH-wildtype glioblastoma (GBM). The limited preclinical models in IDH-mutant gliomas and the poor five-year survival of 7% in GBM limit investigational drugs and the success rate of clinical trials, respectively. These clinical and preclinical obstacles resulted in few approved monotherapies that primarily target the cell cycle and approved combinations with redundant pathways. Therefore, the need for efficacious drugs and combinations targeting alternative pathways is pivotal. Drug repurposing, i.e., redirecting approved drugs to other diseases, has been critical in shortening the lengthy toxicity trials in cancer drug discovery. Among the computational drug repurposing approaches, metabolic modeling enables the simulation of the cellular metabolism using the well-annotated biochemical network with interpretable and accurate target identification. Our review of genome-scale metabolic models (GEMs) in the brain showed glioma GEMs only cover GBM, which are either built on a non-genome scale or lack curation. Here, we present GEMs of the three well-defined glioma subtypes built with the rFASTCORMICS algorithm that predicted repurposable FDA-approved single drugs and combinations for gliomas. Predicted drugs showed comparable efficacy to approved drugs using published in vitro and xenograft drug screenings. The oligodendroglioma metabolic model replicated the metabolic exchanges of the hard-to-culture oligodendroglioma preclinical models. Similarly, two novel predicted combinations of non-cancer drugs were coherent with the known dependencies in IDH-wildtype and -mutant gliomas. Finally, four predicted drugs showed comparable survival as monotherapy or improved survival combined with approved drugs compared to the approved drug arm in phase I/II glioma clinical trials. The previous drug prediction pipeline was also applied to repurpose approved drugs to COVID-19 and melanoma and natural products to breast cancer. Predicted drugs for melanoma, breast cancer and COVID-19 targeting cholesterol synthesis, estrogen metabolism, and cysteine synthesis, respectively, reduced mortality/incidence in their respective diseases. Unlike melanoma-specific cholesterol synthesis, the glioma GEMs accurately captured the in vivo dependency of cholesterol esterification and avoided the in vitro cholesterol synthesis dependency that failed in a clinical trial. Overall, this work provides an overview of how metabolic modeling can be used to detect biomarkers and repurpose drugs, where metabolic modeling was competitive with preclinical methods and could predict new drugs.
