Abstract
AbstractGlioblastoma is currently associated to a dismal prognosis despite intensive treatment involving maximal-safe surgery, radiotherapy and temozolomide (TMZ)-based chemotherapy. Disease progression or relapse is often due to initial or acquired resistance to temozolomide, which may be mediated by the over-expression of the repair enzyme MGMT. To design TMZ-based drug combinations circumventing the initial resistance of MGMT-overexpressing cells, a quantitative systems pharmacology (QSP) model representing TMZ cellular pharmacokinetics-pharmacodynamics and their connection to the most altered pathways in GBM was developed. This digital network representation of TMZ cellular pharmacology successfully integrates, in a mechanistic fashion, multi-type time- and dose-resolved datasets, available in control or MGMT-overexpressing cells.In silicotarget inhibition screening identified an optimal antitumor strategy consisting in priming cancer cells with inhibitors of the base excision repair and of the homologous recombination pathway prior to TMZ exposure. This drug combination was validated in dedicated experiments, thus allowing to re-sensitize cells which were initially resistant to TMZ. Using machine learning, functional signatures of response to such optimal multiagent therapy were derived to assist decision making about administering it to other cancer cell lines or patients. The developed framework can be extended to account for additional patientspecific altered pathways and may be translated towards the clinics by representing the tumor micro-environment and drug whole-body pharmacokinetics. Overall, we successfully demonstrated the relevance of combined QSP and machine learning to design multi-agent pharmacotherapies circumventing initial tumor resistance.One Sentence SummaryAn integratedin vitro-in silicoapproach allowed to design optimal drug combinations re-sensitizing temozolomide-resistant glioblastoma cells.
Publisher
Cold Spring Harbor Laboratory