Computational Analysis of Non-small-cell Lung Cancer Drug Resistance

通過計算的方法分析非小分子肺癌的抗藥性

Student thesis: Doctoral Thesis

View graph of relations

Author(s)

Related Research Unit(s)

Detail(s)

Awarding Institution
Supervisors/Advisors
Award date4 Sep 2017

Abstract

Mutation of the epidermal growth factor receptor (EGFR) is a pathogenic factor in non-small cell lung cancer (NSCLC) leading to EGFR tyrosine kinase inhibitors (TKIs), such as gefitinib, being widely used in the treatment of NSCLC. Patients’ responses to TKIs are largely related to the type of mutation in EGFR. Despite initial promising and dramatic responses to TKIs, drug resistance will almost always develop within 1 to 2 years. Decoding drug resistance in NSCLC can be guided by characterizing the properties of EGFR mutants through mathematical models. This thesis analyzes the drug resistance of EGFR mutations found in NSCLC using different computational methods. This thesis (a) analyzes the relationship between the response level to lung cancer drugs and atomic connectivity dynamics based on trimmed Delaunay triangulation, (b) predicts sensitivity to gefitinib/erlotinib for EGFR mutations in NSCLC based on structural interaction fingerprints and multilinear principal component analysis, and (c) explores, by stability analysis, the mechanisms of acquisition of the EGFR T790M mutation after progression to first line TKIs for metastatic mutant-EGFR NSCLC. In these studies, modeling methods are used to generate structures of EGFR with different mutations and molecular dynamics (MD) simulations are applied to reveal the dynamic state of the molecules. These data, for each EGFR mutant-TKI complex, are the foundation of our analysis, from which different properties of molecules were uncovered through a series of modeling and learning techniques.
Specifically, in (a), the relationship between the number of EGFR residues connected to gefitinib and the drug response level for each EGFR mutation were investigated. Three-dimensional trimmed Delaunay triangulation was applied to construct connections between EGFR residues and gefitinib atoms. The number of EGFR residues connected to gefitinib was then correlated with the drug response level of the corresponding EGFR mutation. In (b), structural protein-ligand interaction fingerprints (IFP) between each EGFR mutant and the binding of the EGFR TKI were examined. Based on the MD simulation data, a 3rd-order tensor was constructed for the IFP of each complex. Multilinear principal component analysis (MPCA) was employed for feature selection, and a classification task was performed to predict the drug response level. In (c), we performed attribute ranking to evaluate the correlation of each personal attribute and the presence of T790M after one or more line(s) of TKI therapy, with or without additional systemic therapy. Computational modeling and molecular dynamics (MD) simulations were employed to investigate the EGFR mutants and the stability of residues around T790 in each mutant-TKI complex was analyzed. These stability data were correlated with the probability of acquiring T790M, based on clinical observations of patients. Overall, these studies will lead to a better understanding of the dynamic features of EGFR mutant-TKI complexes and provide new insights to study and predict drug resistance/sensitivity in the treatment of NSCLCs that have EGFR mutations. Our work can also lead to a better understanding of the mechanism of acquiring the T790M mutation after targeted treatment and will be beneficial to the design of future targeted therapies and innovative drug discovery.