Despite the rapid advancements in targeted drug therapy in cancer treatment over the past two decades, drug resistance remains an issue. Point mutations can often explain the emergence of resistance, while in other cases, it is not clear why resistance occurs. Furthermore, the mechanisms involving the physical forces that contribute to the binding between the drug and the protein are poorly understood. The thesis aims to address these questions by using computer models, primarily quantum mechanical methods in an attempt to shed light on why resistance occurs in leukemia treatment.

We have used DFT to calculate binding energies for drugs in chronic myeloid leukemia (CML) and acute myeloid leukemia (AML),demonstrating that asciminib enhances the effect of nilotinib in T315I-mutated CML, thus allowing treatment with ponatinib to be avoided. In AML treatment, a combination of DFT, MD, NCI, and EDA have been used to show that the mutation R140Q/Q316E in trans in the enzyme IDH2 leads to resistance to enasidenib. The accuracy of various exchange-correlation functionals was evaluated against a limited dataset, yielding an accuracy of ~ 1.5 kcal/mol, highlighting the potential of simple DFT models. In one study, using DFT and MD, we were able to show that resistance when dasatinib binds to a mutated form of ABL1 arises due to a variety of factors, not just the breaking of a hydrogen bond.

Furthermore, EDA has been employed to analyze the dominant physical forces at the binding site, and an extension of this method (DFTB-EDA) has been developed to handle larger systems. Using FEP/REMD in combination with DFT, the binding of imatinib and dasatinib to various proteins has been analyzed to study off-target binding, with the aim of clarifying the drugs' toxicity. We observed that FEP/REMD tended to overestimate the binding energy, while DFT tended to under estimate it.

Overall, the thesis demonstrates that a combination of DFT, EDA, MD, NCI, and FEP is applicable to concrete problems in studying drug resistance in CML and AML, despite the limited accuracy of the methods.This methodology, however, is not limited to these cancer forms but canbe applied more broadly to other conditions.