Session: 632. Chronic Myeloid Leukemia: Clinical and Epidemiological: Poster III
Hematology Disease Topics & Pathways:
Translational Research, Non-Biological, Computational Biology, Therapies, Pharmacology, Technology and Procedures, Machine Learning
The oncogenic BCR-ABL1 tyrosine kinase is the driver of chronic myeloid leukemia (CML) and Philadelphia chromosome-positive acute lymphoblastic leukemia (Ph+ALL). Tyrosine kinase inhibitors (TKIs) targeting ABL kinase are generally effective, but subsets of patients treated with single-agent TKIs develop resistance due to mutations in BCR-ABL1 that impair TKI binding. We have previously reported that BCR-ABL1 compound mutants (exhibiting two mutations within the same BCR-ABL molecule) that include the T315I gatekeeper mutation confer a high degree of resistance to all clinical ABL TKIs used as single agents, including ponatinib and the allosteric inhibitor asciminib. However, combining asciminib with ponatinib provides an effective strategy for overcoming compound mutation-based resistance (Eide et al. Cancer Cell 2019). As the clinical utility of ponatinib is limited by cardiovascular toxicity, including arterial occlusive events (AOEs), we decided to search for alternative molecules for use in combination with asciminib.
To identify functional ponatinib analogs, we performed Quantum Similarity Modeling (QSM) on the reported crystal structure of T315I mutant ABL1 kinase in complex with nilotinib and asciminib (5MO4) (Wylie et al. Nature 2017) to search for other molecules. Compared to conventional computational modeling, QSM identifies novel classes of structurally distinct compounds that are comparable on a quantum level by precisely defining their interaction with the target. Affinity inferred by close complementarity with the shared ligand-protein surface in the region of the surveyed binding site is mapped, using multiple weak local associations. Our in silico QSM platform combines quantum methods with machine learning to investigate extensive chemical spaces. We screened several million compounds against BCR-ABL1 and identified 51 potential candidates predicted effectively to block T315I mutant BCR-ABL1 when combined with asciminib. To prioritize potent and non-toxic drug combinations for further development against compound mutants, we initially profiled all 51 compounds for their efficacy against Ba/F3 BCR-ABLT315I cells, alone and in combination with asciminib (1 nM). Of 51 compounds, LY3009120, a pan-RAF inhibitor that is currently in phase I clinical development for advanced solid malignancies (Sullivan et al. Mol Cancer Ther 2020), showed strong activity against BCR-ABLT315I when combined with asciminib. These data provided proof of principle for the QSM approach.
We next tested the efficacy of all 51 candidates ± asciminib against Ba/F3 cells harboring T315I-inclusive BCR-ABL1 compound mutants, including Y253H/T315I, E255V/T315I, H396R/T315I, G250E/T315I, and T315L as the most resistant mutants. Neither single agent showed any effect. However, LY3009120 strongly inhibited BCR-ABL1 compound mutants when combined with asciminib. No toxicity was observed against Ba/F3 parental cells, confirming that the effects of the combinations are mediated by inhibition of BCR-ABL1. Synergy quantification of the dose-response matrix for the LY3009120/asciminib combination using the Zero Interaction Potency model demonstrated highly synergistic interactions (Synergy score > 10) between the two inhibitors. To directly assess the binding affinity of LY3009120 to the ABL1 kinase domain, we used the cell-based NanoBRET intracellular ABL1 kinase assay on HEK-293 cells expressing luciferase-tagged ABL1. The NanoBRET assay uses energy transfer to quantify the affinity of test compounds by competitive displacement of a cell-permeable fluorescent tracer that is reversibly bound to an ABL1-luciferase fusion protein. We found that LY3009120 competes off the fluorescent tracer at a low micromolar range (EC50 = 0.75 μM), confirming direct binding of LY3009120 to the kinase domain of ABL1. We hypothesize that the binding of LY3009120 to the ABL1 kinase domain induces a conformational change that re-establishes asciminib binding to the myristoyl binding pocket, allowing for synergy. Studies to quantify the binding affinity of LY3009120 and asciminib to BCR-ABL1 mutants are underway, and data will be presented.
In summary, our findings validate QSM as a novel in silico approach to identify TKI combinations. Combining LY3009120 with asciminib may be an effective, low-risk strategy to target BCR-ABL1 compound mutants in patients with clinical TKI resistance.
Disclosures: Deininger: SPARC, DisperSol, Leukemia & Lymphoma Society: Research Funding; Sangamo: Consultancy, Membership on an entity's Board of Directors or advisory committees; Incyte: Consultancy, Honoraria, Research Funding; Fusion Pharma, Medscape, DisperSol: Consultancy; Novartis: Consultancy, Research Funding; Blueprint Medicines Corporation: Consultancy, Membership on an entity's Board of Directors or advisory committees, Other: Part of a Study Management Committee, Research Funding; Takeda: Consultancy, Membership on an entity's Board of Directors or advisory committees, Other: Part of a Study Management Committee, Research Funding.
See more of: Oral and Poster Abstracts