Evolution of Gilteritinib Resistance from Residual Disease to Relapse

Activating mutations in the tyrosine kinase receptor FLT3 are observed in ~35% of all acute myeloid leukemia (AML) cases. Multiple FLT3 inhibitors are currently in clinical development and while most patients initially respond well to FLT3 inhibition, resistance inevitably develops in a period of months. During the initial response to FLT3 inhibitors, residual leukemia cells are able to survive and persist in the marrow microenvironment, which facilitates early resistance. Over time, leukemia cells develop intrinsic mechanisms of resistance, that are not dependent upon the microenvironment, which leads to late resistance and disease relapse.

In this study, we utilized a two-step model to study the temporal evolution of gilteritinib resistance. To model initial extrinsic early resistance, we cultured the FLT3-ITD⁺ AML cell lines, MOLM-14 and MV4;11, with ligands secreted by the marrow microenvironment. We previously reported that fibroblast growth factor 2 (FGF2) is secreted from marrow stromal cells and protects FLT3-ITD⁺ AML cells from quizartinib, a FLT3 inhibitor. FGF2 binds receptor FGFR1, activates MAPK signaling, and promotes ligand-dependent growth (Traer et al. Cancer Res. 2016). We also used FLT3 ligand (FL), which has been shown to reactivate the FLT3 receptor despite the presence of inhibitor. MOLM14 cells were cultured with 100 nM gilteritinib in media alone, or supplemented with 10 ng/ml FGF2 or FL. After 7 weeks, all cultures supplemented with ligand eventually resumed growth (early resistance), whereas MOLM-14 cells without ligand were unable to resume growth.We then removed ligand, which transiently restored sensitivity to gilteritinib. However, after 2 months the cells resumed exponential growth in the absence of extrinsic factors (late resistance). A similar pattern was observed with MV4;11 cells. We analyzed the mechanisms of early and late resistance using a number of orthogonal tools: whole exome sequencing (WES), genome-wide CRISPR/Cas9, proteomics, metabolomics, and small-molecule inhibitor screening.

WES identified NRAS mutations in the majority of late resistant cultures (13/15) and a gatekeeper FLT3 mutation was found in 1, consistent with mutations found in patients treated with gilteritinib on the Admiral trial (McMahon et al., Cancer Discov., 2019). To identify if NRAS mutations were pre-existing, ddPCR was used. NRAS mutations were detected at low level (<0.1%) in early resistant cultures and parental cells, but did not become the dominant mechanism of resistance until the ligand was removed. Moreover, when NRAS mutations were stably expressed in MOLM14 cells, cells remained sensitive to 100 nM gilteritinib and required over a month of continuous culture to become resistant and resume growth, indicating that NRAS mutations alone are not sufficient for gilteritinib resistance.

We further analyzed both early and late resistant lines with genome-wide resensitization CRISPR/Cas screening and global and phospho-proteomics. As expected, CRISPR/Cas screening revealed that NRAS was important for late resistance, which was confirmed by proteomics. In contrast, early resistance revealed multiple hits in metabolic and cell cycle pathways by CRISPR/Cas and global and phospho-proteomics. The unique metabolic signatures (lipid signaling in particular) were verified by metabolomic analyses. CDK proteins were significantly decreased in early resistance, as was the cell cycle. Aurora Kinase B protein (AURKB) was also increased in early resistance, and cells were uniquely sensitive to AURKB small molecule inhibitors and genetic deletion.

We then evaluated primary leukemia cells from 11 patients before and after 1-2 months of gilteritinib to evaluate early resistance. Leukemia cells were isolated by CD33 and CD34 beads and subjected to a targeted proteomic analysis (Figure 1). In agreement with our cell line model, early resistant cells had a distinct protein profile, with significant alteration of cell cycle and lipid metabolism proteins after gilteritinib treatment. Primary patient samples also demonstrated robust sensitivity to AURKB inhibitors only after gilteritinib exposure. Our results suggest that developing drug combinations that selectively target early resistance can improve the depth of initial response and may block development of later resistance mutations, thus improving the durability of response to gilteritinib.