Novel Promising Therapeutic Strategies for Advancing the Treatment of KMT2A-Rearranged AML

BACKGROUND. In recent years, cancer research on genome and transcriptome sequencing tuned the scenario that was previously conceived as targetable in pediatric acute myeloid leukemia (AML), but the actual dearth of new agents approval underlined that performing robust preclinical studies for drug advance still represents a challenge. The optimization of disease models results crucial to prioritize innovative treatments.

METHODS. For this study, we established and characterized AML patient-derived xenografts (PDXs) by immunophenotype, RNA-sequencing and whole-exome sequencing (WES). According to somatic mutations, we determined AML clonal dynamics from patient’ AML to PDXs. In parallel, by deconvolution approach based on single-cell reference transcriptomic profiles and by Gene Set Enrichment Analysis, we dissected RNA-seq data. Based on results emerging from omics data, we selected targeted drugs, tested on a three-dimensional (3D) culture system consisting in a biomimetic scaffold co-cultured with blasts and mesenchymal stromal cells (MSCs), and in AML-PDXs.

RESULTS. We generated 26 AML-PDXs representatives of 14 different AML genetic subtypes. AML cells propagated in PDXs resemble the original AML features in terms of immunophenotype, variants, and transcriptome, offering a comprehensive view of the disease complexity. By WES, we confirmed a high intra-tumoral heterogeneity, we recognized AML "founder" clones characterized by an average of 22 variants (at least one in AML driver genes) always been maintained during PDX passages and we also identified smaller clones, recording the majority of variants in NRAS, CREBBP, FAT1, TET2, FLT3, NOTCH1 and KDM5A genes. Overall, the inference of the evolutionary clonal trajectories informed about the clones-variants to be targeted for their potential to sustain AML relapse. We prioritized peculiar variants and transcriptomic signatures for pharmacological targeting. We selected Asparaginase (ASPN), IACS-010759 (IACS), Disulfiram, Gallein, 5-Azacytidine, Quizartinib, Trametinib, γ-Secretase inhibitor and ICG-001 to be tested either alone or in combination with Venetoclax (VEN) in 3D with ex vivo AML cells and AML patient derived MSCs. We highlighted ICG-001, a CBP/β-catenin inhibitor blocking the β-catenin dependent gene transcription, for its ability to reduce AML proliferation in 3D and the combinations VEN+IACS and VEN+ASPN being highly synergistic in KMT2A-rearranged AML (p<0.001). We conducted phase II-like preclinical investigations on six KMT2A-rearranged AML-PDXs treating them for 4 weeks with VEN+IACS or VEN+ASPN. Only 3 out of 6 models treated with VEN alone underwent remission, whereas it was observed in 6 out of 6 models treated with either of the 2 combinations. At day +15 after stop therapy, 6/6 models treated with VEN relapsed, in comparison with only 1 model treated with VEN+IACS and 1 with VEN+ASPN. Forty-five days after the end of combination treatments 4/6 models persisted in remission. Finally, both novel combinations promoted a significantly prolonged PDXs survival compared to mice treated with VEN alone.

CONCLUSIONS. Overall, AML-PDXs capturing AML heterogeneity allow to target cancer cells in their multiple aspects and to determine a patient-specific drug profiling increasing the chance to treat leukemia. We support the use of pediatric AML-PDXs in preclinical testing for their ability to mimic the heterogeneity of drug response, aiding in the identification of tailored second-line treatments. VEN+ASPN combination represents a novel promising therapeutic strategy for advancing the treatment of KMT2A-rearranged AML.