Combined Targeting Patient-Specific Anti-Apoptotic Molecules and CXCR4-Expressing CAR-T Cells Eliminated High-Inflammation State Leukemia with Poor-Prognostic Mutations

Despite recent advances in molecular specific targeting drugs, prognosis is dismal for relapsed and refractory leukemia with multiple genetic aberrations. We aimed to create therapeutic intervention for high-risk leukemia with poor-prognostic mutations by identifying patient-specific vulnerabilities. To this end, we designed translational research connecting multi-omics analyses, in vitro screening and in vivo using patient-derived xenograft (PDX) models. With xenogeneic transplantation, we first identified leukemia-initiating cell fractions in each patient specimen followed by global gene expression profiling resulting in potential therapeutic targets in genetically-diverse lymphoid and myeloid leukemia. Based on the functional genomics, we set up in vitro treatment with small molecules targeting anti-apoptotic molecules (BCL-2, MCL-1 and BIRC) and molecule related to cell division (AURKB) in 28 B-ALL/MPAL, 9 T-ALL/MPAL and 12 CML cases. There was leukemia lineage-specific responsiveness to targeted inhibition. We found high responsiveness of T-ALL/MPAL to venetoclax (88.9%, eight out of nine samples) and high sensitivity to BIRC inhibitor (AZD5582) in CML (83.3%, 10 out of 12 samples). B-ALL/MPAL showed overlapping responsiveness to BCL-2 and BIRC inhibition. We observed differential sensitivity of small molecules inhibitors between leukemia-lineages even in the presence of identical driver genetic alterations such as BCR-ABL1 translocation, KMT2A-rearrangement and KRAS mutation. Consistent with in vitro sensitivity, we observed highest expression of BCL2 in T-ALL/MPAL cells compared with B-ALL/MPAL and CML by single RNA sequencing. Based on in vitro sensitivity, we performed in vivo treatment with AZD5582 and/or venetoclax combined with conventional therapy using PDX mice. We observed complete elimination of leukemic cells in 8 of 11 B-ALL/MPAL, 3 of 4 T-ALL/MPAL and 2 of 4 CML. However, six cases failed to achieve profound remission. We examined potential relation between mutational profile and in vivo resistance by target DNA sequencing. Each of the six resistant cases harbored mutations in either KRAS and/or TP53. We found specific profiles of serum inflammation cytokines in KRAS-mutated or TP53-mutated cases compared with WT cases. To elucidate if this inflammation state could be regulated by metabolic change in leukemic cells, we performed non-target lipidomics. We found more abundance of poly-unsaturated fatty acids (PUFA) in leukemic cells than normal cells. There was lineage-specific signature of phospholipid class among AML, CML, B-ALL and T-ALL. While AML cells had a large amount of arachidonic acid (AA) containing phospholipids, B-ALL cells were abundant in docosahexaenoic acid (DHA) containing phospholipids. In addition, TP53-mutated leukemic cells were enriched in AA containing phospholipids compared with TP53-WT leukemic cells, which may promote inflammation pathway. To further identify in vivo resistant mechanism, we examined the expression profile of 31 proteins, including apoptotic molecules, immune modulators and cell surface markers by mass cytometry. Protein expression profiling revealed in vivo resistant cases exhibited higher expression of CXCL12, a ligand for chemokine receptor CXCR4, compared with sensitive cases. To overcome KRAS-mutated or TP53-mutated leukemic cells with high-inflammation state, we created CXCR4-expressing CAR-T cells targeting CD19 for B-ALL and CD7 for T-ALL and CML. Reduction of leukemia burden followed by CXCR4 CAR-T cells eradicated leukemic cells in bone marrow and spleen without severe cytokine release syndrome. Altogether, our patient-specific therapeutic strategy may contribute to precision medicine approach and improve clinical outcome in poor prognosis leukemia.