Potential biomarkers and mechanisms in central nervous system involvement of KMT2A rearranged b-ALL: Insights from  Patient-derived xenograft models

Abstract

Abstract Background: Acute lymphoblastic leukemia (ALL) is the most common pediatric cancer, accounting for nearly 70% of childhood leukemia cases, and is driven by genetic mutations that cause uncontrolled proliferation of immature lymphoid cells. Among its subtypes, KMT2A-rearranged B-ALL (KMT2A-r B-ALL) is associated with a high risk of central nervous system (CNS) involvement and poor prognosis, particularly in infants under 1 year old. Despite therapeutic advances, CNS relapse and drug resistance remain major clinical challenges, and overall survival in this subgroup remains low. Currently, no reliable biomarkers exist to predict CNS involvement, and the underlying molecular mechanisms are poorly understood. Recent studies have suggested that integrin signaling and extracellular matrix (ECM) interactions may be involved in CNS infiltration, with mechanisms such as integrin α6-laminin binding, osteolysis, and CXCR4-mediated chemotaxis. However, our RNA sequencing of 145 Taiwanese B-ALL patients found no significant link between integrin-related genes (e.g., ITGA6, ITGA9) and CNS involvement, indicating that other mechanisms may be involved. Moreover, literature has shown leukemic cells infiltrating the subarachnoid space in PDX mice, consistent with autopsy findings of arachnoid involvement in over half of pediatric ALL CNS infiltration cases.Aims: This study aims to identify genetic signatures/biomarkers and associated mechanisms responsible for CNS involvement in pediatric B-ALL, with a specific focus on the KMT2A-r subtype.Methods: Due to limited access to CNS metastatic samples, we used patient-derived xenograft (PDX) models as a tool to mimic ALL CNS involvement. Bone marrow leukemic cells from 11 pediatric KMT2A-r B-ALL patients—including 6 with CNS involvement at diagnosis and 5 without—were intravenously transplanted into NSG™ mice (n > 100 in total) to establish PDX models. CDX (cell line-derived xenograft) models using the human KMT2A-r B-ALL cell line SEM were investigated in parallel. Flow cytometry was used to assess engraftment rates in peripheral blood, spleen, and bone marrow. Whole-head tissue sectioning, CT scans, and MRI imaging of mouse brains were used to monitor CNS infiltration patterns. Bulk RNA sequencing was performed on CNS and bone marrow samples from both PDX and CDX mice. Large-scale bioinformatics analyses using KEGG and Gene Ontology (GO) were performed to identify differentially expressed genes. qRT-PCR, shRNA knockdown (KD) in SEM, and literature-based data mining were performed to validate differentially expressed genes. KD SEM cells expressing luciferase were further implanted to validate relevant genes involved in CNS invasion in vivo. Results: Using the KMT2A-r B-ALL cell line SEM, we delineated the timeline of leukemic cell migration from the calvarial bone marrow into the subarachnoid space and the onset of neurological symptoms. Peak invasion time was 19–21 days after leukemic cell engraftment, and survival endpoints were 35–42 days, with or without neurological symptoms. Histopathological analysis revealed leukemic cells infiltrating the subarachnoid space. Flow cytometry confirmed extensive hCD19+ SEM cells, accounting for 99.36% ± 0.62 of total lymphocytes (mCD45+ and hCD19+ cells, n = 13). In PDX models, survival endpoints were 70–175 days, with or without neurological symptoms. Transcriptomic analysis of PDX CNS and BM samples from 2 cohorts revealed significant enrichment of genes/pathways related to focal adhesion, ECM-receptor interaction, and integrin signaling. By integrating data from the PDX cohorts with qRT-PCR results from SEM CDX models, we identified seven differentially expressed genes with consistent expression patterns—ITGA2B, GPR34, LTF, TTR, SALL1, OLFML3, and TREM2—involved in cell migration, immune modulation, and neuroinvasion. KD of GPR34 in SEM-luciferase+ cells reduced CNS involvement in NSG mice (p < 0.001), suggesting the role of GPR34 in KMT2A-r B-ALL CNS invasion.Conclusion: We combined CDX and PDX models with transcriptomic profiling and functional validation to systematically identify seven candidates as biomarkers and propose potential mechanisms underlying CNS involvement in KMT2A-r B-ALL. Our data may serve as a foundation for developing precision medicine strategies for high-risk KMT2A-r B-ALL patients.