CELL INTELLIGENCE · 肿瘤细胞治疗研究

CD19-directed chimeric antigen receptor (CAR) T-cell therapy has transformed the treatment of relapsed or refractory large B-cell lymphoma (LBCL), including diffuse large B-cell lymphoma (DLBCL), yet primary resistance and relapse remain frequent. Clinical studies link pretreatment B-cell gene expression signatures, CD19 gene expression, malignant-cell CD19 protein expression, immune contexture and myeloid/inflammatory features with CAR T-cell outcomes, but they do not establish that myeloid cells directly induce loss of B-cell identity. In this Mini Review, we use B-cell identity remodeling as a DLBCL-focused, hypothesis-generating research construct describing concordant quantitative or spatially heterogeneous attenuation across B-lineage antigens and transcriptional programs without requiring frank lineage conversion; isolated CD19 attenuation alone is considered antigen attenuation rather than sufficient evidence of broader identity remodeling. We explicitly distinguish direct clinical associations derived largely from broader LBCL cohorts, indirect biological support from DLBCL single-cell and spatial studies, and untested mechanisms through which myeloid or stromal signals might select, stabilize or protect B-cell-identity-low states. Proposed biomarkers are framed as candidate research variables rather than validated clinical thresholds, with PAX5 distinguished from the more context-dependent role of IRF8. Finally, therapeutic hypotheses are organized by antigen dependence and biological compartment, and separate workflows are outlined for mechanistic discovery, prospective translational testing and future clinical implementation. This framework is intended to sharpen testable questions rather than to claim an established causal myeloid-B-cell identity axis.