Lymphly的整合应用揭示B细胞淋巴瘤间共享的分子图谱

Background: We previously created Lymphly, a hierarchical framework that uses discrete genetic events to enhance diffuse large B-cell lymphoma (DLBCL) classification. However, the molecular relationships between DLBCL subtypes and the broader B-cell lymphoma landscape remain poorly delineated. Here, we evaluated if Lymphly can elucidate molecular patterns across B-cell lymphomas to inform therapeutic development and trial design. Methods: Lymphly was used to classify a meta-cohort of ~1,500 DLBCL samples and ~2,000 samples with available molecular data across multiple non-DLBCL B-cell lymphoma diagnoses from internal and publicly available datasets. The samples were assigned canonical (EZB, MCD, BN2, N1) and newly identified subtypes (JS3, JS6), as well as TP53+ and MYC+ statuses. Integrative analyses of molecular and clinical parameters highlighted distinct genomic architectures and risk profiles that may constitute potential therapeutic targets. Results: Lymphly revealed shared genetic and transcriptional features between non-DLBCL lymphomas and specific DLBCL subtypes in the study cohort, indicative of convergent molecular phenotypes between specific Lymphly subtypes and lymphoma diagnoses. Specific Lymphly subtypes dominated certain diagnoses, and samples within each subtype shared similar gene expression profiles. Subtype EZB was enriched among follicular (FL) and Burkitt (BL) lymphoma samples, reflecting their germinal-center origin (GCB). Many of the FL samples also showed BCL2 translocations, which were less common in BL samples. Subtype BN2, featuring mutations affecting NOTCH2 and NF-κB signaling, was enriched among marginal zone and splenic marginal zone lymphoma samples. Subtype MCD, featuring alterations in MYD88 and CD79B, was enriched in primary central nervous system lymphoma and high-grade B-cell lymphomas. Newly classified subtypes JS6 and JS3 revealed additional links between DLBCL and other B-cell lymphoma entities: JS6, a GCB-like entity, was linked to primary mediastinal B-cell lymphoma, while JS3, an activated B-cell-like entity, overlapped with T-cell/histiocyte-rich large B-cell lymphoma and plasmablastic lymphoma. Conclusions: Lymphly demonstrated applicability beyond DLBCL, capturing canonical and newly identified subtypes and shared molecular patterns across B-cell malignancies. Applying DLBCL genetic classification to other B-cell lymphomas revealed potential trajectories of transformation into DLBCL, with Lymphly subtypes reflecting their natural molecular counterparts within the DLBCL spectrum. By integrating an interpretable and a rule-based classification with cross-disease molecular insights, Lymphly is poised to translate these insights into data-driven strategies that optimize decision-making across all stages of diagnostic and treatment development.