Diffuse large B-cell lymphoma: a biologically heterogenous disease
Diffuse large B-cell lymphoma has long been classified as a single disease, but mounting evidence shows that it represents a biologically diverse group of lymphomas with distinct molecular identities. Using single-cell RNA sequencing and ATAC-seq across more than 100 diffuse large B-cell lymphoma (DLBCL) biopsies, the authors explored how genetic alterations translate into gene-expression phenotypes and regulatory programs at single-cell resolution. The study found that established genetic subtypes correspond to reproducible transcriptional phenotypes. Broadly, DLBCL subtypes align with either an activated B-cell–associated group, including MCD, A53, BN2, and N1, or a germinal center B-cell–associated group, including EZB and ST2. Rather than being defined only by mutations, each subtype exhibited a characteristic gene-expression program and transcription factor network that resembled specific stages of normal B-cell differentiation. Beyond subtype classification, the study revealed extensive intratumoral heterogeneity. Most tumors contained multiple genetic subclones with distinct phenotypes, showing that a single biopsy may contain cells in different biological states. These states fell into recurring gene-expression themes related to B-cell differentiation—plasma-cell transition, memory B-cell identity, cell-cycle activity, and MYC-associated cell growth and metabolism. Importantly, these themes varied, independently of subtype, suggesting that genetic subtype alone does not fully capture tumor behavior. Specific genetic alterations were shown to influence these phenotypic programs. For example, CDKN2A loss promoted proliferative programs, while REL amplification suppressed memory B-cell differentiation through NF-κB signaling, demonstrating how genomic changes can drive cellular state rather than simply mark it. Integrating chromatin accessibility and transcriptional data allowed the authors to propose a unifying model in which DLBCL biology is organized along the differentiation axes germinal center B cell, memory B cell, and plasma cell. These axes are controlled by transcription factors that normally regulate B-cell development, including FOXO1, MEF2B, IRF4, PRDM1, BATF, KLF2, and STAT1. Genetic alterations push malignant cells toward one of these regulatory states, creating stable but distinct transcriptional identities. This framework helps explain why DLBCL tumors with different mutations can display predictable patterns of treatment response. The subtype-specific gene signatures reproduced known outcome differences and may help classify tumors that cannot be assigned confidently using only genetic algorithms. In addition, the identification of phenotypically distinct subclones suggests that therapeutic resistance or disease evolution may reflect shifts along these differentiation axes rather than only acquisition of new mutations. Overall, this study reframes DLBCL as a dynamic ecosystem rather than a static diagnostic category. By linking genetic subtype, gene regulation, and cellular differentiation at single-cell resolution, it provides a clearer conceptual framework for understanding lymphoma heterogeneity and offers a foundation for future biologically driven treatment strategies.
Wang B, Wright G, Enssle JC, et al. Axes of biological variation in diffuse large B cell lymphoma. Cancer Cell. 2026. doi.org/10.1016/j.ccell.2025.12.015
Correspondence: Dr. Louis M. Staudt at [email protected]