MULTI-OMIC PROFILING OF DARK ZONE LYMPHOMAS UNCOVERS DYSREGULATED GENE REGULATORY NETWORKS REFLECTING DZ-LZ B-CELL TRANSITION DYNAMICS

Abstract

Introduction: Dark zone (DZ) lymphoma represents a subset of aggressive B-cell lymphomas characterized by a gene expression signature (DZsig) that resembles B cells from the DZ of the germinal center. This category includes Burkitt lymphoma (BL), most high-grade B-cell lymphomas with MYC and BCL2 rearrangements (HGBCL-DH-BCL2), and ∼15% of germinal center B-cell–like (GCB) cell-of-origin (COO) diffuse large B-cell lymphoma (DLBCL). DZsig+ GCB-DLBCL and HGBCL-DH-BCL2 are associated with poor outcomes, yet the biological mechanisms driving these outcomes are poorly understood. We hypothesize that their shared phenotype results from hijacking normal DZ B-cell programs. To investigate this hypothesis, we examined the biology of malignant B cells and the tumor microenvironment (TME) in DZsig+ lymphomas using disaggregated single nucleus (sn) multiome sequencing and CosMx spatial molecular imaging (SMI).

Methods: DZsig and COO status was assigned by digital gene expression profiling. snMultiome sequencing was performed on 7 DZsig+ and 7 DZsig− biopsies. ArchR was used for transcription factor (TF) motif analysis, while SCENIC+ inferred enhancer-driven regulon (eRegulon) activities. Immunohistochemical (IHC) analysis of key TFs was used to validate findings in a cohort of 38 DZsig+ and 112 DZsig− biopsies of DLBCL morphology. To extend these findings, 749 biopsies were analyzed by CosMx SMI on tissue microarrays using a custom 6375-gene probe set. Whole-cell segmentation was performed using Mesmer. Community analysis was performed using spatial graphs constructed via Delaunay triangulation with edge-length thresholding.

Results: High FOXO1 and low B-cell receptor (BCR)/CD40-mediated eRegulon activities (NFκB1, REL, RELB, IRF4) emerged as defining features of DZsig+ lymphomas. These findings are supported by elevated FOXO1 motif accessibility and reduced NFκB1 and NFκB2 accessibility in DZsig+ lymphomas. Consistently, nuclear FOXO1 IHC positivity in our validation cohort was significantly higher in DZsig+ lymphomas and nuclear NFκB1 and NFκB2 IHC positivity was significantly lower in DZsig+ lymphomas compared to DZsig− GCB-DLBCL. SMI profiling revealed that DZsig+ lymphomas have an “immune cold” TME, contrasting with the “immune hot” TME of other GCB tumors. Community-level analysis further revealed a dichotomy, with DZsig+ lymphomas enriched in B-cell abundant clusters, while DZsig− GCB-DLBCL harbored diverse clusters composed of infiltrating T cell, macrophages, and stromal cells.

Conclusion: These findings reveal divergent phenotypes mimicking different states within the GC DZ-light zone (LZ) continuum, where FOXO1 drives the DZ phenotype, and BCR/CD40-mediated signaling suppresses FOXO1 activity in the LZ. In DZsig+ lymphomas, this transition is skewed, with increased FOXO1 and decreased NFκB motif accessibility, sustaining a DZ-like state. An “immune cold” TME suggests the existence of cell-intrinsic survival mechanisms in DZsig+ lymphomas.

Research funding declaration: Research funding from TFRI PPG #1108—New Frontiers Program Project Grant and BC Cancer Foundation

Keywords: bioinformatics; computational and systems biology; genomics, epigenomics, and other -omics; aggressive B-cell non-Hodgkin lymphoma

Potential sources of conflict of interest:

D. W Scott

Consultant or advisory role: Consultancy for Abbvie, AstraZeneca, GenMab, Kite/Gilead, Roche and Veracyte

Other remuneration: Research funding from Roche/Genentech