Abstract P063

Genetic Clustering Reveals Biologic Similarities Across subgroups of Hodgkin Lymphoma, Primary Mediastinal Large B-Cell Lymphoma and Large B-Cell Lymphoma patients

Background: Primary mediastinal large B-cell lymphoma (PMBL) occupies a unique position in the spectrum of aggressive B-cell lymphomas, sharing molecular and clinical features with both Hodgkin lymphoma (HL) and diffuse large B-cell lymphoma (DLBCL) (Rosenwald et al., 2003; Savage et al., 2003; Savage, 2022; Duns et al., 2021). Here we use a data-driven approach integrating all three entities in a unified analytical framework to understand genetic similarities between subgroups of all three diseases.

Methods: Genetic alterations (SNPs, small InDels, CNVs) were assessed by circulating tumor DNA sequencing of HL patients at baseline and relapsed/refractory DLBCL patients. Genetic alterations of PMBL patients were analyzed by tissue DNA sequencing. Non-negative Matrix Factorization (NMF) clustering was used as the primary analytical framework.

Results: NMF clustering suggests that genetically coherent subgroups can be formed that are independent of disease entity. One cluster is defined by high weights for genes involved in JAK/STAT signaling and immune evasion, most prominently SOCS1, STAT6, B2M, TNFAIP3, NFKBIE, GNA13, ITPKB, CIITA, CD58, CD83, XPO1, and histone linker genes (HIST1H1B/C/E), and is strongly enriched for PMBL and HL cases. This signature aligns with constitutive NF-κB and JAK/STAT activation, impaired MHC class I antigen presentation, and a pronounced immune privilege phenotype shared by PMBL and HL (Mottok et al., 2019). A second cluster is characterized by TP53, EZH2, FAT4, CREBBP, DDX3X, EP300, MYC, P2RY8, DTX1, and CD79A/B, a mutation-driven landscape more consistent with GCB-type DLBCL biology (Alizadeh et al., 2000; Li et al., 2018) although including some samples of all three disease entities.

Conclusion: Unsupervised genomic clustering across PMBL, HL, and DLBCL reveals biologically coherent subgroups transcending established diagnostic boundaries.