A genetic comparison of epstein-barr virus-associated polymorphic lymphoproliferative disorder and diffuse large B cell lymphoma

Abstract

Abstract Background: Epstein-Barr Virus (EBV) is a potent transforming oncovirus causally associated with development of cancer including lymphomas. The transformative effects of EBV are enhanced in immunosuppressed patients where diminished host immune surveillance promotes shift in latency program, allowing EBV to express its full oncoviral genome. EBV-associated polymorphic lymphoproliferative disorders (EBV-LPD) are poorly defined in terms of diagnostic histopathologic criteria and lack standardized treatment approaches. Histopathologically and clinically, these lesions lie between hyperplasias and EBV-positive diffuse large B cell lymphomas (EBV-DLBCL). Predicting outcome and propensity for disease progression based on existing classification has been elusive and individual patient management can be haphazard. Here, we compare tumor genetics, immune cell repertoire, and microenvironment in EBV-DLBCL and EBV-LPD to identify features that distinguish these lesions. Design: EBV-DLBCL and EBV-LPD were identified by retrospective search of pathology databases and reviewed to confirm diagnosis. Formalin-fixed, paraffin-embedded (FFPE) diagnostic biopsies underwent RNA-seq and/or WES. Clinical and treatment data were extracted from electronic medical records. Results: 82 tumor samples (48 EBV-DLBCL, 34 EBV-LPD) from 59 patients were analyzed by RNA-seq (N=57) and/or WES (N=67). Samples included pre-treatment (N=62) and on-treatment (N=20) biopsies. All patients had at least one source of immunosuppression including HIV, autoimmune disease, past cancer diagnosis, transplant, or old age (>70y). Pre-treatment biopsies (one per patient) showed significant differences between EBV-DLBCL (N=22) and EBV-LPD (N=10) microenvironments. Deconvolution predicted EBV-LPD samples to have larger fractions of CD4⁺ and CD8⁺ T cells as well as monocytes, and reduced B cells (all FDR-corrected p<0.05). Comparison of single sample gene set enrichment analysis signatures confirmed enrichment of T-cell-associated processes in EBV-LPD, including T-cell activity and trafficking, checkpoint inhibition, and MHC class I antigen presentation (all FDR-corrected p<0.05). In contrast, EBV-DLBCL were enriched in expression of proliferation-related genes. Notably, EBV-LPD T-cell receptor (TCR) immune repertoires showed similar diversity compared to EBV-DLBCL TCRs (p=0.2) despite a 5.64-fold larger total clonotype count (p=5.2 x 10-5), suggesting non-specific clonal T-cell expansion potentially driven by EBV infection, rather than by immune recruitment alone. WES tumor data for a set of 543 genes recurrently mutated in B- and T-cell lymphomas, myeloid neoplasms, and pan-cancer contexts were analyzed. Mutations in these genes were significantly less frequent in EBV-LPD than in EBV-DLBCL. In pretreatment biopsies (one per patient), at least one mutation was detected in 54.5% of DLBCL vs 15.4% of LPD cases (p=0.054). The same trend was observed for all available samples (61% vs. 31%, p=0.031), indicating a modest and restricted role of canonical genetic drivers in the progression of EBV-DLBCL and EBV-LPD tumors, respectively. We also examined intra-patient heterogeneity in 11 cases with multiple tumor biopsies across spatial or temporal sites (6 EBV-DLBCL, 5 EBV-LPD) using mutation data from the focused gene set and overdominant BCR sequences, when available. For EBV-DLBCL, 4 of 6 cases showed concordant mutational and BCR patterns across sites. In contrast, only one LPD case (WES not available) showed BCR concordance; all others exhibited marked heterogeneity in both mutation profiles and dominant BCR clonotypes.Conclusions: Compared to EBV-LPD, EBV-DLBCL are characterized by tumor clonality, presence of known genetic drivers and more frequent concordant mutational and BCR patterns across serial samples, suggesting that analysis of these features, which is currently clinically available, may be useful in distinguishing these lesions from EBV-LPD. In contrast, the EBV-LPD lack canonical lymphoma mutations, show a complex immune-rich environment and lack restricted BCR patterns across serial samples. Although EBV-LPD and EBV-DLBCL microenvironments show similar diversity in T cell immune repertoires, LPD had a 5-fold larger clonotype count, suggesting a more robust local immune response potentially limiting expansion of EBV-infected B cell clones. Incorporation of analysis of the local immune response may be useful in classification of EBV lesions as LPD versus DLBCL.