Session: 621. Lymphomas: Translational – Molecular and Genetic: Poster II
Hematology Disease Topics & Pathways:
Research, Lymphomas, Translational Research, Genomics, B Cell lymphoma, Diseases, Aggressive lymphoma, Lymphoid Malignancies, Biological Processes, Technology and Procedures, Human, Omics technologies
Introduction
Primary central nervous system lymphoma (PCNSL) is a rare B-cell lymphoma with an aggressive course. We recently identified, double expression of BCL2 and MYC (DE) as a biomarker of aggressive disease in PCNSL (Poynton et al, Blood Advances 2024) but the molecular mechanisms underpinning the adverse prognosis of DE status remain unclear. Here, we leverage combined genomic, transcriptomic and spatial profiling to uncover the biological heterogeneity between DE- and non-DE PCNSL.
Methods
To evaluate the biological differences between DE- and non-DE PCNSL, we performed bulk RNAseq on 26 diagnostic PCNSL FFPE tumor samples to obtain gene expression profiles (GEP). These data were combined with prior bulk RNAseq data from 25 PCNSL cases (Fukumura et al, Acta Neuropathol. 2016). Spatial transcriptomic (ST) analysis using the Visium FFPEv2 was performed on a subset of six cases (three DE and three non-DE). The ST data was integrated with a publicly available PCNSL single cell RNAseq dataset (Heming et al, Genome Med. 2022) using cytoSPACE (Vahid et al. Nature Biotech. 2023) to infer cell-type composition and spatial mappings for each ST sample. Whole exome sequencing (WES) analysis was performed on 63 PCNSL cases. DE status was established by immunohistochemistry (IHC) and by RNAseq gene expression profiling.
Results
We first determined the correlation between RNAseq normalised expression of MYC and BCL2 and IHC MYC and BCL2 positivity in a subset of samples with known MYC and BCL2 IHC status and defined RNAseq expression thresholds for determining MYC and BCL2 status which minimised false assignment. Unsupervised GEP clustering of 51 cases (16 DE, 35 non-DE) revealed that DE-PCNSLs primarily clustered together. Differential GEP analysis demonstrated 89 differentially expressed genes (19 upregulated, 70 downregulated) between DE- and non-DE PCNSL cases. Downregulated genes in DE-PCNSLs included genes involved in tumor-microenvironment regulation (BMP7, NDRG4, PPP2R2C and SAMD4A), FOXO3, a suppressor of c-MYC driven lymphomagenesis and PCLO, a component of the presynaptic cytoskeletal matrix which is recurrently mutated in DLBCL and PCNSL.
To determine the spatial patterns of gene expression between DE- and non-DE PCNSL tumors, we performed spatial transcriptomic analysis on six cases (three DE-PCNSLs, three non-DE PCNSLs). Cell type specific localization of tumoral B-cells, non-tumoral B-cells, oligodendrocytes, myeloid cells and T-cells were deconvolved by mapping a PCNSL scRNAseq dataset (n = 13,560 cells) onto spatial transcriptomic data using cytoSPACE. Overall proportions of non-tumoral B-cells, T-cells and myeloid cells were similar between DE- and non-DE PCNSL tumors. As each ST spot concurrently captures GEP of a number of cells, we observed tumoral B-cell-containing spots from DE-PCNSL cases contained a significantly smaller proportion of non-tumoral B-cells (15.9% vs. 23.3%, p<0.01), a significantly smaller proportion of myeloid cells (5.7% vs. 8.5%, p<0.01) but similar proportions of T-cells (4.4% vs. 4.2%, p=0.66) compared to non-DE PCNSL cases. As expected, expression of CD37, MYC and BCL2 were strongly localized to spatial regions rich in tumor B-cells. In DE-PCNSL cases, expression of BMP7 was globally reduced compared to non-DE PCNSL cases and spatially segregated away from tumor B-cell rich regions.
Tumor mutation burden was similar between DE- and non-DE PCNSL cases (3.39/MB vs 2.88/MB, p=0.41). Compared to DE-negative cases, DE-positive cases had higher rates of KMT2D (55% vs. 20%, p=0.04), particularly KMT2D nonsense mutations (17% vs 3%), and IGLL5 mutations (50% vs. 11%, p<0.01). DE- and non-DE cases had similar rates of MYD88, PIM1 and CD79B mutation. PCLO was mutated in 28% of cases with similar mutation rates between DE and non-DE cases.
Conclusions
Our multi-modal approach provides insights into the genetic and molecular differences between DE- and non-DE PCNSL. Distinct patterns of GEPs and spatial proximity of immune cell infiltration rather than significantly different genomic landscapes between these two groups may, in part, underpin the adverse prognostic impact of DE-PCNSL.
Disclosures: Fox: SOBI: Consultancy; Lilly: Consultancy; Atarabio: Consultancy; AstraZeneca: Consultancy; AbbVie: Consultancy, Other: Trial Steering Committee, Research Funding; Janssen: Consultancy; Roche: Consultancy; BeiGene: Research Funding; Takeda: Consultancy; SERB: Consultancy; Ono: Consultancy; MorphoSys: Consultancy; Incyte: Consultancy; Gilead/Kite: Consultancy; Genmab: Consultancy, Other: Trial Steering Committee, Research Funding; BMS: Consultancy. Okosun: BeiGene: Research Funding; Incyte: Consultancy; Genmab: Membership on an entity's Board of Directors or advisory committees, Research Funding; AstraZeneca: Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding.
See more of: Oral and Poster Abstracts