Whole-genome analysis reveals shared early genetic alterations and subtype-specific immune evasion pathways in intravascular large B-cell lymphoma and immune-privileged large B-cell lymphoma

Abstract

Abstract Introduction Intravascular large B-cell lymphoma (IVLBCL) is a rare subtype of aggressive lymphoma characterized by selective growth of large malignant cells within small vessels in multiple organs. Our previous report has showed the enrichment of tumor-derived cell-free DNAs (cfDNAs) in plasma of patients with IVLBCL, which served as a suitable material for genomic analysis. Previous genomic analysis across the coding region of IVLBCL enabled us to categorize the majority of IVLBCLs as MYD88/CD79B double mutant (MCD) type lymphomas according to the LymphGen classification. Among the other lymphoma types with a high prevalence of MCD type, immune-privileged large B-cell lymphoma (IP-LBCL) has been recently proposed as a disease entity in the WHO classification. Although categorized independently, IVLBCL and IP-LBCL share not only clinical characteristics such as recurrent involvements of central nerve system but also genetic features of MCD-type lymphomas. However, the whole-genome level comparison of mutational landscape and clonal architecture of IVLBCL and IP-LBCL has not been performed. Method To comprehensively understand the mutational landscape of IVLBCL and IP-LBCL, we performed whole-genome sequencing of 32 IVLBCLs (28 samples from cell-free DNAs and 4 samples from patient-derived xenografts) and 25 IP-LBCLs (12 central nervous system lymphomas and 13 testicular lymphomas). Peripheral blood mononuclear cells were analyzed as the control. Single nucleotide variants (SNVs) and small insertion-deletions (indels) were analyzed with G-CAT pipeline; copy number alterations (CNAs) with Battenberg; and structural variations (SVs) with GRIDSS. Mutational signatures were analyzed with SigProfiler and its associated packages. Timing analysis of SNVs/indels and CNAs was performed with MutationTimeR package. Results From our whole-genome data of 32 IVLBCLs and 25 IP-LBCLs, we identified the comparable mutational profiles of IVLBCL and IP-LBCL. As for the coding region, we found the consistent recurrence of mutations in MYD88, CD79B, KMT2D and DTX1 genes in IVLBCL and IP-LBCL. Both subtypes had characteristic clustered mutations enriched by SBS84 mutations, which indicates the causative role of activation-induced cytidine deaminase (AID). Copy number analysis showed commonly frequent copy neutral loss-of-heterozygosity (CN-LOH) or gain with LOH of 9p and 3p, gain of 1q, 18q and 19q, and deletion of 6q in both subtypes. Focal copy number changes commonly found in both subtypes include 1q43 deletions (CD58), 6p21 deletions (HLA class I genes), 8q12 deletions (TOX), 9p12 deletions (CDKN2A), 12p13 deletions (ETV6) and 19p13 deletions (MEF2B). Timing analysis of the acquisition of SNVs/indels and CNAs revealed the similar timing patterns between the two subtypes, showing the early acquisition of MYD88 mutations, the majority of kataegic mutations and 9p CN-LOH/gain with LOH. For differently mutated genes between two subtypes, mutations in RAC2, BCL11A, CREBBP and TMEM30A genes, 9p24 focal amplification affecting CD274/PDCDLG2/JAK2 genes, and multiple type of SVs affecting CD274 and PDCDLG2 genes were observed at a higher frequency in IVLBCLs compared to IP-LBCLs. On the other hand, IP-LBCLs had frequent mutations in TMSB4X, ITPKB, and B2M genes, 10q23 deletions (FAS), 15q21 deletions (B2M) and 16p13 deletions (CIITA). While the majority of genes targeted by AID showed a similar distribution of genetic mutations in both subtypes, some genes, such as ETV6, exhibited distinct patterns. Whole genome doubling occurred more frequently in IVLBCL compared to IP-LBCL. Interestingly, these differently affected genes by mutations and CNAs contains genetic alteration associated with immune evasion; focal amplifications or SVs involving CD274/PDCDLG2 genes were specific to IVLBCL, while B2M abnormalities and mutations of HLA class I genes were more frequent in IP-LBCL. Conclusion Our data provide a whole-genome mutational landscape of IVLBCL and IP-LBCL. Sharing the earliest driver mutations/CNAs between both diseases suggested the common genetic mechanism involved in their initiation of clonal evolution. On the other hand, each subtype has their unique genomic alterations, particularly in immune evasion-related pathways. These genomic variations may imply their distinct crosstalk with tumor microenvironments of individual origin sites. Our genetic analysis can contribute to the deep understanding of the pathogenesis of IVLBCL and IP-LBCL.