The molecular landscape of Waldenström macroglobulinemia identified using whole genome sequencing

Abstract

Abstract Waldenström macroglobulinemia (WM) is a lymphoplasmacytic lymphoma characterized by recurrent somatic mutations in MYD88 and CXCR4, which influence response to therapy. Multiomic analysis revealed an expansion of memory B-cells (MBC) in WM, variably impaired in their capacity to differentiate into plasma cells (PCs), allowing classification into two distinct subtypes; MBC-like and PC-like. To deepen understanding of the genomic underpinnings of these subtypes, we performed the largest whole-genome sequencing (WGS) analysis of WM to date. WGS enhanced the capacity to detect and classify somatic mutations, identify single base substitution (SBS) signatures, and characterize structural variants (SVs), providing deeper insights than previously attainable. We performed full-depth WGS (80x) on 47 newly diagnosed WM cases using the Illumina NovaSeq 6000 platform, sequencing both B cells and matched T cells. Our harmonized bioinformatics pipeline (MGP1000) enabled robust somatic variant calling. We characterized single nucleotide variants (SNVs), SBS signatures, copy number variations (CNVs), SVs, and reconstructed clonal phylogenies. Overall, we found mutations in MYD88 (87%), CXCR4 (28%), CD79B (15%), TP53 (11%), ARID1A (9%), H1-4 (9%), MAP3K14 (NIK) (9%), and KMT2D (4%). These mutations affect B-cell receptor (BCR) signaling, genomic stability, and treatment resistance. The variant allele frequency (VAF) of MYD88 ranged from 20% to 100%, while that of TP53 ranged from 50% to 100%, and other mutations averaged 30–50%. MBC-like (n=10) cases had an increased frequency of mutations in MYD88, CXCR4, CD79B, ARID1A, H1-4, KMT2D, and MAP3K14. In contrast, PC-like (n=22) cases were more frequently associated with gain of 1q and deletion of 6q. Across the different subsets, gain 1q was found in 60% and del of 6q in 40% of MBC-like cases, compared to 73% and 55% of PC-like cases, respectively. SBS signatures showed minimal subgroup variation. To contextualize structural variants (SVs) as drivers of WM, we quantified and compared the number and types of SVs with those in other B-cell non-Hodgkin lymphomas (NHL). The average number of deletions per subject was 3.4 in chronic lymphocytic leukemia (CLL), 32.8 in diffuse large B-cell lymphoma (DLBCL), 37.3 in multiple myeloma (MM), and only 8.6 in WM. The average number of duplications was 0.4 in CLL, 8.6 in DLBCL, 7.9 in MM, and 5 in WM. Consistent with this reduced impact, WM also had fewer examples of the complex SV events chromothripsis and chromoplexy. However, subtle differences existed between subtypes, with SVs seen more frequently in MBC-like cases (average deletions per case: 9.7 and duplications: 6.9) compared to PC-like cases (deletions 7.6 and duplications 4.4). The size of the dataset and the use of WGS gave us the ability to understand in more detail the impact of del 6q on the pathogenesis of WM both overall and within subtypes by taking account of its ability to map the extent of the deletions to identify minimally deleted regions (MDRs) containing potential driver genes. Detailed analysis of the MDRs across all cases and subtypes revealed differences in the extent of deletion. The MDRs on 6q ran from 73.5 Mb to 80.0 Mb. As deletion of B-cell specific genes, has been suggested previously to be a key driver of progression in WM, we focused our analysis on these genes from chromosome 6 and identified recurrent deletions of BCLAF1, TNFAIP3, HIVEP2, ARID1B, and CCR6. Notably, chromothripsis, a phenomenon involving extensive clustered chromosomal rearrangements, was detected involving a del 6q in one MBC-like case, suggesting a potential mechanism for how del 6q arises. We further examined gain of 1q, present in 60% of all cases, as a potential marker of high-risk WM. The minimally gained region on this chromosome as being from 148.5 Mb to 149.7 Mb. Another component of such a high-risk subgroup would include deletions of 17p with copy number loss or mutation of the TP53 tumor suppressor gene. We determined the extent of this group in newly diagnosed WM and found TP53 mutations in 12%, copy number loss of 17p in 11% and biallelic inactivation in 6% of cases. To further elucidate disease evolution, we reconstructed clonal phylogenies within and across these subtypes, revealing subtype-specific evolutionary trajectories. Collectively, our findings offer new insights into the genomic architecture of WM and establish a framework for refining subtype-specific therapeutic strategies.