The Landscape of Structural Variant Signatures in Multiple Myeloma Identifies Distinct Disease Subgroups with Implications for Pathogenesis

Multiple Myeloma (MM) initiation and progression is driven by recurrent cytogenetic events, i.e. multiple trisomies or translocations within the immunoglobulin locus. Gene mutations have been extensively studied, and they are generally involved in late phases of disease development. On the contrary, very little is known about non-recurrent structural variations (SV), which are increasingly emerging as critical driver in several cancers.

To determine the extent to which the MM genome is shaped by such events, we performed whole genome sequencing (WGS) on 67 CD138+ purified bone marrow MM samples from 30 patients (median of 2 samples per patient; range 1-4), to which we added 22 previously published cases (Chapman et al, Nature 2011) for a total of 52 patients and 89 tumor samples. We defined SVs as inversions, translocations, internal tandem duplications and deletions, which we analysed using publicly available tools developed at the Wellcome Sanger Institute.

We found a stunning 1887 unique SVs in the whole cohort, with a variable distribution across the entire series (median 29 per patient, range 0-156). To derive a homogeneous catalogue of SV across the different MM patients and biological subgroups, we annotated events according to the recently proposed classification on >2500 cancer genomes (Li Y. et al BioRxiv 2018). IGH and MYC translocations were the most frequent recurrent events and accounted for just 5.3% of the entire SV catalogue. We defined as complex events the following SV classes: chromothripsis, chromoplexy, multiple inversions (distinct between Local_n_Jumps and Local_and_distant_n_jumps in case of translocation involvement), templated insertion between more than 2 chromosomes. According to this classification, in 93% of patients a single, private complex SV was responsible for multiple and simultaneous CNAs across different chromosomes, thus providing a novel pathogenetic explanation for many recurrent CNAs in MM.

Overall, 136 complex events were observed in 43/52 patients (83%). We found 34 instances of chromotripsis (Korbel J.O. et al., Cell 2013) in 18/52 (34%) patients. The vast majority (30/34) were clonal and conserved during evolution, suggesting an early role in MM pathogenesis. In addition, we observed 5 chromoplexy events (Korbel J.O. et al., Cell 2013) acquired in 5 patients. More interestingly, evidence of templated insertion on more than 2 chromosomes was observed in 13 patients (25%). This event is composed by multiple concatenated translocations causing small CNAs (mostly gains) and in 77% it resulted in a translocation involving an important MM oncogene (8 MYC and 2 CCND1), suggesting that this is a novel relevant driver mechanism in MM.

Given the variety of the landscape of SV between patients, we investigated the presence of SV patterns (SV signatures) by the hierarchical dirichlet process (hdp) (https://github.com/nicolaroberts/hdp). Six main SV signatures were extracted: SV signature #1 was characterized by multiple isolated deletions; SV signature #2 was associated with chromotripsis; SV signature #3 was characterized by reciprocal translocation, local_and_distant_n_jumps and templete insertion between 2 chromosomes. Signature #4 was mostly characterized by local_n_jumps; Signature #5 and #6 were associated with temple insertions on multiple chromosomes with or without large tandem duplication, respectively. Different patients showed differential contribution from different signatures, and based on this we observed 5 distinct clusters. Interestingly some of these clusters were associated with distinct and known MM drivers. For example t(4;14)(MMSET;IGH) cases were enriched for SV signature #1. A significant fraction of patients without any recurrent IGH translocation were characterized by high prevalence of chromothripsis. SV signatures #5 and #6 were mostly associated with hyperdiploid patients with MYC translocation and low genomic impairment.

In this study, we describe the landscape of SVs and complex events in MM, suggesting that this notation may represent an important step forward in disentangling the genomic complexity and heterogeneity of MM.