Abstract: Multiple Myeloma (MM) represents a heterogeneous and complex hematological cancer, where karyotypic events are thought to be early drivers of disease pathogenesis. Nevertheless, aside from recurrent translocations between the immunoglobulin loci and known oncogenes, a comprehensive analysis of MM structural variations (SV) has never been performed. Here, we took advantage of whole genome sequencing (WGS) to analyse 97 samples collected at different clinical time points from a cohort of 30 MM patients (median of 2 samples per patients; range 1-4). Bam files were analysed using validated and published tools available at the Wellcome Trust Sanger Institute. For each sample, we analysed the genome-wide profile of copy-number aberrations (CNAs), cancer cell fraction (CCF) and SV using published approaches (Nik-Zainal S. et al. Cell 2012 and Nik-Zainal S. et al. Nature 2016). We defined SVs as inversions, translocations, internal tandem duplications (ITD) and deletions. We found a stunning 2419 SVs in the whole cohort, with a median of 30 unique SVs per sample (range 3-135). Of these, the only recurrent rearrangements were translocations involving MYC (14/30; 46%), CCND1 (7/30; 23%) and MMSET (3/30; 10%), suggesting MM is characterized by an incredibly complex heterogeneity of its genomic architecture. Interestingly, we observed a much higher than previously reported (7/30; 23%) number of translocations involving the light chain immunoglobulin loci (IGL), which always correlated with hyperdiploid karyotype and were mutually exclusive with those involving the heavy chain locus (IGH). Analysis of CCF confirmed that translocations involving CCND1 and MMSET were always fully clonal confirming their early driver role in MM pathogenesis. Integrating SV and CNA data, we found that a SV was responsible for 89.9% of CNAs where a breakpoint could be mapped (i.e. those not involving centromeric or telomeric regions, or whole chromosome deletions or amplifications). Even more interestingly, we found frequent (93% of patients) SVs involving multiple chromosomes and causing multiple CNAs at once, offering a pathogenic explanation to these frequent karyotypic events in MM. We found that such complex events even involved driver translocations, as 3 out of 7 t(11;14) were not simple balanced translocations but were rather associated with CNAs of the locus. MYC translocations were even more frequently associated with CNAs (13/15 cases), often resulting in focal amplifications of the 8q24 region. Focal mono or bi-allelic deletions were also observed on other important oncogenes such BIRC2/3, CDKN2A/B, CDKN2C and TRAF3, and were generally acquired during progression and/or relapse, potentially representing a mechanism of subclonal selection and treatment resistance. Overall, 20/30 (66%) of MM patients showed at least one complex event defined as an event composed by ≥ 3 independent SVs. Chromotripsis was the most frequent: 18 chromotripsis events were observed in 9/30 (30%) patients; the majority (14/18) were clonal and conserved during evolution, suggesting a potential early role in MM pathogenesis. However, chromotripsis was not the only recurrent complex event. In fact, we observed 2 chromoplexy event acquired in 2 patients at relapse. Even more interestingly, in 6 patients we describe a novel complex event characterized by multiple concatenated translocations causing small CNAs on more than 2 different chromosomes, that we named cyclo-template insertion (CTI). Interestingly, all but one CTIs resulted in a translocation involving an oncogene (4 MYC and 1 CCND1), suggesting that this is a novel relevant driver mechanism in MM pathogenesis. We finally used the Bayesian Dirichlet process (Bolli N et al. Nat Comm 2014) to reconstruct the chronological order of the main MM driver events. generating, for each sample, a phylogenetic tree incorporating mutations, CNAs and SVs. We show that during MM life history there are few, albeit often complex, genomic events with the potential to cause major selective subclonal sweeps, and these events were highly variable from patient to patient. Chromotripsis, CTI, CNAs and other SVs accounted for most of these events, confirming their critical role in MM pathogenesis.