Analysis of mutations and structural variants to redefine the genomic landscape of multiple myeloma and its clinical implications

Introduction: In multiple myeloma (MM), karyotypic events such as translocations between the IGH locus and known oncogenes, and recurrent copy-number abnormalities (CNAs) are considered early drivers, being detectable also in pre-malignant stages of the disease. Recently, several recurrent single-nucleotide-variants (SNVs) have been described in MM, but their real driver role and relationship with other genomic events have never been explored on large series. Methods: Here, we combined whole genome (n=30), whole exome (n=849) and targeted (n=373) sequencing data of 1252 MM patients. Eight hundred and four patients were included from the CoMMpass study, generated as part of the Multiple Myeloma Research Foundation Personalized Medicine Initiatives. The driver vs passenger role of each SNV was defined by the dNdS algorithm (Martincorena et al., Cell 2017). The hierarchical dirichlet (HDP) process was used to investigate the main MM genomic subgroups as previously described (Bolli et al. Leukemia 2017). Figure 1. Results: Combining WGS and 879 whole-exome data, we extracted 56 significant driver SNVs [median of 1 per patient (range 0-6)], with KRAS (23%), NRAS (22.1%), DIS3 (9.5%) and FAM46c (4.8%) confirmed as the most recurrent. At least one driver SNV was extracted in 741 patients (84%). We then included additional 373 MM patients investigated by an unmatched targeted sequencing approach (Bolli et al. Leukemia 2017), to create the largest dataset of MM samples to date (n=1252) to investigate the interrelationships of karyotypic events (n=14) and the most frequent SNVs (n=21). To this end, patterns of co-occurrence and mutual exclusivity of recurrent CNAs and SNVs were derived from their distribution and clustered using the HDP. Karyotypic events contributed to clustering more than SNVs, and we extracted five main clusters based on their extended genotype (Figure 1). The first was defined by hyperdiploidy and accounted for 59% of the entire series. del13q, del TRAF3, gain1q21 and del1p13 defined the second cluster (18%). t(11;14)(CCND1;IGH) and mutated NRAS/KRAS defined the third cluster (11%). del13q, gain 1q21, DIS3 mutation, t(4;14) defined the fourth cluster (5.5%). TP53 mutation, del17p13, del13q14, t(11;14), deletion of CYLD defined the last cluster (4%). With a median followup of 621 (range 31-4205) days, the clusters had a distinct clinical outcome, with cluster 5 showing the poorest overall survival and cluster 3 showing a favorable outcome. Conclusion: Our data show that a tentative genomic classification in MM is dominated by karyotypic events, with driver SNVs occurring during later on distinct genomic profiles. Our analysis showed significant clustering, however most events were not entirely segregated within each group, suggesting a context-dependent effect of many of them, and a role for other genomic non-coding drivers. Our analysis supports the use of extended genotyping of MM cases at diagnosis for classification and prognostication.