Heterogeneity of mutational processes in multiple myeloma

Background: The catalog of somatic mutations in a cancer is the aggregate outcome of strength and duration of exposure to one or more mutational processes. Signatures underlying mutational processes have never been investigated in multiple myeloma (MM), a plasma cell malignancy with incompletely understood molecular pathogenesis. While chromosomal abnormalities are involved in MM initiation, they are insufficient for its progression, which is characterized by secondary events, particularly gene mutations. Despite novel treatment options, relapses are frequent and ultimately chemoresistant, thus MM remains an incurable disease. Understanding the molecular processes shaping the MM genome at diagnosis and those contributing to emergence of chemoresistant clones can therefore provide new insights into disease biology and treatment. Aims: We have previously showed how the mutational landscape, clonal architecture and evolutionary progression were heterogeneous across 84 samples from 67 patients with MM (Bolli et al, Haematologica, 2012, Vol 97-S1, a0571). In this study, we set out to extract the mutation signatures characterizing the mutational processes underlying the landscape of mutational changes in MM. Methods:We validated 4421 variants in 67 patients. We showed variability in the numbers and relative contributions of each base substitution (C>A, C>G, C>T, T>A, T>C, and T>G). To provide insight into the processes underlying such a heterogeneous catalogue of mutations, we incorporated the sequence context in which mutations occurred by considering the bases 5’ and 3’ to each mutation to generate 96 possible mutation types. Werepresented the fraction of mutations of each type as a heatmap for each case. We employed nonnegative matrix factorization (NMF) and model selection to extract the quantitative contribution of each mutational signature. Results: Most cases showed a predominance of Signature A, C>T changes at CpG trinucleotides. This is likely to represent spontaneous deamination of methylated cytosine to thymine, and is the dominant process in myeloid malignancies, a major contributor to early mutations in breast cancer, and seen at high rates in the germline. Signature B, consisting of C >T, C>G and C>A mutations in a TpCpX context, contributed 5 to 45% of the variants in most cases, but 100% in the 2 samples characterized by the highest number of variants, suggesting that this signature is associated with a hypermutator phenotype. In another 2 patients, we found clusters of 5-10 mutations from Signature B, all within a region <200bp showing strand specificity, a process known as ‘kataegis’. This signature was first documented in breast cancer, and others and we have speculated to arise from aberrant activity of APOBEC proteins. In serial samples, relative contribution of each signature changed up to 2-fold. Importantly, progression towards aggressive disease was associated with an increased contribution of signature B and the appearance of kataegis, suggesting that this signature may be linked to phenotypic evolution. Another process leading to regional clustering of mutations was somatic hypermutation driven by the AID protein, resulting in clusters of mutations of CCND1 in two cases, both characterized by t(11;14), juxtaposing CCND1 with the IGH locus. Finally, copy number analysis revealed the presence of chromothripsis, a mechanism of genomic instability defined by tens to hundreds of chromosomal rearrangements involving localized genomic regions, associated with poor outcome. Summary / Conclusion:We have described for the first time a catalogue of several heterogeneous mutational processes active in MM, including spontaneous deamination of methylated cytosine, kataegis, somatic hypermutation, and chromothripsis. We have shown that mutational processes are active across different cancer types and retain the same biological features, such as the hypermutator phenotype of Signature B. The extent of such mutational processes can vary from genome-wide to localized clusters, and their relative contributions change over time. This analysis represents advancement in our understanding of MM biology, and of the processes underlying its genomic heterogeneity.