Abstract

Multiple myeloma (MM) is a pathological description for a plasma cell malignancy that exhibits a high degree of genetic diversity between patients. Whole genome analysis methods have elucidated multiple distinct subtypes of disease, however, we have a very limited understanding of what drives each independent subtype. To address this issue, we developed a comprehensive analysis approach to identify MM subtypes and associated genetic features in untreated patient samples from the MMRF CoMMpass study (NCT01454297). In this study we attempt to characterize each sample using whole genome (WGS), exome (WES), and RNA (RNAseq) sequencing. For this analysis we have limited the baseline cohort from the 982 patients with at least one assay completed to the 591 which are completely characterized.

Most cohort level analyses focus on specific genetic aberrations such as copy number (CN) (eg. GISTIC) or somatic mutations (eg. MutSig) to identify genes which are important in the cancer. However, it is well known that cancer cells will dysregulate genes and pathways through a diverse array of mechanisms. We developed an integrated model, leveraging WGS, WES, and RNAseq data, that simultaneously considers; translocations, inversions, CN, copy neutral loss-of-heterozygosity, somatic mutations, constitutional inherited gene defects, gene expression, and in-frame genomically validated gene fusions; in a comprehensive analysis to aid in the discovery of putative oncogenes or tumor suppressor genes.

We first defined two independent subtype classes within the cohort using consensus clustering on the gene expression and copy number data, which identified 12 and 14 subtypes, respectively. We also identified 59 significantly mutated genes using MutSig with a robust resampling approach. Then using our integrated approach we identified 64 putative loss-of-function (LOF) genes which have complete, bi-allelic loss, through anyone of 11 possible combinations of genetic defects in at least 1% of the cohort. These include expected tumor suppressors; TRAF3 (10.1%), DIS3 (6.9%), FAM46C (5.1%), TP53 (4.1%), RB1 (3.2%), and NF1 (1.0%). On chromosome 13 alone we identified 17 LOF genes with LOF events occuring in 15.4% of patients with the most common genes being PSPC1, BRCA2, RB1, DIS3, and TGDS. This process also identified LOF events in genes associated with 5-FU sensitivity, DPYD, and PARP inhibitor sensitivity, BRCA1, BRCA2, and STAG2 suggesting some patients may benefit from these agents. We identified 19 putative gain-of-function (GOF) genes, which had recurrent mutations at the same amino acid position, high level CN gains (6 or more copies), overexpression associated with a structural rearrangement, or inframe fusion transcripts in at least 1% of the cohort. These include expected oncogenes; NRAS, KRAS, BRAF, and myeloma specific genes like WHSC1. Interestingly this approach identified a series of genes that were not detected by MutSig such as IRF4, HIST1H1E, DUSP2, and MAP3K14 plus a number with recurrent suspect activating mutations; BMP2K, PTPN11, ALOX15, and ANKLE1.

Few subtypes of MM have defined genetic events outside of gene expression subtypes defined by the common IgH translocations. Comparing genes identified in our LOF and GOF analysis with our gene expression or CN defined subtypes identified a number of associations. For instance, in the MS gene expression subtype, characterized by t(4;14), we detected the expected associations with WHSC1 and FGFR3 along with a significant enrichment of TRAF3 and DIS3 LOF events and an under representation of NRAS mutations. A gene expression group associated with a hyperdiploid karyotype and 1q gains has significantly fewer NRAS and KRAS GOF events but is enriched for TRAF3 LOF events. GOF events in IRF4 were exclusively found in the CD2 gene expression groups associated with cyclin D translocations. Interestingly, the LOF events in TGDS were exclusively found in a single CN defined subtype characterized by a hyperdiploid karyotype with 1q gains but lacking chromosome 11 trisomies. Finally, the gene expression defined proliferation (PR) subtype, which exhibits poor OS (HR = 3.996, 95% CI = 2.632 - 6.067, p < 0.001) is highly enriched for bi-allelic LOF events in RB1 and MAX. This finding provides the first genetic explanation for why patients with such diverse genetic backgrounds cluster in this high risk subtype.

Disclosures

Lonial:Amgen: Research Funding.

Author notes

*

Asterisk with author names denotes non-ASH members.