Next-generation sequencing (NGS) studies have shown that multiple myeloma is a heterogeneous disease with a complex subclonal architecture and few recurrently mutated genes. Only a minority of variants are potentially actionable and studies on their prognostic value are lacking. Therefore, strategies to investigate the landscape of chromosomal and gene lesions of a large number of myeloma samples in a robust fashion is needed. In this study, we developed a target-enrichment strategy to streamline simultaneous analysis of gene mutations, copy number changes and IGH translocations in multiple myeloma in a high-throughput fashion using NGS.

We designed Agilent SureSelect cRNA pull down baits to target 246 genes implicated in myeloma, in lymphoid malignancies or other cancers, 2538 single nucleotide polymorphisms to detect copy number and allelic ratio at the single-gene and whole-genome level, and we tiled the whole IGH locus to detect IGH translocations and V(D)J rearrangements. As a pilot, we sequenced 13 myeloma cell lines and 10 control haematopoietic cells lines and validated its sensitivity and specificity. We have next applied this baitset to unmatched DNA from CD138-purified plasma cells from 426 patients at diagnosis, including 51 matched samples from a previous whole exome sequencing dataset. We sequenced at an average depth of 329x with 77% of the target region covered at >30x using Hiseq2000 machines (Illumina Inc.). We applied algorithms developed in-house to detect point mutations, insertion-deletions and IGH translocations, filtering out potential artifacts and germline variants. We then annotated as "oncogenic" all variants previously reported as somatic in cancer by the COSMIC database.

418/426 patients had at least one variant, and overall we identified 2207 variants of which 667 were oncogenic. 212/246 evaluated genes were mutated at least once, but only 102 had at least one oncogenic variant. Furthermore, 417/667 (63%) oncogenic variants were accounted for by the top 8 driver genes previously identified (KRAS, NRAS, TP53, FAM46C, BRAF, DIS3, TRAF3, SP140). Additional findings of interest included clustered mutations in SF3B1, EGR1, IRF4, but in <5% of patients each. We identified IGH translocations in 127/375 patients, including 70 t(11;14) and 35 t(4;14). FISH validation suggested our pulldown approach had 93% sensitivity and 98% specificity. Given the high number of variables and patients, we looked at pair-wise interactions between genes and with karyotypic features. We found that BRAF variants are not mutually exclusive with KRAS or NRAS mutations, a finding with implications for pathogenesis and treatment with BRAF inhibitors. Furthermore, samples with IGH translocations, and t(4;14) in particular, had fewer mutations overall, were significantly enriched for mutations in SF3B1 and depleted of mutations in FAM46C, suggesting that the genomic landscape of myeloma may in part vary with karyotypic features.

We then looked at prognostic models. While TP53 mutations had the strongest effect both on overall survival (OS) and progression free survival (PFS), we found a favorable impact of TRAF3 mutations on PFS, and a negative one of NRAS and SP140. More importantly, pairwise interactions served as a starting ground for a rationale subgroup analysis. For example, while FAM46C mutations had no effect on OS across the whole cohort, they predicted better survival in patients without IGH translocations (median 94 months for WT FAM46C vs. not reached for MUT, p=0.042 log-rank test). Similarly, we found a different magnitude of the negative impact of TP53 mutations on OS based on karyotype. Survival was shortened more than 8-fold in cases with both an IGH translocation and a TP53 mutation (median 89 months for WT TP53 vs. 11 for MUT, p=2e-10), while this effect was much less in the absence of an IGH translocation (94 months for WT vs. 50 for MUT, p=0.02).

In conclusion, the large sample size and extent of sequencing provides further insight into the genomic landscape of myeloma and how this impacts the clinical phenotype confirming the utility of the targeted sequencing to both understand the biology as well as its clinical application.


Campbell:14M genomics: Other: Co-founder and consultant.

Author notes


Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract