Identification of Novel Oncogenes and Tumor Suppressor Genes in Newly Diagnosed Multiple Myeloma

Introduction. Identifying driver genes in multiple myeloma (MM) will have a number of key benefits as they can directly affect clinical behavior and define new targets for therapy, subgroups of disease and identify prognostic lesions. Here we focus on identifying key tumor suppressor (TSG) and oncogene (ONC) drivers of MM which could potentially act as novel targets for therapy. As the majority of mutations are present at <5% large datasets are required to identify these drivers. Rare targetable mutations are generally in ONCs and identifying their relevance requires knowledge of the mutational spectrum, sites of recurrent mutation and their downstream effects on protein function. We have taken a number of innovative strategies to identify such genes in the largest set of MM samples established to date.

Methods. We established a set of 1277 newly diagnosed patient samples for which whole exome sequencing was available. Data were derived from the Myeloma XI trial, Dana-Faber Cancer Institute, The Myeloma Institute and the Multiple Myeloma Research Foundation CoMMpass study (IA1 - IA9). Mutations were called using Strelka and Mutect. Two strategies were used to identify drivers of MM pathogenesis 1. Determine significantly mutated genes in the entire cohort and in each subtype (MutSigCV) and 2. Identify ONCs and TSGs using curated cancer gene lists. ONCs were defined by identifying the proportion of recurrent missense mutations (ONC score >0.4) and TSGs by the proportion of nonsense and frameshift mutations (TSG score >0.2). If both thresholds were achieved the larger value was used as defining the gene type. Expressed somatic variants were analyzed using sample matched RNA-seq data.

Results. A total of 26 statistically significantly mutated genes carrying single nucleotide variants (SNV) and indels were identified, by analyzing the dataset overall and within the major etiological subgroups. These results confirm significant mutations in 11 previously identified genes (KRAS, NRAS, DIS3, BRAF, TP53, MAX, TRAF3, CYLD, RB1, FAM46C, HIST1H1E) as well as 9 new significantly mutated genes including UBR5 (3.5%), PRKD2 (3.5%), SP140 (2.4%), TRAF2 (2.1%), PTPN11 (2.3%), RASA2 (1.3%), NFKBIA (1.3%), TGDS (1.3%), and CDKN1B (1.1%). Within cytogenetic sub-groups we found an additional 6 significantly mutated genes including IRF4 and HUWE1 in the t(11;14), ACTG1 in hyperdiploid cases, and FGFR3, IRF4, and MAFB in non-hyperdiploid cases.

Using a curated list of 116 known driver genes plus the 26 significantly mutated genes, we determined their ONC and TSG score using the approach defined by Vogelstein. Using this strategy, we identified an additional 13 driver genes which included EGR1 (4.7%), CCND1 (2.9%) and MAF (1.6%), SF3B1 (1.8%, spliceosome factor), IDH1 (0.6%) and IDH2 (0.4%, increased DNA methylation). Surprisingly, DIS3 (9.9%) and TP53 (5.6%) were also classified as oncogenes due to their high proportion of recurrent missense mutations, 73% and 48% respectively.

We determined if the key recurrent mutations in these oncogenes were expressed in the same cases. As expected, the recurrent variants in KRAS and NRAS were expressed, as were those in BRAF, IDH1, IDH2, EGR1, CCND1, PTPN11, IRF4, FGFR3, and SF3B1 . Recurrent variants in TP53 (R248 (n=4), R175, G199, and Y234 (all n=3)) were also all expressed, as were those in DIS3 (R780 (n=11), M667, H691, R820 (all n=2)).

Understanding the timing of when these variants arise is crucial to targeting them effectively and this can be determined by their cancer clonal fraction (CCF). Examining the CCF of these 40 driver genes, we saw an association with ONCs and a higher CCF and TSGs with a lower CCF, indicating activating mutations were either early events or were selected for, and inactivating mutations were later events. Although NRAS and KRAS mutations are most frequent they are not associated with early events, as indicated by intermediate CCF values (median 0.65). IDH2, EGR1, CCND1 and HIST1H1E had the most clonal mutations (>0.9).

Conclusion. Oncogene activation through mutation is common in MM. We have identified new mutations in MM associated with oncogene activation including PTPN11, IDH1, IDH2, and SF3B1 . Compared to tumor suppressor genes, mutations in oncogenes are more clonal and, therefore, associated with early events in the disease natural history. Fully characterizing driver genes in MM will enhance our ability to manage it effectively.