The varied distribution and impact of RAS codon and other key DNA alterations across the translocation cyclin D subgroups in multiple myeloma

Abstract

We examined a set of 805 cases that underwent DNA sequencing using the FoundationOne Heme (F1H) targeted sequencing panel and gene expression profiling. Known and likely variant calls from the mutational data were analyzed for significant associations with gene expression defined translocation cyclin D (TC) molecular subgroups. The spectrum of KRAS, NRAS, and BRAF codon mutations varied across subgroups with NRAS mutations at Q61 codon being common in hyperdiploid (HRD) and t(11;14) myeloma while being rare in MMSET and MAF. In addition, the presence of RAS-RAF mutations was inversely associated with NFκB pathway activation in all subgroups excluding MAF. In the MMSET subgroup, cases with low FGFR3 expression frequently had RAS-RAF mutations. Conditional inference tree analysis determined that mutation and homozygous deletion of TP53, CDKN2C, and RB1 were key prognostic factors associated with adverse outcome in a non-relapse clinical setting. In conclusion, this study highlights the heterogeneity in the distribution and clinical outcomes of RAS codon and other mutations in multiple myeloma dependent upon primary molecular subgroup.

Keywords: gene expression profiling; multiple myeloma; mutational analysis; translocation cyclin D (TC).

Figure 1. Distribution of RAS-RAF codon and other key alterations across TC-6 subgroups

NRAS mutations at Q61 are commonly seen in D1-HRD, D2, and CCND1-11q13 subgroups but rare among the MMSET and MAF subtypes. FGFR3 mutations are exclusive to the MMSET subtype and other genes are altered at a higher frequency in particular subgroups, e.g. MMSET cases are enriched for BIRC3 alterations and CCND2 expressing subgroups (D2, MMSET, and MAF) are enriched for TRAF3 alterations. Stacked bar plots represent frequencies of key alterations reported as percentages within each TC-6 subgroup. For the small number of cases (1.1%) with co-occurrence of RAS-RAF mutations, only the mutation with highest variant allele frequency (VAF) was used. For the radial alteration plots, each individual slice corresponds to the mutational profile of one specific subject.

Figure 2. Differential gene expression patterns associated with presence of RAS-RAF mutations across TC-6 subgroups

The gene expression levels of DUSP6, DKK1, SPRED2, ETV5, and COBLL1 were highly associated with the presence of RAS-RAF or FGFR3 mutations across all cases (A–E). COBLL1 expression is negatively associated with presence of RAS-RAF mutation while remaining genes are positively associated. The RAS viral oncogene, RRAS2, has a dichotomous expression pattern in the D1-HRD subgroup where high expressers of RRAS2 have few RAS-RAF mutations (F). Density curves are colored according to localized rate of RAS-RAF/FGFR3 mutation and represent probability distributions for mutated and non-mutated cases, separated by dividing horizontal lines. Vertical red lines indicate optimal thresholds of significance according to Fisher exact tests, limited to interior 60% of percentiles, where only highly significant p-values were reported (< 0.005). CCND3-6p21 subgroup not included due to sample size restraints (n =10). Gene expression represents log2 of GCRMA normalized data using the following probe or average of probes: DUSP6 – 208891_at, 208892_s_at, 208893_s_at; DKK1 – 204602_at; SPRED2 – 212458_at; ETV5 – 203349_s_at; COBLL1 – 203641_s_at, 203642_s_at; and RRAS2 – 212589_at, 212590_at.

Figure 3. NFκB signature across TC-6 subgroups and FGFR3 expression in MMSET subgroup

The 11-gene NFκB signature is significantly associated with presence of RAS-RAF or FGFR3 mutations in all TC-6 subgroups excluding MAF where cases without RAS-RAF mutations have higher levels of NFκB activation (A). When examining interaction of FGFR3 expression and RAS-RAF mutation in MMSET, we observed that 56% (10/18) cases with low FGFR3 expression have RAS-RAF mutations compared to 21% of remaining cases (15/71) (B).

Figure 4. Alterations of TP53, CDKN2C, and RB1 showed significant association with overall survival in non-relapse clinical setting

Conditional inference tree analysis of all possible DNA alterations against outcome revealed that alterations in TP53, CDKN2C, and RB1 were significantly associated with an adverse outcome for all cases with disease stages prior to relapse (A). The subset of cases with one or more of these adverse alterations had an 18-month OS rate that was statistically similar to that of GEP70 HR cases (B, C). Combining GEP70 risk and the presence of adverse alterations identified three cohorts with distinct clinical course: poor, intermediate, and standard outcomes (D, E).

Figure 5. CDKN2C and RB1 alterations elicit distinct patterns of gene expression that have significant impact on proliferation and clinical outcome in newly diagnosed MM

Differential expression analysis revealed highly significant patterns of expression associated with alteration in CDKN2C and RB1. Namely that cases with deletion of CDKN2C have lower expression of FAF1 and that cases with alterations of RB1 have higher expression of CDKN2C (Ai). Cases with either of these distinct expression patterns had inferior outcomes (Aii) and higher 50-gene proliferation scores (Aiii). Identical patterns of inferior outcome and elevated proliferation were observed in both the Total Therapy and MRC-IX data sets of NDMM (B, C). In total, alterations in CDKN2C and RB1 offer mutually exclusive paths that culminate in similar high-risk behavior.