Clinical and biological significance of RAS mutations in multiple myeloma

Primary genetic abnormalities in myeloma (MM) such as trisomies of chromosomes 3, 5, 7, 9, 11, 15, 19 and 21 associated with hyperdiploid MM and translocations involving the immunoglobulin heavy chain (IgH) locus on chromosome 14q32 and three main recurrent partners: MMSET/FGFR3, CCND1 and c-MAF are already present in the pre-malignant monoclonal gammopathy of undetermined significance (MGUS) stage.¹ Some patients with these genetic abnormalities may remain as MGUS for many years without transforming to MM, suggesting that they are involved in clonal initiation but do not mediate malignant transformation.

One of the recurrent differences between MGUS and MM is the presence of RAS mutations in the latter. RAS mutations may, therefore, play an important role in malignant transformation of clonal plasma cells and myeloma pathogenesis. However, the clinical and biological significance of RAS mutation in MM has not been clearly established as most of the previous studies have involved small numbers of heterogeneously treated patients.

To establish the clinical and biological significance of RAS mutation in MM, we studied the association of RAS mutation with a comprehensive spectrum of clinical parameters including the newly established ISS staging and survival, as well as a panel of known recurrent genetic abnormalities in MM such as t(4;14), t(14;16), t(11;14), chromosome 13 and 17p13 deletion detected by fluorescent in situ hybridization² and ploidy assessed by DNA content measurement using flow cytometry,³ in a large cohort of newly diagnosed patients enrolled in the Eastern Cooperative Group (ECOG) clinical trial E9486/E9487 (N = 561).⁴ A total of 439 patients, based on sample availability, were included (The ECOG Cohort). For this cohort, DNA from unsorted whole bone marrow was used for RAS mutation studies. We also studied 14 MGUS patients and 82 MM patients (60 newly diagnosed and 22 relapsed) from the Mayo Clinic (Mayo Cohort). Bone marrow samples were obtained after informed consent according to the Declaration of Helsinki. The study was approved by the Mayo Clinic Institution Review Board. CD138 positive plasma cells were enriched using immuno-magnetic beads (AutoMACS; Miltenyi-Biotec, Auburn, CA, USA). RNA and DNA from these enriched cells were used for gene expression and RAS mutation studies, respectively.

Conformation sensitive gel electrophoresis was used to screen samples for KRAS (codons 12, 13 and 61) and NRAS (codons 12, 13 and 61) mutations (Supplementary Methods). RAS mutation was detected in 102 (23%) patients in the ECOG cohort. Seventy-four (17%) patients had mutations in NRAS. The majority of these mutations were in codon 61 (64 of the 74), with 5 mutations detected in codons 12 and 13. We also found mutations in codon 64 (n = 1) and codon 86 (n = 4). Twenty-eight (6%) patients had mutations in K-RAS. Twenty-two of these are in codons 12 and 13 and three in codon 61. In addition, we also found one mutation each in codon 16, 22 and 24. In the Mayo Clinic cohort, RAS mutation was detected in 1 of 14 MGUS patients (7%), 15 of 60 newly diagnosed MM patients (25%) and 10 of 22 relapsed MM patients (45%). Once again N-RAS mutations were more common that K-RAS mutations. Our study confirms the low incidence of RAS mutation in MGUS compared to MM found in a previous study,⁵ consistent with RAS mutation being an important transforming event from MGUS to MM.

Previous studies have produced a broad range of prevalence from 0 to 100%. This variability may be due to a number of factors including small study cohort, patient selection, plasma cell enrichment and sensitivity of detection methods. But despite these differences, the majority of the studies have reported a RAS mutation prevalence of between 20 and 35%, with NRAS mutations (most commonly affecting codon 61) more common than KRAS mutations (Supplementary Table 1). Our results are, therefore, consistent with these other studies. Only two studies including more than 20 patients have reported a prevalence higher than 40%. Both used sensitive methods of detection on CD138-selected plasma cells that have not been used in other studies in MM, and reported prevalence of 76 and 100%. Besides the much higher prevalence of RAS mutation, there are other aspects of the results from these two studies that are inconsistent with findings from others and us. In one study, KRAS mutation was detected in 55% of samples, almost twice the prevalence of NRAS mutation detected.⁶ In the other study, NRAS mutation was detected in all cases tested and usually in sub-clonal populations but KRAS mutation was only detected in one sample.⁷ Once again these studies involved a small number of patients and the causes of these discrepancies are probably multi-factorial.

One potential shortcoming of our study is the use of DNA from unsorted cells in the ECOG cohort. However, our method could detect 1 mutant cell in 10 and the bone marrow plasma cell involvement of the cases analyzed was all above 10%. Furthermore, the prevalence of RAS mutation in the ECOG cohort is similar to that in the Mayo cohort where RAS mutation analysis was performed in CD138 selected plasma cells using the same detection methods. Therefore, it would appear that the use of DNA from unsorted bone marrow did not affect our ability to detect RAS mutations in the ECOG cohort.

Patients with RAS mutation have more aggressive disease features. Using the Wilcoxon rank sum test and Fisher exact test for continuous factors and categorical factors, respectively, RAS mutation is significantly associated with higher plasma cell percentage in the marrow, greater percentage with B₂M greater than 3.5 g/l, greater percentage of ISS stage II and III disease, κ light chain secretion, lower hemoglobin level and more frequent lytic bone lesions. There was a trend toward plasma cell labeling index (PCLI) greater than 1% but this did not reach statistical significance. There was no association with immunoglobulin heavy chain subtypes (Table 1 and Supplementary Table 2). Similar associations were seen when KRAS and NRAS mutations were analyzed separately (data not shown) except that tumors with KRAS mutations are significantly associated with PCLI greater than 1% (50 versus 27%, P = 0.023).

Patients with RAS mutations have significantly shorter overall survival and progression free survival despite similar response to therapy (Figure 1). Interestingly, when we analyzed the impact of KRAS and NRAS mutations on survival separately, only KRAS and not NRAS mutations were significantly associated with shorter overall survival and progression free survival (Figure 1). RAS mutation was an independent prognostic factor for overall survival but not progression free survival in the Cox proportional hazards regression analyses. For overall survival, selection results included ISS stage III, PCLI greater than 1, t(4;14), and 17p13 deletion/TP53 mutation. ISS stage II was forced back into the model for interpretation. Finally, adding RAS mutation improved the model slightly based on likelihood-ratio χ² P-value 0.052. Patients with RAS mutation had a 47% increase in the hazard of death compared with normal RAS patients (P-value of 0.045) after adjusting for all other variables in the model (Table 2).

Although RAS mutations are associated with a more aggressive phenotype and shorter survival, they are rarely found in tumors with adverse genetic factors such as t(4;14), and 17p13 deletion. On the other hand, RAS mutation is significantly associated with t(11;14). There is no association with ploidy categories (Table 3). Interestingly, KRAS but not NRAS mutation was significantly associated with TP53 mutation (15% of those with KRAS mutation versus 3% of those without, P = 0.014).

Patients with RAS mutations have features associated with greater tumor burden such as higher B₂M, greater bone marrow involvement and more advanced ISS stage. It is, however, not commonly seen in tumors with high-risk genetics such as t(4;14) or 17p13 deletion or more proliferative tumors. This suggests that RAS mutation predominantly drives tumor growth or expansion rather than proliferation in MM. It is significantly associated with t(11;14) subtypes of myeloma. Despite t(11;14) being usually associated with better prognosis, patients with RAS mutations have significantly shorter survival. Therefore, prognosis in MM may be mediated by at least three main non-overlapping components: tumor burden, proliferation and poor-risk genetics such as t(4;14) and 17p13 deletion. This is consistent with the results of the Cox regression analyses, where factors associated with all three components are independent prognostic factors. Interestingly, the predominant prognostic impact of RAS mutation is mediated through KRAS mutation. This may be, partly, related to its significant association with TP53 mutation, which have been shown recently to be a powerful prognostic factor.⁸

As RAS mutation is a likely transformation event, its association with t(11;14) genetic subtype but not other primary genetic subtypes such as t(4;14), t(14;16) or hyperdiploidy, suggest that different ‘second hit’ events may cooperate with the different primary genetic events to mediate transformation from MGUS to MM. In this regard, the much higher frequency of chromosome 13 deletion in t(4;14) and t(14;16) suggests that the critically deleted gene on chromosome 13 may cooperate with the genes and pathways deregulated by these translocations to mediate transformation.

In conclusion, RAS mutation represents the single most commonly mutated gene in myeloma, is associated with features consistently with greater tumor burden and likely represents an important transforming second hit especially in t(11;14) myeloma.