Prognostic significance of BRAF and NRAS mutations in melanoma: a German study from routine care

Melanoma is a malignancy originating from melanocytes of the skin with a high propensity to metastasize. Activating mutations of the oncogenes BRAF and NRAS lead to constitutive signaling of the mitogen-activated protein kinase (MAPK) pathway and thereby enhance tumor growth and promote disease progression [1, 2]. Genetic alterations in both genes can be detected in approximately 40 and 20% of cases, respectively [1]. Although specific tyrosine kinase inhibitors targeting BRAF (BRAFi) and MEK (MEKi) are currently available for metastatic melanoma, the overall biologic and prognostic significance of BRAF and NRAS mutations remains unclear and conflicting evidence exists.

In melanoma, the most common mutation of BRAF is a substitution from valine (V) to glutamic acid (E) in codon 600 of exon 15 (V600E). This particular alteration was previously associated with younger age at initial diagnosis, little chronic UV damage, truncal localization of the primary tumor, and a high total body nevus count [3‐7]. Nevertheless, the significance for the further course of the disease such as distant metastasis-free or overall survival (OS) is less clear. One study proposed that mutant BRAF had no impact on the disease-free interval from diagnosis of the culprit tumor to first distant metastasis. However, the median survival in stage IV disease was shorter in patients with BRAF-mutant melanoma and not treated with a BRAF inhibitor than in the wild type (WT) situation [8, 9]. These findings are opposed to other analyses in which patients not treated with BRAF inhibition showed similar survival curves to WT patients [10, 11].

The NRAS gene is most frequently mutated at hotspots in exon 2 (codons 12 and 13) and exon 3 (codon 61) [12]. Mutations of NRAS were previously associated with nodular subtype of the primary tumor and localization in sun-damaged skin [13]. As is the case for BRAF, studies investigating the prognostic relevance of NRAS have revealed discordant results. One of the first examinations dealing with this question identified NRAS as an independent factor indicative of favorable OS [14]. Others have found that NRAS mutation status predicted shorter survival after the diagnosis of stage IV disease [12, 15]. In contrast, no differences in OS after diagnosis of distant metastases were detected in several other investigations [11, 16‐18].

To further elucidate the prognostic impact of mutations in the BRAF and NRAS genes, we analyzed the genotype of 217 patients with melanoma and retrospectively correlated the mutation status to primary tumor and clinical data. To account for a real-life situation and respect potential survival benefits achieved with novel therapy options, patients receiving targeted therapy with kinase inhibitors and immune checkpoint blockade (ICB) with ipilimumab, nivolumab, or pembrolizumab were included in our analysis.

In this study, we assessed genotype-phenotype correlations in 217 patients with melanoma. BRAF mutations were identified in 40.1% and NRAS mutations in 24.4%, while no alterations in either gene were found in the remaining 33.2%. Even though the genotype was not a risk factor in multivariate analysis, BRAF-mutant melanoma patients showed a trend towards better overall and relative survival. In contrast, NRAS-mutant patients were more likely to develop early nodal relapse and metastatic disease than the BRAF-mutant or WT cohort suggesting that they were at highest risk for disease progression and ultimately disease-specific mortality.

Regarding characteristics of the primary tumor, trunk was the most common localization in BRAF-mutant patients. The primaries of this cohort showed less mitotic activity than the NRAS and the WT cohorts, in keeping with a prior study [15]. The nodular subtype was more common in BRAF- and NRAS-mutant groups, while there were no major differences regarding other histologic risk factors such as tumor depth or presence of ulceration. These results are in line with previous investigations in which BRAF mutations preferentially occurred in melanomas of skin with little solar elastosis and less commonly affected the head and neck area [3, 7].

Conversely, NRAS mutations had previously been associated with higher Breslow’s thickness and fast growth of the primary tumor [15, 16]. Although this correlation was not confirmed in our collective, NRAS-mutant tumors accounted for 4 out of 6 local relapses at the primary tumor site and loco-regional recurrence and distant metastasis were significantly more commonly observed in NRAS-mutant patients. The time to nodal recurrence was significantly shortest in the NRAS cohort. These results were further confirmed in a multivariate model for disease progression. Even though NRAS mutational status was not a significant risk factor for OS in multivariate analyses, these results imply that NRAS-mutant tumors appear intrinsically more aggressive and are at high risk for disease progression. This trend for shorter relapse-free survival was also supported by other large-scale analyses [12, 15]. However, in stage IV, the survival differences between NRAS-mutant and non-NRAS-mutant patients proposed previously were not statistically significant in our study, even though NRAS-mutant patients showed a tendency towards shorter survival when metastatic lesions were detected [12, 14]. OS defined as time from the first diagnosis of melanoma to death did not show differences between the genotypes in multivariate analyses, suggesting that the impact of NRAS mutations are more evident in earlier disease stages as proposed by Ellerhorst and colleagues [16]. After 5 and 10 years, the BRAF cohort showed a slight survival advantage of 10–15%, yet without significance for the entire cohort. Post-hoc sample size calculations revealed that at least 1239 patients would have been necessary to significantly assess such differences in survival. Thus, our study was likely underpowered. Nevertheless, these findings raise the question on whether NRAS-mutant patients should undergo a more intensive follow-up after primary diagnosis and whether large trials should stratify for NRAS as independent risk factor.

The strengths of this analysis include the reconstruction of the entire interval from diagnosis of the primary culprit tumor to death or last follow-up as censoring event. Based on our data, we conclude that BRAF and NRAS alterations rather affect early disease stages. Thus, it is indispensable to analyze the complete course of the disease and not only focus on stage IV disease where the differences between the genotypes may be less evident. Furthermore, we assessed the prognostic significance of BRAF and NRAS mutations in the presence of novel therapies such as BRAFi and MEKi or ICB which are broadly available nowadays and led to significant survival benefits in phase II and III trials [29, 30]. However, data in the real-world setting is sparse. Most studies proposing that the mutation status is not prognostic in stage IV disease were conducted either before BRAFi and MEKi were accessible outside of trials or precluded patients who were treated with BRAFi and MEKi [11, 31, 32]. To reflect a real-world situation available outside of clinical trials, we believe that is important to take these therapies into consideration.

Our study does have several limitations. First, the cohort was sampled based on availability of tissue as well as of the mutation status, regardless of other demographic and clinical parameters. Although this approach was performed in several other sound reports [3, 12, 16], sampling based on the sole availability of BRAF and NRAS mutational status may bear potential for bias. While it minimizes a potential survivorship bias, it may enrich for patients who underwent surgery or invasive biopsies in stage III or IV disease to obtain tumor material for the mutation analyses. Our collective may not be representative of the target population, because we included patients, in whom molecular testing was clinically indicated due to disease progression. Indeed, the high overall relapse rate of 85.8% demonstrates that our population was skewed towards high-risk melanoma patients. Thus, our results may not be generalizable to all melanoma patients and, in particular, may not be valid for individuals with lower risk disease. To estimate the effect of this bias, we performed time to event analyses including patients with relapse only. The time to loco-regional relapse was less favorable for NRAS-mutant patients and overall as well as relative survival significantly more favorable for BRAF-mutant patients after this selection. These data were largely consistent with our main findings within the entire population. Nevertheless, we cannot exclude that the lack of inclusion of lower risk disease potentially reinforces selection bias.

In addition to a potential sampling bias, further shortcomings of our study include the retrospective design and low number of patients with rare mutations of BRAF and NRAS. The mutation BRAF V600K had previously been associated with older age, head and neck localization of the primary tumor, shorter distant metastasis-free and shorter OS compared to V600E [11, 18, 33]. We identified 15 patients with V600K-mutant melanoma accounting for 6.9 and 17.2% of the entire and BRAF-mutant population, respectively. This proportion is somewhat lower than reported before and precluded further subgroup analyses due to the small number of cases. Non-V600 mutations of BRAF were reported in 7 (3.2%) patients in our study, two of whom had a concomitant mutation in NRAS. More specifically, the mutations K601E and L597S were identified in 5 (2.3%) patients. The small number limited further statistical analyses and conclusions on this particular subpopulation. Others have provided evidence that the alterations K601E and L597 show distinct clinicopathologic features from general V600E/K mutations. The so far largest series on these mutations suggested that tumors harboring K601E mutations are rather similar to tumors with V600E mutations, whereas those with L597 mutations were akin to BRAF WT tumors [34]. However, data regarding treatment responses to BRAFi and the clinical prognosis is sparse due to the paucity of these mutations and exclusion from major trials [34, 35].

We provide important insights into the significance of BRAF and NRAS mutations in a real-world setting when BRAFi, MEKi, and ICB were largely available. NRAS-mutant melanoma patients showed a significantly higher portion of nodal relapse and the shortest time to loco-regional nodal relapse. Positive SLNB was a significant risk factor and treatment with ICB in stage IV a significant protective factor for OS in multivariate analysis. Even though the mutational status was not a consistent risk factor for OS in multivariate analysis, our results suggest that NRAS-mutant tumors tend to behave more aggressively than their BRAF-mutant and WT counterparts in a high-risk melanoma population.