Clinical Correlates of NRAS and BRAF Mutations in Primary Human Melanoma

Cutaneous melanomas have long been known to harbor activating mutations of the NRAS gene in 15% to 20% of primary tumors (1). Although definitive clinical correlates of mutated NRAS have not been established in the melanoma patient population, at least 1 study has suggested aggressive behavior of this melanoma subset (2). The most common site of mutation is codon 61 in exon 2, although exon 1 mutations are occasionally found (1–3). The codon 61 mutations are heterogeneous, with C181A (Q61K) and A182G (Q61R) being found most frequently. There is in vitro evidence to suggest that codon 61 mutations result in prolongation of the GTP-bound state of N-Ras, leading to enhanced signaling through growth promoting pathways such as the mitogen-activated protein kinase (MAPK) pathway (4).

In this context, considerable excitement was generated some years ago when activating mutations of BRAF, which encodes the kinase immediately downstream of N-Ras, were reported in 60% of cultured melanoma cell lines and 50% of primary human tumors (5, 6). These mutations for the most part consist of T1799A (V600E) in exon 15 and are rarely found in melanomas bearing NRAS mutations. Contrary to expectations, clinical studies have been largely unsuccessful in demonstrating a clear phenotype associated with mutated BRAF (7–10). Adding to the debate surrounding the clinical relevance of BRAF mutations in primary melanoma is the observation that these mutations occur in benign nevi at a similar or greater frequency (11).

In an attempt to further clarify the impact of mutated NRAS and BRAF on clinical melanoma behavior, we determined the mutational status of 223 primary tumors and correlated these findings with pathologic and clinical data. Here, we report that NRAS- and BRAF-mutated tumors are more invasive than tumors wild type at both loci. Furthermore, patients with tumors carrying either mutation are more likely to present with stage III disease.

The study was approved by The University of Texas M. D. Anderson Cancer Center (MDACC) Institutional Review Board and conducted according to the Health Insurance Portability and Accountability Act (HIPAA) guidelines. A written informed consent was obtained from all subjects. Tumor samples consisted of consecutive entries into the MDACC Melanoma Informatics, Tissue Resource, and Pathology Core (Melanoma Tumor Bank) as part of a larger MDACC Melanoma Specialized Program of Research Excellence (SPORE) project.

For each primary tumor, three 5 μm thick, formalin-fixed, paraffin-embedded tissue sections were prepared for laser capture microdissection (LCM) as previously described (12). Microdissection of tumor cells was performed by a board-certified pathologist (VRG) using the PixCell II Laser Capture Microdissection System (Arcturus). Dissected cells were incubated for 72 hours at 37°C in 50 to 100 μL of lysis buffer (10 mmol/L of Tris-HCl, pH = 8.0; 1% Tween-20; 1 mmol/L of EDTA; and 0.04% proteinase K). After a 5-minute high-speed centrifugation, the sample was heated to 95°C for 8 minutes to inactivate the proteinase K. PCR was performed using the GeneAmp Gold PCR Reagent Kit (Applied Biosystems). Primers and conditions for NRAS exon 2 and BRAF exon 15 amplification (Sigma Genosys) are listed below (5, 13):

• NRAS exon 2: Forward: 5′-CCCCTTACCCTCCACAC-3′

•    Reverse: 5′-AGGTTAATATCCGCAAATGAC- 3′

•    95°C 45 seconds, 55°C 45 seconds, 72°C 60 seconds; 40 cycles; magnesium chloride 1.5 mmol/L

• BRAF exon 15 Forward: 5′-TCATAATGCTTGCTCTGA TAGGA-3′

•    Reverse: 5′-GGCCAAAAATTTAATCAGTGGA- 3′

•    94°C 30 seconds, 57°C 60 seconds, 72°C 60 seconds; 40 cycles; magnesium chloride 1.5 mmol/L

PCR products were purified using QIAquick PCR Purification Kit (Qiagen) according to the manufacturer's instructions. Sequencing of the PCR products in both directions was performed by the MDACC DNA Core Facility using an ABI Prism 3100 DNA Genetic Analyzer (Applied Biosystems) using Big Dye v.3.1 dye terminator chemistry (Applied Biosystems). Chromatogram results were interpreted independently by 2 reviewers (JAE and VRG). All mutations were identified on both strands.

Associations with mutation types were evaluated using the 3 genotypes (mutated NRAS, mutated BRAF, and both wild type), as well as mutational status collapsed into 2 categories (either mutation vs. no mutation). Fisher's exact test was used to examine associations between mutation status and the following factors: patient gender, Breslow thickness ≤ 1 mm, Clark level, ulceration, and American Joint Committee on Cancer (AJCC) staging. The associations between genotype and ordinal variables (patient age, Breslow depth, and mitotic rate) were determined using the Kruskal–Wallis test. Overall survival was computed from the date of pathologic diagnosis until the date of death. Patients alive at the end of the study period were censored at the date of last follow-up. Survival curves were constructed using the Kaplan–Meier method and the log-rank test was used to evaluate equality across strata. Associations between independent variables and survival were further investigated using Cox proportional hazards regression models. All P values were 2-sided and P values < 0.05 were considered statistically significant. Analyses were conducted using SAS for Windows (release 9.1, SAS Institute).

A total of 297 primary melanomas were microdissected, representing consecutive entries into the MDACC Melanoma Tumor Bank. Of this initial set, sequences for both BRAF exon 15 and NRAS exon 2 were successfully determined in 223 cases. The remaining 74 samples were excluded from analysis for the following reasons: one or both exons could not be adequately amplified or sequenced (39 cases); the material remaining in the tissue block did not show invasive melanoma (21 cases); the tumor was too small for microdissection (14 cases).

On the basis of the results of PCR and sequencing, each of the 223 cases was assigned one of the following genotypic designations: mutated NRAS (MN), 31 patients (13.9%); mutated BRAF (MB), 109 patients (48.9%); wild type at both loci (WW), 80 patients (35.9%); and mutated at both loci (MM), 3 patients, (1.3%). For analytical purposes, the 3 MM cases were examined separately, such that the final analysis included 220 subjects. The study population consisted of 118 males and 102 females, with a median age of 49 years (range = 18–77 years). The distribution of histologic types included superficial spreading 70%, nodular 19%, lentigo maligna 4%, and unclassified 7%. The rates of specific mutations are shown in Table 1.

To test the hypothesis that mutation-bearing tumors display a more aggressive clinical phenotype, various relevant patient characteristics and histologic features were compared among the 3 genotypes. These included age, gender, Breslow depth, Clark level, mitotic rate, and the presence of ulceration (Table 2). Results show that the median Breslow depth varied significantly among the 3 genotypes, with MN tumors showing the deepest invasion (1.40 mm) and WW the least invasion (0.93 mm), whereas MB tumors were intermediate (1.28 mm; P = 0.006). The same pattern was seen if Breslow depth was examined at a cutoff of 1 mm (P = 0.021). Clark levels followed suit, with mutated and WW tumors showing the most and least aggressive patterns, respectively (P < 0.001). Ulceration also differed significantly among the genotypes, with MB tumors demonstrating the highest rate, whereas rates for MN and WW were considerably lower and similar to each other (P = 0.045). The 3 genotypes did not differ in terms of gender, age, or mitotic index.

All patients were assigned a pathologic stage according to 2009 AJCC criteria (14). As shown in Table 3, there was a trend for patients with either mutation to present with a higher stage than those with wild type tumors (P = 0.104). If the MN and MB categories were collapsed, these findings reached significance (P = 0.027). In particular, 52 subjects presented with stage III disease and patients in this subset were more likely to have mutated tumors (P = 0.031 for MN or MB vs. WW). As further seen in Table 3, the majority of stage IIIC tumors, that is those with larger volume nodal disease, carried one or the other mutation (findings not significant). In spite of this, there was no difference in survival among the stage III patients when stratified according to the presence or absence of mutation (Fig. 1). Furthermore, with a median follow-up of 39 months, overall survival did not differ between the 3 mutational groups as a whole or stratified by stage at presentation (data not shown).

It has been noted previously that NRAS mutated tumors display a propensity for developing on the upper extremities (10). A similar finding was seen in our patient population, with 90% of MN tumors located on either extremity and only 6% on the trunk (Fig. 2). In contrast, MB tumors were localized to the extremities in 41% of cases, with a similar proportion (46%) found on the trunk. The site distribution of WW tumors was similar to that of MB tumors. Differences in location among the 3 genotypes were statistically significant (P < 0.0001).

Three tumors carried mutations in both BRAF exon 15 and NRAS exon 2, designated MM (Table 4). The ages of these 3 patients (58, 58, and 57 years) were somewhat greater than the median age of 49 years for the other groups. Also, interesting was the presence of tandem nucleotide mutations in BRAF exon 15 in 2 of the 3 tumors. Otherwise, the MM tumors were not remarkable for any distinct features and fell across the spectrum of the other 3 genotypes. One MM patient was found to have regional nodal metastasis on presentation and underwent a successful lymphadenectomy. All 3 patients remain alive and disease-free at 40, 44, and 49 months from diagnosis.

Another uncommon mutational finding was the exclusive presence of mutated DNA, whether NRAS or BRAF, as indicated by a solitary mutation peak on the sequencing chromatogram. This can occur as a result of identical mutations on both gene copies or a mutation on 1 copy and deletion of that stretch of DNA on the other. This finding was observed in the tumors of 5 patients, 2 with mutated NRAS and 3 with mutated BRAF (Table 4). Notably, all 5 tumors displayed aggressive features with a median Breslow depth of 3.8 mm and a median mitotic rate of 5 figures/mm². Three of these patients developed regional metastases, but all remain alive and disease-free with a median follow-up of 49.5 months.

Since the recognition of the high frequency of BRAF mutations in cutaneous melanoma, this anomaly has become a favored target for drug design. However, justification for such directed therapy has largely been based on in vitro studies indicating that the BRAF V600E mutation is activating and drives the MAPK pathway. The data that we now present provide clinical justification for the therapeutic targeting of mutated B-Raf, as well as mutated N-Ras, in the treatment of melanoma. Our findings demonstrate a correlation of NRAS and/or BRAF mutations with other factors well known to negatively influence prognosis, such as invasion and ulceration, and are consistent with our previous report that NRAS and BRAF mutations are acquired as melanoma cells progress from the radial to the vertical growth phase (12). In keeping with these observations, patients with mutated tumors are more likely to present with regional metastases. It is interesting to note, however, that in our study population to date, survival does not differ between stage III patients whose primary tumors do or do not carry mutations, even though the mutated tumors tended to produce larger volume nodal disease.

A similar attempt to establish clinical correlates of NRAS and BRAF mutations in melanoma has been published by Edlundh-Rose et al. (10). This group examined 294 tumors, the majority of which were metastases, and reported the characteristics of the primary tumors from which they were derived. Although this approach selected for tumors that eventually metastasized and assumed genotypic agreement between primaries and metastases, many of their findings were consistent with ours, particularly the invasiveness of (presumed) NRAS-mutated primary tumors and lack of survival differences based on genotype. Taken together, these data suggest that the effects of NRAS and BRAF mutations may be limited to early disease stages and that other factors are more influential after regional metastases have occurred. It is also possible that the mutation pattern in our study population differs between the primary and metastatic tumors, as we have not sequenced the exons in question in metastases from these patients. However, the literature would suggest that the mutational status is maintained throughout the various stages of disease progression (15).

The high frequency of ulceration in BRAF-mutated tumors is notable, reaching a rate of 22% compared with around 10% in NRAS-mutated and wild-type tumors. Because BRAF and NRAS are components of the same signaling pathway, it is difficult to reconcile the differences in ulceration rates on the basis of downstream MAPK effectors. Implicit here is that the mechanism of melanoma ulceration, although poorly understood, may be linked to factors unique to BRAF mutation and may ultimately shed light on BRAF-specific molecular processes.

The difference in tumor localization based on genotype is compelling, and here, NRAS mutants distinguish themselves from the 2 other genotypes with their propensity for the extremities. BRAF-mutated tumors have been reported to arise less frequently in skin with chronic sun exposure and actinic damage (16). Although less well documented, melanomas with NRAS mutations are reported as more common on skin with continuous sun exposure (3, 10). Our data generally support these previous reports in terms of sites more likely to be sun-exposed (extremities) or sun-protected (trunk). To further address the suggested association of mutations and sun exposure, we are presently examining the tumor sections for solar elastosis, an accepted marker of chronic sun damage. Alternatively, other differences in the external environment or even differences that occur in melanocytes as they migrate to and develop in central versus peripheral regions of the body may influence the risk of one or another mutation.

In conclusion, our data provide convincing evidence for distinct clinical phenotypes of melanomas bearing NRAS and BRAF mutations, whether considered together or separately, and largely point to an inferior patient outcome. Conversely, patients with tumors wild type at both loci might be expected to exhibit a less aggressive form of this disease. Although melanoma has reliable histopathologic predictors of tumor behavior, genotypic data may assist in decision making for patients with borderline cases, particularly those at risk for complications of sentinel node biopsy or general anesthesia, or patients generally reluctant to undergo invasive procedures. Our ultimate goal is to strengthen and expand these correlative findings as we continue to follow this patient cohort.

Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health.

The authors thank the members of the MDACC Melanoma Informatics, Tissue Resource, and Pathology Core for carefully processing and providing the materials used in this study.

This work was supported by NIH P50 CA093459 and NIH CA16672 (DNA Analysis Facility).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.