In the last decade, the treatment of oncology patients has made a major shift from a one-size-fits-all approach towards personalized medicine. The use of genomic diagnostic technologies integrated in the clinical setting is a key tool for patients-tailored treatments [1, 2]. In widespread melanoma disease, treatments have been historically palliative and mostly ineffective. The identification of oncogenic mutations as drivers in advanced disease provide targeted therapies that have led to improvements in response rates and survival in a subgroup of patients [3–5]. The most relevant mutated gene is BRAF, which is mutated in about 50% to 70% of cutaneous melanomas, and the most common genetic change is a glutamic acid for valine substitution at position 600 (V600E) [6–9]. BRAF mutations may be classified based on their mechanisms to activate the mitogen activated protein kinase (MAPK) pathway into three categories [10, 11]: BRAF class I mutations with kinase activity involving codon 600; class II mutations, with high or intermediate BRAF kinase activity involving codons outside V600 locus and class III mutations, with no or low BRAF kinase activity. Mainly, BRAF class I mutations are targetable alterations with approved therapies [12]. Consequently, BRAF testing is considered mandatory in patients with advanced (unresectable or metastatic) melanomas stage III or stage IV, and is highly recommended in high-risk resected disease stage IIC [13]. Traditionally, these analyzes have been based on the isolated BRAFV600E identification using real-time quantitative polymerase chain reaction (RT-PCR), but clinically validated next-generation sequencing (NGS) panels covering key oncogenic drivers are increasingly being performed routinely. Moreover, when the result of a case is negative for class I BRAF alterations, it is recommended to sequence the other BRAF loci (class II and III alterations) to confirm the wild-type (wt) status, as well as sequencing NRAS and c-kit genes, in order to stratify patients in clinical trials [11, 13]. Thus, as therapies for locally advanced and metastatic melanoma become more complex, and more targetable drivers’ alterations modulating the clinical decisions are found, there is an urgent need to validate and establish diagnostic algorithms for molecular testing at every institution according to established guidelines [14–16].

To achieve a more complete molecular diagnosis, the main objective of the present study was to evaluate the implementation of the diagnostic oncomine solid tumor (OST) NGS panel, based on the ion torrent sequencing technology, in formalin-fixed paraffin-embedded (FFPE) tissue samples to replace the current RT-PCR routine diagnostic tool for BRAFV600 locus test. Thus, we performed a comparison of both approaches (RT-PCR and NGS) for BRAFV600 status. Additionally, we described the spectrum of somatic mutations identified by NGS in a series of advanced melanoma samples in our clinical setting.

As personalized medicine is constantly evolving, the identification and implementation of techniques to identify new actionable genes is of great importance. Integrated genomic diagnostic technologies, such as NGS gene panels have revolutionized clinical genomics and emerged as key drivers of personalized cancer care and cancer discovery, because they can simultaneously screen for multiple genes in a single assay optimizing tumor tissue requirements. NGS gene panels can be commercial or custom designed, depending on the specific gene targeting needs, although a validation is needed in both scenarios [19, 20]. Here, we have presented valuable real data on the implementation of a commercial and validated test with European conformity in vitro diagnostic for multiple molecular-targeted therapies, including BRAF, in a Spanish cohort of advanced melanoma. To this end, we have compared the Idylla BRAF test (RT-PCR), a proven robust and reliable technique for the characterization of BRAF [21, 22], against the OST DNA kit of 22 genes (NGS) in a subset of samples. To the best of our knowledge, this is the first prospective NGS study and the largest in a Spanish cohort of advanced melanoma patients. In addition, the present study has used a commercial NGS with a panel of pan-cancer genes to optimize institutional resources and increase profitability by implementing it in different pathologies, including melanoma.

In our series, the most common location of melanoma was the skin, as it has been described in other Western countries and Japan [23, 24]. We observed the expected frequency of described BRAF mutated melanomas (38%) in the subset of samples where the technique was performed [3, 25, 26]. Furthermore, we obtained a good concordance between RT-PCR and NGS techniques (Kappa = 0.89, 95% CI 0.68 ± 1.05) in agreement with other studies [20–22, 27, 28]. Only two discrepancies were found in the comparison study (BRAFV600K and BRAFL597S), but BRAFL597S mutation was not detected by RT-PCR because it was not covered by the test used. BRAF discrepancies between Cobas (RT-PCR) and NGS techniques have also been reported in selected cases, as in the study by Reiman et al. [20].

Despite the good concordance obtained between RT-PCR and NGS, the turnaround time and cost of the NGS as compared to the RT-PCR test, might lower its applicability for daily diagnosis for some institutions given that it requires experience in the wet lab, specific platforms, bioinformatics and it is a more complex assay. The simplicity of the technology chosen is being considered in busy routine laboratories and many still prefer the solution of PCR techniques. Nonetheless, de Unamuno Bustos et al. [29], demonstrated the cost-effectiveness and lower turn-around-time for NGS-based analysis compared to conventional methods in melanoma samples. Our results provide further information on the great benefits of using NGS tests. The OST panel in our study found 75 mutated samples (65%) and delivered additional information beyond mutations in BRAFV600 locus. BRAF class I mutations, consistent with published literature, were the most common mutation identified (39%) and the most predominant among cutaneous melanomas, but not the only class mutation type identified [8, 26, 30]. Guidelines for negative cases of BRAF class I alterations recommend sequencing the loci of the other known less frequent BRAF mutations (class II and class III BRAF alterations) to confirm BRAF wt status, as well as to test these cases for NRAS and c-kit mutations. Unlike RT-PCR, NGS enables testing of these and other alterations in once without performing further tests and reducing the final turnaround time. The main weakness of our panel is the absence of c-kit gene, reported in mucosal and acral melanomas. However, these are infrequent types of melanoma and represent 11% of the cases included in our series.

NGS allowed identification of BRAF class II alterations, as well as NRAS mutations, which may be relevant for ongoing or future clinical trials. NRAS mutation, described as the second most common oncogenic alteration in melanoma (20%), was associated with shorter survival in early and late stage [26, 31]. We found an NRAS mutation in 14% of melanomas, and a single alteration was identified in most cases (62%). NRAS mutations are usually mutually exclusive with other oncogenic drivers, such as BRAF, KRAS and c-kit mutations [32, 33]. However, we only observed a coexistence of NRAS and BRAF mutations in few cases (two samples), consistent with the results from previous studies [29, 31, 34]. No co-mutations were observed between NRAS and KRAS or BRAF with KRAS mutations.

Sex biased in cancer driver genes and biomarkers have been previously described [35], and our work is one of the few that studied a possible association between sex-biased mutations in melanoma. Although sex was not a significant factor in our study to predict the presence or absence of an oncogenic alteration, we observed that NRAS mutations were predominant in women (P = 0.03), whereas TP53 mutations were more frequent in men (P = 0.03). These results provide further evidence on the profit that BRAF wt and women patients could gain from NGS testing and the total number of patients (53%) that could benefit of targeted therapies. Less common mutations were only observed in men (ERBB4, STK11, FGFR1, FGFR3, KRAS, MET and AKT1). Therefore, these results suggest, as Gupta et al. [36] have previously observed, that there is a gender disparity in the accumulate number of mutations in melanoma with men having a higher and more varied number of somatic mutations than women.

Overall, NGS platforms should be designed and/or selected according to the needs of each institution. There are currently different NGS guidelines providing recommendations to serve as a reference for NGS in house implementation and validation [15, 37]. Minimum samples requirements, such as the presence of at least 20% of tumor cells on FFPE sample is an important criterion to consider, in order to prevent false negative results. Thus, tissue quality control for molecular testing is essential to assess whether the percentage of tumor in a given specimen reaches the detection threshold of the specific molecular test. An important step is to develop an algorithm to select which patients will be assessable for NGS analysis and preserve FFPE samples when necessary for other molecular tests.

At our institution we chose the OST panel to use as the sole platform not only in melanoma but also in other solid tumors, optimizing tumor tissue material, resources and personnel. This approach provides a complete molecular characterization of patients, identifying clinically actionable mutations as well as potential alterations associated with sensitivity to specific drugs.