Prognostic molecular testing in melanoma: ready for prime time?

Prognostication of patients with cutaneous melanoma has long been dependent on clinical variables. Molecular testing has an increasingly important role in the management of malignancies. Early studies on molecular profiling of melanoma were replete with technical challenges. The early molecular profiling platforms required robust RNA and could only be obtained from frozen tissue. Thus, early studies tend to include mostly patients with advanced stages of melanoma where fresh tissues can be obtained with relative ease. With the advancement in molecular techniques such as cDNA-mediated annealing, selection, extension and ligation technique, genetic information can be reliably obtained from formalin-fixed, paraffin-embedded (FFPE) archival tissues [1]. Recent reports using whole-genome RNA profiling, miRNA profiling, gene-expression profiling (GEP) with 31-gene panel and NanoString technology have further advanced the development of molecular testing toward improving prognostication and integration in the routine management of melanoma patients.

Using global genetic profiling approach evaluating the expression of 3000 genes, Jonsson et al. [2] identified four molecular subtypes of melanoma: high-immune, pigmentation, proliferative and normal-like. Each molecular subtype was characterized by gene signatures representing diverse molecular mechanisms. High expression of genes involved in immunological process such as IFNGR1, CXCL12 and IL1R1 was observed in high-immune subtype. In contrast, high expression of cell-cycle associated genes such as E2F1, BUB1 and CCNA2 was noted in proliferative subtype. The pigmentation subtype was characterized by higher expression of genes involved in melanin synthesis and melanocyte differentiation, for example, MITF, TYR, SILV, DCT and EDNRB. High expression of genes involved in epidermis and ectoderm development was observed in normal-like subtype. There was significant difference in overall survival (OS) rates between the four subtypes, with proliferative subtype having with the shortest survival. This molecular-profiling approach was subsequently validated in three independent cohorts [3–5]. Harbst et al. analyzed gene expression in archival tissues from 210 cutaneous melanoma patients and 13 mucosal melanoma patients [3]. Further refinement of the four molecular subtypes was performed to construct a binary classification into high-grade and low-grade subgroups. Significant difference in OS and relapse-free survival (RFS) among the subgroups was observed using the 4-group and 2-group classification schema. Cirenajwis et al. validated the prognostic utility of the 4-group molecular profiling approach in 214 cutaneous melanoma patients using archival tissues [4]. Nsengimana et al. validated the prognostic utility of 2-group molecular profiling approach in 300 archival tumors (224 primary melanoma and 76 metastatic melanoma) [5].

Further gene ontology analysis and mutation spectrum analysis within the context of the gene-expression phenotypes revealed wound response, immune response, DNA repair and cell cycle as important determinant biologic processes in melanoma [3–5]. CDKN2A alterations were more prevalent in the proliferative subtypes, and pigmentation subtype melanomas were enriched for genetic events in CTNNB1, MITF or CCND1 [4]. Hereditary breast cancer signaling and BRCA1 DNA damage response were two of the most significant networks associated with high-grade tumors [3]. High-grade tumors have increased level of genomic imbalance, and increased expression of MITF, its upstream regulator SOX10 and its downstream targets TYR and MLANA [3]. High expression of genes such as CCL21 and CXCL12 is characteristic of the low-grade tumors [5].

In contrast to global gene-expression profiling approach, Gerami et al. evaluated a 31-gene GEP panel in cutaneous melanoma based on a selective approach identifying genes of interest from the analysis of expression data from public databases and from a clinically validated 15-gene prognostic expression profile assay for uveal melanoma [6]. Majority of the genes evaluated were located on chromosomes 1 (CRABP2, TACSTD2, CLCA2, S100A9, SPRR1B and S100A8), 6 (GJA1 and ARG1), 9 (TYRP1 and AQP3) and 12 (MGP, KRT6B and BTG1). Gene ontology analysis revealed a representation of genes involved in epithelial differentiation and development as well as genes representing cellular components such as cell–cell junction and nonmembrane bound organelle classes. Following development phase of the assay utilizing archival tumor tissues from 107 stage I and II primary melanoma patients from three separate institutions, the assay was further refined in a training set composed of 164 stage 0–IV melanoma patients and validated in an independent 104 stage 0–IV melanoma patients. Patients were classified as either class 1 (low risk) or class 2 (high risk). The accuracy of the predictive modeling as determined by area under the receiver operating characteristic curve was 0.91 with an overall risk prediction accuracy of 83%. The gene-expression profiling classification correlated significantly with disease-free survival (DFS) and OS. Its prognostic utility was further confirmed on multivariate analysis with 5-year DFS of 97% in class 1, and 31% in class 2 validation set patients. The negative predictive value was 93% with a positive predictive value of 72%. This assay was further validated in another independent cohort of 217 melanoma patients who had undergone sentinel node biopsy [7]. In this separate validation study, archival primary melanoma tissues from 217 melanoma patients who had undergone sentinel node biopsy were evaluated with the 31-gene GEP assay. GEP classification and sentinel node status were independent prognostic variables for DFS, distant metastasis-free survival and OS. In addition, it was observed that combining GEP classification and sentinel node status further improved metastatic risk prognostication.

Over the last decade, miRNA has attracted significant interests from cancer researchers. The miRNAs are small noncoding RNAs that modulate gene expression of several key cellular processes by inhibiting protein translation and inducing mRNA degradation [8]. Because of their imperfect sequence complementarity with their mRNA targets, each miRNA has the potential to modify the expression of multiple genes. Because they are quantifiable, genome-wide and highly stable in FFPE tissues, miRNAs may serve as ideal biomarkers for cancer [9,10].

Several recent studies have demonstrated the utility of altered miRNA expression as prognostic biomarker for melanoma. Segura et al. reported a panel of six miRNAs (miR-150, miR-342, miR-455-3p, miR-145, miR155 and miR-497) that could predict postrecurrence survival with 80.2% accuracy rate [11]. Caramuta et al. proposed another panel of six miRNAs (miR-191, miR-193b, miR-365, miR-338, miR-193a and let-7i) in association with shorter survival [12]. In a study of 128 primary melanoma tissues, Satzger et al. reported a significant association of high miR-15b expression with RFS and OS on multivariate analyses [13]. Tembe et al. performed global miRNA expression profiling of 45 American Joint Commission on Cancer (AJCC) stage III metastatic melanoma patients [14]. The miRNAs: miR-142-3p, miR-142-5p, miR-146b-5p and miR-150-5p were observed to have differential expression of >1.5-fold in good prognosis compared with poor prognosis patients. Further evaluation in an independent cohort of 86 stage III and IV patients confirmed significant correlation of miR-142-5p and miR-142-3p expression with survival. Hanniford et al. reported that a 4-miRNA signature panel comprising of miR-150, miR-15b, miR-16-5p and miR-374b-3p, in combination with AJCC stage, was predictive of subsequent development of brain metastasis [15].

The miRNAs are thought to be shed from tumor in exosomes which would protect them from degradation by endogenous RNAses [16]. Thus, they are very stable in the bloodstream. Tembe et al. observed an association of serum miR-142-3p expression with stage of disease [14]. Serum miR-150-5p expression correlated with survival in stage IV patients [14]. Fleming et al. analyzed serum levels of 12 miRNAs in 283 melanoma patients and observed miR-15b levels significantly increased over time in patients who recurred [17]. Friedman et al. screened 355 miRNAs using a quantitative reverse transcription-polymerase chain reaction panel in sera from 80 melanoma patients at primary diagnosis (discovery cohort) [18]. Expression of five miRNAs (miR-150, miR-15b, miR-199a-5p, miR-33a and miR-424) were significantly correlated with RFS on multivariate analysis. Further analysis separating the patients into high- and low-risk groups by using expression level cutoffs confirmed significant correlation with RFS with sensitivity of 0.84 and specificity of 0.76. This association was further confirmed in an independent validation cohort.

Recently, a unique nucleic acid bar-coding technique (NanoString^®, WA, USA) has been introduced for molecular testing. This novel technology uses unique digital color-coded barcodes that hybridize to specific nucleic acid targets and allow for direct detection and quantitation of gene transcripts in a single reaction in a highly reproducible and sensitive manner [19]. Utilizing NanoString technology, Sivendran et al. analyzed RNA from 40 stage II–III FFPE primary melanomas with a 53-immune gene panel [20]. The gene panel signature was significantly associated with DFS and RFS. The proposed 53-gene panel was further confirmed in an independent test population of 48 patients. The gene signature was an independent predictor of nonprogression (p < 0.001), RFS (p < 0.001) and disease-specific survival (p = 0.024) in the validation cohort. Multivariate logistic regression showed that the 53-gene signature score was the best predictor of progression.

The future of molecular testing utilizing techniques such as whole-genome RNA profiling, miRNA profiling, 31-gene GEP assay and NanoString technology is bright. Although the reports were based on assays performed on archival tumor samples, they have shown promising utility in risk stratification of cutaneous melanoma. Significant work remains to be done in order for molecular testing to be available on a widespread basis and for these techniques to reach their prime. Further research is needed to improve the availability and ease with which molecular testing is implemented.