It is estimated that melanoma will be the 19th most common worldwide primary site of new cancers in both sexes in 2020, with 324,635 cases [1]. Very recently, immune evasion by cancer cells has become an important therapeutic target. The prognosis varies according to the stage of the disease, from almost 99% 5-year survival rate in localised disease to approximately 27% when distant metastases are present.

As such, being able to predict which patients have the highest risk of developing distant metastases is quite important. Tumour biomarkers can be useful in predicting the risk of metastases and thus prognosis. Some of them can also have a diagnostic use. The use of serum biomarkers, such as lactate dehydrogenase (LDH) or S100b, is recommended in some guidelines, while the use of other serum biomarkers, such as melanoma inhibitory activity (MIA) and vascular endothelial growth factor (VEGF) is limited due to low specificity and limited clinical usability. DNA point mutations in melanoma represent another important biomarker that can guide patient selection and predict treatment response, and are currently recommended by all clinical guidelines. For example, mutations of the Mitogen-Activated Protein kinase (MAPK) pathway, most commonly identified in BRAF and NRAS, have been correlated with shorter survival outcomes and response to selective inhibitors of BRAF mutant protein, such as dabrafenib and vemurafenib. Other gene mutations offer similar correlations and represent potential therapeutic targets, such as NRAS gene mutations. The emergence of circulating tumour DNA (ctDNA) as an alternative source of DNA for genomic studies may become in the future the basis in clinical practice for tumor mutational analysis, staging of disease and consequent prognostication.

In Alexander Breslow’s report in 1970, tumour thickness and cross-sectional tumour area were identified as prognostic variables reflecting tumour burden. The thickness of the primary tumour is considered a significant prognostic factor for stage I and II melanoma; overall, 5-year survival rates in stage III melanoma are based on thickness. When the thickness is less than 1 mm, the 5-year survival rate is 53%; while when it is 1–2 mm, 5-year survival rate is 47%; when 2–4 mm, 5-year survival rate is 40%; and when over 4 mm 5-year survival rate falls to 34%. However, thickness of the primary tumour was not found to be prognostic after tumour metastasis (stage IV). Melanoma ulceration is another important prognostic factor. In literature, there have been two possible explanations of the adverse prognostic value of ulceration in primary melanoma. One possibility is that melanoma ulceration could directly enable dissemination of the tumour. Alternatively, it could be that ulceration is a biological attribute of tumours with a predisposition to disseminate.

Proliferative activity of the tumour and overexpression of c-myc have been found to favour both dissemination and ulceration of the primary melanoma. The hypothesis that melanoma ulceration directly enables the dissemination of the tumour through alterations in the local environment has been indicated in studies on the interactions of melanocytes and keratinocytes. These studies indicate that ulceration may provide melanoma cells with a very effective way to interrupt the keratinocyte-mediated control that prevents melanocyte transformation. Mitotic activity of the primary tumour has also been investigated as a prognostic factor. In DNA replication genes of two pathways are over-represented: replication origins firing (ROF) genes and the separation of sister-chromatids by securin. For example, overexpression of ROF genes in melanoma is associated with poor prognosis. Expression of MCM4 and MCM6 genes is associated with metastasis-free survival and overall survival (OS). Securin is encoded by the hPTTG gene, which acts as an oncogene. Its expression is seen via immunohistochemical staining in the vertical growth phase but not in the radial growth phase of melanoma. Some promising new prognostic tissue biomarkers have also been reported in the literature, including cycloxygenase 1–3 (COX1-3), galectin-3 molecule, matrix metalloproteinases (MMP), and chondroitin sulfate proteoglycan 4 (CSPG4). COX1-3 converts arachidonic acid to prostaglandin. In Becker et al., COX-2 staining intensity was found to correlate to Breslow thickness in melanoma. Galectin 3 and tumour size were found to be inversely related and correlated with OS. MMPs are key to remodeling of the tumour tissue microenvironment. MMP-1 and MMP-3 positive melanoma metastases were associated with reduced disease-free survival (DFS). CSPG4 is believed to be essential in cell adhesion, melanoma migration and metastasis; over 80% of melanomas have been found to be expressing CSPG4. However, it can be found in any disease stage and there is no concrete evidence that it correlates to disease progression. Finally, several recent studies have demonstrated that the receptor for advance glycation end products (RAGE) signaling from both melanoma and non-melanoma cells (fibroblasts, immune cells, endothelial cells) in the tumor microenvironment represents an important element in the process of melanoma tumor growth. The RAGE/ligand axis appears to support the association between chronic inflammation and immunosuppression. Therefore, targeting RAGE in melanoma tumors could be therapeutically beneficial.

The use of serum biomarker assays may identify the presence of residual or recurrent disease prior to imaging studies and relevant radiological evidence. From the therapeutic perspective, this is important, as the prediction or early identification of distant metastasis would enable the timely initiation of systemic therapy in adjuvant or metastatic settings.

LDH catalyzes the conversion of pyruvate to lactate in hypoxic or anoxic conditions. An elevated level of LDH is believed to be due to spillage into the bloodstream when melanoma cells outgrow their blood supply. High levels of LDH are associated with worse prognoses, independently of site or number of metastases. In the American Joint Committee on Cancer (AJCC) melanoma staging system, patients with distant metastasis and elevated LDH levels are considered stage IV M1c. Patients with stage IV disease and normal serum LDH at initial staging have 1-year OS of 65% and 2-year OS of 40%. With elevated LDH levels, 1-year and 2-year OS are 32 and 18%, respectively. Apart from its prognostic value, in patients treated with a combination of dabrafenib and trametinib, LDH was shown to be associated with poorer outcomes. Moreover, when LDH decreases by more than 27.3% from the baseline, this has been associated with radiological response to immunotherapy.

Melanoma inhibitory activity (MIA) is secreted by melanoma cells and is a regulatory growth factor. MIA was proposed as a melanoma biomarker, because it is not expressed in benign human melanocytes or benign melanocytic nevi, but is strongly expressed in malignant melanoma cells. Higher levels of MIA were linked with more advanced stages of melanoma and worse prognosis.

VEGF is elevated in patients with advanced-stage melanoma. This was associated with negative immune effects, such as impaired dendritic cell function. It was also linked with both elevated and decreased T helper 2 (Th2) cytokines. These were found to result in suppression of effective antitumour immunity. VEGF inhibitors can lead to improved dendritic cell function and reverse Th2 dominance, leading to Th1 polarity. These changes should in theory enhance tumour rejection.

Epigenetic alterations: hypermethylation.

Epigenetics is the study of potentially inheritable changes in the phenotype that do not involve alterations in the DNA sequence. Epigenetic DNA changes are crucial in determining which genes are silenced or expressed in processes like cell differentiation, cell growth and immune response, and have been extensively investigated in melanoma. DNA methylation occurs in 5-cytosine exclusively to produce methylcytosine, and the majority of this is seen in CpG dinucleotides. There are large clusters of CpG dinucleotides across all of the genome, referred to as CpG islands, typically in promoter regions. Hypermethylation of these promoter regions in cancer-related genes can facilitate tumour progression. This is because DNA promoter methylation can inactivate tumour suppressor genes (TSGs). Methylation changes in the CpG island promoter regions in TSGs or other tumour-related genes have been observed in malignant cutaneous melanoma. This phenomenon in melanoma has been named the CpG island methylator phenotype (CIMP).

The Ras association domain family 1 A (RASSF1A) is a TSG that can control the cell cycle, promote apoptosis and maintain the genome. In breast, lung and liver cancer cells, RASSF1A expression was lower; studies have demonstrated that this lower expression is due to methylation of its promoter region.

BRAF mutations may be correlated with ultraviolet (UV) exposure. Further research is needed to identify if this mutation could be useful in tracking the transformation of benign naevi to malignant melanoma. BRAF mutations also have implications for prognosis. Recent research has established that BRAF mutations could be linked with shorter DFS in early stages of melanoma. Furthermore, BRAF-mutated melanoma is connected with shorter survival in stage IV disease and V600E expressed in the nucleus (rather than the cytoplasm) is associated with more advanced tumour staging, lymph node metastasis and depth of invasion. With the identification of BRAF and its significant role in melanoma, therapies have been developed to target its specific inhibition.

Selective inhibitors of V600-mutated kinase, dabrafenib and vemurafenib, have been associated with improved OS and DFS.

However, the most concerning challenge currently is resistance to these BRAF inhibitors due to increasing new mutations and upregulation of receptor tyrosine kinase (RTK) or NRAS. Consequently, combination therapies using BRAF and MEK inhibitors such as cobimetinib and trametinib are becoming the standard of melanoma treatment. 

NRAS is a GTPase protein that integrates signals from multiple RTKs. NRAS mutations activate MAPK signaling similarly to BRAF mutations; these activated signalling pathways lead to dysregulated cell cycles along with cell proliferation and further survival pathways.

NRAS mutations have been observed in 15–25% of melanomas and typically develop later in life after UV exposure and on the peripheral extremities. NRAS mutations usually occur independently of BRAF mutations, but 10–20% of melanomas have point mutations in NRAS codons 12, 13 or 61 that may be mutually exclusive with BRAF mutations. If occurring independently, NRAS mutations may be able to bypass BRAF and signal through CRAF instead. NRAS mutations have been observed in melanoma, which suggests that UV radiation may play a significant role in introducing these mutations. However, these mutations may also arise independently of UV radiation as they have been found in congenital melanocytic naevi as well; in these cases, detecting the signature UV radiations may help to diagnose the melanoma . NRAS mutations are also found in melanocytic and dysplastic naevi and melanomas with a high mutation load.

It is unclear if NRAS has any prognostic value if identified, but NRAS expression is associated with higher tumour staging and lower grades of tumour infiltrating lymphocytes. In comparison with wild-type melanoma, NRAS mutations may result in a significantly worse melanoma-specific survival rate. This may be explained histologically by their inducement of thicker lesions, higher mitotic activity and increased lymph node metastasis. NRAS mutations activate the RAF/MEK/MAPK signalling pathway in a similar fashion to BRAF. There has been a focus on inhibiting downstream components of the Ras signaling pathway, in particular the farnesyl transferase inhibitors (FTIs). The toxicity observed is attributed to the fact that FTIs inhibit other proteins that require farnesylation.

GNAQ and GNA11 genes code for G alpha subunits of G proteins that act with G-protein coupled receptors. The conversion of GDP to GTP allows G protein and G-protein coupled receptor signaling and subsequent activation of G proteins; for G proteins to become inactive, GTPase hydrolyses GTP to GDP. These genes cause GTP to be constantly bound to the G protein and result in downstream signaling.

Identifying GNAQ or GNA11 mutations can be useful to diagnose uveal melanoma and differentiate it from other types of melanomas and melanoma of undefined origin. While they can be present in cutaneous melanoma, these cases are rare. In contrast to their diagnostic value, their use as a prognostic marker has limited evidence. Sheng et al. found the median OS to be shorter for patients with GNAQ and GNA11 mutated mucosal melanoma compared to the wild-type subsets. In contrast, other studies have shown GNAQ and GNA11 mutations are not associated with poor patient outcomes or disease progression in uveal melanoma, which could be due to the mutations being initial steps in the development of melanoma. Furthermore, there is no significant difference in OS or DFS in patients holding GNAQ mutations compared to GNA11 mutated melanomas.

NF1 gene encodes a protein (neurofibromin) which acts as a negative regulator of the RAS-dependent pathway. It is known to cause tumours with mostly neuroectodermal origin that consequently can often be found in melanoma. Desmoplastic melanomas have fewer DNA copy number alterations than other melanoma subtypes; nevertheless, the few focal deletions that have been observed targeting CDKN2A and NF1. NF1 mutations are found in up to 45–93% of these melanomas.

Finally, there is a subset of patients who do not express BRAF, NRAS, or NF1 mutations. Most often these tumors may have other MAPK mutations, AKT3 overexpression, or changes in cell cycle pathways and most likely need a novel therapeutic approach.

PI3K/AKT/mTOR is a critical regulator of many physiological processes and essential to the aggressive nature of the tumour, as this pathway promotes cellular growth and survival .

AKT family member mutations are often dysregulated in melanoma and have been identified in up to 43–60% of melanoma cases. PTEN, which classily dampens the PI3K/AKT/mTOR growth-promoting signaling cascade, is noted in 38% of patients with primary melanoma and 58% of patients with metastatic disease. Changes in PTEN and BRAF pathway often co-exist, theoretically allowing dysregulation of both the MAPK and PI3K pathways at the same time. Hence, it might be possible that PI3K inhibitors may afford some benefit to patients with PTEN and/or AKT-mutant melanomas. Rapamycin and its analogues were among the first to be tried, inhibiting mTOR. One of the reasons is that they are known to be well-tolerated clinically, as demonstrated by long-standing use in patients who have undergone organ transplantation. However, mTOR inhibitors have not demonstrated significant clinical activity as single agents in metastatic melanoma patients, nor when combined with RAF inhibitors. One of the potential reasons is the complexities of pathway inhibition in systems with significant cross-talk.

c-KIT is a transmembrane receptor tyrosine kinase. When binding a stem cell factor, it results in activation of several signaling pathways, thereby mediating cancer cell growth, proliferation, invasion, metastasis, and inhibition of apoptosis. The majority of c-KIT mutations are found in mucosal and acral melanomas, as well as in melanomas arising from skin. Past genotyping has shown that they are almost always mutually exclusive with BRAF and NRAS mutations. Expression of KIT is not uniform across the tumor; the highest levels of KIT expression are seen at the leading edge of tissue invasion, indicating a key role it may have in promoting metastasis. The presence of c-KIT mutations has shown to be associated with worse survival as compared with wild-type melanomas.

Unfortunately, because of the relative rarity of c-KIT mutations (1–7%), the availability of targeted therapy to treat this type of melanoma is limited. However, responses to tyrosine kinase inhibitors (imatinib, sorafenib) have been reported in patients with KIT-mutant melanoma.

CDKN2A gene encodes the p16 protein, thus mutations in this gene result in hyperphosphorylation of retinoblastoma protein. Hence, it regulates intracellular oxidative stress in a cell cycle-independent manner.

BAP1 is a tumor suppressor gene and is often associated with metastatic uveal melanoma. There is a known correlation with tumor aggression and worse prognosis in uveal melanoma, as well as greater risk of metastasis. BAP1 tumor predisposition syndrome (BAP1-TPDS) is a cancer syndrome that apart from the aforementioned uveal melanoma predisposes patients to other malignant disorders, including renal cell and basal cell carcinomas, lung and breast/ovarian cancers, meningioma, and malignant mesothelioma.

Therapeutic targeting of BAP1 mutation poses a challenge, as its mechanism in melanoma development is still poorly understood. Not only is the identification of the function of BAP1 responsible for its anticancer role unclear, but also the goal of therapy is complex as it aims to restore one or more functions of BAP1. Some studies are targeting alternative mechanisms of DNA repair. One focuses on poly (ADP-ribose) polymerase (PARP) emerging as a potential target for treatment. The main reason is its role in base-excision and nucleotide excision repair. Therapeutic target of BAP1 has focused on its role in DNA double-strand break repair via homologous recombination. Indeed, there is an ongoing clinical trial of the PARP inhibitor niraparib in BAP1-deficient neoplasms including uveal melanoma (NCT03207347).

Advances in next generation sequencing (NGS) have led to the identification of many important kinase fusions as the primary drive in melanoma, which may represent critical targets for molecular therapy.

NTRK1 fusions typically arise from small deletions, whereas the AGK-BRAF fusion arises through an inversion. Due to the fact that most studies are RNA-based, many of the precise genomic mechanisms are not yet characterised. Most fusions have been identified with a variety of N-terminal partners. Kinases activated by these gene fusions include ALK, RET, ROS1, NTRK1, NTRK3, MET, MAP3K8, MAP3K3, BRAF, and PRKCA. The subsequent activation of downstream RAF/MEK/MAPK, PI3K/AKT/mTOR, and PLC pathways promotes cellular proliferation and migration. These fusions are mutually exclusive of one another, as well as of other driver mutations previously reported, such as NRAS, HRAS, GNAQ, GNA11, NF1, and KIT.

Fusions may be the initiating genomic event in 8–20% of melanomas. From a therapeutic perspective, BRAF or MEK inhibitors may be effective, as they reduce tumor size and proliferation. It seems that NGS has become a useful tool in screening for targetable fusions in advanced melanomas that lack characteristic driver mutations.

Circulating tumour DNA (ctDNA) is highly fragmented single or double-stranded DNA that is shed by tumour cells into the circulation. ctDNA assays constitute a powerful tool for study of the molecular heterogeneity and clonal divergence of a malignancy. Levels of ctDNA can vary depending on tumour vascularity, location and cellular turnover . Generally, undetectable ctDNA and favourable molecular profile carry a better response, progression-free survival and OS compared with detectable ctDNA at baseline or during treatment. ctDNA may also be useful as a biomarker of disease recurrence after melanoma resection. As expected, ctDNA is usually undetectable in early-stage melanomas. In later stages, the presence of BRAF and NRAS mutations is associated with response to immunotherapy. Patients with an undetectable ctDNA either at baseline or during treatment achieve a better objective response as compared to those with detectable ctDNA at baseline which remained detectable during therapy. Moreover, baseline levels of ctDNA were lower in melanoma patients who responded well to targeted therapy. The analysis of methylated ctDNA using methylation-specific PCR in metastatic melanoma has yielded promising associations. Indeed, methylated RASSF1 may serve as an indicator of response to hormonal treatment. ctDNA may be detected as loss of heterozygosity (LOH) in DNA microsatellites. This was seen in a study where the plasma of 76 patients with stage I to IV melanoma was analysed for microsatellite loss using 10 markers. LOH of at least one marker was found in 50% of patients and was correlated with the stage of the disease.

miRNAs are short noncoding RNA molecules (20–200 nucleotides) that regulate gene transcription processes, which in turn affect cell proliferation, apoptosis, cell differentiation and cell survival. Long noncoding RNAs (lncRNAs), with more than 200 nucleotides, also regulate transcriptional, post-transcriptional and epigenetic gene expression modulation. In contrast to ctDNA, miRNAs and IncRNAs are relatively stable, because they are predominantly secreted in vesicles, or in complex with other proteins.

Identification of miRNAs and lncRNAs can provide valuable information in diagnosis and prognosis and offers predictive value in melanoma. It seems that their use is limited due to low specificity and difficulty in attributing whether an increase in levels is due to cancer or due to other conditions, such as inflammation. However, in the era of personalised medicine, the relationship between aberrant miRNA profile and response to therapeutic regimens should be further evaluated. Therapeutic targeting of miRNAs can impact the natural history of melanoma by enhancing sensitivity to both standard therapies and immune checkpoint inhibitors. In particular, elevated levels of miRNA-221 have been identified in early melanomas, compared to healthy individuals. Increased levels were also linked to increased stage of disease. In a recently published study, circulating miRNA-615–3p levels were consistently more efficient in detecting melanoma patients who developed progressive disease whilst treated with immune checkpoint inhibitors, as compared to LDH levels. A panel of five miRNAs was used to classify primary melanoma patients as low or high risk of recurrence. In serial testing, dynamic changes reflected tumour burden. Finally, a study demonstrated that specific circulating miRNA signatures may distinguish melanoma brain metastasis from other types of brain cancer metastases as well as primary glioblastomas. Several lncRNAs were also found at high levels, including SPRY4-IT1, BANCR, HOTAIR, UCA1 and MALAT-1. In particular, levels of UCA1 and MALAT-1 were significantly higher in patients with melanoma compared to controls and were correlated to the stage of the disease. Finally, the potential of targeting ncRNAs for the development of novel therapeutic strategies or for the optimization of the efficacy of standard treatments has been assessed in several studies.