The Genomic Landscape of Melanoma and Its Therapeutic Implications

Melanoma is one of the most aggressive malignancies of the skin. The genetic composition of melanoma is complex and varies among different subtypes. With the aid of recent technologies such as next generation sequencing and single-cell sequencing, our understanding of the genomic landscape of melanoma and its tumor microenvironment has become increasingly clear. These advances may provide explanation to the heterogenic treatment outcomes of melanoma patients under current therapeutic guidelines and provide further insights to the development of potential new therapeutic targets. Here, we provide a comprehensive review on the genetics related to melanoma tumorigenesis, metastasis, and prognosis. We also review the genetics affecting the melanoma tumor microenvironment and its relation to tumor progression and treatment.


Introduction
Melanoma is one of the most aggressive cancer types with an estimated incidence of 325,000 worldwide in the year 2020 and is estimated to continue to grow in the future [1]. Melanoma can be classified into cutaneous melanoma, acral melanoma, mucosal melanoma, and uveal melanoma according to the location of its primary origin. Cutaneous melanoma involving non-glabrous skin is the most common melanoma subtype. It is associated with ultraviolet (UV) radiation exposure and is more common in lighter skin types [1]. Acral melanoma occurs on glabrous skin not commonly exposed to UV radiation and is the most common melanoma subtype observed in patients with darker skin types [2]. Mucosal melanoma is a rare (1.3% of melanomas) but aggressive melanoma subtype originating from melanocytes of mucous membranes [3]. Lastly, uveal melanoma, melanoma arising from the choroid, ciliary body, or iris, represents 5% of melanoma and is more prevalent in populations with lighter skin types [4,5]. The mortality of melanoma remains high, especially for thick lesions and metastatic diseases [6].
The genetic composition of melanoma is complex and varies among different subtypes and individuals, resulting in heterogenous treatment outcomes. Recently, a more in-depth understanding of the genomic landscape of melanoma has been made possible with the aid of next generation sequencing, a sequencing technology making large-scale whole genome sequencing more efficient and practical. Another recent technical advancement is single-cell sequencing, which provides genetic information at the resolution of a single cell. A more thorough understanding of tumor heterogeneity and its microenvironment has been made possible with this technology. The tumor microenvironment, especially immune cells, were shown to profoundly affect survival of melanoma patients [7]. This has become increasingly important as immune checkpoint therapy has become standard treatment for advanced melanoma. However, not all patients respond to immune checkpoint therapy [8]. As our understanding of the complex interaction of tumor cells and the microenvironment increases, opportunities to identify favorable treatment responders have emerged.
In this review, we aim to provide a comprehensive review on the genetic alterations related to melanoma tumorigenesis among different melanoma subtypes, gene mutations related to metastatic melanoma, and the treatment implications of different mutated genes.

Driver Mutations
Driver mutations occur in the early stages of oncogenesis (i.e., process of normal cells turning into malignant cells) and provide growth advantages to the cancer cells [9]. Identifying these mutations provides potential targets for therapy and information regarding prognosis. Four major genetic subtypes of cutaneous melanoma have been proposed by The Cancer Genome Atlas according to frequently detected hot-spot mutations: mutant BRAF, mutant RAS, mutant NF1, and triple wild-type (TWT), characterized by a lack of BRAF, RAS, or NF1 mutations [10]. The most commonly observed BRAF and NRAS mutations are mutually exclusive [10]. These hot-spot mutations (i.e., mutations observed in significantly increased frequencies compared to normal samples) are all regulators of the mitogen-activated protein kinase (MAPK) pathway involving Ras/RAF/MEK/ERK signaling. Within the TWT subgroup, various driver mutations were observed, including KIT, CTNNB1, GNA11, GNAQ, and EZH2 [10]. Similarly, most driver mutations in the TWT subgroup are MAPK-activating mutations, highlighting the critical role of the MAPK pathway in the pathogenesis of melanoma.
Other frequently involved mutated genes and molecular pathways in the carcinogenesis of melanoma are summarized in Table 1.

Cutaneous Melanomas
Cutaneous melanomas include superficial spreading melanoma (SSM) and lentigo maligna melanoma (LMM). These melanoma subtypes arise from areas of skin exposed to ultraviolet (UV) radiation, with SSM more common on skin with intermittent high-intensity UV radiation while LMM more commonly arises on areas with chronic UV radiation. These subtypes are the most common melanoma subtypes and are more commonly seen in patients with lighter skin types [11]. Due to their relation with UV radiation exposure, they frequently exhibit a high burden of somatic mutation showing high levels of UV radiation signatures, especially C > T substitutions [12]. In comparison, the genetic aberrations in acral and mucosal melanomas are dominated by structural variants [12]. About 50% of UV-related melanomas harbor BRAF mutations, with valine (V) to glutamic acid (E) substitution in codon 600 (p.V600E) being the most common mutation [10,[12][13][14]. Clinically, patients with BRAF mutations are younger than other subtypes [10,[13][14][15]. The second most frequently observed is the N/K/H-RAS mutated subtype (20-28%) [10,16]. Contrary to BRAF-mutant melanomas, patients with NRAS-mutant melanoma tend to be older and have chronic sun UV exposure [17]. Following RAS mutation, the third common is the NF1 mutated subtype (14%) [10]. Lastly, 5-10% are TWT harboring less frequently observed driver mutations [10]. Compared to melanomas with hot-spot mutations, TWT melanomas harbor UV signature mutations less frequently [10]. Different from SSM, LMM was shown to have an increased frequency of non-p.V600E BRAF, NF1, and TP53 mutations [15,18].

Telomere and Melanoma Tumorigenesis
Telomeres are DNA sequences located at the end of chromosomes. Its integrity is crucial in maintaining genome stability and is regulated by telomerase, a ribonucleoprotein enzyme that synthesizes telomeric DNA [43]. Telomerase reverse transcriptase (TERT) aberrations are the most common noncoding mutation (i.e., mutation of regions of DNA that do not code for amino acids) in melanoma expressed in early melanoma progression following MAPK pathway activation [44]. TERT promoter is an important regulator of telomerase, whose mutation alters telomere length and promotes tumor cell immortalization [45]. Short telomeres and TERT aberrations are associated with poor survival in melanoma [46][47][48]. In a study including 1019 patients on early melanoma (stage I and II disease) by Rachakonda et al., the hazard ratio for poor melanoma-specific survival was 2.05 (95% CI = 1.33-3.16) for short telomers compared to long telomers [46]. The effect of short telomers on survival was especially prominent in patients under 30 years of age [46]. Patients with TERT promotor mutation and concomitant BRAF or NRAS mutations were shown to have worse disease-free survival compared to patients with TERT promotor mutation or BRAF/NRAS mutations alone [47]. TERT mutation can be used to identify melanoma patients with poor prognosis, including relapsing, disease, and death [47]. Over 60% of UV-related melanomas exhibit TERT promoter mutations [10]. Within the four major genetic subtypes of cutaneous melanomas, TERT promoter mutation is observed in 75% of tumors with BRAF mutation but only observed in 6.7% of TWT melanomas [10]. Although less frequent than UV-related melanomas, 9-41% of acral melanomas [21,39] and 30% of mucosal melanomas [29,40] exhibit TERT alterations. These are more commonly copy number gains (i.e., gene amplification due to increased gene copies) rather than promoter mutations [40]. Different from other melanoma subtypes, TERT alterations are rare in uveal melanomas [41,42].

Cell Cycle Regulators
DNA replication in the process of cell proliferation of the human cell is regulated by several cell cycle regulators. These important players in cell regulation include cyclins, cyclin-dependent kinases (CDKs), CDK inhibitors, and the tumor suppressor retinoblastoma (RB) family of proteins [49]. Mutations in genes encoding these regulators may result in aberrant cell proliferation and cancer [49].
As driver mutations occur at the early stages of tumor development, cell cycle regulators play an important role in sustaining cell proliferation in carcinogenesis, resulting in tumor progression. For example, BRAF mutation alone is not sufficient for inducing melanoma as BRAF mutations are frequently observed in benign nevi [50,51]. Additional genetic alterations in BRAF-mutants are required for cancer development [50]. Commonly, aberrations in cell cycle regulation collaborate with driver mutations in the carcinogenesis of melanoma. Significant players in melanoma cell cycle regulation include the CDKN2A (cyclin dependent kinase inhibitor 2A) gene which encodes the tumor suppressor p16INK4A, p14ARF, CDKs controlling specific check points of the cell cycle, and the tumor suppressor retinoblastoma protein (RB). Aberrations of the p16INK4A:cyclin D-CK4/6:RB pathway is frequently observed in both cutaneous melanomas and acral melanomas with a prevalence of 68.8-82.7% [10,15,17,52,53]. CDKN2A is also the most significant germline mutation with high penetrance in familial melanoma [54].
Mutations of cell cycle regulators CDKN2A and CDKs were less common in acral melanoma compared with cutaneous melanoma [20]. Melanomas with cell-cycle aberrations are associated with a higher frequency of clinical ulceration, which is associated with worse prognosis [15].

Tumor Suppressor Genes
TP53 is one of the major tumor suppressor genes (TSGs) regulating cell division and is the most commonly altered gene in human malignancies [55,56]. Its encoded protein p53 plays a critical role in maintaining genomic integrity by promoting cell arrest and apoptosis. Mutations of TP53 result in rapid tumor progression and metastasis [56]. Despite being the most frequently mutated gene in cutaneous melanoma, BRAF hyperactivity alone is not sufficient for melanoma development without concurrent p53 inactivation [57]. Although frequently observed in various human cancers, the incidence of TP53 mutation (11-30%) is relatively lower in melanoma [58,59], indicating that alterations of other cell cycle regulators, including CDKN2A and its products p14ARF and p16INK4A, are more frequently involved in melanoma tumorigenesis [60]. Melanoma patients with TP53 mutations were shown to have worse prognosis compared to those without the mutation [61,62]. Moreover, TP53 mutation was shown to be associated with poorer response and poorer overall survival to cytotoxic T-lymphocyte antigen-4 (CTLA-4) blockade immunotherapy in metastatic melanoma due to impaired cytotoxic T-cell-induced apoptosis of tumor cells [61].
Similar to TP53, PTEN is also a critical tumor suppressor gene in melanoma development. It inhibits cell cycle progression by blocking the PI3K pathway [63]. PTEN inactivating mutations occur in the later stages of primary melanoma and are associated with poorer overall survival [63]. PTEN loss is also associated with impaired host T cell responses [63,64]. Hence, similar to TP53 mutation, PTEN loss melanomas also show inferior response rate to immune blockade therapies [65,66].

Metastatic Melanoma
Melanoma is an aggressive cancer and has the tendency to metastasize, leading to the patient's death [67]. Alterations of the MAPK pathway occur early in cutaneous melanoma tumorigenesis and continue to be amplified in later stages, predominantly by gene duplication as the tumor progresses [44,68]. Currently, there are no genetic mutations exclusively found in metastatic melanoma [44,68,69]. However, several hub genes associated with metastatic melanoma were identified using bioinformatic approaches. These genes include CDK1, BAP1, CXCL8, THBS1, KIT, DSG1, FLG, and PKP1 [70][71][72][73]. Within the identified hub genes, CDK1 was found to be associated with favorable prognosis [70]. CDK1 was also associated with increased lymphocyte infiltration, which is also an indicator of favorable prognosis [70]. On the contrary, DSG1, FLG, and PKP1 were poor prognostic factors [72]. Loss-of-function mutation of BAP-1 was associated with poor overall survival [71]. Recently, melanoblast-specific genes, such as KDELR3 and KDELR1, were identified to contribute to the tumor's ability to metastasize [74]. These findings may provide potential targets for melanoma treatment. Currently, CDK1 inhibition shows preliminary effects in inducing tumor cell apoptosis in colorectal cancer [75].

Tumor-Infiltrating Lymphocytes
The behavior of tumor cells is influenced by the surrounding tumor microenvironment, which consists of host immune cells, extracellular matrix, blood vessels, and stroma cells [76]. An important factor influencing melanoma prognosis is the interaction between the host immune cells and tumor cells (i.e., tumor stroma immunobiology). The prognosis of melanoma is more heavily influenced by the tumor stroma immunobiology than the genetic subtypes of driver mutations [10]. Tumor-infiltrating lymphocytes (TIL) have first been shown to be an independent positive prognostic factor for melanoma by Larsen and Grude in 1978 and confirmed by many later studies [77][78][79]. Consequently, as shown in The Cancer Genome Atlas (TCGA) program, advanced cutaneous melanomas expressing genes associated with increased immune cell infiltration and signaling are associated with a favorable survival after adjusting for potential contamination by lymph node tissues [10]. However, it is worth noting that the subclass of lymphocytes also affects prognosis. For example, increased infiltration of regulatory T cells is associated with worse prognosis [80]. Nonetheless, lymphocyte subclass analysis was only performed in limited studies, which may lead to potential biases [80]. Interestingly, in contrast with other melanoma subtypes, increased tumor infiltrating in UM is associated with worse prognosis [32], indicating that UM is a unique subset of melanoma.

Targeting the MAPK Pathway
Early studies have shown that BRAF-mutant tumors have worse prognosis (median survival 5.7 vs. 8.5 months) in metastatic UV-related melanomas [13]. However, this has not been confirmed by more recent larger studies [10,13,15]. As the most frequently mutated gene in cutaneous melanoma, BRAF has become a popular target for melanoma treatment. BRAF inhibitors (BRAFi) are currently approved for BRAF V600E and V600K mutant melanoma and successfully improved the progression-free survival and overall survival of patients with these mutations [81,82]. However, the clinical benefits of BRAFi were shown to be short-lived due to drug resistance. The median progression-free survival was 5.1 and 6.8 months for dabrafenib and vemurafenib, respectively [83,84]. MAPK pathway activation by bypassing BRAF is the most common mechanism of BRAFi resistance. Hence, by combining BRAF and MEK (mitogen-activated protein kinase kinase) inhibitors, the survival benefits of these targeted therapies have been prolonged [81,82,85]. Although both BRAF V600E and V600K mutant melanoma benefit from BRAFi, melanomas with BRAF V600K mutations have shorter progression-free survival under BRAF inhibitors but superior response to immune checkpoint inhibitors compared to V600E mutants [10,14,15,18,86]. The progression-free survival for BRAF V600E and V600K mutant melanomas was 6.9 and 5.9 months for vemurafenib and 6.3 and 4.5 months in the Phase III trial of dabrafenib [87,88]. Among the different melanoma subtypes, BRAFi benefits cutaneous melanomas the most, as they have the highest frequency of BRAF V600 mutations. Mucosal melanoma frequently harbors atypical BRAF mutations other than the p.V600 hot spot (such as p.G469A, p.L597Q, p.N581S, p.T599I, and p.G596R), rendering BRAF inhibitors that revolutionized melanoma treatment not applicable in this population [28].
NRAS induces cell-cycle dysregulation and cell proliferation by activating downstream signaling of the MAPK pathway [89]. MEK inhibitors has been used to treat advanced NRAS-mutant melanoma. However, the results were short-lived, and only resulted in improvement in median progression-free survival for approximately 3 months [90][91][92]. Strategies directly targeting NRAS has been attempted by inhibiting farnesyltransferase, a key enzyme in RAS maturation, but have not been successfully applied in a clinical setting due to alternative prenylation via geranylgeranyl transferase type 1 [93][94][95][96].

KIT as a Potential Treatment Target
As KIT mutations are one of the most frequently mutated genes in melanomas, it is a potential target for advanced melanoma treatment. Although currently there are no approved KIT inhibitors for melanoma, KIT inhibitor imatinib has shown to provide a durable response to a subset of patients with KIT mutant metastatic melanoma with a complete response lasting for more than 1.5 years in a subset of patients [97,98]. Studies on the effeteness of other KIT inhibitors including nilotinib and ripretinib are still ongoing [99,100]. KIT mutation is relatively rare in uveal melanoma [36]. Nonetheless, most primary uveal melanomas express KIT protein on tumor immunohistochemical analysis, making KIT a potential target for treatment. However, the results of KIT inhibition by imatinib on UM are disappointing, as no patient showed a response to imatinib in early studies [101,102].

Microphthalmia-Associated Transcription Factor
Microphthalmia-associated transcription factor (MITF) is an important regulator in melanoma development and progression. MITF mutation is frequently observed in melanoma [10]. The role of MITF in melanoma is paradoxical as it is an oncogene but also has properties in suppressing melanoma invasion and metastasis [103,104]. Compared to melanomas with high levels of MITF expression (MITF-high), MITF-low melanomas are shown to be more invasive [104]. Moreover, melanoma with low MITF expression is associated with resistance to BRAF inhibition therapy due to reactivation of ERK in the MAPK pathway [104]. The expression of MITF is negatively correlated with the expression of AXL, a receptor tyrosine kinase. MITF-low/AXL-high melanomas, characterized by low MITF expression and high AXL markers, are associated with intrinsic resistance to RAF/MEK inhibition therapy. Although melanoma can be classified into MITF-low/AXLhigh or MITF-high/AXL-low melanoma at the tumor level, subclones with AXL-high and MITF-high expression co-exist in melanoma at the single cell level [105]. The application of BRAFi selectively selects in favor of AXL-high subclones and contributes to subsequent drug resistance [105].

TERT Promoter Aberrations
TERT promoter mutations are associated with poorer survival and higher metastatic rate in both UV-related and acral melanomas [47,106,107]. Due to the high prevalence of TERT promoter aberrations in various melanoma subtypes, it is a potential target for melanoma treatment. As TERT alterations are common in BRAF V600 mutant melanomas and plays an important role in controlling the apoptosis of cancer cells, combined therapies targeting both BRAF and TERT may potentially improvement the survival of these patients [108]. Similarly, a high frequency of TERT promoter aberrations is also found in NRAS mutant melanomas, making TERT activity inhibition a potential therapeutic target in this population [109]. TERT mutation status has also been used as a potential biomarker indicating favorable response to anti-cytotoxic T lymphocyte-associated antigen 4 (anti-CTLA4) treatment [110].

Cell-Cycle Aberrations
Cell-cycle aberrations are an independent prognostic factor for melanoma-specific survival indicating worse prognosis [15]. As cell-cycle aberrations frequently occur in melanoma, its inhibitors are possible treatment opportunities for these patients. Recently, palbociclib, an FDA-approved CDK4/6 inhibitor for treating metastatic breast cancer, has been demonstrated to show antitumor effects by inhibiting tumor growth in acral melanoma and mucosal melanoma with CDK4 amplifications in patient-derived xenograft models [27,52,111]. However, palbociclib only showed a modest effect (20% patients had tumor shrinkage at 8 weeks) as monotherapy in advanced acral melanoma with CDK4 pathway gene aberrations in a phase II clinical study [112]. Further clinical studies are required to evaluate its effectiveness on melanoma patients. Other CDK4/6 inhibitors include ribociclib and abemaciclib. Similarly, these agents are currently applied to a limited proportion of melanoma patients and further clinical studies are required to evaluate their effectiveness [113]. Cell-cycle inhibitors combined with targeted therapies against the MAPK pathway have been shown to demonstrate a synergic effect in treating melanoma [111,114,115].

Immune Checkpoint Inhibitors
Immune checkpoint inhibitors (ICIs) targeting programmed cell death protein 1 (PD-1) and cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) have become the standard care of advanced melanoma in the past 10 years and revolutionized the treatment of melanoma by significantly increasing the overall survival of melanoma patients [116][117][118][119][120]. Not all melanomas respond to ICIs. Currently, only approximately one-third of melanomas respond to ICI treatment [121][122][123]. Within different melanoma subtypes, the response rate to ICIs is lower in acral melanoma compared to cutaneous melanomas. Genetic markers indicating favorable ICI responders in melanoma include high tumor mutational burden [98,124], BRAF wild type [98], and increased gene expression of inflammation and T-cell population [10,98,125]. Tumor programmed death-1 ligand 1 (PD-L1) expression is also an important factor influencing therapeutic response to anti-PD-1 immunotherapy, with PD-L1-positive melanoma demonstrating a higher response rate to anti-PD-1 therapy compared to their PD-L1-negative counterpart [126]. Overall, 35% of melanoma express PD-L1, with chronic sun-damaged melanoma showing the highest percentage of PD-L1-positive tumors (62%), followed by mucosal melanoma (44%), acral melanoma (31%), and uveal melanoma (10%) [127]. PD-L1 expression is independent of melanoma driver mutation status (BRAF/RAS/NF1) [128]. A more accurate prediction for ICI therapy response may be achieved by combining the above factors.

Conclusions
The frequencies of key driver gene and pathway alterations differ among different melanoma subtypes, reflecting their heterogenic background in pathogenesis and prognosis. In this era of personal medicine, an in-depth understanding of the genetics of melanoma is crucial in developing personalized treatment and clinical decision-making for targeted therapies of specific genes. Currently, BRAF-targeted therapy has been successful in significantly improving overall survival in BRAF-mutant melanomas and c-kit inhibitors have been applied in melanomas with the corresponding mutation. Other potential targeted therapies are under development as more significantly mutated genes and pathways have been identified. More recently, ICI therapy has become the standard treatment for advanced melanoma and has significantly improved the survival of these patients. Since only a proportion of patients respond to this revolutionizing therapy, the identification of potential genetic markers indicating favorable responses has become increasingly important. A deepened understanding of the genetic background of melanoma allows us to utilize the best combination of treatments according to the tumor genetic background and predict treatment response more accurately. However, many recently developed targeted therapies are applied only on a limited number of melanoma patients and larger clinical studies are required to verify their effectiveness in this population. It is also worth mentioning that whole genome sequencing is only applied in a limited number of studies and important regulatory noncoding alterations and single-nucleotide variants may be overlooked under whole exome sequencing or low-pass whole genome sequencing.