CTCs are shed from solid tumors into the circulation, and are considered a vital step in the establishment of distant metastases (16, 20). They have a short half-life due to physical and oxidative stress, anoikis, absence of growth factors and cytokines, and cell loss from implantation into capillary beds (16, 21). Detection of CTCs remains challenging despite advances in cell isolation, enrichment, and analysis methods, with detection rates in advanced melanoma patients ranging from 28 to 87% (16, 22). This is due to several factors, including low concentrations in circulation, marker heterogeneity, and lack of standardized method for isolation (20, 22–26).

Patients with early-stage melanoma commonly have only single positive melanoma markers on CTCs, whilst patients with advanced stages not only had higher CTC counts, but also a greater variety of markers (25). Reverse transcription polymerase chain reaction (RT-PCR) is the most commonly used method for detecting CTC markers (sensitivity 1–10 CTCs/mL whole blood, specificity 99.9%); most detection techniques rely on one to two known markers (25, 27–29). Aya-Bonilla et al. compared isolation and detection methods, finding individual detection rates of 28% for multi-marker immunostaining, 42% for a 5-gene panel via RT-PCR, and 53% for a 19-gene droplet digital PCR (ddPCR) assay; however, by combining all three methods, the detection rate was bolstered to 72% (22). Commonly-used CTC markers include tyrosinase, Melan-A/MART-1, MAGE3, MCAM, gp-100, MITF, and GalNac-T, which have specificities ranging from 85 to 100% but sensitivities from 6 to 95% (Table 1) (24, 27). Other isolation methods include microchips, antibody-coated immunomagnetic beads, and combining RT-PCR with selective fluorescent imaging (26). Therefore, it may be worthwhile adopting a multi-marker approach to improve clinical utility of any one biomarker.

CTCs have also been detected in early-stage melanoma including melanoma in-situ (MIS) (29), with Voit et al. reporting a positive association between detectable CTC counts and progression-free survival (PFS) in stage I-III patients (33). Lucci et al. noted that the detection of one or more melanoma CTCs (per 7.5 mL peripheral blood) in stage III patients was significantly associated with disease relapse within 6 months (24). Fluctuations in CTC counts have been associated with treatment response, disease progression, and OS (16, 29).

New technological advances to enrich and isolate CTCs, and standardization in isolation techniques will likely improve the clinical utility of CTCs to define heterogeneity and occult invasiveness, assess tumor burden, direct treatment decisions, and monitor response to treatment (21, 29).

Circulating tumor DNA (ctDNA) are fragments of single or double-stranded DNA released at low concentrations by tumor cells into the circulation, likely by cell apoptosis and necrosis, although the precise mechanism is uncertain (3, 16–18, 34). ctDNA can be measured in plasma or serum, but it has a short half-life as a result of rapid clearance though the kidneys, liver and spleen (16, 34). This timeframe can be extended if with specific collection tubes which preserve the ctDNA at room temperature up to 7 days (35). ctDNA levels are associated with tumor burden, location, and vascularity, and the rate of cellular turnover (16–18). As patients with early-stage or MIS have low tumor burden and vascularity, ctDNA levels are often not detectable in these groups using conventional technology (3, 16, 17). There are also technical issues in detecting and analyzing ctDNA. Firstly, ctDNA only constitutes a small fraction of the total background cell-free DNA (cfDNA), which itself is only present in small quantities in the circulation (16, 34). Ongoing advances in technology, such as the use of ddPCR, are sensitive at detecting low-level ctDNA, and are readily reproducible (36). The proportion of cfDNA also increases with leukocyte lysis, which can occur during blood collection and clotting, reducing the ability to detect ctDNA (16, 37). Consequently, plasma is the preferred source of ctDNA (16). There is however a lack of consistency in blood collection processes, including types of tubes used, time to plasma separation and processing, and storage and transport conditions (16, 37). These differences make analysis and comparison across different study groups troublesome; thus, cut-off values have not been established, reducing the clinical utility of ctDNA (3, 17). Finally, to distinguish ctDNA from cfDNA, specific somatic mutations in the melanoma tumor must first be identified (3). Tumor heterogeneity thereby continues to hinder detection rates, as ~25% of melanoma patients do not have identifiable tumor-derived ctDNA mutations present in the circulation (3, 36).

Despite the challenges, monitoring ctDNA levels in metastatic melanoma patients is non-invasive, with potential clinical utility demonstrated in determining pre-treatment prognosis, monitoring of treatment response, assessment of clonal evolution and minimal residual disease (MRD), and tracking disease progression (16–18, 34). Monitoring MRD is particularly important to confirm complete responders to systemic therapies due to detection limits of radiological imaging (38).

Low or undetectable pre-treatment levels of ctDNA in patients with metastatic melanoma was associated with lower tumor burden and increased PFS and OS, independent of LDH and tumor stage (16–18, 36). However, if detectable levels of ctDNA following surgical excision was detected, this was associated with a poor PFS and OS (34). In patients who had detectable baseline ctDNA, there was a correlation with tumor burden and LDH, suggesting that ctDNA may reveal tumor burden in the absence of imaging (13, 18, 36). In addition, elevated baseline ctDNA levels correlate with reduced overall response rate, and shortened PFS and OS in melanoma patients treated with targeted or immune therapies (16, 17, 34, 39).

There are however limitations to using ctDNA as a predictive biomarker for patients undertaking second-line treatment immunotherapy after failed initial targeted treatment, as there was no association between low ctDNA levels and PFS (39). Additionally, in patients with brain metastases, the levels of ctDNA may be artificially low as there is less ctDNA released into the circulation (40); accurate detection relies on more invasive collection of cerebral spinal fluid (41).

MicroRNA (miRNA) are short, non-coding RNA fragments which regulate gene transcription and expression, thereby affecting cell differentiation, proliferation, migration, and apoptosis (3, 16, 42, 43). They are secreted into the circulation, usually in lipid-bound vesicles or in a complex bound by protective proteins and are highly stable even under harsh conditions such as extremes of pH and temperature, multiple freeze-thaw cycles, and long-term storage (3, 14, 16).

Detection of miRNA relies on sequence-matching miRNA molecules (probes, oligos, etc.), and selective and sensitive amplification, but standard techniques such as real-time quantitative RT-PCR can be used. Often low concentrations of miRNA in blood and different expression profiles between serum and plasma samples, and across individuals, can make detection challenging (3, 16), and differences in sample collection and preparation, RNA extraction, normalization methods and storage, limit comparisons between studies (3). Many miRNAs have however been linked to melanoma and metastasis (3).

For example, miR-221 has a role in maintaining cell cycle regulation and proliferation, and so its dysregulation promotes melanoma progression (3, 44, 45). Serum levels of miR-221 were generally undetectable in controls and patients with MIS; however, they were significantly elevated in most patients with stage I melanoma, and associated with increasing clinical stage and tumor thickness (46). Other miRNA regulating cell cycle with altered expression in melanoma include miR-9, miR-150, miR-155, miR-205, and miR-222 (3, 45).

miR-199a-5p promotes melanoma metastasis and angiogenesis, and has been shown to be up-regulated 10-fold in patients with stage III melanoma compared to stage Ia disease (47). Even patients with stage Ib to IIb disease had a 6-fold increase in expression compared to healthy and stage Ia groups (47, 48). Up-regulation of miRNA-199a-5p in conjunction with up-regulated miR-877-3p, 1228-3p, 3613-5p and down-regulated miR-182-5p was associated with higher melanoma stage (stage III or higher) at the time of primary melanoma excision (47). Patients who did not have down-regulated miRNA-182-5p were usually within stages Ib-IIb (47). Additionally, dysregulation of miR-199a-5p, miR-150, miR-15b, miR-33a, and miR-424 increased the risk of melanoma recurrence, with increasing levels associated with tumor burden (16).

These studies are a few examples of dysregulated circulating miRNAs associated with differing melanoma stages relative to healthy controls. When used in isolation, individual miRNAs may not yet be able to be used clinically due to inconsistent expression across patients. The use of a collection or panel of miRNA may result in a more sensitive and specific biomarker (3). Margue et al. (49) showed that miR-301-3p, miR-200c-3p, miR-126-5p, miR-374a-5p, and miR-211-5p, when used in a combination, could differentiate patients with stage I and II melanoma from healthy controls. van Laar et al. also used a combination of 38 miRNAs (MEL38), to differentiate healthy controls from patients who had stage I-IV melanoma (50). We too have identified a panel of seven “melanoma-related” miRNA biomarkers that have prognostic value as a liquid biopsy measured in serum (51). In treatment naive stage IV patients, members of this MELmiR-7 panel were able to detect an increase in tumor burden in 100% of cases and the ability of the panel to define OS proved to be superior to LDH and S100B levels (delta log-likelihood = 11, p < 0.001) (51). Levels of circulating miRNA may fluctuate with systemic treatments used to treat late-stage melanoma patients so the use of miRNAs in this context requires further validation.

The use of miRNA as a melanoma biomarker is promising, and whilst several studies focused on an individual miRNA, further research into groups of miRNAs may provide a more sensitive and specific tool to differentiate early-stage melanoma from healthy controls and provide better pre-treatment prognosis.

Extracellular vesicles are membrane-bound particles secreted by all cells, present in all body fluids, and transport cargo (DNA, RNA, miRNA, proteins, lipids, and metabolites) that are representative of the original cell (7, 16, 52, 53). The EVs secreted by tumor cells can deliver tumor-specific molecules to other body sites that may assist tumor survival and proliferation (16). Exosomes, which are nano-sized subset of EVs, released by melanoma cells may also stimulate the migration of endothelial cells, induce angiogenesis to promote distal metastasis, contribute to drug resistance, and may have immunosuppressive actions (54, 55).

Exosomes can be isolated based on physical characteristics such as size, morphology, concentration, cell surface markers or cargo (16, 52, 56, 57). Detection of tumor-derived exosomes with current technologies is difficult, due to their size, low quantities of cargo within each exosome, heterogeneity, and abundance of non-target EVs (7).

In melanoma patients relative to healthy controls, 4–20-fold higher levels of plasma tumor-derived exosomes were detected which persisted despite surgical resection of the primary lesion with no clinical evidence of residual disease (58–60). Exosomes are known to be involved in establishing pre-metastatic niches in cancers including melanoma (61). This persistence of the exosomes, post excision of primary lesions, may be a reason why in some patients, recurrence of metastatic disease occurs many years after initial primary diagnosis. Recently, Wang et al. compared the expression of EV surface markers MCSP, MCAM, CD61, and CD63 in healthy controls and patients who had a history of early-stage melanoma (including MIS) (7). Surface enhanced raman spectroscopy (SERS; an ultra-sensitive detection method) signatures and mapping images showed a distinct separation between the two groups, with higher levels of SERS signals observed in the melanoma patients (7).