Recent studies have reported significant alterations in the miRNA expression profile in the serum and plasma of melanoma patients compared to healthy controls, proposing circulating miRNAs as promising diagnostic melanoma biomarkers. In addition to its diagnostic utility, more recent evidence suggests that miRNA profiling from serum and plasma may be a useful tool for prognosis, and prediction of metastatic outcome and therapeutic response. Prognostic biomarkers should be able to differentiate between stage I/II and stage III/IV melanoma patients indicating which patients are likely to progress to metastatic disease, correlate with poor prognosis and/or stratify patients into groups at high and low risk of recurrence following surgical resection of a non-metastatic melanoma. Lack of accurate prognostic biomarkers could lead to unnecessary lymph node biopsies and lymph node resections which are invasive to the patient.
5.1. Circulating MicroRNAs as Diagnostic Biomarkers in Melanoma
Several efforts have been made to identify potential diagnostic biomarkers that can distinguish between melanoma patients and healthy control individuals (Table 1
). Leidinger et al. [60
] were one of the first research groups to use high throughput screening techniques to identify diagnostic circulating miRNA biomarkers. This work involved screening ~866 human miRNAs in blood cells using a microarray-based approach, and subsequent validation of differentially expressed miRNAs using qRT-PCR. A 16-miRNA diagnostic signature including miR-186, let-7d, miR-18a, miR-145, miR-99a, miR-664, miR-501-5p, miR-378, miR-29c, miR-1280, miR-365, miR-1249, miR-328, miR-422a, miR-30d, and miR-17, that could distinguish between melanoma patients and healthy control individuals with high sensitivity and specify was identified. It is important to note that the majority of melanoma samples used for this study derived from patients with metastatic (stages III and IV) melanoma. It would be interesting to investigate this 16-miRNA signature as a diagnostic biomarker for non-metastatic (stages I and II) melanoma, against which the efficacy of current treatment options and therefore chances of survival are much higher.
More recently, Van Laar and colleagues [61
] performed microarray profiling on plasma samples from 32 patients with stages I–IV melanoma and 16 healthy control individuals. They identified MEL38, a 38-miRNA signature (Table S3
) that is able to distinguish between melanoma patients (stages I–IV) and healthy controls with a high degree of sensitivity and specificity. Instead of using an independent cohort of patients to validate the significance of the circulating miRNAs identified from the discovery series, a classification algorithm was trained on their discovery data and independently validated on miRNA expression data sets representing 473 unique melanoma patients, normal control individuals or cell-line models, downloaded from gene expression data repositories, including National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) and European Bioinformatics Institute (EBI) ArrayExpress. A major disadvantage to this approach is the likely presence of technical differences in experimental design between the studies from which these data sets were generated. As discussed in Section 6
, small changes to experimental design from sample collection and processing to data interpretation and analysis can have a significant influence on the levels of miRNA detected from plasma and serum. None of the miRNAs in MEL38 were identified as differentially expressed in the blood cell screening by Leidinger et al. [60
], which is likely due to the starting material or improvements in detection technology. Nearly half of miRNAs in MEL38 displayed between a 2- and 2.5-fold change which, because the amount of RNA being input into each assay is unknown (discussed in Section 6
), could indeed result from normal human variation as opposed to disease status. This highlights the benefit of using a collection of miRNAs as a signature as if used individually, many of the miRNAs in MEL38 cannot be considered as independent biomarkers, however when used in combination appear to provide a sensitive and specific tool.
miR-424-5p and miR-301a-3p from the MEL38 series displayed a 6.28- fold and 4.86- fold) increase, respectively, in melanoma patients compared to healthy controls, and can therefore be considered as potential independent diagnostic biomarkers or together form a signature with high diagnostic power. In contrast with these findings, Margue et al. [62
] demonstrated that miR-301a-3p is down-regulated in stage III/IV melanoma patients relative to healthy controls, however this study used serum as opposed to plasma, and these two biofluids have been shown to differ in their miRNA expression profile, perhaps due to different sample processing and RNA extraction methods inadvertently enriching for particular miRNAs (discussed in Section 6
). In fact, several miRNAs in the MEL38 discovery series were found differentially expressed in previously published melanoma circulating miRNA signatures, including miR-424-5p [63
], miR-205-5p, miR-301a-3p, and miR-27a-3p [62
], however only miR-424-5p showed the same direction of change between studies indicating one of many difficulties in the field.
Several efforts have been made towards the identification of circulating miRNA biomarkers that can distinguish between stage III/IV melanoma patients and healthy controls. Margue et al. [62
] used “whole” miRNome (1066 miRNAs) and custom quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) arrays to identify and validate potential serum diagnostic miRNA biomarkers that can distinguish between melanoma patients (stages I–IV) and healthy control individuals. MiR-204-5p, miR-182-5p, miR-301a-3p, miR-200c-3p, miR-28-5p, miR-27a-3p, miR-197-3p, and miR-374a-5p were identified as potential independent diagnostic biomarkers down-regulated in the serum of melanoma patients relative to healthy controls, while miR-193b-3p, miR-720, miR-205-5p, miR-126-5p, miR-211-5p, miR-720, miR-206, miR-550a-3p, miR-627-5p, and miR-629-5p were shown to be up-regulated. MiR-301-3p, miR-200c-3p, miR-126-5p, miR-374a-5p, and miR-211-5p could distinguish between non-metastatic melanoma (stage I/II) patients and healthy controls, whilst the remaining miRNAs could only distinguish between metastatic (stage III/IV) melanoma patients and healthy controls. In agreement with these findings, miR-200c-3p was also demonstrated by Fogli and colleagues [64
] to be down-regulated in the plasma of stage III/IV melanoma patients compared to stage I/II melanoma patients and healthy controls, suggesting its potential to be used as both a diagnostic and prognostic biomarker. It has been demonstrated that miR-200c-3p levels are significantly reduced in metastatic melanoma tissue, suggesting that lower serum/plasma levels of this miRNA may be caused by decreased expression and release from the tumor site [65
]. Greenberg et al. [66
] reported miR-29c-5p and miR-324-3p to be lower in the serum of metastatic melanoma patients (stage IV) compared to healthy control individuals. In addition to being indicative of metastatic melanoma; these miRNAs could distinguish between melanoma and both colon and renal cancer. It is important to note that the discovery cohort consisted of only males, and expression analysis studies have revealed potential sex-specific dysregulated miRNA expression [67
]. Diagnostic biomarkers that can distinguish between stage III/IV melanoma patients and healthy control individuals, while providing a less invasive means of diagnosing metastatic melanoma, are unlikely to translate into adequate therapeutic intervention and survival benefit due to the limited efficacy of current treatment options against certain subtypes of stage III/IV melanoma (5-year survival rate of 10–50%), therefore studies focusing on diagnostic biomarkers which aim to identify patients with stage I/II melanoma (5-year survival rate of 80–100%), could improve early intervention.
Indeed, a number of studies have identified potential circulating miRNA biomarkers that can distinguish between stage I/II melanoma patients and healthy controls. It has been demonstrated that miR-15b-5p, miR-149-3p, and miR-150-5p are upregulated, and miR-193a-3p and miR-524-5p are downregulated in the plasma of stage I/II, stage III, and stage IV melanoma patients compared to healthy control individuals [64
]. These miRNAs have the potential to be used as independent diagnostic biomarkers in the earlier stages of melanoma, however, diagnostic sensitivity and specificity was greatly improved when a signature of miR-149-3p, miR-150-5p, and miR-193a-3p was considered. Both the Fleming et al. [68
] and Friedman et al. [63
] studies (discussed in Section 5.2
) support these findings by implicating increased serum miR-150 in a prognostic model of melanoma recurrence. The Fleming et al. study [68
] also agreed with the increasing levels of serum miR-15b, however the Friedman et al. [63
] study disagreed, implicating decreased serum miR-15b in a prognostic model of melanoma recurrence. MELmiR-7 is a serum based 7-miRNA (miR-16, miR-211-5p, miR-4487, miR-4706, miR-4731, miR-509-3p, and miR-509-5p) signature that was shown to be able to discriminate between stage I/II, stage III, and stage IV melanoma patients and non-melanoma controls [69
]. MiR-4487, miR-4706, miR-4731, miR-509-3p, and miR-509-5p showed lower expression in melanoma patients whilst miR-16 and miR-211-5p had higher expression in melanoma patients. The fold changes observed in this study are more suitable for clear and consistent biomarkers (i.e., miR-509-3p displayed a ~6 ΔCq/~64-fold decrease in stage IV melanoma patients relative to healthy controls).
Currently only one study has specifically examined the exosome associated miRNA pool between melanoma patients and controls. Alegre and co-workers [70
] demonstrated that miR-125b levels were significantly lower in the serum exosomes of patients with advanced melanoma compared to serum exosomes of disease-free patients and healthy controls, however no significant difference was observed in miRNAs from whole serum of melanoma patients relative to healthy controls. Conversely, Achberger et al. [71
] demonstrated that plasma levels of miR-125b are upregulated in uveal melanoma patients relative to healthy controls and that miR-125b is also present at higher levels at metastasis compared to primary diagnosis. Indeed, plasma and serum have been shown to have different miRNA profiles, perhaps due to particular miRNAs levels being sensitive to extraction conditions during serum/plasma preparation or exosome enrichment. Achberger et al. [71
] also demonstrated that miR-20a, miR-146a, miR-155, miR-181a, and miR-223 are upregulated in the plasma of uveal melanoma patients relative to healthy control individuals and all, except for miR-181a, are present at higher levels at metastasis compared to primary diagnosis. Whilst promising, only six uveal melanoma patients were used for this study therefore there is clear need to perform a validation study with increased sample size before ratifying the use of these circulating miRNAs as diagnostic biomarkers for melanoma.
5.2. Circulating MicroRNAs as Prognostic Biomarkers in Melanoma
While many studies have focused on identifying circulating miRNA biomarkers that can distinguish between melanoma patients and healthy control individuals, relatively few studies have used staged samples to identify prognostic biomarkers that can distinguish between stage I/II melanoma patients and stage III/IV melanoma patients. There is limited efficacy of current treatment options against certain subtypes of stage III and IV melanoma, resulting in 5-year survival rates of ~50% and 10–25%, respectively, therefore prognostic biomarkers that can identify patients at high risk of recurrence and progression to metastatic disease will allow for early intervention before melanoma cells have metastasized, as there is high efficacy of current treatment options against non-metastatic melanoma subtypes. This section summarizes the potential prognostic circulating miRNA biomarkers identified to date (Table 2
Margue et al. [62
] (Section 5.1
) showed that an increase in levels of serum miR-193b-3p and miR-720 could not only distinguish between melanoma patients and healthy control individuals, but also between non-metastatic and metastatic melanoma patient groups, highlighting the prognostic potential of these two miRNAs. In addition to diagnostic signature, MEL38, Van Laar et al. [61
] identified MEL18, an 18-miRNA serum signature that is able to distinguish between non-metastatic (stage I/II) and metastatic (stage III/IV) melanoma patients (see Table S3
). Guo et al. [72
] demonstrated that miR-16 is downregulated by in the serum of stage I/II melanoma patients relative to healthy controls, and is also downregulated in stage III/IV melanoma patients relative to stage I/II melanoma patients demonstrating the potential for this miRNA to be used as both a diagnostic and prognostic biomarker. MiR-16 levels were shown to negatively correlate with ulceration and tumor thickness and positively correlate with patient survival, implying that serum miR-16 levels could reflect progression status. It is important to note that both the Van Laar et al. [61
] and the Guo et al. [72
] studies failed to use a suitable endogenous normalization method. Further, circulating miR-16 levels are influenced by stress, therefore it would not appear to be suitable to use as a biomarker [73
]. In contrast with the findings by Guo et al. [72
], the Stark et al. study [69
] (Section 5.1
) demonstrated an increase (rather than a decrease) in miR-16 expression in the serum of stage I, II, III, and IV melanoma patients relative to healthy controls. There are a number of variations in the methodology used between these studies that may account for these conflicting results (Tables S1 and S2
). For example, different miRNA isolation kits were used and it has been demonstrated that the miRNA extraction kit can influence the yield and distribution of miRNA detected from serum and plasma samples (Section 6
). Furthermore, the Stark et al. [69
] study used TaqMan microRNA qRT-PCR assays to measure relative expression while the Guo et al. study [72
] used SYBR-Green-based qRT-PCR. Several studies have reported lower sensitivity, specificity and reproducibility of SYBR-Green based qRT-PCR systems relative to TaqMan qRT-PCR systems (Section 6
) and that the TaqMan microRNA qRT-PCR system can minimize extraction kit-dependent variation within the same set of samples [74
]. Different qRT-PCR chemistries can lead to conflicting findings between studies, therefore consideration should be given to the qRT-PCR chemistry used before drawing firm conclusions about the roles of specific miRNA as circulating biomarkers in melanoma.
Several efforts have been made towards the identification of circulating miRNA biomarkers that can stratify patients into groups at high and low risk of recurrence following surgical resection. Fleming et al. [68
] demonstrated that serum levels of miR-150, miR-15b, miR-425, and miR-30d increase with increasing melanoma stages, and that this miRNA signature in combination with stage can effectively distinguish recurrent cases from non-recurrent cases and stratify patients into groups at high and low risk of recurrence with high sensitivity. Further, miR-15b levels were demonstrated to increase over time from pre- to post-recurrence. Friedman et al. [63
] identified a five miRNA signature that includes miR-150-5p and miR-15b-5p in combination with miR-199a-5p, miR-33a-5p, and miR-425-5p that is also able stratify patients into groups at high and low risk of recurrence with high sensitivity and specificity. In agreement with the Fleming et al. [68
] study, miR-150-5p was upregulated and in the serum of melanoma patients at high risk of recurrence, while in contrast, miR-15b-5p was down-regulated in the serum of patients at high risk of recurrence. Results obtained for miR-15b-5p by Fleming et al. [68
] are perhaps the most reliable because they used two endogenous normalizers selected using the NormFinder algorithm, while Friedman et al. [63
] used median normalization. The extraction kit and qRT-PCR chemistry were the same for both studies.
Tian et al. [75
] demonstrated that levels of miR-206 were significantly lower in the serum of melanoma patients relative to healthy controls and that decreased serum miR-206 levels are associated with aggressive disease progression and poor prognosis, suggesting its potential to be used as a prognostic biomarker. miR-221 and miR-21 have been shown to be present at significantly high levels in the serum and plasma, respectively, of melanoma patients relative to healthy controls [76
]. High expression of serum miR-21 matches high miR-21 expression in metastatic melanoma tissues and an increase in miR-21 from controls through to advanced melanoma stages suggests this miRNA has possible prognostic value. Further, when preoperative and postoperative cases were compared, there was a drop in plasma miR-21 levels in post-operative samples [76
]. While no significant difference in miR-221 expression was observed among patients with stages I, II, III, and IV melanoma, miR-221 levels were shown to correlate with tumor thickness [77
] which is a crucial clinicopathologic factor determining melanoma progression, implying that serum miR-221 levels could reflect progression status of melanoma patients. By assigning patients into low expression (less than the median expression level of 2.95) and high expression (≥2.95) groups, Li et al. [78
] demonstrated that miR-221 expression correlates with a number of clinicopathological characteristics including tumor thickness, poor differentiation, higher T classification, higher N classification, metastasis, and advanced clinical stage. Furthermore, patients with high serum miR-221 levels had a significantly lower 5-year rate and recurrence free survival (RFS) rate than those with low serum miR-221 level. In support of these findings, Margue et al. [62
] demonstrated that miR-221-5p was one of four miRNA upregulated in the serum of stage III/IV melanoma patients relative to healthy controls. miR-221 has long been recognized as marker miRNA for the melanocytic lineage [79
] suggesting that this miRNA may be specifically secreted from melanoma cells. Furthermore, miR-221 is an established oncomiR that has been demonstrated to target p27 RNA in melanoma cell lines. miR-221-mediated down-regulation of p27 has been demonstrated to drive cell proliferation and contribute to cancer development and progression in a number of contexts (Section 3
). Taken together these findings would suggest that miR-221 is a potential diagnostic and prognostic circulating miRNA biomarker for melanoma. Whilst promising, all three studies investigating circulating miR-221 as a potential prognostic biomarker for melanoma failed to specify a suitable endogenous normalization method (Table 2
). Further studies, using appropriate endogenous normalizers or other suitable methods (such as the global mean normalization) [81
], should be carried out before ratifying the use of increased circulating miR-221 levels as a biomarker for melanoma diagnosis and prognosis.
It has been demonstrated that the expression of mir-210, a miRNA known to play important roles in the tumor hypoxic state, to be significantly higher in the plasma of stage III/IV melanoma patient’s relative to healthy controls [82
]. In the training cohort, this study demonstrated that miR-210 levels were higher in the serum of stage IV melanoma patients relative to stage III melanoma patients, however they were unable to verify this difference in the validation study. Levels of miR-210 in plasma significantly increased prior to recurrence, and correlated with poor prognosis, suggesting that miR-210 expression could provide early identification of melanoma recurrence. This study confirmed miR-210 expression level was significantly higher in both lymph node metastasis and distant organ metastasis compared to primary tumors demonstrating the development of the hypoxic miR-210 relevance to melanoma metastasis occurrence. Unlike other studies on circulating miRNA biomarkers in melanoma and other cancers, reverse transcription was carried out directly against plasma using a miRNA-specific reverse-transcriptase (RT) primer and Moloney murine leukemia virus reverse transcriptase. The concentration of RNA in serum and plasma is extremely low compared to that of RNA purified and eluted in a relatively small volume of RNase free water, which would lead to high quantitation cycle (Cq) values and broad variation within patient groups due to more stochastic qPCR amplification. This can result in the detection of false positive differences in miRNA expression between patient groups.