Introduction

From the time of the early pioneers of dermatopathology such as Simon and Unna, until the second half of the twentieth century, the pathologic diagnosis of melanocytic lesions was accomplished largely by traditional means: histological examination being the gold standard, ideally in conjunction with adequate clinical history and correlation. Although the microscope used by Unna in those early days would be considered primitive by current standards, the basic technology remains the same today. Light microscopic evaluation by a trained pathologist can provide an accurate diagnosis for the great majority of melanocytic lesions. There is, however, a small subset of melanocytic lesions that eludes appropriate classification by conventional light microscopy alone, preventing accurate prediction of clinical behavior and recommendations for appropriate treatment. Furthermore, inter-observer variability in the classification of melanocytic lesions has been well documented [1,2,3] and uncertainties can lead to both over and undertreatment. Histologically ambiguous melanocytic lesions are often treated more aggressively in order to avoid “missing” a bad actor, as the realm of melanocytic lesions has been one of the well-recognized legal peril [4]. In recent years, two major ancillary tests, comparative genomic hybridization (CGH) and fluorescence in situ hybridization (FISH), have been developed to help guide the diagnosis of ambiguous melanocytic proliferations.

In this overview, we discuss CGH and FISH, and their application to melanocytic lesions. We provide a brief survey of their technical aspects, appropriate use, and limitations.

Comparative genomic hybridzation

The idea that melanomas are characterized by multiple chromosomal copy number abnormalities (CNAs) was suggested a few decades ago by studies that reported chromosomal copy number differences between nevi and melanomas using conventional cytogenetics [5,6,7]. These initial discoveries formed the theoretical basis for further studies using CGH, and later FISH.

Background of the technique

The power of CGH lies in its ability to detect DNA copy number alterations across the entire genome. There are two major variations of this technique: classic and array-based.

Classic CGH

In classic CGH, tumor and reference DNA (from normal human tissue) are differentially labeled using fluorochromes, mixed in a 1:1 ratio, and hybridized onto a substrate of normal metaphase chromosomes. Following a washing step, the metaphase slides are scanned. The resulting wavelength is reflective of differences in relative proportion of tumor vs. normal DNA and can be used to detect genomic areas with gains or losses of DNA material [8,9,10]. The technique was subsequently improved to allow for analysis of archival paraffin-embedded samples [11]. CGH has a significant advantage over standard cytogenetics in that it does not require culturing the cells of interest and thus can be performed on archival formalin-fixed paraffin-embedded (FFPE) tissue. It also obviates any biological selection that may take place during the cell culturing used in traditional cytogenetics [12].

Array-based CGH and SNP arrays

The development of array-based CGH, where arrayed artificial genomic clones replace normal metaphase chromosomes as a substrate, led to increased resolution, reproducibility, and robustness of the assay, and has replaced classic CGH [13,14,15]. Each dot on the array contains genomic DNA from a specific locus and the resolution of the array is proportional with the number of dots. The assay can be performed on some platforms by co-hybridizing tumor and normal DNA onto the array, similar to classic CGH, or on other platforms by hybridizing only labeled tumor DNA. For the latter variation, after scanning, the copy number status at a certain locus is estimated by comparing the signal intensity of the tumor with that of a reference from a control series of non-tumoral tissue.

Recently, single-nucleotide polymorphism (SNP) arrays have been employed as alternatives to CGH. For these arrays, the probed loci are selected to flank known SNPs and each genomic locus is represented by two dots on the array corresponding to the two alleles. SNP arrays have the additional advantage of being able to detect allelic ratio and loss of heterozygosity (LOH), including copy-neutral LOH alterations that are missed by CGH arrays [16]. In addition, SNP arrays can also be used to identify selected point mutations. In recent years, protocols employing molecular inversion probes (MIPs) that use low amounts of tumor DNA from FFPE tissue have been developed. The probes have a footprint of only 40 bp, which allows evaluation of highly degraded DNA. In addition, as at the completion of the procedure the inverted MIPs are hybridized to the microarray, the signal-to-noise ratio is greatly improved compared with conventional SNP arrays in which the tumor DNA is directly hybridized [17, 18].

A typical SNP array plot consists of two tracks (Fig. 1). The upper track shows copy number status (log ratio on the vertical and chromosome locus on the horizontal). Gains and losses are reflected by deflections of the average line above or below 0, respectively (Fig. 1, red arrow). The lower track shows the B-allele frequency for each SNP on the array (B-allele frequency on the vertical and chromosome locus on the horizontal). In the normal state, this track consists of three lines, a middle heterozygous line and two outer lines that are homozygous for the A and B alleles. An LOH event is represented by a split in the middle heterozygous line. This is usually associated with corresponding losses or gains at that locus; however, occasionally, an LOH without numerical abnormalities can be encountered, representing a copy-neutral LOH, which is a form of acquired uniparental disomy (Fig. 1, black arrow).

Fig. 1: A typical SNP microarray plot.
figure 1

The upper track shows copy number status (log ratio on the vertical and chromosome locus on the horizontal). Gains and losses are reflected by deflections of the average yellow line above or below 0, respectively. Red arrow indicates a one copy number gain of chromosome 11p. The lower track shows the B-allele frequency for each SNP on the array (B-allele frequency on the vertical and chromosome locus on the horizontal). In the normal state, this track consists of three lines: a middle heterozygous line and two outer lines that are homozygous for the A and B alleles. Black arrow indicated a copy-neutral LOH of chromosome 14q (note that there are no copy number changes detected in the upper track for 14q).

Full size image

CGH in the evaluation of melanocytic lesions

In 1994, uveal melanomas were the first melanocytic tumors to be studied by CGH and were found to harbor frequent gains of chromosome arms 6p and 8q, and losses of chromosome 3, 6q, and 9p [12]. A much cited early study by Bastian et al. [19] indicated that loses of chromosomes 9 and 10 occurred early in melanoma progression, with gains of chromosome 7 occurring later. A later study found that among 16 primary and 12 metastatic melanomas, the mean number of genetic changes was 6.3 (range 1–14) in primary melanomas and 7.8 (range 1–16) in metastatic lesions. The number of genetic alterations was significantly higher in primary tumors that developed metastases within a year of surgery, in comparison with those that did not [20].

CGH as a diagnostic tool

The potential utility of CGH as a diagnostic tool was suggested most forthrightly by Bastian et al. [21] in a study from 2003, in which the authors found that 96.2% of melanomas displayed numerical aberrations involving segments of chromosomes, in comparison with 13.0% of benign nevi (out of a total of 132 melanomas and 54 benign nevi). This non-overlapping pattern of genomic abnormalities provided the impetus for developing diagnostic strategies based on tests evaluating CNAs such as CGH or SNP arrays (Fig. 2). A more recent study showed a specificity and sensitivity of 94.7% in separating unambiguous nevi from melanomas, using array-based CGH [22]. Array-based CGH was also used to compare copy number changes in melanomas from chronically sun-exposed vs. melanomas arising in sun-protected areas; it was found that melanomas on protected surfaces (acral and mucosal) had a significantly higher degree of chromosomal changes in comparison with melanomas on the skin with various degrees of exposure to the sun [23].

Fig. 2: Examples of SNP-array results in melanoma and nevus.
figure 2

a An example of nodular melanoma. b SNP-array plot of melanoma from a showing numerous gains and losses of segments of chromosomes. c An example of conventional nevus. d SNP-array plot of nevus from c showing no CNAs. e Aggregate frequency of CNAs in a series of 41 melanomas and 18 nevi from University of Michigan Dermatopathology database (frequency of gains (blue) and losses (red) on the vertical and chromosome locus on the horizontal). Top panel shows numerous gains and losses in melanomas. Bottom panel shows a low frequency of isolated abnormalities in nevi with isolated losses of 3p (red) and gains of 11p and 15q (blue).

Full size image

Several studies investigated the role of CGH as a potential ancillary tool for melanocytic tumors that traditionally pose diagnostic challenges using histology alone. These include the blue nevus-like group (cellular blue nevus, atypical cellular blue nevus, and melanoma arising in/resembling blue nevus) and spitzoid tumors (Spitz nevus, atypical Spitz tumor (AST), and spitzoid melanoma). Proliferative nodules in congenital nevi can also be problematic.

CGH in blue nevus-like melanocytic lesions

For the blue nevus-like group, an initial study of 10 histologically unequivocal cellular blue nevi and 1 deep penetrating nevus (by CGH) showed no chromosomal aberrations, in comparison with atypical cellular blue nevi, which showed abnormalities in 3 of 11 lesions, and overt melanomas arising in the blue nevus, which showed abnormalities in all 7 lesions studied [24]. Following this, other investigators have reported a non-overlapping pattern of chromosomal abnormalities between cellular blue nevi and melanoma [25,26,27,28,29]. The melanomas arising in cellular blue nevi tend to display multiple segmental copy number variations, similar to other melanomas. In contrast, cellular and atypical cellular blue nevi demonstrate no abnormalities or a small fraction may show one or two isolated CNAs [25]. An important finding was the association between deletion of 3p21 spanning the locus for BAP1 gene (associated with lack of BAP1 expression by immunohistochemistry) and poor prognosis in melanomas arising in blue nevi [25, 26].

CGH in spitzoid (and related) melanocytic neoplasms

Application of molecular techniques to spitzoid neoplasms is particularly relevant, as these lesions can be a source of diagnostic uncertainty among pathologists [30]. A study in the late 1990s showed clear molecular differences between Spitz nevi and melanomas, with the majority of Spitz nevi being of normal chromosome complement (with the exception of 11p gains in 4 of 17 cases) [31]. A later study of spitzoid neoplasms showed no chromosomal aberrations in seven of eight Spitz nevi; one Spitz nevus showed gains in both 7p and 11p (both of which are now commonly recognized to occur in benign Spitz nevi). In contrast, melanomas with spitzoid features showed multiple segmental chromosomal abnormalities [32].

Initial studies have shown that the majority of nevi have no CNAs by CGH. However, in recent years it has become apparent that certain types of benign nevi or indolent melanocytic proliferations are characterized by isolated genomic abnormalities, which in some cases can be detected by CGH/SNP array (Fig. 3). These findings are important if one is to use CGH to support a diagnosis of melanoma or nevus. Finding isolated CNAs in a melanocytic neoplasm does not imply necessarily that it is a melanoma; in fact, specific abnormalities, when present in isolation in a histologically atypical tumor, may support an indolent biologic behavior. Among spitzoid tumors, much progress has been made in recent years in identifying driver genetic alterations [33,34,35,36]. The first observation was made over 20 years ago when a subset of Spitz nevi was found to harbor HRAS gene mutations and to demonstrate gains of 11p by CGH [33]. On histology, these lesions are intradermal, associated with marked desmoplasia, have an infiltrative pattern, and may show some cytologic atypia and increased proliferation, features now recognized as characteristic for desmoplastic variant of Spitz nevus (Fig. 3a, b). In this context, finding an isolated 11p gain by CGH/SNP array in an atypical spitzoid neoplasm supports a diagnosis of Spitz nevus (Fig. 3c). A more recent study has found that almost half of spitzoid lesions including Spitz nevi, ASTs, and spitzoid melanomas harbor specific rearrangements resulting in kinase fusions of ROS1, NTRK1, NTRK3, ALK, BRAF, MET, and RET [34, 37]. The fusions can be detected by CGH/SNP array if they are the result of unbalanced translocations (Fig. 3j–l) [38].

Fig. 3: Specific CGH/SNP-array abnormalities in benign or indolent melanocytic tumors.
figure 3

a–c Desmoplastic Spitz nevus. a Intradermal melanocytic proliferation with pronounced desmoplastic stromal reaction and infiltrative growth pattern. b Epithelioid cells with a large nuclei, prominent nucleoli, and abundant amphophilic cytoplasm. Two mitotic figures are noted in the center of the field (arrows). c SNP array showing a gain of 11p (arrow) with no additional abnormalities, suggesting a desmoplastic Spitz nevus. d–f BAP1-inactivated tumor. d Large, predominantly intradermal tumor with biphenotypic morphology and a lymphocytic host response. e Top, majority of the lesion is composed of epithelioid cells with large nuclei, prominent nucleoli, and abundant eosinophilic cytoplasm with distinct cell membrane. Bottom, a component or ordinary nevus is identified. f SNP array showing a loss of 3p21 (arrow) with no additional abnormalities, suggesting an indolent BAP1-inactivated tumor/nevus. g–i Proliferative nodule arising in congenital nevus. g Hypercellular non-expansile dermal nodule composed of densely packed uniform melanocytes. h Melanocytes with round to oval hyperchromatic and slightly irregularly shaped nuclei with a moderate amount of eosinophilic cytoplasm, high nuclear to cytoplasmic ratio, and increased mitotic rate. i SNP array showing gains of whole chromosomes 2, 6, 13, 15, 16, and 22 suggestive of a proliferative nodule (arrow). j–l Low-risk atypical Spitz tumor with NTRK1 rearrangements. j Predominately intradermal melanocytic proliferation with a bulbous profile. k Epithelioid cells with intermediate size nuclei and moderate amount of eosinophilic cytoplasm arranged in back to back small nests with occasional mitotic figures (arrow). l SNP array results showing chromosome 1q. A small deletion is noted with breakpoints within LMNA and NTRK1 genes (arrows) resulting in an LMNA-NTRK1 fusion product.

Full size image

Another important discovery was of a subset of melanocytic tumors with epithelioid morphology, atypical features, and inactivation of the BAP1 gene. These lesions in the past were frequently diagnosed as atypical Spitz tumors [39, 40]. Histology shows a dermal proliferation of epithelioid melanocytes with abundant eosinophilic cytoplasm, distinct cellular membranes, large nuclei with vesicular chromatin, and prominent nucleoli, imparting a spitzoid appearance. In contrast to most Spitz nevi, the epidermis is atrophic. An associated banal nevus component and a conspicuous lymphocytic infiltrate are often present. Worrisome histologic features include cytological atypia and lack of maturation (Fig. 3d, e). BAP1 stain shows loss of expression in the large epithelioid cells with preservation of staining in the banal nevus component and can be used as a diagnostic tool. These lesions also harbor BRAF gene mutations. Despite being atypical, these lesions are generally characterized by an indolent biologic behavior. By CGH, BAP1-inactivated nevi are characterized by an isolated deletion on chromosome 3 of various size but always spanning the 3p21 area where the BAP1 gene resides (Fig. 3f).

CGH in congenital melanocytic proliferations and proliferative nodules

Congenital melanocytic proliferations and, more specifically, proliferative nodules developing in congenital nevi can also present diagnostic challenges. These are cellular melanocytic nodules with varying degrees of cytologic atypia and proliferative activity (Fig. 3g, h). There is significant clinical importance in differentiating them from melanomas arising in congenital nevi [41, 42]. By CGH, these lesions demonstrate gains and/or losses of entire chromosomes in contrast to melanomas that harbor CNAs involving partial segments of chromosomes (Fig. 3i) [43, 44].

CGH and difficult-to-classify melanocytic lesions

One of the major limitations of our current knowledge in regards to CGH/SNP arrays is the relative lack of robust studies correlating the patterns of CNAs with outcomes data in the histologically ambiguous melanocytic lesions for which CGH is needed the most. There are studies documenting that in melanoma, the number of CNAs correlate with progression from in situ to invasive and to distant spread, and that high number of focal CNAs and the presence of chromothripsis (an event where numerous genomic rearrangements occur at once [45]) correlate with poor outcome [46, 47]. One group found that that melanocytic lesions that  developed of metastases showed significantly more chromosomal aberrations on array CGH compared with those without metastases [48].

One can extrapolate these results to suggest that the same holds true for histologically borderline melanocytic tumors. A study by Ali et al. [32] on ten spitzoid neoplasms included some clinical follow-up data: a spitzoid neoplasm in this study from a 6-year-old male had one identifiable copy number change, gain in 19p, and clinically had lymph node involvement and local recurrence. A melanoma in that same study showed multiple chromosomal aberrations and a clinical course significant for distant metastases [32]. Future studies may allow for a better correlation with outcome. A second major limitation of CGH is the lack of clearly defined cutoff values or patterns that correlate with poor outcome in cases with small number of CNAs and more research is needed towards this end.

Fluorescence in situ hybridization

Background of the technique

FISH is a technique in which fluorescently labeled probes are directed against specific segments of genomic DNA and hybridized to tissue samples. Samples can be fresh, frozen, or paraffin-embedded tissue sections and a fluorescent microscope is used for detection. Advantages of this technique include limited tissue requirements and the ability to directly visualize cells of interest. The direct visualization allowed by this technique is particularly helpful for tissue specimens that are obscured by inflammation or other non-tumoral cells. Similar to CGH, FISH allows for the detection of chromosomal numerical abnormalities, based on losses or gains of fluorescent signals. A normal/non-neoplastic cell is expected to have two copies of any given chromosome or chromosomal fragment; therefore, identification of more or of less than two fluorescent signals is indicative of chromosomal gains or losses, respectively [49].

The original application of FISH to melanocytic lesions utilized a four-probe dataset, derived from earlier data on DNA copy number alterations. As the technique has evolved, refinements to this original four-probe dataset have been made in order to improve the assay, as discussed below.

FISH in the evaluation of melanocytic lesions

Investigation of melanocytic lesions using FISH began in earnest in the late 1990s. In 1994, Matsuta et al. [50] applied FISH using a centromeric probe specific for chromosome 17 and found that melanomas display more signals in comparison with benign nevi. In a follow-up study, some of the same authors reported their evaluation of 22 melanomas (14 primary and 8 metastatic) for chromosomal copy number changes and observed gains of chromosome 6 and 7, and/or losses of chromosome 9 and 10, which they suggested might play a role in melanoma development and progression [51].

Initial studies and proofs of principle

Similar to CGH, FISH has been used as an ancillary technique to assist in differentiating benign nevi from melanomas by identifying numerical abnormalities at probed loci. Selecting probes on the basis of existing CGH data [19], an initial study by Gerami et al. [52] aimed to develop a multiplex FISH assay by selecting the minimum number of probes capable of providing the highest discrimination between nevus and melanoma from a larger set of probes targeting loci found to be more frequently altered. The first probe set employed probes for 6p25 (RREB1), 6q23 (MYB), 11q13 (CCND1), and centromere 6 (CEP6), and had a sensitivity of 86.7% and a specificity of 95.4% in differentiating unequivocal melanomas from nevi [52].

Using this four-probe set, FISH analysis is performed by evaluating 30 nuclei and calculating the percentage of nuclei with more than 2 signals per nucleus for 6p25 and 11q13, and with less signals for 6q23 compared with CEP6. The test is considered positive if it demonstrates gains for 6p25 or 11q13, or losses of 6q23 (compared with CEP6) relative to validated cutoff values. In a follow-up study, the sample size was expanded and a sensitivity of 83.0% and specificity of 94.0% were reported [53]. Different authors, using the same probe set, studied 40 histologically unequivocal melanocytic tumors (10 metastatic melanoma, 10 primary melanomas, and 20 benign melanocytic nevi) and showed distinction of malignant from benign lesions with 90% sensitivity and 95% specificity [54]. Furthermore, in a study of 144 primary melanomas, it was found that a positive FISH result was an independent adverse prognostic marker, and that FISH-positive primary melanomas had a significantly increased risk of metastasis or melanoma-related death in comparison with FISH-negative cases [55].

Further studies examined the use of this four-probe set on various types of melanocytic lesions. A study on melanocytic deposits in sentinel lymph nodes supported the use of FISH in differentiating nodal nevi from metastatic melanoma [56]. The standard four-probe set was also found to assist in the distinction of nevoid melanoma from mitotically active nevi [57], conjunctival nevi from conjunctival melanoma [58], and epithelioid blue nevus from blue nevus-like melanoma [59]. Similarly, FISH has been suggested as a tool for microstaging of melanoma associated with nevi [60]. A modestly sized study of 12 unequivocal spitzoid melanomas and 6 Spitz nevi showed a sensitivity of 87.5% and a specificity of 100% in differentiating spitzoid melanomas from Spitz nevi [61]. FISH testing was found to have a lower sensitivity for spindle cell melanoma and spitzoid melanoma; a study of 22 pigmented spindle cell nevi and 24 spindle cell melanomas found that FISH testing had a 73% sensitivity and 93% specificity in separating malignant from benign [62]. Another study of 43 unequivocal spitzoid melanomas using the same standard probes against chromosomes 6 and 11 found a sensitivity of only 70%. This sensitivity was improved to 85% when the standard assay was combined with 9p21 [63].

FISH and difficult-to-classify melanocytic lesions

Studies investigating the utility of FISH in ambiguous or difficult-to-classify melanocytic lesions have, in general, found lower sensitivity and specificity than studies comparing only unequivocal lesions. The most robust data regarding ambiguous lesions comes from association with outcomes data. Robust data, however, is difficult to obtain, as many ambiguous melanocytic lesions are treated (altering the natural disease course) and adverse events can be experienced years (sometimes many years) later. What follows is a highlighting of some of the most pertinent available data.

The original study by Gerami et al. [52] included 27 histologically ambiguous lesions, 6 of which developed metastasis. All six cases that developed metastasis were positive by FISH [52]. In a study of 38 atypical spitzoid lesions, the only fatal outcome showed multiple chromosomal alterations by standard FISH [64]. In a separate study with nine histologically challenging lesions, a positive FISH result was obtained in one lesion only. This patient developed lymph node metastasis 14 years later [65]. A study of two spitzoid lesions with fatal outcomes showed a sensitivity of 50% (one of two lesions accurately detected) by FISH using standard probes. The authors of that study concluded that the current panel of FISH probes failed to detect the ASTs that had copy number variations by array CGH, including a metastatic and fatal AST. This group suggested that a comprehensive, genome-wide approach would offer great sensitivity than FISH probes alone [66]. In one of the more sizeable studies, Vergier et al. [67] showed a sensitivity of 85% and a specificity of 90% using a four-probe FISH assay on 43 non-equivocal (non-ambiguous) melanomas and nevi. However, when FISH was evaluated on 90 ambiguous lesions with outcomes data (the presence of metastases at minimum follow-up of 5 years), the sensitivity and specificity of FISH was more modest (43% and 80%, respectively). Sensitivity and specificity increased to 90% and 76%, respectively, when combined with histological review. These authors conclude that a positive FISH result reinforces a diagnosis of melanoma [67].

Gaiser et al. [48] found a somewhat modest sensitivity when FISH was applied to 12 histologically ambiguous lesions in comparison with clinical behavior (the presence or absence of metastases; mean follow-up of 65 months) with a sensitivity of 60% and a specificity of 50%. These authors reached the conclusion that FISH testing did not reach clinically useful sensitivity and specificity [48]. However, it has been suggested that their data may be less robust due to methodologies, specifically in the diagnostic criteria used [68].

Tetzlaff et al. [69] studied 34 histologically ambiguous melanocytic lesions, with follow-up data available for 25, and found that no metastases were experienced in the 17 lesions favored to be benign (with a mean follow-up of 16.8 months). Among lesions favored to be malignant, during a mean follow-up of 14.6 months, a single lymph node metastasis was identified in a 4-year-old girl, who had negative FISH testing. Based on integrated final diagnosis of all features together, the sensitivity of FISH in this study for the diagnosis of melanoma was 50% and specificity was 87.5%. In all cases, the initial diagnosis remained unchanged. The authors concluded that in their experience, in the setting of a lesion with predominately benign findings, a negative FISH test is a reassuring finding. In contrast, a positive FISH result should be carefully interpreted in the context [69].

Several other groups have published experiential data with FISH. In examining 804 ambiguous melanocytic neoplasms, a  group from the University of California San Francisco found that FISH allowed for a more definitive diagnosis in 88% of cases, and was particularly helping if positive [70]. A separate group, who studied 140 histologically ambiguous lesions (with limited follow-up data), also found the use of FISH helpful, while warning about the risk of false-positive results from tetraploidy (see below). Ten lesions with abnormal FISH results were further classified as likely melanoma. No diagnoses were altered based upon negative FISH results [71]. A final group took a more tempered view, highlighting the challenges in studying melanocytic tumors of uncertain malignant potential in the context of a less-than-perfect gold standard [72].

Polyploidy/tetraploidy

One major source of false-positive results is the presence of polyploidy/tetraploidy or balanced chromosomal duplications involving all loci. Tetraploidy has been observed in benign melanocytic lesions, such as Spitz nevi [73]. Rejection of cells showing copy number increases of all probed targets is important in limiting false-positive results [71, 74]. Use of an assay with four separate loci and 9p21 also allows tetraploidy to be recognized with greater confidence, as balanced gains otherwise become statistically unlikely [74].

Beyond the four-probe assay

Incorporating probes for 9p21 (CDKN2A) and 8q24 (MYC) into the four-probe FISH assay was found to improve sensitivity for spitzoid melanomas [63, 74]. A multicenter study described the association of 9p21 homozygous deletion with aggressive clinical behavior in spitzoid melanocytic tumors and also found that 6p25 or 11q13 gains have an increased risk for aggressive clinical behavior compared with FISH-negative ASTs. Interestingly, the presence of isolated 6q23 deletion was not associated with adverse events in spitzoid neoplasms [75]. These studies formed the basis for the addition of 9p21/ Cep9 and 8q24 to the standard four-probe set in some assays. However, even with this improved probe set, not all investigators experienced robust results. One group found low correlation between FISH results and a final diagnosis of melanoma or metastasis [76].

Comparison of FISH and CGH

Concordance between the two methods has been observed to be reasonably high (90% using the original four-probe set) [77] and, from a practical standpoint, both of these techniques have advantages and disadvantages (Table 1). The major advantage of CGH/SNP array is that it allows for interrogation of the entire genome (in comparison with the limited number of genomic loci evaluated via FISH) and is therefore likely more sensitive. In the setting of tetraploidy, CGH/SNP array is also likely to be more specific due to the possible false-positive results seen with FISH [73]. A study by our group showed that for borderline melanocytic lesions FISH has a sensitivity and specificity of only 61% and 84%, respectively, when compared with an SNP array [78].

Table 1 Comparison between CGH /SNP array and FISH for evaluation of melanocytic lesions.
Full size table

There are some circumstances, however, when FISH testing may be preferred. FISH can be performed on superficial specimens with limited material (only two unstained slides are needed) or on specimens in which the tumor cells are infiltrated by a significant component of other cell types, as the cells of interest are directly visualized during the test. In contrast, CGH/SNP array requires 20–30% tumor purity and, especially for small biopsies, 10 tissue sections. FISH is less labor intensive than CGH/SNP array in general, with typically lower turnaround times and cost.

Conclusions

Recent years have seen an increase in the use of ancillary molecular tests in the diagnosis of histologically ambiguous or challenging melanocytic tumors. Although there are certain limitations in our knowledge regarding the exact correlation between various patterns of chromosomal abnormalities and outcome, especially in borderline melanocytic tumors, the existing body of literature suggests that both CGH and FISH can be useful in selected clinical scenarios and are indeed being used (Figs. 4 and 5). A study that surveyed a group of dermatopathologists on the use of these tests showed that 54% reported routine use of molecular testing for ambiguous melanocytic lesions with an additional 37% reporting rare use and only 8% reporting to never have used the tests [79].

Fig. 4: A 90-year-old with lesion on the scalp.
figure 4

Histologic diagnosis: Suspicious for melanoma arising in blue nevus vs. atypical cellular blue nevus. a Large dermal and subcutaneous melanocytic lesion displaying two distinct morphologies. There is a background of spindle cells with melanophages consistent with cellular blue nevus and an expansile cellular nodule. b Top, the cellular nodule contains spindle to epithelioid cells with prominent nucleoli, mild hypechromasia, and rare mitoses (arrow). Bottom, spindle cells with intervening melanophages consistent with cellular blue nevus. c SNP array reveals numerous CNAs including gains of 6p supporting a diagnosis of melanoma. Final diagnosis: melanoma arising in blue nevus.

Full size image
Fig. 5: A 14-year-old boy with a lesion on right superior cheek.
figure 5

Histologic diagnosis: atypical Spitz tumor of uncertain malignant potential. a Dermal proliferation of melanocytes with an infiltrative pattern associated with a sclerotic stroma and lymphoid aggregates. b Cells show a spitzoid morphology with occasional nuclear hyperchromasia and pleomorphism. c p16 immunohistochemistry shows diffuse loss of expression. d FISH with 9p21 (CDKN2A) probe (red) and centromere 9 probe (green). Neoplastic cells show one green signal (white arrow), indicating one copy of centromere 9 but no red signals. On the left side there is a non-neoplastic cell showing two copies of 9p21 (red arrows). Rendered diagnosis: high-risk atypical Spitz tumor concerning for spitzoid melanoma.

Full size image

The use of FISH and CGH will likely increase, as their roles are worked out and refined. The greatest utility for these tests is for lesions situated in the center of the biological spectrum (i.e., in lesions that are not obviously benign or malignant by histologic examination). Results of a study involving a panel of 17 experts that were asked to rate the appropriateness of using molecular testing in various clinical scenarios showed that CGH/FISH assays were considered usually appropriate in most clinical scenarios in which the diagnosis of melanoma is in question but histology is not definitively diagnostic (i.e., borderline melanocytic tumors). As expected, molecular testing was not considered appropriate when pathology is definitive for melanoma or nevus [80].

Unfortunately, borderline melanocytic tumors with ambiguous histology may also demonstrate equivocal molecular results such as a low number of chromosomal abnormalities with uncertain biological significance in relation to outcome [35]. In order to make sense of the molecular results, they need always to be interpreted in conjunction with histological findings on routine light microscopy and with the clinical presentation. In an ambiguous lesion, which is favored to be benign by histology, negative FISH and/or CGH testing is reassuring for an indolent biologic behavior and the lesion may be managed as an atypical nevus (i.e., conservative excision), with the caveat that FISH lacks sensitivity compared with CGH. In an ambiguous lesion, which is worrisome for melanoma, a positive FISH and/or CGH test is supportive of that diagnosis and may warrant more aggressive treatment. However, abnormal testing in a lesion that is ambiguous but favored to be benign and negative testing in a lesion favored to be melanoma should be interpreted with caution, as both false-positive and false-negative results can occur [69].

It should be noted that although this study is focused on CGH and FISH, there are other types of ancillary molecular tests designed to differentiate between melanoma and nevus or evaluate prognosis in melanoma. PCR analysis of RNA expression from genes found to be differentially expressed in melanoma (“gene expression signature analysis”) has also shown promising results. Both 23- and 28-gene expression profile tests have been studied and are available commercially [81, 82]. Analysis of telomerase reverse transcriptase promoter mutations with PCR may be helpful in some cases, particularly when positive [83]. The study of epigenetic alterations, such as DNA methylation, has also shown promise [84]. The advantages, disadvantages, and precise roles of each technique will be clarified, as more data and experience is accrued.

The past few decades have seen substantial advances in how we are able to examine melanocytic lesions, with CGH and FISH being prime examples. As more clinical outcome data are collected, the role of these techniques for ambiguous lesions will likely be further refined.