Melanoma Pathology 2.0 – New Approaches and Classification

Abstract

Cancer is caused by the accumulation of pathogenic alterations of the genome and epigenome that result in permanent changes that disrupt cellular homeostasis. The genes that become corrupted in this process vary among different tumor types, reflecting specific vulnerabilities and dependencies of the cell from which the cancer originated. This also applies to ‘melanoma’, a cancer that constitutes not one, but multiple diseases that can be separated based on their cell of origin, etiology, clinical appearance and course, and response to treatment.

We review the current classification of melanoma within distinct evolutionary pathways and the associated genetic alterations. In addition, we review the application of molecular diagnostics to the diagnosis of melanocytic tumors in the context of histopathologic assessment.

Introduction

The spectrum of genetic alterations in melanocytic neoplasms varies depending on when in life and at which anatomic site the tumors arise. This is strikingly illustrated by the fact that mutations in BRAF or NRAS are common in melanomas on sun-exposed skin and mucosa^1,2 but are virtually absent in those of the uveal tract^3,4. The latter invariably have mutations in the Gαq signaling pathway instead. These differences likely reflect differences in the cells of origin from which these neoplasms arise and/or their microenvironment that likely lead to different vulnerabilities of the cells for transformation. The absence of one type of mutation in one tumor type does not mean that the mutation does not arise in the cell of origin but indicates that it does not provide a selective advantage in that cell type. The patterns of somatic mutations in melanomas, and cancers in general, thus reflect the complex interplay between the processes that generate mutations, the starting state of the cell in which they occur, and the homeostatic rules that constrain that cell.

Without exception, melanomas harbor multiple genetic alterations that corrupt more than one pathway. This reflects the existence of tumor suppressive mechanisms, usually multiple, in long-lived organisms such as humans. Tumor progression to melanoma typically occurs in discrete steps, as the pathways required for full transformation are incrementally corrupted. This can result in partially transformed neoplasms, in which progression is temporarily or permanently halted. Examples include the melanocytic nevus, a clonal proliferation of melanocytes that have acquired a mutation that promotes proliferation but have retained tumor suppressive mechanisms resulting in a stable state. In most nevi this mutation is BRAF V600E, which strongly activates the MAPK pathway^6,7. Other nevus-initiating mutations affect other signaling components of the MAPK pathway and, as these mutations are sufficient for pathway activation, usually only one of them is present in a given nevus. The occasional progression of such a nevus results from one of the constituent cells acquiring an additional mutation that overrides the proliferation-constraining mechanism. This typically involves inactivation of CDKN2A or mutation of the TERT promoter, highlighting oncogene-induced and replicative senescence as important factors that help keep these benign neoplasms in check.

Some not overtly malignant melanocytic tumors harbor additional pathogenic alterations that resulted in a second round of clonal expansion. This points to the redundancy of tumor suppressor mechanisms, as backup mechanisms apparently become engaged when one senescence mechanism becomes corrupted. Genetically intermediate melanocytic tumors are now designated melanocytomas in the revised WHO Classification⁸. Their risk of malignant transformation likely is higher than their corresponding nevus stage but remains to be quantified.

Pathways to melanoma defined in the 2018 WHO Classification of Skin Tumors

The framework outlined above has been the basis of a two-dimensional classification schema in which different melanoma subtypes are distinguished along one axis and their respective precursor stages on the other⁹ and serves as the foundation for the revised melanoma classification in the 2018 WHO Classification of Skin Tumors^8,10.

Classification of melanomas according to their evolutionary pathways attempts to integrate our understanding of causal mechanisms, clinical and histopathological presentation, diagnosis, prognosis and will be a work in progress for some time (Table 1). Here we provide a brief review of these pathways.

Table 1. WHO pathways of melanoma development.

Adapted from ⁸.

	Low CSD melanoma (I)	High CSD melanoma (II)	Desmoplastic melanoma (III)	Spitz melanoma (IV)	Acral melanoma (V)	Mucosal melanoma (VI)	Melanoma arising from congenital nevus (VII)	Melanoma arising in blue nevus (VIII) and uveal melanoma (iX)
Nevus (common mutations)	Conventional nevus (BRAF V600E, NRAS)	No known benign precursor	No known benign precursor	Spitz nevus (HRAS, kinase fusion)	Minority arises from conventional nevus (BRAF V600E) Other benign precursors unknown	No known benign precursor	Congenital nevus (NRAS mutation, BRAF V600E)	Blue and uveal nevi (GNAQ, GNA11, CYSTLTR1, or PLCB4 mutation)
Melanocytoma (common secondary mutations)	Deep penetrating melanocytoma (CTNNB1, APC) BAP1-inactivated melanocytoma PRKAR1A-inactivated melanocytoma Dysplastic nevus (TERT promoter)	Melanoma in situ (NRAS mutation, BRAF non-V600 mutation, KIT mutation)	Occasionally arises from melanoma in situ	Spitz melanocytoma (previously known as atypical Spitz tumor, CDKN2A inactivation)	Melanoma in situ (focal amplifications of CCND1 and other genes, KIT mutation)	Melanoma in situ (focal amplifications of CCND1 and other genes, KIT mutation)	Proliferative nodule (copy number alterations of whole chromosomes) Melanoma in situ	intermediate lesion unknown atypical blue nevus?
Mutations in Melanoma	Mutation of TERT promoter, TP53, inactivation of CDKN2A, PTEN	Inactivation of NF1, CDKN2A, PTEN, mutation of TERT promoter TP53, RAC1	NRAS mutation, NF1 inactivation, amplification of ERBB2, MAP2K1, MAP3K1, BRAF, EGFR, MET, NFKBIE mutation, PIK3CA mutation, PTPN11 mutation	TERT promoter mutation, CDKN2A inactivation	Amplification of TERT, YAP1, EP300, CDK4, MDM2, inactivation of NF1, CDKN2A, ATRX, mutation of NRAS	Amplification of CDK4, MDM2, inactivation of SPRED1, NF1, ATRX, mutation of NRAS, SF3B1	Mutation of TERT promoter	Inactivation of BAP1, mutation of SF3B1, EIF1AX

Pathway 1: Low cumulative sun damage melanoma (low-CSD, superficial spreading melanoma)

The degree of solar elastosis in the dermis reflects the degree of sun exposure and corresponds with patient age, pigmentary phenotype and anatomic site. Melanomas that arise in a background of low cumulative sun-damage (low-CSD) typically arise on the trunk and extremities of adults between age 20 and 60¹¹. Most of these melanomas harbor a BRAF V600E mutation (Figure 1). The defining feature, low cumulative sun-damage, is assessed by the degree of solar elastosis¹¹. Most low-CSD melanomas are superficial spreading melanomas.

Figure 1. Low cumulative sun damage (low-CSD) melanoma arising from a nevus on the back of a 45-year-old man.

The nevus is in the mid-dermis and composed of small melanocytes that mature with descent in the dermis (lower right). The melanoma is composed of large nests of atypical melanocytes with abundant cytoplasmic pigment arrayed in the epidermis and superficial dermis (upper left). Note the minimal amount of solar elastosis in the dermis. Both the nevus and the melanoma harbor a BRAF V600E mutation while the melanoma alone harbors a TERT promoter mutation and copy number gains and losses characteristic of melanoma. This case was previously reported (case 2)⁹⁶. Hematoxylin and eosin stained slide digitally imaged at 40x magnification.

Common or conventional nevi are the main precursor for this melanoma subtype. These nevi typically arise in the first two decades of life and the majority harbor BRAF V600E mutations with a minority harboring NRAS mutations instead. Different types of melanocytomas can develop within these nevi through additional mutations. Deep penetrating melanocytoma (formerly nevus) is defined by additional mutations that activate the beta-catenin pathway¹². BAP1-inactivated melanocytoma (previously classified as a Spitz tumor) is defined by inactivation of the tumor suppressor gene BAP1 ^13–15. PRKAR1A-inactivated melanocytomas are a subset of pigmented epithelioid melanocytoma, formerly referred to as epithelioid blue nevus¹⁶. These melanocytomas are usually clinically stable but can progress to melanoma through further genetic alterations^17,18. Assessment of the presence of additional mutations such as in CDKN2A or the promoter of TERT and chromosomal copy number aberrations can assist with differentiating melanocytoma and low-CSD melanoma.

Within the new paradigm of nevus -> melanocytoma -> low-CSD melanoma, it seems likely that some dysplastic nevi will turn out to be additional forms of melanocytoma that harbor additional progression events beyond the initiating mutation. Clinically, dysplastic nevi present as broad (> 5 mm) irregularly bordered and pigmented macules and thin papules¹⁹. Their histopathologic features include a broad intraepidermal component of single melanocytes and nests of them, nests of melanocytes bridging adjacent rete ridges, and fibroplasia and an inflammatory response in the superficial dermis. In some families, an increased number of dysplastic nevi is associated with a germline alteration such as inactivating mutations in CDKN2A²⁰. This explains why the presence of multiple dysplastic nevi is associated with an increased melanoma risk for the patient; it indicates the presence of a germline that interferes with the suppression of melanocyte transformation in all of the patient’s melanocytes. However, it is to be expected that dysplastic nevi may result from an acquired mutation that would be present in cells of that dysplastic nevus and inform on the progression risk of the individual dysplastic nevus but not reflect any germline predisposition to melanoma in the patient. Disentangling these factors holds the promise to solve the Gordian knot that the conflicting definitions for dysplastic nevi have tied and assist with their evidence-based clinical management.

Pathway 2: High cumulative sun damage melanoma (high-CSD, lentigo maligna melanoma)

Melanomas on skin with marked solar elastosis (high-CSD melanomas) tend to have a lentiginous or single cell growth pattern in the epidermis in contrast to the often pagetoid growth pattern of low-CSD melanomas. They typically have an extended radial growth phase, with a large patch of in situ melanoma developing before an invasive component forms (Figure 2). Both clinically and histopathologically, their borders may be difficult to demarcate and recurrence after excision is more common than for low-CSD melanoma.

Figure 2. Melanoma in situ arising on skin with high cumulative sun damage (high-CSD).

Note the extent of solar elastosis in the dermis. There is a broad junctional proliferation of melanocytes arrayed predominantly as single cells limited to the lower epidermis (lentiginous growth pattern). Hematoxylin and eosin stained slide digitally imaged at 40x magnification.

High-CSD melanomas typically do not harbor BRAF V600E mutations but instead harbor other MAPK kinase pathway mutations such as BRAF V600K, NRAS or KIT mutations, or inactivation of the negative regulators of Ras, NF1 or RASA2. A pre-existing nevus is typically absent and melanoma in situ is considered the predominant precursor lesion.

Pathway 3: Desmoplastic melanoma

Desmoplastic melanoma is defined by a dermal component with a predominance of spindled, unpigmented melanocytes interspersed between thickened collagen bundles resembling a scar. When the desmoplastic component represents >90% of the tumor, it is classified as “pure” as opposed to “mixed” ^21,22. The pure subtype of desmoplastic melanoma infrequently metastasizes to the lymph nodes and has improved survival compared to other forms of cutaneous melanoma of similar tumor thickness. Local complications include extension along nerves due to neurotropism.

Most desmoplastic melanomas can be regarded as a variant of high-CSD melanoma, but they occasionally can arise in other settings. The high-CSD desmoplastic melanomas have an extremely high burden of UV-induced mutations²³. NF1 inactivating mutations often occur with other weakly activating mutations in the MAP-kinase pathway, but BRAF V600E or RAS mutations are infrequent^24,25. Promoter mutations in NFKBIE, encoding NF-κB inhibitor ε, are enriched in desmoplastic melanomas.

Pathway 4: Spitz melanoma (Malignant Spitz Tumor)

Spitz tumors were first recognized as distinct from cutaneous melanoma in the 1940s²⁶. They tend to arise during childhood and characteristically have epithelioid melanocytes with abundant cytoplasm and large nuclei, pagetoid scatter of melanocytes within the epidermis and limited maturation of dermal melanocytes. The set of initiating mutations in Spitz nevi includes activating HRAS mutations, activating gene fusions of receptor tyrosine kinases (ALK, ROS1, NTRK1/2/3, RET, MET, MERTK) and serine/threonine kinases (BRAF, RAF1, MAP3K8) with a broad range of partner genes^27–33.

In the WHO classification, Spitz tumor refers to a spectrum of progression states ranging from nevus to melanoma that, in addition to the histopathological findings, is defined by the initiating oncogenic alterations listed above. This new convention is intended to separate Spitz melanoma from what has been called spitzoid melanoma, a mainly cytologically defined presentation of melanoma that mostly consists of other types of cutaneous melanomas. While this is a step forward in defining Spitz tumors and their progression, there is additional unresolved complexity, which is due to the fact that the various initiating alterations are associated with differences in histopathological appearance^34–39 and perhaps risk of progression to melanoma.

Genetic alterations associated with evolution to Spitz melanoma include inactivation of CDKN2A, PTEN, and TP53 and promoter mutations in TERT^40–42, similar to other subtypes of melanoma. Spitz tumors with homozygous deletion of CDKN2A but without additional driver alterations such as TERT promoter mutation appear to have indolent behavior⁴³ and should be considered intermediate Spitz tumors, with the nomenclature Spitz melanocytoma preferred to the previously used term atypical Spitz tumor. Spitz melanocytomas can spread to regional nodes frequently but exceedingly rarely lead to distant metastases or fatal outcomes⁴⁴. Spitz melanomas tend to occur in younger patients and likely have a better outcomes than other melanomas (Figure 3)⁴⁵.

Figure 3. Spitz Melanoma with ALK fusion.

A. Low magnification view demonstrates an asymmetrically distributed compound melanocytic proliferation. Epidermal hyperplasia as is typical of Spitz nevus is present. B. Medium magnification view of the expansile dermal nests. C. High magnification view of melanocytes with expanded cytoplasm, atypical nuclei and mitotic activity. Spitzoid cytomorphology is present. This Spitz melanoma occurred on the thigh of a 23-year-old woman and demonstrated multiple copy number gains and losses including CDK4 amplification, and was previously reported (case 7)⁴². Hematoxylin and eosin stained slide digitally imaged at 40x magnification.

Pathway 5: Acral Melanoma

Melanomas on the non-hair bearing (glabrous) skin of the hands and feet including the nail unit are referred to as acral melanomas. Recent genetic data indicate that a subset of melanomas on these sites represent low-CSD melanomas, defined by the presence of BRAF V600E and their pattern of DNA copy number changes^46,47. Similar to their non-acral counterparts, acral melanoma of this subset tend to occur in younger patients with European ancestry and may arise from a pre-existing nevus. We postulate that most BRAF V600E mutant acral melanomas are biologically similar from those on non-acral skin.

The remaining acral melanomas, most of which have MAPK activating mutations other than BRAF V600E, clinically and histopathologically fall into the category of acral lentiginous melanoma (ALM) and have distinctive genetic features that separate them from high- and low-CSD melanomas^48–51. They have a low point-mutation burden and instead harbor many focal amplifications and deletions, and structural rearrangements. They have a lentiginous or single cell growth pattern with a broad radial growth phase that may be present for many years before progression to invasive melanoma and often show extension along eccrine ducts (Figure 4). Recent studies in mice identified a melanocyte precursor cell within eccrine glands that may be the cell of origin of acral lentiginous melanoma⁵². The in situ phase often fades into an area colonized by field cells, neoplastic melanocytes that appear close to normal in cytomorphology and are equidistantly spaced in the basilar epidermis with no notable increase in density. They share, however, copy number changes including pathogenic amplifications with the adjacent manifest in situ and invasive melanoma and are considered a very subtle and early form of melanoma in situ⁵³. The field cells can extend several centimeters, explaining why the peripheral margins of acral melanoma in situ can be difficult to assess and partial biopsies have to be assessed with caution. The presence of the occult field cells likely also explains an increased local recurrence rate after surgical excision.

Figure 4. Acral melanoma arising within a broad patch of acral melanoma in situ.

A. Low magnification view demonstrates an ulcerated nodule of invasive melanoma arising within a melanoma in situ. B. Medium magnification view of the adjacent melanoma in situ shows the lentiginous growth pattern, similar to the high-CSD melanoma in situ in Figure 2. Hematoxylin and eosin stained slide digitally imaged at 40x magnification. C. The copy number profile shows the log2 ratio of the tumor genome compared to a normal reference on the y-axis along the genome on the x-axis. A log2 ratio of 0 represents a normal or unaltered copy number state. The acral melanoma depicted here has numerous copy number losses (chromosomal regions with log2ratio <−0.5) and gains (log2ratio>0.5) including focal amplifications of TERT, CDK4 and MDM2 (log2ratio>2).

Acral lentiginous melanoma harbors a diversity of genetic driver alterations, many of which are promising drug targets, including KIT mutations, non-V600E BRAF mutations, and kinase fusions. It can be difficult to distinguish BRAF V600E mutant acral melanoma from ALM by histopathologic and clinical features alone. Molecular testing for V600E and other BRAF mutations should be performed when BRAF targeted therapy is clinically indicated.

Pathway 6: Mucosal Melanoma

Mucosal melanomas arise from melanocytes in the sinonasal and genitourinary mucosa. Due to their internal location, mucosal melanomas are often diagnosed either as large primary tumors or metastases. Similar to acral melanomas, mucosal melanomas harbor a low point mutation burden and also harbor many focal amplifications and deletions and structural rearrangements^51,54–56. BRAF V600E mutations are uncommon. There are indications of differences to acral melanoma and perhaps variation among mucosal melanomas in that genitourinary mucosal melanomas more commonly have hotspot mutations in SF3B1 as compared to sinonasal ones⁵⁴. Furthermore, some mucosal melanomas have GNAQ mutations, further indicating a possible relationship between this category of melanocytic neoplasms thought to originate from dermal rather than epithelial melanocytes^57,58. Similar to acral lentiginous melanoma, mucosal melanomas have diverse driver alterations, including some that could be targeted therapeutically. SPRED1 was recently identified as a tumor suppressor in mucosal melanoma and it is often inactivated in the setting of mutant KIT⁵⁵.

Pathway 7: Melanoma arising in congenital nevus

Congenital nevi are defined as nevi that are present at birth, but nevi that arise shortly after birth are often subsumed under this term. However, there may be differences between bona fide congenital nevi, which are often very large and those arising after birth because, giant congenital nevi are most frequently caused by mutations in NRAS, whereas smaller congenital nevi more commonly harbor BRAF V600E mutations^59,60. This suggests that the cell of origin of giant congenital nevi can be more effectively transformed by NRAS mutation whereas the cell of origin of conventional acquired nevi are more effectively transformed by BRAF mutations, supporting the idea that their cells of origin may differ.

The risk of melanoma arising within a congenital nevus increases with the size of the nevus, likely reflecting the total number of melanocytes with the initiating mutation. The risk of melanoma developing in a congenital nevus is estimated to be up to 2.5-8% in giant congenital melanocytic nevi greater than 20 cm in diameter^61,62.

Rapidly growing nodules within a giant congenital nevus are not uncommon⁶³. Most are benign, even if the densely cellular nodules may appear worrisome histopathologically. Assessment of copy number in these tumors can be helpful as proliferative nodules typically harbor copy number gains and losses of entire chromosomes, while melanomas harbor copy number gains and losses of parts of chromosomes (Figure 5)⁶⁴. Specific secondary alterations apart from copy number changes to explain the development of these atypical proliferative nodules have not yet been identified.

Figure 5. Pseudo-melanomatous atypical proliferative nodule arising in a congenital nevus.

A. Low magnification view shows a large nodule arising within a giant congenital nevus on the neck of a 3-year-old boy. B. Medium magnification view of the nodule demonstrates sheets of melanocytes with scant cytoplasm and nuclei with an open chromatin pattern. Pyknotic nuclei are present. Hematoxylin and eosin stained slide digitally imaged at 40x magnification. C. Copy number profile shows gains and losses of whole chromosomes. A notable exception is the copy number change of chromosome 1 which affects only the long chromosomal arm.

Pathway 8 and 9: Melanoma arising in Blue Nevus and Uveal Melanoma

Melanoma in blue nevi and uveal melanoma arise from melanocytes that are not associated with any epithelium and share similar genetic alterations. They both harbor activating mutations in the Gαq pathway, predominantly in GNAQ and GNA11 but rarely in their upstream receptor CYSLTR1 ^65–68 or their downstream effector PLCB4⁶⁹. They do not harbor a UV signature^{3,4,65,66,69,70}. Mutations in the Gαq pathway are sufficient to form blue nevi in the skin and uveal nevi in the eye. Progression to melanoma occurs through a distinct set of additional genetic alterations, which include bi-allelic inactivation of BAP1, hotspot mutations in SF3B1 (similar to a subset of mucosal melanomas), or mutations in EIF1AX^71–73. Specific DNA copy number changes are also shared by both types of melanoma and include monosomy 3, resulting in loss of heterozygosity of BAP1, and gain of 8q, which contains the oncogene MYC^74,75. Uveal melanomas are divided into low risk (class 1) and high-risk (class 2) tumors by their expression profiles⁷⁶. Class 2 signatures are associated with BAP1 inactivation and portend a higher rate of liver metastasis and lethal outcome^77,78. Uveal melanomas disseminate hematologically with a propensity to metastasize to the liver^79,80. Blue nevus like melanomas often disseminate through lymphatics, similar to other cutaneous melanomas although cases with a heavy burden of liver metastasis have been reported, suggesting that some may display a similar liver tropism^75,81.

Blue nevi may originate from melanocytes that developed via the ventromedial pathway in which stem cells from the neural crest migrate along developing nerves and can give rise to Schwann cells or melanocytes⁸². This may explain the sometimes segmental distribution of blue nevi (Nevus of Ito and Ota), which are often associated with perineural involvement.

Melanomas arising in blue nevus need to be distinguished from other highly pigmented melanomas (i.e. with β-catenin activation or PRKAR1A-inactivation). If a remnant of blue nevus cannot be clearly identified, the presence of an activating Gαq mutation or BRAF V600E mutation can help classify lesions into the blue or low-CSD pathway, respectively.

Final thoughts on the WHO pathway classification

We anticipate that this classification will continue to evolve as additional knowledge is uncovered. Criteria to classify individual lesions into respective pathways have to be refined and may change. In fact, as discussed above, since the development of this classification, additional data suggests that a subset of acral melanomas likely belong in the low-CSD pathway. As compared to previous classifications, the genetic drivers and level of cumulative sun-damage as assessed by the degree of solar elastosis in the skin of the primary melanoma are given more weight and classification has become less dependent on other histopathologic features.

Nodular melanoma refers to melanomas in vertical growth phase that do not have an associated benign precursor or radial growth phase. Rather than constituting a melanoma type/pathway on their own, nodular melanomas occur in most of the above pathways. They do not seem to show discernible differences in genetic alterations from other melanomas in their respective pathway, but it has been proposed that they arise through an ‘inverted’ sequential order in which pathogenic mutations arise⁸³. The initial mutations would not lead to neoplastic proliferation but once such ‘poised’ melanocytes acquire oncogenic drivers in the MAPK pathway they would be fully transformed, skipping over earlier progression stages.

Molecular diagnostics

Molecular diagnostics in melanoma are used for both diagnosis and treatment selection. Histopathologic and genetic features that define the boundaries between nevus, melanocytoma, and melanoma are still being delineated. At the current time, the integration of molecular diagnostics with clinical and histopathologic features of a tumor is critical to providing the best diagnosis.

Immunohistochemistry

A mutation-specific antibody has been developed for BRAF V600E⁸⁴. It does not recognize V600K/D/R or other BRAF mutants. Mutation specific antibodies for RAS exist for RAS Q61R and RAS Q61L that recognize specific Q61 but not Q61H mutant forms of H-, K-, and NRAS and other important variants such as those at codons 12 or 13^85,86. With appropriate validation, mutation-specific antibodies have good sensitivity and specificity and short turn-around time.

To detect or screen for kinase fusions, immunohistochemistry to detect the kinase domain of the native protein can be used. The kinase domain is present in the normal and the chimeric protein generated by structural rearrangements, but as the normal proteins of ALK and ROS1 are not expressed in normal melanocytes, immunoreactivity with either antibody in a melanocytic tumor is indicative of the respective kinase fusion. A caveat is alternative transcript initiation of ALK (ALK^ATI) that results in the expression of a truncated form of ALK, which can lead to immunoreactivity in the absence of a fusion⁸⁷. The normal protein of other kinase genes are often expressed in normal melanocytes, making immunohistochemistry less suitable to detect rearrangements. While immunohistochemistry for the kinase domain of NTRK1/2/3 using the pan-Trk antibody can be used to screen for NTRK1/2/3 fusions, only very strong homogeneous expression predicts an underlying fusion⁸⁹. Depending on the fusion partner, the antibodies against kinase domains can demonstrate different staining patterns. ETV6-NTRK3 has mostly nuclear expression whereas MYO5A-NTRK3 has mostly cytoplasmic expression³⁹. Immunohistochemistry currently is not useful to detect BRAF or MAP3K8 fusions.

In deep penetrating melanocytoma, a range of different mutations result in β-catenin accumulation, which can be detected by immunohistochemistry. Conventional melanocytic nevi express β-catenin in a gradient pattern with higher expression levels in neoplastic melanocytes near the epidermis and other epithelial structures. However, in deep penetrating melanocytoma, there is strong uniform beta-catenin expression from top to bottom, sometimes with notable nuclear staining^12,90.

Immunohistochemistry can also be used to infer loss of tumor suppressors. Tumor suppressor proteins that can be evaluated by immunohistochemistry include BAP1, PRKAR1A, p16, and NF1. However, missense mutations can disrupt protein function but maintain immunoreactivity and thus lead to false negative results (i.e. positive immunoreactivity despite loss of a functional protein). In addition, the absence of expression of a tumor suppressor may indicate that the tumor suppressor has not been induced, rather than inactivated genetically.

Using immunohistochemistry to assess for molecular alterations in melanocytic tumors requires an understanding of the expected staining patterns and should be integrated with clinical and histopathologic features.

DNA and RNA based mutation detection

For detecting mutations in DNA, multiple platforms are in use, including Sanger sequencing, allele-specific PCR assays, and targeted next-generation sequencing (NGS). Each method has its advantages and disadvantages. Laboratories are gradually shifting to targeted NGS as the range of relevant genetic alterations in cancer expands. Depending on the targeted NGS platform, other genetic alterations including DNA copy number changes and loss of heterozygosity can also be identified.

Identification of structural rearrangements such as kinase fusions requires specific techniques as the genomic breakpoints usually occur in intronic regions that can span hundreds of kilobases. NGS of DNA can detect fusions if entire introns are targeted for capture, but RNA sequencing is a more robust and cost-effective method of detection. Detection of fusion transcripts by RT-PCR is highly sensitive and specific, but RT-PCR assays must be designed to detect specific fusions and is not practical for detecting the broad spectrum of kinase fusions that occur in melanocytic neoplasms.

Most melanomas demonstrate multiple DNA copy number changes, with some differences in pattern among those that develop within different pathways. A small number of specific gains or losses can be assessed by fluorescence in situ hybridization (FISH). The advantage of FISH is that it allows for assessment of tumors with heterogenetic subclones, small size or significant inflammation. Copy number changes, and depending on the platform also allelic imbalance, can be assessed by comparative genomic hybridization (CGH) with the advantage of providing an assessment of the entire genome but at the cost of lower sensitivity if the tumor cell content is low or there is significant tumor heterogeneity.

While nevi typically do not demonstrate copy number alterations, Spitz nevi with HRAS mutations often demonstrate copy gain of chromosome 11p, which contains the mutant HRAS. Spitz nevi with kinase rearrangements may demonstrate copy number alterations of the chromosomal regions that flank the rearranged genes.

Gene Expression Profiling

Gene expression tests for melanocytic tumors aim to assist with diagnosis or prognosis. The expression of specific transcripts is measured across all cells in the tumor, including neoplastic melanocytes, stromal cells and inflammatory cells. MyPath melanoma (Myriad Genetics, Salt Lake City Utah) provides a numerical score binned into ranges that correspond with likely benign, indeterminate or malignant based on the expression of 23 genes⁹¹. Sensitivity and specificity are 91.5% (CI 86.4-95.2%) and 92.5% (CI 90.0-94.5%) respectively. Initial findings suggest that the test has different performance across different subtypes of nevi and melanoma^92,93 and may have decreased performance for indeterminate or intermediate tumors⁹⁴. The clinical utility of this test remains uncertain.

There are multiple prognostic gene expression tests for cutaneous melanoma. Improved risk-stratification can guide the use of sentinel lymph node biopsy, surveillance, and adjuvant therapy. However, these tests have not been assessed in clinical trials to determine what clinical utility they provide beyond AJCC staging parameters, and the Melanoma Prevention Working Group does not recommend their routine use⁹⁵.

Future directions

The pace of technologic development has led to a remarkable expansion of our understanding of the genetic progression of cancer and melanoma and melanocyte biology and resulted in improved treatments and refined diagnostic methods. One of the critical hurdles ahead is the development of prognostic biomarkers that can identify patients with primary melanomas at high risk of progression so that they can receive the appropriate adjuvant therapies while their residual tumor burden is still at a minimum.

Figure 6. Melanoma arising within a blue nevus on the scalp of a 23-year-old man.

A. Low magnification view demonstrates a partly necrotic and hemorrhagic nodule that extends into the subcutis. The melanoma harbored GNA11 Q209L and SF3B1 R625C mutations. B. Medium magnification view of region outlined in A shows blue nevus on the left, melanoma centrally, and hemorrhage and necrosis on the right. Hematoxylin and eosin stained slide digitally imaged at 40x magnification. C. Copy number profile reveals copy gains and losses, including monosomy 3, characteristic of uveal and blue melanomas.