Pathology and Genomics of Pediatric Melanoma: A Critical Re-examination and New Insights

Abstract

The clinicopathologic features of pediatric melanoma are distinct from those of the adult counterpart. For example, most childhood melanomas exhibit a uniquely favorable biologic behavior, save for those arising in large/giant congenital nevi. Recent studies suggest that the characteristically favorable biologic behavior of childhood melanoma may be related to extreme telomere shortening and dysfunction in the cancer cells. Herein, we review the genomic profiles that have been defined for the different subtypes of pediatric melanoma and particularly emphasize the potential prognostic value of TERT alterations for these tumors.

INTRODUCTION

The pathologic diagnosis of pediatric melanoma is contingent upon integrating certain clinical information, is increasingly reliant on ancillary and molecular diagnostic testing, and often can only be made after consultation with individuals or institutions experienced in this difficult diagnosis.[1–4] Pediatric melanoma differs in unique ways from adult melanoma and warrants special recognition.[5,6] Further, even between prepubescent and postpubescent melanoma there are key differences with respect to those subtypes most commonly observed and their biological behaviors.[7–12] Because of these reasons, patient age is used to classify pediatric melanoma into i) congenital, ii) childhood (prepubescent; arbitrarily aged ≤10 years), and iii) adolescent (postpubescent; arbitrarily aged 11 to 19 years) subtypes.[13]

Genome-wide sequencing studies have defined the genomic profiles of these subtypes of pediatric melanoma.[14] These studies, combined with epigenetic analyses, point to dysregulation of the telomerase reverse transcriptase (TERT) gene as an intriguing explanation for the varied clinical behavior observed between childhood and adult melanoma.[15–17] The purpose of this review is to provide an update on the current knowledge in the molecular biology of pediatric melanoma. It summarizes critically important clinicopathologic factors in the diagnosis of pediatric melanoma, discusses the genomics of pediatric melanoma with an emphasis on the potential prognostic value of telomerase reactivation in pediatric melanoma, and provides an integrated overview of clinical, pathologic, and genomic aspects of specific pediatric melanoma subtypes.

CLINICOPATHOLOGIC CONSIDERATIONS

Certain clinical data and pathologic considerations are essential in evaluating pediatric melanocytic lesions. These include: age of the patient, anatomic site of the lesion, knowledge about preexisting conditions, such as presence of congenital or numerous acquired melanocytic nevi or genetic disorders, detailed information about the clinical nature and evolution of the lesion, and acute awareness of potential histologic mimics of melanoma.

Patient Age

Prepubertal and postpubertal pediatric melanoma are significantly different, with the latter sharing many biological properties with adult melanomas.[7–10] In prepubescent children, most cases of melanoma encountered are pediatric spitzoid melanomas/atypical Spitz tumors. These are considered low-grade/borderline malignant neoplasms that rarely result in clinically apparent distant metastases.[7,18,19] A second, biologically distinct form of melanoma in prepubescent children that arises in association with large/giant congenital nevi, or de novo, is highly aggressive and rapidly lethal.[14,20,21]

In postpubescent pediatric patients (adolescents), two major subtypes of melanomas are most commonly encountered. The first subtype is conventional adult melanoma, which shares similar morphologic (superficial spreading and nodular) and molecular features with adult melanoma (details in the next section).[14,17] The second subtype is spitzoid melanoma, which exhibits a heterogeneous clinical behavior and an unpredictable clinical course in adolescents.[22]

Histologic Mimics

The following histologic mimics may occur with some frequency in the pediatric population and are prone to be diagnosed erroneously as pediatric melanoma: (1) Spitz nevi and atypical variants[1], (2) proliferative nodules in congenital nevi,[23,24] (3) acquired melanocytic nevi from particular anatomic sites, such as the scalp,[25,26] genital area,[27,28] conjunctiva,[29] and acral surfaces;[30] and (4) melanocytic nevi with pagetoid melanocytosis, Reed nevus, deep penetrating nevus, epithelioid blue nevus/pigmented epithelioid melanocytoma and cellular blue nevus.[2,31]

Predisposing Conditions

Pre-existing conditions, such as congenital nevi,[32,33] familial melanoma,[34,35] an atypical nevus phenotype,[34] neurocutaneous melanosis,[36,37] xeroderma pigmentosum (DNA repair defects),[38,39] or immunosuppression [13,40,41] correlate with an increased risk for pediatric melanoma.

General Histopathologic Considerations

The histopathologic criteria used for diagnosis of pediatric melanoma are essentially the same as those used for adult cases. Melanomas can be distinguished from nevi by some combination of the following characteristics: large size (lesion diameter >6 mm and especially >1 cm), asymmetry, poorly demarcated lateral borders, ulceration, lack of maturation, confluence and high density of melanocytes in “sheet-like” arrangements, high dermal mitotic rate (>6 mitoses/mm²), mitoses in the deepest portion of the dermal component, dusty cytoplasmic melanin, and severe cytologic atypia.[1,2]

PEDIATRIC MELANOMA TUMORIGENESIS

Despite the inherently complex genomic profile of malignant melanocytes, alterations in a defined set of molecular pathways are thought to drive melanoma tumorigenesis.[42,43] These critical pathways are dysregulated stepwise, often in the following sequence for adults [44]: (1) activation of the RAS/RAF/mitogen-activated protein (MAP) kinase signaling pathway by an oncogenic driver, (2) telomerase reactivation, (3) inhibition of the p16-retinoblastoma (p16/RB) pathway, and (4) alterations of the PTEN and the phosphatidylinositol 3-kinase (PTEN-PI3K) pathway. However, the same order of events might not occur in childhood melanomas, which are biologically distinct.[15,17]

Melanoma Oncogenes and Tumor Suppressor Genes

Adult melanomas are divided into 4 mutually exclusive molecular subgroups based upon the oncogenic driver that initiates tumorigenesis: (1) activating mutations in BRAF, (2) activating mutations in the RAS family (most commonly NRAS), (3) inactivating mutations in NF1 (negative regulator of the RAS pathway), and (4) the “triple wild-type” category, encompassing melanomas lacking BRAF, N/H/KRAS, or NF1 mutations.[45,46] Such oncogenic aberrations result in constitutive activation of the MAP kinase pathway.[46,47] The PI3K-AKT pathway is also commonly dysregulated in melanomas. In BRAF-mutant melanomas, this is accomplished through alterations of PTEN (negative regulator of PI3K-AKT),[48–50] while in NRAS-mutant melanomas this pathway is directly activated through NRAS, which activates both the MAP kinase and the PI3K-AKT pathways.[51]

In pediatric melanoma, BRAF is the predominant driver of conventional adult-subtype melanoma whereas NRAS is almost unique to melanomas arising in association with large/giant congenital melanocytic nevi (CMN).[14,17] In contrast, kinase fusions are observed in 39% of spitzoid melanomas.[15,52,53]

Without additional alterations, tumor progression is restricted by replicative or oncogene-induced senescence.[43,54] For example, induction of p16^INK4a (CDKN2A) in nevi inhibits cell cycle progression via a negative interaction with cyclin-dependent kinase 4 (CDK4) and CDK6, thereby activating the RB family of tumor suppressors to induce senescence.[54] Melanoma cells may escape senescence by abrogation of the p16-RB pathway,[43] often through loss, inactivating mutation or epigenetic alteration of CDKN2A (~60%) or, less commonly, through mutation of CDK4 or other related genes.[55]

Telomerase Reactivation in Pediatric Melanoma

Somatic cells have limited replicative capacity because telomere length is progressively shortened through iterative cycles of replication,[56] which activates cell cycle checkpoints resulting in replicative senescence. Acquiring a means to maintain telomere length is therefore an essential step in the development of cancer cells, known for their capacity for uncontrolled proliferation.[57] In 85%–90% of cancer cell lines and tissues, telomere shortening is countered by aberrant expression of the TERT oncogene. TERT, which is normally silenced in somatic cells, encodes the catalytic subunit of the telomere-lengthening enzyme telomerase.[58,59]

TERT is aberrantly expressed in cancer by several possible mechanisms. Promoter mutations activating TERT transcription are seen in approximately 70% of melanomas in adults.[44,60,61] These mutations occur mainly at two mutually exclusive positions that are 124 and 146 base pairs upstream of the start codon.[60,61] In melanoma, TERT expression can also be deregulated through aberrant promoter methylation,[16,17] or by structural rearrangement of the gene; the latter occur in a manner mutually exclusive of promoter mutation.[17,62] Such lesions in TERT occur commonly in malignant adult melanomas[60–62] and biologically aggressive adolescent melanomas,[17] but are not observed in low-grade pediatric melanomas.[15–17] Thus, a combination of TERT promoter mutation, hypermethylation, or rearrangement, in conjunction with other clinicopathologic parameters, holds promise as a prognostic marker for pediatric melanomas.

TERT promoter mutations are acquired before alteration of CDKN2A in adult invasive melanoma;[44] however, this is not consistently seen in pediatric melanoma.[15,17] Biallelic loss of CDKN2A characterizes a subset of low-risk/borderline-risk spitzoid or nevoid melanomas which have not yet activated telomerase.[15,17]. It is possible that alterations in cell cycle checkpoint genes have allowed melanocytes to escape from replicative senescence and divide beyond the number of cell divisions typically allotted. Eventually, however, telomere shortening would induce telomeric crisis and cell death. This may explain the favorable outcome in otherwise morphologically malignant pediatric tumors which do not express TERT and are marked by chromosomal instability. These tumors are unable to maintain sufficiently long telomeres and therefore cannot sustain continued proliferation.

SUBTYPES OF PEDIATRIC MELANOMA

Conventional Adult Subtype Melanoma

Approximately 40%–50% of pediatric melanomas are histologically similar to adult melanomas.[2,4] For reasons that are unknown, this conventional adult subtype of melanoma rarely occurs in prepubescent children. In our experience, adult melanoma type can rarely occur in peri-adolescence, in patients as young as 10 to 11 years of age, even in the absence of pre-existing conditions (Figure 1). These patients are typically Caucasians with fair skin, poor tanning ability, and numerous nevi.[63,64] More than two-thirds of pediatric patients with adult conventional subtype melanoma harbor variants of Melanocortin 1 receptor (MC1R), which is associated with fair, sun-sensitive skin and an increased susceptibility to melanoma.[65] Germline mutations in the melanoma susceptibility gene CDKN2A, which is responsible for 25%–50% of familial melanomas, are seen only in a small subset of sporadic early onset melanoma cases (1.6%).[14,35,41,66]

Figure 1.

Photomicrographs of hematoxylin and eosin-stained (H&E) sections and immunohistochemical stains for p16 and BRAF V600E in an adult subtype melanoma from a peri-adolescent patient.

Malignant melanoma, nodular type (A; H&E, scanning magnification), composed of confluent nests of epithelioid malignant melanocytes (B, H&E 20×). Immunohistochemical analysis shows loss of p16 expression (C) and positive staining for BRAF V600E (D) in the melanoma cells. The melanoma harbored a BRAF^V600E and a TERT promoter mutation.

As seen in adult “superficial spreading” melanomas, these melanomas often develop via a radial or horizontal growth phase, which is initially intra-epidermal (in situ) or microinvasive melanoma. Metastatic disease is unlikely to occur until the tumor enters the vertical growth phase. Epithelioid melanoma cells are characteristically disposed in a pagetoid array and, given time, usually invade the papillary dermis. Although adult melanomas can be diagnosed by prominent pagetoid melanocytosis, a subset of nevi in children may display this feature thus diminishing its clinical utility. “Nodular” melanomas are defined by an almost exclusively bulky or cohesive invasive component and little or no associated lateral intraepidermal component (Figure 1). Melanomas of acral (glabrous) skin are rare in childhood.[2,4,67] Lentiginous melanocytic proliferation and upward migration of melanocytes are common in childhood nevi, particularly in glabrous skin. These changes must not be over-interpreted unless architectural disorder is prominent and cytological abnormalities occur throughout the breadth of the lesion. In one European study, only three of 145 Caucasian children with melanomas had involvement of the soles.[4] In general, lentigo malignant melanomas, typically seen in chronically sun damaged skin, do not occur in childhood except in xeroderma pigmentosum patients.[1,4]

Pediatric conventional adult subtype melanomas are genetically also very similar to adult melanomas.[14] Whole-genome and -exome sequencing of 15 pediatric conventional adult subtype melanomas showed that they were similar to adult tumors by (1) having an ultraviolet light-induced mutational burden with an abundance of C>T transitions at dipyrimidine sites, (2) often harboring activating mutations in the BRAF and the TERT promoter, and (3) frequently harboring inactivating mutations in CDKN2A and PTEN tumor suppressor genes upon BRAF activation.[14,68] In a follow-up study of adolescent melanomas, other mechanisms of TERT dysregulation (i.e., promoter hypermethylation and structural rearrangement) were found in a subset of tumors.[17]

Activated BRAF drives 85% or more of pediatric conventional melanomas and the presence of this signature should raise a strong suspicion of this subtype (Figure 1),[14,17] although we have observed BRAF mutations in some low-risk nevoid melanomas and rarely in pediatric spitzoid melanomas.[15,17] As the natural history of disease for pediatric conventional and adult melanomas is similar,[21,39,69–71] their management should likely be the same. The classification and stage grouping criteria, based on tumor thickness, mitotic rate, ulceration, and nodal metastases, according to the American Joint Committee on Cancer staging system used for adult cutaneous melanoma, are also applied to the pediatric counterpart.[72]

Spitzoid Melanoma

Spitzoid tumors including spitzoid melanoma are a distinct histologic variant of melanocytic neoplasms that are more commonly seen in individuals younger than 20 years. Despite a high incidence (~50%) of sentinel lymph node metastases in spitzoid neoplasms, these patients often have excellent outcomes and rarely develop distant disease.[15,22,73]

Reliable classification of spitzoid lesions such as Spitz nevus, atypical Spitz tumor, or spitzoid melanoma remains a challenge, especially in the pediatric population.[1,2,74,75] The principal question of what constitutes a “spitzoid” melanoma[75,76] can be addressed by clinical, histopathologic, and genetic criteria. Clinically, pediatric spitzoid melanoma is commonly amelanotic and presents as a pink, red papule, nodule, or polyp that clinically resembles a benign non-melanocytic lesion, such as pyogenic granuloma or wart.[6,10,77,78] Thus, the conventional ABCDE (asymmetry, border irregularity, color variation, diameter >6 mm, and evolution) criteria recommended as a visual guide for the detection of melanoma in adults may not reliably capture this melanoma subtype.[78]

Histopathologic features of spitzoid melanoma are first of all cytologic – the presence of prototypic enlarged epithelioid and/or spindle cells – but also architectural, such as the silhouette of the lesion often integrating a well-defined, dome-shaped configuration with epidermal hyperplasia, vertical disposition of melanocytes, and a wedge-shaped dermal architecture (Figure 2)[1,2,74–76,79].

Figure 2.

Photomicrographs of H&E stained sections and interphase fluorescence in situ hybridization (FISH) with RET and ROS1 dual color, break-apart probe from patients with pediatric spitzoid melanocytic tumors.

A-C: Pediatric spitzoid melanoma with predominantly spindle cell features (A, H&E 4×; B, H&E 40×) harboring RET fusion. FISH image (C) shows split red and green signals, consistent with RET rearrangement.

D-F: Pediatric spitzoid melanocytic tumor with epithelioid and spindle cell features (D, H&E 4×; E, H&E 20×) harboring ROS1 fusion. FISH image (F) shows separate red and green signals in several nuclei, consistent with ROS1 rearrangement.

An often neglected yet critical point is that not all melanocytic neoplasms with spindle/epithelioid cell features are spitzoid. Increasingly, genetic criteria can aid in the classification of spitzoid melanomas. For example, specific kinase gene fusions often occur in such melanomas but not in other spindle and epithelioid cell melanocytic neoplasms or conventional melanomas. [52,53] These events are found in approximately 50% of all spitzoid tumors (from nevus to melanoma), and involve the receptor tyrosine kinases NTRK1, NTRK3, ALK, ROS1, RET, or MET, or the serine-threonine kinase BRAF (Figure 2).[52,53,80,81] As these kinase fusions are present across the entire spitzoid spectrum, they are unlikely to determine the biology of these tumors.[15,52] Activating mutations in HRAS or loss/inactivation of BAP1, which occurs in a subset of Spitz nevus and atypical Spitz tumors, is mutually exclusive from kinase fusions.[82] Activating BRAF mutations are uncommon in spitzoid tumors, but occur in a subset of epithelioid spitzoid melanocytic tumors with BAP1 loss.[83]

Another critical question is how spitzoid melanoma can be distinguished from atypical Spitz tumors. Essential histopathologic criteria for melanoma have been elaborated in an earlier section. In challenging cases, genetic criteria can aide in the evaluation of these neoplasms. A normal chromosomal copy number status or some changes limited to an isolated alteration (e.g., extra copies of 11p or gains involving 7q) point to the diagnosis of Spitz nevus or atypical Spitz tumor, respectively. On the other hand, multiple copy number changes as seen in conventional melanoma suggest the diagnosis of spitzoid melanoma.[84,85] These changes can be detected by fluorescence in situ hybridization (FISH) or comparative genomic hybridization (CGH). Even if the diagnosis is spitzoid melanoma, the majority of childhood tumors, at least in short-term follow-up, never develop clinically detectable metastasis beyond local lymph nodes. In contrast, rare spitzoid neoplasms may result in distant metastases and death.[15,22,86] Thus, the crucial question of how the biological potential of a particular spitzoid lesion can be predicted remains unresolved.

Recent, preliminary studies suggest that the presence of TERT alterations may be useful in predicting clinical behavior.[15] In a series of 56 pediatric and adult patients (median age, 9 years) with atypical Spitz tumors or spitzoid melanomas, TERT promoter mutations were only detected in tumors from patients who developed visceral metastasis (two adolescents, ages 11 and 14; two adults).[15] The presence of these mutations correlated significantly with high risk parameters described by Barnhill et al for risk stratification of spitzoid lesions: Age ≥10 years, mitotic rate (i.e., >5 per mm²) and ulceration.[15,87].

One must be wary that the presence of promoter mutation alone cannot be used to predict elevated TERT mRNA levels, since other mechanisms of TERT upregulation are also possible in cancer cells. TERT mRNA quantification in biopsy samples has great prognostic potential but unfortunately is of limited practical application because such samples are often contaminated with stromal or non-tumoral tissue or have degraded RNA in old archival material. Thus, detecting any one of the three DNA-level TERT alterations mentioned earlier may be useful as a substitute for detection of elevated TERT mRNA levels.

The prognostic parameters used for staging in adult melanomas, such as Breslow thickness and nodal metastases, do not carry the same weight in predicting outcome in pediatric spitzoid melanoma/atypical Spitz tumor as in conventional melanoma.[7,19,72,73] In fact, because of their usually excellent prognosis, the question remains whether childhood spitzoid melanomas merit the designation of melanoma. It is true that the presence of a malignant process in these tumors is supported by the genomic instability suggested by frequent observations of multiple copy number changes that are detected by CGH or FISH, and the frequent biallelic loss/inactivation of CDKN2A, involved in replicative and oncogene-induced senescence. However, in the absence of TERT dysregulation, continuous cellular propagation in these cases eventually leads to extreme telomere shortening and eventually telomeric crisis,[88,89] which is characterized by end-to-end chromosomal fusions, chromosome breakage, and anaphase bridges. These cellular events may explain the changes in gene copy number and the morphologic features of malignancy commonly observed in such neoplasms.[88,89] Although replicative lifespan is limited in these non-immortal malignant cells, there still remains the potential to acquire the capacity to maintain telomere lengths, either through TERT expression or through a telomerase-independent Alternative Lengthening of Telomeres mechanism. It is well-known that clinically localized melanomas can recur with distant metastasis several years or even decades after removal of the primary tumor.[90] Although the incidence of such further progression is likely confined to a rare subset of cases, given the typically excellent outcomes of patients with pediatric atypical Spitz tumors/spitzoid melanomas, anecdotal reports of late recurrence[91] suggest that long-term follow-up is required to appreciate the full potential of this neoplastic process.

Melanoma Arising in Large/Giant Congenital Melanocytic Nevi

CMN are defined as pigmented lesions being present at birth. Large (predicted adulthood diameter, >20–40 cm) or giant (predicted adulthood diameter >40 cm or covering ≥5% of the skin surface) CMN carry an approximately 6% lifetime risk of malignant transformation.[32,33,92,93] More than half of the melanoma cases that are associated with these lesions occur before the age of 10 years.[2,32,33,92] Melanomas arising in large or giant CMN represent a highly aggressive subtype that is often associated with rapid progression and a fatal outcome.[14,20,94]

These are commonly of the small-cell type of melanoma, which can lead them to be misconstrued as a benign lesion. They are generally dermal and composed of epithelioid cells, spindle cells, or small round cells resembling a malignant small round “blue” cell tumor (Figure 3).[1,2,4] Cells can take on the appearance of lymphoblastic lymphoma or comparable “blastic” tumors.[1,2] Mitotic figures and nuclear debris are also commonly observed.

Figure 3.

Photomicrographs of H&E sections and TERT mRNA in situ hybridization (ISH) in a melanoma and in a proliferative nodule arising in association with giant congenital nevi.

A-C: Melanoma consisting of malignant small round cells in the dermis within a giant congenital nevus (A, H&E 4×; B, H&E 40×). TERT mRNA ISH shows numerous punctate intracellular signals (C), consistent with high TERT mRNA levels.

D-F: Proliferative nodule arising within a giant congenital nevus in deep soft tissue, characterized by cellular features, mitotic activity (D, H&E 2×; E, H&E 40×), and no significant levels of TERT mRNA detected by ISH (C).

At the molecular level, these melanomas often harbor an activating mutation in NRAS at codon 61, acquired at the precursor lesion stage.[14,20,95] Alterations in PTEN are uncommon and TERT promoter mutations are not observed, at least in children.[14,16] Instead, TERT upregulation may be mediated through aberrant promoter methylation.[16,96]

It is important to realize that most nodular proliferations developing in large or giant CMN, particularly in younger children, are in fact benign.[1,2,4] However, it may be extremely difficult to histopathologically distinguish melanoma from benign proliferative nodules developing in CMN (Figure 3). Ulcerated proliferative nodules in neonates, often on the scalp, composed of small cells with a high mitotic activity, can be especially challenging to distinguish.[97] In rare instances, large/giant CMN are associated with contiguous involvement of lymph nodes that must be distinguished from nodal metastasis. These lesions can be distinguished from melanoma by CGH and TERT promoter methylation analysis. CGH analysis has revealed gains and losses of whole chromosomes in proliferative nodules, in contrast to the regional copy number variation that is observed in melanoma.[97,98] Unlike melanomas arising in CMN, which express high levels of TERT mRNA, proliferative nodules do not express TERT mRNA at detectable levels (Figure 3).[16] These melanomas are characterized by aberrant TERT promoter methylation, which can be detected by high-throughput bisulfite methylation sequencing or combined bisulfite restriction enzyme analysis.[96] Such epigenetic alterations are absent in proliferative nodules.

Congenital melanoma

Melanoma at birth or in early life is rare. There are three types of congenital melanoma.[1,2] The first type is maternal melanoma metastatic to the fetus.[99] In this case, neonates usually have widespread visceral metastases at birth and die within days to months. The placenta usually shows gross and histopathological evidence of metastatic melanoma. Because of the theoretical risk of transmission from mother to fetus, placentas should be examined for gross and histopathological evidence of metastatic disease with maternal history of invasive melanoma. When present, placental metastases rarely lead to congenital melanoma; some placental deposits also appear benign and may represent developmental arrests or “benign seeding” of the placenta.[100] Nonetheless, the finding of placental involvement by melanocytic deposits warrants careful examination and follow-up of the newborn.

The second type of congenital melanoma is primary congenital melanoma, which may develop de novo or in association with a congenital nevus.[1,2] In our experience, congenital melanoma has a predilection for the scalp.[1] The third type of congenital melanoma is that which develops in large/giant CMN with or without neurocutaneous melanosis; this type likely constitutes the largest group of congenital melanoma. These melanomas are predominately dermal but might exhibit lentiginous and junctional nested arrangements of melanoma cells.

Leptomeningeal Melanoma Associated with Neurocutaneous Melanosis

Leptomeningeal melanoma can arise in approximately two thirds of patients with neurocutaneous melanosis.[36,37] They are intracranial in approximately 50% of patients and most commonly involve the frontal and temporal lobes. They are rapidly fatal in almost all patients (median age of death 3 years).

Summary

Identifying molecular markers which predict aggressive behavior of pediatric melanomas is critical for risk stratification and choosing appropriate treatment strategies. This is particularly true in view of the diversity of disease observed in this age group. Recent preliminary evidence suggests that alterations in TERT and aberrant telomere lengths might be particularly useful as prognostic markers in pediatric melanomas. These potential markers need to be validated in large patient cohorts, using long-term follow-up data, before their clinical implementation.

Abbreviations Key

TERT

telomerase reverse transcriptase

FISH

fluorescence in situ hybridization

CGH

comparative genomic hybridization

CMN

congenital melanocytic nevi

PI3K

phosphatidylinositol 3-kinase