Pathology of Melanoma: Interpretation and New Concepts

Natural history of melanoma: histologic aspects

Although some controversy exists about how melanoma develops and evolves, a pathway proposed by Ackerman is considered by many to be the most valid. Accordingly, an oncogenic stimulus such as ultraviolet irradiation affects melanocytes in the epidermis at the dermoepidermal junction (DEJ) resulting in neoplastic transformation. These altered melanocytes begin to proliferate initially as solitary units at the DEJ and manifest histologically as scattered, single hyperchromatic cells with a cleft or halo surrounding them. Clinically, there may be only a very light tan macule or it may be undetectable ( Fig. 27.1 ).

Over time, these melanocytes become more numerous and coalesce forming small nests confined to the DEJ. There may also be involvement of adnexal structures such as acrotrichia and acrosyringia. At this point, there is often a tan to brown macule ranging from 3 to 4 mm in diameter but it is usually not significantly atypical ( Fig. 27.2 ). As the lesion progresses, more nests of atypical melanocytes develop at the DEJ and solitary atypical melanocytes ascend to the upper portions of the epidermis. In time, nests also extend above the DEJ. Clinically, such lesions may appear darker brown and shades of black may be seen which correlates with the presence of melanin in the upper levels of the epidermis and cornified layer ( Fig. 27.3 ). Some of the architectural features of melanoma may begin to be visible at this time, namely asymmetry and irregular borders.

As the lesion progresses, there is more involvement of the epidermis with nests and single atypical melanocytes throughout and involving the adnexal structures. Neoplastic melanocytes also enter the papillary dermis, initially in small foci, but over time it becomes more extensively involved, filled and expanded ( Figs 27.4 and 27.5 ). At this stage, lesions are usually greater than 6 mm in diameter and demonstrate the ‘ABCDE’ clinical features of melanoma in most cases. Most lesions at this stage are still macular or only slightly elevated although skin markings are often obliterated.

Subsequently, lesions may progress to involve deeper structures such as the reticular dermis and subcutaneous fat. This may be accompanied by additional clinical features including nodule formation, ulceration and colors such as red, blue and white, the latter associated with regression. These features signify more advanced lesions and worse prognosis. Finally, metastasis may develop. Usually, regional lymph nodes are involved first although local ‘satellite’ cutaneous metastasis may be the initial manifestation of metastatic disease.

Different clinical and histologic manifestations are seen depending on which sites are involved. In cutaneous metastases of melanoma, histologically there are aggregates of atypical neoplastic melanocytes in the dermis usually without contiguity with the epidermis ( Fig. 27.6 ). There may also be involvement of the superficial papillary dermis in so-called ‘epidermotropic’ metastases. Blood vessels and lymphatics also often contain neoplastic cells and there may be involvement of nerves. Clinically, these may appear as intracutaneous or subcutaneous nodules that range in color from that of normal skin to jet black. When lymph nodes are involved, clinically there are usually one or more firm nodules that develop in a lymph node chain or group that may be fixed to underlying structures. Histologically, there is usually replacement of the lymph node architecture by atypical neoplastic melanocytes.

Although this pathway describes melanoma that develops de novo, approximately 25% of melanomas develop in association with nevi ( Fig. 27.7 ). This has led some investigators to propose a pathway that describes ‘precursor’ lesions. However, melanoma may arise in virtually any melanocytic nevus, especially giant congenital nevi, so it remains important for clinicians to evaluate all nevi carefully for changes that may signal the development of melanoma within them.

Clark et al. described a model of progression of melanoma where a ‘radial’ growth phase is followed by a ‘vertical’ growth phase. The radial growth phase is characterized by malignant melanocytes proliferating in the epidermis and papillary dermis at which time the cells purportedly do not have the ability to metastasize. These same cells were shown to have limited capacity for growth in cell culture. Melanomas in this phase have been considered to be biologically less aggressive than those of the so-called ‘vertical growth phase’ in which there is more extensive involvement of the dermis. Melanocytes from lesions in this phase have the capacity for immortality in cell culture. However, there are significant exceptions to this model, including nodular melanomas that have a very short or no radial growth phase and metastasizing thin melanomas without a vertical growth phase.

Recent studies using comparative genomic hybridization have demonstrated that cells from acrolentiginous melanoma demonstrate genotypic features of metastatic melanoma while confined to the epidermis. In addition, it is not clear whether some ‘radial’ growth phase melanoma cells that were cultured were nevus cells. Finally, the determination of when the ‘radial’ growth phase ends and the ‘vertical’ growth phase begins is a subjective determination based on evaluation of histologic sections which can be prone to sampling and interpretation errors. As such, the assignment of radial and vertical growth phases for melanoma has fallen into disfavor.

Molecular studies suggest that some ‘dysplastic’ nevi may represent intermediate lesions in a multi-step melanoma tumorigenesis pathway. Alterations of some tumor suppressor genes, oncogenes, mismatch repair proteins, extracellular matrix proteins, and growth factors are common to both lesions. However, such work remains controversial and speculative and most workers still consider ‘dysplastic’ nevi to be markers for the development of melanoma in some patients rather than true precursors.

Appropriate biopsy technique

Appropriate clinical examination, lesion selection and the resulting method of choice for sampling are important steps for accurate pathological diagnosis (see Chapter 40 ).

Examination of a patient with a pigmented lesion should include a full skin evaluation and determination of family and personal risk factors for melanoma. Patients with many nevi, especially dysplastic nevi, present a particular clinical challenge in that they may have numerous lesions that fulfill the ‘ABCDE’ criteria for the clinical diagnosis of melanoma. These patients present with the ‘signature’ nevus phenomenon in which a given individual develops multiple similarly appearing nevi that display a biologically benign phenotype even though they may appear clinically atypical. In this subset of patients, it is important to recognize lesions that appear significantly different and atypical and perform a biopsy of such lesions.

The American Academy of Dermatology Guidelines/Outcomes Committee for Cutaneous Melanoma and the National Comprehensive Cancer Network have published information outlining initial management of melanocytic lesions suspected of being melanoma. Classic teaching is that excision of the entire lesion should be performed. However, there are a number of problems with this recommendation. First, many lesions are located in cosmetically sensitive areas and primary excision may be mutilating or impossible. Second, and perhaps more important, there are many lesions that may simulate melanoma clinically (such as pigmented basal cell carcinoma, squamous cell carcinoma and seborrheic keratosis among others). Given the large number of such lesions, excision of all lesions that could theoretically represent melanoma would be impossible. There are no good studies to quantify what qualifies a lesion to be suspected as, but not diagnostic of, melanoma, by a clinician. We recommend that lesions with a greater than 50% chance of being melanoma based on clinical and/or dermoscopic examination be considered for primary excision where possible. However, in the other 50%, other time-honored accepted biopsy techniques can be utilized. These include saucerization or deep shave, broad punch and incisional techniques. The choice of an incorrect or inappropriate technique can compromise microscopic evaluation for diagnosis.

Optimizing cosmetic outcome with a superficial shave or small-diameter punch biopsy is not recommended when suspicion of melanoma is high. These techniques may fail to sample diagnostic areas, especially in clinically suspected melanomas. Many of the histologic criteria of melanoma are architectural, such as breadth, symmetry and circumscription, and therefore biopsies should be of sufficient breadth and depth to allow for assessment of these features.

A full-thickness incisional biopsy is an accepted form of sampling for a lesion that would normally be a candidate for primary excision yet is on an anatomic site that would prove difficult for excision or is a large lesion. This method involves performing an ellipse within the lesion that attempts to sample a significant portion of it, especially encompassing areas that are elevated or black. Most of these provide a sample that is diagnostic, although clinicians should realize that a subsequent re-biopsy may be required if the sample is not adequate for either diagnosis.

Punch biopsies represent either an excision, if the entire lesion is removed with the punch, or a form of an incision, if only a part of the lesion is removed. While a punch excision is obviously an excellent biopsy technique, punches taken of broad lesions may be prone to sampling error and should be performed with caution especially when less than 5–6 mm in diameter ( Fig. 27.8 ). Multiple small punches may also not be optimal and may be associated with sampling error.

Some have postulated that partial incisional or shave biopsies of intact primary melanomas may increase the rate of sentinel lymph node (SLN) micrometastases. Martin et al. evaluated 2164 patients with melanoma and concluded that biopsy technique did not affect incidence of SLN metastasis, locoregional recurrence, or overall survival for patients who underwent excisional versus incisional versus shave biopsy.

Since the most important prognostic factor for melanoma is thickness (Breslow depth) of the lesion, a non-representative biopsy could potentially yield false information on which treatment planning is based. A retrospective review of 145 cutaneous melanomas demonstrated that initial diagnostic biopsies performed using non-excisional saucerization or punch technique provided specimens that allowed for 88% accuracy of assessment of Breslow depth when compared with the Breslow depth determined on the re-excision specimen. Saucerization biopsies were more accurate than punch biopsies less than 5 mm in diameter for melanomas less than 1 mm thick. Therefore, saucerization biopsy is preferable to superficial shave or punch biopsy for primary cutaneous melanoma when an initial sample is taken for diagnosis.

Additionally, Hsu and Cockerell reviewed 1123 histologically proven cutaneous melanomas and found significant diagnostic discrepancy between initial punch biopsies and re-excision specimens. While excisional biopsy and saucerization shave biopsy demonstrated near 100% accuracy, punch technique was only 86.5% accurate. Due to the inherent intralesional heterogeneity of cutaneous melanoma, small punch biopsies may not always obtain representative specimens and may subject the patient to a significant risk of misdiagnosis.

Histopathologic criteria for melanoma

Currently the histologic diagnosis of melanoma is based on a constellation of cytological and architectural microscopic features referred to as ‘criteria’ for diagnosis. Histological criteria represent a constellation of findings observed microscopically that are highly specific and sensitive for the diagnosis of melanoma. Other histological findings that may be observed in melanoma are referred to as ‘clues’ to diagnosis and are used along with criteria to establish a definitive diagnosis.

A ‘unifying concept’ regarding the histologic classification of melanomas based on both architectural and cytologic features has been proposed by Ackerman. The three cardinal architectural characteristics of melanoma include large dimensions of the lesion (diameter >6 mm), asymmetry, and poor circumscription. While a valuable criterion, asymmetry may be difficult to appreciate in early lesions. Other features include predominance of single melanocytes over nests of melanocytes in the epidermis, extensive involvement of adnexal epithelium by atypical melanocytes, presence of atypical melanocytes throughout the entire thickness of the epidermis (pagetoid spread), and irregularity in size and shape of nests of melanocytes ( Fig. 27.9 ). Variability in the size and spacing between nests of melanocytes, patchy inflammation, and irregular pigment distribution throughout a lesion are other findings that may be present. Pagetoid spread is a highly sensitive histological finding of melanoma, especially when observed in the setting of other criteria or extensive solar elastosis. Characteristic findings of melanoma in the dermis include coalescence of nests of melanocytes into sheets and lack of maturation, which refers to failure of cells to become smaller with increasing depth.

Cytologically, melanocytes in melanoma may assume a number of different morphologies and may be small round, large round, pagetoid, balloon, oval, spindle, multinucleate or dendritic. There may be variable cytologic atypia ranging from virtually none in some melanoma in situ, some desmoplastic melanomas (DMM) and early acrolentiginous or mucosal melanoma, to frank anaplasia. When atypical, there is cellular pleomorphism with variability in cell size and shape, prominent chromatin, large nucleoli and reversal of the nuclear:cytoplasmic ratio. There may be increased mitotic activity although absence of mitotic activity may also be observed. Mitotic figures, especially if atypical, numerous and in deeper portions of a lesion, is a sign of melanoma with rare exceptions. Other features include lack of maturation and the presence of necrotic melanocytes. Necrosis en masse of large areas of a melanocytic lesion also is virtually diagnostic of melanoma.

Pathologic features of different clinical forms of cutaneous melanoma

Melanoma in situ

Melanoma in situ is a term used to define melanoma that is confined to the epidermis. It may be applied to any of the classically described subtypes of melanoma. It is a useful term in that it conveys information about prognosis by its very nature, namely that it is curable following appropriate early surgical therapy. The most common subtype, lentigo maligna, is present on sun-damaged skin of older individuals.

Melanoma in situ is characterized histologically by an increased number of atypical melanocytes arranged singly and in nests distributed irregularly at the DEJ and above it ( Fig. 27.10 ). They involve a broad area of the epidermis, and in contrast to junctional nevi and lentigines, the epidermis is often thin and there may be effacement of epidermal retia. Atypical melanocytes are often present in cutaneous adnexal structures. Solar elastosis is a common finding and is an important clue to the diagnosis of melanoma in situ in the setting of melanocytic hyperplasia. There may be a lichenoid infiltrate of lymphocytes in the superficial papillary dermis that may obscure the melanocytic proliferation in the epidermis and impart features similar to a benign lichenoid keratosis, which is a histologic pitfall that may lead to misdiagnosis.

The histologic evaluation of margin status following excision of these lesions may be difficult because of diffuse melanocytic proliferation that is present in the background skin of these patients as a consequence of longstanding sun damage. In contrast to persistent melanoma, the number of melanocytes per unit area is less and nests of melanocytes are not seen. In some cases, however, the distinction may be impossible without clinical correlation.

Nodular melanoma

This variant of melanoma refers to lesions with a predominant dermal component, often with a relatively minimal intraepidermal component. In some cases, the dermal component extends laterally further than the epidermal component ( Fig. 27.11 ). While some individuals consider this to represent a true variant, others consider it as a more advanced form of melanoma of any subtype which has become nodular as a consequence of evolution. As virtually all melanomas progress through an in situ stage, if left untreated, they may progress to involve the dermis and may form papules or nodules clinically. There is one form of nodular melanoma known as ‘early’ nodular melanoma, however, that involves the dermis at a relatively early point in time in contrast to others that tend to remain confined to the epidermis for a significant part of their evolution.

Acrolentiginous melanoma

These lesions tend to have a greater number of dendritic melanocytes with dendrites extending to the upper levels of the epidermis on volar skin. Initially, the degree of cytologic atypia may be minimal. Later, melanocytes may become more spindle or pagetoid in appearance with prominent pleomorphism and subsequently involve the deeper dermis ( Fig. 27.12 ).

Desmoplastic melanoma

Desmoplastic melanoma (DMM) represents less than 1% of all variants of melanoma. It usually presents as a slightly pigmented or skin-colored to pink indurated papule, nodule or plaque on sun-exposed skin ( Fig. 27.13 ). Typically they affect the head and neck region, appear later in adult life and are more commonly seen in men. Up to 70% of cases are misdiagnosed clinically, often described as ‘fibroma’, ‘squamous cell carcinoma’ or ‘scar’. As a result of its banal histologic appearance and its clinically non-descript morphology, it is often misdiagnosed both clinically and histologically and is a common source of litigation.

There are several histologic variants of DMM shown and described in Figure 27.14 . All tend to be large, poorly demarcated neoplasms involving the full thickness of the dermis, sometimes with a slightly raised central nodule. Most lesions have a component of melanoma in situ in the overlying epidermis but this may be absent in 20% of cases ( Fig. 27.15 ). They can arise in association with a pre-existing melanoma of another type, often lentigo maligna, although they may arise de novo. Perineural involvement is evident in up to 35% and is responsible for recurrence and spread along nerves ( Fig. 27.16 ). Cytologically, neoplastic cells consist of spindle-shaped melanocytes with variable pleomorphism and hyperchromatic nuclei although there may be minimal to absent atypia. Neoplastic cells may appear small and monomorphous with scant cytoplasm and have a low mitotic index (<1/mm ² ). Two common histologic clues used to assist in the diagnosis of DMM are the presence of pigment within cells and the presence of nodular aggregates of lymphocytes scattered throughout the dermis ( Fig. 27.17 ).

Traditionally, DMM have been thought of as having a better prognosis when compared to other types of melanoma of equivalent Breslow depth. Several reports have contradicted this finding, suggesting that case-matched control patients with melanoma matched for tumor thickness have similar survival rates to patients with DMM. In light of these findings, several authors have proposed that DMM with prominent fibrosis (pure subtype) are unlikely to disseminate to regional lymph nodes and are associated with a favorable outcome when compared with those with mixed desmoplasia or other melanoma variants.

Immunohistochemistry ( Table 27.1 ) is often necessary to distinguish DMM from other entities included in the histologic differential diagnosis as the spindle cells may be poorly differentiated. S100 protein is the most useful stain as it is almost uniformly positive although occasionally only weakly. Stains for MART-1 and HMB-45 antigen are negative in most DMM although a few reports have demonstrated focal positive staining.

Table 27.1

Commonly Used Immunohistochemical Markers for Melanoma

Marker	Cell Type/Tumor Recognized (Staining)	End Protein Product/Recognized Protein/Epitope	Staining Pattern	Note
S100 protein	Melanocytes, cells derived from neural crest cells, and tumors derived from: Langerhans cells, nerve sheath cells, adipocytes, antigen-presenting dendritic cells	21 kDa acidic calcium binding protein	Nuclear and cytoplasmic	DMM = usually positive
HMB-45	Melanocytes, clear cell sarcoma of soft parts, angiomyolipomas, lymphangioleiomyomatosis, sweat gland tumors, renal cell carcinoma with t(6,11)(p21;q12)	Glycoprotein gp100	Cytoplasmic	DMM = typically negative
MART-1/MelanA	Melanocytes, angiomyolipomas, lymphangioleiomyomatosis, adrenal cortical tumors, gonadal steroid tumors	MART-1: Melanoma antigen recognized by T cells (melanA represents the gene)	Cytoplasmic and/or nuclear	DMM = typically negative
Tyrosinase	Melanocytes, some angiomyolipomas, pigmented nerve sheath tumors	Enzyme in melanin synthesis	Cytoplasmic	DMM = typically negative
MITF	Melanocytes, reports of staining histiocytes, fibroblasts, lymphocytes and mast cells. Spindle cell tumors, angiomyolipomas, rarely breast and renal cancer	Microphthalmia-associated transcription factor	Nuclear	DMM = typically negative
NKIC3 (CD63)	Melanocytes, histiocytes, eosinophils, dendritic cells, mast cells, endothelial cells, cellular neurothekeomas, medullary thyroid cancer, granular cell tumor, neuroendocrine tumors, neurofibromas, schwannomas. Reports of staining in several internal malignancies including bladder and colorectal	25–110 kD glycoprotein	Peripheral cytoplasmic membrane staining	Cellular neurothekeomas tend to be NKI/C3 positive, and S100, HMB-45, MART-1 negative

DMM = desmoplastic melanoma

Sensitivity/specificity of melanocyte markers

Sensitivity:

S100 protein (97–00%) > HMB-45 (69–93%) = MART-1 (75–92%) = Tyrosinase (84–94%) = MITF (81–100%)

Specificity:

HMB-45 (95–100%) = Mart-1 (88–100%) = Tyrosinase (94–100%) > S100 protein (75–87%)

Verrucous melanoma

Verrucous melanoma is a variant described as a hyperkeratotic pigmented lesion, usually on an extremity (71%). Clinically, they are usually misdiagnosed as benign lesions such as seborrheic keratosis, verruca, nevus and Spitz nevus.

This lesion is often considered a variant of superficial spreading melanoma and is characterized by having an exophytic papillomatous growth pattern ( Fig. 27.18 ). Pseudo-epitheliomatous hyperplasia and overlying hyperkeratosis are prominent features which may obscure the underlying melanoma cells. Blessing et al. reviewed 20 cases and found 10% had been given erroneous diagnosis; however, eight patients suffered metastases and seven died of their disease. The Breslow depth and Clark's level may be difficult to determine given the papillomatous architecture. Kuehnl-Petzoldt et al. reported that reliable Clark's levels could be assigned in only two-thirds of these neoplasms. It is important to recognize this variant as it may easily be confused with squamous cell carcinoma or other benign epithelial proliferations.