Main

Spitz nevus is an acquired melanocytic lesion, which generally develops in children and adolescents, although it is not rare in adults. It is classically believed to be benign, and conservatively managed. However, since its description in 1948 by Sophie Spitz,1 its diagnosis has been a cause for concern because it is not always possible to render an unequivocal diagnosis of Spitz nevus vs melanoma.

Although certain features are now systematically evaluated to distinguish between Spitz nevi and melanomas, diverse experts may vary in defining the relative importance of each of the following criterion: size, shape, circumscription of junctional nests, upward spread in the epidermis, extention into the subcutis, eosinophilic globules, mitotic figures and their superficial or deep location within the tumor, giant cells, presence of ulceration and stromal alterations.2, 3, 4, 5 Furthermore, most lesions do not show many of the conventional criteria that are used routinely for the histopathologic diagnosis of the tumor, thus emphasizing the lack of objective criteria for predicting the biologic behavior of such lesions.3, 6, 7 The classification of some cases as metastasizing Spitz nevi1 illustrates the difficulty of accurately distinguishing some Spitz nevi from melanoma based solely on histological criteria. Thus, the diagnostic term ‘atypical Spitz tumor’ is used to describe lesions that deviate from the typical appearance of Spitz nevi and have an uncertain biological significance.7

Given that the diagnosis of a subset of lesions remains challenging, the purpose of this study was to evaluate the expression of a wide panel of markers using routine immunohistochemical methods to provide a universally available set of proteins to distinguish between Spitz nevus and melanoma. We compared the protein expression signatures in both series of typical Spitz nevi and vertical growth phase melanomas in an attempt to establish a powerful tool allowing the prediction of biological behavior of the melanocytic lesions. The results provide a set of nine markers differentially expressed in Spitz nevi vs melanomas useful in differential diagnostic of both melanocytic lesions.

Materials and methods

Tissue Samples and Tissue Arrays Construction

We studied a total of 90 formalin-fixed paraffin-embedded samples from the files of the Pathology Department, 12 de Octubre University Hospital, Madrid, collected from 1995 to 2005. Twenty eight corresponded to typical Spitz nevi, concordant with the features of nests of spindle and/or epithelioid melanocytes, size <7 mm, maturation, symmetry, no pagetoid spread and no mitosis or pushing margins at the base,3 and 62 primary vertical growth phase non-spitzoid melanomas. Vertical growth phase was defined after pre-established criteria:8 all tumor cell nests in the epidermis being smaller than at least one of the dermis, the presence of dermal tumor cell nests of 25 or more cells in width and mitotic activity in intradermal neoplastic cells. Melanoma patients were treated after standard criteria: surgical resection with adequate margins. Adjuvant therapies with interferon or chemotherapy for metastatic disease were used when clinically indicated.

Representative sections stained with hematoxylin–eosin were evaluated for histological features by the authors, and blocks were selected on the basis of suitable tissue (at least 1-mm thick)9 to construct the tissue arrays.

A tissue microarray containing Spitz nevi and two more with VGP melanomas were assembled as described earlier,9 including at least two 1.5 mm-diameter cylinders of highly cellular areas from the primary paraffin blocks. In addition, normal tissues (mostly reactive lymphatic nodes, tonsil and skin) were placed in the arrays as internal controls.

Immunohistochemistry and Scoring System

Immunohistochemical staining was performed on 3 μm tissue sections from the tissue microarray blocks using 22 different primary antibodies against biologically relevant proteins (source and concentration listed in Table 1). Immunodetection was performed with either the LSAB visualization system or Envision (both from Dakocytomation, Glostrup, Denmark) using diaminobenzidine chromogen as substrate. All sections were counter-stained with hematoxylin (DAKO).

Table 1 Antibodies used in the study indicating clone, source, dilution, visualization method, scoring, threshold and positive controls

Negative controls were obtained by omitting the primary antibody. The selection of the threshold and internal controls for the antibodies were earlier established to facilitate reproducibility of the method according to the published literature9 (see Table 1). All cases were independently scored by two pathologists and were recorded as positive taking into account the protein expression in neoplastic cells and the specific cutoff for each marker. Any discrepancies in the percentage of positive cells were resolved by simultaneous re-evaluation of the sections.

Statistical Analysis

Fisher's exact test was used to assess whether a specific protein expression was associated with a given lesion, either Spitz nevus or melanoma. Statistical significance was concluded for values of P<0.05. In addition, adjusted P-values were calculated using Hochberg and Benjamini10 method to avoid the influence of multiple testing. Clinical and histopathological data were compiled after all statistical studies were performed (Table 2).

Table 2 Clinical and histopathological features of Spitz nevi and melanomas

Results

Clinical and Histopathological Features

The series comprises 90 patients (52% females and 48% males) ranging from 3 to 91 years. The average ages for Spitz nevi and melanoma patients were 21.6 and 62 years, respectively. Location of the lesions was known in 71 out of 90 cases (79%), the most common being the extremities for Spitz nevi (46%) and the trunk for melanomas (31%). All Spitz nevi patients were alive with no metastasis, whereas 56% (35 out of 62) vertical growth phase melanomas developed metastasis and 34% (21 out of 62) died.

Most cases of Spitz nevi showed epithelioid or spindle melanocytes with no significant cytological atypia (Figure 1). Neither mitotic figures nor pagetoid extension were seen in any of the 28 cases of Spitz nevi. The histological examination of melanomas (Figure 2) revealed that the most common melanoma subtype, according to the World Health Organization Classification for melanoma, was superficial spreading melanoma (29 of 62 cases, 47%). Twenty one out of 62 cases of melanoma (34%) were ulcerated and 69% (43 of 62 cases) had Breslow's index >1 mm. Most of them presented with metastasis when diagnosed (35 of 62 cases, 56%). Table 2 summarizes clinical and histopathological data of this series.

Figure 1
figure 1

Two cylinders from typical Spitz nevi (H&E; original magnification: × 100 (a, b) and × 400 (c, d)).

Figure 2
figure 2

Two cylinders from primary VGP melanomas (H&E; original magnification: × 100 (a, b) and × 400 (c, d)).

Expression Profiling in Spitz Nevi and Melanoma

The results of protein expression in Spitz nevi and melanoma are detailed in Tables 3 and 4 and illustrated in Figure 3. Immunostaining for melanocytic specific markers showed S-100 positivity in all cases, melan-A expression in the majority of Spitz nevi (96%) and melanomas (92%), whereas HMB-45 was only expressed in 44%, 12 out of 27 cases of Spitz nevi compared with 95% (59 of 62) in melanoma cases.

Table 3 Contingency analysis (Fisher's exact test) comparing the level of protein expression in Spitz nevi and vertical growth phase (VGP-) melanomas
Table 4 Statistically significant markers helping in differential diagnostic between Spitz nevi and melanomas
Figure 3
figure 3figure 3

Representative immunohistochemical staining in VGP melanomas and Spitz nevi for significant proteins including p21 (ad), survivin (eh) and caveolin (il) (original magnification: × 100 and × 400).

Regarding cell-cycle regulators, cyclin D1 showed high levels of expression in Spitz nevi (17 of 23, 74%) and was significantly decreased in primary vertical growth phase melanomas (10 of 61, 16%). No significant changes were found in p16, p53 or Rb, whereas strikingly higher nuclear positivity was noted for p21 in the majority of Spitz neoplasm (91%, 21 of 23) compared with malignant tumors, in which p21 was only expressed in 27% (17 of 62) of cases (Figure 3).

The proliferative activity was assessed using Ki-67 and topoisomerase IIα immunostainings. Spitz nevi exhibited significant lower rate of Ki-67 and topoisomerase IIα, as there were very few cells staining positively (<20%) for Ki-67 and only few cases (15%, 4 of 26) stained positively for topoisomerase IIα in Spitz nevi; however, 37% (23 of 62) and 79% (49 of 62) of melanomas showed nuclear positivity for Ki-67 hazardly distributed and topoisomerase IIα, respectively. In addition to the observed differences in proliferative activity between Spitz nevi and melanomas, in Spitz nevi most of the mitotic figures were confined to the junctional or superficial component of the neoplasm, whereas in melanomas, they were scattered throughout the entire thickness of the neoplasm. Nuclear expression of the apoptotic inhibitor survivin was negative in all nevi, and frequently expressed (42 of 62, 68%) in melanomas (Figure 3).

In the group of membrane receptors, caveolin showed significant loss of expression in vertical growth phase melanomas (19 of 62, 31%) compared with Spitz nevi (16 of 26, 62%) as shown in Figure 3.

Concerning cell–matrix transition proteins, both osteonectin (SPARC) and protein kinase Cα (PKCα) earlier well recognized to have an important function in local invasion,11 have shown to be significantly overexpressed in Spitz nevi (25 of 26, 96% for both markers) compared with melanomas (23 of 53, 43% and 26 of 54, 48%, respectively).

Discussion

In 1910, Darier and Civatte12 described an unusual melanocytic tumor developing rapidly on the nose of a young child. One century later, although guidelines have been proposed for diagnosis, the inability to accurately interpret many of such Spitz tumors and to know their biological potential still makes the diagnosis difficult.7 The most typical or banal Spitz tumors share particular features with conventional acquired nevi. However, the problem with the indiscriminate use of ‘nevus’ is that it connotes a completely benign status and a priori presents no risk to the patient. This is not the case for a subset of spitzoid tumors as it has clearly been shown.1, 13, 14 Despite all the efforts, some proportion of lesions seems impossible to distinguish from melanoma, and rare lesions behave aggressively so that an intermediate or third category lesions difficult to classify as clearly benign or malignant has raised.6 Several authors have favored the term ‘Spitz-like or atypical Spitz tumor’ defining a spectrum of lesions extending from benign nevus to malignant spitzoid melanoma.3, 6, 7, 14 Such a complex context raises the possibility of inaccurate diagnoses that could actually be classified as either benign or malignant, thus avoiding imprecise terms in the histopathologic reports.15

Recently, it seems that comparative genomic hybridization and fluorescence in situ hybridization on lesional tissues represent helpful adjunctive techniques in difficult cases of spitzoid melanocytic neoplasms,16, 17, 18, 19, 20, 21 although these high-resolution molecular techniques are still not widely available and are sometimes difficult to interpret. Thus, further studies are necessary to evaluate these techniques as diagnostic tools. Histopathologic examination remains the gold standard to distinguish Spitz nevi from melanoma, albeit no objective criteria exist to predict the biologic behavior of a specific lesion. Earlier, several studies, mostly based on the expression of single or few proteins, have tried to shed light on the eternal controversy surrounding spitzoid lesions,5, 15, 22, 23 but no markers are yet known to reliably differentiate between the two neoplasms. Thus, neither HMB-45, AgNOR, cyclin D1, c-myc, c-fos, telomerase, cdc-7, anti-leptin receptor, BCL-2, p53 nor p16 proved helpfulness in distinguishing Spitz nevi from melanoma.13, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33

In this study, we performed a comparative analysis of protein expression using a wide panel of antibodies between classical Spitz nevi and vertical growth phase melanomas. Using tissue microarray analysis, a high-throughput means that allows the evaluation of several markers in a large number of melanocytic tumors, we identified a specific protein expression profile differentiating between typical Spitz nevus and non-spitzoid melanoma (Table 2).

This study primarily shows that Spitz nevi displays evidence of cell-cycle deregulation specially in an increased cyclin D1, p16, p53 and Rb expression, which in part accounts for the overlapping immunoprofile seen in Spitz nevi and melanoma.2, 29, 31 However, one of the most striking data observed in this study is that p21, an inhibitor of the cell-cycle progression at the G1/S transition, is found to be significantly overexpressed in Spitz nevi compared with melanoma expression, in agreement with earlier studies.2, 31, 34 These data support the important function of p21 in Spitz nevus induced senescence that may be crucial for protection against malignant progression. It seems likely that elevated levels of p21 reflect unknown regulatory changes within the p21 pathway independent of p53. The absence of correlation between p53 gene mutation and p21 expression has already been shown in different tumors.35 Moreover, no relationship has been found between expression levels of wild-type p53 and p21 protein in melanoma cell lines.36 On the other hand, our results strongly suggest a function of cyclin D1 in the development of melanocytic neoplasms, confirming the earlier observations that it is highly expressed in Spitz nevi29 and there is a diminished expression in vertical growth phase melanomas toward the deep reticular consistent with earlier reports, which show that cyclin D1 is confined to the superficial component of the tumor.37

Unlike melanoma, the cyclin D1 overexpression observed in Spitz nevi does not seem to correlate to cell proliferation. Thus, as earlier observed, Ki-67, a cell-cycle-associated nuclear antigen, nuclear survivin, a mitotic regulator by preserving microtubule stability,38 and topoisomerase IIα, exclusively expressed in the proliferating compartments of normal tissues and tumors,39 are significantly increased in melanoma vs nevus.8, 40, 41, 42, 43 Although melanomas show higher proliferation rates (37 vs 0% in Spitz nevi) when assessed with Ki-67, few cases of Spitz nevi show scattered positive cells at the top of the lesion, whereas melanoma has nuclear labeling throughout the lesion.41, 42

The structural membrane-scaffolding protein caveolin, with additional function in the transduction of signals from the plasma membrane, showed a loss of expression in vertical growth phase melanomas when compared with nevi. These results are consistent with earlier reports9 and support the reduction of its expression and caveolae formation in oncogenically transformed cells.44

In addition, some of the protein markers included in our study have been earlier shown to be involved in mediating cell–matrix interactions and have been implicated in melanoma invasion and progression.11 This study also reveals a high expression of the serin/threonine kinase PKCα and the secreted protein acidic and rich in cysteine, osteonectin/SPARC, in Spitz nevi (96% for both proteins) and a statistically significant difference compared with melanomas (48 and 43%, respectively; P<0.05). To our knowledge, this is the first report focusing on these proteins to differentiate Spitz nevi from melanoma, although osteonectin has earlier been shown to predict metastasis in melanoma.11, 45 These observations primarily suggest that the invasion of the deep dermis is not necessarily related to malignancy, and identify two new markers that can be used in distinguishing melanomas from Spitz nevi.

None of these findings is separately sufficient to rule out the diagnosis of melanoma, but it seems that the combination of a protein expression panel including cyclin D1, p21, osteonectin (SPARC), PKCα, Ki-67, topoIIα and survivin might have a great diagnostic usage in the differentiation of typical Spitz nevus from its malignant mimics. The significantly different expression observed in Spitz nevi and melanomas for some proteins would indicate the usage of a multi-marker expression panel in the differential diagnosis (Table 3). Specially, higher expression of cyclin D1, p21, osteonectin (SPARC) and PKCα, and lower expression of Ki-67, topoIIα and survivin are strongly related to benign behavior.

In summary, we have shown the possibility that a routinely used technique such as immunohistochemistry may aid in differentiating Spitz nevus from melanoma. However, further larger studies including atypical Spitz tumors, lesions that deviate from the typical appearance of Spitz nevi and have an uncertain biological behavior7 with long-term follow up, and spitzoid melanomas seem to be mandatory to confirm and build a useful and valuable panel of marker expression in classifying melanocytic lesions.