Main

Merkel cell carcinoma, a primary neuroendocrine carcinoma of the skin, is a rare neoplasm that probably originates from a primitive pluripotential epidermal cell.1 Similar to the small cell neuroendocrine carcinomas from other locations, Merkel cell carcinoma is a highly malignant neoplasm, being considered the most aggressive primary cancer of the skin.2, 3, 4 The prognosis of Merkel cell carcinoma, however, is relatively favorable when compared to extracutaneous small cell carcinoma.5 In addition to morphological similarities, Merkel cell carcinoma and extracutaneous small cell carcinoma also exhibit genetic analogies such as their cell cycle and apoptosis-related molecular profiles.6, 7

In a previous work of ours,7 the only significant difference between Merkel cell carcinoma and extracutaneous small cell carcinoma was a higher Ki67 proliferation rate in the latter, a finding that would partly explain their different behaviors. Moreover, among Merkel cell carcinomas an elevated Ki67 proliferation rate was associated with a trend to metastasize and kill the patient.7, 8 Nevertheless, differences in proliferation rate do not entirely explain the variable clinical courses of skin tumors. Cutaneous squamous cell carcinoma, like extracutaneous small cell carcinoma, shares with Merkel cell carcinoma some genetic abnormalities, but commonly exhibits a relatively indolent behavior in spite of its very high proliferation rate.9 As altered cell cycle or apoptosis mechanisms do not seem to have a decisive impact on Merkel cell carcinoma biologic behavior, other factors must condition the aggressive evolution of this neoplasm. The present work intended to explore the prognostic impact exerted on Merkel cell carcinoma by the expression of various molecules, which are involved in tumor invasion and angiogenesis and belong to cell activation signaling pathways such as the mitogen-activated protein kinase (MAPK) and AKT pathways. Specifically, we focused on the study of matrix metalloproteinases (MMPs), a family of zinc-dependent endopeptidases capable of degrading the extracellular matrix,10 and tissue inhibitors of metalloproteinases (TIMPs). Additionally, we investigated some other carcinogenesis-related molecules so far unexplored in Merkel cell carcinoma and assessed the influence of neuroendocrine differentiation on the evolution of Merkel cell carcinoma.

Differential expression of the aforesaid molecules was immunohistochemically evaluated using tissue microarray technology, a simple method allowing the application of numerous antibodies to case series with economy of time and reagent consumption, facilitating maximum preservation of paraffin-embedded tissue as well.11 Previous results demonstrated excellent agreement rates between whole section and tissue array analyses.12 A further aim of the study was to look for potentially valuable targets of new treatments, as the survival benefit offered by adjuvant therapy to patients with advanced stage Merkel cell carcinoma is very limited at the present time.

Materials and methods

Samples

Paraffin blocks corresponding to 31 consecutive cases of Merkel cell carcinoma were retrieved from the files of the Departments of Pathology, Hospital Germans Trias i Pujol, Hospital de la Santa Creu i Sant Pau, Hospital del Mar and Hospital de la Creu Roja, Barcelona, Spain. All tissue specimens had been fixed in buffered formalin and routinely processed. The clinical course as regards development of regional and distant metastases and survival was recorded for all Merkel cell carcinoma cases. Of the 31 Merkel cell carcinoma patients, 15 were free of disease on last follow-up whereas 16 had developed metastases. Four lymph nodes showing Merkel cell carcinoma metastases were also included in the study and four cases of lung or urinary bladder extracutaneous small cell carcinoma were selected as controls. Pathological diagnoses were based on microscopic study of hematoxylin-eosin-stained sections. Diagnostic confirmation was provided by immunohistochemical paranuclear dot-like positivity for neuroendocrine markers and low molecular weight keratins. Immunohistochemical investigation of keratin 7, keratin 20, and TTF-1 expression helped to distinguish Merkel cell carcinoma from metastatic extracutaneous small cell carcinoma in doubtful instances. Merkel cell carcinoma paraffin blocks containing necrotic or poorly preserved tissue were excluded. None of our cases showed combined cutaneous squamous cell carcinoma and Merkel cell carcinoma.

Tissue Microarray

Hematoxylin-eosin-stained slides and Ki67 immunostainings of all cases were reviewed by two researchers (MTFF and MG), who marked well-preserved areas. Tissue corresponding to these areas was randomly sampled from paraffin blocks, with no special preference for the different parts of the tumor (eg, superficial zone vs infiltrating border). Two cylindrical cores, each measuring 0.6 mm in diameter, were obtained from every donor block using a tissue microarray workstation MTA-1 (Beecher Instruments, Silver Spring, MD, USA). Cylinders, including those from extracutaneous small cell carcinoma control cases, were arrayed in a new paraffin block. In order to validate whether the tissue array was representative, the proliferation rates (percentage of Ki67-positive tumor cells) obtained with the whole sections were compared with the tissue array values. The concordance was complete in 87% of cases, whereas a difference under 8% was registered in 10% of instances. A difference of 18%, observed in just one case, was attributed to the presence of extensive necrosis in the whole section.

Antibodies and Immunohistochemical Studies

Table 1 summarizes the antibodies utilized and their suppliers, as well as the parameters used for evaluation. Five micrometers thick sections were deparaffinized, hydrated, immersed in buffered citrate, and autoclaved. Afterwards, sections were incubated for 30 min in rabbit serum. Incubations with primary antibodies were carried out for 22 h at room temperature. Slides were washed and incubated with biotinylated rabbit anti-mouse Ig antibodies at a 1:700 dilution and then incubated in PBS/6% hydrogen peroxide for 15 min at room temperature before addition of avidin–biotin peroxidase complex (Dakopatts, Glostrup, Denmark). The chromogen 3,3′-diaminobenzidine tetrachloride (Serva, Heidelberg, Germany) was applied, and counterstaining was performed with Harris hematoxylin. A non-immune mouse serum was used as a negative control. Quantification of the results was adapted to the staining patterns of each antibody. For markers such as P53 where the staining was uniformly strong, only the percentage of positive cells was evaluated. Conversely, intensity of staining was the only parameter considered for markers like PTEN, whereas in positive cases the staining was diffuse. Nevertheless, for most markers, both percentage of positive cells and intensity of staining were evaluated. The composite scores used for intensity and percentage are specified in the footnote of Table 1. The results of synaptophysin were firstly evaluated considering percentage and intensity, but a binary score was also assessed using 10% of positive cells as cutoff point. Owing to the difficulty on accurately quantifying the percentage of vascular endothelial growth factor (VEGF)-positive cells, the cases were classified as negative, high expression (>50%), or low expression (≤50%). Cytokeratin 20 expression was evaluated as a binary variable using 1% as cutoff point to differentiate cases with negative or very low expression from positive cases. Finally, as the abnormal expression of the mismatch repair gene proteins (MLH-1, MSH-2, MSH-6, or PMS-2) consists in a total absence of staining, this variable was also binary (absence/presence). For every case, all parameters (Table 1) were first independently evaluated by each of two authors (MTFF and EM) and then jointly re-evaluated under a double-headed microscope for final score agreement.

Table 1 Antibodies, suppliers, dilutions, and techniques used for immunohistochemistry and parameters evaluated
Full size table

Statistical Study

Immunohistochemical expression of each marker was assessed in the tissue array (two cylinders for each case) and the results were averaged. Statistical analysis was carried out using SPSS (V11.5) software (SPSS Inc., Chicago, IL, USA), with two-sided tests. Cases of Merkel cell carcinoma were classified according to evolution as ‘good prognosis’ or ‘aggressive’ (Merkel cell carcinomas with good prognosis corresponded to disease-free patients and aggressive Merkel cell carcinomas to cases developing metastases). The statistical significance of differences in immunohistochemical expression levels of the different markers was evaluated using Student's t-test and Mann–Whitney test. Differences between groups were considered to be statistically significant when the P-value was less than 0.05. Fisher's exact test was used to determine the statistical significance of differences between proportions in binary variables and in binary-transformed variables (the latter were those variables for which a statistically significant difference or trend was found; they were subsequently recoded). The Spearman correlation coefficient with 95% confidence limits was used to test the association strength between the different antibodies expression levels.

Results

The significant results provided by the statistical analysis are summarized in Tables 2 and 3. No statistically significant association was found between the scores of synaptophysin expression and prognosis. Nevertheless, we observed that all four cases with synaptophysin expression in less than 10% of tumor cells had good prognosis. By using this binary variable, a significant association (P=0.0301, Fisher's exact test) was identified between low synaptophysin expression (Figure 1a) and good prognosis. As for other markers of neuroendocrine differentiation, chromogranin showed the lowest scores and the highest scores corresponded to CD56 (Figure 1b), but no statistically significant differences were found in regard to prognosis. This was also the case with cytokeratin 20.

Table 2 Immunohistochemical markers for which a trend or statistically significant (P<0.05, bold) association with prognosis of Merkel cell carcinoma was found
Full size table
Table 3 Immunohistochemical markers with expression levels showing statistically significant correlation (Spearman's ρ, P<0.05) in Merkel cell carcinoma
Full size table
Figure 1
figure 1

(a) Merkel cell carcinomas showing more than 10% synaptophysin-positive cells had a significant worse prognosis. (b) In most cases, CD56 was the most sensitive neuroendocrine differentiation marker. However, neither the percentage of positive cells nor the intensity of staining correlated with prognosis.

Full size image

The cell cycle markers P53, P21, Bcl-2, and Bcl-6 were not shown to have prognostic value. Overexpression of P53 was observed in cases from both groups. More than 5% of cells were positive for P21 in most Merkel cell carcinomas (Figure 2a), but the percentage of positive cells was usually under 50%. Cases of Merkel cell carcinoma associated with bad prognosis tended to show higher P21 expression levels (P=0.058). Bcl-2 overexpression was almost universal. An inverse correlation between P53 and Bcl-2 expression was not observed. Bcl-6 expression was completely negative in all the lesions studied. All cases of Merkel cell carcinoma expressed KIT (CD117) to greater or lesser extent, but variations in intensity were not significantly associated with prognosis.

Figure 2
figure 2

(a) In most cases of Merkel cell carcinoma, P21 was expressed in more than 25% of cells. Higher expression levels of P21 tended to be associated with a worse prognosis and correlated with (b) higher levels of stromal NF-kappaB.

Full size image

A low percentage of cells were PTEN-positive in all but one case of Merkel cell carcinoma, in which both the primary lesion and metastases showed strong and diffuse (score=3) PTEN expression. This same case of Merkel cell carcinoma exhibited marked P53 overexpression. Nuclear expression of NF-kappaB transcription factor in Merkel cell carcinoma cells ranged from absent to faintly positive. Conversely, the stroma of many tumors exhibited a strong NF-kappaB immunoreactivity, but cases with bad prognosis had significantly higher scores (P=0.0342, rank sum test). Levels of NF-kappaB expression in tumor stroma (Figure 2b) correlated significantly with P21 expression (Spearman's ρ=0.4489, P=0.0128).

Survivin was expressed in many cases and was not significantly associated with prognosis. Positive staining for P38 (Figure 3) was present in most Merkel cell carcinomas in both groups, but P38 scores ranked significantly higher in cases with bad prognosis (P=0.0263). The expression levels of P38 were significantly correlated with those of synaptophysin (Spearman's ρ=0.3760, P=0.0444), P53 (Spearman's ρ=0.3980, P=0.0266), and MMP7 (Spearman's ρ=0.3679, P=0.0496).

Figure 3
figure 3

Phosphorylated P38 was expressed in most cases of Merkel cell carcinoma. Intensity of immunostaining and percentage of positive cells were significantly correlated with a worse prognosis.

Full size image

As regards the expression of prostaglandin E synthases 1 and 2, there was great variability and no significant association with tumor progression was identified. Only two cases showed high hypoxia-inducible factor (HIF)-1α expression levels, which were not related to tumor progression. In spite of the relatively weak intensity of the staining obtained for VEGF, we found a significant association (P=0.0189) between higher scores and aggressive behavior. The expression of VEGF receptor-2 (FLK) was very variable, ranging from high to absent. No significant correlation was found with the clinical course of Merkel cell carcinomas.

Expression of standard CD44 and its variants V3 and V6 was variable and not statistically associated with evolution of Merkel cell carcinoma. No correlation was found between β-catenin expression and evolution of prognosis; aberrant nuclear expression was seen only in one case, which did not develop metastases.

All MMPs, but MMP19, were expressed in tumor cells of Merkel cell carcinomas. MMP13 showed only a slight staining in two cases of Merkel cell carcinoma belonging to the good prognosis group and one instance of Merkel cell carcinoma with bad prognosis. MMP7 overexpression (Figure 4) was significantly associated with bad prognosis (P=0.0107), as was the case with MMP10/2 (P=0.024) (Figure 5), but no statistically significant differences in expression levels of the other MMPs studied were found. Expression levels of VEGF were significantly correlated with those of both MMP7 (Spearman's ρ=0.4489, P=0.0128) and MMP10/2 (Spearman's ρ=0.5602, P=0.0024). Additionally, expression levels of MMP7 were significantly correlated with those of P21 (Spearman's ρ=0.3730, P=0.0463) and P38 (Spearman's ρ=0.3679, P=0.0496). As regards TIMPs, all Merkel cell carcinomas with good prognosis, except two, stained for TIMP3 (Figure 6), which was also detected with a significantly higher score in every case of Merkel cell carcinoma with bad prognosis (P=0.0394). Only six cases of Merkel cell carcinoma were positive for TIMP1 and five cases for TIMP2, all of them with low scores. The differences among TIMP1, TIMP2, or TIMP4 expression levels were not statistically significant. Finally, none of the cases showed alterations in the expression of the DNA mismatch repair gene proteins associated with hereditary non-polyposis colon cancer (MLH-1, MSH-2, MSH-6, or PMS-2).

Figure 4
figure 4

MMP7 expression levels were elevated in most cases of Merkel cell carcinoma and were significantly correlated with adverse outcome.

Full size image
Figure 5
figure 5

MMP10 expression levels in Merkel cell carcinomas were significantly higher in the cases with metastatic spread.

Full size image
Figure 6
figure 6

TIMP3 expression was present in most Merkel cell carcinomas and was significantly correlated with an aggressive evolution.

Full size image

In summary, significant associations were found between metastatic tumor spread and high expression levels of MMP7, MMP10/2, TIMP3, VEGF, P38, stromal NF-kappaB, and synaptophysin. There were statistically significant correlations between expression levels of MMP7 and those of VEGF, P21, and P38. The following expression levels were also found to be correlated: MMP10/2 and VEGF, P38 and synaptophysin, P38 and P53, and P21 and stromal NF-kappaB.

Discussion

Merkel cell carcinoma is probably the most aggressive cutaneous malignancy.2, 3, 4, 13 Its histogenetic origin has been widely debated, although current evidence supports the hypothesis that Merkel cell carcinoma, as well as normal Merkel cells, originates from epithelial cells in either the epidermal basal layer or the adnexa or eventuates from a primitive cutaneous epithelial precursor.14 This hypothesis is supported by the existence of features common to both Merkel cell carcinoma and cutaneous squamous cell carcinoma, such as their tendency to appear in sun-damaged skin of elderly people and their sharing of identical UV-type-specific p53 and Ha-ras mutations.8 Conversely, Merkel cell carcinoma exhibits similarities with extracutaneous small cell carcinoma that are not only of a morphological or immunohistochemical kind but also include some intrinsic sensitivity to ionizing radiation and chemotherapy, as well as a high metastatic potential.13 Therefore, Merkel cell carcinoma seems to occupy an intermediate position between cutaneous squamous cell carcinoma and extracutaneous small cell carcinoma, so that Merkel cell carcinoma is much more aggressive than the former but less so than the latter.2, 3, 4, 5

Neuroendocrine differentiation is considered to carry a negative prognostic connotation in some carcinomas,15 whereas other carcinoma types have yielded contradictory results16, 17 or shown just a trend toward an association between neuroendocrine differentiation and high tumor grade.18 We found a statistically significant association of low (<10% of tumor cells) expression of synaptophysin with good outcome in Merkel cell carcinoma. Moreover, chromogranin A expression scores ranked higher in the group of cases of Merkel cell carcinoma with aggressive behavior, although this difference did not reach statistical significance. Nevertheless, we found the highest extension and intensity scores when using CD56 antibody, which corroborates the contention that this is a useful marker for Merkel cell carcinoma,19 although it would be convenient to use it in combination with other markers that allow confirmation of the neuroendocrine nature of the neoplasm. Our results differ from those of a previous study,20 in which the expression of neuroendocrine markers in Merkel cell carcinoma cells was seen to be associated with good prognosis. In addition, very low cytokeratin 20 expression was associated with good prognosis in our series.

Most Merkel cell carcinomas in our series were positive for Bcl-2, P53, and KIT (CD117) but, as in previous studies,21, 22 no association with prognosis was found. Bcl-6, normally present in the epidermal prickle layer and also detected in well-differentiated cutaneous squamous cell carcinoma,23 was steadily negative in our series of Merkel cell carcinomas. This supports the assumption that Bcl-6 expression is associated with squamous cell maturation.23 The tumor suppressor gene p21, which to the best of our knowledge has never been previously studied in Merkel cell carcinoma, was also widely positive in most of our instances of Merkel cell carcinoma, and its expression levels tended to be associated with a bad prognosis.

In regard to the PI3-K pathway, alterations of NF-kappaB and PTEN are frequently found in human tumors of diverse origin. Activation of NF-kappaB providing resistance to chemotherapy and radiotherapy has been described in cutaneous squamous cell carcinoma,24 but seems to play a minor role in Merkel cell carcinoma. Nevertheless, we detected NF-kappaB overexpression in ‘reactive’ intratumoral myofibroblasts, probably a reflection of the stromal transformation associated with carcinogenesis.25 The correlation between P21 and stromal NF-kappaB expression levels that we found can be explained by the ability of NF-kappaB to induce P21 expression.26 In our cases of Merkel cell carcinoma, there was downregulation of the tumor suppressor PTEN. In contrast to some findings in extracutaneous small cell carcinoma of the lung,27 it has been previously shown that the incidence of PTEN mutations in Merkel cell carcinoma is low.28 Thus, we can hypothesize that PTEN gene inactivation is caused by other transcriptional or post-transcriptional mechanisms, such as the promoter methylation seen in non-small cell carcinoma of the lung.29

All our cases of Merkel cell carcinoma showed high rates of nuclear survivin expression, independently of their benign or aggressive evolution. Survivin is overexpressed in many cancers, usually translocated to the nucleus, and associated with metastatic spread, resistance to chemotherapy and bad prognosis.30, 31, 32, 33 In some tissues neuroendocrine differentiation seems to induce survivin activation.31, 32 Our results suggest that survivin contributes to Merkel cell carcinoma development but has no independent impact on the metastatic spread of the neoplasm. Of interest, in Merkel cell carcinoma PTEN downregulation might contribute to survivin overexpression, an association also described in high-grade tumors from other locations.33

MAPKs are involved in cell differentiation, proliferation, and apoptosis.34 Three main MAPK pathways are known in humans: the extracellular signal-regulated kinase (ERK) pathway, the C-Jun NH2-terminal kinase (JNK) pathway, and the P38 pathway.34 Whereas P38 and JNK are better known by their role in the inflammatory and immune responses,34, 35 many studies have demonstrated ERK activation in human cancers.35, 36 Nevertheless, all three MAPK pathways seem to be involved in carcinogenesis and recent evidence suggests that P38λ, one of the four P38 isoforms (P38α, P38β, P38κ, and P38λ),36 is instrumental in malignant transformation linked to K-ras mutations. In a study of Merkel cell carcinoma, the ERK pathway was completely inactive in 42 of 44 cases,37 but no information on the P38 pathway was provided. In the present study, there was immunopositivity for phosphorylated P38 in 26 of 31 cases of Merkel cell carcinoma. Additionally, expression levels of P38 were seen to be significantly correlated with those of synaptophysin, P53, and MMP-7, and cases of Merkel cell carcinoma with bad prognosis showed significantly higher expression levels of P38.

Aberrantly increased cyclooxygenase-2 (COX-2), found in many carcinomas, is an early event in tumorigenesis.38 Both phosphorylated P38 and UV exposure can increase cellular levels of COX-2. In spite of the influence exerted by both mechanisms in most cases of Merkel cell carcinoma, low levels of COX-2 in Merkel cell carcinoma were found in a previous study.39 In view of this, we decided to explore the expression of two enzymes downstream of COX-2 (microsomal prostaglandin E synthases 1 and 2) and found low expression levels that were unrelated to biologic behavior, even though microsomal prostaglandin E synthase 1 expression tended to be more extensive in Merkel cell carcinomas with good prognosis.

HIF-1, a transcription factor, is overexpressed in some human cancers40 and can influence target genes such as the VEGF.41 In our cases of Merkel cell carcinoma we could not demonstrate activation of the HIF pathway, the intensity of VEGF staining being faint in most cases. Even so, our aggressive Merkel cell carcinomas showed significantly higher VEGF expression levels and, interestingly enough, VEGF expression in our series was statistically correlated with that of two MMPs (MMP-7 and MMP-10/2), which also showed higher expression scores in aggressive cases. On the other hand, expression of VEGF receptor 2 (FLK-1) was variable in our cases of Merkel cell carcinoma. These results suggest a likely role of VEGF signal pathway in Merkel cell carcinoma tumor progression and the potential benefit of using anticancer treatments targeted on these molecules.

Albeit a previous study42 reported the expression of cell adhesion molecule CD44 in Merkel cell carcinoma and suggested that its loss might be related to tumor dissemination; our results do not confirm this contention. We found highly variable (from absent to strongly positive) expression of standard CD44 and CD44 variants V3 and V6 in our cases of Merkel cell carcinoma, independently of their evolution.

The physiological roles of β-catenin include not only cell adhesion and maintenance of tissue architecture and polarity, but also transcriptional activation.43 Mutations of the β-catenin gene, CTNNB1, may lead to expression loss or aberrant nuclear accumulation. Whereas β-catenin loss causes decreased cell adhesion and facilitates metastatic spread, nuclear translocation has a proliferative effect.43 Both alterations have been described in tumors arising in different organs, including neuroendocrine lung carcinomas,44 and are usually associated with poorer survival.43, 44, 45 Decreased or aberrant expression of β-catenin was a frequent finding in our cases of Merkel cell carcinoma and confirms previous reports of altered adhesion molecule profiles in Merkel cell carcinoma and their likely role in the development of this neoplasm.46

MMPs and their tissue inhibitors, a large group of molecules involved in a variety of physiological processes, seem to have an important function in metastatic spread, although their expression in Merkel cell carcinoma has been only occasionally explored.47 The concentration of most MMPs and TIMPs is regulated at transcriptional and post-transcriptional levels and immunohistochemistry is a good method to evaluate their expression.48 In a previous work on 23 cases of Merkel cell carcinoma, a statistically significant correlation was found between high MMP1 and MMP3 expression levels and metastatic potential, whereas MMP7 was not evaluated.47 Upregulation of MMP7 (matrilysin), a predictor of poor outcome in several neoplasms, is correlated with stage in non-small cell lung carcinoma49 and has been described in high-grade neuroendocrine lung tumors.50 Moreover, the MMP7-181A/G polymorphism seems to be related to a higher risk of developing certain carcinomas.51 In our study, we found high expression levels of most metalloproteinases, but only those of MMP7 and MMP10 were significantly related to bad prognosis. There was a significant correlation between the expression levels of MMP7 and those of VEGF, P21, and P38. Upregulation of MMP10 (stromelysin 2), also an unfavorable prognostic factor in many neoplasms,49 was significantly associated with an aggressive outcome and correlated with VEGF expression levels in the present study.

Regarding TIMPs, TIMP1 has been shown to be overexpressed and associated with poor outcome in many human malignancies,52 but this was not the case in our series. As for TIMP2, it does not seem to play a role in Merkel cell carcinoma tumorigenesis. In contrast, we found that TIMP3 overexpression was significantly associated with an adverse outcome. TIMP3 expression loss can be due to hypermethylation or mutant p53-mediated transcriptional repression. In many neoplasms, TIMP3 downregulation is related to tumor invasion and increased angiogenesis, as the binding of VEGF to VEGF receptor-2 is unblocked and downstream signaling is activated. A recent article claims that TIMP3 deficiency in the host, but not in the neoplasm, is responsible for enhancing tumor growth and angiogenesis.53 Additionally, in the work by Massi et al,47 TIMP3 was seen to be expressed in 91.3% of Merkel cell carcinoma cases. In other studies, 70–100% of lung extracutaneous small cell carcinomas cells were immunohistochemically positive for TIMP3,54 and TIMP3 overexpression in carcinoma cells was related to a worse prognosis.55, 56 Finally, we observed a tendency of TIMP4 downregulation to be related to aggressiveness. Low TIMP4 expression has also been described in lung cancer54 and seems to be caused by methylation.57

None of our Merkel cell carcinomas lacked MLH-1, MSH-2, MSH-6, or PMS2 immunoreactivity. Therefore, the tumors in our series were unrelated to Muir–Torre syndrome or hereditary non-polyposis colon cancer syndrome, which are secondary to germline mutations in the aforesaid genes.58

In conclusion, our identification of expression profiles associated with aggressive behavior of Merkel cell carcinoma may be helpful in predicting the clinical course of Merkel cell carcinoma and potentially useful for the development of much needed therapies against the most aggressive of skin neoplasms.