Melanoma (MM) is the human malignancy that has undergone the greatest increase in incidence during the last few decades.

Population Studies suggest that approximately 6 to 14% of melanoma cases occur in a familial setting [1]. In Spain, familial Melanoma is considered when there is at least one invasive melanoma and one more case of melanoma and/or pancreatic cancer among firstdegree relatives on the same side of the family [2]. In these families two major melanoma susceptibility genes have been identified. The oncogene CDK4 has been found in a few melanoma families (estimated 2%) [3]. Germline mutations in CDKN2A are found in approximately 20 to 40% of melanoma families [4]. Several studies have reported that patients with melanoma and a CDKN2A mutation have an earlier age of onset and an increased risk of multiple primary melanomas (MPM) [5-7]. Not only melanomas occurring in mutation carriers but familial melanoma has been demonstrated to share some characteristics in previous studies [8].

Survivin represents a multifunctional protein that suppresses apoptosis and regulates cell division at the G2-M phase. It is a nuclear shuttle protein that is actively exported from the nucleus [9]. Survivin seems to exist in 2 subcellular pools (in the cytoplasm and nuclear). This is consistent with its function in the regulation of both cell viability and cell division. Growing evidence suggests that survivin expression in cancer cell nuclei may represent an important prognostic marker to predict disease outcome. Current reports in this research area are however inconsistent and propose opposing conclusions regarding the significance and prognostic value of survivin nuclear expression [10,11]. Survivin has been recently identified as a metastasis-associated gene for Melanoma [12].

The purpose of this study was to further characterize and expand the knowledge of the clinical and histopathologic characteristics of familial melanoma to provide more information to clinicians and also contribute to the understanding of the complex interplay of genetic and environmental factors in the pathogenesis of melanoma.

The study was approved by the institutional review board.

All familial melanoma patients from whom the paraffin block of the tumour was available were eligible for the study, and two sporadic melanoma patients from whom the paraffin block was also available were also eligible for the study.

We compiled 189 paraffin blocks and the corresponding slides of 189 MMs (62 familial MMs and 127 sporadic MMs). The following variables were evaluated:

Epidemiological data

Included sex, age, age at diagnosis, histopathological subtype, melanoma site, Breslow thickness, presence of metastases and follow up.

Phenotype data

Included eye and hair colour, phototype and nevi count.

Analysis of Histological Features

All histopathological evaluations were carried out on routinely stained HE sections. Cases were classified as superficial spreading melanoma (SSM), lentigo malignant melanoma (LMM), nodular melanoma (NM), acral lentiginous melanoma (ALM) according to the WHO classification [13]. Breslow thickness and Clark level of invasion were evaluated for each tumour. Based on the work of Viros and coworkers [14]. we evaluated the following histological features:

Solar elastosis, type of cells, inflammatory infiltrate, regression, mitotic rate, pagetoid invasion, nest formation, lentiginous hyperplasia and cellular atypia.

Immunohistochemical analysis TMAs

For immunohistochemical evaluation of all the tumours we constructed tissue microarrays (TMAs). We selected a minimum of 2 areas per tumour and a total of 4 TMAs blocks were performed. Each TMA block was cut into four micrometer sections.

Immunohistochemical studies were performed from tissue microarrays with the automated immunohistochemical system TechMate 500® (Dako Co, Carpinteria, CA), using the EnVision system (Dako). Briefly, 4 μm sections were deparaffinized and hydrated through graded alcohols and water. Peroxidase was blocked for 7.5 minutes in ChemMate peroxidase-blocking solution (Dako). Then, the slides were incubated with the primary antibodies for 30 minutes and washed in ChemMate buffer solution (Dako). The peroxidaselabelled polymer was then applied for 30 minutes. After washing in ChemMate buffer solution, the slides were incubated with the AEC substrate chromogen solution, washed in water, counterstained with hematoxylin and mounted. The primary antibody used in the study was Survivin (Abcam, Cambridge, UK; 1/500 dilution).

CDKN2A mutation analysis

Blood samples were taken from all patients belonging to the familial MM group. The PUREGENE DNA Isolation Kit (Gentra Systems, Minneapolis, MN, USA) was used to isolate genomic DNA from lymphocytes according to the manufacturer’s instructions.

Promoter (-34G>T variant), intronic (IVS2-105) and coding regions of the CDKN2A gene (exons 1α, 2 and 3 of the p16INK4A protein and exon 1β corresponding to p14ARF protein) were amplified by PCR using primers and conditions previously described [15].

Statistical analysis

Descriptive analysis of the sample was performed, including percentages for categorical variables, and mean, minimum, maximum and standard deviation values for continuous variables. Comparisons of continuous variable means were performed using Student’s exact t-test when variables followed a normal distribution. Comparisons of discrete variable means were performed using the Mann–Whitney non parametric test. Comparisons between categorical variables were performed with χ2 tests and Fisher corrections were required. Kaplan Meyer analysis and Cox proportional hazards regression models were used to analyze associations between available variables and overall survival. All variables on univariate analysis were incorporated into a multivariable model. Statistical analyses were performed with SPSS (version 18.0; SPSS Inc., Chicago, IL).

Descriptive results

We included 189 MMs, 62 (38.8%) belonging to the Familial MM group.

Patients included 88 men (46.6%) and 101 women (53.4%) with a median age of 62.83 year ± 18.35. In the Familial MM group 17 patients (25.8%) were carriers of a CDKN2A mutation and 39 (20.6%) of all patients had multiple MM (MMM). For the analysis, the first tumour from 10 MMM patients (25.6%) was included and successive tumours in 29 patients (74.4%). In the familial patients, 37 (61.7%) were the index case and 23 (38.3%) were not. We included two related patients in the familial group.

The median Breslow thickness was 1.53 mm ± 3.072 (min 0.2 max 3mm). Taking into account MM histopathological subtype 128 (67.7%) were SSMM, 14 (7.9%) LMM, 35 (18.5%) ALM and 11 (5.8%) NM.

Comparison of familiar and sporadic MMs

Of 189 patients included in the study, 62 (38.8%) had a positive family history of melanoma. Patient characteristics and demographic data are presented in Table 1. Patients in the familial melanoma group had a significantly younger age of melanoma onset (median: 44.26 vs 56.87), p<0.05. Melanomas belonging to the sporadic melanoma group had a significantly higher Breslow thickness (mean: 2.35 vs 1.22), p<0.05. There was a significantly higher number of in situ melanomas in the familial group than in the sporadic group (35.5% vs 19.7%), p<0.05 (OR. 1.662, 95% CI 1.111-2.485). We found differences, although not significant, in the median Breslow thickness in the MMM group depending on whether the MM was the first or successive (0.64 mm vs 0.33 mm), p=0.065. Although not significant, successive MM were more frequently of the in situ type than first MM in those patients (48.3% vs 20%), p=0.152. We found differences, also not significant, in the median Breslow thickness in the familial group depending on whether the patient was the index case or was not (1.4 mm vs 0.73 mm), p=0.713. Non index cases were more frequently of the in situ type than index cases (52.2% vs 25%), p=0.033 (Table 2). Taking into account melanoma site, a significantly higher percentage of melanomas in the lower limbs was found in the familial melanoma group (30.6% vs 13.4%), and no melanoma was found on palms in this group, p<0.05. Comparing the histological subtype of MM between the two groups we found significant differences among SSMM; we found that 82.3% of the familial melanoma tumours were of SSMM type vs the 61.4% of the sporadic melanomas, p<0.05.

Table 1: Patient and tumour characteristics. Familial versus Sporadic group.

Table 2: Multiple and familial melanoma. Mean Breslow thickness and percentage of in situ type melanomas

Phenotypic characteristics and histopathological features of tumours are given in Table 3.

Table 3: Phenotypic characteristics and histopathological features. Familial versus Sporadic group.

Interestingly a higher number of familial tumours had cytoplasmatic positivity (83.9 vs 70.1), p=0.05. Multivariate regression analysis showed that the most representative variables associated with familial melanoma were earlier age at diagnosis (OR 1.036; 95% CI 1.017-1.055), lower Breslow thickness (OR 1.288; 95% CI 1.013-1.683) and in situ melanomas (OR 2.645; 95% CI 1.211-5.778) (Table 4).

Table 4: Multivariate analysis for characteristics associated with familial melanoma.

Comparison of CDKN2A mutated and non mutated MMs

We studied p16 mutations and we found that 17 patients were carriers of a CDKN2A mutation, all of them belonging to the familial group. Patient characteristics and demographic data are presented in Table 5. Carriers of a CDKN2A mutation had a significantly earlier age of melanoma onset (mean: 37.51 vs 54.19 year), p<0.05. Melanomas belonging to non-carriers had a significantly higher Breslow thickness (2.12 vs 0.74), p<0.05. There was a significantly higher number of in situ melanomas in the mutation carrier group (58.8% vs 21.3%), p<0.05. As regards melanoma site, we found a higher number of lower limb melanomas (47.1% vs 16.6%), p<0.05. All the tumours in the carrier group were of the SSMM type, p<0.05.

Table 5: Patient and tumour characteristics Carriers of a CDKN2A mutation versus non-carriers.

Phenotypic characteristics and histopathological features of tumours are given in Table 6. Tumours of CDKN2A mutated patients had significantly more big round cells than tumours of non-carriers (35.3% vs 4.7%), p<0.05 (Figure 1). No differences were observed in the other histopathological characteristics evaluated. No differences were observed in immuno-positivity for nuclear survivin between the two groups, but a higher number of tumours from CDKN2A mutated patients had cytoplasmatic positivity (93.8 vs 72.7%), p=0.052.

Figure 1: a and b: Big round cells in p16 mutated tumour.

Table 6: Phenotypic characteristics and histopathological features Carriers of CDKN2A mutation versus non-carriers

Multivariate regression analysis showed that the most representative variables associated with CDKN2A mutation carriers were earlier age at diagnosis (OR 1.060; 95% CI 1.016-1.105), in situ melanomas (OR 6.961; 95% CI 1.895-25.567), the presence of multiple melanomas (OR 8.920; 95% CI 2.399-33.166) and immunopositivity of the tumours for cytoplasmic survivin (OR 9.072; 95% CI 1.025-85.010) (Table 7).

Table 7: Multivariate analysis for characteristics associated with p-16 melanoma.

Kaplan Meyer analysis and Cox proportional hazards regression models were used to analyze associations between available variables and overall survival. Mean time of follow up for our patients was 73.78 months (SD 41.53 months, minimum 0 months and maximum 217 months). All the variables in the univariate analysis were included into a multivariate model. On univariate analysis sex, Breslow thickness, Clark level, histopathological subtype, Sporadic or Familial MM group, cellular atypia, inflammatory infiltrate and number of mitoses were associated with overall survival (Table 8). On multivariable analysis, sex carried significant prognostic value for overall survival (hazard ratio (HR) 4.802, 95% confidence interval (CI) 1.779-12.956). Breslow thickness was a significant independent predictor of overall survival (HR:1.571, 95% CI 1.366-1.806). Cellular atypia was also a prognostic value on overall survival on the multivariable model.

Table 8: Kaplan Meyer analysis and Cox proportional hazards regression models results.

We compared the clinical and histopathological characteristics of 62 patients with familial melanoma and 127 patients with sporadic melanoma in our Mediterranean area. We also studied the presence of CDKN2A mutation in patients belonging to the familial melanoma group and compared clinical and histopathological characteristics of 17 patients with CDKN2A mutation and 168 non-carriers. The characterization of familial cases has been shown to result in early detection of new melanomas by elaborating specific educating and surveillance plans for all the members [16].

Several significant differences were found between familial and sporadic groups. When the variables were included in a multivariable analysis, the earlier age of onset, lower Breslow thickness and a higher proportion of in situ melanomas maintained their significance in the model.

Of the 62 familial melanoma patients, 17 were found to be carriers of a CDKN2A mutation. Several significant differences were found between carriers and non-carrier groups. When the variables were included in a multivariable analysis the earlier age of onset, a higher proportion of in situ melanomas, the higher risk of multiple melanomas and the cytoplasmic survivin immunostaining maintained their significance.

Our results agree with previous reports; other studies reported a younger age at diagnosis of familial melanoma, an increased risk of multiple primary melanomas and a higher proportion of in situ and superficial spreading malignant melanoma [6,7]. Nagore and coworker’s also studied a Mediterranean population of familial cases and they found no familial melanoma on hand or foot and no histological ALM. Similarly, in our series we found no familial melanoma on palms and a lower proportion of ALM than in the sporadic group, and no cases of ALM in carriers [6]. An increased proportion of superficial spreading type melanomas in familial melanoma have been reported before [6]. It has been suggested that this is because a relatively large proportion of melanomas in patients with familial melanoma arises from nevi as familial melanoma is associated with increased nevi number. Melanomas that are associated with nevi are usually of the superficial spreading type. Our results are in accordance with this hypothesis as we found a higher proportion of superficial spreading melanomas in the familial and p-16 group and also a larger number of nevi, but we did not analyse the proportion of melanomas arising from nevi.

A higher frequency of melanomas located on the lower limbs was found in familial melanomas and in p-16 melanomas compared with sporadic and non-carriers. In a study by our group Carrera and coworkers found that a higher proportion of the early melanomas on the limbs studied belonged to familial cases [17].

The finding of thinner tumours in familial and p-16 cases could be explained by these patients being high risk patients already under close surveillance, although we and others found that patients with familial melanomas still manifested earlier ages at diagnosis even when only the first cases in each family were considered, which points to an earlier occurrence in this group because of a genetic predisposition [18]. In accordance to that, we found no statistical differences in median Breslow thickness in the familial group depending on whether the patient was the index case or not. The higher frequency of multiple melanomas in individuals with a familial form of cutaneous melanoma is an unquestionable fact, widely reported in the literature and intimately related with the presence of mutations in the CDKN2A gene [14]. In our study, the presence of multiple primary melanomas was a predictive factor of being a familial case, and the strongest predictive factor of being a CDKN2A carrier, even after multivariate analysis. Our results suggest a higher expression of cytoplasmic survivin in familial and p-16 tumours, a characteristic that is maintained on the multivariate analyses for p-16 tumours. Vetter and co-workers found that cytoplasmic survivin is mainly expressed in metastatic melanomas [12]. We know that somatic loss of p16 is seen in the majority of human melanomas and that this accelerates melanomagenesis [19]. The fact that Vetter and co workers found cytoplasmic survivin mainly expressed in metastatic melanomas could be related to p16 loss, and would be in agreement with our finding that germline p16 mutated melanomas stained for cytoplasmic survivin. The mean Breslow thickness in those p16 germline mutated tumours positive for cytoplasmic survivin was 0.8mm, and we were not able to value survivin as a prognostic marker in this subset of patients because none died or developed metastases.

Our results in the familial and p-16 mutated population are in accordance with the inherited increased susceptibility with, most of them reported before in other founder populations including: (1) younger age of onset; (2) high risk of MPM; (3) lower Breslow thickness and a higher proportion of in situ melanomas and (4) higher proportion of cytoplasmic surviving [7].

Taken together, our findings agree with the hypothesis of at least two pathways for cutaneous melanoma pathogenesis proposed by Whiteman and co-worker’s, one associated with increased numbers of nevi, intermittent sun exposure, younger age at diagnosis and location on the trunk (nevus pathway) [20]. The second pathway is associated with chronic sun exposure, fewer nevi, older age at diagnosis, and location in the head and neck region (ultraviolet B pathway). In our familial and p-16 population, the increased number of nevi, the high proportion of superficial spreading type melanomas and low proportion of lentiginous melanomas, the low proportion of tumours located on the head and neck area (with no cases in the p-16 population) and the higher proportion of sunburns during childhood suggest that melanomas in these patients develop predominantly though the nevus pathway.

In summary, our study shows that familial occurrence of cutaneous melanoma was significantly associated with earlier age of onset, lower Breslow thickness and a higher proportion of in situ melanomas; and also with higher risk of multiple melanomas and the cytoplasmic survivin immunostaining in the case of carriers of CDKN2A mutations. As reported in other founder populations with other CDKN2A mutations our findings are in concordance with the so-called divergent pathways hypotheses: familial melanomas tend to follow the nevus pathway.

These findings were based on patients with melanoma in the Mediterranean area. More studies are necessary to determine whether our results apply to other populations.

We thank the research nurses from the Melanoma Unit, Daniel Gabriel and Pablo Iglesias for collecting phenotype data of patients and to Maria Eugenia Moliner the clinical nurse of the Melanoma Unit for all her support.

National Cancer Institute (NCI) of the US National Institute of Health (NIH) (CA83115).