Correlation of histopathological patterns in cutaneous melanoma with BRAF mutations

Introduction: Mutations on BRAF gene located on chromosome 7q are the most frequently found in cutaneous melanomas (60%-80%). The only study correlating histopathological patterns of cutaneous melanomas with the presence of BRAF mutations was undertaken by Viros et al. in 2008. The authors observed that morphological features of melanomas are associated with BRAF mutations. Objectives: To correlate histopathological patterns in cutaneous melanoma with the presence of BRAF mutations in order to corroborate the results of the study performed by Viros et al. Methods: Paraffin embedded surgical specimens of 20 primary cutaneous melanomas with BRAF mutation and 20 specimens without BRAF mutation were evaluated independently by two dermatologists that carried out a blind experiment. The features analyzed were nesting, circumscription, presence of isolated melanocytes in the lesion, size and shape of neoplastic cells, and tumor cell pigmentation. Results: “Nesting” was the most prevalent variable for the determination of melanomas with BRAF mutations according to both observers (r = 0.46; p = 0.04). Conclusion: As far as mutational status is concerned, it was not possible to confirm any predictive value for histopathological patterns such as circumscription, presence of isolated melanocytes in the lesion and cytological features. Difficulties in the interpretation of some histological criteria were demonstrated by the variation in the observers’ conclusions. It is difficult to state if genetic alterations such as BRAF mutations may serve as biomarkers for melanoma classification. abstract key words Malignant melanoma


Introduction
The current World Health Organization (WHO) has classified the melanomas in four types, namely, superficial spreading, lentigo maligna, nodular, and acral lentiginous (6) .However, controversy remains as to whether these variations reflect inherent biologic differences or secondary effects that depend on the skin architecture of different anatomic sites (8) .The first to call this classification into question was Ackerman, who considered all malignant melanomas to evolve in a similar way: all are at first horizontally (superficially) spreading within the epidermis and all may eventually extend vertically into the dermis.This repetitive sequence of histological changes in melanoma is accompanied by equally repetitive clinical features at all anatomic sites.Some clinical features, however, are more common on certain anatomic site than others (1) .
Recently molecular biology techniques have permitted the identification of molecular markers involved in melanoma pathogenesis such as the BRAF gene.Mutations on the BRAF gene located on chromosome 7q are the most frequent mutations (60-80%) found in human melanomas.The BRAF gene encodes a serine/threonine kinase downstream of RAS in the RAS/RAF/MAPK pathway that is involved in transduction of mitogenic signals from membrane receptors to the nucleus (3) .
The only study correlating histomorphological patterns of melanomas closely with BRAF mutations was undertaken by Viros et al., in 2008.The authors showed that distinctive morphological features of cutaneous melanoma, such as increased upward migration, nest formation of intraepidermal melanocytes, thickening of the involved epidermis, sharper demarcation to the surrounding skin, and larger, rounder, and more pigmented tumor cells, seem to be associated consistently with BRAF mutations.The authors suggested that a genetically refined morphological classification of primary melanomas may improve existing melanoma classifications by forming subgroups that are genetically more homogeneous and likely to differ in important clinical variables such as outcome and pattern of metastasis (14) .The authors admitted, however, that their results needed to be confirmed in independent samples and in regard to interobserver agreement in the assessment of morphological features.
In this study, we attempted to correlate the assessment of histopathological patterns in cutaneous melanoma with the presence of BRAF mutations in order to reassess the results of Viros et al. especially in regard to interobserver agreement.

Material and methods
A total of 40 cases of primary cutaneous melanoma were included in this study.Twenty cases with BRAF mutation and 20 cases without BRAF mutation were compared.

Histopathological patterns
All histopathological evaluations were carried out independently on routinely stained HE sections by two dermatopathologists.The criteria used to examine the histopathological features were as follows: • nesting: intraepidermal melanocytes were defined as arranged in nests rather than single cells if they formed clusters of five or more cells whether located in the basal epidermis or in higher layers of the epidermis and classified as: √ nests predominant: more than 25% of the melanocytes arranged in nests (Figure 1A); √ nests not predominant: intraepidermal melanocytes present almost exclusively as single cells with only rare nests or with no more than 25% of cells in nests (Figure 1B); • circumscription: lateral circumscription was assessed by examining the transition of the intraepidermal growth portion of the tumor to normal skin at the tumor periphery: √ well circumscribed: when transition from involved epidermis to adjacent normal skin was abrupt and easily determined within one or two rete ridges (Figure 2A); √ poorly circumscribed: continuous decrease of the number of intraepidermal melanocytes making it difficult to establish the transition too normal skin (Figure 2B); • scatter of intraepidermal melanocytes: the proportion of intraepidermal melanocytes present above the basal layer was assessed as: √ scatter predominant: when equal proportions of intraepidermal melanocytes present at the dermo-epidermal junction and in higher epidermal layers or most (> 50%) of the intraepidermal melanocytes situated in the upper layers of the epidermis (Figure 3A); √ scatter not predominant: essentially all melanocytes situated at the dermo-epidermal junction, with only rare melanocytes in higher epidermal layers or the majority of melanocytes (75%-100%) situated at the dermo-epidermal junction, with some present in higher epidermal layers (Figure 3B); • size and shape of tumor cells and nuclei : these features were assessed in the most cellular portion of the tumor using a 20X lens.Nuclei of small lymphocytes were used as a size reference.Tumor cells were considered: √ 1: small if the greatest diameter was < 10 µm (Figure 4A); √ 2: large if the greatest diameter was > 10 µm (Figure 4B).

Statistical analysis
Interobserver agreement was estimated by the kappa coefficient of agreement as follows: • < 0, 20: poor agreement; • 0, 21-0, 4: fair agreement; • 0, 41-0, 6: moderate agreement; • 0, 61-0, 8: good agreement; • 0, 81-1: very good agreement.Fischer's exact test was applied to estimate the correlation of morphological features with presence of BRAF mutation.Odds ratio was used to estimate the risk of mutation according to the morphological features.The multivariate logistic regression model was applied to identify which morphological feature was the most powerful one to determinate the presence of BRAF mutation.Two-sided tests were applied and the significance level was 5%.The sample size was estimated considering an alpha error (type I) of 5% and an alpha error (type II) of 10%, with the estimated power of the test of 90%.

Results
This study analyzed 40 specimens of cutaneous melanoma, 20 BRAF mutated and 20 BRAF wild-type.Four morphological features were evaluated independently by two dermatopathologists (A and J) blinded to the knowledge of the genetic status of the tumors.According to the conventional melanoma classification, 29 were superficial spreading and 11 nodular melanomas.Seventeen superficial spreading melanomas and three nodular melanomas presented the BRAF mutation.
Table 1 summarizes the frequency (n and %) of the morphologic features evaluated by the two observers according to the presence of BRAF mutations.The odds ratio of BRAF mutation according to the morphological features and the observers is given in Table 2.
The multivariate logistic regression analysis, considering BRAF mutation as a dependent variable and the morphologic features as the independent variables, showed that "nesting" was the variable with highest power of determination for melanomas with BRAF mutations (r = 0,46; p = 0,04) for both observers.

Discussion
BRAF represents the most common oncogene mutated in melanoma and according to some authors it is the most important biomarker for this tumour (2) .The frequency of BRAF mutations significantly exceeds the frequency of known mutations of other major genes in cutaneous malignant melanomas, such as N-Ras, p16, and p53 (11) , however, it is not clear, whether BRAF gene mutations are associated with progression of melanoma (5) .The frequent occurrence of BRAF mutations in melanoma suggests that specific BRAF inhibitors could be useful therapeutic agents for advanced melanoma (8) and an improved classification of melanomas that combines an analysis of known genetic factors with histomorphological features might divide melanomas into subgroups that are likely to differ in terms of their clinical outcome and responses to target therapies when they become available (14) .
A recent study showed that melanomas with BRAF mutations presented with an increased upward migration and nest formation of intraepidermal melanocytes, thickening of the involved epidermis, and sharper demarcation to the surrounding skin, and they had larger, rounder, and more pigmented tumor cells (14) .According to the authors, using simple combinations of features, BRAF mutation status could be predicted with up to 90, 8% accuracy in the entire cohort of 302 specimens, as well as with the categories of the current WHO classification (14) .
Despite numerous studies that correlate the presence of BRAF mutations with clinical and prognostic histopathological data in cutaneous melanoma (4,7,9,10,11,13) no other studies attempted to correlate the presence of these mutations with histopathological criteria used for diagnosis of melanoma.
In our study, we tried at first to use the same quantitative classification of criteria as suggested by Viros et al., but the application of these criteria turned out to be very subjective.We found it particularly difficult to classify the presence of intraepidermal melanocytes and scatter of melanocytes into four categories, as well as lateral circumscription into three categories.Therefore, we decided to simplify the classification as given in the section on Methods in this article.Nevertheless, it remained difficult to analyze scatter when the epidermis was very thin (Figure 6) or ulcerated (Figure 7) and to judge predominance of nests or single melanocytes when the tumor was composed largely of sheets of melanocytes.
In the study by Viros et al., interobserver agreement was determined by having an additional observer score in a randomly selected subset of 50 cases.The authors stated in their article that Kappa statistics indicated moderate to excellent interobserver agreement.The data provided by them shows, however, that Kappa coefficient ranged from 0.4 for circumscription, over 0.6 for nesting and 0.7 for scatter to 0.8 for solar elastosis.According to our criteria, interobserver agreement in their study ranged from fair (for circumscription and nesting) over moderate (for scatter) to good (only for the presence of sun damage).
In our own study, all cases were analyzed independently by two dermatopathologists and the coefficient of agreement ranged from 0.49 (for scatter), 0.54 (for cell size and nuclei size) over 0,58 (for tumour cell pigmentation) and 0.7 (for nesting) to 0.74 (for circumscription), indicating that the interobserver agreement was either moderate or good, but never excellent.
Prominent nesting was the only morphological feature, which was significantly associated with BRAF mutations for both observers.This feature found in a melanoma rises the risk for BRAF mutations 7 times (odds ratio = 7[A]; odds ratio = 7, 4 [J]), although the confidence interval was broad, considering the sample size.Nesting was identified as the feature with highest power of determination for the presence of BRAF mutation by multivariate logistic regression analysis, but the coefficient of determination was moderate.
Although our results show that predominance of nests over single melanocytes suggests the presence of BRAF mutations and that prominent nesting raises the risk for BRAF mutations around 7 times, too few cases were analyzed to interpret this finding as definitive.We could not confirm any predictive value in regard to BRAF mutational status of a melanoma for the histopathological patterns of scatter, circumscription, size of cells or nuclei and pigmentation.
Difficulties in the interpretation of some criteria were de monstrated by the variation in the interobserver agreement.At this point, it is difficult to state whether genetic alterations such, as BRAF mutations can serve as biomarkers for classification of melanoma.

Table 1
Morphologic features of Melanomas by Mutation Status of BRAF Fisher exact's test.