Specific markers for discrimination of melanoma from dysplastic nevi were scarce. The misdiagnosis of melanoma is the second most common reason for cancer malpractice claims in the United States 
. To overcome these limitations, we utilized TMA to evaluate diagnostic usefulness of biomarkers for distinguishing melanoma from dysplastic nevi. Our data demonstrated that 4 of 12 markers were diagnostically useful, either singly or in combination for distinguishing melanoma from dysplastic nevi. This method may represent a simple and easy way to implement the translation of tissue microarray data into clinical practice. Four markers described here could be used to assist in the histological diagnosis of melanoma, thereby providing important information to clinical pathologists.
We found that Bim, BRG1, Cul1 and ING4 were differently expressed between melanoma and dysplastic nevi using the univariate logistic regression (p < 0.01). Univariate analyses alone may not be the best approach in choosing which markers to combine in a predictive panel of markers. The differential expressions of Bim, BRG1, Cul1 and ING4 between melanoma and dysplastic nevi were confirmed by the multiple logistic regressions (p < 0.05). These results imply that it is possible to combine this 4-marker panel to distinguish melanoma from dysplastic nevi. For the single biomarker accuracy, ING4 has high sensitivity of 93.46% and specificity of 66.67%. We found that applying the multiple-biomarker strategies improved specificity and AUC. Combined 4-markers showed substantial improvement specificity over single marker, from the highest 66.67% to 81.8%, and AUC reached 84.3%. Our experience, as well as that of others, has shown that a panel of markers is most helpful for differentiating between melanoma and dysplastic nevi. Furthermore, the reliability of 4 markers to distinguish melanoma from dysplastic nevi was confirmed by ANN. ANN analysis as a statistical modeling tool has demonstrated the ability to assimilate information from multiple sources and detect subtle and complex patterns 
. In our study, combination of 4 markers improved AUC prediction of 86.2% for dysplastic nevi and 86.0% for melanoma.
Utility of markers were measured by the classification tree, we identified ING4, Cul1 and BRG1 were the most important classification parameters in ranking top-performing biomarkers. Classification tree was separated by the most powerful prediction variable ING4, and the 10 fold of cross-validation error was 0.03, suggesting that the tree was stable and reliable. CART is an alternative to logistic regression and has several advantages as a tool for developing clinical decision rules. A decision tree, on the other hand, is easily understood by physicians. The most clinically useful information gained by using the prediction CART tree is that ING4 is priority marker, then Cul1 and BRG1 in clinical application. We analyzed primary and metastatic melanomas separately, the ING4 and Cul1 were the best in the classification tree. Furthermore, the CTHRC1 marker may be a useful marker for primary melanoma and BRMS1 serves an important marker for metastatic melanoma distinguishing form dysplastic nevi. The classification tree requires additional research to validate the diagnostic value of Bim, BRG1, Cul1 and ING4 in an independent data set.
Biological interpretation is required to understand why the proposed markers are significantly different and as a utilization in patients with melanoma and dysplastic nevi. Several of the markers incorporated into our study have been previously demonstrated to have a role in driving melanoma progression. ING4, a tumor suppressor, mediates chromatin modification and has a suppressive effect on tumorigenesis and innate immunity 
. It inhibits melanoma angiogenesis by suppressing NF-κB pathway and is involved in melanomagenesis and induces growth suppression and apoptosis in melanoma cell line 
. Cul1, a member of Cullin family, plays an important role in protein degradation and protein ubiquitination, is increased in early stages of human melanoma and promotes melanoma cell proliferation through regulating p27. 
BRG1, the catalytic subunit of the SWI/SNF chromatin remodelling complex, is a novel binding partner of the tumor suppressor p16INK4a, which is one of the most important melanoma susceptibility genes identified to date 
and is involved in melanoma initiation 
. Bim is a novel member of the Bcl-2 family that promotes apoptosis 
. Among all the BH3-only protein members, Bim has been shown to have the ability to interact with all Bcl-2 members, suggesting that it may serve as a key factor in the event of apoptosis and thus inhibition of Bim function may be involved in tumorigenesis.
In summary, we describe a multi-marker immunohistochemical panel of Bim, BRG1, Cul1 and ING4 which may aid in differential diagnosis for melanoma from dysplastic nevi.