Identifying effective melanoma biomarkers is currently a challenging task, to improve early diagnosis and to reduce melanoma-related mortality. In the present study, the expression of 27 cytokines/chemokines was investigated in melanoma serum- and tissue- samples, compared to healthy controls. Serum samples for protein-expression analysis were from 232 patients enrolled at the IDI-IRCCS hospital in Rome, (Italy). RNA expression data of the same 27 molecules were from 511 melanoma- and healthy-tissue samples, from the GENT2 public database. A 3-step approach was carried out for the statistical analyses, namely, a single-molecules Mann-Whitney analysis, a paired-molecules analysis by Pearson correlation, and a profile analysis by the Support Vector Machine (SVM). Protein expression analysis showed significant differences in melanoma samples compared to ctrls, and within melanoma samples as function of Breslow thickness and as function of gender.
Gene expression analysis showed very significant differences and identified extremely effective marker genes, namely, IL-1Ra, IL-7, MIP-1a, and MIP-1b. SVM analysis of the gene expression of these 4 genes demonstrated that their combination represents a very effective signature able to discriminate melanoma patients, with AUC = 0.98. Validation, on additional 1019 independent samples, fully confirmed these observations.
We show here, for the first time, that combined analysis of the gene expression of IL-1Ra, IL-7, MIP-1a, and MIP-1b discriminates melanoma from control tissues in a very effective manner. We therefore propose such 4-molecules profile as an effective melanoma marker and a possible therapeutic target.