A Novel 23-Gene Expression Signature is Predictor of Inferior Prognosis in Melanoma

Abstract

Background: Melanoma is a serious form of skin cancer that begins in melanocytes. Metastasis, somatic mutations and gene expression profiles are important prognostic factors for melanoma patients. However, accurate prediction of patient prognosis remains an unsolved problem for the disease. This study was to develop a novel gene profile to accurately classify melanoma patients into subgroups with different survival probabilities. Methods: Survival-related genes were determined by Kaplan–Meier survival analysis and multivariate analysis using the expression and clinical data of 467 melanoma patients from The Cancer Genome Atlas (TCGA) database and validated in an independent Gene Expression Omnibus (GEO) dataset. Feature selection was performed by the Least Absolute Shrinkage and Selection Operator (LASSO) method. A prognostic 23-gene score was established and compared with two known gene-expression risk scores. The stratification of melanoma patients was performed by unsupervised hierarchical clustering of 23 gene expression levels to identify clusters of melanoma patients with different survival probabilities. Results: The LASSO model comprising 23 genes was considered as the optimal model. The 23-gene score was associated with increased mortality in melanoma patients regardless of clinicopathological characteristics. Hierarchical clustering analysis of the 23 genes revealed three subgroups of melanoma patients. The cluster3 melanoma tumours were associated with higher 23-gene score and poorer overall survival than cluster1 and cluster2 tumours. The 23-gene score had higher area under curve (0.76) than the 8-gene risk score and IRGs score (0.58 and 0.59) in the prediction of overall survival of melanoma patients. Conclusions: The 23-gene score is superior to the two established prognostic gene signatures in the prediction of prognosis of melanoma patients.