The use of integrative genomics to define molecular signatures of melanoma histologic subtypes.

Abstract

8553 Background: Studies of c-kit inhibitor imatinib for the treatment of acral melanoma emphasize the importance of delineating and targeting the unique pathways driving tumorigenesis of histologic subtypes. There are no agents specifically tailored to treat the two most common subtypes: superficial spreading (SSM) and nodular melanoma (NM). We integrated gene expression and SNP array data to identify genes and pathways that differentiate NM and SSM. Methods: Gene array (U133A 2.0 Affymetrix) and SNP array (6.0 Affymetrix) were performed on 18 melanoma cell lines (2SSM, 4NM, 12 MET) and 4 melanocyte controls. Gene lists and correlations between DNA copy number and mRNA expression were made using the Partek Genomics Suite. Array data were validated in an expanded group of NM and SSM lines using qRT-PCR, western blot, and an in silico analysis of a public melanoma data set (Jaeger CCR 2007). Immunohistochemistry (IHC) of candidate genes was performed using human tissues from the IMCG. Results: Principal component analysis of gene expression showed high variance among NMs, with all but one falling outside 2 standard deviations of the mean expression for all other melanomas (SSM and MET). mRNA array revealed 114 genes differentially expressed between NM and SSM (p < 0.05, Bonferroni correction). Integrative genomics yielded 14 genes in which expression was significantly correlated with copy number (p < 0.05, Spearman's rank). qRT-PCR verified differential expression between NM and SSM in 7/8 (88%) genes studied, and the same trend between NM and SSM expression of all 7 genes was observed in the external validation set of human tissue. Pathway analysis of the most differentially expressed genes between NM and SSM showed highest representation of metabolic processes (65 genes, 55.1%, p = 0.05). Metabolism-related proteins MTAP and ALDH7A1, and reported tumor suppressor EPB41L3 were assessed by western blot which verified higher expression in NM vs. SSM in all 3 proteins. IHC in human specimens also showed higher levels of ALDH7A1 in NM vs. SSM. Conclusions: Integrative genomics revealed molecular differences between NM and SSM. Data suggest NM and SSM have distinct biology, which may justify different strategies of drug development. No significant financial relationships to disclose.