SAT-041 Bioinformatic Analysis of Human Cutaneous Melanomas Identifies Increased Mortality and Genetic Alterations in GH-responsive Tumors

Abstract

The impact of growth hormone (GH) on systemic glucose, growth and the myriad of downstream targets indicate the potential to enhance the progression of cancers. There is, however, continued debate as to GH’s impact on cancer with varied effects across subjects and cancer types. To bridge this gap in fundamental understanding, 16 human cancer types were selected and the RNA expression data for all matching samples (primary or metastatic) were queried in The Cancer Genome Atlas (TCGA). Each cancer type was stratified into GH-responsive [GH receptor (R) expression in the top quartile of each cancer type] and baseline (the remaining 75% of samples). Patients with GH-responsive cutaneous melanomas had a significant reduction in post-diagnosis lifespan (log-rank test, P = 0.0073). To explore the deleterious effects of GH in melanoma progression, we developed a novel tool, gdc-file-matcher (https://github.com/pxslip/gdc-file-matcher), to extract dataset metadata. Using the extensive metadata available in TCGA, we identified a significant negative correlation (Pr(>|t|) = 0.021) between GHR expression and post-diagnosis lifespan in male melanoma patients who died within 1 year of diagnosis that drove the reduction in lifespan. The same trend, though not significant, was also observed in female melanoma patients. To validate the stratification approach, differential expression profiles were generated using a generalized linear model in EdgeR. The expression profiles were validated against expression levels of adiponectin, multiple ABC transporters, and metastatic markers in implanted B16-F10 melanomas in wild-type C57BL6J mice and transgenic bGH mice. The expression profiles of the implanted in vivo melanoma samples validated the stratification approach as indicative of GH action in human melanomas. To understand the molecular mechanisms resulting in increased mortality rates of GH-responsive tumors, the differentially expressed genes were investigated. The gene expression profiles showed clear evidence for a greater metastatic potential in GH-responsive tumors. In addition, upregulated genes showed an increased potential for angiogenesis and drug resistance along with a downregulation of DNA repair genes, P53 signaling cascades and other protective pathways. Thus, we demonstrate induced melanomas in the C57BL6J background are a reliable mouse model for translational research in GH-related cancer and are likely able to be expanded as a model for additional melanoma therapy research. Most critically, the survival and gene expression data from this in vivo validated informatic analysis demonstrate a clear role for Pegvisomant or alternate forms of GH modulation in the treatment of GH-responsive melanomas.