Stehn, Christopher; Colman, Howard; Boucher, Kenneth; Grossman, Allie H; Holmen, Sheri L (University of Utah)
Faculty Advisor: Holmen, Sheri (University of Utah, Surgery)
Despite therapeutic advances in the treatment of melanoma, development of brain metastases continues to be a major cause of treatment failure. Prognosis for patients with brain metastases is exceedingly poor, therefore the development of sensitive and specific biomarkers to predict which melanoma patients are at highest risk for disease progression are needed. To accomplish this goal, we developed a novel combined molecular/clinical/pathologic predictor of brain metastasis risk. We first analyzed multiple gene expression datasets including The Cancer Genome Atlas (TCGA; n = 437) and an independent series from the European Genome-Phenome Archive (n = 183) and identified a list of 60 consensus genes that is robustly predictive of development of melanoma brain metastases (p < 0.05; FDR 5%). Next, we performed a similar analysis of association of miRNAs and melanoma brain metastasis risk which identified a set of miRNAs with significant predictive power. An optimized combined set of 15 mRNA and miRNA markers was a better predictor of brain metastasis risk than either mRNA or miRNA list alone when applied to the TCGA data set. The combined predictor was most sensitive in separating patients with no metastases from those with either brain metastases or systemic metastases. Current efforts are focused on optimizing miRNA and mRNA separation of patients specifically with brain metastases from those with other metastases using a machine learning linear classifier, and with integrating the expression classifier with other clinical and pathologic predictive factors including: age, stage, thickness, location, histology, ulceration, and gender. The sensitivity and specificity of the resulting clinical/molecular predictor will be validated in an independent retrospective patient dataset, and subsequently implemented in a prospective brain metastasis screening trial to determine real-world utility of this approach in preparation for prospective brain metastasis adjuvant/chemoprevention trials utilizing both immunotherapy and targeted therapy approaches.