Toward an miRNA signature for predicting outcome in patients with melanoma.

Abstract

9039 Background: Melanoma remains a clinical challenge, as an aggressive and often unpredictable tumor. The incidence continues to increase, but treatment options are limited. One significant challenge is predicting which patients with stage I or II disease will have a poor outcome. We sought to develop a tool to assist in risk stratifying patients with melanoma. Methods: This is a prospective trial in which patients with melanoma at least 0.5mm deep were enrolled between 2007-2011. Tissue was collected from primary tumors, bulky nodes and metastatic deposits. RNA was extracted, labeled and hybridized to a miRNA microarray (Affymetrix). Hierarchical cluster method was used to find similarities among samples. A clinical patient database was maintained. Results: This study looks at 58 intermediate and high risk subjects. The median depth of melanoma was 3mm. The median age was 69 (range 26-95) and most patients (48/58, 83%) were treated with surgery alone, with ten receiving adjuvant therapy. All patients underwent wide excision, and 43 patients had sentinel node biopsy. Twenty-seven patients developed a recurrence (47%), and 19 (33%) patients have died, 14 from melanoma. The median length of follow-up is 21 months (range 1-76 months). A total of 112 probes are differentially expressed at a p value of 0.005. The largest number of differentially expressed probes relate to location of the tumor. Other significant differences are origin of the specimen (primary versus node), mitotic rate and subject age. High expression of miR-193b was correlated with death from melanoma, p=0.05. Parameters with no differential expression include gender, tumor depth and subtype, and ulceration. Conclusions: Microarray analysis of miRNAs is a reliable platform to obtain prognostic information about melanoma. The miRNA signatures found in this patient population reinforce some well described differences among patients with melanoma, and fail to find a difference between other groups thought to be clinically important. Tumor genetic profiling may provide better risk stratification of patients with melanoma than clinical parameters alone.