Abstract
MicroRNAs (miRNA) are small non-coding RNA molecules involved in post-transcriptional gene regulation. Deregulation of miRNA expression occurs in cancer, and miRNA expression profiles have been associated with diagnosis and prognosis in many cancers. Osteosarcoma (OS), an aggressive primary tumor of bone, affects ~10,000 dogs each year. Though survival has improved with the addition of chemotherapy, up to 80% of canine patients will succumb to metastatic disease. Reliable prognostic markers are lacking for this disease. miRNAs are attractive targets of biomarker discovery efforts due to their increased stability in easily obtained body fluids as well as within fixed tissue. Previous studies in our laboratory demonstrated that dysregulation of genes in aggressive canine OS tumors that participate in miRNA regulatory networks is reportedly disrupted in OS or other cancers. We utilized RT-qPCR in a 384-well-plate system to measure the relative expression of 190 miRNAs in 14 canine tumors from two cohorts of dogs with good or poor outcome (disease-free interval >300 or <100 days, respectively). Differential expression analysis in this subset guided the selection of candidate miRNAs in tumors and serum samples from larger groups of dogs. We ultimately identified a tumor-based three-miR Cox proportional hazards regression model and a serum-based two-miR model, each being able to distinguish patients with good and poor prognosis via Kaplan-Meier analysis with log rank test. Additionally, we integrated miRNA and gene expression data to identify potentially important miRNA-mRNA interactions that are disrupted in canine OS. Integrated analyses of miRNAs in the three-miR predictive model and disrupted genes from previous expression studies suggest the contribution of the primary tumor microenvironment to the metastatic phenotype of aggressive tumors.