Single-nucleotide polymorphisms (SNPs) in 3'-untranslated regions of cancer-related genes might affect regulation by microRNAs and contribute to cancer patients' outcome.

We used public databases to identify SNPs within miRNA-binding sites in deregulated genes in non-small cell lung cancer (NSCLC). A total of 13 SNPs in 10 genes were included and genotyped by SNaPshot assay in 576 NSCLC patients. Associations between SNPs, overall survival (OS) and chemotherapy response were evaluated by Cox regression and logistic regression. We then examined the functionality of the significant polymorphisms.

Two SNPs (TYMS rs2790 and MICA rs9266825) were significantly associated with OS. In the combined analysis, an increasing number of unfavorable loci was associated with a poorer prognosis (P for trend<0.001) and patients having 2-3 unfavorable loci had a 1.61-fold elevated risk of death (95% confidence interval: 1.20-2.15), compared with those carrying 0-1 unfavorable loci. A significant effect of SNPs on platinum-based chemotherapy response was observed among 296 advanced NSCLC patients without surgical operation: rs2790, rs4246215 and rs1882. Further analysis using mRNA expression data from the HapMap suggested that these significant loci (FEN1 rs4246215, HDAC2 rs11391, MICA rs1882 and rs9266825) were closely associated with host genes expression. In vitro functional study for TYMS rs2790 was carried out. Luciferase assay showed a lower expression level for rs2790 G allele as compared with A allele, and the hsa-miR-1248 had an effect on modulation of TYMS gene.

Our data indicate that miRNA-binding site SNPs in deregulated genes may serve as candidate prognostic markers of NSCLC clinical outcome.