MicroRNA-binding site polymorphisms in genes involved in colorectal cancer etiopathogenesis and their impact on disease prognosis.

According to the Vogelstein's model of colorectal carcinogenesis, genetic variations in highly penetrant genes may be involved in the colorectal cancer (CRC) pathogenesis. Similarly, aberrant function and/or altered expression of microRNAs (miRNAs) often occur in CRC. In this context, polymorphisms in miRNA-binding sites (miRSNPs) may affect miRNA/target gene interaction, resulting in differential mRNA/protein expression and increased susceptibility to common diseases. To explore this phenomenon, we have mined the 3' untranslated regions (3'UTRs) of genes known to be frequently mutated in CRC to search for miRSNPs and tested their association with CRC risk and clinical outcome. Eight miRSNPs (rs1804191, rs397768, rs41116 in APC; rs1137918, s227091, rs4585 in ATM; rs712, rs1137282, rs61764370 in KRAS; rs8674 in PARP1 and rs16950113 in SMAD7) were tested for their association with CRC risk in a case-control study (1111 cases and 1469 healthy controls). The role of these miRSNPs was also investigated in relation to clinical outcome on a subset of patients with complete follow-up. rs8679 within PARP1 was associated with CRC risk and patients' survival. In the dominant model, carriers of at least one C allele were at a decreased risk of cancer (P = 0.05). The CC genotype in rs8679 was also associated with an increased risk of recurrence/progression in patients that received 5-FU-based chemotherapy (log-rank test P = 0.03). Carriers of the homozygous variant genotype TT for rs712 in KRAS gene were associated with a decreased risk of rectal cancer (odds ratio (OR) = 0.65, 95% confidence intervals (CI) 0.43-1.00, P = 0.05) while individuals with colon cancer carrying the heterozygous GT genotype showed a longer overall survival (OS) (P = 0.04). We provide the first evidence that variations in potential miRNA-binding target sites in the 3' UTR of PARP1 gene may modulate CRC risk and prognosis after therapy. Further studies are needed to replicate our finding and assess miRSNPs as predictive biomarkers in independent populations.