Drosha is an RNAse III enzyme, which mediates the processing of pri-miRNAs into pre-miRNAs with DGCR8 (8). In a previous in vitro functional investigation, a reduction in miRNA processing efficacy, which was induced by the knockdown of DROSHA, was found to reduce the levels of mature forms of tumor-suppressive miRNAs and facilitate the invasion of breast cancer cells (17). Several studies have indicated the role of Drosha in breast cancer. A case-control study demonstrated that two SNPs in DROSHA, rs644236 and rs7737174, may contribute to the risk of breast cancer in postmenopausal women (18). Jiang et al also suggested rs2291109 as a predictor for breast cancer risk, however, the association was not confirmed (19). In addition, patients with breast cancer carrying the DROSHA rs874332 C allele are at increased risk of mortality (20). As rs874332 is located in the 3′-UTR of DROSHA mRNA and a predicted miRNA binding site, it is possible that rs874332 may be correlated with the translational repression and mRNA destabilization of DROSHA through an miRNA-mRNA interaction. However, data from Sung et al (21) involving east Asian women, including 5,066 cases and 4,337 controls, failed to identify an association between the SNPs in DROSHA and breast cancer risk. In addition, DROSHA rs6877842 has been reported to reduce the risk of recurrence in patients with renal cell carcinoma by 36%, and haplotypes of DROSHA (rs6877842/rs10719) have been associated with survival rates (22). rs10719 may also affect the risk of malignant peripheral nerve sheath tumors through increasing the expression level of DROSHA (23).

DGCR8, as a component of the multiprotein complex with the RNAse III enzyme, Drosha, is a double stranded RNA-binding protein, which is involved in the processing of pri-miRNAs into pre-miRNAs (9). Impaired miRNA processing through the knockdown of DGCR8 also facilitates the invasion of breast cancer cells (17). The rs9605062 in DGCR8 may upregulate the level or timing of gene expression (20), and it has been reported that rs9606250 is significantly associated with poor disease-free survival (DFS) rates in breast cancer (20). In addition, the interruption of miRNA binding of rs417309, located at the binding sites of miR-106b and miR-579 in the 3′-UTR of DGCR8, has been found to increased the risk of breast in the Chinese population (19). Another two linked SNPs, rs2073778 and rs720012, in DGCR8 have also been shown to be significantly associated with tumor progression in bladder cancer (24).

XPO5 is located in the nuclear membrane, and mediates the transport of pre-miRNAs to regulate miRNA expression (10). The XPO5-mediated nuclear export of pre-miRNAs may be a rate-limiting step in miRNA biogenesis. The over-expression of XPO5 has been shown to result in enhanced miRNA activity (25), whereas the loss of XPO5 leads to reduced expression and function of pre-miRNAs (26). Among the SNPs in XPO5, rs11077 has received the most attention. Located in the 3′-UTR of XPO5, rs11077 may affect mRNA stability, alter the expression of XPO5 and, consequently, affect the expression of miRNAs, including those specific for drug metabolism, altering the response to chemotherapy and affecting survival rates of patients with advanced non-small-cell lung cancer (NSCLC) and small-cell lung cancer (27,28). In addition, rs11077 has been associated with poor progression in hepatocellular carcinoma and renal cell carcinoma (22,29). A mutation in rs11544382 in a functionally conserved region of XPO5 may also alter the protein structure of XPO5, resulting in altered nucleocytoplasmic transport activity (30), and this SNP has been associated with an increased risk of breast cancer (30).

Dicer is an enzyme responsible for the cleavage of miRNA precursors, and has been implicated in the oncogenic process of several types of cancer. Increasing evidence has supported the role of DICER rs1057035 in cancer susceptibility. This SNP has been associated with a decreased risk of oral cancer (31), cervical carcinoma (32) and hepatocellular carcinoma (33). This polymorphism is located in the 3′-UTR of DICER and a predicted binding site of miR-574-3p, which may affect the binding of miR-574-3p, and result in decreased mRNA expression levels of DICER (31). Of note, this SNP has been shown to be associated with a 1.72- and 2.08-fold increased risk of progression and cancer-associated mortality, respectively, among patients with breast cancer (20).

The GEMIN4 protein is referred to as an important molecule in the RISC complex, which is involved in the maturation process of miRNAs, and the recognition and repression of target mRNAs (7). The protein expression level of GEMIN4 is closely associated with the biogenesis of associated miRNAs (7). Rs7813 in the exons of GEMIN4 has been frequently identified as a predictive biomarker in several types of cancer, including renal cell carcinoma (22), prostate cancer (34), malignant peripheral nerve sheath tumor (23) and ovarian cancer (35). Another non-synonymous SNP, rs2740348, which is located in the functional region of the GEMIN4 gene has been demonstrated to decrease the risks of prostate cancer and renal cell carcinoma by 36 and 33%, respectively (34,36). Notably, Wan et al found that rs2740348 and rs7813 were significantly associated with cell growth and DNA repair in a heptacellular carcinoma cell line (37), suggesting that the amino acid changes caused by these SNPs may have a physiological significance on the development of cancer.

AGO2 is important in miRNA-mediated gene silencing, as a component of the RISC complex that directly binds miRNAs and mediates the cleavage of target mRNAs (7). Emerging evidence from in vitro analysis and clinical samples has indicated that the abnormal expression or enzymatic function of AGO2 is associated with cancer development and progression. In breast cancer cell lines, the overexpression of AGO2 induces the transformed phenotype (38). Sung et al indicated that AGO2 rs3864659 may have a protective effect on breast cancer risk (18). In addition, two further SNPs in AGO2 rs11786030 and rs2292779 have been significantly associated with poor DFS and poor overall survival (OS) rates in breast cancer (20). Variations in the genomic structure of AGO2, including changes in copy number or frameshift mutations, have also been reported to be associated with several types of cancer, including multiple myeloma, gastric cancer and colorectal cancer (39,40).

The pre-miR-27a, rs895819, has been frequently investigated in the development of cancer, however, the results remain contradictory rather than conclusive (41–46). To integrate all individual studies and comprehensively analyze the role of rs895819 in tumorigenesis, several meta-analysis have been performed. Previous overall meta-analysis suggested no association between the pre-miR-27a rs895819 polymorphism and cancer susceptibility (47–49). In a stratified analysis, according to the type of cancer, individuals with the variant G allele were consistently found to be at a reduced risk of breast, renal cell and nasopharyngeal cancer, but at an increased risk of digestive tract cancer (47). In addition, subgroup analysis according to ethnicity revealed that the rs895819 AG genotype was associated with a decreased risk of cancer in Caucasian individuals (48). As this SNP is located at the terminal loop of pre-miR-27a, it may have an effect on the secondary structure of pre-miR-27a (42). The substitution of G for A in rs895819 may reduce the size of the loop and alter the minimum free energy, consequently inhibiting cleavage and resulting in low expression levels of mature miR-27a (42).

rs11614913 in the mature sequence of miR-196a2 has been increasingly identified a predictor for various types of cancer (50–52). The results from several meta-analyses, each containing thousands of subjects, have demonstrated that rs11614913 may contribute to the risk of developing breast cancer (53), lung cancer (54,55), hepatocellular carcinoma (56) and cancer of the digestive system (57,58). In addition, the SNP was positively correlated with improved recurrence-free survival (RFS) in patients with stage II and stage III NSCLC (59). The polymorphism may negatively affect endogenous processing of either miR-196a2 precursor to its mature form, and the levels of mature miR-196a2 are lower in CC carriers, compared with TT carriers (60). Furthermore, binding assays have revealed that this SNP can affect the binding of mature miR-196a2 to its target mRNA (60).

miR-146a, first identified in the mouse, has been shown to be important in tumorigenesis, by promoting cell proliferation and colony formation in NIH/3T3 cells (61). However, it also exhibits an antitumor property, by suppressing metastatic ability, in breast cancer and prostate cancer (62,63). The G-C substitution (rs2910164), located in the middle of the stem hairpin on the passenger strand of the precursor miR-146a, has a lower transcriptional activity due to decreased nuclear processing efficiency, leading to low level of mature miR-146a in cells (64). Although three meta-analyses consistently found that rs2910164 was not involved in overall cancer risk, stratified analysis by ethnicity has shown a close association between rs2910164 and overall cancer risk in the Caucasian population (50–52). Jiang et al reported that the rs2910164 GG homozygote was a protective genotype, in terms of susceptibility to acute-on-chronic hepatitis B liver failure (65). However, results from another meta-analysis showed the C variant to be associated with decreased hepatocellular carcinoma risk (56). As for patients with NSCLC, variants of rs2910164 were found to be positively correlated with RFS (59). However, in the development of cervical squamous cell carcinoma, the G allele of rs2910164 was associated with a significantly increased risk, as well as reduced tumor differentiation and a decline in lymph node status (66).

It is known that the secondary structure of miRNA is critical to mRNA-miRNA interactions and gene regulation (67). The rs3746444 polymorphism may affect miR-499 maturation and regulate the expression of its target genes through directly altering its secondary structure. Zhou et al provided evidence that rs3746444 may contribute to the susceptibility to cervical squamous cell carcinoma (66). Of note, Liu et al demonstrated that rs3746444 has a protective effect in the development of head and neck cancer (68). In addition, the T allele of rs3746444 was associated with a decreased risk of breast cancer among Asian individuals, however, a follow-up meta-analysis suggested risk was increased in Caucasians individuals, suggesting ethnic differences in the consequences of SNPs (69). A further meta-analysis failed to identify any significant correlation between the miR-499 polymorphism and risk of hepatocellular carcinoma (56).

The expression level of miR-218 is associated with infection with high-risk human papilloma virus (HPV), and is involved in the pathogenesis of cervical cancer (70). The rs11134527 in miR-218 has been shown to upregulate the expression of miR-218, and inhibit the expression of its target gene, LAMB3, by interfering with the mRNA-miRNA interaction. The overexpression of LAMB3 induces carcinogenesis by increasing carcinoma cell migration and disturbing tumor microenvironment, therefore, this polymorphism has been implicated in the infective process of high-risk HPV, thus contributing to cervical carcinogenesis (66,71). Another study evaluated the role of rs11134527 in hepatocellular carcinoma, which noted that the AG genotype of rs11134527 was associated with family history and elevated levels of serum α-fetoprotein, suggesting that the AG genotype may be associated with genetic predisposition in patients with hepatocellular carcinoma (72).

To date, the majority of the studies in this field are case-control studies, based on a candidate gene approach (15). Although several positive results have been reported, inconsistent findings and non-replication of previous results have frequently occurred (15). This may be attributed to several reasons, including the heterogeneity of patient groups, different experimental designs, insufficient sample size or unclear disease biology (15).

Heterogeneity in clinical confounding factors and endpoint phenotypes between initial and replication studies can undermine the opportunity to compare among them. It is essential to account for all confounding factors, which may predispose to a given phenotype, in order to estimate the residual phenotype that is likely due to genetics. As a small sample size can provide imprecise or incorrect estimates of the magnitude of an observed effect, sufficient sample size is necessary to accurately distinguish a suggested effect from a lack of effect (82). As several initial studies have been reported in populations of European descent, the challenge remains to extend investigations to include other ethnic populations (82).

Well-planned investigations are required, providing sufficient statistical power and stringency to detect and quantify a modest impact of the investigated SNPs. Follow-up epidemiological association investigations are important to validate previous findings in multiple independent large and homogenous samples. The National Cancer Institute-National Human Genome Research Institute working group on replication in association studies has published a comprehensive set of guidelines, providing a number of essential criteria for establishing positive replication studies (83).

In addition, further functional investigations are required to clarify the underlying mechanism. Several miRNAs are found in CpG islands, and miRNA expression can also be affected by DNA methylation and histone deacetylase inhibitors, providing another example of the bivalent roles of how miRNAs in malignancy (84). For example, two well-defined tumor suppressors, miR-124 and miR-34, are subject to epigenetic silencing by aberrant DNA hypermethylation, affecting cell cycle pathways in tumors, whereas the downregulation of miR-34 affects the Notch pathway, which is involved in cell invasion and apoptosis (85–87). Furthermore, DNA methylation profiles in miRNA promoter regions can be useful as a diagnostic and prognostic marker. For example, miR-23b, an miRNA with tumor suppressor activity in prostate cancer, is downregulated through DNA hypermethylation of its promoter region, and its expression level is correlated with OS and RFS (88). In addition to SNPs, structural variations, including insertions, deletions, inversions and copy number variants, with important implications on tumor biology (79,80). Evaluating the link among genetic variants, epigenetic modifications and disease predispositions is currently an active area of investigation (79,80).

Of note, there are several ways in which the processes of miRNA production, stability and maturation can be orchestrated (89–91). A semi-miRNA of12 nucleotides in length, which correspond to the 5′ region of the miRNA, let-7, is generated along the miRNA pathway, and may be involved in the control of gene expression by regulating the activity of mature miRNAs in vivo (92). Novel mechanisms for miRNA biogenesis have been described, and may be important as cancer drivers (93). Winter et al provided the first evidence that a small number of miRNAs are generated from single-stranded loop regions of human pre-miRNA hairpins, termed loop-miRs (94). In addition, an alternative miRNA processing pathway has been found in Drosophila melanogaster and Caenorhabditis elegans, which bypasses DROSHA and uses a splicing technique to generate miRNA precursors from short intronic sequences (95,96). The genetic polymorphisms and the functional implications of these novel pathways require further investigation.