A Genetic Polymorphism in pre-miR-27a Confers Clinical Outcome of Non-Small Cell Lung Cancer in a Chinese Population

Background Recent evidence indicates that microRNAs (miRNAs) can function as tumor suppressors and oncogenes. Single nucleotide polymorphisms (SNPs) at miRNA genes can influence the maturation of miRNAs or miRNA-mediated transcriptional regulation. Our objective was to investigate the association of SNPs in deregulated miRNAs with clinical outcome in patients with non-small cell lung cancer (NSCLC) in a Chinese population. Methods Deregulated miRNAs in NSCLC and their SNPs were identified through public databases. A single SNP, rs895819 in pre-miR-27a, was found suitable for selection. TaqMan assays were performed for genotyping and to assess the effect on the overall survival (OS) and chemotherapy response in 576 NSCLC patients. Results Log-rank test and Cox regression analysis indicated that the G allele of rs895819 was associated with shorter survival and increased risk of death in NSCLC [dominant model: 22.0 vs. 46.0 months, P<0.001; adjusted hazard ratio (HR) = 1.71, 95% confidential interval (CI): 1.12–2.26]. Further stepwise regression analysis suggested that this SNP was an independently unfavorable factor for the prognosis of NSCLC and the effect remained significant in subgroup analysis stratified by clinical parameters and treatment status. Moreover, multivariate logistic regression analysis showed that the subjects with AG/GG genotypes of rs895819 had significantly decreased response rate to platinum-based chemotherapy compared to those with the AA genotype. Conclusion Our results suggest that the pre-miR-27a rs895819 polymorphism may influence NSCLC patients’ clinical outcome. Further large sample studies should be used to validate our findings.


Introduction
Lung cancer is the leading cause of cancer-related deaths worldwide, due to its high incidence, malignant behavior and lack of effective in treatment strategy [1]. In China, there has been a significant increase in the incidence of lung cancer in both urban and rural areas over the last two decades, especially in non-small cell lung cancer (NSCLC) [2,3]. Although platinum-based combination chemotherapy is the first-line treatment for patients with advanced NSCLC, the response rates vary among patients, ranging from 26% to 60% [4]. The five-year over survival (OS) rate is only 15% in the US and is even lower in China [1]. Established methods for predicting prognosis include the tumor, node, and metastasis (TNM) staging system, however, recent studies indicate that the discovery and application of specific prognostic biomarkers can improve the treatment and outcome of patients with NSCLC [5]. Despite a considerable amount of research [6,7,8], very few stable biomarkers have been identified for risk assessment or predication of clinical outcome and further investigations are necessary.
MicroRNAs (miRNAs) are a class of small endogenous noncoding RNAs that act as negative gene regulators by directly cleaving target mRNA or by inhibiting translation [9,10]. The loss and gain of function of specific miRNAs or aberrant miRNA expression are thought to be key events in the tumorigenesis of many types of cancer [11,12]. In NSCLC, miRNA expression profiles and specific miRNAs have been correlated with patients' survival [13,14]. In addition, there is increasing evidence that single nucleotide polymorphisms (SNPs) play a significant role in cancer susceptibility and outcome. The high degree of phylogenetic conservation in miRNA sequences determines that the functional genetic variations in miRNAs may affect various biological processes. Therefore, SNPs in miRNA genes could influence the primary transcripts (pri-miRNAs), precursor RNAs (pre-miRNAs), mature miRNAs or miRNA-mediated transcriptional regulation [15].
We utilized public databases to identify deregulated miRNAs in NSCLC and SNPs in these miRNAs sequences, including primary, precursor and mature miRNAs. Our searches identified a single SNP, rs895819 in pre-miR-27a, for further investigation. The rs895819 polymorphism is located in the terminal loop of pre-miR-27a and involves an A.G nucleotide transition. Hsa-miR-27a has been shown to function as an oncogene by targeting prohibitin [16] and is therefore a target for anticancer drugs [17,18]. Metens-Talcott et al. [19] showed that transfection of antisense miR-27a in breast cancer cells led to an increase in expression of the putative Sp-repressor Zinc finger and BTB domain containing 10 (ZBTB10), with a concurrent decrease in the expression of Spdependent survival and angiogenic genes. These include vascular endothelial growth factor (VEGF) and survivin [19]. Overexpression of VEGF and surviving are associated with progression or poor survival of NSCLC [20,21]. Sun et al. [22] reported rs895819 modifies gastric cancer susceptibility by modulating the expression levels of miR-27a and its target gene, ZBTB10. Shi et al. [23] suggested that this polymorphism could predict renal cell cancer risk in a Chinese population. Meta-analysis has shown that the pre-mir-27a polymorphism may play a role in breast cancer susceptibility and cancer development in Caucasian [24].
To date, there have been no reported studies on the relationship between pre-mir-27a polymorphisms and survival in cancer patients. Based on current knowledge of the biological functions of hsa-miR-27a and the role of polymorphisms in predicting cancer risk, we hypothesized that rs895819 might be associated with clinical outcome in NSCLC patients. Therefore, we evaluated the associations between rs895819 and NSCLC overall survival (OS), as well as response to platinum-based chemotherapy.

Ethics Statement
This study was approved by the institutional review board of Nanjing Medical University. All participants were voluntary and provided written informed consent prior to taking part in this research.

Study Subjects
All subjects were recruited from the First Affiliated Hospital of Nanjing Medical University (Jiangsu, China) between January 2004 and September 2012. They were all newly diagnosed, histopathologically confirmed and without a prior history of cancer or previous chemo-or radiotherapy. In total, 612 patients with NSCLC were recruited, all of whom were unrelated ethnic Han Chinese population (CHB). A structured questionnaire on demographics and environmental exposure, including age, sex and smoking consumption, was conducted by trained interviewers through face-to-face interviews with the patients. In addition, 5 ml venous blood was collected from each patient for genomic DNA extraction. Subjects with a low frequency (,1 cigarette per day) and duration (,1 year) of smoking were defined as nonsmokers; all others were classified as smokers.
The response to platinum-based (cisplatin or carboplatin) chemotherapy in patients with advanced NSCLC was assessed following the the first two or three cycles and defined according to Response Evaluation Criteria in Solid Tumors (RECIST) criteria 1.1 [25]. The patients were divided into the following two groups based on their responses: those with complete response (CR) and partial response (PR) were classed as responders; those with stable disease (SD) and progressive disease (PD) were classed as nonresponders. Follow-up was performed every three months from the time of enrollment until death or the last scheduled follow-up. The maximun follow-up time was 102.0 months (last follow-up in February 2013) and the medial follow-up time was 18.0 months. We selected the patients with complete follow-ups and adequate DNA sample. Patients who completed all the follow-ups and provided adequate DNA sample were selected. Of the original 612 patients, 576 were enrolled and genotyped in our study.

Genotyping
Each blood sample was collected in an EDTA anticoagulant tube and stored at 280uC until DNA extraction. Genomic DNA was extracted following the standard protocols, with proteinase K digestion and phenol/chloroform extraction. Genotyping was performed by TaqMan allelic discrimination assays using an ABI 7900 system (Applied Biosystems, Foster City, CA, USA). The prime and probes are as follows: forward, GGCGGAACTTAGC-CACTGT, reverse, CAGGGCTTAGCTGCTTGTG; FAM-ACTTGGTGTGGACC-MGB and HEX-ACTTGGCGTG-GAC-MGB. The TaqMan assays were performed in a final reaction volume of 5 ml containing 0.25 ml primer, 0.125 ml probe, 2 ml PCR mixture reagent and 25 ng DNA. The PCR reaction consisted of an initial step at 95uC for 10 min followed by 55 cycles of denaturing at 95uC for 15 s and annealing at 60uC for 60 ss. SDS allelic discrimination software (ABI) was used to analyze the PCR genotyping results. Two blank (water) controls were included in each 384-well assay. At least 10% of samples were randomly selected for repeat analysis, yielding 100% concordance. A further 60 samples were selected randomly for direct sequencing to confirm the TaqMan results. Again, the results showed 100% concordance.

Statistical Analysis
Hardy-Weinberg equilibrium was assessed by a goodness-of-fit x 2 test. OS was calculated as the time between the first treatment and death or the last follow-up date. Association between the genotype and survival rate was estimated by the Kaplan-Meier method and log-rank test. The median survival time (MST) was also calculated. Cox proportional hazards models were performed to estimate the hazard ratios (HRs) and their 95% confidential intervals (CIs) for OS. Cox stepwise regression model was also conducted to determine predictive factors to NSCLC prognosis. The P value for the heterogeneity test was based on the x 2 -based Q test. Odds ratios (ORs) and their 95% CIs were calculated as a measure of difference in the response rate using logistic regression analysis (responders vs. non-responders). All statistical analyses were performed using SPSS 18.0 software (SPSS Inc., Chicago, Illinois, USA), and P,0.05 based on a two-side test was considered statistically significant.

Characteristics of the Study Population
The demographic and clinical characteristics of the 576 patients enrolled in this study, and their association with OS are given in Table S1. The median age at diagnosis was 60 years (range, 29-86 years), and included 380 males (66.0%) and 267 smokers (46.4%). Among these patients, 381 (66.2%) were adenocarcinomas, 166 (28.8%) were squamous cell carcinomas, and the others (29 patients, 5.0%) were large cell, undifferentiated or mixed-cell carcinomas. During the follow-up period, 206 patients died from pre-miR-27a Polymorphism and NSCLC Outcome NSCLC. Smoking status, clinical stage and surgical operation were significantly associated with survival time (all log-rank P,0.001, Figure 1); however, chemotherapy and targeted therapy were not. Of note, patients with diabetes had a 42% significantly decreased risk of death, compared with those without diabetes (54.9 vs. 42.0 months; HR = 0.58, 95% CI: 0.35-0.97).

SNP Selection
The IGDB.NSCLC database (http://igdb.nsclc.ibms.sinica. edu.tw) [26] was used to identify deregulated miRNAs in NSCLC. A total of 28 up-regulated miRNAs in amplified regions and 28 down-regulated miRNAs in deleted regions were found. Patrocles database (www.patrocles.org) [27] identified 5 miRNAs with 5 putative SNPs within the pre-miRNAs or mature miRNAs sequences of these 56 miRNAs (Table 1). These included rs2043556 within the pre-miR-605 sequence and rs895819 within pre-miR-27a Polymorphism and NSCLC Outcome the pre-miR-27a sequence, which both had a minor allelic frequency (MAF) .5% in CHB. The rs2043556 was excluded because the high GC content can lead to failure in the TaqMan probe assay. Therefore, only one SNP, rs895819 A.G, was finally selected.

Association between the pre-miR-27a Polymorphism and Survival in NSCLC
The genotype frequencies of the rs895819 were in Hardy-Weinberg equilibrium (P = 0.538). The Log-rank test showed a significant association between rs895819 and survival in patients with NSCLC in the codominant and dominant models (both P,0.001; Figure 1 Table 2).
In order to identify independent prognostic factors for NSCLC survival, we performed multivariate stepwise Cox regression analysis using selected demographic characteristics, clinical features and the pre-miR-27a polymorphism. The final predictive model included the SNP rs895819 (P,0.001), along with smoking status (P = 0.018), diabetes mellitus (P = 0.042) and surgery (P,0.001) ( Table 3).

Stratification and Interaction Analysis
The association between pre-miR-27a polymorphism and NSCLC survival was further evaluated by stratified analysis of age, sex, smoking status, diabetes mellitus, histology, clinical stage, surgical operation, chemotherapy and targeted therapy. As shown in Table 4, an increased risk of death was observed in the following

Association between the pre-miR-27a Polymorphism and Response to Platinum-based Chemotherapy
The association between rs895819 and treatment response was further evaluated. Only the 296 patients with inoperable advanced stage NSCLC (IIIb-IV), who had received platinum-based chemotherapy as first-line therapy, and had a complete evaluation of chemotherapy response, were included to avoid potential confounding effects from surgery, clinical stage and treatment type. Detailed treatment information is provided in Table S2. The multivariate logistic regression model showed that patients with a G allele of rs895819 had significantly decreased response rate to platinum-based chemotherapy compared to those with the AA genotype (AG/GG vs. AA: adjusted OR = 0.54, 95% CI = 0.32-0.91) and consequently, an increased risk of death (adjusted HR = 1.59, 95% CI: 1.09-2.31). We detected the similar results using an additive model (adjusted OR = 0.54, 95% CI = 0.35-0.83; adjusted HR = 1.59, 95% CI: 0.97-1.63; Table 5).

Discussion
In the present study, we found that the variant allele G of rs895819 within pre-miR-27a was associated with a significantly increased risk of deaths for NSCLC. The effect was even stronger in older subjects, males, those with adenocarcinomas, without diabetes, and in those who did not undergo surgical operation and have received chemotherapy. Furthermore, a significantly decreased response rate to platinum-based chemotherapy was observed in advanced NSCLC patients with AG/GG genotypes. This is the first reported study on the association between the pre-miR-27a rs895819 polymorphism and the clinical outcome in patients with NSCLC.
Although miRNAs represent only a small part of the genome, they regulate almost one-third of human genes [28]. An increasing number of studies have indicated that miRNAs are involved in critical biological processes, including development, differentiation, apoptosis and proliferation, and may therefore play important roles in carcinogenesis as tumor oncogenes or suppressors [29,30]. Genetic polymorphisms in miRNAs have been reported to be associated with many tumors, including gastric cancer [31], breast cancer [32] and colorectal cancer [33]. Hu et al. [14] found that a variant homozygote of rs11614913 in pre-miR-196a2 was associated with poor survival in patients with NSCLC. Hong et al. [34] suggested that the pre-miR-149 rs2292832 T.C polymorphisms were significantly associated with disease-free survival and OS in NSCLC. Moreover, genetic variations within the miRNA-machinery genes or miRNA-binding sites in 39-untranslated region of target genes have also been reported to be associated with cancer susceptibility or prognosis [35,36,37].
It has been suggested that a genetic variation causing structural change in a critical region of miRNA may affect the process and the maturation of miRNA [38,39]. The pre-miR-196a2 rs11614913 C allele was associated with significantly increased expression of mature miR-196a and could therefore affect the binding of miR-196a2 to its target mRNA LSP1 [14,40]. The SNP rs11671784 has been reported to impair the maturation of miR-27a, resulting in reduced expression of mature miR-27a and an increased level of its target HOXA10 [41]. The SNP rs895819 has also been indentified in pre-miRNA regions of hsa-miR-27a, with a MAF .5% in CHB. Preliminary functional assays by Sun et al. [22] revealed that the rs895819 G allele conferred a higher level of miR-27a, which accompanied significantly reduced ZBTB10 mRNA. We used the RNAfold program to predict the most stable secondary structures of pre-miR-27a with two different alleles. However, neither a conformational effect nor a free energy effect was predicted (data not shown). In contrast, on the basis of the model of Zeng and Cullen, mutations in the stem of pre-miRNAs had a marked effect on miRNAs processing [39]. Although the SNP within miR-27a was located in the terminal loop, the mutant within terminal loop may slightly impair pri-miRNA processing and DGCR8 binding based on the ssRNA-dsRNA Junction Anchoring Model [42]. Such a mechanism has been demonstrated for hnRNP A1 [43].
We found that the G allele of rs895819 within pre-miR-27a was associated with a significantly decreased response rate to platinumbased chemotherapy and a significantly increased risk of deaths for NSCLC. Hsa-miR-27a has been found to be overexpressed in several types of tumors, including breast cancer [44] and ovarian cancer [45]. It has been shown to function as an oncogene by targeting prohibitin [16] and is a target for anticancer drugs [17,18]. Down-regulated miR-27a might can also reverse multidrug resistance of esophageal squamous cell carcinoma [46]. Although the expression of miR-27a has been reported to be significantly reduced in both tissue and serum samples of NSCLC patients [26,47], a relatively higher level of miR-27a expression was observed with the rs895819 G allele compared to the A allele [22]. These evidence may partly support our findings. Further indepth functional studies are required to elucidate the mechanism of miR-27a and this variant in NSCLC.
We also observed an increase in OS in patients with both NSCLC and diabetes mellitus compared to those without diabetes (54.9 vs. 42.0 months; P = 0.034). Multivariate stepwise Cox regression analysis showed that this was an independent prognostic factor for NSCLC, which is consistent with several cohort studies in Caucasians [48,49]. Oral antidiabetic drugs, especially metformin, are commonly administered as front-line treatment for patients with diabetes. Both basic and animal experiments have indicated that a metformin-mediated reduction in insulin resistance may be related to a reduced risk of tumor development [50,51,52]. Metformin administered alone or in combination with chemotherapy has also been reported to block tumor growth in vitro and in vivo experiments [53,54,55]. The findings of Tan et al. [56] demonstrated that metformin may improve chemotherapy outcomes and survival for NSCLC patients with diabetes. These studies support our results. When interpreting our results, several limitations should be concerned. Firstly, all of the subjects were recruited from a single institution, which can introduce selection bias. Nonetheless, the genotype distribution was in Hardy-Weinberg equilibrium. Secondly, our sample size was relatively small. However, we achieved 89% study power (two-sided test, a = 0.05) to detect an HR of 1.71 for the rs895819 G genotype in dominant model. Thirdly, due to lack of lung cancer tissue corresponding to the blood specimen analyzed, we were unable to detect the expression of miR-27a and its targets, and explore the association between them and SNP rs895819. Despite these limitations, this is the first study to investigate the association between the pre-miR-27a rs895819 polymorphism and clinical outcome of NSCLC, and provided valuable information which might serve to guide future studies and clinical practice.

Conclusions
In conclusion, our results provided the first insight into the contribution of pre-miR-27a rs895819 polymorphism to the clinical outcome of NSCLC. The findings further highlight that polymorphisms in miRNA sequences may play an important role in lung cancer. Although the association appeared to be statistically significant in the present study, the findings should be further validated by large, well-designed studies.