The role of microRNA-608 polymorphism on the susceptibility and survival of cancer: a meta-analysis

The role of rs4919510 polymorphism in microRNA-608 (miR-608) and cancer susceptibility and prognosis remain controversial and debatable. We conducted a meta-analysis of twenty-four eligible publications on the association of rs4919510 polymorphism with cancer risk and/or prognosis. Odds ratios, hazard ratios, and 95% confidence interval were used to investigate the association between this polymorphism and susceptibility, overall survival, and recurrence-free survival of cancer. Overall, eighteen case-control studies and nine cohort studies evaluated the susceptibility and prognostic value of rs4919510 polymorphism in cancer, respectively. Pooled analysis showed that rs4919510 polymorphism was not associated with cancer risk in all five genetic models. When stratifying by different cancer sites, rs4919510 polymorphism was detected to have a significant association with a decreased risk of colorectal cancer in homozygous model (P = 0.006) and recessive model (P = 0.001), subgroup analysis also emerged a weakened correlation between rs4919510 polymorphism and an increased risk of papillary thyroid cancer in heterozygote model (P = 0.04). Furthermore, the prognosis of rs4919510 variant in cancer patients showed that rs4919510 GG genotype was significant association with poor recurrence-free survival in homozygous models (P = 0.04). The meta-analysis suggested that the microRNA-608 rs4919510 polymorphism maybe associate with a significantly decreased risk for colorectal cancer. Further investigations on larger populations are required to evaluate and confirm this relationship.

AGING sing, or by affecting the miRNA-mRNA interactions [7]. Therefore, SNPs in miRNAs may be considered as biomarkers for the diagnosis and/or prognosis of cancer.
The microRNA-608 (miR-608) gene lies within an intron of SEMA4G on the human chromosome 10q24 locus. Due to the loss of heterozygosity, the 10q24 locus has been reported to be correlated with certain types of human cancers, including brain, colorectal, prostate, and breast cancer [8][9][10][11]. The rs4919510 polymorphism, located in the mature miR-608 sequence, is a C/G single-nucleotide variation. The rs4919510 variation was considered to influence miRNA activities by binding to mir608 target sites within CD4 antigen, growth hormone receptor (GHR), retinoic X receptor beta (RXRB), and tumour protein p53 (TP53) genes with lower free energies than the wild type allele [12].
It has been hypothesised that rs4919510 polymorphism in miR-608 plays a part in carcinogenesis and clinic outcomes of cancer. Recently, several studies have revealed that the rs4919510 variation in the miR-608 gene could influence the susceptibility and/or prognosis of certain types of cancer, including colorectal [10,[13][14][15], breast [11,[16][17][18][19], thyroid [20,21], gastric [22,23], lung [24,25], head and neck cancers [26][27][28], and neuroblastoma [29], as well as esophageal squamous cell carcinoma [30], and hepatocellular carcinomas [31,32]. However, the results of the previous studies are inconsistent. In addition, an earlier meta-analysis found that rs4919510 polymorphism in miR-608 was significantly associated with decreased cancer risk in recessive model [33]. In recent years, there was a lot of new literature published. In this study, a phenotype meta-analysis was performed to evaluate the effects of miR-608 rs4919510 genetic polymorphism on susceptibility and prognosis of different types of cancer.

Characteristics of eligible studies
Through the primary literature retrieval in Pubmed, Embase, and CNKI, 90 studies were identified for cancer susceptibility and/or prognosis assessment based on rs4919510 polymorphism of miR-608 gene. After review of titles and abstracts, we identified 39 potential articles eligible to be included for the evaluation. After retrieving the literature, two overlapping studies were found [25,34], and we excluded the earlier study. Other two of the selected studies did not meet the requirements of Hardy-Weinberg Equilibrium (HWE) [35,36], and hence twenty-four articles were assessed for eligibility to be included in our meta-analysis. The flow chart for the studies selection process is shown in Fig. 1.
There were two separate groups from different ethnicities (African-Americans and Caucasians) in Ryan et al's study [10]. Thus, we treated them as separate studies. Another report by Zheng et al. [27] also included two independent studies of prognosis. Finally, eighteen independent case-control studies from seventeen published articles were evaluated for the association of rs4919510 polymorphism with cancer risk [10, 11, 13, 15, 17-23, 25, 26, 28-31] and nine studies from seven published articles were evaluated for the prognostic value of rs4919510 polymorphism in cancer [10, 14-16, 24, 27, 32]. All eligible studies were published between 2012 and 2018 and carried out in the USA, China, Chile, Iran, and Lithuania. The studies presented data for several types of cancer including colorectal, gastric, lung, breast, thyroid, neuroblastoma and head and neck cancers as well as hepatocellular, and oesophageal squamous cell carcinomas. The characteristics of the eligible studies are presented in Table 1 and Table 2.

The association between rs4919510 polymorphism and cancer susceptibility
Overall, there were eighteen studies encompassing 10,345 cases and 14,160 controls included in our metaanalysis. The associations of the rs4919510 polymorphism in miR-608 with the risk of different types of cancers are shown in Table 3. No statistically significant association was observed in the overall population under all comparison models. In the subgroup analysis of ethnicity, a significant association with increased cancer risk was found for Asians under heterozygote comparison (OR = 1.08, 95% CI = 1.01-1.17, Fig. 2). When the analysis was stratified by different cancer types, rs4919510 polymorphism was detected significant association with a decreased colorectal cancer risk in two genetic models (GG vs. CC: OR= 0.74, 95% CI = 0.60-0.92, Fig. 3; GG vs. CC/CG: OR= 0.73, 95% CI = 0.61-0.88). Moreover, significant associations were observed between rs4919510 polymorphism and papillary thyroid cancer in heterozygote model (OR = 1.25, 95% CI = 1.01-1.54).

The association between rs4919510 polymorphism and prognostic significance of cancer
Through retrieval in databases, nine studies reporting the association between rs4919510 polymorphism and the prognostic value of cancer were available. Survival data of the studies included our meta-analysis are summarised in Table 4. No significant association was observed between rs4919510 polymorphism and overall survival of cancer under univariate analysis or multivariate analysis. A significant result existing in AGING homozygote comparison model indicates the GG allele maybe a significantly risk role in recurrence-free survival of cancer (GG vs CC: HR = 1.51, 95% CI = 1.02-2.23). However, the negative association was observed under multivariate analysis in homozygote model.

Publication bias
In our meta-analysis, both Begg's funnel plot and Egger's test were conducted to assess the publication bias. As shown in Fig. 4, the results revealed no publication bias for eligible studies under heterozygous model. The symmetry was then confirmed in Egger's linear regression test (P = 0.21).

Heterogeneity and sensitivity analysis
Significant heterogeneities in the data of microRNA-608 rs4919510 polymorphism were observed in the overall meta-analysis as well as subgroup analysis, the results are shown in Table 3. Sources of heterogeneity were further studied by sensitivity analysis, the result showed that no individual study influenced the pooled OR value (Fig. 5), indicating that the results of our meta-analysis are statistically credible and robust.

DISCUSSION
Aberrant miRNA function and expression could influence multiple physiological processes and participate in the occurrence, development, and prognosis of many human diseases including malignant tumours [7,37]. Furthermore, SNPs or mutations in miRNAs may directly exert their effects by binding to target mRNAs [38]. The rs4919510 polymorphism was located in the sequence of mature miR-608 and its G/C variation was common in several populations. A previous report found that the variant miR-608 was predicted to bind to mir608 target sites within immunity and defense, DNA repair, cell growth-related, and cell death-related genes with lower free energies than that of AGING the wild type allele, for example, CD4, GHR, RXRB, and TP53 [12]. Therefore, it was assumed that the rs4919510 polymorphism possibly elicits different biological activities of these target mRNAs and is associated with the pathogenesis and prognosis of cancer as a predictive biomarker.
Our meta-analysis results showed that no significant association was observed between rs4919510 variant in miR-608 and risk of cancer in the overall population based on seventeen eligible case-control studies. However, an association was found between rs4919510 polymorphism and cancer risk in an Asian population based on the heterozygote model (P = 0.03), meanwhile, similar positive results were observed in PTC patient (P = 0.04). Therefore, rs4919510 polymorphism might be potential indicators of PTC risk.
A previous study found that rs4919510 polymorphism was not significantly associated with BC risk, but variant genotypes (GG/CG) influenced HER2-positive BC risk [11]. The results our own group proved rs4919510 alleles of miR-608 were not associated with the presence of BC a Northern Chinese population [17]. Negative results have also been reported in Chile populations [19]. Conversely, another research results revealed that rs4919510 GC genotype significantly decreased the risk of BC compared to CC genotype in an Iran population [18]. In a published pooled analysis, Wang et al [39] reported that there is no relationship between rs4919510 polymorphism and BC risk, our findings were partially in line with results from this meta-analysis.
Ryan et al. [10] and Kupcinskashas et al.'s [13] studies indicated no significant associations between rs4919510 polymorphism and CRC risk in Caucasian and African American populations. Opposite results were proved by Ying et al. who found that rs4919510 GG genotype was significantly associated with a decreased susceptibility to colorectal cancer in homozygous and recessive genetic models [15]. Remarkably, our meta-analysis results showed that rs4919510 GG genotype was significantly associated with a decreased susceptibility to CRC in homozygous and recessive genetic models.

AGING
The results suggested that the presence of GG models might play a protect role in tumorigenesis in colorectal cancer. The strength of his conclusion should be confirmed on the base of more studies and sample size.
Qiu et al. [26] showed that carriers of rs4919510 CC genotypes had a reduced risk of nasopharyngeal carcinoma versus individuals who carried GG/CG genotypes in Chinese population. However, negative results were also reported in Chinese populations by Miao et al. [28]. Moreover, null significant association between this polymorphism and gastric cancer risk in Kupcinskas et al. [22] and Jiang et al.'s [23] studies. The other studies by Wang et al. [31] and Zhang et al. [30] showed that rs4919510 polymorphism was not associated with HCC and ESCC susceptibility respectively. AGING Table 3. Meta-analysis results for the overall and subgroup analyses of rs4919510 polymorphism and cancer susceptibility.

AGING
To date, there are three meta-analyses concerning the correlation of rs4919510 genetic susceptibility to cancer. Hu et al.'s study [40] indicated that no significant associations between rs4919510 polymorphism and overall risk of cancer or the risk of specific types of cancer. Another pooled analysis observed that rs4919510 polymorphism was signifycantly associated with elevated cancer risk in Chinese [41], however, the meta-analysis included some published article that the control population was not conformed the requirements of Hardy-Weinberg Equilibrium. Liu et al. [33] showed that rs4919510 variant was significantly associated with decreased cancer risk based on recessive model. Our study included more case-control studies than those two pre-vious researches and the results challenged of their findings and provided stronger evidence.
Polymorphism of rs4919510 in miR-608 has been little explored for its effect on clinical outcomes for different types of cancer. Lin et al. [42] found that rs4919510 variation was associated with an increased renal cell carcinoma risk. Xing et al. [14] reported that the rs4919510 C allele was associated with a reduced risk of colorectal cancer and had a greater RFS and OS in a Chinese population. Interestingly, a study on colorectal cancer by Ryan et al. [10] showed that the rs4919510 GG genotype was associated with an increased risk of death in Caucasians and with a reduced risk of death in African Americans. However, negative results were The squares and horizontal lines correspond to the study specific OR and 95% CI. The area of the squares reflects the weight (inverse of the variance). The diamond represents the summary OR and 95% CI.
AGING reported in Asian populations in different types of cancer [15,16,24]. Furthermore, Zheng et al. [27] found that rs4919510 polymorphism was not associated with RFS but with OS of nasopharyngeal carcinoma in Asians.
As far as we know, this is the first meta-analysis regarding the rs4919510 polymorphism in miR-608 and its clinical prognosis significance in cancer. The results indicated that rs4919510 variant was significantly asso-ciated with the RFS of cancer under homozygote models (P = 0.04) in Asians by univariable analysis, however, which was no longer remained significant after adjustment for age, gender, histology and clinic stage and so on in multivariable model. No significant association was observed between this polymorphism and the OS of cancer. This might result from the different types of cancers, differences in clinical trials including different ethical backgrounds, different baseline characteristics, or various follow-up time periods  AGING used in each study. Furthermore, the exact functional mechanism by which rs4919510 polymorphism in miR-608 may affect cancer prognosis, there are several plausible hypotheses such as the polymorphism altering the transcription of target genes and reducing the sensitivity of the cancer cell to chemotherapeutic drugs [7,43].
Our study provides added evidence that the rs4919510 gene polymorphism may be associated with cancer susceptibility and prognosis. However, some potential limitations of our meta-analysis should be acknowledged. First, the sample size was relatively small in some studies, which may reduce the statistical power of the analysis and needs to be interpreted with caution. Second, in our meta-analysis, some cancers only had one study included, such as hepatocellular carcinoma, esophageal squamous cell carcinoma, lung cancer and neuroblastoma, which may lead to heterogeneity in quantitative analysis. Third, we perform stratified analysis by population, most of which are Chinese, and type of cancer, most of which are BC and CRC, rather than other baseline characteristics. Fourth, there was significant heterogeneity in some comparison models in our meta-analysis. Even with the use the appropriate meta-analytical techniques with the random-effects model, the heterogeneity might affect the precision of the overall results.
In conclusion, this meta-analysis suggests that the rs4919510 polymorphism in miR-608 is associated with cancer risk and prognosis. Further well-designed, prospective investigations including rs4919510 polymorphism and cancer susceptibility and prognosis with larger sample sizes are required in order to validate our findings.

Literature search and selection criteria
This meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) criteria [44].  [45]. The Newcastle-Ottawa Scale (NOS) was used for assessing the quality of case-control study, the results AGING were showed in Table 1. The NOS score of all articles are not less than 6 scores that mean each selected literature was a high-quality study.
Studies that fit into our meta-analysis fell into one of the following categories: (1) evaluation of the association of rs4919510 polymorphism and cancer susceptibility and/or prognosis; (2) provision of sufficient data to estimate the odds ratio (OR) or hazard ratio (HR) and 95% confidence intervals (CI) according to rs4919510 polymorphism; (3) case-control design; (4) controls without any evidence of malignant disease. The following were the exclusion criteria: (1) repeats studies, reviews, and abstracts; (2) the study did not have a control group; (3) design based on family; (4) the control population was not conformed the requirements of Hardy-Weinberg Equilibrium. The diagnosis of each cancer in eligible study was made based on the histopathological evidence and those with secondary tumour or metastasized cancer from other organs were excluded. Two investigators (Z-MD and J-RL) retrieved the literature and extracted data from each eligible study independently.

Data extraction and synthesis
For each eligible study investigating the association between rs4919510 polymorphism and cancer risk, the following information was extracted by the two independent researchers: first author, year of publication, country, tumour type, ethnicity, source of control, size of the study population, genotyping methods, as well as Hardy-Weinberg value of controls. For studies examining the association between rs4919510 polymorphism and its prognostic value in cancer, the following information was collected from the eligible articles: first author, year of publication, origin of the study population, type of cancer, method of survival analysis, HRs with 95% CIs, as well as the follow-up time. Different ethnic groups were categorised as Caucasian, Asian, and African. Disagreements were resolved by discussion among all authors.

Statistical analysis
ORs and 95% CIs were used to measure cancer risk associated with the rs4919510 polymorphism. Statistical heterogeneity between studies was measured by the Chi-square-based Q statistic and evaluated with the I 2 test; higher I 2 values means higher levels of heterogeneity (I 2 = 75-100%: extreme heterogeneity; I 2 = 50-75%: large heterogeneity; I 2 = 25-50%: moderate heterogeneity; I 2 < 25%: no heterogeneity) [46]. When the p value was greater than 0.10, the fixed-effects model was used, otherwise the random-effects model was used. The Egger's test and Begg's funnel plot were adopted to assess publication bias. Sensitivity analysis determined whether the individual data in fact have a major effect on the results of the review.

AUTHOR CONTRIBUTIONS
ZJD, JRL and ZMD designed the research, analyzed the data, wrote the paper. XML, ZMD and KL reviewed drafts of the paper, and participated in its design and coordination. WL and GW contributed reagents/materials/ analysis tools; DZM, XHL and YXZ prepared figures and/or tables. KL and QH reviewed drafts of the paper.

CONFLICTS OF INTEREST
The authors declare no conflict of interest.