CCDN1 rs603965 polymorphism may serve as a genetic biomarker of brain tumor: A meta‐analysis of 5,769 subjects

Abstract Introduction Some studies already tried to assess the associations between cyclin D1 (CCND1) polymorphisms and brain tumor. However, the results of these studies were not consistent. Thus, we performed the present meta‐analysis to explore the relationship between CCND1 polymorphisms and brain tumor in a larger pooled population. Methods PubMed, Web of Science, Embase, and CNKI were searched for related articles. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated to assess the potential associations. Results Totally nine studies with 5,769 subjects were analyzed. A significant association with brain tumor susceptibility was observed for the rs603965 polymorphism in GG versus GA + AA (dominant comparison, p = 0.003, OR = 0.72, 95% CI 0.57–0.89, I 2 = 64%), AA versus GG + GA (recessive comparison, p = 0.004, OR = 1.46, 95% CI 1.13–1.88, I 2 = 67%), and G versus A (allele comparison, p = 0.0004, OR = 0.77, 95% CI 0.66–0.89, I 2 = 66%) in overall population. Further subgroup analyses by ethnicity yielded similar positive results in both Asians and Caucasians. Moreover, in stratified analyses by type of disease, we noticed that the rs603965 polymorphism was significantly associated with the susceptibility to glioma, but such positive results were not detected in pituitary adenoma or meningioma. Additionally, a significant association with tumor grade was also observed for the rs603965 polymorphism in G versus A (allele comparison, p = 0.02, OR = 0.74, 95% CI 0.59–0.95, I 2 = 26%). Conclusions Our findings suggested that CCND1 rs603965 polymorphism may serve as a potential genetic biomarker of brain tumor, especially for glioma.


| INTRODUCTION
Brain tumor refers to cancer that originates from the brain, and it accounts for 1.8% of new cancers and 2.3% of cancer related deaths all over the world (Siegel, Miller, & Jemal, 2017). Despite rapid progress in chemotherapy, radiation therapy, and minimally invasive surgical treatment achieved in the past few decades, the 5-year survival rate of brain tumor is still extremely low (McNeill, 2016;Zhang et al., 2017). Therefore, early diagnosis is of greatest importance, and identify its potential biomarkers is vital for further improving the prognosis of patients with brain tumor.
Although the exact etiologies of brain tumor are still poorly understood, the obvious family aggregation tendency of brain tumor suggests that inherited factors are involved in its development (Lapointe, Perry, & Butowski, 2018;Suvà & Louis, 2013). Cyclin D1 (CCND1) is encoded by the CCND1 gene located on chromosome 11q13. It controls the G1 -S phase transition of the cell cycle and plays a crucial role in the regulation of cell proliferation and differentiation (Casimiro, Velasco-Velázquez, Aguirre-Alvarado, & Pestell, 2014;Qie & Diehl, 2016). Genetic variations in the CCND1 gene may lead to alternations in gene expression or changes in CCND1 protein structure, which may subsequently affect biological functions of CCND1 and ultimately impact individual susceptibility to multiple malignancies including brain tumor.
Recently, some studies already tried to assess the associations between CCND1 polymorphisms and brain tumor. However, the results of these studies were not consistent (Cander et al., 2012;Chen et al., 2012;Gazioglu et al., 2007;Liu et al., 2015). Previous studies failed to reach a consensus regarding associations between CCND1 polymorphisms and brain tumor partially because of their relatively small sample sizes. Thus, we performed the present meta-analysis to explore the relationship between CCND1 polymorphisms and brain tumor in a larger pooled population.
Included studies of this meta-analysis must meet all the following criteria: (a) study on associations between CCND1 polymorphism and susceptibility to brain tumor/ biological characteristics of brain tumor; (b) providing genotypic distribution of investigated polymorphisms in cases and controls; (c) full text in English or Chinese available. Studies were excluded if one of the following criteria was fulfilled: (a) not about CCND1 polymorphism and brain tumor; (b) insufficient data to estimate associations between CCND1 polymorphism and brain tumor; (c) reviews, editorials, or comments. In the case of duplicate reports by the same authors, we only included the study with the largest sample size.

| Data extraction and quality assessment
From eligible studies, we extracted the following information: name of the first author, year of publication, country and ethnicity of participants, sample size, and the genotypic distribution of CCND1 polymorphism in cases and controls. The probability value (p value) of Hardy-Weinberg equilibrium (HWE) was also calculated.
We used the Newcastle-Ottawa scale (NOS) to evaluate the quality of eligible studies (Stang, 2010). The NOS has a score range of zero to nine, and studies with a score of more than seven were thought to be of high quality.
Two reviewers conducted data extraction and quality assessment independently. When necessary, we wrote to the corresponding authors for extra information. Any disagreement between two reviewers was solved by discussion until a consensus was reached.

| Statistical analyses
In this study, statistical analyses were performed by using Review Manager Version 5.3.3. Pooled odds ratios (ORs) and 95% confidence intervals (CIs) were calculated to estimate potential associations between CCND1 genetic polymorphism and brain tumor in dominant, recessive, overdominant and allele genetic models, and a p value of 0.05 or less was defined as statistically significant. Between-study heterogeneities were evaluated by using the I 2 statistic. Random-effect models (REMs) would be used for analyses if I 2 was greater than 50%. Otherwise, analyses would be conducted with fixed-effect models (FEMs). Subgroup analyses were subsequently carried out by type of disease and ethnicity of participants. Stabilities of synthetic results were tested in sensitivity analyses. Publication biases were assessed by funnel plots.

| Characteristics of included studies
The literature search procedure was shown in Figure 1. Totally 270 articles were found by using our searching strategy. After excluding irrelevant and duplicate articles, 19 articles were retrieved for further evaluation. Another 10 articles were subsequently excluded after reading the full text. Ultimately, a total of nine eligible studies involving 2,079 cases and 3,690 controls were enrolled in analyses (see Figure 1). Characteristics of included studies were summarized in Table 1.

| CCND1 rs603965 polymorphism and susceptibility to brain tumor
Totally 5,769 subjects were enrolled in pooled analyses. A significant association with brain tumor susceptibility was observed for the rs603965 polymorphism in GG versus GA + AA (dominant comparison, p = 0.003, OR = 0.72, 95% CI 0.57-0.89, I 2 = 64%), AA versus GG + GA (recessive comparison, p = 0.004, OR = 1.46, 95% CI 1.13-1.88, I 2 = 67%), and G versus A (allele comparison, p = 0.0004, OR = 0.77, 95% CI 0.66-0.89, I 2 = 66%) in overall population. Further subgroup analyses by ethnicity yielded similar positive results in both Asians (dominant, recessive, and allele comparisons) and Caucasians (recessive comparison). Moreover, in stratified analyses by type of disease, we noticed that the rs603965 polymorphism was significantly associated with the susceptibility to glioma, but such positive results were not detected in pituitary adenoma or meningioma (see Table 2).

| CCND1 rs603965 polymorphism and biological characteristics of brain tumor
Totally 868 subjects were enrolled in pooled analyses. A significant association with tumor grade was observed for the rs603965 polymorphism in G versus A (allele comparison, p = 0.02, OR = 0.74, 95% CI 0.59-0.95, I 2 = 26%). Nevertheless, no any positive results were detected for rs603965 polymorphism when it comes to tumor size or tumor invasiveness (see Table 2).

| Sensitivity analyses
Sensitivity analyses were performed to test the effects of individual study on pooled results if the original analysis was based on at least four studies. No any alterations of results were detected in overall and subgroup analyses, which suggested that our findings were statistically stable.

| Publication biases
We used funnel plots to evaluate potential publication biases. The shape of funnel plots was symmetry for every comparison, which indicated that our findings were unlikely to be impacted by severe publication biases.

| DISCUSSION
As far as we know, this is so far the most comprehensive metaanalysis about CCND1 polymorphism and brain tumor. The pooled results showed that the rs603965 polymorphism was significantly associated with the susceptibility to brain tumor. Moreover, it was also significantly associated with tumor grade. The positive results obtained with the susceptibility to brain tumor were largely driven by the glioma subgroup, yet no statistically significant associations were observed in pituitary adenoma and meningioma subgroups. The stabilities of synthetic results were evaluated by sensitivity analyses, and no alterations of results were observed in any comparisons, which suggested that our findings were statistically stable.
There are several points that worth noting about this metaanalysis. First, the present meta-analysis aimed to explore associations between all CCND1 polymorphisms and brain tumor. However, only the rs603965 polymorphism could be analyzed because this was the only CCND1 polymorphism that was investigated by multiple different studies. Previous experimental studies found that the rs603965 polymorphism was located at the conjunction of exon 4 and intron 4 of CCND1 gene, the G to A variation in this locus was associated with a shorter transcription and a 55 amino acids loss of the C-terminal domain of CCND1 (Jeon, Kim, Jeong, Bae, & Jung, 2018;Vodicka et al., 2018), which may greatly alter the activity of CCND1 and give rise to the development of brain tumor. Second, the etiologies of different types of brain tumor may be somewhat different. Therefore, although it is biologically possible that CCDN1 rs603965 polymorphism may be implicated in the development of brain tumor, its effects on different types of brain tumor may be different, and this may partially explain the observed heterogeneities in the current meta-analysis. Third, it is notable that in 2014, Qin et al. (2014) also performed a meta-analysis to estimate associations between CCDN1 rs603965 polymorphism and brain tumor. The current meta-analysis should be considered as a valuable supplementary work to this previous meta-analysis because of the following two points, (a) Three related studies were published in the last 4 years. Therefore, an update meta-analysis is warranted and the sample sizes of our analyses were also significantly larger than of previous metaanalysis, which could significantly reduce the risk of obtaining false positive or false negative results; (b) This meta-analysis also tried to investigate potential associations between CCDN1 rs603965 polymorphism and biological characteristics of brain tumor. Although the sample size was still relatively small, the results obtained were quite interesting, and we call on future studies to report related data. Fourth, in the current meta-analysis, we noticed that studies conducted in Asians were mainly about glioma, while studies conducted in Caucasians were mainly about pituitary adenoma. This phenomenon suggested that the disease spectrum of brain tumor in different ethnicities may be quite different, and the exact reason underlying this phenomenon is also worth investigating. Fifth, the functional significance of the rs603965 polymorphism is well established, yet no significant associations were observed for pituitary adenoma or meningioma. Since the sample sizes of pooled analyses in the current meta-analysis were still relatively small, it is possible that our study was still not statistically adequate to detect the actual associations between the rs603965 polymorphism and other types of brain tumor. Therefore, further studies with larger sample sizes still need to confirm our findings.
Some limitations of this meta-analysis should also be noted when interpreting our findings. First, due to lack of raw data, our pooled analyses were based on unadjusted analyses, but we have to admit that failure to perform further adjusted analyses for confounding factors may impact the reliability of our findings (Qian, Zhang, Qian, He, & Li, 2017). Second, grey literatures like research materials that were not formally published in academic journals were not considered to be eligible for analyses in this meta-analysis since it is hard to evaluate their quality. However, since grey literatures were not analyzed, although funnel plots suggested that severe publication biases were unlikely to be existed in the current meta-analysis, it is still possible that our findings may be impacted by potential publication biases (Liu & Jiang, 2017). Third, associations between CCND1 rs603965 polymorphism and brain tumor may also be modified by gene-environmental interactions. However, most studies did not explore the effects of these potential interactions, which impeded us to conduct relevant analyses accordingly (Montelli Tde et al., 2011). Considering the above mentioned limitations, our findings should be interpreted with caution.

| CONCLUSION
In summary, our meta-analysis suggested that the CCND1 rs603965 polymorphism may serve as a potential genetic biomarker of brain tumor, especially for glioma. However, further well-designed studies with larger sample sizes are still warranted to confirm our findings.