Introduction

Glioma is the most common and the worst prognosis on primary central nervous system tumors, making up approximately 30 % of all brain and central nervous system tumors and 80 % of all malignant brain tumors [1, 2]. Even with aggressive treatments such as surgery, radiation, and chemotherapy, the median survival is very low, particularly for patients with malignant glioma. The etiology and pathogenesis of glioma have remained poorly understood to date. Several occupations, environmental carcinogens, N-nitroso compound diet, and therapeutic X-irradiation have been established as risk factors for glioma [3, 4]. However, occupational and environmental exposures are not enough to account for the development of glioma, in that the genetic predisposition is also related to the glioma development [2]. Single nucleotide polymorphisms (SNPs) of Fc-epsilon RI-alpha, X-ray cross-complementing group 1 Arg399Gln, and epidermal growth factor gene are demonstrated to exert risk effects on glioma [5–7].

The mammalian p53 family consists of p53, p63, and p73 [8]. The human p53 is the most frequently mutant gene in human tumors and located on chromosome 17p13 with the function of tumor suppression through cell cycle arrest and apoptosis [9]. It encodes a protein, namely p53, which plays vital roles in DNA repair, cell signaling transduction, and the cell cycle control to maintain the genomic stability [10, 11]. The p53 can be activated and upregulated by stresses such as DNA damages, abnormal cell proliferation signals, and ultraviolet light [10]. The p53 protein, as a powerful tumor suppressor, is an attractive cancer therapeutic target, which also holds promise as a drug target for novel therapies in glioma [12]. Nevertheless, the mutated p53 may exert tumor-promoting effects in response to cellular stress. Through the use of linkage studies and genome-wide association studies, several p53 SNPs including p53 codon 72 Arg/Pro have been indicated in the development of malignant tumors [13–16]. Regarding the susceptibility to glioma, a number of case–control studies have investigated the role of p53 codon 72 Arg/Pro polymorphism in the glioma pathogenesis [17–26], while the findings were inconsistent and ambiguous across different individual studies among ethnicities. Therefore, we performed the current meta-analysis of all published case–control studies thus far and aimed to provide a better understanding on the molecular mechanism of glioma.

Materials and methods

Eligible studies

We searched the PubMed, Embase, Web of Science, and China National Knowledge Infrastructure (CNKI) databases to identify potentially relevant studies concerning the relationship of p53 codon 72 Arg/Pro polymorphism and the susceptibility to glioma. Case–control studies with sufficient available data for calculating the odds ratio (OR) and 95 % confidence interval (95 % CI) were retrieved by use of the following items: “p53,” or “p53 codon 72,” or “p53 codon 72 Arg/Pro,” or “rs1042522”; and “glioma,” or “brain tumor,” or “astrocytomas,” or “glioblastoma,” or “oligodendroglioma,” or “oligoastrocytoma,” or “neuroepithelial tumor”; and “polymorphism,” or “polymorphisms,” or “mutation.” Reviews and family-based and case-only studies were all excluded. The irrelevant studies and studies with overlapping data were also not included. The references of all eligible publications were hand-checked for additional relevant case–control studies.

Data extraction

Two investigators extracted data independently by reading through the full texts. Disagreements were resolved by consensus. For each study, the following data were extracted: the first author’s name, publication year, ethnicity, country of origin, study design, genotyping method, inclusion criteria for patients and controls, sample size, definition of glioma, glioma subtypes, source of controls, the mean age of cases and controls, the Hardy–Weinberg equilibrium (HWE) for the genotype distribution in controls, and the frequency of genotype in cases and controls (Arg/Arg, Pro/Pro, and Arg/Pro).

Statistical analysis

The pooled ORs with corresponding 95 % CIs were calculated to assess the association between p53 codon 72 Arg/Pro polymorphism and the glioma risk. Five gene contrast models were evaluated including the comparisons for Pro allele vs. Arg allele, Pro/Pro vs. Arg/Arg, Pro/Arg vs. Arg/Arg, Pro/Arg+Pro/Pro vs. Arg/Arg, and Pro/Pro vs. Arg/Arg+Pro/Arg. We conducted stratified analysis by ethnicity, source of controls, and glioma subtypes to further estimate the gene association. The HWE for genotype distribution in controls was estimated by the chi-square test, and a P value less than 0.05 was considered statistically significant. Sensitivity analysis by sequential omission of each study one at a time was carried out to investigate the impact of individual study on the pooled effect. The between-study heterogeneity was checked by use of I 2 statistic test and chi-square-based Q statistic test [27, 28]. The value of P less than 0.05 indicated significant heterogeneity among the included studies, and thus the random-effect model was applied (DerSimonian and Laird method) to calculate the pooled ORs with corresponding 95 % CIs [29]. Otherwise, the fixed-effect model (the Mantel–Haenszel method) was adopted when there was no obvious heterogeneity among individual studies [30]. Egger’s linear regression test and Begg’s graphical methods were both used to estimate the publication bias risk in our study [31, 32]. The meta-analysis was done by use of STATA 12.0 (StataCorp LP, College Station, Texas).

Results

The characteristics of eligible studies

There were 15 relevant publications retrieved after a comprehensive search of the PubMed, Embase, Web of Science, and CNKI databases [17–26, 33–37]. Among them, five were excluded due to overlapping data, case-only design, and not related to p53 codon 72 mutation, respectively [33–37]. According to the source of controls, the publication by Malmer et al. was regarded as two independent case–control studies. Totally, 10 individual publications with 11 case–control studies were included into this meta-analysis involving 3,281 cases and 4,626 controls [17–26]. The genotype distribution in controls of case–control studies were all in agreement with HWE. Based on ethnicity, studies were performed as follows: one with 301 cases and 302 controls were among Asians [26]; eight with 2,802 cases and 4,112 controls were among Caucasians [17–22, 24]; and two with 178 cases and 212 controls were among the mixed population [23, 25]. Regarding the source of controls, eight including 2,124 cases and 3,256 controls were population-based case–control studies [17, 19, 20, 22–26], while two with 1,022 cases and 1,253 controls were hospital-based case–control ones [20, 21], and still one could not be characterized due to insufficient information [18]. The characteristics of all included studies were presented in Table 1 in details.

Table 1 Characteristics of all case–control studies in this meta-analysis
Full size table

Meta-analysis results

As shown in Table 2, the overall pooled ORs indicated that the p53 codon 72 variant exerted no risk effect on the glioma development under all comparisons (ORPro allele vs. Arg allele = 1.04, 95 % CI = 0.90–1.20, P OR = 0.581; ORPro/Pro vs. Arg/Arg = 0.95, 95 % CI = 0.80–1.14, P OR = 0.614; ORPro/Arg vs. Arg/Arg = 1.01, 95 % CI = 0.79–1.29, P OR = 0.993; ORPro/Arg + Pro/Pro vs. Arg/Arg = 1.03, 95 % CI = 0.82–1.29, P OR = 0.799; ORPro/Pro vs. Arg/Arg + Pro/Arg = 1.02, 95 % CI = 0.86–1.22, P OR = 0.785) (Table 2 and Fig. 1). Sensitivity analysis did not alter the pooled results among total included studies (data not shown).

Table 2 Summary meta-analysis results for the association of the of p53 codon 72 Arg/Pro polymorphism and glioma risk
Full size table
Fig. 1
figure 1figure 1figure 1

Forest plots for the association between the P53 codon 72 Arg/Pro polymorphism and glioma risk. a Pro allele vs. Arg allele. b Pro/Pro vs. Arg/Arg. c Pro/Arg vs. Arg/Arg. d Pro/Arg+Pro/Pro vs. Arg/Arg. e Pro/Pro vs. Arg/Arg+Pro/Arg

Full size image

Stratified analysis among Caucasians

The included studies were primarily conducted in Caucasians, Asians, and the mixed population. The stratified analysis in Caucasians showed that the p53 codon 72 polymorphism was not related to the glioma risk under all gene models (ORPro allele vs. Arg allele = 1.09, 95 % CI = 0.93–1.27, P OR = 0.297; ORPro/Pro vs. Arg/Arg = 0.94, 95 % CI = 0.76–1.16, P OR = 0.550; ORPro/Arg vs. Arg/Arg = 1.17, 95 % CI = 0.92–1.48, P OR = 0.206; ORPro/Arg + Pro/Pro vs. Arg/Arg = 1.16, 95 % CI = 0.92–1.46, P OR = 0.207; ORPro/Pro vs. Arg/Arg + Pro/Arg = 0.93, 95 % CI = 0.76–1.14, P OR = 0.473) (Table 2).

Stratified analysis by source of controls

No significant association was observed either in population-based studies or hospital-based studies (Table 2). Interestingly, the pooled ORs for all gene comparisons by use of the fixed-effect model suggested that the p53 codon 72 variant was negatively associated with the glioma pathogenesis, although with a lack of evidence for statistical significance (ORPro allele vs. Arg allele = 0.92, 95 % CI = 0.80–1.05, P OR = 0.231; ORPro/Pro vs. Arg/Arg = 0.82, 95 % CI = 0.58–1.17, P OR = 0.280; ORPro/Arg vs. Arg/Arg = 0.93, 95 % CI = 0.78–1.12, P OR = 0.442; ORPro/Arg + Pro/Pro vs. Arg/Arg = 0.92, 95 % CI = 0.77–1.09, P OR = 0.310; ORPro/Pro vs. Arg/Arg + Pro/Arg = 0.85, 95 % CI = 0.60–1.19, P OR = 0.344) (Table 2).

Stratified analysis by glioma subtypes

As presented in Table 2, we carried out stratified analysis among patients with astrocytomas, glioblastoma, and oligodendroglioma. The Pro/Pro genotype seemed to play a protective role in the development of glioblastoma (ORPro/Pro vs. Arg/Arg = 0.68, 95 % CI = 0.48–0.95, P OR = 0.026). Nevertheless, no significant association was found under other gene contrast models regarding the risk for glioblastoma (Table 2). In addition, the pooled ORs showed a negative relationship between the p53 codon 72 variant and the risk for astrocytomas and oligodendroglioma, but without statistical significance (Table 2).

Heterogeneity analysis

There was significant between-study heterogeneity among all included studies in comparisons of Pro allele vs. Arg allele, Pro/Arg vs. Arg/Arg, and Pro/Arg + Pro/Pro vs. Arg/Arg (for Pro allele vs. Arg allele: I 2 = 68.5 %, P H < 0.001; for Pro/Arg vs. Arg/Arg: I 2 = 81.7 %, P H < 0.001; and for Pro/Arg + Pro/Pro vs. Arg/Arg: I 2 = 80.6 %, P H < 0.001) (Table 2). The stratified analyses indicated that the heterogeneity was not attributed to the hospital-based studies on the susceptibility to glioma (Table 2).

Publication bias

The analyses of Egger’s linear regression test and Begg’s graphical methods showed that there was no publication bias risk in the current meta-analysis (Fig. 2).

Fig. 2
figure 2

Begg’s funnel plot for the publication bias estimate (Pro/Pro vs. Arg/Arg+Pro/Arg, P Egger = 0.806)

Full size image

Discussion

Dysregulation in cell cycle and apoptosis control is a crucial cause for malignant tumors. The p53 is a well-established tumor suppressor through cell cycle arrest and apoptosis, which has been considered as one of the most attractive targets for molecular cancer therapy [9, 38]. Mutation in p53 gene can damage the biological activity of its encoding protein and thus may contribute to various carcinogenesis. P53 codon 72 Arg/Pro polymorphism is the most extensively studied locus of p53 gene. The polymorphism at codon 72 of the p53 gene results in either a proline or an arginine. They play diverse roles in the development of cancer, suggesting different functional activities of proline and arginine [39, 40]. The mutant p53 codon 72 Arg/Pro was not associated with the risk of breast cancer among Turkish women, while the homozygote and heterozygote Pro genotypes were significantly related to the tumor stages of this cancer [41]. Conversely, Buyru et al. demonstrated that the Arg/Arg homozygote carriers in a Turkish population were more susceptible to breast cancer [40]. The varying geographical distribution, environmental exposures, study designs, and source of controls may account for the discrepancy.

The effect of p53 codon 72 Arg/Pro polymorphism on the glioma susceptibility remains rather obscure owing to the contradictory findings among different ethnicities. Parhar et al. firstly reported a positive association between the p53 codon 72 Arg/Pro polymorphism and the susceptibility to glioma, particularly the high-grade astrocytomas [18]. Besides, a previous meta-analysis by Shi et al. demonstrated that the p53 codon 72 polymorphism was associated with an elevated risk of high-grade glioma development in Europeans [42]. However, the sample size in overall cases was too limited (<1,000) to precisely estimate the relationship of p53 codon 72 variant with glioma risk due to insufficient statistical power. Our updated meta-analysis of 7,907 subjects suggested that the p53 codon 72 Arg/Pro variant may exert no effect on the glioma development among all studies (ORPro allele vs. Arg allele = 1.04, 95 % CI = 0.90–1.20; ORPro/Pro vs. Arg/Arg = 0.95, 95 % CI = 0.80–1.14; ORPro/Arg vs. Arg/Arg = 1.01, 95 % CI = 0.79–1.29; ORPro/Arg + Pro/Pro vs. Arg/Arg = 1.03, 95 % CI = 0.82–1.29; ORPro/Pro vs. Arg/Arg + Pro/Arg = 1.02, 95 % CI = 0.86–1.22). Interestingly, the Pro/Pro genotype seemed to play a protective role in the development of glioblastoma (ORPro/Pro vs. Arg/Arg = 0.68, 95%CI = 0.48–0.95). However, we failed to identify significant association of the p53 codon 72 Arg/Pro polymorphism with glioma risk in stratified meta-analyses of population-based, hospital-based, astrocytoma, and oligodendroglioma studies among Caucasian. Sensitivity analysis confirmed the stability and credibility of all findings. Besides, stratified analyses among Asians and the mixed population were not carried out due to limited sample size. Moreover, we did not conduct the stratified analysis by glioma grade due to insufficient available data of individual publications. More case–control studies are recommended to further elucidate the impact of mutant p53 codon 72 Arg/Pro on the susceptibility to different grades of glioma.

The p53 codon 72 Arg/Pro polymorphism exerted diverse effects on the glioma onset across ethnicities. The p53 codon 72 polymorphism was identified as a risk factor for the development of glioma in Indian population [25]. Jin and the colleagues demonstrated a 34 % increased risk of glioma among the Chinese Han carriers of Pro/Pro haplotype [26]. The study by El Hallani et al. supported a critical role of p53 codon 72 functional variation in the oncogenesis of glioblastoma in younger patients, but not the elder patients, among Caucasians [24]. Nevertheless, no statistically significant correlation was observed between the p53 codon 72 variant and the glioma risk among Caucasians [17]. In our stratified meta-analysis by ethnicity, the p53 codon 72 polymorphism did not modify the risk for glioma under all gene models in Caucasians. We failed to perform stratified analyses among Asians and the mixed population due to insufficient available case–control publications.

The association of p53 codon 72 variant with the glioma risk varies among different subtypes of this disease. Lima-Ramos et al. found no statistically significant differences of the p53 codon 72 genotype distributions in patients with distinct histological subtypes of glioma including astrocytoma, glioblastoma, as well as oligodendroglioma [22]. On the contrary, the p53 codon 72 Arg/Pro variant was elucidated for the first time by Parhar and the colleagues [18] to exert increased risk effects on the high-grade astrocytomas. Diverse occupations, environmental factors, and genetic backgrounds may be responsible for the conflicting findings of independent studies mentioned above. In the present meta-analysis, the Pro/Pro genotype seemed to negatively correlate with the risk of glioblastoma, suggesting a protective role of p53 codon 72 Arg/Pro polymorphism in glioblastoma development. However, the significant correlation was not observed under other gene models except for the comparison of Pro/Pro vs. Arg/Arg. Furthermore, the p53 codon 72 variant was found to be negatively associated with the risk for astrocytomas and oligodendroglioma, although without statistical significance. In relation to the astrocytomas and oligodendroglioma risk, the findings may be due to chance since the overall sample sizes for studies on astrocytomas and oligodendroglioma were insufficient to make precise estimates for the gene association.

According to the source of controls, we stratified the included publications into population-based or hospital-based case–control studies. Interestingly, the pooled ORs for hospital-based studies suggested a negative association of the p53 codon 72 variant with the glioma risk, although without statistical significance (ORPro allele vs. Arg allele = 0.92, 95 % CI = 0.80–1.05; ORPro/Pro vs. Arg/Arg = 0.82, 95 % CI = 0.58–1.17; ORPro/Arg vs. Arg/Arg =0.93, 95 % CI = 0.78–1.12; ORPro/Arg + Pro/Pro vs. Arg/Arg = 0.92, 95 % CI = 0.77–1.09; ORPro/Pro vs. Arg/Arg + Pro/Arg = 0.85, 95 % CI = 0.60–1.19). Nevertheless, the hospital-based subjects could not be representative of the whole population, which may lead to bias. Thus, this finding must be interpreted with caution and needs further elucidation. El Hallani et al. reported that the p53 codon 72 polymorphism alters the risk for glioblastoma among the younger patients (<45 years), but not the elder patients (>45 years), in Caucasians [24]. We did not make stratified analysis based on the mean age of cases due to limited available data. The role of p53 codon 72 variant in the glioma development across different age stages can be determined by more high-quality, independent studies in the future.

In conclusion, the present meta-analysis by pooling all available published data shows that the polymorphism of p53 codon 72 Arg/Pro may play a protective role in the development of glioblastoma. However, the precise association of p53 codon 72 mutation with the risk for glioma needs to be determined in more case–control studies across diverse ethnicities. In addition, the role of p53 codon 72 Arg/Pro polymorphism in glioma patients of various ages can be further elucidated in future studies.