Association of PPARG rs 1801282 C>G polymorphism with risk of colorectal cancer: from a case-control study to a meta-analysis

The functional single nucleotide polymorphisms in peroxisome proliferator-activated receptor gamma (PPARG) gene were predicted to be correlated with the susceptibility of colorectal cancer (CRC). The aim of the present study was to explore the relationship between PPARG rs1801282 C>G polymorphism and the risk of CRC. First, we conducted a case-control study with 387 CRC cases and 1,536 controls. We used the SNPscan method to determine the genotypes of PPARG rs1801282 C>G polymorphism. We found PPARG rs1801282 C>G polymorphism had a tendency of decreased risk to CRC risk (CG vs. CC: adjusted OR, 0.67, 95% CI = 0.43–1.04 for CG vs. CC, P = 0.073; GG vs. CC: adjusted OR, 0.68; 95% CI, 0.44–1.05; P = 0.078). The stratified analysis revealed PPARG rs1801282 C>G polymorphism also had a tendency of decreased risk to colon cancer (CG vs. CC: adjusted OR = 0.54, 95% CI = 0.27–1.08, P = 0.083). The results of subsequent meta-analysis suggested that PPARG rs1801282 C>G polymorphism might be a protective factor for CRC, especially in Asians, colon cancer and rectum cancer subgroups. In conclusion, our study indicates that PPARG rs1801282 C>G polymorphism might decrease the risk of overall CRC. Larger sample size and well-designed case-control studies are needed to confirm the potential association.


INTRODUCTION
Colorectal cancer (CRC) is the fifth most common type of malignancy among males and the fourth most common type among females in China, accounting for 215,700 and 160,600 cases in 2015, respectively [1].The incidence of CRC is rapidly increasing in developing countries including China [1,2]; however, the etiology of CRC remians unknown.Risk factors, such as family history of CRC, advanced age, inflammatory bowel

Meta-Analysis
diseases, benign adenomatous polyps, being physically inactive, drinking, smoking, high intake of dietary fat and low intake of vegetables and fruits, may play important roles in the development of CRC [3][4][5][6][7][8][9].Accumulating evidence suggested that besides individual lifestyle and environmental factors, some genetic factors may be relevant to the etiology of CRC.
The gene of peroxisome proliferator-activated receptor gamma (PPARG), a ligand-activated transcription factor, is located in 3p25.PPARG shares conservative domain with other steroid receptors (e.g., the vitamin D, estrogen, progesterone, retinoid and thyroid receptors), which recognize to peroxisome proliferator-activated receptor (PPAR) response elements in the region of promoter, and then bind to them.Subsequently, these steroid receptors regulate the transcription of some target genes.It is well known that PPARG may be involved in controlling adipocytes differentiation, regulating energy homeostasis, influence of cellular cholesterol homoeostasis, and the development of type 2 diabetes mellitus (T2DM) and obesity [10][11][12].
Many investigations evidenced the potential roles of PPARG gene in determining CRC susceptibility.Understanding the variants in this gene correlated with CRC susceptibility may be helpful for CRC prevention and diagnosis.Recently, some case-control studies focused on the relationship of PPARG polymorphisms with the risk of CRC.A common single nucleotide polymorphism in PPARG gene [rs1801282 C>G (Pro12Ala)] have been established, which were associated with receptor activity, insulin sensitivity, body mass index (BMI), and risk of T2DM [13,14].Many studies focused on the association of PPARG rs1801282 C>G polymorphism with risk of CRC.Several meta-analyses demonstrated that PPARG rs1801282 C>G polymorphism was associated with the decreased risk of CRC in Caucasians [15,16].However, there were only three case-control studies with relatively small sample sizes focused on the relationship between PPARG rs1801282 C>G polymorphism and CRC in Asians [17][18][19].The evidence may be limited.
The biological significance of PPARG indicates that functional polymorphisms in PPARG gene may influence the susceptibility of CRC.Thus, the attempt of the present study was to assess the relationship of rs1801282 variations in PPARG with CRC risk.The results of our case-control study might be limited by sample size.With the aim to overcome this limitation, a comprehensive pooled-analysis was subseqently carried out to determine the association of PPARG rs1801282 C>G polymorphism with CRC risk.

Study characteristics
Table 1 summarized the distribution of demographic variables and risk factors in CRC cases and controls.
We found there was no significant difference in the distributions of age (cases: 60.21 ± 12.48, vs. controls: 60.82 ± 8.82; P = 0.272), sex (P = 0.213), smoking (P = 0.505) and alcohol consumption (P = 0.058) between cases and controls.CRC patients have relatively lower body mass index (BMI) than that of the control subjects (P < 0.001).When it comes to TMN stages, according to AJCC criteria from 2010, 196 and 191 CRC patients were classified as stage I/II and III/IV, respectively.The primary information of PPARG rs1801282 C>G polymorphism was listed in Table 2.The genotype distributions of PPARG rs1801282 C>G polymorphism in controls were in accordance with HWE (P = 0.544).

Association of PPARG rs1801282 C>G polymorphism with CRC risk in a stratification group by site of tumor
To assess the effect of PPARG rs1801282 C>G polymorphism in different tumor site, a stratified analysis was conducted.The stratified analysis revealed PPARG rs1801282 C>G polymorphism also had a tendency of decreased risk to colon cancer (CG vs. CC: adjusted OR = 0.54, 95% CI = 0.27-1.08,P = 0.083; Table 4).

Meta-analysis of PPARG rs1801282 C>G polymorphism and CRC risk
Next, a comprehensive meta-analysis was carried out to determine the relationship between PPARG polymorphisms and CRC risk.In total, 219 abstracts were retrieved from Pubmed and EMBASE databases.www.impactjournals.com/oncotarget The detailed selecting process is summarized in Figure 1.There were several subgroups in our present study and some publications [17,[19][20][21][22], we treated them separately.The detailed characteristics and PPARG rs1801282 genotypes of included studies are listed in Table 5.Finally, our present study and previously published studies involving 12,761 cases and 21,113 controls were recruited in this pooled-analysis.
Using the exclusion method in turn, one-way sensitivity analysis was performed to determine whether an included study could affect the final decision.The results showed that our findings were stable and reliable (Figure 5).

DISCUSSION
PPARG is a nuclear hormone receptor, and mainly exists in colorectum, adipose tissue, and immune system [23].PPARG plays a very important role in the inflammatory response, adipose cell differentiation, modulation of metabolism, and cellular apoptosis [24][25][26][27].PPARG regulates and/or interacts with multifarious signaling pathways, including those associated with p21, p53, NF-kappa-β, STAT, BCL2, cyclooxygenase-2 (COX-2) and cyclin D1 [24-26, 28, 29].PPARG is highly expressed in tumour cells, treatment with PPARG ligands can induce cell apoptosis and differentiation [30][31][32].PPARG mutation may increase CRC risk [22].The possible association of PPARG rs1801282 C>G polymorphism with CRC risk has been extensively studied; however, findings of those investigations were conflicting, especially in Asians.To obtain a more precise assessment of these potential associations, we conducted a case-control study.Then, given the accumulating evidences and to shed some light on this issue, we performed a pooled-analysis of this potential relationship from Pubmed and EMBASE databases.For PPARG rs1801282 C>G polymorphism, individuals carrying the GG and GG/CG genotype had a tendency of decreased risk to CRC risk.In colon cancer subgroup, the results of logistic regression analyses indicated that tendency was also noted.The results of subsequent meta-analysis suggested that PPARG rs1801282 C>G polymorphism was associated with decreased susceptibility of CRC,  especially in Asians, colon cancer and rectum cancer subgroups.Adiposity and a sedentary lifestyle have been consistently related to CRC risk, and are vital determinants of hyperinsulinemia and insulin resistance.High concentrations of insulin or C-peptide (an insulin marker) have manifested direct association with CRC risk [33,34].A common functional polymorphism (Pro12Ala; rs1801282) in PPARG is C→G missense substitution causing a proline to alanine substitution in codon 12 of exon 2. Functional studies on PPARG rs1801282 polymorphism have revealed that G variant may alter the binding affinity of the protein to PPARG-responsive DNA elements compared to the C variant and the differential expression of PPARG-target genes has indicated the role of PPARG rs1801282 C>G polymorphism in transcriptional activity of PPARG [13,35].The PPARG rs1801282 C→G substitution produces protein with higher activity [13,36].Presence of the rs1801282 C>G polymorphism was reported to be associated with improved insulin sensitivity, lower body mass index (BMI), and a reduced risk of T2DM [37,38].Thus, it is possible that PPARG rs1801282 C>G polymorphism may be a protective factor for colorectal cancer through insulinrelated mechanisms.In our case-control study and metaanalysis, we uniformly found that PPARG rs1801282 G allele might decrease CRC risk.These results were consistent with the protective effect of this polymorphism, and suggest this polymorphism may confer a lower CRC risk.Several meta-analyses have been undertaken to assess the relationship between PPARG rs1801282 C>G polymorphism and CRC risk [15,16,39].In the present study, we also conducted a meta-analysis on this association including largest sample size (25 studies with 33,874 subjects).Overall, our findings of meta-analysis were consistent with those results.While in subgroup analyses, we found there were significant associations between PPARG rs1801282 C>G polymorphism and decreased risk of CRC among Asians, colon cancer and rectum cancer subgroups.These results of subgroup analysis were not similar to previous meta-analyses.In our meta-analysis, more studies and more participants were recruited.Thus, our findings may be more reliable than before.
Our study has some limitations.Firstly, our case-control study was hospital-based and might be   Secondly, the sample size of patients with CRC was moderate.Thirdly, some factors, such as diet, physical activity, use of non-steroidal anti-inflammatory drugs, other functional SNPs in PPARG gene, and etc., were not considered.In the future, well-designed studies are needed to further investigate the association thoroughly.Finally, the relationship between PPARG polymorphisms and CRC risk involves a complex mechanism; thus, gene-gene and gene-environment interactions should be considered in future studies.
In conclusion, our study indicates that PPARG rs1801282 G allele might decrease the risk of overall CRC.In the future, more case-control studies with large sample size are needed to evaluate the effect of genegene and gene-environment interactions of the PPARG rs1801282 C>G with CRC risk.

Study population and patient selection
Our study consisted of 387 CRC patients (236 men and 151 women) and 1,536 cancer-free controls (989 men and 547 women) in an Eastern Chinese Han population.The CRC cases were consecutively recruited from the Colorectal Surgery of Union Hospital, Fujian Medical University (Fuzhou city, China), between October 2014 and May 2016.Histologically, adenocarcinoma was confirmed via pathology.The major exclusion criteria were: patients with a history of another malignancy and hereditary nonpolyposis CRC.The controls were matched with age and gender and without any history of personal malignancy.All cancer-free controls were recruited from the Affiliated People's Hospital of Jiangsu University and the Affiliated Union Hospital of Fujian Medical University.The variables and risk factors of all participants were collected by two doctors with a pre-structured questionnaire.All participants wrote the informed consent.Data on CRC clinicopathological characteristics were extracted from the medical records.This case-control study is approved by the ethics committee of Fujian Medical University and Jiangsu University (Fuzhou city and Zhenjiang city, China).The experimental protocol was performed in strict accordance with the approved guidelines.

DNA extraction and genotyping
Every participant donated 2ml Ethylenediamine tetraacetic acid (EDTA)-anticoagulated intravenous blood.Genomic DNA from lymphocyte was extracted by Promega DNA Blood Mini Kit (Promega, Madison, USA).As described in previous studies, the genotyping of the rs1801282 C>G polymorphism in PPARG gene was performed by a custom-by-design 48-Plex SNPscan Kit (Genesky Biotechnologies Inc., Shanghai, China) [40,41].This genotyping method was based on double ligation and multiplex fluorescence PCR [42].For quality control, 4% of all sample sizes (seventy-seven samples) were randomly selected and were genotyped again by the same genotyping method.The results of genotyping were not changed.

Statistical analysis
Hardy-Weinberg equilibrium (HWE) was determined by an online Chi-square test (http://ihg.gsf.de/cgi-bin/hw/hwa1.pl).The association of PPARG rs1801282 C>G polymorphism with CRC risk was evaluated using crude and adjusted odds ratios (ORs) with their 95 % confidence intervals (CIs) when appropriate.All statistical analyses were performed by SAS 9.4 for Windows (SAS Institute, Cary, USA).An unpaired Student's t-test was harnessed to check the differences for continuous variables between CRC cases and controls.And χ2 test was used to assess the differences in the included risk factors [e.g., smoking, drinking and body mass index (BMI)], demographic variables, and the frequencies of various allele and genotype between CRC cases and controls.A P < 0.05 (two-tailed) was defined as the level of significance.

Meta-analysis
To further assess the association of PPARG rs1801282 C>G polymorphism with CRC risk, we performed a comprehensive meta-analysis.Firstly, we carried out a systematic search through PubMed and EMBASE databases with the terms of 'Peroxisome proliferator activated receptor gamma' or 'PPARG' and 'polymorphism' or 'mutation' or 'variant' and 'cancer' or 'carcinoma' or 'malignancy' and 'colorectal' or 'colon' or 'rectal'.All included publications were published up to 7 October 2016.The major included criteria were: (a) case-control or cohort study based on PPARG rs1801282 C>G polymorphism with sufficient genotype data and (b) the distribution of genotype in controls was in accord with HWE.The combined ORs and their 95% CIs were applied to determine the relationship of rs1801282 C>G polymorphism in PPARG gene with CRC risk.The between-study heterogeneity assumption was assessed using Chi-square-based statistic I 2 test and Cochran's Q-test [43].When I 2 > 50% or P < 0.1, we used the random-effects model (DerSimonian and Laird method) to estimate the pooled OR [44,45].Otherwise, the fixed effects model (the Mantel-Haenszel method) was applied [46].Potential publication bias in metaanalysis was evaluated through Begg's funnel plot and the Egger's linear regression test [47] (P < 0.1 was defined representative of statistical publication bias).The statistical analyses of meta-analysis were performed by STATA version 12.0 (Stata Corporation, College Station, TX, USA).And all P-values were two-sided (P < 0.05).The power value of this meta-anlysis (α = 0.05) was evaluated by the Power and Sample Size Calculator (http://biostat.mc.vanderbilt.edu/twiki/bin/view/Main/PowerSampleSize) [48].

Figure 1 :
Figure 1: Flow chart of study selection procedure.

Figure 2 :
Figure 2: Forest plot of association between PPARG rs1801282 C>G polymorphism and CRC risk in random model (GG+CG vs. CC).

Figure 3 :
Figure 3: Begger's funnel plot of the meta-analysis of between PPARG rs1801282 C>G polymorphism and CRC risk in random model (GG+CG vs. CC).

Figure 4 :
Figure 4: Filled funnel plot of the meta-analysis of between PPARG rs1801282 C>G polymorphism and CRC risk in random model (GG+CG vs. CC).

Figure 5 :
Figure 5: Sensitivity analysis on association between PPARG rs1801282 C>G polymorphism and CRC risk in random model (GG+CG vs. CC).

Table 1 : Distribution of selected demographic variables and risk factors in colorectal cancer cases and controls
a Two-sided χ 2 test and student t test; BMI, body mass index; FPG, fasting plasma glucose; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; b six subjects have missing data.

Table 4 : Overall and stratified analyses of PPARG rs1801282 C>G polymorphism with colorectal cancer by region Genotype Overall colorectal cancer cases (n = 387) vs. controls
a Adjusted for age, sex, smoking status, alcohol use and BMI status in a logistic regression.