Introduction

Glioma is the most common type of malignant brain tumors in the central nervous system (CNS), accounting for approximately 80% of primary brain tumors1. The incidence of brain cancer is the highest in European (5.5/100,000 persons), North America (5.3/100,000 persons), Australia (5.3/100,000 persons), Western Asia (5.2/100,000 persons) and Northern Africa (5.0/100,000 persons)2. In China, there were 1,016,000 newly diagnosed cases of brain and CNS tumor in 20153. Glioma occurs varied in age, sex, race, histologic type and geographic characteristics4. And, glioma has poor overall survival (OS), with less than 5 year survival of patients after diagnosis5. The etiology of glioma is multifactorial, which is the results of environmental exposure and genetic factors4. Single nucleotide polymorphism (SNP) is the most studied mutations involved in genetic predisposition of glioma. Recently, increasing studies are focused on the role of Peroxisome proliferator-activated receptors (PPARs) polymorphisms on cancer.

PPARs is a subfamily of nuclear receptor transcription factors and consists three isoforms (PPARα, PPARδ and PPARγ). PPARD encodes PPARδ, a nuclear hormone receptor that implicated in varieties of biological processes, including epidermal cell proliferation, migration, lipid and glucose metabolism6,7,8. PPARD is highly expressed in brain, heart, skeletal muscle, adipose tissue and pancreatic islets9. In mice, PPARD agonists increase leptin secretion and improve type 2 diabetes10,11. The overexpression of PPARD was observed in various human cancers, such as colorectal, pancreatic and lung cancer12,13,14,15. Previous studies revealed that PPARD polymorphisms were associated with lipid levels, metabolic traits, obesity and risk of coronary heart diseases (CHD) and cancers16,17,18,19. PPARD rs2016520 is located in the 5′-untranslated region of exon, which has been widely studied in multiple physiological and pathological process20,21,22,23. However, little is known on the relationship of PPARD polymorphisms with glioma risk and prognosis.

PPARG is located in human chromosome 3p25 and encodes a nuclear receptor (PPARγ) activated by fatty acid metabolites or synthetic medicines24,25,26. PPARG is mainly expressed in suprabasal keratinocytes, adipocyte tissue, vascular endothelial cells, macrophage cells and smooth muscle cells27,28. PPARG regulates adipocyte differentiation and controls genes expression involved in lipid and glucose homeostasis29. And, PPARG has anti-inflammatory effect by restraining the production of inflammatory mediators30. It has been reported that PPARG is implicated in the pathology of obesity, diabetes, atherosclerosis and cancer. Wang et al. indicated that PPARG could arrest cell growth in human oral cancer31. Fan et al. pointed that anti- PPARG therapy is a potential strategy to improve endocrine-resistant breast cancer32. Nevertheless, the role of PPARG in glioma has not been elucidated.

Therefore, we conducted a case-control study to investigate the association of PPARD and PPARG polymorphisms (rs2016520, rs67056409, rs1053049, rs2206030, rs2920503, rs4073770 and rs1151988) with glioma risk and prognosis in the Chinese Han population.

Methods

Study population

This study consisted of 568 glioma patients and 509 healthy controls, recruited from the Second Affiliated Hospital of Xi’an Jiaotong University, Shaanxi Province, China. All glioma patients were newly diagnosed and histologically confirmed according to the World Health Organization (WHO) classification33. The exclusion criteria of glioma patients are as follows: (1) patients have history of cancer or CNS diseases; (2) patients are under 18 years old. The controls were healthy individuals without history of cancer or serious diseases who randomly enrolled from the same hospital. We obtained demographic and clinical information of study population from medical records and follow-up. This study was performed in accordance with the Declaration of Helsinki, and it was approved by the ethics committee of the Second Affiliated Hospital of Xi’an Jiaotong University. Informed consents were required form all participants before this study.

SNP selection and genotyping

Combined previously studies, we selected four SNPs of PPARD (rs2016520, rs67056409, rs1053049 and rs2206030) and three SNPs of PPARG (rs2920503, rs4073770 and rs1151988), with minor allele frequencies (MAF) greater than 5% in the HapMap Chinese Han Beijing population. We extracted DNA from peripheral blood samples using the blood DNA kit (GoldMag Co. Ltd., Xiʹan, China). SNP genotyping was performed in the Agena MassARRAY system (Agena, San Diego, CA, USA). Primers for polymerase chain reaction (PCR) amplification and extension were designed by the Agena MassARRAY Assay Design 3.0 Software (San Diego, CA USA). PCR primers of selected SNPs were listed in Supplemental Table 1. In addition, we used Agena Typer 4.0 Software (San Diego, CA, USA) to manage and analyze data.

Statistical analysis

We conducted all statistical analysis using Microsoft Excel and SPSS version 21.0 software (SPSS, Chicago, IL, USA). Student’s t-test and chi-square test were used to compare the differences in age and sex between glioma patients and healthy controls. The Hardy-Weinberg equilibrium (HWE) was checked for controls with Fisher’s exact test. We assessed the association of PPARD and PPARG polymorphisms with glioma risk by calculating odd ratios (OR) and 95% confidence intervals (CI) using logistic regression. Multifactor dimensionality reduction (MDR, version 3.0.2) was used to analyze SNP-SNP interactions on glioma risk. Then, we plotted patient survival curves by the Kaplan-Meier method and log-rank test. The association of PPARD and PPARG polymorphisms with OS and progression free survival (PFS) of glioma patients was evaluated by calculating hazard ratios (HR) and 95%CI using univariate and multivariate analysis. In multivariable survival analysis, we assessed the associations of PPARD and PPARG polymorphisms with glioma prognosis adjusted by age, sex, WHO grade, surgery, radiotherapy and chemotherapy. All tests were two-sided, and P < 0.05 was regarded as statistical significance. Additionally, our results were adjusted for multiple comparison using false discovery rate (FDR) correction.

Results

Characteristics of study population

The characteristics of 568 glioma patients and 509 healthy controls were presented in Table 1. The mean ages of the cases and controls were 39.68 ± 16.96 and 41.32 ± 15.69 years old, respectively. No significant variation in age or sex was found between the two groups (age: P = 0.102, sex: P = 1.000). Among glioma patients, 438 (77%) people were astrocytoma. According to WHO grading standards, 35 (6%) patients were grade I, 320 (56%) patients were grade II, and others are in high-grade glioma (III + IV). In addition, surgery method, radiotherapy and chemotherapy of patients were shown in Table 1.

Table 1 Comparison of glioma patients and controls by characteristics.
Full size table

Association of PPARD and PPARG polymorphisms with glioma risk

In Table 2, PPARD and PPARG polymorphisms were accord with HWE in controls (P > 0.05).

Table 2 Association of PPARD and PPARG polymorphisms with glioma risk.
Full size table

HaploReg (https://pubs.broadinstitute.org/mammals/haploreg/haploreg.php) predicted that PPARD and PPARG polymorphisms were related to the regulation of SiPhy cons, Promoter histone marks, Enhancer histone marks, DNAse, Motifs changed, GRASP QTL hits, NHGRI/EBI GWAS hits, Selected eQTL. After adjustment for age and sex, PPARD polymorphisms (rs2016520, rs67056409 and rs1053049) were significantly associated with glioma risk (P < 0.05). Rs2016520 and rs1053049 of PPARD had a decreased glioma risk in allele (rs2016520: OR = 0.82, 95%CI = 0.68–0.99, P = 0.041; rs1053049: OR = 0.78, 95%CI = 0.64–0.95, P = 0.012), dominant (rs2016520: OR = 0.78, 95%CI = 0.62–1.00, P = 0.047; rs1053049: OR = 0.75, 95%CI = 0.59–0.96, P = 0.020) and additive (rs2016520: OR = 0.81, 95%CI = 0.67–0.99, P = 0.037; rs1053049: OR = 0.78, 95%CI = 0.64–0.95, P = 0.012) models. We found that the allele distribution of rs67056409 were significantly different between cases and controls (P = 0.046), and subjects had lower risk of glioma in additive model (OR = 0.82, 95%CI = 0.67–0.99, P = 0.041). There were no significant association between glioma risk and other genetic polymorphisms (rs2206030, rs2920503, rs4073770 and rs1151988). However, FDR analysis revealed that the significant associations between genetic polymorphisms and glioma risk were not reliable.

We further did stratification analysis of PPARD and PPARG polymorphisms with glioma risk (Tables 3 and 4). For the subjects older than 40 years old, rs2016520, rs67056409 and rs1053049 of PPARD significantly decreased risk of glioma in multiple models (P < 0.05). Rs1053049 had a strong relationship with decreased risk of glioma in the subgroup of male (allele: OR = 0.76, 95%CI = 0.59–0.99, P = 0.041; dominant: OR = 0.71, 95%CI = 0.51–0.98, P = 0.040; additive: OR = 0.76, 95%CI = 0.59–0.99, P = 0.040). Then, we divided glioma patients to astrocytoma and others, we found that rs2016520, rs67056409 and rs1053049 of PPARD were significantly associated astrocytoma risk compared with other glioma (P < 0.05). We also explored the effect of WHO grade on the relationship of genetic polymorphisms with glioma risk. The results showed PPARG rs2920503 was strongly related to higher risk of high-grade glioma (III + IV) in co-dominant (OR = 2.04, 95%CI = 1.13–3.68, P = 0.018) and recessive (OR = 2.03, 95%CI = 1.15–3.57, P = 0.014) models. After FDR correction, the protective effects of rs2016520, rs67056409 and rs1053049 on glioma risk were still significant among the individuals older than 40 years old (FDR- P < 0.05).

Table 3 Association of PPARD and PPARG polymorphisms with glioma risk stratified by age and sex.
Full size table
Table 4 Association of PPARD and PPARG polymorphisms with glioma risk stratified by pathological classification and WHO grade.
Full size table

MDR analysis

We used MDR analysis to assess the impact of the interaction among seven SNPs. The results obtained from MDR analysis for one- to seven- locus modes were presented in Table 5. A seven-locus model including polymorphisms of PPARD (rs2016520, rs67056409, rs1053049 and rs2206030) and PPARG (rs2920503, rs4073770 and rs1151988) was the best model of SNP-SNP interaction for glioma risk (cross-validation consistency = 10/10, accuracy = 0.660, sensitivity = 0.751, specificity = 0.570, P < 0.001).

Table 5 MDR analysis of SNP-SNP interaction.
Full size table

Clinical factors and glioma prognosis

After obtained follow-up data of glioma patients, we investigated the impact of clinical factors on glioma prognosis (OS and PFS). As shown in Table 6, surgery and chemotherapy had significant correlations with OS and PFS of glioma (P < 0.05, FDR- P < 0.05). The prognosis of patients had gross-total resection (GTR) was better than those had sub-total resection (STR) or near-total resection (NTR) (OS: Log-rank P = 1.54E-07, HR = 0.63, 95%CI = 0.53–0.77, P = 1.88E-06, FDR- P = 1.32E-05; PFS: Log-rank P = 1.91E-09, HR = 0.59, 95%CI = 0.49–0.71, P = 6.78E-08, FDR- P = 4.75E-07). Additionally, glioma patients who undergone chemotherapy lived longer than those not (OS: Log-rank P = 1.38E-05, HR = 0.69, 95%CI = 0.58–0.83, P = 7.47E-05, FDR- P = 0.000261; PFS: Log-rank P = 0.005, HR = 0.79, 95%CI = 0.66–0.95, P = 0.011, FDR- P = 0.039). There were no significantly associations between other clinical factors (sex, age, WHO grade and radiotherapy) and glioma prognosis (P > 0.05).

Table 6 The impact of clinical factors on glioma patient OS and PFS.
Full size table

Association of PPARD and PPARG polymorphisms with glioma prognosis

Then, we assessed the association of PPARD (rs2016520, rs67056409, rs1053049 and rs2206030) and PPARG (rs2920503, rs4073770 and rs1151988) polymorphisms with glioma prognosis. In Supplemental Table 2, univariate analysis did not show a strong relationship of PPARD and PPARG polymorphisms with glioma prognosis (P > 0.05). Moreover, we did not observe significantly association of PPARD and PPARG polymorphisms with OS and PFS of glioma patients (P > 0.05, Supplemental Table 3).

Discussion

In this case-control study, we examined the association of PPARD and PPARG polymorphisms with glioma risk and prognosis in the Chinese Han population. After FDR correction, we found that PPARD polymorphisms were significantly associated with glioma risk, and the effects were dependent on age (P < 0.05, FDR- P < 0.05). Moreover, surgery method and chemotherapy had strongly effects on glioma prognosis (Log-rank P < 0.05, P < 0.05, FDR- P < 0.05).

PPARs are involved in the regulation of metabolic homeostasis, whose activity are controlled by fatty acid ligands34. After activation, PPARs heterodimerize with retinoid X receptors (RXRs) to affect the expression of downstream genes. It is reported that PPARs might had a functional crosstalk concerning the control of their expression35. Previous studies on the role of PPARs signaling in cancer mainly based on the availability of PPARs agonists and antagonists36. In brain tumor stem cells, PPARγ agonists inhibit cell growth and induce cell cycle arrest37. In mice, expression of PPARδ is related to prognosis and metastatic ability of breast cancer cells38. Polymorphisms of PPARD and PPARG are associated with risk and prognosis of many diseases, including cardiovascular disease, diabetes, brain diseases, medulloblastoma and other cancers39,40,41. In our study, we firstly observed that PPARD polymorphisms (rs2016520, rs67056409 and rs1053049) were significantly associated with glioma risk. Similar association has been reported in colorectal cancer39. It suggests that PPARD polymorphisms could be involved in the susceptibility of glioma development. And, stratified analysis showed the effects of PPARD polymorphisms on glioma risk were age-dependent. It provides a scientific basis on individualized treatment of glioma. The effects of PPARD polymorphisms on glioma risk might related to SiPhy cons, Promoter histone marks, Enhancer histone marks, DNAse, Motifs changed, GRASP QTL hits, NHGRI/EBI GWAS hits, Selected eQTL. However, our results should be confirmed in further studies, including next-generation technology, PCR, western-blot analysis, etc.

Glioma is likely to have unfavorable prognosis caused by rapid proliferation and diffuse brain invasion. Despite surgery, chemotherapy and radiotherapy treatments improve, the prognosis of glioma remains poor42. Recent studies reported that some lipophilic molecules have antiproliferation and/or differentiation effects on glioma cells, and PPARs mediated some activities of these processes43. PPARγ has been observed in transformed neural cells of human and PPARγ agonist interferes with glioma growth and malignancy43,44,45. In this study, we firstly confirmed the effects of surgery method and chemotherapy on prognosis of glioma patients. Then, we explored the association of PPARD and PPARG polymorphisms with OS and PFS of glioma patients. No significant associations were observed by univariate and multivariate analysis. It demonstrated that PPARD and PPARG polymorphisms might not contribute the prognosis of glioma.

There are some limitations in the present study. First, we selected and genotyped several polymorphisms of PPARD and PPARG, more genetic polymorphisms should be studied in the future. Second, we could not evaluate more factors on the association of genetic polymorphisms and glioma risk due to the limited sample size and information. Third, the molecular mechanisms of PPARD and PPARG on glioma risk and prognosis are not elucidated in our study.

Conclusion

In conclusion, we found genetic polymorphisms of PPARD were associated with glioma risk in the Chinese Han population, which suggests the role of PPARD in the carcinogenesis of glioma.

It provided information on exploring the mechanism and targeted therapy of glioma, it also promotes the development of precision medicine on glioma. Further studies in larger samples with more ethnic groups are needed to validate our results and explore the mechanism of PPARD and PPARG in glioma.