Introduction

Glioma is a tumor that originates from neuroectodermal mesenchymal cells and accounts for about 40–50% of brain tumors. It is the most common intracranial malignant tumor (McNeill 2016). According to the 2016 World Health Organization glioma classification standard, gliomas mainly include astrocytoma, oligodendrocytoma, oligoastrocytoma, et al. (Gupta and Dwivedi 2017). Glioma has a high recurrence rate and high mortality. And in the clinical treatment, it often appears insensitive to radiotherapy or resistance to chemotherapy, which will lead to poor clinical treatment effects and poor prognosis (Van Meir et al. 2010). At present, the specific pathogenesis of glioma is not very clear. Therefore, glioma has always been one of the most difficult problems in neurosurgery. Studies have shown that in addition to the effects of high-dose ionizing radiation, genetic susceptibility genes may play a certain role in the pathogenesis of glioma (Tanyıldız et al. 2016). At present, some studies on the association between genetic polymorphisms and glioma have been reported worldwide (Chen et al. 2015; Custódio et al. 2011; He et al. 2016; Jiao et al. 2016; Shamran et al. 2014). Although these studies have let us to gain some new insights into the pathogenesis of glioma, we have not found an effective, specific, and unified method for prevention and treatment. Therefore, finding new and effective genetic markers is still very important, which will help us to judge the prognosis of glioma patients early and then conduct targeted interventions for treatment.

N-myc downstream regulated gene-1 (NDRG1) was cloned and isolated in 1997 for the first time and has been found in many cancers (Azuma et al. 2012), such as pancreatic cancer (Stein et al. 2004), prostate cancer (Kovacevic et al. 2011), and esophageal squamous cell carcinoma (Rabouille and Klumperman 2005). NDRG1 has also been found to be involved in embryogenesis and development, cell growth and differentiation, lipid synthesis, stress response, immune function, and myeloid formation (Kovacevic and Richardson 2006). Most importantly, NDRG1 may play an inhibitory role in the development of glioma and may be a potential prognostic indicator for glioma (Sun et al. 2009).

There is evidence that relatives of patients with glioma have a higher risk of glioma (Hemminki et al. 2009). And some studies have shown that the gene polymorphisms, one of genetic variation, are considered as a risk factor for glioma (Wrensch et al. 2005). The genome-wide association study of glioma has reported the association between gene polymorphisms and the risk of glioma, such as CDKN2B, RTEL1, and PHLDB1 et al. (Shete et al. 2009). However, no research on the association between NDRG1 gene polymorphisms and glioma risk or patient prognosis has been found. Therefore, we explored the association between 5 candidate SNPs on NDRG1 (rs2272646 A/G, rs3779941 C/A, rs3808599 G/C, rs2977497 T/C, rs3802251 C/T) and glioma risk or patient prognosis in Chinese Han population through the experimental design of ‘case–control’.

Materials and Methods

Study Subjects

In this study, 1061 participants (558 glioma patients and 503 healthy individuals) were recruited at the department of Neurosurgery at Tangdu Hospital (Xi’an, China) during the same period, and then, we conducted a study on the association between NDRG1 SNPs and the risk of glioma. At the same time, we also explored the impact of NDRG1 SNPs on the prognosis of patients with glioma. 558 cases were composed of glioma patients in the Department of Neurosurgery at Tangdu Hospital (Xi’an, China), and 503 healthy individuals were collected from the physical examination center of Tangdu Hospital (Xi’an, China) during the same period. All glioma patients meet the WHO diagnostic criteria for central nervous system tumors, while none of the healthy individuals has a history of cancer or central nervous system disease. All participants did not have any blood diseases. This study adopts the ‘case–control’ research method as a whole. In order to get the basic demographic and epidemiological information of all participants (age, gender, WHO grade, surgical operation, radiotherapy, chemotherapy, astrocytomas), we collected useful information through medical records, questionnaire surveys, and follow-up. Finally, after obtaining the informed consent of all participants, we collected peripheral blood samples from each of them for subsequent DNA extraction (blood collection for glioma patients must be done before radiotherapy, chemotherapy, and surgery). The study has been approved by the ethics committee of the Northwest University, and the follow-up work was carried out after obtaining the informed consent of all patients.

Selection and Genotyping of SNPs

Combining the relevant information of NDRG1 gene polymorphisms in the dbSNP database, we selected candidate SNPs with an allele frequency ≥ 5%. Then, five SNPs on NDRG1 were selected for our study (rs2272646 A/G, rs3779941 C/A, rs3808599 G/C, rs2977497 T/C, rs3802251 C/T). We extracted and purified the whole genome DNA according to the experimental procedures from the kit instructions (GoldMag Co. Ltd. Xi’an, China). Subsequently, we used NanoDrop 2000 to test the purity and concentration of DNA samples, and the test results showed that the OD260/280 of all DNA was between 1.8 and 2.0, indicating that the purity of DNA samples was good, which was conducive to subsequent studies. Afterwards, the extracted DNA was stored in a low-temperature refrigerator (− 80 °C) until needed for the next experiment. The primers required for this study were all designed by MassARRAY Assay Design software, and finally the MassARRAY (Agena, San Diego, CA, USA) system was used by us for genotyping.

In order to ensure the reliability and reproducibility of the experimental results, we randomly select 5% of DNA samples for repeatability testing. And the repetition rate of experimental results is > 99%.

Statistical Analysis

The Association Between SNPs on NDRG1 and the Risk of Glioma

The difference in demographic characteristics in this study was tested by SPSS 17.0 statistical software. The p value represents whether it is statistically significant (p < 0.05: statistically significant). After testing whether all candidate SNPs meet the Hardy–Weinberg balance (HWE), the correlation between the candidate SNPs and the risk of glioma was studied. The study included overall analysis and subgroup analysis (age, gender, astrocytomas). Using wild-type alleles as a reference, the plink 1.07 online tool software was used to estimate multiple genetic models (codominant, dominant, recessive, and logarithmic addition). The analysis results of this part were all estimated based on the odds ratio (OR) and 95% confidence interval (CI) obtained by the logistic regression model adjusted by age and gender (OR 1: the factor has no effect on the occurrence of the disease; OR < 1: reduce the risk of disease; OR > 1: increase the risk of disease). Finally, we used multi-factor dimensionality reduction (MDR) to evaluate the interactions of candidate ‘SNP–SNP’ in the risk of glioma.

Prognosis Analysis of 558 Patients with Glioma

The overall prognosis analysis is based on SPSS 17.0 software for statistical analysis. Univariate survival analysis used the Kaplan–Meier method to calculate the median survival time and 1-year, 2-year, and 3-year survival rates of patients. The Log-rank test was used to compare survival risks. The Cox hazard proportional regression model was used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs), and then, we evaluated the impact of NDRG1 genotype on the overall survival and progression-free survival of glioma patients. We also used Kaplan–Meier method and Log-rank test to draw the corresponding survival curves of glioma patients.

All tests in this study were two-sided tests, and p < 0.05 was considered statistically significant.

Results

Sample Overview

This study adopted a ‘case–control’ experimental design. The average age of glioma patients was 40.52 ± 18.08 years, including 307 males (55%) and 251 females (45%); the average age of healthy individuals was 40.75 ± 13.99 years, including 280 males (56%) and 223 females (44%). Table 1 summarizes the demographic (age and gender) and clinical information (WHO grade, astrocytoma, surgical operation, radiotherapy status, and chemotherapy status) of the participants. We found that there was no statistical difference between the case group and the control group in gender (p = 0.853) and age (p = 0.817).

Table 1 Characteristics of patients with glioma and healthy individuals
Full size table

Genotyping and Candidate SNPs-Related Information

5 candidate SNPs (rs2272646 A/G, rs3779941 C/A, rs3808599 G/C, rs2977497 T/C, rs3802251 C/T) on NDRG1 were successfully genotyped. Detailed information about these five candidate SNPs is summarized in Table 2. All candidate SNPs were in line with HWE (p > 5%). The results of HaploReg showed that the SNPs in this study may be regulated by many factors, including Enhancer histone marks, DNAse, Motifs changed, GRASP QTL Hits, Selected eQTL Hits, and Promoter histone marks.

Table 2 The basic information and HWE about the selected SNPs of NDRG1
Full size table

Evaluation of the Correlation Between NDRG1 SNPs and Glioma Risk

Overall Analysis

The association between SNPs on NDRG1 and glioma risk under multiple genetic models was tested based on logistic regression, and the results were adjusted by age and gender (Table 3). The results showed that among the five candidate SNPs, rs3808599 or rs3802251 and the risk of glioma may have a certain association. Specifically, rs3808599 on NDRG1 can reduce the risk of gliomas in homozygous (GG vs. CC, OR 0.41, CI 0.19–0.89, p = 0.024) and recessive models (GG vs. GC-CC, OR 0.42, CI 0.19–0.90, p = 0.025); rs3802251 on NDRG1 can also reduce gliomas risk in allelic (C vs. T, OR 0.79, CI 0.67–0.94, p = 0.008), homozygous (CC vs. TT, OR 0.63, CI 0.44–0.90, p = 0.011), dominant (CC-CT vs. TT, OR 0.73, CI 0.56–0.95, p = 0.017), and log-additive models (OR 0.79, CI 0.66–0.94, p = 0.008). We did not find any evidence of the association between the remaining three candidate SNPs and glioma risk.

Table 3 Analysis of the association between glioma and SNPs of NDRG1
Full size table

Age and Gender

The results showed (Table 4) that rs3808599 on NDRG1 reduced the risk of glioma among the participants ≤ 40 years old under the homozygous model (GG vs. CC, OR 0.30, CI 0.11–0.83, p = 0.020) and the recessive model (GG vs. GC-CC, OR 0.29, CI 0.11–0.82, p = 0.019); the rs3802251 on NDRG1 can also reduce the risk of glioma among males of the participants under heterozygous (CT vs. TT, OR 0.69, CI 0.47–1.00, p = 0.049) and dominant models (CC-CT vs. TT, OR 0.68, CI 0.48–0.97, p = 0.0033). We also found that rs3802251 only showed the ability to reduce the risk of glioma in participants > 40 years old under the allelic model (C Vs. T, OR 0.78, CI 0.61–0.99, p = 0.049), but the p value was infinitely close to the critical value (0.05). If it is inferred from the above results that rs3802251 has a significant association with the risk reduction of glioma among participants > 40 years old, the reason may be insufficient. Therefore, it is very necessary to carry out necessary verification experiments in the future. In addition, we did not find evidence that there is an association between the five candidate SNPs and the risk of glioma in the female participants.

Table 4 The SNPs of NDRG1 associated with risk of glioma in the subgroup tests (age and gender)
Full size table

Astrocytoma

The results showed (Table 5) that rs3802251 on NDRG1 has a certain association with astrocytoma patients in allelic (C vs. T, OR 0.81, CI 0.67–0.97, p = 0.023), homozygous (CC vs. TT, OR 0.67, CI 0.46–0.99, p = 0.043), dominant (CC-CT vs. TT, OR 0.75, CI 0.57–0.99, p = 0.042), and log-additive models (OR 0.81, CI 0.68–0.98, p = 0.031), and it showed a risk in reduction effect (OR < 1).

Table 5 The SNPs of NDRG1 associated with risk of glioma in the subgroup tests (astrocytoma)
Full size table

WHO Grade

The results showed (supplemental Table 1) that there may be no association between the five candidate NDRG1 SNPs and the WHO grade of glioma in Chinese Han population.

MDR Analysis

MDR analysis was used to evaluate the interactions between ‘SNP–SNP’. Figure 1 can describe the interaction between 5 candidate SNPs. The blue line indicated that the candidate SNPs may have a redundant role in regulating the risk of glioma. All experimental results have been shown in Table 6: The best single-point model for predicting the risk of glioma is rs3802251 (testing accuracy = 0.539, CVC = 10/10, p = 0.0094); the two-site model is rs3779941, rs3802251 (testing accuracy = 0.511, CVC = 4/10, p = 0.0003); the three-site model is rs3779941, rs3808599, rs3802251 (testing accuracy = 0.504, CVC = 4/10, p < 0.0001); the four-site model is rs2272646, rs3808599, rs2977497, rs3802251 (testing accuracy = 0.511, CVC = 5/10, p < 0.0001); and the five-site model is rs2272646, rs3779941, rs3808599, rs2977497, rs3802251 (testing accuracy = 0.540, CVC = 10/10, p < 0.0001). Therefore, our analysis concluded that the impact of the five candidate SNPs on the risk of glioma may be interdependent.

Fig. 1
figure 1

Dendrogram analysis of SNP–SNP interaction (NDRG1). The colors in the tree diagram represent synergy (yellow) or redundancy (blue)

Full size image
Table 6 SNP–SNP interaction models analyzed by the MDR method
Full size table

Haplotype Analysis of NDRG1

The results of linkage disequilibrium (LD) and haplotype analysis of NDRG1 polymorphism showed that the LD block (Fig. 2) was composed of two SNPs (rs2272646 and rs3779941). In addition, in the haplotype analysis, we also adjusted for the effects of covariates (age and gender). Haploidy frequency (case group/control group) was shown in Supplemental Table 2. The logistic regression results show that there is no haplotype significantly related to the risk of glioma.

Fig. 2
figure 2

Haplotype block map for 2 SNPs in NDRG1 gene. The numbers inside the diamonds indicate the D′ for pairwise analyses

Full size image

Prognosis Analysis of 558 Patients with Glioma

Overall

A follow-up survey was conducted on 558 glioma patients in this study, and the follow-up time was 1–36 months. Based on the follow-up records, we conducted a univariate analysis between overall survival (OS) or progression-free survival (PFS) and clinical factors in 558 glioma patients. These clinical factors include: gender, age, WHO grade, surgical operation, radiotherapy status, and chemotherapy status (Table 7 and Fig. 3). Our results showed that from the perspective of surgical resection methods, glioma patients with total tumor resection (OS: log-rank p < 0.001, HR = 0.62; PFS: log-rank p < 0.001, HR = 0.60) had a better prognosis than patients with non-total resection, and the result was statistically significant (p < 0.001). For the radiotherapy, glioma patients who have undergone gamma knife radiotherapy were associated with an increased risk of PFS (PFS: log-rank p = 0.039, HR = 1.40, p = 0.041). For the chemotherapy, compared with glioma patients who have not undergone chemotherapy, patients after chemotherapy have a better prognosis (OS: log-rank p < 0.001, HR = 0.70, p < 0.001). However, we did not find any evidence that other clinical factors (gender, age, WHO grade) were related to the prognosis of glioma patients.

Table 7 Univariate analysis of the influence of clinical factors on glioma patient OS and PFS
Full size table
Fig. 3
figure 3

Kaplan–Meier curves for overall survival and progression-free survival according to the glioma patients with different clinical factors (a OS according to surgical operation; b OS according to chemotherapy status; c PFS according to surgical operation; d PFS according to radiotherapy status)

Full size image

Astrocytoma Patients

The results showed that female astrocytoma patients have a potential association with progression-free survival (PFS: log-rank p = 0.029, HR = 1.23, p = 0.050). As shown in Table 8 and Fig. 4a and b, patients with total tumor resection had a better prognosis than patients with non-total resection (OS: log-rank p < 0.001, HR = 0.62, p < 0.001; PFS: log-rank p < 0.001, HR = 0.58, p < 0.001). For radiotherapy (Table 8 and Fig. 4c), compared with astrocytoma patients who are not undergoing radiotherapy, the results showed that no matter what kind of radiotherapy was given, it was associated with an increased risk of PFS in astrocytoma patients (log-rank p = 0.031, Conformal radiotherapy: HR = 1.59, p = 0.023; Gamma knife: HR = 1.50, p = 0.029). For chemotherapy, astrocytoma patients who have undergone chemotherapy have a better prognosis (OS: log-rank p < 0.001, HR = 0.62, p < 0.001). Table 8 summarizes the experimental results after univariate analysis.

Table 8 Univariate analysis of the influence of clinical factors on astrocytoma patient OS and PFS
Full size table
Fig. 4
figure 4

Kaplan–Meier curves for overall survival and progression-free survival according to the astrocytoma patients with different clinical factors (a OS according to surgical operation; b PFS according to surgical operation; c PFS according to radiotherapy status)

Full size image

SNPs and the Prognosis of Glioma Patients (Univariate Analysis)

We evaluated the impact of five candidate SNPs on the survival rate of glioma patients. The results are shown in Table 9 and Fig. 5, and we found that rs3779941 has a potential impact on the OS and PFS of glioma patients (OS: log-rank p = 0.006; PFS: log-rank p = 0.040). At the same time, we also found an evidence that the genotype CC of rs3779941 was associated with the increased risk of OS in glioma patients (OS: HR = 3.07, 95% CI 1.27–7.44, p = 0.013).

Table 9 Univariate analysis of the association between SNPs in NDRG1 and glioma patient OS and PFS
Full size table
Fig. 5
figure 5

Glioma patient survival based on NDRG1 rs3779941 polymorphism. Kaplan–Meier survival curves are plotted for overall and progression-free survival (a OS based on NDRG1 rs3779941 polymorphism; b PFS based on NDRG1 rs3779941 polymorphism)

Full size image

SNPs and the Prognosis of Glioma Patients (Multivariate Analysis)

After Cox multivariate analysis (multivariate: gender, age, WHO grade, radiotherapy, surgical operation, chemotherapy), the results showed that the rs3779941 polymorphism was associated with prognosis of glioma patients (Table 10). Specifically, the genotype CC of rs3779941 was a risk factor that increases the risk of OS (OS: HR = 2.59, 95% CI 1.05–6.37, p = 0.039) in glioma patients, but there was no association with PFS (HR = 1.77, 95% CI 0.72–4.35, p = 0.212). There did not seem to be any association between the remaining candidate SNPs and the OS or PFS of glioma patients.

Table 10 Multivariate analysis of the association between SNPs of NDRG1 and glioma patient OS and PFS (overall and astrocytoma)
Full size table

SNPs and the Prognosis of Astrocytoma Patients (Multivariate Analysis)

Finally, we also performed the association analysis between NDRG1 gene polymorphisms and the prognosis of astrocytoma patients. The results showed (Table 10) that the genotype CC of rs3779941 was a risk factor that increased the risk of OS in astrocytoma patients (OS: HR = 2.63, 95% CI 1.06–6.56, p = 0.038), but there was no association with PFS (HR = 1.78, 95% CI 0.72–4.42, p = 0.211). There did not seem to be any association between the remaining candidate SNPs and the OS or PFS of astrocytoma patients.

Discussion

Glioma is the tumor with the highest incidence and the worst prognosis among primary brain tumors, posing a great threat to human health. With the development of sequencing technology and genome-wide association studies (GWAS), more and more studies have proved that in addition to external factors such as high-dose ionizing radiation, genetic susceptibility genes also play a certain role in the occurrence and development of glioma (Ostrom et al. 2014; Tanyıldız et al. 2016), such as POLR3B, VTI1A, ZBTB16, ETFA, etc. (Kinnersley et al. 2018). Up to now, there is no report about the association between NDRG1 gene polymorphisms and the occurrence and prognosis of glioma. However, studies have shown that NDRG1 is necessary to inhibit the occurrence of glioma (Ma et al. 2015). This study was the first to explore the relationship between the five polymorphisms of NDRG1 (rs2272646 A/G, rs3779941 C/A, rs3808599 G/C, rs2977497 T/C, rs3802251 C/T) and the genetic risk of glioma or the prognosis of patients in the Chinese Han population. And as far as we know, this study is the first to find that the NDRG1 SNPs (rs3779941, rs3808599, and rs3802251) are potentially associated with glioma susceptibility or prognosis. In addition, in the prognosis analysis of glioma patients, we found that surgical operation, radiotherapy, and chemotherapy are the key factors for the prognosis of glioma patients in the Chinese Han population.

It has been found that NDRG1 plays an important role in regulating the pathogenesis/molecular mechanisms of tumor cells. Based on previous studies (Byun et al. 2018; Ito et al. 2020), NDRG1 is found to be responsible for the regulation on cellular proliferation, apoptosis, invasion, migration, and metastasis in tumor tissues. At present, NDRG1 has been proposed as a tumor suppressor gene in a variety of cancers, including breast cancer (Bandyopadhyay et al. 2004), colon cancer (Guan et al. 2000), and prostate cancer (Bandyopadhyay et al. 2003). And NDRG1 is also necessary for inhibiting the occurrence of glioma (Ito et al. 2020; Ma et al. 2015). Sun et al. found that the expression level of NDRG1 in high-grade glioma tissue is relatively lower than that in normal brain tissue or low-grade glioma tissue (Sun et al. 2009). These research results prompted that NDRG1 may be an internal regulator that can affect the occurrence and development of glioma, and the expression of NDRG1 may play a very important role in the progression and prognosis of glioma.

In this study, we found that rs3808599 on NDRG1 can reduce the risk of glioma in homozygous and recessive models, whether in the overall participants (homozygous: OR 0.41; recessive: OR 0.42) or in the participants ≤ 40 years old (homozygous: OR 0.30; recessive: OR 0.29); rs3802251 on NDRG1 can significantly reduce the risk of glioma in the overall participants, male participants, and astrocytoma patients under variety of genetic models; the prognostic analysis after the follow-up investigation found that rs3779941 on NDRG1 was significantly associated with the prognosis of glioma patients in our study. At the same time, we also found that the three candidate SNPs (rs3779941, rs3808599 and rs3802251), which are potentially associated with the risk or prognosis of glioma in this study, are all located in the intron region. And there have been several studies suggesting that mutants located in the intron region can disrupt transcriptional regulatory motifs by affecting gene expression, which will affect the occurrence and development of diseases (Chang et al. 2019; Vaz-Drago et al. 2017; Zhao et al. 2012). Combined with the results of this study, we speculate that rs3779941, rs3808599, and rs3802251 may affect the occurrence and development of glioma by affecting the gene expression of NDRG1 and disrupting the transcriptional regulatory motifs.

More importantly, studies have shown that NDRG1 can inhibit the proliferation and invasion of glioma cells, and overexpressed NDRG1 will inhibit the growth of glioma tumors in vivo (Ma et al. 2015). It follows that NDRG1 inhibits the proliferation and invasion of glioma cells and other cellular behaviors, which rs3779941, rs3808599, and rs3802251 of NDRG1 may play a certain role. Our study has provided new ideas for the diagnosis and prognosis analysis of clinical glioma. Perhaps from the viewpoint of ‘how rs3779941, rs3808599 and rs3802251 affect the expression of NDRG1 in glioma cells’, we will have the opportunity to understand the specific molecular mechanism of NDRG1 in risk/prognosis of glioma.

Many studies have found that WHO grade is an important factor in the survival rate of glioma patients, and it is often used by doctors to predict the survival rate of glioma patients (Desjardins et al. 2018; Liu et al. 2018; Rasmussen and Hansen 2017). However, our results showed that the five candidate NDRG1 SNPs were not associated with the WHO grade in glioma patients. Our results may indicate that the five candidate NDRG1 SNPs did not play any role in the effect of WHO grade on the survival rate of Chinese Han glioma patients. We speculate that the reason for the result may be due to the difference of the genetic background or the small sample size. But this is only speculation, and we need further experiments to verify it. In this study, we also found that NDRG1 rs3802251 was associated with the risk of astrocytoma in multiple genetic models (Table 5). However, the upper limit of 95% CI is all close to 1.0, and the p value is still < 0.05. We speculate that the reason may be the lack of samples in the subgroup. Therefore, it is necessary to expand the sample size to verify the correlation between NDRG1 rs3802251 and the susceptibility of astrocytoma, which will make the results of this study more reliable.

There are inevitably several shortcomings in our research. On the one hand, enlarging the sample size and selection range is necessary in the following research. On the other hand, this study is only a preliminary research. Therefore, in order to clearly clarify the molecular mechanism of how the NDRG1 SNPs affect the risk or prognosis of glioma, it is necessary to further explore how these variants (rs3779941, rs3808599 and rs3802251) affect the expression of NDRG1. It will help us to understand the mechanism of NDRG1 genetic polymorphism in the occurrence and development of glioma. Despite the abovementioned deficiencies in this study, the results of our study have provided data supplement for the risk assessment of glioma in Chinese Han population.

Conclusion

In summary, the results of this study showed that the NDRG1 gene polymorphisms have a potential association with the risk or prognosis of glioma in the Chinese Han population, which provides new ideas for the risk assessment and prognosis evaluation of glioma in the Chinese Han population.