Genetic Variations of CARMN Modulate Glioma Susceptibility and Prognosis in a Chinese Han Population.

Background: This study aimed to evaluate the relationship between CARMN polymorphisms and glioma risk and prognosis in a Chinese Han population.
Methods: Seven single nucleotide polymorphisms (SNPs) in CARMN were genotyped among 592 glioma patients and 502 healthy controls. Log-additive models were used for risk assessment by the odds ratios (ORs) and 95% confidence intervals (CIs). Univariate and multivariate Cox regression analysis was applied to calculate Hazard ratios (HRs) and 95% CIs for prognosis assessment.
Results: CARMN rs13177623 was a protective factor for glioma susceptibility (OR = 0.78, p = 0.043). In addition, rs13177623, rs11168100, rs12654195 and rs17796757 were associated with the risk of glioma among the subgroup stratified by age or gender. We also found that G rs13177623G rs12654195 haplotype was related to the decreased risk of glioma (OR = 0.61, p = 0.005). Importantly, rs13177623 [overall survival (OS): HR = 0.83, p = 0.047, and progression free survival (PFS): HR = 0.82, p = 0.031], rs12654195 (OS: HR = 0.64, p = 0.005 and PFS: HR = 0.65, p = 0.007) and rs11168100 (OS: HR = 0.71, p = 0.035) were associated with a better prognosis for glioma, especially in grade I-II glioma. In patients with grade III-IV glioma, rs17796757 polymorphism presented an improved OS.
Conclusion: Our results firstly reported the contribution of CARMN variants (rs11168100, rs12654195, rs13177623, and rs17796757) to the susceptibility and prognosis of glioma in a Chinese Han population, which provided a novel insight on the relationship between CARMN gene and glioma tumorigenesis.

Introduction

Glioma is the most common intracranial malignant tumor derived from glial cells, accounting for the majority of all primary brain and central nervous system tumors.1 It is characterized by the significant mortality and morbidity of approximately 101,600 new cases and 61,000 deaths in China each year.2 Malignant glioma is a devastating type of brain and other nervous system tumors because of its high malignancy, extremely high mortality rate, and recurrence risk.3 Despite improvements in therapeutics including surgery in combination with chemo- and/or radiotherapy, the five-year relative survival rate following diagnosis of a malignant brain still grim.4 The etiology of glioma remains poorly understood to date, but environmental exposure and genetic factors are identified to increase glioma risk. In recent years, the role of inherited genetic variants in glioma has been highly addressed, which revealed single nucleotide polymorphisms (SNPs) in genes contribute to the susceptibility and prognosis of glioma.5–7

Long non-coding RNAs (lncRNAs) are a class more than 200 nucleotides non-protein coding RNA, that regulate gene or miRNA expression at the transcriptional, post-transcriptional and epigenetic levels.8 LncRNAs participate in different stages of glioma formation, invasion, and progression.7 Recent evidence indicates that genetic variations in functional lncRNAs may play important roles in the occurrence and development of glioma, such as genetic polymorphisms in lncRNA-PTENP1 and lncRNA H19.9,10

Cardiac mesoderm enhancer-associated non-coding RNA (CARMN) is a newly identified lncRNA, also named MIR143HG, and has been reported to be the precursor of miR-143 and miR-145, which linked to gliomagenesis.11,12 MicroRNA-145-5p downregulation has been shown to play important roles in the oncogenesis and progression of many cancer types including glioblastoma.13 Furthermore, miR‑143 inhibited glioma cells migration and invasion through cytoskeletal rearrangement.14 Ropivacaine suppressed glioma progression by regulating circSCAF11 and miR-145-5p.15 These suggested CARMN, the host gene of miR-143 and miR-145, might have an important role in the occurrence and development of glioma. Nevertheless, no association studies between CARMN polymorphisms and glioma have been published to date.

Considering the effect of genetic variants on glioma, we hypothesized that CARMN polymorphisms might contribute to glioma development and prognosis. Here, we conducted a case–control study to evaluate the role of CARMN polymorphisms in glioma and found that four SNPs were significantly related to glioma risk and patients survival in a Chinese Han population.

Materials and Methods

Study Subjects

In this study, 592 glioma patients and 502 healthy controls enrolled from the department of Neurosurgery at Xi’an Children’s Hospital and Tangdu Hospital. All included patients had recently diagnosed and histopathologically confirmed glioma according to the World Health Organization (WHO) classification. All subjects had a Han Chinese ethnic background. All glioma patients were newly diagnosed and confirmed by histopathology. The blood samples were collected before radiotherapy and chemotherapy or surgery. Patients with a self-reported cancer history, serious systemic diseases or other complex diseases were excluded. Age and gender matched healthy controls were enrolled from annual checkup at the same hospitals. The controls had no any cancers or chronic diseases and no brain and central nervous system diseases. Demographic and clinical information was collected from structured questionnaires and/or medical records. All the patients were followed up every 3 months by return visit, telephone and letter. During the follow-up period, the survival time was recorded until death or the last follow-up. This study was approved by the institute ethics committee of the Xi’an Children’s Hospital (No. 20200014) and in accordance with the Helsinki Declaration. Written informed consent was obtained from each participant.

Genotyping of CARMN Polymorphisms

Peripheral blood samples (5 mL) were collected from all of the study participants. Genomic DNA was extracted using the commercially available GoldMag-Mini Whole Blood Genomic DNA Purification Kit (GoldMag Co. Ltd., Xiʹan, China), and stored at −80°C until analysis. The candidate variants in CARMN were selected based on a minor allele frequency (MAF) of > 5% in Chinese populations of the 1000 Genomes Project data (), a pairwise linkage disequilibrium (LD) r2 ≥ 0.80, in conformance with Hardy–Weinberg equilibrium (HWE, p > 0.05) and the genotyping call rate > 95%. Seven CARMN SNPs (rs11168100, rs12654195, rs13177623, rs17796757, rs353299, rs353300 and rs353303) were included for genotyping in the current study.16 Agena MassARRAY platform (Agena, San Diego, CA, USA) was applied to determine the genotypes of CARMN polymorphisms as described previously.17,18 The MassARRAY platform is based on MALDI-TOF (matrix-assisted laser desorption/ionization-time of flight) mass spectrometry in a high-throughput and cost-effective manner. Primers for amplification and extension were designed by Agena on-line design software (), as shown in . The steps for SNPs genotyping were based on manufacturer’s protocol, as following: 1) targeted regions for the multiplex assay were amplified by PCR; 2) PCR products were treated through shrimp alkaline phosphatase (SAP) to neutralize unincorporated nucleotides; 3) single base extension reaction were then performed to extend the PCR fragments by one base into the SNP site; 4) the mass of the resultant extended fragments were measured by MALDI-TOF, resulting in a spectrum of distinct mass peaks for the multiplex reaction. The process of genotyping was in double-blinded by two laboratory personnel. For quality control, 10% random sample was repeated genotyping, and the reproducibility was 100%.

Statistical Analyses

SPSS 18.0 (SPSS, Chicago, IL, USA) and PLINK 1.07 package was used for statistical analyses. Differences between cases and controls in demographic characteristics were evaluated by χ2 test or independent samples t-test where appropriate. The frequencies of allele and genotype of CARMN polymorphisms in cases and controls were calculated by χ2 test. HWE was tested for controls with the χ2 test. The association between CARMN genetic variants and glioma risk was estimated by the odds ratios (ORs) and 95% confidence intervals (CIs) after adjusting for age and sex using logistic regression under allele genotype, dominant, recessive and log-additive models, respectively. The pairwise linkage disequilibrium (LD) were measured by the Lewontin’s coefficient D’ using the Haploview v4.2 software, and haplotype association tests for glioma susceptibility were carried out using logistic regression analysis. Univariate and multivariate Cox regression analysis was applied to calculate Hazard ratios (HRs) and 95% CIs for evaluating the association of CARMN polymorphisms with glioma prognosis. Survival analysis of glioma patients was assessed by Kaplan–Meier survival curves and the log rank test. A two-sided p values of <0.05 were regarded as statistically significant.

Results

Participant Characteristics

The subjects included 592 glioma samples (40.53 ± 13.90, 326 males and 266 females) and 502 cancer-free controls (40.46 ± 18.08, 275 males and 227 females). The frequency distribution of age and sex was matched between cases and controls (p = 0.934 and p = 0.924, respectively). Other clinical details of patients with glioma such as WHO grade, surgical method, radiotherapy, chemotherapy and survival condition were presented in Table 1.

Table 1

Features of Glioma Patients and Health Controls

Features	Cases (n = 592)	Controls (n = 502)	p
Age (Mean ± SD, years)	40.53 ± 13.90	40.46 ± 18.08	0.934a
≥ 40	329	249
< 40	263	253
Gender
Male	326	275	0.924b
Female	266	227
WHO grade
I–II	378
III–IV	214
Surgical method
STR & NTR	185
GTR	407
Radiotherapy
No	58
Conformal radiotherapy	159
Gamma knife	375
Chemotherapy
No	349
Yes	243
Survival condition
Survival	41
Lost	24
Death	527

Notes: ap values was calculated by independent samples t-test. bp values was calculated by Chi-square tests.

Abbreviations: WHO, World Health Organization; NTR, near-total resection; STR, sub-total resection; GTR, gross-total resection.

Details of CARMN Genetic Polymorphisms

Seven genetic polymorphism in CARMN was genotyping and the call rate was > 99.7%. Details of CARMN genetic polymorphisms were displayed in . The genotype frequencies of all variants in the controls were in HWE (p > 0.05), which suggesting selected samples could represent the whole population. We used HaploRegv4.1 to annotate the potential function of these selected SNPs (). The results found that six intronic SNPs were associated with the regulation of promoter and/or enhancer histones, DNAse, proteins bound, or changed motifs, suggesting they might exert biological functions in this way in patients.

Genetic Effects CARMN Variants of on Glioma Susceptibility

The allele and genotype distribution for CARMN variants was summarized in Table 2 and . Logistic regression analysis adjusted for age and sex was performed to examine the role of CARMN variants in glioma risk. We found that CARMN rs13177623 was a protective factor for glioma susceptibility, and GA-AA genotype of rs13177623 had a reduced glioma risk compared with GG genotype (OR = 0.78, 95% CI: 0.61–0.99, p = 0.043; Table 2). There was no statistically significant association between other CARMN variants (rs353299, rs353303, rs12654195, rs11168100, rs17796757 and rs353300) and risk for glioma (all p values > 0.05, ) in the overall participants.

Table 2

Correlation Between CARMN Variants and the Susceptibility to Glioma

SNP ID	Model	Genotype	Control	Case	OR (95% CI)	p
rs13177623	Allele	G	718	889	1
		A	286	295	0.83 (0.69–1.01)	0.059
	Genotype	GG	256	338	1
		GA	206	213	0.78 (0.61–1.01)	0.055
		AA	40	41	0.78 (0.49–1.24)	0.286
	Dominant	GG	256	338	1
		GA-AA	246	254	0.78 (0.61–0.99)	0.043
	Recessive	GG-GA	462	551	1
		AA	40	41	0.86 (0.55–1.35)	0.512
	Additive	GG+GA+AA	—	—	0.84 (0.69–1.01)	0.062

Notes: p values were calculated by logistic regression analysis with adjustments for age and gender. Bold p < 0.05 means the data is statistically significant.

Abbreviations: SNP, single nucleotide polymorphism; OR, odds ratio; CI, confidence interval.

We further investigated the correlation of CARMN variants with glioma risk by stratifying for age, sex and pathological grade. Stratified analyses by age (Table 3) displayed that rs13177623 had a lower risk of glioma (OR = 0.67, 95% CI: 0.48–0.94, p = 0.022) among the subgroup at age ≥ 40 years. CARMN rs11168100 and rs12654195 were associated with decreased the risk of glioma (OR = 0.47, 95% CI: 0.26–0.85, p = 0.012 and OR = 0.55, 95% CI: 0.31–0.96, p = 0.034, respectively), while rs17796757 increased the risk (OR = 1.50, 95% CI: 1.02–2.19, p = 0.038) among the subjects at age < 40 years. In stratified analyses by sex, rs13177623 was significantly associated with decreased risk in males under the allele (OR = 0.77, 95% CI: 0.60–0.99, p = 0.045) and dominant (OR = 0.72, 95% CI: 0.52–0.99, p = 0.043) models. However, no significant association was observed in females (all p > 0.05). These results suggested that CARMN rs13177623 polymorphism might be male specific for glioma risk. When stratified by the WHO grade, patients with III-IV glioma had a significantly lower frequency of rs13177623 GA genotype compared with patients with I-II glioma (OR = 0.66, 95% CI: 0.46–0.95, p = 0.027, ).

Table 3

Correlation of CARMN Variants with Glioma Risk Stratified by Age and Gender

SNP ID	Model	OR (95% CI)	p	OR (95% CI)	p
Age (year)		≥ 40		< 40
rs11168100	Allele	1.01 (0.79–1.30)	0.946	0.86 (0.66–1.12)	0.254
	Homozygote	1.32 (0.72–2.44)	0.372	0.54 (0.29–1.02)	0.057
	Heterozygote	0.85 (0.60–1.20)	0.344	1.38 (0.94–2.02)	0.101
	Dominant	0.91 (0.65–1.27)	0.583	1.15 (0.80–1.64)	0.462
	Recessive	1.44 (0.80–2.58)	0.226	0.47 (0.26–0.85)	0.012
	Additive	1.02 (0.79–1.31)	0.901	0.92 (0.70–1.20)	0.517
rs12654195	Allele	1.10 (0.86–1.41)	0.467	0.89 (0.69–1.15)	0.377
	Homozygote	1.52 (0.83–2.80)	0.173	0.66 (0.36–1.19)	0.164
	Heterozygote	0.94 (0.66–1.33)	0.710	1.46 (0.99–2.15)	0.057
	Dominant	1.02 (0.73–1.42)	0.923	1.22 (0.85–1.75)	0.284
	Recessive	1.58 (0.88–2.81)	0.123	0.55 (0.31–0.96)	0.034
	Additive	1.11 (0.86–1.43)	0.432	0.97 (0.74–1.26)	0.799
rs13177623	Allele	0.89 (0.69–1.16)	0.406	0.77 (0.58–1.02)	0.064
	Homozygote	1.59 (0.73–3.45)	0.241	0.56 (0.29–1.08)	0.084
	Heterozygote	0.67 (0.48–0.94)	0.022	1.05 (0.71–1.55)	0.810
	Dominant	0.74 (0.53–1.03)	0.076	0.92 (0.64–1.33)	0.663
	Recessive	1.90 (0.89–4.06)	0.098	0.55 (0.29–1.04)	0.067
	Additive	0.89 (0.68–1.17)	0.419	0.85 (0.65–1.12)	0.252
rs17796757	Allele	0.97 (0.75–1.24)	0.784	1.18 (0.91–1.54)	0.202
	Homozygote	1.24 (0.68–2.27)	0.486	1.09 (0.59–2.02)	0.778
	Heterozygote	0.79 (0.56–1.12)	0.186	1.50 (1.02–2.19)	0.038
	Dominant	0.86 (0.61–1.19)	0.358	1.41 (0.98–2.02)	0.063
	Recessive	1.39 (0.78–2.48)	0.266	0.89 (0.50–1.61)	0.709
	Additive	0.97 (0.76–1.26)	0.842	1.19 (0.90–1.56)	0.218
Gender		Male		Female
rs13177623	Allele	0.77 (0.60–0.99)	0.045	0.92 (0.69–1.22)	0.554
	Homozygote	0.66 (0.36–1.22)	0.188	0.96 (0.47–1.99)	0.920
	Heterozygote	0.73 (0.52–1.02)	0.067	0.85 (0.59–1.24)	0.405
	Dominant	0.72 (0.52–0.99)	0.043	0.87 (0.61–1.24)	0.440
	Recessive	0.76 (0.42–1.37)	0.360	1.03 (0.51–2.09)	0.940
	Additive	0.78 (0.60–1.00)	0.049	0.92 (0.69–1.22)	0.555

Notes: p values were calculated by logistic regression analysis with adjustments for age and gender. Bold p < 0.05 means the data is statistically significant.

Abbreviations: SNP, single nucleotide polymorphism; OR, odds ratio; CI, confidence interval.

We also examined the impacts of the haplotypes on glioma susceptibility. Linkage disequilibrium (LD) is a nonrandom allele association, and generated by mutation and recombination. LD is measured by the LD coefficient D’: D’ = 1 is defined as complete linkage disequilibrium; D ‘= 0 is called linkage equilibrium; and D ‘< 1 indicated that gene recombination had occurred. If there is a linkage disequilibrium between SNPs, these SNPs can form a linkage disequilibrium block. As shown in Figure 1, three LD blocks (rs13177623–rs12654195, rs11168100–rs353303 and rs353300–rs353299) were constructed from the seven variants in CARMN by coefficient D’ 0.97. In addition, we found that G rs13177623G rs12654195 haplotype was related to the decreased risk of glioma (OR = 0.61, 95% CI: 0.43–0.86, p = 0.005, Table 4).

Figure 1

The linkage disequilibrium structure of seven SNPs in the CARMN gene. Three LD blocks (rs13177623–rs12654195, rs11168100–rs353303 and rs353300–rs353299) were constructed from the seven variants in CARMN by coefficient D’ 0.97. The numbers in squares are D′ values.

Table 4

Correlation of CARMN Haplotypes with Glioma Susceptibility

Blocks	SNPs	Haplotype	Frequency		Crude Analysis		Adjusted by Age and Gender
			Case	Control	OR (95% CI)	p	OR (95% CI)	p
Block 1	rs13177623|rs12654195	AG	0.248	0.281	0.84 (0.70–1.02)	0.080	0.84 (0.70–1.02)	0.079
	rs13177623|rs12654195	GG	0.916	0.947	0.61 (0.43–0.86)	0.005	0.61 (0.43–0.86)	0.005
	rs13177623|rs12654195	GT	0.666	0.663	1.02 (0.85–1.22)	0.867	1.02 (0.85–1.22)	0.863
Block 2	rs11168100|rs353303	AG	0.408	0.413	0.98 (0.82–1.16)	0.797	0.98 (0.82–1.16)	0.796
	rs11168100|rs353303	TA	0.313	0.330	0.92 (0.77–1.11)	0.387	0.92 (0.77–1.11)	0.386
	rs11168100|rs353303	AA	0.723	0.744	0.90 (0.74–1.09)	0.272	0.90 (0.74–1.09)	0.270
Block 3	rs353300|rs353299	TT	0.854	0.861	0.95 (0.75–1.21)	0.677	0.95 (0.75–1.21)	0.678
	rs353300|rs353299	TC	0.338	0.351	0.95 (0.79–1.13)	0.538	0.94 (0.79–1.13)	0.534
	rs353300|rs353299	CC	0.515	0.508	1.03 (0.87–1.22)	0.732	1.03 (0.87–1.22)	0.727

Notes: p values were calculated using logistic regression analysis with and without adjustment by gender and age. Bold p < 0.05 indicates statistical significance.

Abbreviations: OR, odds ratio; CI, confidence interval.

Genetic Effects CARMN Variants of on Glioma Prognosis

During follow-up, there were 527 patients died of glioma, 41 patients survived and 24 patients lost. We next explored the contribution of CARMN variants to the overall survival (OS) and progression free survival (PFS) of glioma patients. The Kaplan–Meier survival curves indicated that the genotype of rs12654195 variant might be associated with OS (Log-rank p = 0.026) and PFS (Log-rank p = 0.027) of glioma patients, as shown in Figure 2. In addition, rs17796757 polymorphism had the effect on OS (Log-rank p = 0.039) of patients with grade III–IV grade III–IV glioma, while rs12654195 variant on OS (Log-rank p = 0.008) and PFS (Log-rank p = 0.011) of patients with grade I-II glioma ().

Figure 2

Kaplan–Meier survival curve for significant association of rs12654195 with OS (A) and PFS (B) of glioma patients.

The results of univariate Cox proportional hazard model revealed that GG genotype of rs12654195 had a better OS (HR = 0.71, 95% CI: 0.52–0.96, p = 0.025) and PFS (HR = 0.69, 95% CI: 0.51–0.95, p = 0.021) of glioma patients compared with TT genotype (Table 5). In patients with grade III–IV glioma, rs17796757 was significantly related to the improved OS (AT vs AA, HR = 0.71, 95% CI: 0.52–0.95, p = 0.024). In patients with grade I-II glioma, GT genotype and TT genotype of rs12654195 presented an increased OS (HR = 0.75, 95% CI: 0.59–0.94, p = 0.011, and HR = 0.66, 95% CI: 0.44–0.99, p = 0.043, respectively) and PFS (HR = 0.76, 95% CI: 0.61–0.96, p = 0.020, and HR = 0.66, 95% CI: 0.44–0.99, p = 0.046, respectively).

Table 5

Univariate Analysis of the Association Between CARMN Variants and OS and PFS of Glioma Patients

SNP ID	Genotype	OS					PFS
		Total	Events	SR (1-/3-Year)	HR (95% CI)	p	Total	Events	SR (1-/3-Year)	HR (95% CI)	p
rs12654195	TT	260	237	0.267/0.073	1		259	236	0.118/0.077	1
	GT	271	240	0.369/0.097	0.86 (0.72–1.03)	0.097	269	239	0.212/0.091	0.89 (0.74–1.06)	0.184
	GG	61	50	0.344/0.132	0.71 (0.52–0.96)	0.025	59	48	0.305/0.153	0.69 (0.51–0.95)	0.021
III–IV grade
rs17796757	AA	98	93	0.245/0.041	1		97	92	0. 124/0.049	1
	AT	90	80	0.356/0.097	0.71 (0.52–0.95)	0.024	88	79	0.170/0.097	0.81 (0.60–1.09)	0.161
	TT	26	24	0.385/0.051	0.75 (0.48–1.17)	0.208	26	24	0.247/0.041	0.82 (0.52–1.28)	0.383
I–II grade
rs12654195	TT	158	144	0.224/0.072	1		157	143	0.104/0.076	1
	GT	184	158	0.418/0.123	0.75 (0.59–0.94)	0.011	184	158	0.244/0.114	0.76 (0.61–0.96)	0.020
	GG	36	28	0.333/0.187	0.66 (0.44–0.99)	0.043	35	27	0.314/-	0.66 (0.44–0.99)	0.046

Notes: Log-rank p values were calculated using the Chi-Square test. Bold p < 0.05 indicates statistical significance.

Abbreviations: OS, overall survival; PFS, progression free survival; SR, survival rate; HR, hazard ratio; CI, confidence interval.

Further, the correlation of CARMN variants and PFS or OS was evaluated using a multivariate Cox proportional hazard model, adjusted for age, gender, WHO grade, surgical method, radiotherapy and chemotherapy (Table 6). We found rs13177623 GA genotype carriers had an improved OS (HR = 0.83, 95% CI: 0.69–1.00, p = 0.047) and PFS (HR = 0.82, 95% CI: 0.68–0.98, p = 0.031) for glioma. Rs12654195 (GG vs TT, OS: HR = 0.64, 95% CI: 0.47–0.87, p = 0.005 and PFS: HR = 0.65, 95% CI: 0.48–0.89, p = 0.007) and rs11168100 (TT vs AA, OS: HR = 0.71, 95% CI: 0.51–0.98, p = 0.035) homozygous carriers were also associated with a better prognosis for glioma. For the subgroup of patients with grade III–IV glioma, rs17796757 polymorphism presented an increased OS (AT vs AA, HR = 0.70, 95% CI: 0.51–0.95, p = 0.025). For the subgroup of patients with grade I–II glioma, the heterozygous of rs13177623 and rs11168100 were significantly associated with improved OS (HR = 0.78, 95% CI: 0.62–0.98, p = 0.030 and HR = 0.73, 95% CI: 0.58–0.92, p = 0.008, respectively) and PFS (HR = 0.79, 95% CI: 0.63–0.99, p = 0.044 and HR = 0.76, 95% CI: 0.60–0.96, p = 0.020, respectively). In addition, improved OS and PFS for grade I–II glioma was also seen for the homozygote (OS: HR = 0.62, 95% CI: 0.42–0.94, p = 0.024, and PFS: HR = 0.62, 95% CI: 0.41–0.94, p = 0.024) and heterozygous (OS: HR = 0.70, 95% CI: 0.56–0.88, p = 0.002, and PFS: HR = 0.72, 95% CI: 0.57–0.91, p = 0.006) of rs12654195 variant.

Table 6

Multivariate Analysis of the Association Between CARMN Variants and OS and PFS of Glioma Patients

SNP ID	Genotype	OS		PFS
		HR (95% CI)	p	HR (95% CI)	p
rs13177623	GG	1		1
	GA	0.83 (0.69–1.00)	0.047	0.82 (0.68–0.98)	0.031
	AA	0.72 (0.51–1.03)	0.070	0.75 (0.53–1.05)	0.096
rs12654195	TT	1		1
	GT	0.87 (0.72–1.04)	0.129	0.84 (0.70–1.01)	0.059
	GG	0.64 (0.47–0.87)	0.005	0.65 (0.48–0.89)	0.007
rs11168100	AA	1		1
	AT	0.88 (0.73–1.06)	0.167	0.84 (0.70–1.01)	0.067
	TT	0.71 (0.51–0.98)	0.035	0.74 (0.54–1.01)	0.060
III–IV grade
rs17796757	AA	1		1
	AT	0.70 (0.51–0.95)	0.025	0.75 (0.55–1.03)	0.079
	TT	0.70 (0.44–1.10)	0.123	0.75 (0.47–1.18)	0.213
I–II grade
rs13177623	GG	1		1
	GA	0.78 (0.62–0.98)	0.030	0.79 (0.63–0.99)	0.044
	AA	0.88 (0.56–1.38)	0.579	0.85 (0.54–1.33)	0.469
rs12654195	TT	1		1
	GT	0.70 (0.56–0.88)	0.002	0.72 (0.57–0.91)	0.006
	GG	0.62 (0.42–0.94)	0.024	0.62 (0.41–0.94)	0.024
rs11168100	AA	1		1
	AT	0.73 (0.58–0.92)	0.008	0.76 (0.60–0.96)	0.020
	TT	0.82 (0.53–1.27)	0.375	0.77 (0.50–1.21)	0.264

Notes: p values were calculated by Cox multivariate analysis with adjustments for gender, age, WHO grade, surgical method, use of radiotherapy and chemotherapy. Bold p < 0.05 indicates statistical significance.

Abbreviations: OS, overall survival; PFS, progression free survival; HR, hazard ratio; CI, confidence interval.

Discussion

The present study explored the possible correlation of seven polymorphisms in CARMN with the risk and prognosis of glioma among a Han Chinese population. Our results revealed that rs11168100, rs12654195, rs13177623, and rs17796757 variants were associated with the susceptibility to glioma and the OS and PFS of patients. In addition, we also found that G rs13177623G rs12654195 haplotype was a protective factor for glioma susceptibility. To the best of our knowledge, this is the first to assess the role of CARMN polymorphisms in glioma risk and prognosis.

CARMN gene, located on chromosome 5q32, is affiliated with the non-coding RNA class.19 The expression of CARMN was significantly dysregulated in various cancers and involved in carcinogenesis. For example, Lin et al reported that CARMN inhibited tumor proliferation and metastasis by suppressing MAPK and Wnt signaling pathways in hepatocellular carcinoma.20 CARMN suppressed miR-21 through methylation to inhibit cell invasion and migration.21 CARMN have reported expressing stably homologous miRNAs: miR-143 and miR-145.22 Previous studies have demonstrated miR-143/145 regulate the proliferation, migration and invasion of glioma cells and could be potential therapeutic target for anti-invasion therapies of glioma patients.11,23 Recently, LncRNA CARMN inhibited the proliferation of glioblastoma cells by sponging miR-504.24 These suggested that CARMN could be of pathogenic importance in glioma.

Our study was the first to evaluate the correlation of CARMN variants with susceptibility and prognosis of glioma. We found CARMN rs13177623 was related to the decreased risk of glioma. Previous studies have indicated that the incidence rates of glioma tended to be associated with age and gender.25 Age stratified analysis showed rs13177623 had a lower risk of glioma at age ≥ 40 years, while rs11168100, rs12654195 and rs17796757 were associated with the susceptibility to glioma at age < 40 years. These indicate that the contribution of CARMN polymorphisms to glioma risk was associated with age exposures. In stratified analyses by gender, rs13177623 was significantly associated with decreased risk in males, but not in females, which suggesting the effect of rs13177623 polymorphism on glioma risk presented sex difference. Moreover, our study also evaluated the effect of CARMN polymorphisms on the prognosis of glioma patients. We found that rs13177623, rs12654195 and rs11168100 were associated with a better prognosis for glioma, especially in grade I–II glioma. In patients with grade III–IV glioma, rs17796757 polymorphism presented an improved OS. Previous studies supported that SNPs differentially might influence the expression and function of lncRNAs.26–28 Therefore, CARMN variants might contribute to the risk and prognosis of glioma by affecting the function of CARMN. However, further functional study is necessary to explore the role of these polymorphisms in the etiology of glioma.

Inevitably, our study had several limitations. Firstly, the inherent selection bias cannot be exclude because this study based on a hospital-based case–control study. Therefore, we recruited subjects matched by age, gender, and residential area to reduce the bias. Secondly, we did not assess the potential function of these polymorphisms in CARMN. Further functional experiments should be required to investigate the role of CARMN variants in glioma occurrence and development. Thirdly, some environmental factors such as occupational exposure and dietary were not available; the interaction of these factors with CARMN genotypes should be performed in a larger survey.

Conclusion

In summary, we firstly reported the contribution of CARMN variants (rs11168100, rs12654195, rs13177623, and rs17796757) to the susceptibility and prognosis of glioma in a Chinese Han population. Our study provides a novel insight on the relationship between CARMN gene and glioma tumorigenesis. These findings add to the growing body of evidence linking lncRNAs polymorphisms to glioma etiology. In addition, further studies are required to validate our results.