Dysregulated Expression and Methylation Analysis Identified TLX1NB as a Novel Recurrence Marker in Low-Grade Gliomas

Low-grade gliomas (LGGs) are the most common CNS tumors, and the main therapy for LGGs is complete surgical resection, due to its curative effect. However, LGG recurrence occurs frequently. Biomarkers play a crucial role in evaluating the recurrence and prognosis of LGGs. Numerous studies have focused on LGG prognosis. However, the multiomics research investigating the roles played by gene methylation and expression in LGG recurrence remains limited. In this study, we integrated the TCGA and GEO datasets, analyzing RNA and methylation data for recurrence (R) and nonrecurrence (NR) groups. We found a low expression of TLX1NB and high methylation in recurrence patients. Low expression of TLX1NB is associated with poor survival (OS: p = 0.04). The expression of TLX1NB is likely to play a role in the prognosis of LGG. Therefore, TLX1NB may represent an alternative early biomarker for the recurrence of low-grade gliomas.


Introduction
Low-grade glioma (LGG) is an uncommon type of the primary central nervous system tumor classified by the WHO as Class I and II [1,2]. Usually, LGG is inactive, and the main therapy strategy is complete surgical resection because this treatment can be curative. However, even if the tumor is resected, the tumor cells resistant to irradiation and chemotherapy may grow gradually; thus, the neoplasm can still relapse at some time point during the clinical course [3][4][5][6]. Due to recurrence and metastasis, the prognosis of LGG remains controversial [7,8]. Because these tumors have a long asymptomatic natural history, whether patients with limited lesions and few symptoms are given active or delayed treatment and the timing of postoperative radiotherapy and chemotherapy have not been determined [9]. Thus, it is of great importance to understand the underlying molecular mechanisms govern-ing LGG recurrence and to identify novel recurrenceassociated biomarkers [5,10].
Comprehensive multiomics provides a deeper and more comprehensive understanding of specific genes and enables us to identify functional genes for biomarker mining [11]. The Cancer Genome Atlas (TCGA) offers multidimensional maps of the key genomic changes in cancer, including alterations in DNA, RNA, copy number, and methylation [12].
The objective of the present study was to explore possible clinical biomarkers of recurrence in LGG through multifaceted analysis of recurrence patterns in intracranial LGG.

2.
1. Data Sources. RNA read count data, expression matrix data, copy number variation data, and DNA methylation data (Illumina Human Methylation 450 k Array) of LGG were obtained from The Cancer Genome Atlas (TCGA, https:// cancergenome.nih.gov). Clinical information on LGG, including survival and recurrence data, was downloaded from the GDC Data Portal (https://portal.gdc.cancer.gov/) [12]. The gene expression matrix of GSE35158 was downloaded from the Gene Expression Omnibus (GEO) database.

Preprocessing and Analysis of RNA Data.
We defined patients with recurrence as the case group and patients without recurrence as the control group. The RNA read count data were preprocessed via DESeq2 1.26.0 (R package) [18]; log2 fold change > 1 and adjusted p value < 0.05 were selected as the threshold. All differentially expressed genes were obtained. Bar plots of different genes were generated by ggplot2 3.3.0 (R package) [19]. Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis, and data visualization were performed by cluster-Profiler 3.14.3 (R package) [20].

Preprocessing and Analysis of DNA Methylation Data.
The DNA methylation data were preprocessed and normal-ized using minfi 1.32.0 (R package) [21]. Differentially methylated regions with a change value >0.1 or <-0.1 and a p value < 0.05 were selected as the methylation change threshold. All different methylated regions were obtained. Annotation was performed via the online tool wANNOVAR [22] (http:// wannovar.wglab.org).

Overlap Statistics between RNA and Methylation and
Copy Number Variation (CNV) Analysis. Upregulated RNA and hypomethylated genes or downregulated RNA and hypermethylated genes were extracted. Copy number variation analysis of target genes was performed by ggpubr 0.2.5 (R package) (https://github.com/kassambara/ggpubr).

Survival
Analysis and Statistics. Survival plots of selected genes were generated by survival 3.1-12 (R package) and survminer 0.4.6 (R package) [23], and a p value < 0.05 was selected as the significance level.    2 International Journal of Genomics 2.6. GSEA Analysis. We calculated the correlation score between the target gene and all other genes and then ranked the correlation score from high to low. Finally, we obtained the gene list. We selected oncogenic gene sets (C6) as the input dataset, performed GSEA analysis by clusterProfiler 3.14.3 (R package) [20], and visualized pathways of interest by enrichplot 1.6.1 (R package) [20].

Differentially Expressed Genes between Recurrence and
Nonrecurrence Samples. We selected the samples with recurrence information and RNA sequencing data and eventually obtained 65 recurrence and 188 nonrecurrence data for fur-ther RNA expression analysis. We set log2 fold change > 1 and adjusted p value < 0.05 as a significant change threshold, and finally, we obtained 329 upregulated and 125 downregulated genes at the RNA level (Figure 1(a) and Table 1). Through GO and KEGG functional analyses, the differentially expressed genes were determined to be primarily involved in neuroactive ligand−receptor interaction pathways (Figure 1(b)) and DNA-binding transcription activator activity (Figure 1(c)).

Identification of Methylation Regions between Recurrence
and Nonrecurrence Samples. A total of 485,577 loci of DNA methylation were obtained from the TGCA database. After NA values were removed, 467,971 loci were retained for  3 International Journal of Genomics further analysis. Fifty-nine candidate DMRs were obtained from the analysis results. After annotation, removing unannotated DMRs, four hypermethylated regions were obtained for further cross-overlap analysis ( Table 2).

Integrated Analysis for RNA Expression, DNA
Methylation, and Copy Number Variation. Genes were selected following the criterions: high methylation at DNA level and low expression at RNA level or low methylation at DNA level and high expression at RNA level. TLX1NB was identified by this approach (Figure 2(a)). The copy number variation of TLX1NB was investigated, and no difference between nonrecurrence and recurrence was found (Figure 2(b), p > 0:05). These results indicated that the downregulation of TLX1NB may not be associated with copy number alterations. To evaluate the methylation effect on TLX1NB, we compared methylation and RNA expression levels between recurrence and nonrecurrence groups. We found methylation of TLX1NB upregulated at 24% in the recurrence group compared with the control (Figure 2(c)), meanwhile at RNA expression level, the expression of TLX1NB downregulated at 80% (Figure 2(d)).

Gene Set Enrichment Analysis of TLX1NB.
To further investigate the potential functions of TLX1NB, GSEA was performed based on the LGG expression data. TLX1NB was negatively correlated with typical tumor driver genes, such as P53, EGFR, TBK1, and STK33 (Figure 3). These results implied that TLX1NB is involved in typical cancer pathways and tumor proliferation processes.
To understand the expression of TLX1NB in LGG patients in different subtypes, we employed dataset GSE35158 to analyze 80 patients' expression profiles. We observed that there was no difference in the TLX1NB expression between grade II and grade III patients (Figure 4(a), p = 0:25, Wilcoxon's test); however, there was a significantly  International Journal of Genomics increased TLX1NB expression level in the neuroblastic subtype, which was associated with Hu immunopositivity and a mature neuronal gene set in GSE35158 (Figure 4(b), p = 0:0041, Kruskal-Wallis test). We also found decreased expression of TLX1NB in the PTEN deletion group (Figure 4(c), p = 0:038, Kruskal-Wallis test). To explore the relationship between TLX1NB, PTEN promoter methylation, and IDH deletion, we analyzed TLX1NB expression and PTEN promoter in GSE35158 and TLX1NB expression and IDH mutation in TCGA and GSE35158. No significant change of TLX1NB expression was observed between PTEN methylation group and nonmethylation group in GSE35158 ( Figure 4(d)). TLX1NB was significantly upregulated in the IDH mutation group in the TCGA dataset (Figure 4(e)); however, no difference was observed in the GSE35158 dataset (Figure 4(f)). The result for IDH mutation was controversial. So, we resume that TLX1NB has the potential to be a new biomarker in LGG associated with recurrence and prognosis.

TLX1NB Predicts Survival Level in
LGG. From the previous results, we could assume that low expression of TLX1NB is associated with adverse consequences of LGG. Thus, we analyzed the prognosis of LGG patients in TCGA. Five hundred twenty-eight patients were divided into two groups: the

Discussion
According to CBTRUS (Central Brain Tumor Registry of the United States), glioma is the most common CNS tumor, accounting for~27% of CNS tumors [1] and leading to severe disability and mortality. According to the WHO central nervous tumor histological grading standards, such tumors such as grade I to II astrocytomas, papillary glioneuronal tumors, and vascular central gliomas are collectively referred to as lowgrade gliomas [2].
LGGs account for 15% to 30% of all gliomas.
LGGs are more common in children and young people with an average age of onset of 30 to 45 [4]. Although surgical resection is the main method of treatment, the tumor cannot be completely removed in the true sense due to the invasive growth of gliomas, the principle of tumor location, and the maximum protection of nerve function, which affects the survival of the patient [5]. Therefore, research that affects the risk of glioma recurrence may provide evidence to guide clinical treatment and facilitate the development of personalized strategies.
Regarding recurrence, previous studies revealed that for instance, high-level methylation of the MGMT promoter leads to hypermutation at recurrence, 1p/19q codeletion or IDH mutation is associated with longer overall survival and better treatment response, and TP53 mutation is associated with a worse prognosis [5,10,14,16,24,25]. However, the number of recurrence studies focusing on gene expression remains notably limited. In this study, we comprehensively analyzed TCGA and GEO data and observed that TLX1NB may be a potential recurrence biomarker. Previous studies have demonstrated that TLX1NB can be a prognostic lncRNA biomarker in lung adenocarcinoma [26]; however, there are no other studies revealing the role of TLX1NB in CNS cancer.
DNA copy number variations (CNVs) are an important component of genetic variation, affecting a greater fraction of the genome than single-nucleotide polymorphisms (SNPs) [27]. Therefore, we investigated the CNVs of TLX1NB and found no difference between the recurrence and nonrecurrence groups.
TLX1NB is adjacent to the TLX1 gene, and we hypothesize that its role could be associated with TLX1 function as a cis-regulatory factor [28]. Therefore, we correlated TLX1NB and TLX1 and found that TLX1NB was positively coexpressed with TLX1NB (Supplementary Figure 1). TLX1, a member of the NK-linked or NK-like (NKL) subfamily, is involved in the specification of neuronal cell fates. Thus, we hypothesize that TLX1NB and TLX1 together drive function. Our gene set enrichment analysis showed that TLX1NB was associated with typical cancer driver genes, such as P53, EGFR, STK33, and TBK1. These results   International Journal of Genomics demonstrate the role played by TLX1NB in cancer pathways. We also found that TLX1NB was negatively correlated with the transcription activating gene set (Supplementary Figure 2a) and the IL2 gene set (Supplementary Figure 2b).

Conclusion
We investigated the relationship between TLX1NB in LGG.
TLX1NB is a predictor of LGG survival; reduced expression of TLX1NB worsens the prognosis and survival of LGG patients. TLX1NB probably affects LGG through the tumor activating pathway and could be a meaningful biomarker for LGG.

Data Availability
The raw data can download following the link in our manuscript, and analyzed data has been showed in table 1 and table  2. Regarding the intermediate process data we are willing to provide anytime when any researcher raise their request.