KDELR1 Is an Independent Prognostic Predictor and Correlates With Immunity in Glioma

Background Gliomas are the most malignant central nervous system tumors. With the development of sequencing technology, more potential biomarkers related to the treatment, prognosis, and molecular classification of glioma have been identified. Here, we intend to investigate the potential biological function and clinical value of a new biomarker in glioma. Methods KDELR1 expression data and the corresponding clinical information were downloaded from public databases and then preprocessed using R language. Correlation, Kaplan–Meier survival, and Cox regression analyses were performed to explore the clinical significance of KDELR1 in glioma patients. Furthermore, the immune infiltration and microenvironment parameters were evaluated via TIMER and CIBERSORT. Immunohistochemistry was conducted to confirm the KDELR1 expression and its correlation with immunity infiltration and prognosis. Results KDELR1 was upregulated in glioma samples compared with normal brain tissues, and its expression was significantly correlated with age, the World Health Organization (WHO) grade, recurrence, necrosis, microvascular proliferation, molecular classification, isocitrate dehydrogenase (IDH) mutation, and 1p/19q codeletion status. In addition, survival analysis showed that glioma patients with KDELR1 overexpression had shorter overall survival (OS) and disease-free survival times, and Cox regression analysis revealed that KDELR1 acted as an independent prognostic factor of OS in glioma patients. Gene set enrichment analysis indicated a significant enrichment of metabolism-associated pathways. KDELR1 expression was positively associated with immune infiltration (including infiltration by CD8+ T cells, CD4+ T cells, macrophages, and so on) and microenvironment parameters (including stromal, immune, and ESTIMATE scores) in gliomas. The expression of KDELR1 and its correlation with the tumor grade and prognosis were confirmed by immunohistochemistry in clinical samples (n = 119, P < 0.05). Conclusions Taken together, these findings suggest that KDELR1 is correlated with the tumor grade, molecular classifications, and immune infiltration; highlighting that KDELR1 is a novel and promising biomarker for molecular classification, treatment, and prognostic assessment may further indicate the treating effect of immune therapy.

Background: Gliomas are the most malignant central nervous system tumors. With the development of sequencing technology, more potential biomarkers related to the treatment, prognosis, and molecular classification of glioma have been identified. Here, we intend to investigate the potential biological function and clinical value of a new biomarker in glioma.
Methods: KDELR1 expression data and the corresponding clinical information were downloaded from public databases and then preprocessed using R language. Correlation, Kaplan-Meier survival, and Cox regression analyses were performed to explore the clinical significance of KDELR1 in glioma patients. Furthermore, the immune infiltration and microenvironment parameters were evaluated via TIMER and CIBERSORT. Immunohistochemistry was conducted to confirm the KDELR1 expression and its correlation with immunity infiltration and prognosis.
Results: KDELR1 was upregulated in glioma samples compared with normal brain tissues, and its expression was significantly correlated with age, the World Health Organization (WHO) grade, recurrence, necrosis, microvascular proliferation, molecular classification, isocitrate dehydrogenase (IDH) mutation, and 1p/19q codeletion status. In addition, survival analysis showed that glioma patients with KDELR1 overexpression had shorter overall survival (OS) and disease-free survival times, and Cox regression analysis revealed that KDELR1 acted as an independent prognostic factor of OS in glioma patients. Gene set enrichment analysis indicated a significant enrichment of metabolism-associated pathways. KDELR1 expression was positively associated with immune infiltration (including infiltration by CD8+ T cells, CD4+ T cells, macrophages, and so on) and microenvironment parameters (including stromal, immune, and ESTIMATE scores) in gliomas. The expression of KDELR1 and its correlation with the tumor grade and prognosis were confirmed by immunohistochemistry in clinical samples (n = 119, P < 0.05).

INTRODUCTION
Gliomas are the most common primary malignant tumors of the central nervous system (CNS), accounting for approximately 40%~60% of intracranial tumors, and have high morbidity and mortality (1). Low-grade gliomas (LGGs) and high-grade gliomas (HGGs) describe grade I/II and grade III/IV gliomas, respectively. Grade IV glioma, or glioblastoma multiforme (GBM), has a poor prognosis, with a median survival time of only 14-16 months from the first diagnosis (2,3). Currently, new strategies like precise surgery, adjuvant radiotherapy, and chemotherapy (4,5) have been conducted, while the prognosis of glioma has not improved significantly to date, and the recurrence rate is still high (6).
Compared with the World Health Organization (WHO) classification of CNS tumors published in 2007 and 2016, the main change in the 2021 classification is laying more emphasis on genetic parameters into the strategy of glioma diagnosis, breaking with the century-old standard of diagnosis based entirely on microscopy (3,7) . The Cancer Genome Atlas (TCGA) team has subclassified GBMs into four subtypes: proneural, mesenchymal, neural, and classical (8). In addition, Phillips et al. divided HGGs into proneural, mesenchymal, and proliferative subtypes (9). Notably, patients with mesenchymal subtype glioma usually have a worse prognosis than patients with the proneural subtype (10). In addition, mutations in the telomerase reverse transcriptase (TERT) promoter and IDH were found to be beneficial for the classification and risk stratification of grade II/III glioma patients (11,12).
Given the limited effectiveness of traditional treatments, a series of new treating strategies have been tried in gliomas with the help of advancing genomics. In recent years, high hopes have been placed on tumor immunotherapy, which has not only gradually become a research hotspot in the scientific studies (13) but also achieved amazing curative effects in the clinical treatment of some patients. Unfortunately, only limited patients with gliomas prolong the survival time after immune treatments (14). At present, relevant basic studies have revealed that the expression (15) or combinations (16)(17)(18) of immune-related indicators can be used to predict the prognosis of GBM patients, which suggests that finding a molecular target that can effectively predict the immune status of patients is urgent for guiding treatment (19).
KDEL Endoplasmic Reticulum Protein Retention Receptor 1 (KDELR1), a 24.5-kDa seven-transmembrane protein encoded by ERD2, which is located on 19q13.33, is responsible for the retrieval of soluble endoplasmic reticulum (ER) residents from the Golgi back to the ER (20). Some studies reported that KDELR1 mainly regulates the retention of soluble ER residents and the transporting processes in the secretory pathway (21,22). Some recent studies showed that KDELR1 might participate in maintaining cellular homeostasis (23) along with a potential role that involves the regulation of integrated stress responses (ISRs) in T cells (24), which is a possible way of regulating the immune homeostasis. However, the roles of KDELR1 in the biological functions and molecular mechanisms of gliomas are not clear.
In this study, genetic and clinical data related to KDELR1 were downloaded from public databases, including TCGA, Gene Expression Omnibus (GEO), and Chinese Glioma Genome Atlas (CGGA). After data preprocessing was completed, correlation, survival, and Cox regression analyses were performed to explore the potential function of KDELR1. We aimed to develop a novel diagnostic biomarker for glioma, thus assisting disease stratification and precise treatment.
To explore the expression levels of KDELR1 in brain or CNS cancer (especially gliomas) and normal samples, the KDELR1 gene was submitted to Oncomine (https://www.oncomine.org) with the criteria of P < 0.01, fold-change (FC) > 1.5 and gene rank = all (24,25). In addition, Gene Expression Profiling Interactive Analysis 2 (GEPIA2; http://gepia.cancer-pku.cn/ index.html), based on gene expression and clinical data from TCGA and GTEx, was used for the survival analyses of OS and disease-free survival (DFS) (26). Gene set enrichment analysis (GSEA) was used to analyze enrichment in high-risk and lowrisk groups defined by KDELR1 expression levels (27).

Analysis of Immune Infiltration and the Microenvironment
The R package "ESTIMATE" was used to analyze the communities of immune and stromal cells according to the characteristics of gene expression and then to obtain immune, stromal, and ESTIMATE scores (28). The significance of immune cells in the prognosis of glioma patients was explored in the Tumor Immune Estimation Resource (TIMER) site (29). TIMER was also used to provide an analysis of the clinical correlation between immune cell infiltration and patient survival.

Clinical Specimen Collection and Immunochemistry Staining
Samples from 119 patients (5 patients with WHO grade I, 33 patients with WHO grade II, 28 patients with WHO grade III, and 53 patients with WHO grade IV glioma) were collected from Shanghai OutDo Biotech. Co., Ltd., Shanghai, China. Informed consent was obtained from all patients, and the experimental protocols were approved by the Ethics Committee of Shanghai OutDo Biotech Co., Ltd.

Statistical Analysis
GraphPad Prism 7.0, SPSS Statistics 20, and R language were used for statistical analysis. The box plots of the expression level of KDELR1 across different groups were generated and calculated by GraphPad Prism 7.0. Kaplan-Meier survival analysis and Cox regression analysis were performed and generated via the R language. Immunoreactivity scores were analyzed and generated via SPSS Statistics 20. Differences with P<0.05 were considered statistically significant.

KDELR1 Expression Is Significantly Associated With Clinical Features and the Molecular Subtypes of Gliomas
We initially evaluated KDELR1 transcription levels in different human tumors by analyzing TCGA RNA-seq data using the TIMER database ( Figure 1). KDELR1 mRNA expression was markedly higher in GBM tissue than in normal brain tissue. In addition, KDELR1 was found to be highly expressed in bladder urothelial carcinoma (BLCA), breast-invasive carcinoma (BRCA), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), glioblastoma (GBM), head and neck squamous cell carcinoma (HNSC), kidney renal clear cell carcinoma (KIRC), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), pheochromocytoma and paraganglioma (PCPG), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), stomach adenocarcinoma (STAD), thyroid carcinoma (THCA), and uterine corpus endometrial carcinoma (UCEC) tissues and was significantly lower in head and neck squamous cell carcinoma that were positive for human papillomaviruses (HNSC-HPV+) than in the respective control tissues. These results demonstrated that KDELR1 was abnormally expressed in multiple tumors. Overall, these results indicated that KDELR1 expression was higher in GBM tissues than in normal counterparts.
The Oncomine search yielded seven analyses indicating KDELR1 upregulation and only one analysis indicating KDELR1 downregulation between brain or CNS cancer and normal samples with the criteria of P < 0.01, FC > 1.5, and gene rank = all ( Figure 2). These results indicate that KDELR1 might be upregulated in CNS cancers, such as gliomas, compared with the corresponding normal samples.

Overexpression of KDELR1 Is Positively Associated With Older Age, Recurrence, Necrosis, and Microvascular Proliferation in Gliomas
To further clarify the relationship between KDELR1 expression and the clinical features of glioma patients, the glioma samples were classified into two or more groups according to each clinical feature in each dataset. Several datasets showed that KDELR1 expression was higher in glioma patients aged ≥45 years than in those aged <45 years, including mRNA-array_301 and mRNAseq_325 of CGGA, TGGA_glioma, GSE4271, and GSE13041 (GPL96) (P < 0.05; Figures 3A-E). In addition, recurrent glioma samples showed a higher expression level of KDELR1 than primary samples in the CGGA mRNA-array_693 dataset (P < 0.05; Figure 3F). Moreover, microvascular proliferation and necrosis are the diagnostic criteria of GBMs, and the analysis of the GSE4271 dataset showed that

KDELR1 Expression Is Positively Related to the WHO Grade and Pathological Classification of Gliomas
Using the information from databases including GEO, CGGA, and TCGA, we wanted to determine whether the expression level of KDELR1 was related to the WHO grade and pathological classification of gliomas. The results showed that KDELR1 expression significantly increased with the WHO grade in several cohorts, including mRNA-array_301, mRNA-array_325, and mRNA-array_693 of CGGA, TCGA_glioma, GSE4271, GSE4290, and GSE4412 (P < 0.05; Figures 4A-G). Moreover, the expression level of KDELR1 was gradually upregulated along the sequence from control samples to oligodendrogliomas, astrocytomas, and GBMs in the GSE4290 dataset (P < 0.05; Figure 4H). Similarly, KDELR1 expression was upregulated from control samples to oligodendrogliomas, astrocytomas, and GBMs in the GSE68848 dataset (P < 0.05; Figure 4I). Taken together, these results indicate that KDELR1 expression is significantly related to the clinical features of gliomas and plays an important positive role in glioma progression.

KDELR1 Expression Is Strongly Correlated With Molecular Classification and Biomarkers in Gliomas
With new developments in sequencing technology, more biomarkers of important clinical value have been identified and applied to clinical practices such as molecular classification, which improves risk stratification and treatment accuracy. To date, several similar molecular classifications of gliomas have been proposed by different research teams; the relationships between their corresponding subtypes and KDELR1 expression were analyzed in this study. The results from both CGGA mRNA-array_301 and TCGA_glioma showed that KDELR1 was frequently highly expressed in the classical and mesenchymal subtypes and weakly expressed in the proneural and neural subtypes of GBM (P < 0.05; Figures 4J, K). Similarly, mesenchymal-subtype gliomas had a significantly higher expression level of KDELR1 than proneural-subtype gliomas (P < 0.05; Figures 4L-O). Further, KDELR1 expression was found to be strongly correlated with molecular biomarkers such as IDH mutation and the 1p/19q codeletion status in the three datasets from CGGA (P < 0.05; Figure 5). The results showed that glioma patients with IDH mutation had lower KDELR1 expression than gliomas with IDH wildtype (P < 0.05; Figures 5A-C). EstimateScore, ImmuneScore, StromalScore, and TumorPurity were analyzed between the two groups in CGGA301 and CGGA325 through the ESTIMATE algorithm ( Figures 6A-H). ImmuneScore in IDH mutation samples was higher than in IDHwildtype samples (Figures 6B, F). In summary, KDELR1 is strongly related to certain molecular biomarkers and the glioma classification, which indicates that KDELR1 could play a critical role in the development and molecular classification of gliomas.

KDELR1 Acts as a Poor Prognostic Factor in Glioma Patients
The above results showed that KDELR1 overexpression is positively associated with unfavorable clinical features, such as higher WHO grades, the mesenchymal subtype, recurrence, older age, and GBM, which indicates that KDELR1 might be an unfavorable prognostic factor in gliomas. When KDELR1 was submitted to the online tool GEPIA2, survival analyses indicated that glioma samples with KDELR1 overexpression had shorter OS and PFS (progression-free survival) times than those with low KDELR1 expression (P < 0.05; Figures 7A, B). Similarly, survival analyses revealed that the high-KDELR1-expression group had a shorter OS time than the low-KDELR1-expression group in six different datasets, including mRNA-array_301, mRNA-array_325, and mRNA-array_693 of CGGA, GSE4271, and GSE68848 (P < 0.05; Figures 7C-G).

Cox Regression Analysis of KDELR1 as an Independent Predictor of Survival in Glioma Patients
As KDELR1 was associated with the prognosis of glioma patients, we wanted to determine whether it was an independent predictor of OS in gliomas using univariate and multivariate survival analyses. In the CGGA mRNA-array_301 dataset, univariate analysis showed that KDELR1 expression as well as age, the WHO grade, primary/ recurrent/secondary type, histology, TCGA subtype, radiotherapy, chemotherapy, IDH mutation, and 1p/19q codeletion status were significantly associated with OS, and further multivariate analysis showed that KDELR1 expression and the chemotherapy status were independent prognostic predicators of OS in gliomas ( Table 1). Moreover, similar results in the CCGA mRNA-array_325 and TCGA_glioma datasets showed that KDELR1 expression acts as an independent prognostic predicator in gliomas ( Tables 2, 3). Collectively, Cox regression analyses showed that KDELR1 is an independent prognostic predictor in glioma.

KDELR1 Is Correlated With Immune Infiltration and the Microenvironment In Glioma
To explore KDELR1 and immune infiltration in LGG and GBM, KDELR1 was analyzed using the TIMER database. KDELR1 was significantly correlated with dendritic cells in GBM and B cells, CD8+ T cells, CD4+ T cells, macrophages, neutrophils, and dendritic cells in LGG ( Figure 8A). Moreover, the results showed that B cell, CD8+ T cell, CD4+ T cell, macrophage, neutrophil, and dendritic cell infiltration significantly affected Violin plot showed the four differential scores (EstimateScore, ImmuneScore, StromalScore, TumorPurity) between IDH wildtype and mutation glioma samples by using the ESTIMATE algorithm in CGGA301 and CGGA325 datasets, respectively.    the prognosis (P<0.05) and was correlated with KDELR1 expression in LGG patients but not in GBM patients ( Figure 8B). We analyzed the proportions of 22 immune cells in the two groups by the CIBERSORT algorithm, which revealed that there were significant differences in the proportions of CD8+ T cells, activated NK cells, monocytes, M1 macrophages, M2 macrophages, and neutrophils between the groups with highand low-KDELR1-expression levels ( Figure 9A). The ESTIMATE algorithm was performed to assess the immune levels of glioma patients and showed significant differences (p<0.001) in the immune score, stromal score, and ESTIMATE score between the patients with high and low KDELR1 expression. Specifically, the immune score, stromal score, and ESTIMATE score were all higher in the patients with high KDELR1 expression ( Figures 9B-D). A similar analysis was conducted in the IDH mut and IDH wide subgroups; the abundances of several immune cells (including CD4 naive T cells, gamma delta T cells, monocytes, M0 macrophages, M1 macrophages, and neutrophils) were different among the two groups ( Figures 9E, F).

Functional Enrichment Analysis of KDELR1
To explore the potential function of KDELR1, a total of 100 genes co-expressed with KDELR1 were identified using the GEPIA2 database, and these genes were uploaded to DAVID online. Further Gene ontology (GO) enrichment analysis revealed that these 100 co-expressed genes may be associated with chaperonemediated protein folding, cell redox homeostasis, the regulation of mitochondrial membrane potential, Wnt signaling, planar cell polarity, extracellular exosome, focal adhesion, endoplasmic reticulum membrane, protein binding, glycoprotein binding, and protein disulfide isomerase activity ( Figures 10A-C).
Interestingly, Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis further suggested that KDELR1 may be involved in protein processing in the endoplasmic reticulum, Nglycan biosynthesis pathway, and Epstein-Barr virus infection ( Figure 10D). In addition, GSEA indicated that the gene sets specific to the high-KDELR1-expression group were mainly enriched in metabolism-associated pathways ( Figure 10E).

Verification of KDELR1 Expression in Glioma Tissues by Immunohistochemistry (IHC)
To verify the results of the bioinformatics analysis, IHC staining was performed on tissue microarray slides containing the samples of 119 gliomas ( Figure 10. The expression level of KDELR1 was significantly higher in high-grade glioma than in low-grade glioma tissues (p<0.001), which was consistent with the results of bioinformatics analysis at the RNA level. The relevance between the expression of KDELR1 and immune infiltration was confirmed by the IHC of CD4 and CD8; a positive trend of immune infiltration was consistent with data obtained from the algorithm ( Figure 11).

DISCUSSION
Gliomas are the most common primary CNS tumor, accounting for more than 80% of primary brain tumors (30). Among them, GBM is prone to recurrence and has a median survival time of less than 2 years (4). Using bioinformatics analyses, in this study, we found that KDELR1 expression levels were higher in glioma samples than in the corresponding normal tissues. In addition, further evaluation confirmed that KDELR1 was closely related to the clinical features of glioma. These findings suggest that KDELR1 may be a promising biomarker for the precise diagnosis, molecular characteristics, treatment, and prognostic evaluation of gliomas. It has been established that several clinical features, including age, recurrence, and pathology, are responsible for clinical prognosis in glioma patients (31)(32)(33). A previous study also suggested that older glioma patients would have a poorer prognosis than younger patients, which indicates that KDELR1 expression levels might possess an indirect clinical value in prognosis (34). Consistently, our findings also reveal that older patients had higher KDELR1 expression than younger patients. In pathologic analysis, we found that HGGs had higher KDELR1 expression than LGGs. In addition, previous studies (35)(36)(37) have demonstrated that mesenchymal-subtype gliomas are associated with poor prognosis, which are confirmed by our finding that mesenchymal -subtype gliomas had a higher expression level of KDELR1 than proneural subtype gliomas. Subsequently, we further explored whether the expression of KDELR1 is related to recurrence. As expected, KDELR1 was more frequently highly expressed in the recurrent group compared with the initial diagnosis group. Taken together, these findings indicate that KDELR1 might act as a novel promising biomarker for the diagnosis, treatment, and prognosis of gliomas. Recently, with the rapid advances in next-generation technology, research on the biomolecular markers and associated signaling pathways that are involved in the occurrence and development of gliomas has made substantial progress (38). After the IDH mutation status was confirmed to be related to the prognosis of patients with GBMs, a subsequent study reported that the chromosome 1p/19q codeletion status, O6-methylguanine-DNA methyltransferase (MGMT) promoter region genotype, a-thalassemiamental retardation syndrome X (ATRX), and amplification of the epidermal growth factor receptor (EGFR) played more important roles in the prognosis and treatment prediction of gliomas (39,40). Our study demonstrates that KDELR1 is downregulated in the 1p/19q codeletion group compared with the 1p/19q non-codeletion group. Thus, KDELR1 might be a negative prognostic factor in gliomas from the 1p/19q codeletion perspective. On the other hand, our results indicated that the IDH mutation group had a lower expression level of KDELR1 than the IDH-wildtype group.
In 2010, a study classified GBMs into proneural, neuronal, classical, and mesenchymal subtypes according to the status of Platelet-derived growth factor alpha receptor (PDGFRA), isocitrate dehydrogenase 1 (IDH1), Epidermal growth factor receptor (EGFR), and Neurofibromin type 1 gene (NF1) (8). Additionally, in a study of 107 HGG samples, Phillips et al. divided the samples into three subtypes: proneural, mesenchymal, and proliferation. Proliferation-and mesenchymal-subtype tumors tend to express high levels of genes related to cell proliferation and angiogenesis (9), respectively, and these types often occur in elder (more than 50 years old) patients who have poor prognosis (41). Notably, our research found that KDELR1 is highly expressed in the mesenchymal subtype and expressed at low levels in proneuralsubtype gliomas. Based on this evidence, we further inferred, from the molecular classification, that KDELR1 was strongly associated with an unfavorable clinical outcome.
Our previous results have indicated that the increased expression of KDELR1 is strongly related to negative survival factors such as older age, a higher WHO grade, recurrence, IDH wild type, and 1p/19q non-codeletion status; therefore, we further investigated whether the expression of KDELR1 is related to the survival time of patients with gliomas. Survival analyses showed that gliomas with KDELR1 overexpression were associated with shorter OS and PFS times than gliomas with low KDELR1 expression. This finding may provide proof that KDELR1 can be used in predicting clinical prognosis.
We further focused on the relation between KDELR1 and immune infiltration; CD8+ T cells, CD4+ T cells, and macrophages were found with high expression in HGGs along with the overexpression of KDELR1. The IHC expression of CD4 and CD8 in four different grades is consistent with the bioinformation results, and CD8 showed more participants in the tumoral immune microenvironment.
In summary, our findings show that KDELR1 is upregulated in gliomas compared with normal brain tissues and that its expression is significantly associated with clinical features such as the WHO grade, recurrence, molecular classification, IDH mutation, and 1p/19q codeletion status. Moreover, the survival and Cox regression analyses of different datasets suggested that KDELR1 expression in gliomas could be an independent, unfavorable prognostic factor for survival time. On the other hand, KDELR1 expression was associated with immune infiltration (including the infiltration of CD8+ T cells, CD4+ T cells, macrophages, and so on) and microenvironment parameters (including stromal, immune, and ESTIMATE scores) in gliomas. Collectively, these results indicate that KDELR1 could be a promising novel biomarker for molecular classification, immune treatment, and prognostic assessment in glioma.
However, there were some limitations in this study. First, the research on KDELR1 in glioma is still in the early stage, and our research is limited to the bioinformatics database analysis and experimental verification of IHC. Thus, further studies based on surgical samples will be imperative to perform in vitro or in vivo assays, validating these findings, and as a biomarker, diagnostic tests will be conducted in future research. Second, bioinformation data collected from the online database were established and categorized based on the 2016 WHO CNS classification system. Since the 2021 WHO classification system has been published, a prospective study is underway to collect samples and improve this work.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Ethics Committee of Shanghai OutDo Biotech Co., LTD. and the Ethics Committee of Huashan Hospital. The patients/participants provided their written informed consent to participate in this study.