High APLN Expression Predicts Poor Prognosis for Glioma Patients

Apelin (APLN) is an endogenous ligand of the G protein-coupled receptor APJ (APLNR). APLN/APLNR system was involved in a variety of pathological and physiological functions, such as tumorigenesis and development. However, its prognostic roles in patients with central nervous system (CNS) cancers remain unknown. The present study was designed to explore the expression profile, prognostic significance, and interaction network of APLN/APLNR by integrating data from Oncomine, GEPIA, LOGpc, STRING, GeneMANIA, and immunohistochemical staining. The results demonstrated that APLN and APLNR mRNA expression were significantly increased in CNS cancers, including both low-grade glioma (LGG) and glioblastoma (GBM), when compared with normal CNS tissues. The high APLN, but not APLNR, expression was significantly correlated with overall survival (OS), recurrence free survival (RFS), and progression free survival (PFS) of LGG patients. However, neither APLN nor APLNR expression was significantly related to prognostic value in terms of OS, disease free interval (DFI), disease specific survival (DSS), or progression free interval (PFI) for GBM patients. Additionally, immunohistochemistry staining confirmed the increased APLN expression in tissues of LGG patients with grade II than grade I. These results showed that an elevated APLN level could predict poor OS, RFS, and PFS for LGG patients, and it could be a promising prognostic biomarker for LGG.

Recently, APLN/APLNR was shown to be involved in several kinds of cancers, such as lung cancer, gastroesophageal cancer, colonic cancer, and hepatocellular carcinoma [13]. In non-small-cell lung cancer (NSCLC), APLN mRNA was significantly upregulated in tumor tissue compared with normal lung, and high level of APLN protein was reported to be associated with the increased microvessel densities and worse overall survival [14]. In patients with hepatocellular carcinoma, APLN mRNA was markedly higher in tumors than in nontumor tissues [15]. APLN/APLNR mRNA and protein expression in ovarian epithelial cancer and/or granulosa tumor cell lines were significantly higher than those of noncancer ovarian cell lines [16]. In patients with cholangiocarcinoma, APLN and APLNR genes were obviously upregulated in tumor tissues compare to nonmalignant liver tissues [17]. Kälin et al. found that APLN/APLNR mRNAs were upregulated within microvascular proliferations in glioblastoma specimens, compared with normal CNS tissues [18].
Glioma is the most common primary malignant brain tumor in adults, with 5-year survival of 20%-30% [19]. Glioma is subclassified and graded from I to IV according to histological criteria described by the World Health Organization (WHO) [20]. Low-grade glioma (LGG, grade I-II) is considered as infiltrative neoplasm that most frequently occurs in the cerebral hemispheres of adults [20]. Glioblastoma (GBM, grade IV) represents the most aggressive subtype of gliomas. Traditional histopathological methods do not adequately predict clinical outcomes of gliomas [21]. Therefore, the prognostic biomarkers are needed to predict and improve the clinical decision-making process. The prognosis of APLN and APLNR in gliomas remains unknown. The current study was designed to examine the prognostic values, gene expression profile, and interaction network of APLN/APLNR in LGG and GBM.

Materials and Methods
2.1. Oncomine Analysis. Oncomine (https://www.Oncomine .org) is an online gene expression array database with publicly accessible services. The database, consisting of 715 datasets and 86733 normal and tumor samples, provides genome-wide expression analyses for researchers [22,23]. The normal samples were from normal human tissues. The thresholds (p value ≤0.05, fold-change ≥2, gene rank ≤10%, and data type: mRNA) were set, and the APLN and APLNR genes in gliomas and normal tissues were compared and analyzed using Oncomine.

GEPIA Analysis. Gene Expression Profiling Interactive
Analysis (GEPIA, (http://gepia.cancer-pku.cn) is a webbased tool used for gene expression analysis based on the TCGA and the GTEx databases. GEPIA, containing 9736 tumors and 8587 normal samples, provides key interactive and customizable functions, including differential expression analysis, profiling plotting, and correlation analysis [24]. The normal samples were provided by normal human tissues. We examined the expression of APLN/APLR in LGG and GBM cancers by GEPIA.
2.4. GeneMANIA and STRING Analysis. GeneMANIA (http://genemania.org) is a web tool for generating hypotheses about gene function, analyzing gene lists and prioritizing genes for functional assays [33,34]. Analysis of APLN correlated genes was performed using GeneMANIA. Furthermore, we examined the protein interactions of APLN using STRING. The STRING database (http://string-db.org) aims to search for protein-protein interactions (PPI), including physical as well as functional associations. The latest version of STRING includes 5090 organisms, 24584628 proteins, and 3123056667 total interactions [35].
2.5. Immunohistochemical Staining and Evaluation. Tissue microarray, purchased from Wuhan Servicebio Technology Co., Ltd. (Wuhan, China), was routinely dewaxed, rehydrated, and immunohistochemistry staining was carried out following the standard procedure. The anti-apelin-36 (1 : 150 dilution, Phoenix Biotech Co., Ltd., Beijing, China) was used as primary antibody. The secondary antibody (HRP-labeled antirabbit antibody, 1 : 200 dilution, Servicebio Co., Ltd., Wuhan, China) was incubated with the sections, and the staining was visualized using diaminobenzidine (DAB) solution and counterstained with hematoxylin. PBS, instead of the primary antibody, was utilized as a negative control for excluding unspecific binding of secondary antibody. The protocol was approved by the Committee of Medical Ethics and Welfare, Henan University, School of Medicine.
2.6. Statistical Analysis. Kaplan-Meier curves of OS, DFI, DSS, PFI, and RFS were constructed by setting the quarter (upper 25% vs lower 75%) of APLN/APLNR expression as the cut-off, while Kaplan-Meier curves of PFS were constructed by setting the median (upper 50% vs lower 50%) as the cut-off. Univariate and multivariate Cox regression analyses were performed to evaluate the prognostic value   3 Oxidative Medicine and Cellular Longevity of APLN in terms of OS, RFS, and PFS for LGG patients using SPSS 18.0. Risk factors (p < 0:2) assessed by univariate analysis were utilized for the subsequent multivariate analysis. Kaplan-Meier curves of PFS, subgrouped by different IDH status and treatment therapies in GSE107875 dataset, were performed by GraphPad Prism 8.0, followingly a logrank test was used to evaluate the difference between different subgroups. For immunohistochemistry analysis, the statistical difference of APLN expression between two groups was evaluated by Student's t-test. p < 0:05 was considered statistically significant.
The LGG cases from GSE107850 dataset are subdivided according to treatment therapy and type of surgery. Elevated APLN expression displayed worse PFS in LGG patients with biopsy (p = 0:0429), but not in patients with partial resection (p = 0:4375) or total resection (p = 0:2321). The increased APLN expression is associated with inferior PFS in patients with temozolomide (p = 0:0396), but not in patients with radiation therapy (p = 0:063; Figure S4). For high APLN expression group (upper 50%), LGG patients with mutant IDH status had inferior PFS (p = 2:5E − 3) in comparison to patients with IDH wildtype status, but not for low APLN expression group (lower 50%, p = 0:7659; Figures S5(a, b)), indicating that combination of IDH mutation status and APLN could guide the prognostication more specifically for LGG patients. Selection of the most suitable treatment to each LGG patient is the goal of precision medicine. However, no matter whether APLN is high or low, there is no significant outcome difference between RT and TMZ treatment groups (Figures S5(c, d)).   10 Oxidative Medicine and Cellular Longevity

Protein Expression of APLN in an Independent
LGG Cohort. To further validate the expression of APLN in LGG, APLN protein expression was examined on a tissue chip through an immunohistochemical assay. Overall, the samples from 147 LGG patients were included in the study. APLN positive staining was observed mainly in cytoplasm of cancer cells, and weak staining was showed in the membrane (Figures 4(a)-4(c)). The staining scores of APLN were significantly higher in grade II LGG cancers than grade I (p < 0:05), as shown in Figure 4(d). In addition, staining scores of APLN in LGG cases with IDH1 mutation were significantly higher than in patients without IDH1 mutation (p < 0:05, Figure 4(e)). However, there were no significant differences of staining scores of APLN in LGG cases with different Ki67 expression (p = 0:150, Figure 4(f)), and with/without p53 mutation (p = 0:549, Figure 4(g)).

Discussion
LGG is infiltrative neoplasm which generally occurred in the cerebral hemispheres of adults, and encompassed astrocytomas, oligodendrogliomas, and oligoastrocytomas [20]. In the current study, the definition of glioma was according to the WHO clasasification [20]. A part of LGG may be speedily developed to grade IV glioma, becoming GBM [40]. The survival of LGG patients ranged from 1 to 15 years, while the median survival time of GBM patients ranged from 12-15 months [41,42]. Although different treatment approaches such as surgical resection, chemotherapy, and radiotherapy were applied, the survival rate for glioma patient was still low, and the discovery of novel biomarkers were required. APLN/APLNR was reported to be involved in the regulation of tumor growth, cancer cell migration, neoangiogenesis, and apoptosis in different types of cancers. In the present study, we demonstrated that APLN and APLNR expression were upregulated in both LGG and GBM clinical cases, compared with corresponding normal tissues. The result was in line with the previous reports that APLN/ APLNR gene expression were increased in gastric cancer, liver cancer, cholangiocarcinoma, lung cancer, ovarian cancer, prostate cancer, etc., compared with the equivalent normal tissues or cells [6]. Harford-Wright et al. reported that APLN/APLNR may act as a paracrine signal that sustains tumor cell expansion and progression in glioblastoma [43] indicating a crucial relation between APLN/APLNR and glioblastoma.
APLN/APLNR had been showed as prognostic markers in several types of cancers. High APLN level in tumor tissue predicted worse outcome for patients with gastric cancer and muscle-invasive bladder cancer [44,45]. The elevated serum APLNR was correlated with inferior OS in patients with clear-cell renal cell carcinoma [46]. However, the relationship between APLN/APLNR and prognosis of brain tumor is still unknown. In current study, the correlation of APLN expression and prognostic value was evaluated and the results indicated that high APLN, but not APLNR, was significantly correlated with poor OS, RFS, and PFS in LGG patients. However, neither high APLN nor high APLNR was significantly correlated to prognostic values of OS, DFI, DSS, or PFI in GBM patients. Additionally, immunohistochemical staining confirmed APLN expression in cytoplasm of glioma cells, which was consistent with the distribution of APLN in colon adenocarcinomas [47] and breast carcinoma [48]. The APLN expression was significantly higher in grade II LGG cases than grade I. These results suggested that high APLN expression could predict poor prognosis for LGG patients.
Further analysis was performed to explore whether the correlation between high APLN expression and poor OS/ RFS/PFS of LGG was influenced by clinical characteristics, including seizure history, sample type, targeted molecular therapy, histological type, chemo/radio-therapy, and type of surgery. Our results showed that the significant correlation between elevated APLN expression and inferior OS or RFS in LGG patients was restricted to patients with seizure history, primary tumor, targeted molecular therapy, oligo-dendroglioma, biopsy, and TMZ therapy. This result will facilitate clinicians to manage the personalized treatment for LGG.
In this study, GeneMANIA analysis showed that the top 20 correlated genes were screened. These genes were enriched in G protein subunits (including GNAZ, GNAT2, GNAT3) and chemokine receptors (CCR6, CCR8). The results indicated a correlationship between APLN and G protein, which was consistent with the previous report that APLN-activated APLNR couples to G protein, such as Gαi and Gαq [49,50]. Based on the PPI network, we obtained top 10 hub proteins, and they were mainly classified as pathways of angiotensin (ACE2, MAS1) and adipokines (LEP, SERPINA12, ADIPOQ). The angiotensin and LEP played a regulatory role in glioma cells [51,52].

Conclusions
In conclusion, both APLN and APLNR expression was significantly elevated in LGGs and GBMs compared with normal tissues, but only the high APLN expression was correlated with poor OS, RFS, and PFS in LGG patients, providing a clue that APLN might serve as a prognostic biomarker for LGG. Further studies are required to investigate the molecular mechanism of APLN in tumorigenesis and progression of LGG.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
The authors declare that they have no conflicts of interests. Figure S1. Kaplan-Meier curves of recurrence free survival (RFS) in LGG patients with seizure history (A), primary tumor (B), targeted molecular therapy (C), and oligodendroglioma (D) from TCGA data. Figure S2. Kaplan-Meier curves of overall survival (OS) in LGG patients without seizure history (A), with recurrent tumor (B), with astrocytoma (C), and with oligoastrocytoma (D) from TCGA data. Figure  S3. Kaplan-Meier curves of recurrence free survival (RFS) in LGG patients without seizure history (A), with recurrent tumor (B), without targeted molecular therapy (C), with astrocytoma (D), and with oligoastrocytoma (E) from TCGA data. Figure S4 Kaplan-Meier curves of PFS in LGG patients with different treatment therapies (RT vs. TMZ, B) in high APLN expression group (C) and low APLN expression group (D). The high APLN expression was defined by upper 50% expression, and low APLN expression group was defined by lower 50% expression in GSE107850 dataset. RT, radiation therapy; TMZ, temozolomide.