Prognostic Signicance of STC2 and DNA methylation in Head and Neck Squamous Cell Carcinoma: A Study Based on the TCGA and GEO Databases

Background: Head and neck squamous cell carcinoma (HNSC) is a popular malignancy type that brings about poor prognosis with a low survival rate worldwide. Stanniocalcin 2 (STC2) is a glycosylated peptide hormone and shows the potential to become a new biomarker for the evaluation of malignant tumors. The purpose of this study was to explore the prognostic implications of STC2 and DNA methylation in HNSC and the role of STC2 expression in immune cell inltration. Methods: STC2 gene expression data were collected from Gene Expression Omnibus (GEO) and the Cancer Genome Atlas (TCGA) databases. Univariate and multivariate analyses were employed to screen prognostic risk factors. The relationship between STC2 expression and TP53 mutation in HNSC was explored. TCGA data were utilized to analyze how STC2 expression affected immune cell inltration in HNSC. Results: STC2 was highly expressed in HNSC patients (P < 0.01), especially those with a lower overall survival rate (P < 0.0001). TP53 mutation might be a risk factor of STC2 overexpression in HNSC (P = 0.0015). STC2 expression was negatively correlated with STC2 methylation (Spearman: -0.43, P < 0.001). Hypermethylation or hypomethylation at the eight CpG sites most related to STC2 expression was identied as independent factors for HNSC prognosis. STC2 was positively correlated with cancer-associated broblasts inltration and associated with the inltration of various immune cells. Conclusion: STC2 can be regarded as a vital prognostic biomarker of HNSC due to its essential roles in immune cell inltration.

reveals aberrant STC2 expression in assorted tumor types, which can be considered a new biomarker for prognosis prediction. Recent studies also linked STC2 to a poor prognosis of patients with gastric, colorectal, pancreatic, breast cancers, and other malignancies [6,[12][13][14][15]. All these suggest that aberrant STC2 expression is closely related to the occurrence, progression, clinical manifestations, and prognosis of tumors. Kita Y, Law AY et al. further con rmed that STC2 could be a predictor for lymph node metastasis of esophageal squamous cell carcinoma and that it promoted epithelial-mesenchymal transition (EMT) of ovarian cancer cells under hypoxia [11,16]. But little is known about the value of STC2 in predicting HNSC prognosis and related manifestations. Therefore, we mined the Gene Expression Omnibus (GEO) and the Cancer Genome Atlas (TCGA) databases to explore whether high STC2 expression and DNA methylation were associated with HNSC progression and whether STC2 expression was involved in immune cell in ltration in HNSC.

Immune in ltration analysis
Timer2.0 offers comprehensive assessments for the association between immune in ltration and STC2 gene features in HNSC [20]. We analyzed the crosstalk between STC2 expression and cancer-associated broblasts (CAFs) using the TIMER, TIDE, XCELL, MCPCOUNTER, and EPIC algorithms to assess immune in ltration levels. The Partial correlation coe cient (partial corr.) and the corresponding P-value were yielded via the Spearman rank correlation test following purity adjustment. The data were visualized as heat maps and scatter plots. In addition, the correlation module was utilized to evaluate the association between STC2 and tumor-in ltrating immune cells, including T cells (general), CD8 T cells, B cells, M1 cells, and M2 cells, in HNSC, as described in our previous studies [21,22].

Enrichment analysis of STC2-related genes
We searched the STRING website (https://string-db.org/) [23] for proteins interacting with STC2 protein, with an interaction cutoff of 0.4. The STC2 protein-protein interaction (PPI) data were clustered using a clustering method. The top 102 genes positively associated with STC2 in HNSC were analyzed using UALCAN, and genes with extremely low expression (median TPM < 0.5) were ruled out. Then KEGG (Kyoto encyclopedia of genes and genes) pathway analysis and GO (gene ontology) enrichment analysis were performed on these genes in R [R-3.6.1,64 bit](https://www.r-project.org/).

Clinical and prognostic implication of STC2 overexpression in HNSC
We analyzed STC2 expression in HNSC using GSE143112, GSE51985, and GSE59102 cohorts. The results showed high STC2 expression in HNSC (or combined with LSCC) (Fig. 1), which was subsequently con rmed with the TCGA cohort using the GEPIA and UALCAN web-tools ( Fig. 2a,b). STC2 expressions in stage 1, 2, 3, 4 HNSC signi cantly increased versus normal controls (all P < 0.001) (Fig. 2c). The Kaplan Meier curves and Cox proportional hazards regression model analysis for its correlation with survival prediction revealed that high STC2 expression was a prognostic risk factor of worse OS in HNSC (Table 1, Fig. 2d). The cBioPortal cancer genomic analysis offered clear alterations that the frequency of TP53 mutation was about 70% in HNSC and that STC2 expression was markedly upregulated in the mutated TP53 group (P = 0.0015) (Fig. 3a,b). This suggested a close relationship between aberrant STC2 expression and TP53 mutation in HNSC. So we further mined the HPA database for STC2 protein expression pro les. By comparing the intensity of IHC staining between STC2-positive HNSC tissues versus normal controls, we found that STC2-positive staining was stronger in HNSC tissues than normal ones (Fig. 3c,d), further demonstrating STC2 overexpression in HNSC.  (Fig. 4a,b), and the latter were identi ed at 33 CpG sites (Fig. 4c). Next, we estimated methylation levels at the CpG sites affecting STC2 mRNA expression using Spearman's rank correlation coe cient. As illustrated in Fig. 4d, four CpG sites (cg19673350, cg18335607, cg02135395, and cg06439005) were most negatively correlated with STC2 expression, and another four (cg01984743, cg26605683, cg15420720, and cg03846076) most positively correlated with STC2 expression were con rmed. Kaplan-Meier survival curves were plotted based on the methylation levels at the two groups of CpG sites (Fig. 5). The Kaplan-Meier analysis for the negative-correlation group revealed that HNSC patients with CpG demethylation was associated with worse OS. The survival curves for the positivecorrelation group revealed that CpG hypermethylation predicted worse OS of HNSC patients. These results further implied that high STC2 expression predicted a poor prognosis of HNSC patients.

Relationship between STC2 and immune cells
Tumor-in ltrating immune cells are important components in the tumor microenvironment, closely associated with tumor initiation, development, and metastasis [24][25][26][27]. We searched the TIMER2.0 database and assessed the association between STC2 mRNA expression and immune cell in ltration.
CAFs as a component of the tumor-stromal compartment predominantly regulate the function of multiple tumor-in ltrating immune cells [28]. We estimated the relationship between the CAF in ltration level and STC2 expression in HNSC using the TIMER, TIDE, XCELL, MCPCOUNTER, and EPIC algorithms. The results showed that the CAF in ltration level was signi cantly and positively correlated with STC2 expression (Fig. 6a). Each STC2 expression matrix where STC2 expression was linked to CAFs/tumorin ltrating immune cells was analyzed with one algorithm after adjusting for tumor purity (Fig. 6b). By using the TIDE algorithm, the analysis showed that STC2 expression was positively associated with the CAF in ltration level (r = 0.413, P = 1.18e-21). This correlation was con rmed by analysis using other algorithms. Furthermore, we analyzed associations of STC2 expression with tumor-in ltrating immune cells

in HNSC, including T (general) cells, CD8 T cells, B cells, M1 cells, M2 cells. STC2 expression was negatively associated with the in ltration levels of T (general) cells, CD8 T cells, and B cells and positively
associated with those of M1 and M2 cells (Fig. 7). From the above results, it can be concluded that STC2 expression is closely linked to immune cell in ltration in HNSC, and high STC2 expression can predict worse OS of patients. We therefore speculated that immune in ltration could be one of the mechanisms contributing to STC2's impacts on HNSC prognosis.

Enrichment analysis of STC2-related genes
We initially screened 102 genes closely related to aberrant STC2 expression in HNSC using TCGA data with the UALCAN web-tool. We then implemented KEGG and GO enrichment analyses to explore the underlying mechanisms of STC2's prediction of HNSC prognosis. The results showed that these hub genes were most enriched in "metabolic pathways" and "biosynthesis of antibodies" based on KEGG analysis (Fig. 8a). Meanwhile, "oxidation-reduction process" and "positive regulation of cell proliferation" were top GO terms signi cantly associated with these genes (Fig. 8a). The top 10 proteins interacting with STC2 were illustrated in Fig. 8b.

Discussion
HNSC is the sixth most common non-skin cancer worldwide, with an overall incidence of 600,000 cases per year and a mortality rate of 50% [29]. STC2 gene encodes a secreted glycoprotein abundant in various tissues with an autocrine/paracrine function. Yoshiaki Kita et al. suggested that high STC2 expression in tumor tissues sensitively predicted shorter ve-year survival [16]. The upregulation of STC2 expression has been proven in other tumor types, for example, colorectal and prostate cancers [30,31]. It may be associated with clinical features of HNSC progression, such as tumor proliferation and lymph node invasion, and a worse prognosis [16]. All these are consistent with our ndings. Our results also demonstrated that the STC2 gene was differentially expressed between normal and stage 1-4 HNSC patients, respectively, and this difference was more evident in patients with higher clinical stage. Therefore, we speculated that STC2 was involved in HNSC progression.
TP53 protein regulates the expressions of a large number of target genes and encodes p53. P53 has been considered an oncogene early on as it is preferentially expressed in cancerous cells [32]. Since 1989 when Levine et al. found wild-type p53 as a tumor suppressor gene [33], it has captured more attention and has been con rmed as a regulator of several cellular activities, such as apoptosis, proliferation, differentiation, and cell-cycle control [34]. Its multi-function role has been observed in DNA damage, hypoxia, oxidative stress, DNA mutations, nutrient deprivation, and other stress responses. Alterations in TP53 gene expression, needless to say, are most certain to be detected in a signi cant proportion of human cancers. Most studies identi ed TP53 mutations in exons 5-8 with missense changes, but some ascertained functional mutations in exons 5-8 of TP53, which actively contributed to cancer progression [35,36]. Evidence about TP53 mutations generally supports their associations with shorter recurrencefree survival or OS of HNSC patients [37][38][39]. This relationship has subsequently been ascertained by a clinical trial comparing TP53-mutated versus wild-type TP53 patients with HNSC [39]. Our study found that the frequency of TP53 mutations in HNSC was approximately 70%, and STC2 expression was pronouncedly increased in the mutated TP53 group. Therefore, the association between TP53 mutations and high STC2 expression is more likely to affect HNSC occurrence and development.
Mammalian DNA methylation is essential in the epigenetic regulation of gene transcription. Dysregulation of genomic DNA methylation has been found to participate in carcinogenesis [40]. A previous study reported that STC2 silencing in human cancer cells could be associated with the methylation of CpG islands of STC2. Methylation of the STC2 promoter is a potential mechanism to hinder STC2 expression in cancer cells [41]. Hence, we explored the relationship between STC2 expression and methylation in HNSC, and consistently, the results suggested a negative correlation between them. Moreover, the eight CpG sites revealing a positive and negative correlation with STC2 expression were classi ed into two groups, respectively. Kaplan-Meier survival curves for the negative-correlation group indicated the association between a lower methylation level and high STC2 expression. The survival curves for the positive-correlation group showed CpG hypermethylation predicted higher STC2 expression, suggesting worse OS of patients.
Numerous studies of STC2 expression within the tumor microenvironment found that STC2 enhanced tumor cell migration and invasion [42,43]. Clinical studies con rmed STC2 as a new biomarker of multiple cancers due to its involvement in tumor neovascularization [44,45]. STC is known for its potent anti-in ammatory effect [46,47], of which STC1 exerts the effect via inducing uncoupling proteins to reduce oxidative stress [48]. STC2, homologous to STC1, is activated by oxidative stress to forestall cell apoptosis. The present study revealed a positive correlation between STC2 expression and the in ltration levels of CAFs, indicating that STC2 may promote brogenesis in HNSC. Previous studies have reported that STC2, as an immune-related protein, is a potential oncoprotein in the process of hepatocarcinogenesis and is considered a promising biomarker and molecular target [49,50]. It is reported that STC2 can inhibit CD 8 + T cell in ltration [51]. These are consistent with our conclusions.
Our study offered KEGG and GO enrichment analyses based on initially screened genes related to aberrant STC2 expression in HNSC and identi ed "metabolic pathways" and "biosynthesis of antibodies" as highly relevant pathway terms of STC2's role in head and neck carcinogenesis. Also, "oxidationreduction process" and "positive regulation of cell proliferation" were highly associated with the functions of the STC2 gene in this process.
Besides, three limitations are deserving to be discussed. First, inevitably, the raw data from the GEO and TCGA databases were originally used for other purposes. So we have been unable to validate the prognostic value of STC2 gene knockdown or protein downexpression in HNSC. Second, although we preliminarily analyzed the biological processes of STC2 in HNSC using enrichment analysis, the speci c mechanism of how STC2 expression/methylation patterns affect HNSC progression requires more research support. Finally, the details of how STC2 promotes immune cell in ltration in HNSC are worthy of standardized investigation. However, the current results are encouraging, considering STC2 as a promising biomarker for the predicting HNSC prognosis. Competing interests:       Kaplan-Meier survival curves based on the methylation levels at eight CpG sites.  The correlation of STC2 expression with the in ltration levels of tumor-in ltrating immune cells.