A Computational Approach to Justifying Stratifin as a Candidate Diagnostic and Prognostic Biomarker for Pancreatic Cancer

Pancreatic cancer (PC) is considered a silent killer because it does not show specific symptoms at an early stage. Thus, identifying suitable biomarkers is important to avoid the burden of PC. Stratifin (SFN) encodes the 14-3-3σ protein, which is expressed in a tissue-dependent manner and plays a vital role in cell cycle regulation. Thus, SFN could be a promising therapeutic target for several types of cancer. This study was aimed at investigating, using online bioinformatics tools, whether SFN could be used as a diagnostic and prognostic biomarker in PC. SFN expression was explored by utilizing the ONCOMINE, UALCAN, GEPIA2, and GENT2 tools, which revealed that SFN expression is higher in PC than in normal tissues. The clinicopathological analysis using the ULCAN tool showed that the intensity of SFN expression is commensurate with cancer progression. GEPIA2, R2, and OncoLnc revealed a negative correlation between SFN expression and survival probability in PC patients. The ONCOMINE, UCSC Xena, and GEPIA2 tools showed that cofilin 1 is strongly coexpressed with SFN. Moreover, enrichment and network analyses of SFN were performed using the Enrichr and NetworkAnalyst platforms, respectively. Receiver operating characteristic (ROC) curves revealed that tissue-dependent expression of the SFN gene could serve as a diagnostic and prognostic biomarker. However, further wet laboratory studies are necessary to determine the relevance of SFN expression as a biomarker.


Introduction
The pancreas is a pear-shaped organ located in the abdomen, and it plays an essential role in converting foods to become fuel for body cells. However, in some cases, the growth of the pancreas becomes uncontrollable due to some reasons and thus becomes cancerous. Pancreatic cancer (PC) is one of the deadliest cancers and is the seventh most common cause of cancer-related deaths in both men and women [1]. According to GLOBOCAN, in 2018, the estimated number of PC cases and deaths were 458,918 and 432,242, respectively, corresponding to 2.5% of all PC is often difficult to diagnose at an early stage; as a result, the majority of PC cases are diagnosed at an advanced stage, and only 10-20% of cases are surgically treatable [7]. This trend is due to the lack of distinct clinical signs and symptoms, due to a lack of accurate biomarkers, and due to the limited resolution of imaging techniques, resulting in a high mortality rate in PC [7,8]. The 5-year overall survival rate for PC has remained low at 3% in recent years, as more than half of PC patients are diagnosed at an advanced stage [9,10]. Compared with the screening programs for other cancers, such as lung, breast, colon, and cervical cancers, those for PC are difficult to implement due to the lack of specificity of a particular test [11]. The most common biomarker that has been approved by the US Food and Drug Administration (FDA) for PC diagnosis is the carbohydrate antigen (CA) . However, CA has not been considered to be the most effective screening tool due to its low sensitivity and specificity and poor predictive value of 0.5-0.9% in asymptomatic patients [12,13]. Meanwhile, CA 19-9 expression may increase in other medical conditions, such as acute cholangitis, pancreatitis, obstructive jaundice, and liver cirrhosis [11]. Currently, there are no biomarkers with an adequately high accuracy that could be used to screen sporadic PC; therefore, there is an urgent need to identify biomarkers for PC [14].
Stratifin (SFN) encodes the 14-3-3σ protein, which is a member of a highly conserved family of 14-3-3 proteins found in all eukaryotic organisms [15]. SFN was first identified as human mammary epithelial marker 1 before being rediscovered as a key regulator of cell cycle checkpoints [16,17]. Decreased SFN expression has been found in various cancers, including breast [18], lung [19], liver [20], endometrium [21], head and neck [22,23], vulva [24], and prostate cancers [25,26]. Conversely, upregulation of the SFN gene expression has been observed in other cancers, including pancreatic [27][28][29], colorectal [30], and esophageal squamous cell carcinoma [31]. The expression of the SFN gene varies in different cancers, and it performs a double-edged function [32]. Therefore, the role of SFN expression is likely context dependent. On the basis of its tissue-dependent expression pattern, SFN can be used as a diagnostic and prognostic biomarker in PC. However, there has been insufficient evidence demonstrating that SFN expression can be used as a biomarker for PC.
This study compared the expression pattern of SFN in PC patients and healthy individuals based on data obtained from online databases. Moreover, clinicopathological features, coexpression, prognostic values, gene ontologies, signaling pathways, and network analysis were performed. The workflow for this study is depicted in Figure 1.  Figure 1: Schematic of the overall workflow in this study. 2 BioMed Research International 10%, and 2, respectively. ONCOMINE is a web-based datamining platform that is aimed at facilitating the identification of cancer-related genes by analyzing genome-wide expression [33,34]. The pancancer view for SFN was determined using the UALCAN (http://ualcan.path.uab.edu/) platform. UALCAN is a comprehensive, user-friendly, and interactive web-based resource used to study cancer OMICS data [35]. Then, GENT2 (http://gent2.appex.kr/gent2/) [36] was adopted by using the GPL570 platform (HG_U133_ Plus_2) to investigate the SFN expression levels in different types of cancer.  [37]. UALCAN was utilized to obtain SFN expression data in PC and then compared with those in normal tissues.

SFN Expression in relation to Clinicopathological
Parameters in PC. The UALCAN web tool with default settings was used to assess the mRNA expression of the SFN gene in PC patients based on their clinicopathological features. In this investigation, SFN expression was analyzed based on clinicopathological parameters, such as cancer stages, race, age, nodal metastasis status, and tumor grade. Only the statistically significant results were taken into account in the analysis.

Association between SFN Expression and Survival
Probability in PC Patients. The impact of SFN expression on the survival probability of PC patients was investigated using the GEPIA2, R2 (http://r2platform.com), and OncoLnc (http://www.oncolnc.org/). The R2 genomics platform is a publicly available web-based platform that allows researchers to integrate, analyze, and visualize clinical and genomics data [38]. OncoLnc is an online tool for estimating survival relationships and for accessing clinical data for mRNAs, miRNAs, and lncRNAs (long noncoding RNAs) [39]. The R2 platform was utilized to generate a Kaplan-Meier plot (OS) for the SFN gene against the mixed tumor pancreas Hussain-130-rma-sketch-hugene10t and mixed pancreatic adenocarcinoma Sadanandam-47-MAS5.0-u133p2 datasets by setting the optimum cut-off values. The Kaplan-Meier plot was drawn by splitting the patient population at the median. A P < 0:05 was considered significant.

Coexpression Analysis of the SFN Gene in PC Cancer.
The SFN gene's coexpression profile in PC was determined, and the corresponding heat map was obtained from the Collisson Pancreas dataset through the ONCOMINE web tool. From this dataset, the cofilin 1 (CFL1) gene was the most positively correlated with SFN expression in PC. To confirm the relationship between SFN and CFL1, we used the TCGA (PAAD) dataset from the UCSC Xena server (https:// xenabrowser.net/) [40]. Furthermore, correlation data were obtained from the UCSC Xena server, and a scatter plot was drawn by using ggplot2 [41]. The GEPIA2 was utilized to confirm the positive correlation between SFN and CFL1 transcripts in the PC.
2.6. Enrichment Analysis of the SFN Gene. The Enrichr (https://maayanlab.cloud/Enrichr/) web tool was used to extract the gene ontologies and signaling pathways of the SFN gene, as well as the corresponding bar graphs. Enrichr is a user-friendly web-based enrichment analysis tool that graphically presents the collective functions of genes [42,43]. Gene ontologies were analyzed using GO Biological  [44]. We also used the GeneMANIA (https://genemania.org/) web platform to create an interaction network of closely linked genes. GeneMANIA is used to predict the function of a gene or gene lists and to identify the physical interaction, genetic interactions, coexpression, pathway, colocalization, and shared protein domain [45].
2.8. TF and miRNA Network Analyses. TFs are proteins that regulate gene expression by binding to certain DNA sequences [46], and miRNAs are a type of noncoding RNAs that play crucial functions in gene regulation [47]. TF and miRNA networks were constructed based on the ChEA [48] and TarBase [49]  Cell color is determined by the best gene rank percentile for the analyses within the cell. NOTE: An analysis may be counted in more than one cancer type.

mRNA Expression in Human Cancers.
We analyzed the expression pattern of SFN in numerous cancer studies by using the ONCOMINE platform. The results showed that SFN was upregulated in seven cancer types, namely, bladder, head and neck, kidney, liver, lung, ovarian, and pancreatic cancers ( Figure 2(a)). In the pancancer view based from the ULCAN tool, we found that SFN was upregulated in 16 cancer types, downregulated in 6 cancer types, and equally expressed in 2 cancer types ( Figure 2(b)). We also confirmed the upregulation of SFN in different cancers using the GENT2 tool.

SFN Expression in PC versus Healthy
Tissues. SFN was significantly upregulated in different PC types, including pancreatic adenocarcinoma, pancreatic carcinoma, and pancreatic ductal adenocarcinoma, compared with its expression in normal tissues (Figures 3(a)-3(c) and Table 1). Using the GEPIA2 and UALCAN platforms, we further assessed the upregulation of SFN. Our findings indicated that SFN expression was significantly higher in PC tissues than in normal tissues (Figures 3(d) and 3(e)).

SFN Expression in relation to Clinicopathological
Parameters in PC. We looked at variations in SFN gene expression levels in PC patients based on their clinicopathological features. In terms of individual cancer stages, the increase in SFN expression correlated with that in PC progression ( Figure 4(a)). In terms of patients' race, SFN expression is increased in Asian patients (Figure 4(b)). In

Association between SFN Expression and Survival
Probability in PC Patients. To evaluate the prognostic value of the SFN gene, we determined the survival probability of PC patients using GEPIA2, R2, and OncoLnc. The results obtained from these tools revealed a negative correlation between survival probability and SFN expression (i.e., high SFN expression results in low survival probability). GEPI A2 provided data on the overall and disease-free survival probability of PC patients (Figures 5(a) and 5(b)), whereas R2 and OncoLnc provided information on overall survival probability (Figures 5(c)-5(e)). The analysis results underscored the prognostic relevance of a high SFN expression in PC patients.

12
BioMed Research International E2F1, MYC, and MYBL2 (Figure 9(a)). Modification of these TFs might play a significant role in altering the SFN gene expression in PC. In the miRNA analysis, we obtained a network showing the direct interaction of 19 miRNAs with SFN ( Figure 9(b)). These miRNAs can modify the SFN expression at the posttranscriptional stage.
3.9. ROC Curve Analysis of the SFN Gene. In the ROC curve, the area under the curve (AUC) is used to discriminate between classes. In the GSE16515 dataset, the AUC for the SFN gene expression was 0.965 ( Figure 10(a)) and the AUC for the survival of patients was 0.637 ( Figure 10(b)). These AUC results indicate that the SFN gene might be used as a diagnostic and prognostic marker.

Discussion
PC is one of the most aggressive cancers affecting human health, and it is considered the "silent disease," as it does not show noticeable symptoms at an early stage [52]. Given 13 BioMed Research International that it displays characteristics similar to those of other diseases, such as ulcer, gastritis, and pancreatitis, it is mostly diagnosed at an advanced stage [53]. As early detection remains difficult, finding a potential novel biomarker that aids in early detection is desired. In this study, we utilized bioinformatics approaches to assess the importance of the SFN gene as a biomarker in PC prediction.
Upregulated SFN gene expression in PC and other cancer types was observed in ONCOMINE, UALCAN, and GENT2. The upregulated SFN expression in PC cells was compared with that in normal pancreatic cells using the data from ONCOMINE, GEPIA2, and UALCAN. A study on the molecular profiling of stroma in pancreatic ductal adenocarcinoma has revealed the upregulated expression of SFN, along with other genes [54]. This upregulated SFN expression in PC is supported by other studies [27,[55][56][57]. Gene expression levels in cancers can vary under different clinicopathological conditions, as cancer is a heterogeneous and complex disease. We analyzed the SFN expression based on patients' age, race, tumor grade, tumor stage, and nodal status. The results showed that SFN was highly upregulated among Asians, among 41-60-year-old individuals, among those with a positive nodal status, and among grade 3 tumor patients. Interestingly, in the case of cancer stages, SFN expression increased proportionally with cancer stage progression. Then, the prognostic value of SFN in PC was evaluated using the GEPIA2, OncoLnc, and R2 platforms. High SFN expression significantly (P < 0:05) correlated with low overall and disease-free survival. Our current findings agreed with those of another study in which SFN was considered an independent prognostic biomarker in pancreatic ductal adenocarcinoma [54]. Moreover, it has been demonstrated that the elevated 14-3-3σ protein levels likely contribute to the poor prognostic outcome of human pancreatic tumors, as they promote resistance to radiation and anticancer treatments [15].
Gene coexpression provides information that aids in the identification of functionally linked genes. Coexpression analysis using the ONCOMINE platform revealed 13 genes, among which CFL1 was highly coexpressed with SFN. Furthermore, CFL1 coexpression in PC was confirmed by GEPIA2 and UCSC Xena. CFL1 is a small, ubiquitous, actin-binding protein that plays important roles in cytokinesis, endocytosis, apoptosis, cell proliferation, and migration, as well as in tumor development, infiltration, and metastasis [58,59]. Moreover, it has been reported that this protein is necessary for the invasion and spread of numerous human malignant solid tumors [60,61]. Recent studies have found a positive association between high CFL1 gene expression and PC progression [59,62].
Enrichment analysis for the SFN gene was performed by utilizing the Enrichr web platform. The most prominent pathways, including cell cycle control, Chk1/Chk2-mediated inactivation of cyclin B, DNA damage response, aldosteroneregulated sodium reabsorption, estrogen-responsive protein efp control cell cycle, and p53 pathway, were obtained from BioPlanet 2019, Reactome 2016, WikiPathway 2021, KEGG 2021, Biocarta 2016, and Panther 2016, respectively. SFN was initially found to be a p53-inducible gene that responds to DNA-damaging agents [63]. A study has reported that SFN inhibits the initiation of mitosis by sequestering the mitotic initiation complex (cdc2-cyclin B1) and preventing it from entering the nucleus [64]. In this manner, SFN causes G2 arrest, allowing damaged DNA to be repaired. It has been demonstrated that SFN directly controls the G2/M checkpoint of the cell cycle by protecting p53 against MDM2mediated ubiquitination and degradation [65][66][67]. These findings indicate that SFN acts as a negative regulator of cell cycle progression and might be considered a tumor suppressor. However, SFN plays a double-edged function in human cancers, and its function may vary among organs and tissues [32,68]. Meanwhile, accumulation of 14-3-3σ has been observed in PC, but it cannot perform its major ascribed functions, such as sustaining a G2 checkpoint and performing an antiapoptotic action, due to multiple alterations in its interaction with downstream partners [69]. Network analysis based from the STRING database revealed the functional interaction partners of SFN, namely, TP53, FOXO1, LRRK2, RAF1, CDK2, BAD, CDC25B, AKT1, ANPEP, and YWHAZ. It has been reported that overexpression of CDC25B is associated with pancreatic ductal adenocarcinoma and that its inhibitor prevents PC cell growth by blocking the G2/M phase transition via the inhibition of cdc2 dephosphorylation [70]. According to the NCBI, defects in the ANPEP gene enhances angiogenesis, tumor growth, and metastasis [71]. Meanwhile, overexpression of the YWHAZ gene has been demonstrated to be a prognostic and therapeutic target in gastric cancer [72,73]. In GeneMANIA, SFN shares consolidated pathways with MST1R and YWHAG. MST1R expression has been shown to play an oncogenic function in human pancreatic intraepithelial neoplasia, as well as in primary human and animal metastatic cell lines [74]. In PC, the overexpression of the YWHAG gene is associated with poor overall survival compared with low YWHAG expression [75]. Furthermore, our network analysis revealed some TFs and miRNAs that might play important roles in determining how SFN gene expression is regulated at the transcriptional and posttranscriptional levels.
In ROC analysis, SFN expression showed excellent (AUC = 0:917) diagnostic value of pancreatic cancer. A meta-analysis study showed that the sensitivity and specificity of CA 19-9 were 78.2% and 82.8%, respectively [76]. However, the CA 19-9 level may be augmented in other medical conditions, such as acute cholangitis, pancreatitis, obstructive jaundice, and liver cirrhosis [11]. In our study, SFN also exhibited as a good (AUC = 0:637) prognostic marker in pancreatic cancer. In these aspects, SFN might be considered as an auxiliary biomarker of CA 19-9 in PC. Of course, there are some limitations in our study. First, due to the lack of enough datasets, the sample size for analysis was relatively small. Second, the absence of in vivo and in vitro experiments is another flaw of our study. Third, this study cannot explain how the tissue-specific upregulation SFN gene is related to pancreatic cancer. Therefore, further wet laboratory molecular studies are needed.

Conclusion
Data from the online bioinformatics platforms utilized in this study showed that SFN expression in PC was upregulated relative to that in normal tissues. Moreover, a negative correlation between SFN expression and survival probability was found in PC. In our network analysis, SFN-associated proteins, TFs, and miRNAs were identified. Based on these findings, we can conclude that the high tissue-dependent SFN expression might be used as a biomarker for diagnosis, prognosis, and therapeutic purposes. However, further wet laboratory-based studies are needed to bolster the significance of SFN overexpression in PC.

Data Availability
Any data or information used in this current study is available from the corresponding author on reasonable request.

Conflicts of Interest
The authors declare that they do not have conflicts of interest.