The Cancer Genome Atlas and experimental validation-based analysis of the SPRR2A gene’s expression and clinical relevance in endometrial cancer

Introduction

Endometrial cancer (EC) is women’s most common kind of gynaecological cancer, making up between 20 and 30 per cent of all gynaecological malignancies and 7 per cent of all cancers in women overall.1 Since there are no specific indicators, patients must undergo uterine curettage for a correct EC diagnosis.2 CA125 is currently the most commonly used serum tumour marker for EC. However, CA125 is also a serum tumour marker for ovarian epithelial cancer.3 Thus, current biomarkers are inadequate for early and accurate diagnosis, so the search for new specific biological markers is vital to guide clinical management. New specific biological markers may supplement CA125.

Although a good 5-year relative survival rate (70%–92%) is linked to early-stage EC,4 the incidence is significantly lower in individuals who have risk factors such as lymph node metastases, lymph-vascular space invasion, non-endometrioid (serous or clear cell) histology, or stage II or stage III endometrioid EC.5 Since the patient’s survival mostly depends on early detection of cancer, early screening and accurate diagnosis are key to improving prognosis. To improve disease diagnosis, novel and diverse biomarkers for exhaustive early detection are needed.

More cancer biomarkers have been discovered in recent years, including microarrays and high-throughput sequencing.6–8 These data have uncovered a wealth of valuable biological information, providing an important research tool in the search for specific and sensitive molecular markers for diagnosis and prognosis. Among these, to analyse and study huge volumes of human tumour tissue to find molecular mutations at the DNA, RNA, protein and epigenetic levels, a key data source is The Cancer Genome Atlas (TCGA) database. TCGA, a comprehensive resource for cancer research, integrates multi-omic data, including genomic, transcriptomic, epigenomic and proteomic data, providing a valuable resource for the scientific community. To screen differentially expressed genes (DEGs) between EC and normal tissues, we examined the levels of protein-coding gene expression in EC using data from TCGA database by the R package DESeq2. The enrichment analysis of the Gene Ontology (GO) and the KEGG pathway database was carried out to create a functional enrichment study of DEGs. In addition, we constructed the network of protein-protein interactions (PPI) of DEGs. Finally, the hub gene SPRR2A was identified by survival analysis. This finding provides a theoretical basis for the precise treatment and diagnosis of endometrial carcinoma.

Materials and methods

The experimental human EC cell lines (HEC-1-A, HEC-1-B, RL95-2, AN3CA, Ishikawa) and the control human endometrial epithelial cells (hEEC) and DMEM (Dulbecco's Modified Eagle Medium) medium supplemented with 10% FBS were purchased from Procell company (China). SYBR Green I (10000 ×) was purchased from Solarbio company (China). SPRR2A primer sequences were synthesised by Shanghai Bioengineering Service Co.

Results

Differential gene screening

The expression profile of protein-coding genes in the EC dataset from TCGA was collected for differential gene analysis by DESeq2 (figure 1A). After eliminating nulls, the screening produced 18 693 IDs in total. The threshold of |log2(FC)| > 2 and p.adj <0.05 was reached by 2573 IDs. The top 20 up/downregulate DEGs have been listed in online supplemental table 1.

Figure 1

Request permissions

(A) Volcano plots of differentially expressed genes of EC (DEGs). The X-axis represented the fold change (FC) of expression level between EC and normal tissue. The Y-axis represented the adjusted p value of the differential expression. Genes with higher or lower expression levels in EC were shown in red and blue respectively (LogFC ≥2 or ≤ −2, P.adj<0.05). The top 10 genes (SPRR2A, SPRR2D, SPRR3, SPRR1A, SPRR1B, KRT16, CASP14, IVL, RPTN, PI3) in the PPI network were displayed in the volcano plot. (B) Analysis of GO and KEGG involved in the differentially expressed genes. The X-axis represented the significance of enrichment analysis according to adjusted p-value. The Y-axis represented the description information of the top three entries enriched to each pathway, which included biological processes (BP), cellular composition (CC), molecular functions (MF) and KEGG. EC, endometrial cancer; DEGs, differentially expressed genes; GO, Gene Ontology; PPI, protein-protein interaction.

Hub gene screening and prognostic analysis

Based on the STRING database and Cytoscape software, we constructed a PPI network for the DEGs in EC. There were 396 nodes and 955 edges made up of the PPI network (online supplemental figure 1). Nodes represent proteins produced by protein-coding genes. Edges represented associations between proteins. The top 10 genes (SPRR2A, SPRR2D, SPRR3, SPRR1A, SPRR1B, KRT16, CASP14, IVL RPTN, PI3) in PPI network ranked by MCC method had been identified as potential hub genes (figure 2A). The detailed analytical data are provided in online supplemental files 1–3 and 5. The abbreviations and functions of potential Hub genes are shown in table 1. According to the median gene expression, these 10 potential hub genes were divided into two groups: high and low expression. Then, potential hub genes underwent Kaplan-Meier analysis, which revealed that SPRR2A (ENSG00000241794) at a higher level was connected to a poorer prognosis of EC. The difference was significant (p<0.05) (figure 2B). The other nine genes (SPRR2D, SPRR3, SPRR1A, SPRR1B, KRT16, CASP14, IVL RPTN, PI3) underwent Kaplan-Meier analysis; the differences were not significant (p>0.05) (online supplemental figure 2).

Figure 2

Request permissions

(A) Network of protein-protein interactions (PPI). The top 10 genes in the STRING network were ranked by MCC method according to score. The higher the score, the darker the colour and the higher the ranking. The top 10 genes included SPRR2A, SPRR2D, SPRR3, SPRR1A, SPRR1B, KRT16, CASP14, IVL RPTN and PI3. (B) Survival analysis of hub gene SPRR2A. The X-axis represented the observation time (months). The Y-axis represented the overall survival (OS) rate. Each point on the curve represented the patient’s overall survival (OS) rate at that point in time. Patients were stratified into high and low expression groups based on median SPRR2A expression. Genes with high or low expression groups in EC were shown in red and blue lines respectively. The survival probability is plotted against time (months), with the risk table below indicating the number of patients remaining at specific time points. The HR for high vs low SPRR2A expression was 1.52 (95% CI 1.00 to 2.30), with a P value of 0.048, suggesting a potential poor prognostic impact of SPRR2A expression on EC patient survival. EC, endometrial cancer; MCC, maximum clique centrality.

Table 1

View inline•Open as popup

Abbreviations and functions of potential hub genes

We identified prognostic variables and independent risk factors for the prognosis of EC by examining clinical data linked to prognostic data. Poor predictive factors for EC included the late FIGO stage, high SPRR2A expression, pathological type II, older age, deep myometrial infiltration and high histological grade, according to univariate Cox analysis (p<0.05). FIGO stage, pathological type, depth of myometrial infiltration and high histological grade, were identified as independent risk factors for the prognosis of EC using multifactorial Cox analysis (p<0.05) (table 2).

Table 2

View inline•Open as popup

Univariate and multivariate Cox regression study of EC prognosis

SPRR2A gene expression in cancerous and non-cancerous tissues

We examined RNA-seq data gathered from the TCGA-UCEC database to compare the expression of the SPRR2A gene between EC and non-cancer tissues. The data included 23 pairs of matching cases and 589 unpaired cases (35 cases of normal tissue in the control group and 554 cases of EC tissue in the experimental group). The statistical description is provided in online supplemental file 4.The outcome demonstrated that EC tissues had a greater level of SPRR2A expression than normal tissues, and this difference was significant (p<0.05) (figure 3).

Figure 3

Request permissions

Differential expression of SPRR2A in TCGA database. (A) The expression of SPRR2A in paired cases. (B) The expression of SPRR2A in unpaired cases. The X-axis represented groups. The Y-axis represented the expression of SPRR2A. Markers of significance include **p<0.01 and ***p<0.001. TCGA, The Cancer Genome Atlas.

The relative RNA expression of SPRR2A was significantly upregulated in human EC cell lines (HEC-1-A, HEC-1-B, RL95-2, AN3CA and Ishikawa cell lines) compared with the normal hEEC line hEEC. The difference in SPRR2A RNA’s relative expression was significant (all p<0.05) (figure 4A). These findings indicate that SPRR2A is markedly overexpressed in EC cell lines relative to normal endometrial epithelial cells. There was a significant difference in the relative RNA expression levels of SPRR2A between EC tissues and adjacent tissues. As shown in figure 4B, RT-qPCR detection showed that the relative RNA expression level of SPRR2A in EC tissues was significantly higher than that in adjacent tissues, indicating that SPRR2A may play an important role in the development of EC.

Figure 4

Request permissions

Relative RNA expression of SPRR2A in endometrial cancer. (A) Relative RNA expression of SPRR2A in endometrial cancer cell lines. X-axis: cell lines (hEEC, HEC-1-A, HEC-1-B, RL95-2, AN3CA, Ishikawa). Y-axis: relative RNA expression of SPRR2A. (B) Relative RNA expression of SPRR2A in endometrial cancer tissues and adjacent tissues. X-axis: tissue types (adjacent, tumour). Y-axis: relative RNA expression of SPRR2A. Statistical significance: *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

Correlation between SPRR2A expression and clinical characteristics

The characteristics of 544 patients with EC, including clinical and gene expression data, were collected from TCGA database. Based on the mean value of SPRR2A expression, the patients with EC were divided into high- and low-SPRR2A expression groups (table 3).

Table 3

View inline•Open as popup

Correlation between SPRR2A expression and clinicopathologic characteristics of patients with EC

The expression levels of SPRR2A in EC were analysed across multiple clinical subgroups (figure 5). In terms of age, there was no significant difference in SPRR2A expression between patients aged ≤60 and >60. For histological grade, a significant difference was observed. SPRR2A expression was significantly lower in G1 tumours compared with G2 and G3 tumours. In terms of clinical stage, there was no significant difference in SPRR2A expression between stages I and II and stages III and IV. For tumour invasion, there was no significant difference in SPRR2A expression between tumours with invasion <50% and ≥50%. For histological type, there was no significant difference in SPRR2A expression among endometrioid, mixed and serous types.

Figure 5

Request permissions

Expression levels of SPRR2A across different clinical subgroups of endometrial carcinoma. (A) Comparison of SPRR2A expression between patients ≤60 years and >60 years. (B) Analysis of SPRR2A expression across different histological grades (G1, G2, G3). (C) Evaluation of SPRR2A expression in early-stage (stages I and II) versus late-stage (stages III and IV) disease. (D) Assessment of SPRR2A expression based on tumour invasion percentage (<50% vs ≥50%). (E) Investigation of SPRR2A expression in endometrioid versus mixed/serous histological types. Data are presented as log2(TPM+1) values, with statistical significance assessed using non-parametric tests. Statistical significance: ns P＞0.05, *p<0.05.

Discussion

Biomarkers of EC are valuable for the early screening of high-risk women, risk stratification, development of individualised treatment plans and assessment of prognosis. Next-generation sequencing technology has proven to be an effective method for identifying new biomarkers, which gives us new ways of screening EC-related biomarkers.

We chose transcriptome information from the TCGA10 associated with EC for our investigation and screened DEGs between EC and normal tissue using DEseq2.9 The top 20 up/downregulate DEGs in online supplemental table S1 were prioritised due to their significant dysregulation and potential mechanistic role in EC. For instance, CLDN6 is abnormally overexpressed in EC and is significantly associated with disease progression and poor prognosis. It promotes tumour migration and proliferation through the adhesion signalling and oestrogen receptor α pathway.11–13 ZIC1/ZIC2 promotes the proliferation and metastasis of EC cells by regulating lncRNA SNHG12 and the Notch signalling pathway.14 HOXB13 is regulated by FTO (fat mass and obesity-associated protein) demethylation, which activates the WNT (wingless-type MMTV integration site family) pathway to promote EC metastasis.15 OGN (osteoglycin), as an immune-related DEG (IRDEG), its low expression is associated with poor prognosis of EC patients and epithelial-mesenchymal transition (EMT).16 Then we performed an enrichment analysis of the KEGG and GO to functionally annotate the DEGs between EC and normal tissue. GO is a community-based bioinformatics library that describes the biological function and significance of genes and their production by using ontology.17 Through GO enrichment analysis, we learnt that DEGs mostly participate in biological processes, including keratinisation, epidermal development, keratinocyte differentiation, epidermal cell differentiation, muscle tissue development, calcium regulation, etc. Genome sequences and other high-throughput data may be scientifically interpreted using KEGG, an integrated database resource. The KEGG Orthology database stores the molecular activities of genes and proteins together with their ortholog groupings.18 The functions of the pathways were mostly focused on passive transmembrane transport activity, receptor-ligand activity, ion channel activity, DNA-binding transcriptional activator activity and enzyme inhibitor activity. Some pathways are closely associated with EC. For instance, the neuroactive ligand-receptor (NLR) interaction pathway regulates cell proliferation, migration and apoptosis in various cancers. As a common malignant tumour of the female reproductive system, the progression of EC is closely related to hormonal signals and the inflammatory microenvironment.19 20 Recent studies have found that the NLR pathway may affect the invasiveness and therapeutic resistance of EC by mediating the cross-talk between oestrogen receptor and G protein-coupled receptors.21 22

To build the PPI network of EC DEGs, we used Cytoscape and the STRING database. The top 10 genes ranked by the CytoHubba MCC method were obtained as potential hub genes. PPI networks use mathematical graphs with edges and nodes to represent proteins and the dynamics between protein partners.23 Biological applications of PPI networks include the prediction of protein function,24 prioritisation and prediction of potential genes or targets,25 studies of post-genome-wide associations,26 identification of genetic features or patterns connected to illness, as well as the forecasting of disease phenotypic trends. Consequently, these 10 hub genes screened by the PPI network could be potential targets for EC.

Kaplan-Meier analysis of 10 hub genes showed that SPRR2A (ENSG00000241794) was associated with the prognosis of EC. Consequently, we selected SPRR2A as the hub gene and used RT-qPCR to confirm the differential expression in EC and hEEC. The qPCR result showed a higher expression level of SPRR2A in human EC cell lines compared with hEEC. A study discovered that SPRR2A was an independent predictive factor for developing regional recurrence after therapy of head and neck squamous cell carcinoma.27 These findings align with our observations.

In the univariate Cox analysis, high SPRR2A expression was identified as a poor prognostic factor suggesting a significant association between SPRR2A levels and patient survival when other clinical variables were not considered. However, in the multivariate Cox analysis, SPRR2A expression was no longer an independent prognostic factor. This discrepancy can be attributed to the adjustment for potential confounding factors in the multivariate analysis, like clinical stage, histological grade and tumour invasion. Multivariate Cox regression is a more comprehensive and rigorous statistical approach that allows simultaneous evaluation of multiple variables, thereby identifying truly independent prognostic markers. When multiple variables are considered together, the prognostic impact of SPRR2A may be overshadowed by stronger predictors. We observed significant differences in SPRR2A expression across histological grades. Its expression was lower in G1 tumours than in G2 and G3 tumours. This suggests that SPRR2A is associated with tumour differentiation and may affect patient survival through tumour dedifferentiation. Although SPRR2A is not an independent prognostic factor, its expression pattern is linked to tumour biology. SPRR2A still has the potential to be a therapeutic target or a biomarker for EC, and it is worthy of in-depth study.

SPRR2A and other SPRR family members are small proline-rich proteins (SPRRs). This family includes two SPRR1 genes, seven SPRR2 genes and one SPRR3 gene. These genes are structural proteins of the keratinised envelope that serve as a defence barrier. According to many investigations, barrier epithelia from the lung, skin and gut were involved in inflammatory processes, stressful situations, microbial contamination and restorative processes.28 Consequently, because of SPRRs’ function for defence barrier, inflammatory reactions and damage healing are significantly influenced by SPRRs.29–32

Cells go through a process called EMT, in which they stop being epithelial and start to resemble mesenchymal cells. EMT was linked to several tumour-related processes, including malignant development, tumour initiation, tumour cell migration, tumour stemness, intravasation to the circulation, metastasis and therapeutic resistance.33–35 A study about cholangiocarcinoma clarified that SPRR2A promotes local invasiveness by inducing EMT, while inhibiting metastasis by suppressing mesenchymal-epithelial transition (MET) and MUC1 expression. The epithelial migration phases of wound healing, which involves EMT, and the epithelial restoration observed during the reverse process, MET, are both mirrored in the sequential events of cholangiocarcinoma advancement.36 Cell migration is a p53-related process which is also linked to SPRR2A. Tumour cell migration, invasion and metastasis are all regulated by EMT, which is inhibited by p53.28 Another study also pointed out that SPRR2A upregulation inhibits p53 acetylation and its target genes, leading to the temporary maintenance of mesenchymal properties in damaged cells.37 A research elucidated the underlying molecular pathways of SPRR2A-induced EMT. Through its SH3-domain networks, SPRR2A regulates ZEB-1 signalling and promotes both normal and malignant wound healing in BECs (bronchial epithelial cells).38 In summary, SPRR2A is closely related to EMT and subsequently affects tumour progression.

Our study explored the relationship between SPRR2A expression and various clinical pathological factors in EC. The significant difference in SPRR2A expression across histological grades indicates that SPRR2A may be associated with tumour differentiation. This phenomenon is reflected in other tumours. The expression of SPRR2A shows significant differences among neuroendocrine neoplasms (NENs) with different histological grades (G1, G2, G3). According to existing studies, the expression level of SPRR2A in low-grade (G1) tumours is significantly lower than that in intermediate and high-grade (G2/G3) tumours, and this pattern may be associated with the proliferative activity and differentiation status of tumours.39–41 For example, in gastroenteropancreatic neuroendocrine tumours, G1 tumours generally exhibit a lower proliferation index (Ki-67) and more conservative molecular characteristics, while G2/G3 tumours are accompanied by higher proliferative activity and a dedifferentiated phenotype, which may drive the upregulation of SPRR2A.39 40 42 In addition, some studies have pointed out that the changes in the expression of SPRR2A may be related to the dysregulation of the tumour microenvironment or specific signalling pathways (such as TGF-β or HOX genes),40 43 but the specific mechanism still needs further verification. However, the lack of significant differences in other factors like age, clinical stage, tumour invasion and histological type implies that SPRR2A may not be directly linked to these aspects in this study. Overall, these findings provide preliminary insights into the potential clinical significance of SPRR2A in EC, particularly its relationship with tumour differentiation, which warrants further investigation to clarify the exact molecular mechanisms involved and its potential as a therapeutic target or prognostic biomarker.

In response to the comparison of SPRR2A with existing biomarkers such as CA125, we acknowledge that CA125 is currently the most commonly used serum tumour marker for EC. However, it is not specific to EC as it is also a marker for ovarian epithelial cancer. Our study highlights that SPRR2A offers several advantages. The overexpression of SPRR2A in EC tissues, confirmed by both TCGA-UCEC data analysis and RT-qPCR validation, suggests it may serve as a more specific indicator for EC. Moreover, the association of high SPRR2A expression with poorer prognosis provides additional prognostic value beyond that of CA125. We propose that SPRR2A could potentially complement CA125, enhancing diagnostic accuracy and offering a more comprehensive assessment of EC.

Regarding pre-analytical variables, factors such as sample storage and RNA quality are crucial for the reproducibility of SPRR2A measurement. RNA degradation can occur if samples are not stored at - 80°C or are subjected to repeated freeze-thaw cycles, which may compromise the accuracy of SPRR2A expression measurements. Furthermore, RNA purity and integrity are essential for reliable RT-qPCR analysis. To ensure reproducibility, we emphasise the importance of standardised protocols for sample collection, processing and storage, as well as the use of high-quality reagents and equipment. Validation studies in independent cohorts will also be necessary to confirm the generalisability of SPRR2A as a biomarker across different clinical settings.

Despite the significant findings of this study, several limitations should be acknowledged. First, the sample size of our experimental validation was relatively small, which may limit the statistical power and generalisability of our results to the broader population of EC patients. This calls for validation in larger independent cohorts to confirm the diagnostic and prognostic potential of SPRR2A. Second, our study primarily focused on the expression levels of SPRR2A and its correlation with clinical features. The specific biological functions and molecular mechanisms of SPRR2A in EC progression were not fully elucidated. Further functional studies, such as gene knockdown or overexpression experiments in vitro and in vivo, are required to explore how SPRR2A affects tumour cell growth, invasion and metastasis at the molecular level. Third, we relied on public datasets for data analysis. The heterogeneity of patient populations and differences in treatment protocols across studies may have introduced potential biases that are difficult to control for. Additionally, the clinical follow-up time and patient outcomes data from the TCGA database may not be comprehensive enough to fully establish the role of SPRR2A as an independent prognostic biomarker. Finally, our study did not explore the potential therapeutic applications of SPRR2A or its utility as a drug target in EC. Future research could investigate whether modulating SPRR2A expression could offer therapeutic benefits, thereby providing more comprehensive insights into its clinical applications.

Conclusions

This study combined bioinformatics analysis and RT-qPCR experiments to confirm that SPRR2A expression is significantly upregulated in EC compared with normal endometrium. High SPRR2A expression was linked to poor prognosis in EC patients, and its expression levels differed across histological grades. These findings indicate that SPRR2A has potential as a diagnostic and prognostic biomarker for EC. However, the correlation between SPRR2A expression and EC prognosis requires validation through clinical case collection and survival follow-up. The specific molecular mechanisms and related pathways also warrant further investigation.