Clinical utility of OGN in pan-cancer: diagnostic biomarker and immune microenvironment regulator

Introduction

Cancer poses a serious threat to global public health. There were approximately 19.7 million new cancer cases and 9.7 million cancer-related deaths worldwide in 2022 (1). China also revealed a heavy cancer burden domestically, with 4.825 million new cases and 2.574 million deaths in 2022 (2). Urgent exploration of cancer mechanisms and development of effective prevention strategies are therefore essential.

Small leucinerich proteoglycans (SLRPs), a class of important glycoprotein rich in isoleucine, are synthesized by cells and released into the extracellular matrix (ECM) (3). Osteoglycin (OGN) is a class III member of the SLRP family, and the OGN protein contains a cysteine rich region and multiple glycosylation sites. It consists of six leucine rich repeat (LRR) motifs, and the precursor protein undergoes varying degrees of glycosylation modification in the cytoplasm to play different roles (4).

OGN is expressed in both normal vascular stroma and perivascular fibroblasts, and is involved in angiogenesis, cell growth, cell proliferation and migration processes (5). OGN can reverse tumor epithelial mesenchymal transition (EMT), thereby reducing cell invasion and metastasis (6). Several researches have summarized that OGN expression is significantly down regulated in various tumor tissues compared with adjacent cancer tissues, such as digestive tract tumors, reproductive system tumors, head and neck cancer (7-11). OGN is reported highly related to the overall survival (OS) and prognosis of tumor patients. In colorectal cancer (CRC), OGN reduces vascular endothelial growth factor (VEGF) signaling and vascular abnormalities, which helps improve vascular function and immune cell delivery, thereby enhancing the ability of CD8⁺ T cells to enter tumors (12). Studies on myocardial and cardiovascular diseases also suggest that OGN interacts with the inflammatory pathway of neutrophils/macrophages, which provides a deducible mechanism basis for OGN to regulate the recruitment, polarization and antigen presentation of myeloid cells in tumors (although direct verification in tumor scenarios is required) (5). These researches illuminate that OGN may affect tumor growth by altering the tumor immune microenvironment.

Existing evidence suggests that OGN has biological functions in three dimensions: matrix remodeling, angiogenesis regulation, and immune microenvironment shaping. However, its expression patterns, pathogenic pathway coupling (such as EGFR/PI3K-AKT, HIF-1 α/VEGF, TGF-β, etc.) and systemic relationship with clinical outcomes or immune responses in different tumor lineages have not been fully elucidated. Given the complexity of OGN multi isomer or multi site glycosylation and tissue-specific expression, relying solely on single cancer studies is unable to capture their cross tumor commonalities and differences. Pan-cancer integrated analysis is expected to (I) depict the expression profile of OGN in multiple cancer types; (II) analyze its association with tumor immune features; (III) identify signaling pathways and therapeutic response markers that co-occur or interact with OGN, providing evidence for the translational application of OGN as a prognostic or predictive biomarker and even intervention target. We present this article in accordance with the TRIPOD reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1499/rc).

Methods

Clinical and expression data downloading and analysis

Normal tissue expression profiling

Transcriptomic data for OGN in healthy human tissues were retrieved from the Genotype-Tissue Expression (GTEx) database (v8) (13,14). Tissue-specific baseline expression was visualized using boxplots generated with the R packages dplyr (v1.0.7) and ggplot2 (v3.3.5) (15-17). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Cancer cell line analysis

OGN expression patterns across tumor cell lines were extracted from the Cancer Cell Line Encyclopedia (CCLE) (18). Differential expression was analyzed using ggpubr (v0.4.0) and ggplot2 (19).

Pan-cancer analysis

The Cancer Genome Atlas (TCGA) datasets were queried via TCGA biolinks (v2.18.0) to obtain tumor/normal paired expression profiles [log2(FPKM+1) transformed] and clinical metadata (20-22). And expression box diagram of OGN across pan-cancer were plotted by R/Bioconductor packages ggsignif, ggplot2, and RColorBrewer (17,23,24). Comparative analysis employed Welch’s t-test with significance thresholds (P<0.05, FDR-corrected). Protein-level validation utilized immunohistochemistry data from the Human Protein Atlas (HPA; v21.0) using anti-OGN (HPA013132, Sigma-Aldrich; 1:500 dilution) (25,26).

Differential expression and pathway analysis

Multi-group DEG screening (|log2FC|>1, adj.P<0.05) was performed using DESeq2 (v1.30.1) with batch correction via limma (27-30). GSEA (v4.1.0) examined enriched pathways in MSigDB collections (C2/C5) with 1,000 permutations (31-33).

Survival analysis of OGN across pan-cancer

To evaluate the prognostic significance of OGN expression across cancers, we conducted a comprehensive pan-cancer survival analysis. Patients from each cancer type in TCGA were stratified into “OGN-high” and “OGN-low” groups based on the “optimal cut-off value” determined by the “surv_cutpoint” function in the R package “survminer” (version 0.4.9). This cut-off maximizes the survival difference between the two groups based on OGN expression.

We assessed four distinct survival endpoints to capture different aspects of patient prognosis: OS, defined as the time from diagnosis to death from any cause; disease-specific survival (DSS), defined as the time to death due to the specific cancer; disease-free interval (DFI), defined as the time to the first recurrence of the disease or new tumor; and progression-free interval (PFI), defined as the time to first disease progression or death.

Survival curves for each endpoint were generated using the Kaplan-Meier method and compared between OGN-high and OGN-low groups using the Log-rank test, implemented in the R package “survival” (version 3.2-10). A P value of less than 0.05 was considered statistically significant. Additionally, a pooled pan-cancer survival analysis was performed by combining clinical data from all TCGA cancer types to evaluate the overall effect of OGN (34-36).

Diagnosis value analysis of OGN across pan-cancer

Biomarker performance

ROC curve analysis (pROC v1.17.0.1) stratified diagnostic performance by AUC: non-informative (0.5–0.6), moderate (>0.6–0.75), or high discriminative power (>0.75) (37-39).

Molecular correlates

Programmed cell death ligand-1 (PD-L1), tumor mutation burden (TMB) and microsatellite instability (MSI) associations were tested via Pearson correlation (|r|>0.3, P<0.01) using Hmisc (v4.5-0). Radar plots (ggsci v2.9) visualized MSI relationships (40-43).

Immunity estimation of OGN across pan-cancer

The association between OGN and the tumor immune microenvironment was investigated from three perspectives: overall immune score, immune cell infiltration, and correlation with immune-related genes.

First, the ESTIMATE algorithm (R package “estimate”, version 1.0.13) was applied to the gene expression profile of each tumor sample to calculate the StromalScore, ImmuneScore, and ESTIMATEScore, which infer the presence of stromal cells, immune cells, and combined tumor purity, respectively (44). The differences in these scores between OGN-high and OGN-low groups (defined as in section 2.2) across cancer types were visualized using violin plots generated with “ggplot2” (version 3.3.5) (45).

Second, the relative infiltration levels of 22 distinct immune cell types were quantified using CIBERSORTx [accessed (Date of Access)] (46,47). The analysis was run using the LM22 signature matrix and 1,000 permutations for statistical accuracy. Only samples with a CIBERSORTx output P value <0.05 were retained for subsequent analysis to ensure reliable deconvolution results.

Third, we analyzed the correlation between OGN expression and immune-related genes. A comprehensive list of immune-related genes was downloaded from the ImmPort Portal database (https://www.immport.org/home). This list encompassed genes involved in key immune processes, including the T-cell receptor (TCR) signaling pathway, B-cell receptor (BCR) signaling pathway, natural killer cell cytotoxicity, chemokines, and chemokine receptors. The strength of the linear association between OGN and each immune-related gene across all TCGA samples was assessed using Pearson correlation analysis (48,49). Correlation coefficients and corresponding p-value were calculated using the R package “psych” (version 2.1.9). The results were visualized as a correlation heatmap using the R package “pheatmap” (version 1.0.12), with the input data reshaped via “reshape2” (version 1.4.4) (50-52).

Cell signaling score evaluation of OGN across pan-cancer

PROGENy (v1.14.0) quantified 14 oncogenic pathway activities (e.g., WNT, TGF-β). Spearman correlations against OGN expression were plotted as a heatmap (reshape2 v1.4.4) (53).

Statistical analysis

All statistical analysis were performed by R language software (version 4.3.2). We use Mann-Whitney U test to estimate the statistical significance if the data is not normally distributed, otherwise, we performed Student’s t-test. We used Pearson correlation analysis to evaluate the expression correlation between proteins, transcripts, TMB, PD-L1, MSI, cell signaling score and immune cells score. All results were considered significant at a P value of <0.05.

Results

Expression of OGN in various tissues and cell lines

To assess the basic expression level of OGN, we analyzed data from normal human tissues and different tumor cell lines from GTEx and CCLE databases respectively. Differential expression level of OGN across various normal human tissues, revealing a widespread distribution with distinct expression levels. Tissues such as the ovary, breast, colon and lung exhibit notable OGN expression, while others show relatively lower levels (Figure 1A). To further explore the expression levels of OGN in different tumor cell lines, we quantified the expression of OGN by Transcripts Per Kilobase Million (TPM) to standardized measure for comparison. Unlike normal tissues, the vast majority of tumor cells have low expression of OGN (Figure 1B). These results highlight the differential basal expression of OGN in normal physiological contexts and cancerous states, laying a foundation for further exploration of its role in tumorigenesis.

Figure 1 Basal expression of OGN in normal tissue and cancer cell lines was accessed through GTEx and CCLE, respectively. (A) Expression of OGN in normal human tissues was displayed with boxplot through GTEx. (B) Basal level of OGN was accessed through CCLE with boxplot. CCLE, Cancer Cell Line Encyclopedia; GTEx, Genotype Tissue Expression; OGN, osteoglycin; TPM, transcripts per kilobase million.

Differences in transcription and expression levels of OGN gene

We analyzed the expression of the OGN gene in pan-cancer using the TCGA dataset. As shown in Figure 2A, we used box plots to present the different expression of OGN between normal and tumor tissues, revealing significant differences in OGN expression across multiple cancer types. Further studies on paired cancer samples (Figure 2B) compared OGN expression between non-tumor and tumor tissues across various cancer types. The expression level of OGN in tumor tissue is statically lower than that in paired normal tissue were observed in most cancer types, including colon adenocarcinoma (COAD), breast cancer (BRCA), esophageal carcinoma (ESCA), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), prostatic adenocarcinoma (PRAD), stomach adenocarcinoma (STAD) and thyroid cancer (THCA). Regarding the association between OGN expression and tumor stages, we compared OGN expression in non-tumor tissues and different tumor stages (I–IV). The results revealed that the expression level of OGN in tumor tissues was significantly lower than that in normal tissues, but there was no difference in OGN expression among different tumor stages (Figure 2C). We further illustrated the differences in OGN expression at different tumor stages through box plots (Figure 2D). In conclusion, in the TCGA dataset, OGN expression decreases in most tumor tissues, but the OGN expression had no relationship with tumor stages. In addition to mRNA data from TCGA, we analyzed the immunohistochemical microarrays from the HPA database to analyze the OGN expression level across different tumors. As showed in Figure 3, the OGN expression varies greatly among different tumors. Moreover, the expression level of OGN protein in tumor tissue is generally lower than that in normal tissue, and this difference is more obvious in breast cancer, CRC, ovarian cancer, prostate cancer, and urothelial cancer. Furthermore, these findings provide a comprehensive landscape of OGN protein expression in human malignancies, laying the foundation for further functional investigations into its role in cancer pathogenesis.

Figure 2 Expression of OGN across pan-cancer was performed in TCGA dataset. (A) Expression of OGN in nontumor and tumor across pan-cancer was shown with boxplot. (B) Expression of OGN in nontumor and tumor was performed in paired cancer samples. (C) Expression comparison of OGN in nontumor and different tumor stage (stage I–IV) was showed with colorful box diagram. (D) Expression of OGN in different patients’ stage was displayed with boxplot. *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001; ns, not statistically significant. OGN, osteoglycin; TCGA, The Cancer Genome Atlas.

Figure 3 Protein level of OGN across pan-cancer was accessed with immunohistochemistry from HPA. These pan-cancer types included (A) breast cancer (images available from https://images.proteinatlas.org/13132/29960_B_3_4.jpg; https://images.proteinatlas.org/13132/29957_A_5_4.jpg; https://images.proteinatlas.org/13132/29957_A_6_4.jpg; https://images.proteinatlas.org/13132/29957_A_6_5.jpg) (B) cervical cancer (images available from https://images.proteinatlas.org/13132/29960_B_7_3.jpg; https://images.proteinatlas.org/13132/29960_B_9_3.jpg; https://images.proteinatlas.org/13132/29957_B_4_3.jpg; https://images.proteinatlas.org/13132/29957_B_6_5.jpg; https://images.proteinatlas.org/13132/29957_B_5_3.jpg); (C) colorectal cancer (images available from https://images.proteinatlas.org/13132/29960_A_7_3.jpg; https://images.proteinatlas.org/13132/29960_A_9_3.jpg; https://images.proteinatlas.org/13132/29960_A_8_3.jpg; https://images.proteinatlas.org/13132/29957_A_1_2.jpg; https://images.proteinatlas.org/13132/29957_A_1_5.jpg; https://images.proteinatlas.org/13132/29957_A_3_3.jpg); (D) lung cancer (images available from https://images.proteinatlas.org/13132/29960_A_3_4.jpg; https://images.proteinatlas.org/13132/29958_B_1_7.jpg; https://images.proteinatlas.org/13132/29958_B_3_1.jpg; https://images.proteinatlas.org/13132/29958_B_1_2.jpg); (E) ovarian cancer (images available from https://images.proteinatlas.org/13132/29960_A_4_7.jpg; https://images.proteinatlas.org/13132/29960_A_6_7.jpg; https://images.proteinatlas.org/13132/29960_A_5_7.jpg; https://images.proteinatlas.org/13132/29957_B_1_2.jpg; https://images.proteinatlas.org/13132/29957_B_1_3.jpg; https://images.proteinatlas.org/13132/29957_B_1_8.jpg); (F) prostate cancer (images available from https://images.proteinatlas.org/13132/29960_A_3_5.jpg; https://images.proteinatlas.org/13132/29960_A_2_5.jpg; https://images.proteinatlas.org/13132/29960_A_1_5.jpg; https://images.proteinatlas.org/13132/29957_A_7_5.jpg; https://images.proteinatlas.org/13132/29957_A_8_4.jpg; https://images.proteinatlas.org/13132/29957_A_9_5.jpg); (G) stomach cancer (images available from https://images.proteinatlas.org/13132/29960_A_6_1.jpg; https://images.proteinatlas.org/13132/29960_A_5_1.jpg; https://images.proteinatlas.org/13132/29960_A_4_1.jpg; https://images.proteinatlas.org/13132/29959_B_2_3.jpg; https://images.proteinatlas.org/13132/29959_B_1_8.jpg; https://images.proteinatlas.org/13132/29959_B_1_6.jpg); (H) testis cancer (images available from https://images.proteinatlas.org/13132/29960_A_6_6.jpg; https://images.proteinatlas.org/13132/29960_A_5_6.jpg; https://images.proteinatlas.org/13132/29960_A_4_6.jpg; https://images.proteinatlas.org/13132/29959_A_2_3.jpg; https://images.proteinatlas.org/13132/29959_A_2_5.jpg; https://images.proteinatlas.org/13132/29959_A_2_8.jpg); (I) skin cancer (images available from https://images.proteinatlas.org/13132/29960_B_8_1.jpg; https://images.proteinatlas.org/13132/29960_B_7_1.jpg; https://images.proteinatlas.org/13132/29958_B_8_7.jpg; https://images.proteinatlas.org/13132/29958_B_8_5.jpg; https://images.proteinatlas.org/13132/29958_B_9_3.jpg); (J) urothelial cancer (images available from https://images.proteinatlas.org/13132/29960_A_6_5.jpg; https://images.proteinatlas.org/13132/29960_A_4_5.jpg; https://images.proteinatlas.org/13132/29959_A_4_2.jpg; https://images.proteinatlas.org/13132/29959_A_4_6.jpg; https://images.proteinatlas.org/13132/29959_A_5_8.jpg). Patient id, Patient identifier from HPA. HPA013132, antibody id of OGN in HPA. Scale bar, 200 and 100 μm. HPA, Human Protein Atlas; OGN, osteoglycin.

Diagnostic significance of OGN across pan-cancer

To evaluate the diagnostic potential of OGN in pan-cancer, we analyzed its performance with ROC curves, correlation with MSI and association with PD-L1 expression. As shown in Figure 4A, the area under the ROC curve (AUC) was calculated to assess OGN’s diagnostic utility across 13 cancer types (AUC: 0.703–0.990), 69.2% (9/13) of which had especially high diagnostic accuracy (AUC >0.900), including BRCA, bladder urothelial carcinoma (BLCA), COAD, kidney chromophobe cell carcinoma (KICH), kidney renal papillary cell carcinoma (KIRP), LUAD, LUSC, rectum adenocarcinoma (READ) and THCA. An AUC of 0.966 was observed in COAD, indicating a perfect diagnostic value (AUC: 0.75–1.0), and THCA exhibited an AUC of 0.920, also falling into the perfect diagnostic category. Cancers with AUC values in the medium diagnostic range (AUC: 0.6–0.75) were not explicitly listed here, but values below 0.5 (no diagnostic value) were absent in the presented data, suggesting OGN may have diagnostic potential in specific cancer types.

Figure 4 Diagnostic significance evaluation of OGN across pan-cancer was performed with AUC of ROC curves, correlation with MSI, and with PD-L1. (A) AUC of ROC curves was plotted to estimate the diagnosis value of OGN <0.6, no diagnostic value; 0.6–0.75, medium diagnostic value; >0.75, perfect diagnostic value). (B) Radar plot displayed the correlation significant between OGN and MSI. (C) Pearson correlation analysis between OGN and PD-L1 was showed with scatter plot in MESO, SARC, STAD, TGCT, and THYM. *, P<0.05; **, P<0.01; ***, P<0.001. AUC, area under the curve; MSI, microsatellite instability; OGN, osteoglycin; PD-L1, programmed death-ligand 1; R, Pearson correlation coefficient; ROC, receiver operating characteristic.

Correlation of OGN expression with MSI, PD-L1 expression and TMB

MSI, TMB, and PD-L1 are important clinical indicators for solid tumor immunotherapy, and show close relationship to the efficacy of immune checkpoint inhibitors. Therefore, we first used radar plot analysis (Figure 4B) to analysis the correlations between OGN and MSI across several cancer types. Notably, positive associations were observed in THCA, uterine corpus endometrial cancer (UCEC), glioblastoma multiforme (GBM) and KIRP, indicating that OGN expression may be linked to MSI status in these malignancies. The strength and direction of these correlations provide insights into potential mechanisms underlying OGN-mediated tumorigenesis related to genomic instability. Pearson correlation analysis (Figure 4C) examined the relationship between OGN and PD-L1 expression in mesothelioma (MESO), sarcoma (SARC), STAD, testicular germ cell tumors (TGCT), and thymoma (THYM). A slight association was identified in MESO with an R value of −0.39 (P=0.0002), while inverse relationships were observed between OGN and PD-L1 in other tumors. To further assess the correlation of OGN with TMB. As shown in Figure 5, significant associations were observed between OGN expression and TMB in several cancer types. In LUAD, a notable correlation was found (P=4.3e−09), indicating that OGN may play a crucial role in the diagnostic evaluation of LUAD in relation to TMB. For cervical squamous cell carcinoma (CESC), BRCA and head and neck squamous cell carcinoma (HNSC), although the specific correlation values were partially obscured, the presented data suggest potential links between OGN and TMB in these malignancies. The fragments per kilobase of exon model per million mapped fragments (FPKM) was used to quantify OGN expression, providing a standardized measure for cross-cancer comparison. These findings highlight the potential of OGN as a diagnostic biomarker in pan-cancer, particularly in relation to TMB, and offer new insights into the molecular mechanisms and clinical evaluation of various cancers. However, further validation in independent cohorts and functional studies are needed to confirm these associations and elucidate the underlying biological pathways. Collectively, these results demonstrate that OGN exhibits cancer-type specific diagnostic value, correlates with TMB and MSI status in select malignancies, and may interact with PD-L1 expression in certain tumor contexts, providing a foundation for further exploration of OGN as a diagnostic biomarker or therapeutic target.

Figure 5 Diagnostic significance evaluation of OGN across pan-cancer was also accessed by its correlation with TMB. FPKM, fragments per kilobase of exon model per million mapped fragments; OGN, osteoglycin; TMB, tumor mutation burden.

Prognostic significance of OGN in pan-cancer

To evaluate the clinical prognostic value of OGN at the pan-cancer level, we performed Log-Rank test survival analysis using data from TCGA datasets. As shown in Figure 6A, expression of OGN was significantly associated with OS in several cancer types. For example, in LIHC, LUAD, MESO and thymoma (THYM), survival curves indicated distinct separation between patients with high and low OGN expression. Specifically, certain cancer types like THYM showed an absolute benefit in the OS of OGN high expression group, while others such as CESC, there is no significant difference between different OGN expression levels groups. Figure 6B illustrates the association between OGN expression and DSS. LIHC, LUAD, SARC, and THYM demonstrate significant differences in DSS based on OGN levels. For instance, high OGN expression in LUAD was linked to better DSS (P=1.7e−03), while the trend in liver hepatocellular carcinoma (LIHC) showed a contrasting pattern, highlighting cancer-type specificity. We next assessed the correlation between OGN expression and disease recurrence or progression (Figure 7). In Figure 7A, low OGN expression was associated with reduced Disease Free Internal (DFI) in cholangiocarcinoma (CHOL, P=1.2e−02), glioblastoma and low-grade glioma (GBMLGG, P=1.6e−03), KICH (P=5.3e−04), LIHC (P=1.2e−04), PRAD (P=3.5e−04) and THYM (P=7.6e−04). For example, CHOL patients with low OGN expression showed a shorter time to disease recurrence compared to those with high expression. Figure 7B depicts the impact of OGN on PFI across pan-cancer. Notable associations were observed in CHOL, LUSC, PRAD and UCEC, that high OGN expression was linked to longer PFI (e.g., PRAD, P=1.3e−03). These results indicate that OGN may serve as a potential biomarker for predicting disease recurrence and progression in specific cancer types. These survival analyses demonstrate that OGN expression exhibits pan-cancer prognostic significance, with distinct associations with OS, DSS, DFI, and PFI across different tumor types. These results indicate the potential role of OGN as a cancer-specific prognostic marker and warrant further investigation into its mechanistic role in tumor progression and patient outcomes.

Figure 6 Pan-cancer prognostic significance estimation of OGN was also accessed by OS and DSS with Log-Rank test survival analysis. (A) Effect of OGN on patients’ OS was displayed with survival-curves. (B) Effect of OGN on patients’ DSS was also displayed with survival-curves. PANCAN, overall cancer patients from TCGA. Log-rank, log-rank test was used to evaluate the prognostic significance of OGN. DSS, disease-specific survival; OGN, osteoglycin; OS, overall survival; TCGA, The Cancer Genome Atlas.

Figure 7 Pan-cancer prognostic significance estimation of OGN was also accessed by DFI and PFI with Log-rank test survival analysis. (A) Effect of OGN on patients’ DFI was displayed with survival-curves. (B) Effect of OGN on patients’ PFI was also displayed. PANCAN, overall cancer patients from TCGA. Log-rank, log-rank test was used to evaluate the prognostic significance of OGN. DFI, disease-free interval; OGN, osteoglycin; PFI, progression-free interval; TCGA, The Cancer Genome Atlas.

Effect of OGN expression on immune microenvironment across pan-cancer

Since OGN is an important component of the extracellular matrix and has been reported to be involved in regulating the tumor immune microenvironment, we here investigated its correlation with immune cells across pan-cancer through a heatmap (Figure 8). The Pearson correlation test was conducted to assess statistical significance. Different colors represented varying Pearson correlation coefficients, with warmer tones indicating higher coefficients. The results indicated that there is a tumor heterogeneity in immune cell types associated with OGN expression. In COAD, macrophage M1 and M2 were positively associated with OGN expression, while in MESO, natural killer (NK) cells had a prominent correlation with OGN. T-cell-gamma delta subsets were associated with OGN expression in uterine carcinosarcoma (UCS). We further compared immune scores between low and high OGN expression groups across pan-cancer using violin plots. Statistical analysis via Student’s t-test revealed significant differences in immune scores between high and low OGN group across various tumors including COAD, BRCA, CHOL, diffuse large B-cell lymphoma (DLBC), ESCA, HNSC, GBM, LUAD, LUSC, prostate adenocarcinoma PRAD and READ, indicating that OGN expression is associated with altered immune cell infiltration or activity in these cancer types (Figure 9). Furthermore, we also performed expression correlation analyses to explore how OGN influences immune-related gene pathways. Pearson correlation tests showed that OGN expression was significantly associated with genes in the T-cell receptor (TCR) signaling pathway (Figure 10A), B-cell receptor (BCR) signaling pathway (Figure 10B), and NK cell cytotoxicity pathway (Figure 10C) across pan-cancer. Further analysis also revealed similar correlations for chemokines (Figure 11A) and chemokine receptors (Figure 11B). These results suggest that OGN may modulate immune cell recruitment and function by regulating chemokine signaling networks.

Figure 8 Pan-cancer correlation analysis between expression of OGN and immune cells was displayed with heatmap. Pearson correlation test was performed to estimate this statistical significance. Different color displayed different Pearson correlation coefficient, and the warmer tones represent the higher Pearson correlation coefficient. *, P<0.05; **, P<0.01. OGN, osteoglycin.

Figure 9 Effect of different OGN level on immune score across pan-cancer were accessed with violin plots. Low, lower expression group of OGN; high, higher expression group of OGN. Student’s t-test was used to evaluate this statistical significance. OGN, osteoglycin.

Figure 10 Effect of different OGN level on expression of immune related genes across pan-cancer were accessed with expression correlation analysis. (A) Effect of OGN on TCR signaling pathway related genes’ expression was showed. (B) Effect of OGN on BCR signaling pathway related genes’ expression was showed. (C) Effect of OGN on natural killer cell cytotoxicity related genes’ expression was showed. Pearson correlation test was performed to estimate this statistical significance. Different color displayed different Pearson correlation coefficient, and the warmer tones represent the higher Pearson correlation coefficient. *, P<0.05; **, P<0.01. OGN, osteoglycin.

Figure 11 Effect of different OGN level on expression of immune related genes across pan-cancer were accessed with expression correlation analysis. (A) Effect of OGN on chemokines related genes’ expression was showed. (B) Effect of OGN on chemokine receptors related genes’ expression was showed. Pearson correlation test was performed to estimate this statistical significance. Different color displayed different Pearson correlation coefficient, and the warmer tones represent the higher Pearson correlation coefficient. *, P<0.05; **, P<0.01. OGN, osteoglycin.

Functional pathway analysis of OGN in pan-cancer

Extensive research has shown that surface OGN is closely related to the transformation of tumor epithelium to mesenchyma (6,54), and other studies suggest that OGN regulates glucose metabolism (55). Moreover, OGN has been proved to involve in tumorigenesis, tissue repair and scar formation (4). However, to date, there is no research on functional analysis of the OGN gene across pan-cancer. To characterize the molecular pathways associated with OGN, we performed correlation analysis between OGN expression and cancer-related pathway scores. Pearson correlation heatmaps showed distinct association patterns, with warmer tones indicating stronger positive correlations. As showed in Figure 12A, OGN expression was significantly linked to JAK-STAT, hypoxia, NFκB, MAPK, PI3K, EGFR and WNT pathway activity in specific cancer types. Additionally, a grouped volcano plot (Figure 12B) highlighted differentially expressed genes (DEGs) between low and high OGN groups, with fold change (FC) and P value thresholds, suggesting that OGN modulates broad gene expression programs involved in cancer progression and immune regulation. These findings demonstrate that OGN expression is associated with immune cell-mediated responses and dysregulated signaling pathways in pan-cancer, providing insights into its potential role as an immunomodulatory and oncogenic factor. Furthermore, we conducted functional enrichment analysis via gene set enrichment analysis (GSEA). The results revealed significant biological pathways associated with OGN-related DEGs. The top 10 enriched hallmarks (Figure 13A) highlighted key processes, including EMT, a hallmark of cancer progression linked to metastasis and therapeutic resistance. This strong enrichment in EMT is consistent with previous reports that OGN can reverse EMT in cancers like CRC, and our findings suggest this role is conserved across multiple cancer types. Other enriched pathways included angiogenesis-related processes and tumor microenvironment remodeling, suggesting OGN may influence tumor cell invasion and vascularization. The significant enrichment of the PI3K/AKT/mTOR and JAK-STAT signaling pathways aligns with reported OGN functions in inhibiting cell proliferation via the PI3K/Akt/mTOR pathway in breast cancer and its potential immunomodulatory roles, respectively, indicating these may be key mechanistic pathways through which OGN exerts its pan-cancer effects. A statistical plot (Figure 13B) visualized the enrichment of these hallmarks across cancer types, using an advanced Venn diagram. Black dots denoted the presence of enriched hallmarks in specific cancers (e.g., PAAD, KIRC), indicating OGN-DEGs were significantly associated with conserved biological programs across tumor types. For instance, EMT and angiogenesis-related hallmarks showed consistent enrichment in multiple cancers, implying OGN may play a pan-cancer role in promoting tumor progression via these pathways. These findings suggest OGN modulates critical tumorigenic processes, providing mechanistic insights into its potential as a therapeutic target or biomarker in cancer.

Figure 12 Function analysis of OGN was performed across pan-cancer. (A) Effect of different OGN level on cancer related pathways score were accessed with correlation heatmap. Cancer related pathways included Androgen (line 1), EGFR (line 2), Estrogen (line 3), Hypoxia (line 4), JAK-STAT (line 5), MAPK (line 6), NF-κB (line 7), p53 (line 8), PI3K (line 9), TGF-β (line 10), TNF-α (line 11), Trail (line 12), VEGF (line 13), and WNT pathway (line 14). Pearson correlation test was performed to estimate this statistical significance. Different color displayed different Pearson correlation coefficient, and the warmer tones represent the higher Pearson correlation coefficient. (B) Effect of different OGN level on gene expression profile was accessed and displayed with grouped volcano plot. *, P<0.05; **, P<0.01. EGFR, epidermal growth factor receptor; FC, fold change; NF-κB, nuclear factor kappaB; OGN, osteoglycin; TGF-β, transforming growth factor-beta; TNF-α, tumor necrosis factor alpha; VEGF, vascular endothelial growth factor.

Figure 13 Function analysis of OGN was predicted by its DEGs GSEA enrichment. (A) Top ten GSEA enrichment hallmarks of OGN’ DEGs was displayed. (B) Statistical graph of GSEA enrichment hallmarks of OGN’ DEGs was presented with present with advanced Venn diagram. The small black dots on different hallmarks represent whether they were enriched in this cancer type. DEGs, differentially expressed genes; GSEA, gene set enrichment analysis; OGN, osteoglycin.

Discussion

Cancer is a progressive malignant disease worldwide, with an increasing incidence, has become a major public health threat. OGN, a member of the small leucine-rich proteoglycan (SLRP) family, plays critical roles in extracellular matrix organization, cell signaling, and tissue homeostasis (4). Initially identified for its osteogenic properties, OGN exists in various tissues (4,56) and the bloodstream (57). Moreover, OGN has emerged as a multifaceted regulator in cancer biology. Dysregulated OGN expression has been reported in various malignancies, including CRC (58), gastric cancer (59), breast cancer (9) and thyroid carcinoma (10), in which it is reported as inversely correlated with tumor progression and patient survival. For instance, in CRC, OGN inhibits the progression of CRC by suppressing the EGFR/AKT/ZEB-1 pathway (12). Similarly, in thyroid cancer, OGN expression declines progressively with tumor malignancy, suggesting its tumor-suppressive role (10). Despite these insights, the pan-cancer landscape of OGN remains poorly characterized. Previous studies have focused on individual cancer types, lacking systematic analysis across diverse malignancies. In this study, we found that OGN is not only expressed differently in various tissues, but also has differential expression in different cell lines. Our analysis results show that OGN expression levels are generally low in various tumor cell lines, indicating that OGN may play a negative regulatory role in the occurrence and development of tumors.

The different expression of OGN between normal tissues and cancer is a remarkable feature. We used data from multiple databases, including GTEx, CCLE, TCGA, and HPA, consistently demonstrate that OGN is widely expressed in normal human tissues like the breast, liver, kidney, and muscle, while its expression is predominantly down regulated in a majority of tumor cells. This indicates that OGN is likely to play a negative regulatory role in the occurrence and development of tumors. Jiang et al. reported that over expression of OGN can significantly inhibit the vitality, DNA synthesis, and cell invasion of ovarian cancer cells, alter EMT markers, and exert inhibitory effects on tumor growth (56). Similarly, our study showed that OGN expression level in ovarian cancer tumor tissue is statistically lower than that in normal tissue. However, it’s up regulated in certain cancers like THCA, may reflect distinct tumorigenic pathways. This phenomenon has also been validated in pituitary tumors. Pit-1 can activate the expression of OGN by binding to the Pit-1 response element in the OGN promoter (60). Our data show OGN is also highly expressed in neurological tumors, such as osteosarcoma, diffuse glioma and high grade spindle cell sarcoma. In addition, we found that OGN is lowly expressed in the bloodstream, while T-lymphoblast leukemia/lymphoma expresses OGN highly. Based on this, we speculate that the way OGN regulates tumors in the blood system or intracranial tumors may play different roles through different pathways different from other solid tumors. These evidences indicate OGN’s cancer-type-specific expression patterns and functional diversity.

Previous studies have explored the relationship between OGN and tumor prognosis. Liu et al. found that the expression level of OGN in breast cancer is lower than that in normal tissues, and the blood OGN level of patients who reach pathological complete response (pCR) after neoadjuvant therapy is lower than that of patients with residual lesions, that showed OGN be a predictive indicator of pCR after neoadjuvant therapy for breast cancer (61). Dings et al. proposed that the serum OGN level is a prognostic indicator of good OS rate of pancreatic cancer, they concluded that serum OGN indicate that the abundance of specific subgroups of CAF is of great significance for the prognosis of PDAC (62). OGN has also been proved to be one of the prognostic indicators for urothelial carcinoma and breast cancer (61,63). Here, we first evaluate the prognostic value of OGN across pan-cancer, the results strongly support that OGN expression exhibits pan-cancer prognostic significance, with distinct positive associations with OS, DSS, DFI, and PFI across different tumor types. These prove that there is a clear connection between OGN expression and disease recurrence or progression. Furthermore, we assess OGN’s diagnostic utility across pan-cancer with ROC curves, and found that OGN have diagnostic potential in specific cancer types, which is in contrast to Mark P. G. Dings’ conclusion that serum OGN is not a biomarker for detecting PDAC (62). These findings imply that OGN could potentially serve as a therapeutic marker for specific tumor types.

Immunotherapy has been widely used in clinical practice in the past decade, and the exploration of immunotherapy related biomarkers remains a current research hotspot. MSI, TMB, and PD-L1 expression levels are important reference indicators for clinical guidance of immunotherapy. Given that OGN plays an important role in forming the cancer microenvironment, we further investigated the association between OGN and key cancer hallmarks, including MSI, TMB and PD-L1. Our research demonstrated that OGN showed a relationship with MSI, TMB and PD-L1 in kinds of caner types, indicating that OGN is likely to be involved in tumor immune regulation. Hu et al. concluded that over expression of OGN in tumor cells leads to recruitment of more CD8⁺ T cells (12). However, our data show that OGN expression in CRC is statistically associated with M2 macrophage cell rather than CD8⁺ T Cells. OGN is also reported be positive with M2 macrophage and CD8⁺ T Cells in gastric cancer (64), while our study shows that OGN is positively express with M2 macrophage and CD4⁺ T Cells. Additionally, OGN is necessary for the activation of pancreatic cancer fibroblasts and marks a unique group of inflammatory fibroblasts (iCAF). Serum OGN indicates the abundance of specific subsets of CAF is of great significance for the prognosis of PDAC (62). Consistent with these literature reports, our research results indicate that OGN is associated with immune cell infiltration in multiple tumor, although there are differences in immune cell subsets between different research results, which may be due to sample size induced analysis segregation. Our findings show that immune cell infiltration is associated with immune scores between different OGN expression groups across pan-cancer. The functional interplay between OGN and tumor immune microenvironment signaling networks requires further mechanistic validation.

It is an undeniable fact that ECM is involved in various signal transductions (65). OGN, as an important structural component of ECM, is also involved in the complex network of cellular signal transduction. It has been shown to participate in a wide range of processes such as epithelial repair (66), fibrosis (67), cardiovascular events (68,69), glucose metabolism (55), and tumorigenesis. To date, the molecular mechanisms underlying OGN’s dual roles (tumor-suppressive vs. oncogenic) across various cancer types are unclear. Some explorations have been made. Studies by Cui et al. have indicated that the invasion and metastasis ability of Hca-F cells decreased after transfection with OGN plasmid, and a decrease in the ability of tumor cells to metastasize to surrounding lymph nodes was observed (70). Hu et al. demonstrated that OGN reduced Zeb-1 expression via EGFR/Akt pathway, resulting in reversing the epithelial to mesenchymal transition invasiveness in CRC (6). Consistently, we used different expression gene enrichment to analyze the function of OGN, and found that the EMT is the most prevent hallmarks across pan-cancer. Additionally, the role of microRNA on OGN is not yet unified. Jiang et al. concluded that MiR-1290 targets OGN and inhibits its expression, and OGN can partially reverse the effect of miR-1290 and inhibit tumor cell invasion (56). In contrast to this finding, Alvarez-Díaz et al. reported that Vitamin D inhibits OGN activity, while miR-22 restores OGN activity (71), Yang et al. proved that miR-5003-3p could up regulate the level of OGN. Transcription factor C/EBP (LIP) promotes cell death by inducing OGN expression, resulting in promoting cancer cell death (72). OGN inhibits cell proliferation and invasiveness in breast cancer via PI3K/Akt/mTOR Signaling Pathway (73). Furthermore, we further found OGN expression was significantly linked to JAK-STAT, hypoxia, NFκB, MAPK, PI3K, KRAS, EGFR and WNT pathway activity in specific cancer types. This proves that OGN, as an important hotspot molecule, is involved in numerous cellular signaling pathways.

Interestingly, OGN demonstrates opposite prognostic associations among distinct cancers. This phenomenon may reflect the dualistic nature of extracellular matrix regulators in different target organs, whose biological effects depend heavily on the tumor microenvironment and pathway context. It is well known that PD-L1 expression is often driven by stromal cells and tumor-associated macrophages (TAMs) rather than tumor cells themselves in CRC, leading to a mismatch between PD-L1 level and immune escape capacity. Similarly, OGN expression and its prognostic value may be influenced by non-tumor components such as fibroblasts, endothelial cells, and infiltrating immune cells within the tumor microenvironment. In stroma-rich or immune-active tumors, OGN stabilizes the ECM and enhances immune infiltration, supporting tumor suppression. In contrast, within fibrotic or immunosuppressive niches, OGN may facilitate fibroblast activation, angiogenesis, or immune exclusion through pathways such as TGF-β or WNT. Moreover, OGN’s extensive post-translational modifications produce structurally diverse isoforms with potentially distinct signaling functions. These findings suggest that OGN acts not as a fixed tumor suppressor but as a context-dependent modulator of tumor-stroma and immune interactions, explaining its divergent prognostic significance across cancer types.

Collectively, our results expand the current understanding of OGN from a tissue-specific regulator to a multi-dimensional factor that integrates survival analysis, immune phenotype, and cellular signaling pathways. Nevertheless, the observed correlations should be interpreted cautiously. Despite the comprehensive pan-cancer bioinformatic analyses performed in this study, several limitations should be acknowledged to provide a balanced interpretation of the findings. All analyses in the present work are based on retrospective mining of public datasets such as TCGA, GTEx, CCLE, and HPA. Although these repositories provide extensive multi-omics data, they are inherently limited by batch effects, incomplete clinical annotation, and technical differences in sequencing platforms. The expression profiles of OGN across cancers are inferred from bulk transcriptomic data, which cannot fully distinguish tumor cells from stromal or immune components. Consequently, part of the observed differential expression or immune correlations may reflect changes in the tumor microenvironment composition rather than cell-intrinsic regulation of OGN. In addition, survival analyses rely on pre-existing clinical metadata with non-uniform follow-up durations and treatment histories, which may introduce confounding bias. Functional validation through in vitro and in vivo studies, such as CRISPR-mediated OGN knockout, glycosylation-site mutagenesis, and immune co-culture assays, is required to delineate whether OGN acts as an active regulator or merely a passive indicator of the tumor microenvironmental state. Addressing these gaps could establish OGN as a novel detection biomarker and therapeutic target, particularly in cancers with unmet clinical needs.

Conclusions

Our integrated pan-cancer analysis highlights OGN as a context-dependent regulator that connects extracellular matrix remodeling with immune and angiogenic signaling. OGN expression is broadly down regulated in cancers and correlates with key biological hallmarks, including EMT suppression, vascular regulation, and JAK-STAT/PI3K signaling activity. These findings suggest that OGN may serve as a potential biomarker for prognosis and immune modulation across multiple cancer types.

However, given the retrospective and correlation-based nature of this study, the causal roles of OGN remain to be experimentally verified. Future research integrating transcriptomic, proteomic, and functional approaches is essential to determine whether OGN directly influences these pathways or reflects broader microenvironmental changes. A deeper mechanistic understanding of OGN may ultimately support its development as a diagnostic indicator or therapeutic target in precision oncology.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1499/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1499/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1499/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Han B, Zheng R, Zeng H, et al. Cancer incidence and mortality in China, 2022. J Natl Cancer Cent 2024;4:47-53. [Crossref] [PubMed]
3. Colineau L, Laabei M, Liu G, et al. Interaction of Streptococcus pyogenes with extracellular matrix components resulting in immunomodulation and bacterial eradication. Matrix Biol Plus 2020;6-7:100020. [Crossref] [PubMed]
4. Nulali J, Zhan M, Zhang K, et al. Osteoglycin: An ECM Factor Regulating Fibrosis and Tumorigenesis. Biomolecules 2022;12:1674. [Crossref] [PubMed]
5. Rienks M, Papageorgiou A, Wouters K, et al. A novel 72-kDa leukocyte-derived osteoglycin enhances the activation of toll-like receptor 4 and exacerbates cardiac inflammation during viral myocarditis. Cell Mol Life Sci 2017;74:1511-25. [Crossref] [PubMed]
6. Hu X, Li YQ, Li QG, et al. Osteoglycin (OGN) reverses epithelial to mesenchymal transition and invasiveness in colorectal cancer via EGFR/Akt pathway. J Exp Clin Cancer Res 2018;37:41. [Crossref] [PubMed]
7. Liu B, Hu Y, Wan L, et al. Proteomics analysis of cancer tissues identifies IGF2R as a potential therapeutic target in laryngeal carcinoma. Front Endocrinol (Lausanne) 2022;13:1031210. [Crossref] [PubMed]
8. Ren N, Liang B, Li Y. Identification of prognosis-related genes in the tumor microenvironment of stomach adenocarcinoma by TCGA and GEO datasets. Biosci Rep 2020;40:BSR20200980. [Crossref] [PubMed]
9. Röwer C, Ziems B, Radtke A, et al. Toponostics of invasive ductal breast carcinoma: combination of spatial protein expression imaging and quantitative proteome signature analysis. Int J Clin Exp Pathol 2011;4:454-67.
10. Sponziello M, Lavarone E, Pegolo E, et al. Molecular differences between human thyroid follicular adenoma and carcinoma revealed by analysis of a murine model of thyroid cancer. Endocrinology 2013;154:3043-53. [Crossref] [PubMed]
11. Xu S, Chen C, Liu H, et al. Osteoglycin Inhibits the Progression of Lung Adenocarcinoma by Modulating ICAM1-Mediated Cell Adhesion via the PI3K/AKT Pathway. Mol Carcinog 2025;64:1564-79. [Crossref] [PubMed]
12. Hu X, Li YQ, Li QG, et al. Osteoglycin-induced VEGF Inhibition Enhances T Lymphocytes Infiltrating in Colorectal Cancer. EBioMedicine 2018;34:35-45. [Crossref] [PubMed]
13. The Genotype-Tissue Expression (GTEx) project. Nat Genet 2013;45:580-5. [Crossref] [PubMed]
14. Gibson G. Human genetics. GTEx detects genetic effects. Science 2015;348:640-1. [Crossref] [PubMed]
15. Wickham H, François R. dplyr: A Grammar of Data Manipulation. 2014. Available online: https://cran.r-project.org/web/packages/dplyr/index.html
16. Poncet P. lplyr: dplyr verbs for lists and other verbs for data frames [R package lplyr version 0.1.12]. 2017. Available online: https://cran.r-project.org/src/contrib/Archive/lplyr/
17. Ginestet C. ggplot2: Elegant Graphics for Data Analysis. Journal of the Royal Statistical Society 2011;174:245-6.
18. Nusinow DP, Szpyt J, Ghandi M, et al. Quantitative Proteomics of the Cancer Cell Line Encyclopedia. Cell 2020;180:387-402.e16. [Crossref] [PubMed]
19. Kassambara A. ggpubr: ‘ggplot2’ Based Publication Ready Plots [R package ggpubr version 0.1.0]. 2017. Available online: https://CRAN.R-project.org/package=ggpubr
20. Colaprico A, Silva TC, Olsen C, et al. TCGAbiolinks: an R/Bioconductor package for integrative analysis of TCGA data. Nucleic Acids Res 2016;44:e71. [Crossref] [PubMed]
21. Wickham H, Rstudio. tidyverse: Easily Install and Load the ‘Tidyverse’ Packages. 2016. Available online: https://CRAN.R-project.org/package=tidyverse
22. Wickham H. Package tidyverse. 2017. Available online: https://CRAN.R-project.org/package=tidyverse
23. Neuwirth E. RColorBrewer (version 1.0-5). 2011. Available online: https://CRAN.R-project.org/package=RColorBrewer
24. Ahlmann-EltzeCPatilI.ggsignif: R Package for Displaying Significance Brackets for ‘ggplot2’.2021. doi:
25. Thul PJ, Lindskog C. The human protein atlas: A spatial map of the human proteome. Protein Sci 2018;27:233-44. [Crossref] [PubMed]
26. Colwill KRenewable Protein Binder Working Group. Gräslund S. A roadmap to generate renewable protein binders to the human proteome. Nat Methods 2011;8:551-8. [Crossref] [PubMed]
27. Dowle M, Srinivasan A, Gorecki J, et al. Package ‘data.table’. 2019. p. 596. Available online: https://CRAN.R-project.org/package=data.table
28. Wang S, Liu X. The UCSCXenaTools R package: a toolkit for accessing genomics data from UCSC Xena platform, from cancer multi-omics to single-cell RNA-seq. J Open Source Softw 2019;4:1627.
29. Gagolewski M. stringi: Fast and portable character string processing in R. Journal of Statistical Software 2022;103:1-59.
30. Wickham H, editor. Data manipulation with dplyr. R user conference; 2014. Available online: https://user2014.r-project.org/abstracts/talks/45_Wickham.pdf
31. Schauberger P, Walker A, Braglia L, et al. Package ‘openxlsx’. 2018. Available online: https://CRAN.R-project.org/package=openxlsx
32. Slowikowski K, Schep A, Hughes S, et al. Package ‘ggrepel’. 2018. Available online: https://CRAN.R-project.org/package=ggrepel
33. Weesie J. reshape2 f wide j long g. Available online: https://stata-press.com/journals/stbcontents/stb38.pdf
34. Lin H, Zelterman D. Modeling Survival Data: Extending the Cox Model. Technometrics 2002;44:85-6.
35. Kassambara A. Drawing Survival Curves Using ggplot2 [R package survminer version 0.2.0]. 2017. Available online: https://CRAN.R-project.org/package=survminer
36. Ahlmann-Eltze C. ggsignif: Significance Bars for ‘ggplot2’. 2017. Available online: https://CRAN.R-project.org/package=ggsignif
37. Robin X, Turck N, Hainard A, Tiberti N, Müller M. pROC-package. 2010. Available online: https://CRAN.R-project.org/package=pROC
38. Mandrekar JN. Receiver operating characteristic curve in diagnostic test assessment. J Thorac Oncol 2010;5:1315-6. [Crossref] [PubMed]
39. Obuchowski NA, Bullen JA. Receiver operating characteristic (ROC) curves: review of methods with applications in diagnostic medicine. Phys Med Biol 2018;63:07TR01. [Crossref] [PubMed]
40. Peterson BG, Carl P. Exploratory Data Analysis in Finance Using PerformanceAnalytics. International User and Developer Conference, Ames, Iowa, 8-10, 2007.
41. Wei T, Simko V, Levy M, et al. Package ‘corrplot’. Statistician. 2017;56:e24.
42. Harrell FE, Dupont C. Hmisc: Harrell Miscellaneous. 2015. Available online: https://CRAN.R-project.org/package=Hmisc
43. Xiao N, Li M. Package ‘ggsci’. 2016. Available online: https://CRAN.R-project.org/package=ggsci
44. Yoshihara K, Shahmoradgoli M, Martínez E, et al. Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun 2013;4:2612. [Crossref] [PubMed]
45. Moon KW. Make a Plot by ggplot2. In: Moon KW, editor. Learn ggplot2 Using Shiny App. Cham: Springer; 2016: 15-20.
46. Craven KE, Gökmen-Polar Y, Badve SS. CIBERSORT analysis of TCGA and METABRIC identifies subgroups with better outcomes in triple negative breast cancer. Sci Rep 2021;11:4691. [Crossref] [PubMed]
47. Newman AM, Liu CL, Green MR, et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 2015;12:453-7. [Crossref] [PubMed]
48. Zhang M, Wang X, Chen X, et al. Novel Immune-Related Gene Signature for Risk Stratification and Prognosis of Survival in Lower-Grade Glioma. Front Genet 2020;11:363. [Crossref] [PubMed]
49. Bhattacharya S, Dunn P, Thomas CG, et al. ImmPort, toward repurposing of open access immunological assay data for translational and clinical research. Sci Data 2018;5:180015. [Crossref] [PubMed]
50. Revelle W. Package ‘psych’. 2015. Available online: https://CRAN.R-project.org/package=psych
51. Kolde R. Package ‘pheatmap’. 2018. Available online: https://CRAN.R-project.org/package=pheatmap
52. Wickham H. reshape2: Flexibly Reshape Data: A Reboot of the Reshape Package. 2014. Available online: https://CRAN.R-project.org/package=reshape2
53. Schubert M, Klinger B, Klünemann M, et al. Perturbation-response genes reveal signaling footprints in cancer gene expression. Nat Commun 2018;9:20. [Crossref] [PubMed]
54. Yang M, Hu H, Wu S, et al. EIF4A3-regulated circ_0087429 can reverse EMT and inhibit the progression of cervical cancer via miR-5003-3p-dependent upregulation of OGN expression. J Exp Clin Cancer Res 2022;41:165. [Crossref] [PubMed]
55. González-Salvatierra S, García-Fontana B, Martínez-Heredia L, et al. Exploring the role of osteoglycin in type 2 diabetes: implications for insulin resistance and vascular pathophysiology. Am J Physiol Endocrinol Metab 2023;325:E649-60. [Crossref] [PubMed]
56. Jiang B, Xiao S, Zhang S, et al. The miR-1290/OGN axis in ovarian cancer-associated fibroblasts modulates cancer cell proliferation and invasion. J Ovarian Res 2024;17:52. [Crossref] [PubMed]
57. Woessner MN, Hiam D, Smith C, et al. Osteoglycin Across the Adult Lifespan. J Clin Endocrinol Metab 2022;107:e1426-33. [Crossref] [PubMed]
58. Wang Y, Ma Y, Lü B, et al. Differential expression of mimecan and thioredoxin domain-containing protein 5 in colorectal adenoma and cancer: a proteomic study. Exp Biol Med (Maywood) 2007;232:1152-9. [Crossref] [PubMed]
59. Lee JY, Eom EM, Kim DS, et al. Analysis of gene expression profiles of gastric normal and cancer tissues by SAGE. Genomics 2003;82:78-85. [Crossref] [PubMed]
60. Hu SM, Li F, Yu HM, et al. The mimecan gene expressed in human pituitary and regulated by pituitary transcription factor-1 as a marker for diagnosing pituitary tumors. J Clin Endocrinol Metab 2005;90:6657-64. [Crossref] [PubMed]
61. Liu J, Lei B, Yu X, et al. Combining Immune-Related Genes For Delineating the Extracellular Matrix and Predicting Hormone Therapy and Neoadjuvant Chemotherapy Benefits In Breast Cancer. Front Immunol 2022;13:888339. [Crossref] [PubMed]
62. Dings MPG, Manoukian P, Waasdorp C, et al. Serum levels of iCAF-derived osteoglycin predict favorable outcome in pancreatic cancer. Int J Cancer 2023;152:511-23. [Crossref] [PubMed]
63. Ning X, Deng Y. Identification of key pathways and genes influencing prognosis in bladder urothelial carcinoma. Onco Targets Ther 2017;10:1673-86. [Crossref] [PubMed]
64. Yu X, Yu B, Fang W, et al. Identification hub genes of consensus molecular subtype correlation with immune infiltration and predict prognosis in gastric cancer. Discov Oncol 2021;12:41. [Crossref] [PubMed]
65. Gopinath P, Natarajan A, Sathyanarayanan A, et al. The multifaceted role of Matricellular Proteins in health and cancer, as biomarkers and therapeutic targets. Gene 2022;815:146137. [Crossref] [PubMed]
66. van der Koog L, Woest ME, Gorter IC, et al. Fibroblast-derived osteoglycin promotes epithelial cell repair. NPJ Regen Med 2025;10:16. [Crossref] [PubMed]
67. Huang S, Lin Y, Deng Q, et al. Suppression of OGN in lung myofibroblasts attenuates pulmonary fibrosis by inhibiting integrin αv-mediated TGF-β/Smad pathway activation. Matrix Biol 2024;132:87-97. [Crossref] [PubMed]
68. Kameda T, Sugihara T, Obinata D, et al. Androgen receptor and osteoglycin gene expression predicting prognosis of metastatic prostate cancer. Sci Rep 2024;14:30654. [Crossref] [PubMed]
69. Al-Sharify D, Nielsen SH, Matthes F, et al. Increased proteolytic cleavage of osteoglycin is associated with a stable plaque phenotype and lower risk of cardiovascular events. Atherosclerosis 2022;355:8-14. [Crossref] [PubMed]
70. Cui X, Song B, Hou L, et al. High expression of osteoglycin decreases the metastatic capability of mouse hepatocarcinoma Hca-F cells to lymph nodes. Acta Biochim Biophys Sin (Shanghai) 2008;40:349-55. [Crossref] [PubMed]
71. Alvarez-Díaz S, Valle N, Ferrer-Mayorga G, et al. MicroRNA-22 is induced by vitamin D and contributes to its antiproliferative, antimigratory and gene regulatory effects in colon cancer cells. Hum Mol Genet 2012;21:2157-65. [Crossref] [PubMed]
72. Wassermann-Dozorets R, Rubinstein M. C/EBPβ LIP augments cell death by inducing osteoglycin. Cell Death Dis 2017;8:e2733. [Crossref] [PubMed]
73. Xu T, Zhang R, Dong M, et al. Osteoglycin (OGN) Inhibits Cell Proliferation and Invasiveness in Breast Cancer via PI3K/Akt/mTOR Signaling Pathway. Onco Targets Ther 2019;12:10639-50. [Crossref] [PubMed]