High SOX17 expression is correlated with favorable prognosis and CD8(+) TILs expression in patients with pancreatic cancer after curative resection Authors: Jun Yu, Yuan Pan, Yi Chen, Zhihui Shu, Lin Ma, Bin Wu, Jingdong Li, Jianshui Li, Chuan Lan Published: 2026 ABSTRACT Highlights•1.SOX17 is an independent favorable prognostic marker in pancreatic cancer patients after curative resection.•2. High SOX17 expression correlates with increased CD8+ TIL infiltration, suggesting an immunomodulatory role.•3. SOX17 expression is inversely associated with FAP+ CAFs, a stromal component linked to poor prognosis.•4. Single-cell and bulk RNA-seq analyses validate the association between SOX17 and immune-stromal components.•5. GPCR signaling pathway is implicated as a potential mechanism through which SOX17 influences prognosis.Keywords: The SRY-box transcription factor 17 (SOX17), Prognostic factor, CD8+ tumor-infiltrating lymphocytes (TILs), Cancer-associated fibroblasts (CAFs), Pancreatic cancer Introduction Pancreatic cancer (PC) is one of the most aggressive malignancies of the digestive system, accounting for approximately 2% of all cancers [1]. Over the past 30 years, the worldwide incidence of PC has doubled [2]. In the United States, although PC ranks as the tenth most commonly diagnosed cancer, it is projected to become the third leading cause of cancer-related deaths [3,4]. Owing to the high rates of recurrence and mortality, the five-year overall survival (OS) rate remains only 2% to 9%, even after standardized treatment [5]. Surgery remains the primary treatment modality for PC [6]. However, for patients with advanced or recurrent disease, other therapies, such as chemotherapy, radiotherapy, genomic therapies, stromal targeting, and immunotherapies remain unsatisfactory [7], highlighting the urgent need to identify novel therapeutic targets for PC. The SRY-box transcription factor (SOX) family comprises >20 members of the human genome and plays crucial roles in cell development, fate determination, and differentiation [8]. Among them, several members served as prognostic markers and therapeutic targets in various kinds of tumors [9]. As a key member of the SOX transcription factor family, SOX17 plays a critical role not only in regulating cardiovascular system development, lymphangiogenesis, and the formation of various endoderm-derived organs [10,11], but also in tumorigenesis and cancer progression in humans [12]. For instance, in colorectal cancer, studies showed that SOX17 may remodel the vascular endothelial cell barrier and inhibit tumor cell invasion by expressing high vascular endothelial growth factor receptor 2, which makes SOX17 an potential indicator in clinic scenario [13,14]. Besides, other recent studies indicated that SOX17 can mediate intratumoral epigenetic reprogramming. Its absence leads to an increased infiltration of CD8+ T cells and inhibitions of adenocarcinoma formation in mice, then promote immune escape in primary colorectal cancer [15,16]. In addition to being involved in the malignant biological behavior of tumors, SOX17 also serves as a prognostic predictor for multiple types of cancer in clinical practice [17]. Pancreatic cancer features a distinctive fibro-dense immunosuppressive tumor microenvironment composed of crucial cellular components such as cancer-associated fibroblasts (CAFs) and tumor-infiltrating lymphocytes (TILs) [[18], [19], [20], [21], [22]]. The interactions among these components and pancreatic cancer cells significantly influence tumor progression and treatment outcomes [23]. Given that SOX17 can affect tumor development by modulating tumor vascular endothelial cells and immune cell infiltration, investigating its role in pancreatic cancer holds important research significance. Therefore, this study aimed to investigate the clinical value and correlations with tumor cellular components of SOX17 in patients with PC after curative resection. Materials and methods Patients and samples This study included 168 postoperative patients diagnosed with pancreatic cancer between January 2016 and December 2023 at the Affiliated Hospital of North Sichuan Medical College. The inclusion criteria was as follows: 1) All patients underwent either pancreatoduodenectomy or antegrade radical distal pancreatectomy based on the location of their tumor lesions, with postoperative pathology confirming pancreatic cancer; 2) The clinicopathological and follow-up data were complete. The exclusion criteria were as follows: 1) age < 18 years; 2) patients with a history of other malignant tumors within the past 5 years; 3) patients with distant metastasis at the time of diagnosis; 4) patients with R1 or R2 resection (microscopic or macroscopic residual tumor); 5) patients with incomplete clinicopathological and follow-up data. Pathological grading of PC was performed according to the 2019 WHO classification, and TNM staging was based on the 8th edition of the American Joint Committee on Cancer staging system. No patient in this study underwent neoadjuvant chemotherapy. After surgery, patients received 6 cycles of adjuvant chemotherapy with fluorouracil- or gemcitabine-based regimens. Postoperative follow-up comprised blood tests and enhanced computed tomography or magnetic resonance imaging every two months in the first year and every six months thereafter. Overall survival (OS) was defined as the time from surgery to death, and disease-free survival (DFS) was defined as the time from surgery to recurrence or death. This retrospective study was approved by the Ethics Committee of North Sichuan Medical College (No.2024037). All procedures complied with the relevant national and institutional guidelines for human experimentation, as well as the 1964 Helsinki Declaration and its subsequent amendments. Hematoxylin-Eosin staining and immunohistochemical staining of SOX17 Formalin-fixed, paraffin-embedded tissue sections from patients with PC were deparaffinized using eco-friendly clearing agents and ethanol, followed by rinsing with distilled water. Sections were stained with hematoxylin for 2.5 min, rinsed for 30 s with distilled water, differentiated with a hydrochloric acid-ethanol solution for 7 s, and immersed in tap water for 15 min. Excess water was removed before staining with eosin for 6 min. Dehydration was performed using 95% ethanol, and the slides were mounted with resin and sealed. Hematoxylin and eosin (HE) staining was performed on samples from all 168 patients to confirm the presence of tumor tissues. Antigen retrieval was performed using different pH antigen repair solutions, such as EDTA buffer (pH 9.0), citric acid buffer (pH 6.0), or specific reagents (MVS-0100; Maixin Biotechnology, Fuzhou, China), under high-pressure autoclaving for 15 min. Immunohistochemical staining was performed using an UltraSensitive SP Kit (Mouse/Rabbit, KIT-9710; Maixin Biotechnology, Fuzhou, China). Endogenous peroxidase activity was blocked by incubating the sections with blocking agents (reagent 1) at room temperature for 15 min, followed by incubation with a nonspecific staining blocker (reagent 2) for 15 min. After blocking endogenous peroxidase activity, the sections were incubated overnight with an anti-SOX17 antibody (1:2000 dilution; DF7769; Abcam, Shanghai, China). The subsequent steps included incubation with secondary reagents (reagents 3 and 4), DAB staining, hematoxylin counterstaining, and differentiation. Three researchers independently scored staining intensity as negative, weak, moderate, or strong. Negative or weak staining was classified as low expression, whereas moderate or strong staining was classified as high expression. Immunohistochemical staining of TILs, CAFs, and PD-L1 expression Samples were processed as described above. The primary antibodies were anti-CD8 (1:100 dilution, AF5126, Affinity Biosciences, Jiangsu, China), anti-Foxp3 (1:100 dilution, AF6544, Affinity Biosciences, Jiangsu, China), anti-FAP (1:50 dilution, AF5344, Affinity Biosciences, Jiangsu, China), anti-α-SMA (1:200 dilution, AF1032, Affinity Biosciences, Jiangsu, China), and anti-PD-L1 (1:1000 dilution, 28,076-1-AP, Wuhan Sanying Biotechnology, Wuhan, China). After overnight incubation at 4 °C with primary antibodies, slides were warmed to 37 °C for 30 min, incubated with biotin-labeled secondary antibodies (reagent 3) at room temperature for 15 min, and further treated with streptavidin-biotin peroxidase complex (reagent 4) for 15 min. Diaminobenzidine (DAB; DAB-0031, Maixin Biotechnology, Fuzhou, China) was used as the chromogen. Counterstaining was performed using hematoxylin, followed by differentiation with a hydrochloric acid-ethanol solution. Three independent researchers, blinded to the patient clinical data, evaluated the stained sections under a × 200 optical microscope. For CD8 and Foxp3 staining, the number of positive cells in a field of 1 mm in diameter was counted in three fields containing a large number of cancer cells, and the results were expressed as the average of the counts in triplicate (number of cells per field). Immunohistochemical evaluation of FAP and α-SMA expression in CAFs was performed using a four-tier intensity scoring system: 0 (no staining); 1 (weak staining); 2 (moderate staining); and 3 (strong staining). Scores of 0–1 were defined as negative expressions, whereas scores of 2–3 were considered positive expressions. CD8+ and Foxp3+ TILs were quantified, and patients were stratified into high- and low-expression groups based on a 50% cutoff. A similar scoring system was used for PD-L1 expression in PC tissues. Single-cell and bulk RNA-seq analyses To validate the correlations between SOX17 expression and components of tumor microenvironment. Single-cell RNA sequencing (scRNA-seq) data containing 27 primary PC samples from GEO dataset GSE205013 were analyzed. DoubletFinder was used to control the quality of the doublets and remove them. For further quality control, cells were filtered using the parameters nFeature_RNA > 200 (gene count) and mt < 20 (mitochondrial percentage). UMAP analysis was performed to examine the cellular subpopulations. Samples were divided into low- and high-expression groups based on median SOX17 expression. To validate the function of SOX17 in pancreatic cancer, bulk RNA-seq analyses of other datasets were conducted. Differences in SOX17 expression in pancreatic cancer and normal tissues, as well as in OS and DFS between SOX17 high- and low-expressed groups, were analyzed using data from the Gene Expression Profiling Interactive Analysis (GEPIA) database. Pancreatic cancer data (n = 178) from The Cancer Genome Atlas (TCGA) date set were divided into SOX17 high- and low-expression groups based on the median SOX17 expression. Immune and stromal scores were calculated using the xCell package and visualized using boxplots generated using the ggplot2 package. The correlations between CD8 T cells and SOX17 expression were estimated by the CIBERSORT and single sample Gene Set Enrichment Analysis (ssGSEA) algorithms via R packages. Differentially Expressed Genes (DEGs) were identified using edgeR, and correlations between SOX17 expression and immune cell enrichment scores were analyzed. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed using the "ClusterProfiler" R package to identify the biological processes (BP), molecular functions (MF), cellular components (CC), and potential signaling pathways associated with SOX17 expression. Complementarily, We applied Gene Set Enrichment Analysis (GSEA) to detect whether the enrichment of GO and KEGG pathways is statistically significant in SOX17 expression. The significance thresholds for GSEA were |NES| > 1, p < 0.05, and FDR < 0.25. Statistical analysis Continuous variables were expressed as mean ± standard deviation for normally distributed data or as median (interquartile range) for non-normally distributed data. The Student’s t- test or Mann-Whitney U test was used to evaluate differences in continuous variables, while the chi-square test or the Fisher exact test was applied to categorical variables. Cox proportional hazards regression was used to identify prognostic factors, with variables from the univariate analysis (P < 0.05) included in the multivariate analysis. Kaplan-Meier curves were used to estimate OS and DFS, and log-rank tests were used for comparison. Wilcoxon rank-sum and chi-square tests were used for the correlation analysis. Statistical significance was set at P < 0.05. All analyses were conducted using the IBM SPSS Statistics version 23. Results SOX17 was an independent prognostic factor for patients with pancreatic cancer after curative resection In total, 168 patients with PC who underwent surgical resection were enrolled in this study (Fig. 1). Among them, 123 (73.2%) patients died, and 111 (66.1%) suffered from tumor recurrence. The median follow-up period was 26 months. The presence of tumor tissue in all 168 patient samples was confirmed by hematoxylin and eosin (HE) staining. After immunohistochemical (IHC) evaluation, 74 patients were assigned to the low SOX17 expression group, and 94 patients were assigned to the high SOX17 expression group (Fig. 2A). Baseline data, including age, sex, total bilirubin levels, tumor size, pT, pN, tumor differentiation, and tumor location, showed no significant differences between the two groups. However, CA19– 9 levels were higher in the low SOX17 expression group compared with high SOX17 expression group (Table 1). Flow diagram of the study participant selection process. SOX17 expression in pancreatic cancer and its association with patient prognosis. (A) Representative immunohistochemical staining of SOX17 in resected pancreatic cancer specimens. Red arrows indicate high SOX17 expression in cancer cells, while white arrows indicate low SOX17 expression. (B) Kaplan-Meier curves showing overall survival and disease-free survival stratified by SOX17 expression level. The log-rank test was used to compare survival differences between the high- and low-expression groups. SOX17, The Sex Determining Region Y-related high-mobility group box 17. TABLE 1Clinicopathological features comparisons based on SOX17 expression.TABLE 1 dummy alt textCharacteristicTotal(n = 168)SOX17 expressionp valueLow (n = 74))High (n = 94)Age(y)63.99±9.5064.18±10.5763.85±8.630.827Sex0.943Male114(67.9%)50(29.8%)64(38.1%)Female54(32.1%)24(14.3%)30(17.9%)CA19–9(U/mL)98.60(21.45–457.83)215.50(34.98–461.18)53.70(17.40–464.25)0.04Total bilirubin(umol/L)41.40(13.08–166.30)72.45(12.83–175.93)27.45(13.38–159.73)0.225Tumor size(mm)30.00(20.25–40.00)30.00(20.75–40.00)30.00(20.00–40.25)0.746pT0.608T1/2125(74.4%)57(33.9%)68(40.5%)T3/443(25.6%)17(10.1%)26(15.6%)pN0.259N0110(65.5%)45(26.8%)65(38.7%)N158(34.5%)29(17.3%)29(17.3%)Differentiation0.699Poor23(13.7%)9(5.4%)14(8.3%)Moderate103(61.3%)48(28.6%)55(32.7%)Well42(25%)17(10.1%)25(14.9%)Location0.35Head100(59.5%)47(28.0%)53(31.5%)Neck,body,and tail68(40.5%)27(16.1%)41(24.4%)Data are presented as mean ± standard deviation, n (%), or median (25–75%). CA19– 9, Carbohydrate antigen-199. Clinicopathological features comparisons based on SOX17 expression. Data are presented as mean ± standard deviation, n (%), or median (25–75%). CA19– 9, Carbohydrate antigen-199. Kaplan-Meier analysis demonstrated that high SOX17 expression was correlated with favorable OS (log-rank P < 0.05) and DFS (log-rank P < 0.05) (Fig. 2B). Univariate and multivariate Cox regression analysis showed that high SOX17 expression and positive FAP+ CAF expression were independent prognostic factors for OS and DFS (TABLE 2, TABLE 3). Interestingly, CD8+ TILs were identified as an independent prognostic factors for DFS but not for OS (Table 3). TABLE 2Univariate and multivariate analysis of overall survival according to SOX17 expression in patients with pancreatic cancer.TABLE 2 dummy alt textVariablesUnivariate analysisMultivariate analysisHR(95%CI)PHR(95%CI)PAge>65 years1.186(0.828–1.699)0.352Female1.194(0.811–1.757)0.369CA19–9>37 U/mL1.128(0.779–1.632)0.523Total bilirubin>34.2 umol/L0.851(0.594–1.221)0.381Tumor size >30 mm1.387(0.966–1.991)0.076pN10.998(0.692–1.437)0.989Poor differentiation0.922(0.558–1.524)0.752Neck,body, and tail0.878(0.607–1.270)0.489FAP+ CAFs positive expression3.018(2.005–4.542)<0.0012.828(1.858–4.304)<0.001α-SMA+ CAFs positive expression0.615(0.430–0.879)0.0080.774 (0.537–1.116)0.170PD-L1 positive expression1.541(1.055–2.251)0.0251.483 (0.996–2.207)0.052CD8+ TILs high expression0.587(0.404–0.853)0.0050.704 (0.481–1.029)0.070Foxp3+ TILs high expression1.313(0.918–1.879)0.136High SOX17 expression0.593(0.412–0.855)0.0050.662 (0.445–0.986)0.042SOX17, The SRY-box transcription factor 17; HR, hazard ratio; CA19– 9, Carbohydrate antigen-199; FAP, fibroblast activation protein; α-SMA, alpha smooth muscle actin; PD-L1, programmed cell death ligand-1; TILs, tumor-infiltrating lymphocytes; Foxp3, forkhead box protein 3. Univariate and multivariate analysis of overall survival according to SOX17 expression in patients with pancreatic cancer. SOX17, The SRY-box transcription factor 17; HR, hazard ratio; CA19– 9, Carbohydrate antigen-199; FAP, fibroblast activation protein; α-SMA, alpha smooth muscle actin; PD-L1, programmed cell death ligand-1; TILs, tumor-infiltrating lymphocytes; Foxp3, forkhead box protein 3. TABLE 3Univariate and multivariate analysis of disease-free survival according to SOX17 expression in patients with pancreatic cancer.TABLE 3 dummy alt textVariablesUnivariate analysisMultivariate analysisHR(95%CI)PHR(95%CI)PAge>65 years1.122(0.767–1.642)0.552Female1.808(1.220–2.678)0.0031.255(0.799–1.972)0.324CA19–9>37 U/mL0.848(0.578–1.243)0.398Total bilirubin>34.2 umol/L0.746(0.509–1.095)0.135Tumor size >30 mm1.258(0.856–1.848)0.242pN10.794(0.536–1.177)0.251Poor differentiation1.430(0.860–2.377)0.168Neck,body, and tail0.878(0.607–1.270)0.489FAP+ CAFs positive expression2.638(1.750–3.975)<0.0012.311(1.512–3.534)<0.001α-SMA+ CAFs positive expression0.762(0.521–1.114)0.161PD-L1 positive expression1.217(0.819–1.809)0.330CD8+ TILs high expression0.422(0.285–0.623)<0.0010.526(0.338–0.817)0.004Foxp3+ TILs high expression1.415(0.968–2.070)0.073High SOX17 expression0.602(0.404–0.896)0.0120.635(0.416–0.970)0.036SOX17, The SRY-box transcription factor 17; HR, hazard ratio; CA19– 9, Carbohydrate antigen-199; FAP, fibroblast activation protein; α-SMA, alpha smooth muscle actin; PD-L1, programmed cell death ligand-1; TILs, tumor-infiltrating lymphocytes; Foxp3, forkhead box protein 3. Univariate and multivariate analysis of disease-free survival according to SOX17 expression in patients with pancreatic cancer. SOX17, The SRY-box transcription factor 17; HR, hazard ratio; CA19– 9, Carbohydrate antigen-199; FAP, fibroblast activation protein; α-SMA, alpha smooth muscle actin; PD-L1, programmed cell death ligand-1; TILs, tumor-infiltrating lymphocytes; Foxp3, forkhead box protein 3. Association between SOX17 expression and TILs CD8+ and Foxp3+ TILs were quantified, and patients were stratified into high- and low-expression groups using a 50% cutoff. Correlation analysis revealed that high SOX17 expression was positively associated with CD8+ TILs (W = 2690.5, P = 0.012; Fig. 3A), but not with Foxp3+ TILs (P = 0.81; Fig. 3B). Correlation of CD8 and Foxp3 expression in TILs with SOX17 expression. (A) Representative immunohistochemical staining of CD8 in TILs (brown, cytoplasmic) showing correlation with SOX17 expression levels. (B) Representative immunohistochemical staining of Foxp3 in TILs (brown, nuclear) showing correlation with SOX17 expression levels. Foxp3, forkhead box protein 3; SOX17, SRY-box transcription factor 17; TIL, tumor-infiltrating lymphocyte. Association between SOX17 expression and CAFs and PD-L1 expression Previous studies have associated FAP+ CAFs with poor overall survival (OS) and disease-free survival (DFS), while α-SMA+ CAFs have been linked to better OS and DFS. In this study, FAP was expressed positively in 107 patients and negatively in 61 patients, and α-SMA was expressed positively in 98 patients and negatively in 70 patients. Correlation analysis revealed a significant negative association between SOX17 expression and FAP expression in CAFs (χ2 = 6.47, P = 0.011; Fig. 4A and B), but no significant correlation was found between SOX17 and α-SMA expression in CAFs (P = 0.495). Additionally, PD-L1 was positively expressed in 110 patients and negatively expressed in 58 patients; however, no significant correlation was observed between SOX17 expression and PD-L1 expression (Fig. 4C). Correlation of PD-L1 expression in cancer cells and FAP/α-SMA expression in CAFs with SOX17 expression. (A) Representative immunohistochemical staining of PD-L1 in cancer cells showing correlation with SOX17 expression level. (B) Representative immunohistochemical staining of FAP in CAFs showing correlation with SOX17 expression level. (C) Representative immunohistochemical staining of α-SMA in CAFs showing correlation with SOX17 expression level. PD-L1, programmed cell death ligand-1; SOX17, The Sex Determining Region Y-related high-mobility group box 17; FAP, fibroblast activation protein; CAF, cancer-associated fibroblast; α-SMA, alpha smooth muscle actin. Validation of the function of SOX17 expression in pancreatic cancer by single-cell and bulk RNA-seq analyses Following the removal of doublet cells, the total cell count across the samples was 76,847. After further quality control, this number was reduced to 68,064 cells. The samples were classified into 12 distinct cellular subpopulations, with the proportions of each group and their corresponding markers presented in Fig. 5A and B After excluding 17 samples lacking SOX17 expression, the remaining 10 samples were categorized into high- and low-expression groups. Consistent with expectations, SOX17 expression was found to correlate with specific cellular subpopulations (Fig. 5C). Notably, samples with high SOX17 expression exhibited increased infiltration of CD8+ T cells (Fig. 5D and E), which was in strong agreement with the previous correlation analysis. Validation of the correlation between SOX17 and CD8+ T cells through single-cell analysis. (A) UMAP plot illustrating the major cell populations in the GSE205013 dataset. (B) Heatmap showing marker genes associated with each cell population. (C) Correlation between SOX17 expression and distinct cell populations. (D) Correlation between SOX17 expression and CD8+ T cells. (E) Stacked bar plot of cell proportions, indicating that high SOX17 expression is associated with increased CD8+ T cell infiltration. SOX17, SRY-related HMG-box 17; PD-1, programmed cell death protein 1; PD-L1, programmed cell death ligand 1. In the TCGA dataset, high SOX17 expression was correlated with higher ImmuneScore and StromalScore (Fig. 6A), while SOX17 expression was also positively correlated with CD8+ TILs (Fig. 6B-D). Besides, the survival analysis showed that high SOX17 mRNA expression was also associated with favorable OS and DFS, consistent with IHC findings (Fig. 7A). Differential expression analysis identified 52 upregulated and 410 downregulated genes between the high- and low- SOX17 expression groups, visualized in heatmaps and volcano plots (Fig. 7B and C). Enrichment analysis revealed that both GO and KEGG terms were significantly enriched in the G protein-coupled receptor (GPCR) signaling pathway (Fig. 7D). The same pathway was also highlighted by gene set enrichment analysis (Fig. 7E). Analysis of immune miscroenvironment based on SOX17 expression. (A) The boxplot of the ImmuneScore and StromalScore of PC patients based on SOX17 expression; (B) Correlation analysis of CD8 T cells and SOX17 expression by CIBERSORT algorithm; (C) Correlation between SOX17 expression and immune cell infiltration; (D) Correlation analysis of CD8 T cells and SOX17 expression by ssGSEA algorithm. SOX17, The sex determining region Y-related high-mobility group box 17; PC, Pancreatic cancer; ssGSEA, single sample gene set enrichment analysis (ssGSEA). Validation of SOX17 mRNA expression in relation to patient prognosis and potential enriched pathways. (A) Kaplan–Meier curves for overall survival and disease-free survival based on SOX17 mRNA expression levels. (B) Volcano plot showing the distribution of DEGs between high and low SOX17 expression groups. (C) Heatmap illustrating the expression of cytokines and receptors in the high and low SOX17 expression groups. (D) GO and KEGG enrichment analyses. (E) GSEA pathway enrichment analysis. SOX17, SRY-related HMG-box 17; DEGs, differentially expressed genes; GO, Gene ontology; KEGG, Kyoto Encyclopedia of genes and genomes; GSEA, Gene set enrichment analysis. Discussion In this study, we found that high SOX17 expression was associated with a favorable prognosis in patients with PC after curative resection. The correlation analyses showed that high SOX17 expression was positively correlated with CD8+ TILs and negatively with FAP+ CAFs. According with that, scRNA-seq showed that the high SOX17 expression group had a higher proportion of CD8+ T cells. Meanwhile, analysis of the TCGA dataset confirmed that the group with high SOX17 expression exhibited a greater number of immune-infiltrating cells. Furthermore, immune enrichment correlation analysis indicated an elevated presence of CD8+ T cells in this high-expression group. The tumor microenvironment (TME) in PC is highly complex and comprises a variety of stromal components, including cancer-associated fibroblasts (CAFs), immune cells, endothelial cells, neurons, and a dense extracellular matrix (ECM), which collectively foster tumor progression, therapy resistance, and immunosuppression [23]. Among these, CAFs are a key stromal population that promote tumor growth, angiogenesis, invasion, and chemoresistance through the secretion of growth factors and inflammatory mediators [24]. In PC, CAFs are heterogeneous and can be classified into three main subtypes: myofibroblastic CAFs (myCAFs), inflammatory CAFs (iCAFs), and antigen-presenting CAFs (apCAFs), each contributing distinctively to disease progression [25]. MyCAFs, located proximally to tumor cells, are primarily involved in ECM remodeling; iCAFs reside farther from the tumor nest within a dense stromal network and secrete inflammatory chemokines (e.g., IL-6, IL-8, IL-11) and tumor-promoting factors [26,27]; whereas apCAFs express MHC-II and CD74, enabling them to activate CD4+ T cells in an antigen-specific manner [28].Notably, FAP and α-SMA serve as critical markers of CAF activation. FAP is highly expressed in most epithelial malignancies but largely absent in normal stroma [29]; its presence correlates with poor prognosis, and its depletion can induce hypoxic tumor necrosis [30,31]. Conversely, α-SMA expression is associated with improved survival, while its loss enhances tumor invasiveness [22]. The immune landscape within PC is similarly shaped by the immunosuppressive TME. While CD8+ T-cell infiltration generally predicts better outcomes, high levels of Foxp3+ regulatory T cells (Tregs) are inversely correlated with survival [32]. Immune checkpoint molecules such as PD-1 and PD-L1 further contribute to T-cell dysfunction and tumor immune escape [33]. Importantly, CAFs actively suppress anti-tumor immunity by impairing CD8+ T-cell recruitment, stromal penetration, and cytotoxic activity [34]. They can also directly inhibit T cells through upregulation of immune-checkpoint ligands; for example, CAFs from melanoma biopsies upregulate PD-L1 and PD-L2, which engage PD-1 on T cells and dampen their function [35]. Similarly, elevated PD-L1 and PD-L2 have been observed in CAFs isolated from resected PC specimens [36]. Despite these insights, current chemotherapeutic and immunomodulatory regimens remain largely ineffective in PC, underscoring the need to identify additional drivers of disease progression. The SOX family of transcription factors, characterized by their highly conserved HMG domains, plays critical yet often context-dependent roles in tumor biology [37]. Individual SOX members can act as either oncogenes or tumor suppressors, largely influenced by their specific expression patterns and cellular environments [38]. In oncology, SOX17 has emerged as a clinically relevant molecule with dual value: as a prognostic biomarker and as a regulator of tumor-immune interactions. For example, in gastric cancer, SOX17 serves as an independent prognostic factor, largely through its association with APC promoter methylation [[39], [40], [41]]. Dysregulated SOX17 expression has been documented across a spectrum of malignancies-including breast, endometrial, lung, esophageal, and gastric cancers [[42], [43], [44], [45], [46]]. Notably, while low SOX17 levels typically predict poorer outcomes in many cancers, high SOX17 expression has been linked to enhanced cisplatin sensitivity in endometrial cancer, underscoring its context-specific role [47]. Furthermore, SOX17 helps maintain cellular differentiation; its expression is higher in moderately differentiated endometrial cancer cells than in undifferentiated counterparts, supporting its tumor-suppressive function in certain settings [[48], [49], [50]]. At the molecular level, even single mutations in SOX17 can disrupt protein interactions (e.g., with Oct4), interfere with canonical WNT/β-catenin signaling, and consequently inhibit metastatic behavior [51]. It should be noted that abnormal expression of SOX17 is highly heterogenous across multiple cancer types and the expression pattern of SOX17 is also associated with cancer prognosis in certain cancer types. A pan-cancer analysis by Xu et al. [52] found that patients with kidney renal clear cell carcinoma and pancreatic adenocarcinoma with higher expression levels of SOX17 exhibited promising OS rates, suggesting that higher expression of SOX17 is a potentially useful prognostic marker for these tumors, which is largely consistent with our study. In contrast, for ocular melanomas, mesothelioma, and aforementioned cancers, higher expression levels of SOX17 were significantly associated with poor overall survival. Moreover, of particular relevance is the association between SOX17 and CD8+ T cell-mediated anti-tumor immunity. Studies have revealed that SOX17 expression inversely correlates with CD8+ T cell infiltration in the tumor microenvironment. Mechanistically, SOX17 can directly repress the expression of key genes in the IFN-γ signaling cascade within cancer cells, thereby attenuating the feed forward loop of effector-like CD8+ T cell recruitment and activity. This creates an immune-evasive niche, particularly evident in early colorectal adenomas and cancers where SOX17 upregulation suppresses adaptive immune responses [15,16]. The protein's role in modulating the immunogenicity of cancer stem cells (CSCs), specifically by transcriptionally repressing LGR5 to form a less immunogenic LGR5-CSC population, further highlights its capacity to impair CD8+ T cell recognition and clearance of tumor cells [16]. Consequently, SOX17 expression patterns may serve as a predictor of response to immunotherapies, which predominantly rely on functional CD8+ T cell activity. However, this study is the reliance on mouse models, which may not fully recapitulate the complexity of human cancers. Additionally, the study’s focus on the early stages of tumorigenesis leaves gaps in understanding the role of SOX17 in later stages of cancer progression and metastasis. Our study reached the opposite conclusion, which also provides a reference for exploring the role of SOX17 in the tumor immune microenvironment of PC. Emerging evidence also highlights SOX17’s involvement in G protein-coupled receptor (GPCR) signaling-a pathway increasingly recognized for its role in tumor progression and immune modulation [53,54]. GPCRs regulate diverse intracellular signaling cascades and influence immune-cell migration and function, thereby potentially affecting responses to immune checkpoint inhibitors [55]. In line with this, our bioinformatic analysis suggests that SOX17 may affect patient prognosis in pancreatic cancer (PC), potentially involving the GPCR signaling pathway. Nevertheless, this finding requires validation through further in vitro studies. In summary, this positions SOX17 not only as a prognostic indicator but also as a promising immunomodulatory target. Targeting SOX17 could potentially overcome therapy resistance in PC. In conclusion, SOX17 represents a multifaceted molecule with significant clinical potential: its expression patterns offer prognostic information across multiple cancers, and its functional involvement in key pathways-including immune evasion, differentiation maintenance, and GPCR signaling-makes it an attractive candidate for targeted therapeutic strategies. Its specific role in regulating CD8+ T cell infiltration and function further underscores its importance in shaping the tumor-immune microenvironment, providing a compelling rationale for developing SOX17-directed therapies to enhance anti-tumor immunity. Further investigation is warranted to fully elucidate its mechanisms and translate these insights into improved clinical outcomes. Ethics approval and consent to participate The studies involving humans were approved by the Ethics Committee of North Sichuan Medical College (number: 2024,037). The studies were conducted in accordance with the local legislation and institutional requirements. The human samples used in this study were acquired from primarily isolated as part of your previous study for which ethical approval was obtained. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements. The IRB (the Ethics Committee of North Sichuan Medical College) waived patient informed consent. Consent for publication Not applicable. Availability of data and materials The scRNA-seq dataset analyzed during the current study was GSE205013 from the GEO dataset. This data had been deposited in the Genome Sequence Archive for Human (GSA-Human) in the National Genomics Data Center under accession no HRA004625 (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA004625). Funding This work was supported by the Youth Innovation Project of Sichuan Medical Association (grant no 20250075) and the Institutional Research and Development Fund Project of North Sichuan Medical College (grant no CBY22-QNA07). This work has also been supported by the Sichuan Provincial Clinical Key Specialty Construction Project (Project No 2023GPWKP003). CRediT authorship contribution statement Jun Yu: Writing – original draft, Methodology, Data curation, Conceptualization. Yuan Pan: Writing – original draft, Methodology, Conceptualization. Yi Chen: Methodology, Data curation. Zhihui Shu: Writing – original draft, Formal analysis, Data curation. Lin Ma: Writing – original draft. Bin Wu: Writing – original draft. Jingdong Li: Writing – original draft. Jianshui Li: Writing – review & editing. Chuan Lan: Writing – review & editing, Writing – original draft. Declaration of competing interest The authors declare no competing interests. Acknowledgements Not applicable. Contributor Information Jianshui Li, Email: ljs00916@foxmail.com. Chuan Lan, Email: lanyouyou@nsmc.edu.cn. References Associated Data Data Availability Statement The scRNA-seq dataset analyzed during the current study was GSE205013 from the GEO dataset. This data had been deposited in the Genome Sequence Archive for Human (GSA-Human) in the National Genomics Data Center under accession no HRA004625 (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA004625).