Exosomal P4HA3: a promising biomarker for diagnosis and prognosis in gastric cancer

Introduction

Gastric cancer (GC) remains a formidable global health challenge, ranking as the sixth most commonly diagnosed cancer and the fifth leading cause of cancer-related mortality worldwide (1). The insidious onset and absence of distinctive early symptoms often lead to delayed diagnosis, contributing to a poor prognosis, particularly in advanced stages where the five-year survival rate remains dismal (2,3). Despite recent advancements in targeted and immunotherapeutic approaches that have improved median survival times (MST), the overall survival (OS) rate for GC patients remains unsatisfactory (4,5). Consequently, there is a critical and unmet need to develop effective, non-invasive strategies for the early detection and dynamic monitoring of GC.

Current early screening methods for GC primarily rely on endoscopic biopsy, imaging examinations, and serum biomarker detection. However, endoscopic biopsy is invasive, costly, and carries risks such as bleeding and perforation (6). Traditional imaging techniques often lack the sensitivity required for detecting early-stage lesions (7). Therefore, non-invasive circulating biomarkers are increasingly regarded as promising tools for early GC detection and surveillance (8). Commonly used serum biomarkers include pepsinogen, carcinoembryonic antigen (CEA), and carbohydrate antigen 19-9 (CA19-9). Nonetheless, the clinical utility of these markers is limited by their suboptimal sensitivity and specificity (9). For instance, the positive rates for CA19-9 pre- and post-operation were reported to be only 7% and 10% (10), respectively, underscoring the urgent need for novel biomarkers with improved diagnostic performance.

In this context, exosomes—nanoscale extracellular vesicles (30–150 nm in diameter) secreted by various cell types—have emerged as a promising new class of biomarkers. These vesicles facilitate intercellular communication by transferring bioactive molecules, including proteins, nucleic acids, and lipids, from donor to recipient cells, thereby playing pivotal roles in cancer initiation and progression (11,12). Detectable in nearly all types of cells, exosomes are closely linked to tumor behavior by delivering oncogenic cargo that modulates the tumor microenvironment and metastatic niche (13,14). For example, in bladder cancer, exosomal long non-coding RNA LNMAT2 promotes lymphangiogenesis and lymphatic metastasis (15), while in colon cancer, exosome-derived Wnt1 enhances tumor cell proliferation by activating the Wnt pathway activation (16), in liver cancer, CPS1 derived from serum exosomes may be a promising diagnostic and prognostic biomarker (17). These findings highlight the significant potential of exosomal cargo [e.g., long non-coding RNA (lncRNA), messenger RNA (mRNA), and microRNA (miRNA)] as biomarkers reflecting tumor development (18). However, the roles of exosome-related genes in GC remain poorly characterized, and their potential diagnostic and prognosis value is yet to be fully elucidated.

Prolyl 4-hydroxylase subunit alpha 3 (P4HA3), a key enzyme involved in collagen synthesis and stability, has been implicated in the pathogenesis of various cancers, including lung cancer, renal cell carcinoma, and colon cancer. It promotes tumor growth, invasion, and metastasis by activating the PI3K/AKT signaling pathway and facilitating epithelial-mesenchymal transformation (EMT) (19). Despite these findings, the expression, function, and mechanisms of exosomal P4HA3 in GC remain largely unexplored.

Therefore, this study aims to investigate the potential of P4HA3 exosomes and tissues as a novel diagnostic and prognostic biomarker for GC. Utilizing a comprehensive approach that integrates bioinformatics analyses of public databases with experimental validation in vitro and in vivo, we seek to characterize the role of P4HA3 in GC tissues progression and its underlying mechanisms. We present this article in accordance with the MDAR and ARRIVE reporting checklists (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0194/rc).

Methods

Data acquisition and processing

Gene expression data [in fragments per kilobase million (FPKM) format], and corresponding clinical information for GC were downloaded from public databases, including the Gene Expression Omnibus [GEO; https://www.ncbi.nlm.nih.gov/geo/, accessed on (October 30, 2023)] and The Cancer Genome Atlas [TCGA; https://cancergenome.nih.gov, accessed on (October 26, 2023)]. The specific datasets utilized in this study were TCGA-Stomach Adenocarcinoma (TCGA-STAD), GSE153413 (comprising 10 GC and 5 normal plasma exosome samples) and GSE103236 for subsequent analyses, with GSE103236, which serve as external validation cohort. FPKM data from TCGA were converted to transcripts per million (TPM) for further analysis. To mitigate potential batch effects arising from the integration of multiple datasets, we applied the ComBat algorithm from the “sva” package (version 3.46.0) in R. Clinical characteristics of the eligible GC patients after quality control were summarized in Table S1. To minimize bias from non-GC-related mortality, we excluded individuals with an OS time of ≤30 days or missing survival data, as these short-term fatalities were likely attributable to postoperative complications or other non-cancer causes (20), a filtering criterion consistent with previous oncological studies.

Patient and sample collection

Tissue specimens

A total of 50 archived pairs of GC samples and matched adjacent non-tumor tissue samples were retrospectively retrieved from the pathology tissue bank of Guilin People’s Hospital. These samples were collected from patients who underwent surgical resection between January 2019 and December 2022. All GC tissue samples were pathologically confirmed according to the TNM staging system established by the Union for International Cancer Control (UICC). After retrieval, the formalin-fixed, paraffin-embedded (FFPE) tissue blocks were used for immunohistochemical (IHC) analysis.

Plasma specimens

Peripheral blood samples were obtained from 50 GC patients and 50 healthy volunteers at Guilin People’s Hospital from August 2022 to August 2024. For each participant, 5 mL of venous blood was drawn into EDTA-anticoagulant tubes. Plasma was isolated by centrifugation at 3,500 ×g for 10 min at 4 ℃. The supernatant (plasma) was carefully aliquoted and stored at −80 ℃ until exosome extraction.

The baseline clinical characteristics of the plasma cohort were summarized in Table S2. The associations between P4HA3 expression levels plasma-derived exosomes and the clinical features were provided in Table S3. None of the enrolled subjects had undergone any anticancer treatment prior to initial blood draw. GC staging was assessed in accordance with the 7th edition of the UICC TNM classification system.

Ethical approval and informed consent

The study protocol was approved by the Ethics Committee of Guilin People’s Hospital (approval No. 2022-146KY). All procedures involving human participants were performed in accordance with the ethical standards of the Declaration of Helsinki and its subsequent amendments. Written informed consent was obtained from all individual participants or their legal representatives for the collection and use of plasma samples. For the retrospective analysis of archived tissue specimens, the requirement for individual informed consent was waived by the same ethics committee.

All animal experiments were approved by the Experimental Animal Ethics Committee of Guilin Medical College (approval No. GLMC202405186). All procedures were performed in accordance with the institutional guidelines for the care and use of laboratory animals.

Bioinformatics analysis

Screening of exosome-related and prognostic genes

The exosome-related dataset GSE153413 was obtained from the GEO database. The raw count matrix was normalized using the “DESeq2” package (version 1.38.0) in R (version 4.2.1). Differential gene expression analysis was performed in parallelusing three independent methods: the “Deseq2” package, the “limma” package (version 3.52.0) with the voom transformation, and the “edgeR” package (version 3.38.0). Genes with an adjusted P value (Benjamini-Hochberg false discovery rate, FDR) <0.05 and an absolute log₂ fold change (|log₂FC|) ≥1 were considered statistically significant. The final list of exosome-related differentially expressed genes (DEGs) was defined as the intersection of the significant genes identified by all three methods to ensure robustness. A heatmap visualizing these genes was generated using the “pheatmap” package (version 1.0.13).

DEGs in GC tissues were identified from the TCGA-STAD dataset using the same thresholds (FDR <0.05, |log₂FC| ≥1). The prognostic significance of these DEGs was assessed using the “survival” package (version 3.4.0), and 323 genes associated with OS were identified. The upregulated genes from the exosomal dataset (GSE153413) were then intersected with the prognostic genes from TCGA-STAD, resulting in five candidate genes (NPAS3, ZNF697, PALLD, P4HA3, HP). P4HA3 was selected for further validation.

Differential expression and prognostic analysis

Based on the intersection analysis, P4HA3 was selected as the lead candidate for further investigation. The expression of tissue P4HA3 were compared between GC patients and normal controls. The association between tissue P4HA3 expression and clinicopathological stages was also assessed. To evaluate its prognostic value, patients within the TCGA cohort were stratified into high- and low-expression groups based on the median tissue P4HA3 expression value. Kaplan-Meier survival curves were generated, and the log-rank test was used to compare OS differences between the two groups. The prognostic accuracy of tissue P4HA3 for predicting 1-, 3-, and 5-year survival was evaluated using time-dependent receiver operating characteristic (ROC) analysis with the “timeROC” package (version 0.3), and area under the curve (AUC) values were calculated.

To independently validate the differential expression of tissue P4HA3, we analyzed the GSE103236 dataset. The consistent upregulation of tissue P4HA3 in GC samples within this independent cohort confirmed our initial findings from the TCGA discovery set.

Functional enrichment and immune infiltration analysis

To investigate the biological function of tissue P4HA3, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using the “clusterProfiler” package (version 4.18.0). The analysis was conducted on genes showing a strong positive correlation with tissue P4HA3 expression in the TCGA-STAD dataset (Pearson correlation coefficient r>0.7). Terms with an FDR <0.05 were considered significantly enriched.

The relative abundances of tumor-infiltrating immune cells were estimated using the CIBERSORT algorithm. The association between tissue P4HA3 expression and immune/molecular subtype was analyzed using the TISIDB database [http://cis.hku.hk/TISIDB/, accessed on (November, 06, 2023)].

Pan-cancer and drug sensitivity analysis

Tissue P4HA3 expression across multiple cancer types was analyzed using data from TCGA through the “pheatmap” package (version 1.0.12). The mutation frequency and specific mutation patterns of tissue P4HA3 in various cancers were analyzed using the cBioPortal database [https://www.cbioportal.org/, accessed on (November, 25, 2023)].

The correlation between P4HA3 expression and drug sensitivity of cancer cells to drugs was analyzed using the CellMiner database [https://discover.nci.nih.gov/cellminer/, accessed on (November, 18, 2023)]. The analysis was restricted to clinically trialed and FDA-approved drugs. The Pearson correlation coefficient between tissue P4HA3 expression and drug activity (IC₅₀) was calculated.

Exosome isolation and characterization

Exosome isolation

Exosomes were isolated from plasma samples using the ExoQuick exosome extraction kit (Thermo Fisher Scientific, Waltham, MA, USA; Catalog: 4484450) according to the manufacturer’s protocol. Briefly, thawed plasma was centrifuged at 3,500 ×g for 20 min at 4 ℃ to remove cell debris. The supernatant was mixed with the exosome precipitation reagent and incubated at 4 ℃ for 30 min, followed by centrifugation at 10,000 ×g for 70 min to pellet the exosomes. The final exosome pellet was resuspended in phosphate buffered saline (PBS) and stored at −80 ℃ for downstream applications.

Exosome characterization

Isolated vesicles were characterized using a multi-method approach to confirm their identity as exosomes.

Transmission electron microscopy (TEM): for morphological analysis, exosomes were fixed in 2% paraformaldehyde, applied to a formvar-carbon coated grid, and negatively stained with 1% uranyl acetate. Imaging was performed using a HT7700 TEM operated at 80 kV.

Nanoparticle tracking analysis (NTA): the size distribution and concentration of exosomes were determined using a ZetaView NTA system. Samples were diluted 1:1,000 in PBS to achieve an optimal concentration for analysis.

Western blotting: protein lysates from exosomes were subjected to western blot analysis to detect specific exosomal markers. The membranes were probed with primary antibodies against the positive markers CD81 (Abcam, ab109201), and TSG101 (Abcam, ab125011), and the negative marker Calnexin (Abcam, ab133615). Antibody dilutions and incubation conditions were optimized according to the manufacturers’ protocols.

Cell culture and transfection

The human GC cell line HGC-27 was obtained from Chinese Academy of Sciences Cell Bank (Shanghai, China). Cells were cultured in RPMI-1640 medium (Sigma-Aldrich, St. Louis, MO, USA) with 10% fetal bovine serum (FBS; Gibco, Grand Island, NY, USA) and 1% penicillin-streptomycin at 37 ℃ in a humidified atmosphere containing 5% CO₂. Cell line authentication was performed using short tandem repeat (STR) profiling, and tests for mycoplasma contamination were negative.

The human GC cell line SGC-7901 was also obtained from Chinese Academy of Sciences Cell Bank (Shanghai, China) and cultured under the same conditions as HGC-27 cells. This cell line was selected for secondary validation based on its relatively high endogenous P4HA3 expression level.

Stable knockdown of P4HA3 by lentiviral transduction

To establish a GC cell line with stable P4HA3 knockdown, HGC-27 cells were transduced with lentiviral vectors encoding short hairpin RNA (shRNA) targeting P4HA3. The lentivirus was produced by co-transfecting 293T cells with the recombinant shRNA-P4HA3 shuttle plasmid and packaging plasmids using the RNAi-Mate transfection reagent. Viral supernatants were collected 72 hours post-transfection, concentrated by ultracentrifugation, and titrated.

For transduction, HGC-27 cells were seeded in 6-well plates and infected with the lentivirus at a predetermined optimal multiplicity of infection (MOI) in the presence of 1 µg/mL Polybrene. After 18 hours, the virus-containing medium was replaced with fresh complete medium. To select stable cell pools, puromycin (2 µg/mL) was added to the culture 72 hours post-transduction, and selection was maintained until non-transduced control cells were completely eliminated. The knockdown efficiency of P4HA3 in HGC-27 cells was confirmed at the protein level by Western blot analysis.

To verify whether cells secrete P4HA3 via exosomes, exosomes were also isolated from the serum-free conditioned medium of HGC-27 cells stably transfected with si-NC or si-P4HA3 using the same ExoQuick reagent kit mentioned above. The expression of P4HA3 and the exosomal marker TSG101 in these exosomes was then detected by Western blot analysis.

The lentiviral transduction procedure and shRNA constructs (LV-2005 and the negative control) used for HGC-27 cells were also applied to establish P4HA3-knockdown and control pools in SGC-7901 cells. Knockdown efficiency was confirmed by Western blot analysis. Exosomes were also isolated from the conditioned medium of SGC-7901 cells using the identical protocol described above, for the purpose of analyzing exosomal P4HA3 secretion.

Plasmid construction and P4HA3 overexpression

For P4HA3 overexpression in HGC-27 cells, the full-length human P4HA3 coding sequence was cloned into the pcDNA3.1 (+) vector (Thermo Fisher Scientific, USA) to generate the pcDNA3.1-P4HA3 plasmid. The empty pcDNA3.1 (+) vector was used as a negative control (Vector). HGC-27 cells were seeded in 6-well plates and transfected at 60–70% confluence using Lipofectamine 3000 reagent (Thermo Fisher Scientific, USA), following the manufacturer’s protocol. In brief, 2.5 µg of plasmid DNA was diluted in Opti-MEM medium, mixed with the transfection reagent, and added to the cells. The cells were incubated with the complex at 37 ℃ under 5% CO₂ for 48 hours before being harvested for subsequent Western blot analysis to confirm overexpression.

All functional assays were performed with a minimum of three independent biological replicates.

Functional assays in vitro

Exosome uptake assay

For in vitro and in vivo tracking, purified exosomes were labeled with the lipophilic fluorescent dye PKH67 (Sigma-Aldrich, USA; Catalog: PKH67GL) according to the manufacturer’s protocol. Briefly, 100 µg of exosome protein was resuspended in 100 µL of PBS and mixed with an equal volume of a PKH67 dye solution (diluted 1:500 in Diluent C). The mixture was incubated for 10 minutes at room temperature in the dark. The labeling reaction was stopped by adding an equal volume of 1% bovine serum albumin (BSA) in PBS. Labeled exosomes were purified from unbound dye using 100 kDa molecular weight cut-off ultrafiltration centrifuge tubes (Millipore, USA) by centrifugation at 10,000 ×g for 70 minutes, followed by three washes with PBS.

HGC-27 cells were seeded into confocal dishes (Nest, China) at a density of 5×10⁴ cells per dish and cultured overnight to approximately 70% confluence. The culture medium was then replaced with fresh complete medium containing PKH67-labeled exosomes at a final concentration of 20 µg/mL. Cells were incubated with the labeled exosomes for 6, 12, and 24 hours at 37 ℃ in the dark. After incubation, cells were washed three times with ice-cold PBS to remove non-internalized exosomes, fixed with 4% paraformaldehyde for 15 minutes at room temperature, and counterstained with an anti-fade mounting medium containing DAPI (Beyotime, Shanghai, China) to visualize nuclei.

Cellular uptake of PKH67-labeled exosomes (green fluorescence) was observed and Imaged using a Nikon A1R confocal laser scanning microscope (Nikon, Tokyo, Japan) equipped with appropriate filter sets. Image acquisition and analysis were performed using NIS-Elements imaging software (Nikon).

Quantitative real-time polymerase chain reaction (qRT-PCR)

Total RNA was extracted from exosome samples or cells using TRIzol™ reagent (Thermo Fisher Scientific, USA). RNA purity and concentration were determined using a spectrophotometer (NanoDrop, Thermo Fisher Scientific), with all samples having an A260/A280 ratio between 1.8 and 2.0. Complementary DNA (cDNA) was synthesized from 1 µg of total RNA using the RevertAid RT Reverse Transcription Kit (Thermo Fisher Scientific, USA). The thermocycling conditions were as follows: 95 ℃ for 30 sec, followed by 40 cycles of 95 ℃ for 5 seconds and 60 ℃ for 34 sec. Gene expression was normalized to GAPDH, and relative quantification was calculated using the 2^-ΔΔCt method. The sequences of the primers used in this study are listed in Table S4.

To assess the relationship between exosomal P4HA3 expression and clinicopathological features, correlation analyses were conducted using the “ggstatsplot” package in R. Variables including age, gender, clinical stage, and levels of conventional tumor markers were included to evaluate potential associations with exosomal P4HA3 expression.

Cell proliferation and migration assays

Cell counting kit-8 (CCK-8) assays

Cells were seeded into 96-well plates at a density of 5×10³ cells per well. After transfection, 10 µL of CCK-8 reagent (APExBIO, USA) was added to each well at 24 and 36 h time points, followed by incubation for 4 hours. Absorbance was measured at 450 nm using a microplate reader (Varioskan LUX, Thermo Fisher Scientifc, USA). The CCK-8 assay was also performed in SGC-7901 cells following P4HA3 knockdown, as described.

Scratch assay

Cells were seeded into 6-well plates and grown to 90–100% confluence. A standardized wound was created in the cell monolayer using a sterile 200 µL pipette tip. After washing with PBS to remove detached cells, serum-free medium was added. Images of the scratch were captured at 0, 24, and 36 h using an inverted microscope. The migration rate was quantified by measuring the change in the scratch area using ImageJ software (NIH, USA). The percentage of wound closure was calculated using the formula: % Closure = (Area at 0 h − Area at Tn)/Area at 0 h × 100%.

Apoptosis analysis by flow cytometry

Cells apoptosis was evaluated using an Annexin V-FITC/PI Apoptosis Detection Kit (Beyotime, China). Briefly, 48 hours post-transfection, cells were harvested, washed with cold PBS, and resuspended in 1X Binding Buffer. The cells were then stained with 5 µL of Annexin V-FITC and 10 µL of PI for 15 min at room temperature in the dark. Apoptotic cells were analyzed within 1 hour using a flow cytometer (FACScan, BD Biosciences, USA). Data were analyzed using FlowJo software (Tree Star, USA). Apoptosis was similarly assessed in SGC-7901 cells by flow cytometry using the same method.

Western blot analysis

WB analysis was conducted following a previously reported protocol (21) with adjustments. Total protein was extracted from cells using RIPA lysis buffer (Beyotime, China) containing protease and phosphatase inhibitors. Protein concentration was determined using a BCA assay kit (Thermo Fisher Scientific, USA). Equal amounts of protein (20–30 µg) were separated by 10% SDS-PAGE and transferred to PVDF membranes (Millipore, USA). After blocking with 5% non-fat milk for 1 hour at room temperature, the membranes were incubated overnight at 4 ℃ with primary antibodies. The primary antibodies and dilutions used were as follows: anti-P4HA3 (1:500, Abcam, AB101657), anti-COL1A1 (1:1,000, CST, #72026), anti-E-cadherin (1:2,000, CST, #3195), anti-N-cadherin (1:1,000, CST, #13116), anti-Vimentin (1:1,000, CST, #5741), and anti-GAPDH (1:5,000, CST, #2118). After washing, membranes were incubated with HRP-conjugated secondary antibodies for 1 hour at room temperature. Protein bands were visualized using an enhanced chemiluminescence (ECL) substrate and imaged with a ChemiDoc XRS+ imaging system (Bio-Rad, USA). Band intensities were quantified using ImageJ software (NIH, USA) and normalized to GAPDH.

IHC

IHC staining was performed on formalin-fixed, FFPE tissue sections from 50 paired GC and adjacent normal tissue samples to evaluate P4HA3 protein expression. Tissue sections (4 µm thick) were deparaffinized in xylene and rehydrated through a graded ethanol series. Antigen retrieval was performed by heating the slides in sodium citrate buffer (pH 6.0) for 15 min in a microwave oven. After cooling, endogenous peroxidase activity was quenched with 3% hydrogen peroxide for 15 min at room temperature. Non-specific binding was blocked by incubating the sections with 3% BSA for 1 hour at room temperature.

The sections were then incubated overnight at 4 ℃ with a primary antibody against P4HA3 (1:50, Thermo Fisher Scientific, PA5-48623). After washing with PBS, the sections were incubated with a horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG secondary antibody (1:50, Abcam, ab6702) for 45 min at room temperature. The antigen-antibody complex was visualized using a 3,3’-diaminobenzidine (DAB) horseradish peroxidase chromogenic kit (ZSGB-BIO, China). Finally, the sections were counterstained with hematoxylin, dehydrated, cleared, and mounted.

IHC scoring criteria

P4HA3 protein expression was evaluated independently by two experienced pathologists who were blinded to the clinical data. Staining intensity and the percentage of positive tumor cells were assessed. The staining intensity was scored as follows: 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). The proportion of positive cells was scored as: 0 (<5%), 1 (5–25%), 2 (26–50%), 3 (51–75%), and 4 (>75%). The final immunoreactivity score (IRS) for each case was calculated by multiplying the intensity score by the proportion score, resulting in a total score ranging from 0 to 12. An IRS score of ≥4 was defined as high expression, and ≤3 as low expression.

Slides were scanned using an Aperio ScanScope XT slide scanner system (Leica Biosystems, Germany) with a 40× objective. Image analysis was performed using ImageScope software (Leica Biosystems).

Animal experiments

Xenograft tumor model

Female BALB/c nude mice (4–5 weeks old, n=3 per group) were purchased from Beijing Weitong Lihua Biotechnology Co. Ltd. (Beijing, China). Mice were housed under specific pathogen-free (SPF) conditions with a 12 h light/dark cycle and provided with sterile food and water ad libitum. After one week of acclimatization, mice were randomly assigned to two groups (si-NC and si-P4HA3) using a random number table. This sample size (n=3 per group) is consistent with common practice in preliminary xenograft studies to minimize animal use in accordance with the “3R” principles.

Tumor monitoring and analysis

Tumor size was monitored every 48 hours. Tumor length (L) and width (W) were measured with a digital caliper by an investigator who was blinded to the group allocations. Tumor volume was calculated using the formula: Volume = (L × W²)/2. The predefined humane endpoints for the study included a tumor volume exceeding 1,500 mm³, a body weight loss of >20%, or signs of severe distress; no animals reached these criteria before the experiment concluded.

At the endpoint, all mice were euthanized by cervical dislocation under deep anesthesia with 5% isoflurane. Upon confirming the loss of consciousness via absent pedal reflex, cervical dislocation was carried out to ensure death. All raw measurement data were compiled in Table S5. Tumors were carefully excised, photographed, weighed, and measured. A portion of each tumor was snap-frozen in liquid nitrogen for molecular analysis, and another portion was fixed in 4% paraformaldehyde for histological examination.

Statistical analysis

All analyses were performed using SPSS software (version 22.0; IBM, NY, USA) and GraphPad Prism (version 10.0; GraphPad Software, USA). Data are presented as mean ± standard deviation (SD) for normally distributed continuous variables or as number (percentage) for categorical variables. The normality of data distribution was assessed using the Shapiro-Wilk test. The specific statistical tests applied were as follows:

Comparisons between two groups: for continuous variables satisfying normality, an unpaired, two-tailed Student’s t-test was used. The Mann-Whitney U test was applied for non-normally distributed data.

Comparisons among multiple groups: for multi-group comparisons of continuous data, one-way analysis of variance (ANOVA) was employed for normally distributed data, followed by Tukey’s honest significant difference (HSD) post hoc test for pairwise comparisons. If the data violated the assumptions of normality or homogeneity of variances, the Kruskal-Wallis test was used, followed by Dunn’s test for post hoc analysis.

Analysis of categorical data: for comparisons of proportions and countable data, the Chi-squared (χ²) test was used. Fisher’s exact test was applied when the expected frequency in any contingency table cell was less than 5.

A P value of less than 0.05 was considered statistically significant (*P<0.05, **P<0.01, ***P<0.001, ****P<0.0001).

Results

Identification of P4HA3 as a key related and prognostic gene in GC

To identify exosome-derived genes with clinical relevance in GC, we first performed differential expression analysis on the plasma exosome dataset GSE153413. This analysis identified 269 significantly upregulated genes and 21 downregulated genes (|log₂FC| ≥1, FDR <0.05; Figure 1). Concurrently, analysis of the TCGA-STAD tissue dataset also showed a widespread dysregulation of mRNA expression in GC tissues (Figure 2A), from which we identified 323 genes significantly associated with OS (FDR <0.05).

Figure 1 Screening of exosome-related differentially expressed genes in gastric cancer. (A) Identification of 269 upregulated genes in plasma exosomes from the GSE153413 dataset. (B) Identification of 21 downregulated genes in plasma exosomes. (C) Expression level of P4HA3 in exosomes. (D) Heatmap showing differential expression of exosome-related genes. GC, gastric cancer; NC, negative control.

Figure 2 Identification of prognostic exosome-related genes. (A) Heatmap of differentially expressed genes in gastric cancer tissues from TCGA database. (B) Venn diagram showing the intersection of upregulated genes from the GSE153413 dataset and prognostic genes from the TCGA-STAD cohort, identifying five candidate genes: NPAS3, ZNF697, PALLD, P4HA3, and HP. (C) Differential expression of the five intersection genes in exosomes (GSE153413) and GC tissues (TCGA). FC, fold change; GC, gastric cancer; TCGA, The Cancer Genome Atlas.

Intersection of the 269 exosome-upregulated genes with the prognostic genes yielded five candidate genes: NPAS3, ZNF697, PALLD, P4HA3, and HP (Figure 2B). We applied a criterion of consistent and significant upregulation in both GC tissues and plasma exosomes for prioritization. While P4HA3 and HP met this criterion, NPAS3 and PALLD were excluded due to their discordant expression patterns between compartments. As shown in Figure 2C, although significantly upregulated in plasma exosomes (NPAS3: log₂FC =5.82, adj. P<0.05; PALLD: log₂FC =4.59, adj. P<0.05), both genes were significantly downregulated in GC tissues(NPAS3: log₂FC =−1.28, adj. P<0.05; PALLD: log₂FC =−1.47, adj. P<0.05). On the other hand, P4HA3 and HP showed consistent upregulation in GC exosomes and tissues (P4HA3: exosome log₂FC =5.91, FDR <0.05; TCGA log₂FC =2.11, adj. P<0.001).

It is well-established that HP is a primary etiological factor for GC, with extensive research delineating its impact on exosomal cargo and the tumor microenvironment (22). However, a significant subset of GC cases, particularly those of the diffuse subtype or in younger populations, arise through H. pylori-independent pathways. The current study was therefore strategically designed to identify novel exosome-derived biomarkers that may operate independently of the canonical H. pylori-driven oncogenesis. Our unbiased bioinformatics approach identified P4HA3, whose marked upregulation in plasma exosomes and GC tissues, along with its strong association with aggressive tumor phenotypes and poorer prognosis, which is verified in external datasets, suggests a fundamental role in GC progression that may extend beyond the context of H. pylori infection. While future studies investigating the potential interplay between H. pylori status and P4HA3 expression will be invaluable for understanding GC heterogeneity, our present findings position exosomal P4HA3 as a promising diagnostic and prognostic biomarker worthy of further investigation.

Tissue P4HA3 expression correlates with advanced disease and poor prognosis

To comprehensively validate the clinical relevance of our candidate gene, we performed a detailed analysis of tissue P4HA3 expression and its prognostic value. Consistent with our screening results, P4HA3 expression was significantly elevated in GC tissues from the TCGA-STAD cohort compared to normal samples (P<2.22e−16; Figure 3A). Notably, tissue P4HA3 expression levels increased progressively with advancing clinical stage (stage I vs. stage III & IV, P<0.05; Figure 3B), suggesting a potential role in disease progression.

Figure 3 Expression and prognostic value of P4HA3 in GC. (A) Expression of P4HA3 in GC and normal tissues from the TCGA (P=2.22e−16). (B) Association between tissue P4HA3 expression levels and clinical stage of GC. (C) Kaplan-Meier survival analysis of the prognostic value of tissue P4HA3 in the TCGA cohort. (D) ROC curves evaluating the accuracy of tissue P4HA3 in predicting 1-, 3-, and 5-year overall survival of GC patients. (E) Validation of tissue P4HA3 overexpression in an independent external cohort (GSE103236). AUC, area under the curve; FC, fold change; GC, gastric cancer; ROC, receiver operating characteristic; TCGA, The Cancer Genome Atlas.

To quantify its prognostic impact, patients were stratified into high- and low-expression based on the median tissue P4HA3 expression. Kaplan-Meier analysis revealed that patients with high tissue P4HA3 expression had a significantly poorer OS (Figure 3C), which was consistent with the results in the Kaplan-Meier plotter public database [n=631, HR =1.41 (1.14–1.75), P=0.0018, Figure S1]. Time-dependent ROC analysis demonstrated that tissue P4HA3 exhibited moderate accuracy in predicting 1-year (AUC =0.70), but its predictive power was limited for 3-, and 5-year survival (AUC =0.54 and 0.59, respectively; Figure 3D), indicating its utility may be more relevant for short-term prognosis.

To independently corroborate the core finding of tissue P4HA3 dysregulation, we analyzed an external validation cohort, GSE103236. The upregulation of tissue P4HA3 in GC was successfully validated in this independent dataset (P<0.001; Figure 3E), strengthening the reliability of our initial observation from the TCGA cohort.

Collectively, these findings established P4HA3 as a robust biomarker whose upregulation is strongly associated with GC progression and unfavorable patient outcomes. To elucidate the potential biological mechanisms underpinning these clinical observations, we next sought to characterize the functional pathways and tumor immune microenvironment associated with tissue P4HA3 expression.

Biological functions and immune landscape associated with P4HA3

To gain insight into the biological processes driven by tissue P4HA3, we performed GO and KEGG pathway enrichment analyses. Genes showing a strong positive correlation with tissue P4HA3 expression (Pearson r>0.7) in the TCGA-STAD cohort were selected as the input. This stringent correlation threshold was chosen to focus on genes with a high likelihood of coregulation. Notably, sensitivity analyses confirmed that the core enriched pathways remained robust even when more lenient thresholds (e.g., r>0.5 or r>0.6) were applied, supporting the specificity of our findings (Figures S2,S3). GO analysis of the r>0.7 gene set revealed significant enrichment in biological processes related to extracellular matrix (ECM) organization, collagen fibril organization, and cell-substrate adhesion (Figure 4A). KEGG pathway analysis indicated predominant involvement in protein digestion and absorption, the PI3K-AKT signaling pathway, and ECM-receptor interaction (Figure 4B). These enrichment results are consistent with the established role of tissue P4HA3 as a collagen-modifying enzyme and suggest its potential involvement in remodeling the tumor microenvironment and facilitating EMT (23).

Figure 4 Functional enrichment analysis of tissue P4HA3. (A) GO enrichment analysis for genes correlated with tissue P4HA3 expression. (B) KEGG pathway enrichment analysis for genes correlated with tissue P4HA3 expression. BP, biological process; CC, cell composition; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; MF, molecular function.

Given the critical role of the tumor immune microenvironment in cancer progression, we leveraged the CIBERSORT algorithm to deconvolute the immune cell infiltration landscapeand assess its association with tissue P4HA3 expression. As shown in the Figure 5A, our deconvolution analysis revealed that high tissue P4HA3 expression was associated with distinct immune infiltration patterns. Interrogation of immune cell-specific expression databases indicated that P4HA3 mRNA is predominantly expressed in natural killer (NK) cells, with its level higher than in many other immune cell types. Specifically, tumors with high tissue P4HA3 expression exhibited a markedly greater abundance of NK cells and macrophages compared to the low-expression group (P<0.05; Figure S4). Furthermore, significant upregulation was also observed in effector memory CD4⁺ T cells and effector memory CD8⁺ T cells. The specific enrichment of NK cells suggests a potential, yet undefined, role for cellular P4HA3 in modulating innate immune responses within the GC microenvironment.

Figure 5 Immune and molecular subtyping analysis of tissue P4HA3. (A) Correlation analysis between tissue P4HA3 expression and levels of immune cell infiltration. (B) Tissue P4HA3 expression across different immune subtypes (C1: wound healing, C2: IFN-γ dominant, C3: inflammation, C4: lymphocyte depletion, C5: immunologically quiet, C6: TGF-β dominant). (C) Tissue P4HA3 expression across different molecular subtypes. *, P<0.05; **, P<0.01. CIN, chromosomal instability; EBV, high load of Epstein-Barr virus; GS, genomic stability; HM-indel, highly mutated gene insertion or deletion; HM-SNV, highly mutated single nucleotide variation.

To further explore the immunomodulatory role of P4HA3, we analyzed its distribution across established immune and molecular subtypes of GC using the TISIDB database. Tissue P4HA3 mRNA expression was significantly elevated in the TGF-β-dominant immune subtype (C6) compared to other subtypes [e.g., C1(wound healing); P<0.001; Figure 5B], which is characterized by stromal activation and immune exclusion. At the molecular level, tissue P4HA3 was most highly expressed in the genomically stable (GS) subtype, with a trend of elevated expression in the chromosomally unstable (CIN) subtype (Figure 5C). This association with the TGF-β-dominant and GS subtypes aligns with its proposed role in ECM remodeling and suggests a potential link to immune evasion mechanisms (24), though functional studies were needed to establish causality.

Pan-cancer analysis and genomic alterations of P4HA3

To determine whether the dysregulation of tissue P4HA3 extends beyond GC, we conducted a pan-cancer analysis of its expression profile across 33 cancer types from TCGA. Tissue P4HA3 mRNA expression was significantly upregulated (log₂FC >1, FDR <0.05) in a majority of malignancies compared to their corresponding normal tissues (Figure 6A,6B). Notably high expression was observed in sarcoma (SARC), cholangiocarcinoma (CHOL), stomach adenocarcinoma (STAD), and colon adenocarcinoma (COAD) (Figure S5). This conserved overexpression pattern across diverse cancers suggests that tissue P4HA3 may play a broad oncogenic role in human tumorigenesis (25).

Figure 6 Pan-cancer analysis of tissue P4HA3 expression and genomic alterations. (A) Pan-cancer expression analysis of tissue P4HA3 mRNA across 33 cancer types from TCGA. (B) Comparison of P4HA3 expression between tumor and adjacent normal tissues in selected cancer types. (C) Schematic of the P4HA3 gene showing the landscape and locations of somatic mutations. (D) Mutation frequency and the corresponding types of P4HA3 alterations across different cancers. TCGA, The Cancer Genome Atlas.

We next interrogated the genomic alteration landscape of tissue P4HA3 using the cBioPortal database. The most prevalent type of genetic alteration across cancers was gene amplification (Figure 6C), with the highest mutation frequencies observed in esophageal carcinoma (ESCA, >8%), ovarian cancer (OV), and GC (STAD) (Figure 6D). While the functional consequences of these amplifications on protein stability and activity remain to be fully elucidated, their prevalence, particularly in gastrointestinal cancers, underscores the potential clinical relevance of tissue P4HA3.

Given the emerging link between tumor mutational burden (TMB) and response to immunotherapy, we explored the correlation between tissue P4HA3 alteration status and TMB. Although no significant genome-wide correlation was found, these data contribute to the growing characterization of the molecular landscape in which tissue P4HA3 operates and may inform future studies on its role in cancer biology and therapy.

Exosomal P4HA3 as a non-invasive biomarker

To translate our findings into a clinically relevant context, we investigated the potential of P4HA3 as a non-invasive biomarker by analyzing its expression in plasma-derived exosomes from an independent cohort of 50 GC patients and 50 healthy volunteers. Western blot analysis validated the successful isolation by detecting the presence of the exosomal markers CD81 and TSG101, and the absence of the negative cellular marker Calnexin (Figure 7A). The isolated vesicles exhibited the characteristic cup-shaped morphology under transmission electron microscopy (Figure 7B), and nanoparticle tracking analysis confirmed a size distribution peaking within the typical exosomal range (mean diameter: 78.8 nm, mode: 80.0 nm, concentration: 5.76E+9 particles/mL; Figure 7C). Critically, qRT-PCR analysis revealed that the level of P4HA3 mRNA in plasma exosomes was significantly elevated in GC patients compared to healthy controls (P<0.001; Figure 7D), corroborating our prior findings from tissue and bioinformatics analyses.

Figure 7 Isolation, characterization, and clinical relevance of plasma exosomal P4HA3. (A) Western blot analysis of exosomal markers (CD81, TSG101) and a negative control marker (Calnexin) in plasma-derived exosomes from GC patients and healthy controls. (B) Representative TEM image of isolated exosomes (scale bar =100 nm). (C) Size distribution and concentration of exosomes as determined by NTA. (D) qRT-PCR analysis of P4HA3 mRNA levels in plasma exosomes from GC patients and healthy controls. (E-G) Correlation between exosomal P4HA3 levels and (E) clinical stage, (F) T stage, and (G) N-stage. GC, gastric cancer; N, node; NTA, nanoparticle tracking analysis; qRT-PCR, quantitative real-time polymerase chain reaction; T, tumor; TEM, transmission electron microscopy.

We next assessed the clinical relevance of exosomal P4HA3 by examining its correlation with key clinicopathological parameters. Exosomal P4HA3 levels showed a significant positive correlation with advanced clinical stage (r=0.54, P<0.01; Figure 7E), deeper tumor invasion (T stage; r=0.49, P<0.05; Figure 7F), and greater lymph node metastasis (N stage; r=0.38, P<0.05; Figure 7G). These consistent correlations suggest that exosomal P4HA3 is not merely a marker of tumor presence, but is quantitatively associated with disease aggressiveness.

To assess the independent association of tissue P4HA3 with clinical stage, we performed a multivariate linear regression analysis using the TCGA-STAD cohort (n=375), adjusting for age and gender. This well-powered analysis demonstrated that tissue P4HA3 expression is associated with advanced clinical stage (P<0.05; Figure S6). The forest plot shows the coefficient estimates with 95% confidence intervals for each variable in the model.

To functionally link the presence of P4HA3 in plasma exosomes to its biological effects on recipient cells, we next investigated whether GC cells could internalize these vesicles. Purified exosomes were fluorescently labeled with the lipophilic dye PKH-67 (green) and co-cultured with HGC-27 cells. Confocal microscopy imaging after 6, 12, and 24 hours of incubation revealed a time-dependent uptake of PKH67-labeled exosomes, which extensively accumulated in the cytoplasm of recipient cells by 24 hours (Figure 8). Nuclei were counterstained with DAPI (blue). This clear demonstration of efficient exosome uptake by GC cells provides a crucial mechanistic foundation for the subsequent investigation of exosomal P4HA3’s functional roles.

Figure 8 Uptake of fluorescently labeled exosomes by HGC-27 cells. Representative confocal microscopy images showing the time-dependent uptake of PKH67-labeled exosomes (green) by HGC-27 cells. Nuclei were counterstained with DAPI (blue). Scale bar =100 μm. DAPI, 4'-6-diamidino-2-phenylindole.

P4HA3 knockdown suppresses GC cell malignant phenotypes in vitro

To investigate the functional role of the P4HA3 protein in GC cells, we established a stable P4HA3-knockdown cell line using lentiviral transduction (Figure S7A,S7B). Three distinct lentiviral shRNAs targeting P4HA3 (LV-2001, LV-2003, and LV-2005) were constructed and used to transduce HGC-27 cells, followed by puromycin selection to obtain stable knockdown pools. Western blot analysis confirmed efficient depletion of P4HA3 protein by all three shRNAs, with LV-2005 showing the most potent effect (P<0.001; Figure S7C,S7D). Further more, P4HA3 was successfully demonstrated to be overexpressed in HGC-27 cells by Western blot (Figure S7E,S7F). On the other hand, to investigate whether P4HA3 is secreted via the exosomal pathway, we examined the level of P4HA3 in exosomes derived from cell culture supernatants. The results demonstrated that a clear P4HA3 band was detected in exosomes from si-NC cells, whereas the P4HA3 signal was significantly attenuated in exosomes from si-P4HA3 cells. The expression of the exosomal marker protein TSG101 remained consistent across all groups, which confirmed equal loading and successful exosome isolation (Figure S8). These findings indicate that P4HA3 can be secreted by GC cells via exosomes, and its secretion level is directly regulated by intracellular P4HA3 expression.

Based on this validation, we selected the most effective shRNA (LV-2005, hereafter referred to as si-P4HA3) for all subsequent functional assays. We first assessed the effect on cell migration using a scratch wound-healing assay. The migratory capacity of HGC-27 cells was markedly attenuated following P4HA3 knockdown. The percentage of wound closure was significantly lower in the si-P4HA3 group than in the si-NC group after 24 and 36 hours (P<0.01; Figure 9A).

Figure 9 Modulation of P4HA3 expression affects malignant phenotypes and EMT in gastric cancer cells. (A) Cell migration assessed by scratch wound healing assay in HGC-27 cells transfected with control (si-NC) or P4HA3-targeting (si-P4HA3) shRNA. (B) Cell viability measured by CCK-8 assay at 24 and 36 hours post-transfection. (C) Apoptosis was detected by flow cytometry using Annexin V-FITC/PI staining at 24 h post-transfection. The lower-right quadrant (Annexin V+/PI-) and the upper-right quadrant (Annexin V+/PI+) represent early and late apoptotic cells, respectively. Data are presented as mean ± SD. (D)Western blot analysis of COL1A1 and EMT marker proteins (E-cadherin, N-cadherin, Vimentin) in HGC-27 cells transfected with an empty vector (Vector) or a P4HA3-overexpressing plasmid (P4HA3-OE). ns, no significance; **, P<0.01; ***, P<0.001. CCK-8, Cell counting kit-8; EMT, epithelial-mesenchymal transformation; NC, negative control; SD, standard deviation.

We next evaluated the impact of P4HA3 depletion on cellular proliferation. The CCK-8 assay revealed that cell viability was significantly impaired in the si-P4HA3 group compared to the si-NC group at both 24 and 48 hours post-transfection (P<0.01; Figure 9B).

Furthermore, we investigated whether the suppression of proliferation and migration was associated with the induction of apoptosis. Flow cytometric analysis using Annexin V-FITC/PI staining showed that the percentage of apoptotic cells (early and late apoptosis combined) was significantly higher in the si-P4HA3 group compared to the si-NC control (P<0.001; Figure 9C).

To assess the generalizability of these findings, we performed parallel validation in a second GC cell line, SGC-7901, which expresses a relatively high basal level of P4HA3. Consistent with the results in HGC-27 cells, knockdown of P4HA3 in SGC-7901 cells also significantly suppressed cell proliferation and induced apoptosis (Figure S9).

Collectively, these functional assays demonstrate that knockdown of intracellular P4HA3 effectively suppresses the malignant phenotypes of GC cells by inhibiting proliferation, impairing migration, and promoting apoptosis in vitro.

P4HA3 overexpression upregulates COL1A1 and induces EMT marker changes

Building on our bioinformatics enrichment analysis, which suggested a potential role for P4HA3 in ECM organization and EMT, we sought to experimentally validate these computational predictions and examine the associated changes in COL1A1 and EMT markers using cellular models. We first confirmed the strong positive correlation between P4HA3 and COL1A1 expression in GC tissues observed in the TCGA-STAD cohort (Pearson r=0.84, P<0.001; Figure S10) at the protein level in our cellular model. Concurrently, P4HA3 overexpression in cells induced a marked shift in EMT marker expression characteristic of a more aggressive phenotype, evidenced by the downregulation of the epithelial marker E-cadherin and the upregulation of the mesenchymal markers N-cadherin and Vimentin (P<0.05; Figure 9D).

These results, demonstrating that forced expression of P4HA3 in GC cells is sufficient to drive molecular changes consistent with EMT, provide complementary evidence to our knockdown data. When considered together, the loss-of-function and gain-of-function experiments at the cellular level lend support to the notion that P4HA3 contributes to GC progression, potentially through a process involving COL1A1 upregulation and EMT.

Therapeutic implications and in vivo validation

To consolidate the clinical relevance and therapeutic potential of P4HA3, we integrated evidence from patient tissues, in vivo models, and drug response profiling.

First, IHC analysis of our patient cohort provided compelling evidence at the protein level. P4HA3 protein expression in tumor tissues was significantly stronger in human GC tissues compared to matched adjacent normal tissues (Figure 10A,10B), reinforcing the transcriptomic findings and solidifying P4HA3 as a consistently upregulated protein in GC.

Figure 10 P4HA3 knockdown inhibits tumor growth in vivo. (A,B) Representative images and expression of IHC staining for P4HA3 in gastric cancer and matched adjacent normal tissues (scale bar, 100 μm at 40× magnification). (C) P4HA3 mRNA levels in xenograft tumor tissues measured by RT-qPCR. (D,E) Representative photographs of subcutaneous tumors resected from nude mice. (F) Tumor weight from the xenograft mouse model. (G) Tumor volume growth curves over time. ***, P<0.001. IHC, immunohistochemical; NC, negative control; RT-qPCR, reverse transcription-quantitative polymerase chain reaction.

We next substantiated the tumor-promoting function of P4HA3 using an in vivo xenograft model. Mice inoculated with HGC-27 cells subjected to stable P4HA3 knockdown (si-P4HA3) prior to transplantation developed significantly smaller and lighter tumors compared to the control (si-NC) group, as confirmed by RT-qPCR analysis of tumor tissues (P<0.001; Figure 10C). Representative images (Figure 10D,10E) and quantitative measurements of tumor weigh (P<0.01; Figure 10F) and volume over time (P<0.001; Figure 10G) consistently demonstrated the potent inhibitory effect of silencing P4HA3 in cancer cells on tumor growth in vivo.

Finally, to delineate the drug sensitivity profile of P4HA3 expression levels, we analyzed its correlation with FDA-approved and or clinically trialed agents using the CellMiner database. Among the 25 drugs significantly associated with P4HA3 gene expression, we selected the four agents with the strongest positive correlation and the four with the strongest negative correlation for detailed visualization in Figure 11. Specifically, elevated P4HA3 levels were associated with increased sensitivity to ritonavir, gefitinib, sorafenib, and nandrolone phenylpropionate, but with reduced sensitivity to azathioprine, colchicine, vincristine, and geldanamycin. These results position P4HA3 expression as a potential predictive biomarker for drug response, which may guide individualized therapeutic strategies.

Figure 11 Association between P4HA3 expression and drug sensitivity. Scatter plots showing the correlation between P4HA3 expression levels and sensitivity to (A-D) 6-mercaptopurine, colchicine, vincristine, geldanamycin; and (E-H) norvir, gefitinib, sapitinib, and nandrolone phenylpropionate.

Discussion

GC continues to pose a substantial global health burden, with its high mortality rate largely attributable to limitations in timely detection. Early diagnosis of GC remains challenging, contributing to poor prognosis. Identifying sensitive, non-invasive biomarkers is critical for improving patient outcomes (26,27). Exosomes, as nanoscale extracellular vesicles detectable in readily accessible body fluids, offer significant advantages over traditional tissue biomarkers, including non-invasiveness and dynamic monitoring capability (28). These vesicles function as critical mediators of intercellular communication and have emerged as promising sources of biomarkers for cancer diagnosis and prognosis (29).

P4HA3, a member of the prolyl 4-hydroxylase family, encodes a protein involved in collagen synthesis and stability. Accumulating evidence links P4HA3 to the pathogenesis and progression of various cancers, though its function appears context-dependent. In renal cell carcinoma and colon cancer, elevated tissue P4HA3 expression promotes tumor growth, invasion and metastasis via the PI3K/AKT signaling pathway (30). Conversely, Long et al. reported that tissue P4HA3 suppresses pituitary adenoma progression, suggesting a context-dependent function of tissue P4HA3 in tumor biology (31). Despite these findings, the function of tissue P4HA3 in GC—particularly its expression in exosomes, regulatory mechanisms, and immunotherapeutic potential—remains poorly characterized before our investigation. This underexploration is likely due to the predominant focus on H. pylori-related mechanisms in GC, while exosomal P4HA3 may operate through alternative pathways, and exosome-based biomarkers are still emerging.

Our study demonstrates that P4HA3 is significantly overexpressed in both GC tissues and plasma exosomes, and is associated with unfavorable clinical outcomes. The consistent upregulation of exosomal P4HA3 across multiple analytical platforms, along with its correlation with advanced clinical stage, supports its potential role as a non-invasive biomarker in GC (32).

Functional enrichment analysis provided insights into the biological processes driven by tissue P4HA3. Our analyses revealed that P4HA3-associated genes are involved in ECM organization, collagen fibril organization, and cell-substrate adhesion. KEGG pathway analysis indicated predominant involvement in protein digestion and absorption, the PI3K-AKT signaling pathway, and ECM-receptor interaction. These findings are consistent with the established role of tissue P4HA3 as a collagen-modifying enzyme and suggest its potential involvement in remodeling the tumor microenvironment and facilitating EMT.

Immune profiling based on transcriptomic data showed that high P4HA3 expression in GC tissues is significantly elevated in the TGF-β dominant immune subtype, which is characterized by stromal activation and immune exclusion. Since TGF-β suppresses antitumor immunity and promotes cancer invasion, this immune context may underlie the pro-tumorigenic effects of P4HA3 (33). Additionally, TGF-β stimulation upregulates P4HA3 expression (23), suggesting a potential feedback loop that drives immune evasion in advanced GC.

Pan-cancer analysis of tumor tissue expression confirmed that P4HA3 dysregulation extends beyond GC, with significant upregulation observed in various malignancies including colorectal cancer, lung cancer, and breast cancer. Genomic alteration analysis revealed that gene amplification represents the most prevalent type of genetic alteration, with the highest mutation frequencies observed in esophageal carcinoma, ovarian cancer, and GC. While the functional consequences of these amplifications require further elucidation, their prevalence underscores the potential clinical relevance of P4HA3. The emerging link between TMB and response to immunotherapy (34,35) suggests that the mutation status of P4HA3 may have implications for treatment strategies.

Our experimental validation consistently demonstrated the oncogenic role of P4HA3 in GC cells across multiple models. Lentiviral transduction knockdown significantly suppressed cell proliferation, impaired migratory capacity, and induced apoptosis in GC cells. These findings were further corroborated by immunohistochemical and xenograft mouse models, which consistently demonstrated the tumor-promoting effects of P4HA3. Drug sensitivity analysis also suggests that P4HA3 expression levels in cancer cells could inform therapeutic strategies, potentially guiding the use of targeted agents and optimizing treatment selection for GC patients. The concordance of results across cellular and animal models as well as drug sensitivity analysis strengthens confidence in our conclusions.

Building on our enrichment analysis, which suggested P4HA3’s potential involvement in EMT-related pathways, we experimentally showed that overexpression of P4HA3 in GC cells promotes a pro-metastatic phenotype concomitant with COL1A1 upregulation and shifts toward an EMT-like state. Specifically, overexpression of P4HA3 drove molecular changes characteristic of EMT, including downregulation of E-cadherin and upregulation of N-cadherin and Vimentin. Taken together, these findings suggest that P4HA3 may contribute to GC progression by facilitating a process that involves COL1A1 and EMT. This proposed linkage provides mechanistic rationale for considering P4HA3 as a potential therapeutic target. The hypothesized relationship between P4HA3, COL1A1, and EMT is summarized in the schematic model (Figure S11). Targeting this P4HA3-associated pathway may represent a strategy to disrupt EMT and metastatic dissemination, warranting further exploration.

While our findings nominate P4HA3 as a key player in GC progression, translating this knowledge into clinical utility requires clarifying its distinct potential applications. The high expression of tissue P4HA3, strongly linked to advanced stage and poor prognosis, primarily supports its role as a prognostic biomarker and, mechanistically, a promising therapeutic target. Our drug sensitivity analysis (Figure 11) suggests that P4HA3 expression levels may predict response to certain agents, a hypothesis requiring direct experimental validation. On the other hand, the detectable and disease-associated exosomal P4HA3 in plasma highlights its immediate potential as a non-invasive diagnostic and dynamic monitoring biomarker. To rigorously evaluate its more advanced role as a companion diagnostic—that is, a test to guide specific therapy—future studies must directly link P4HA3 levels with treatment outcomes. To substantiate the therapeutic relevance of P4HA3, the following functional studies are essential: First, in vivo pharmacodynamic studies in xenograft models are needed. This would involve treating mice bearing P4HA3-knockdown versus control tumors with standard chemotherapy (e.g., cisplatin) or pathway-targeted agents (e.g., PI3K/AKT inhibitors). A significantly enhanced therapeutic effect in the knockdown group would provide direct evidence that P4HA3 expression confers treatment resistance, validating it as a combination therapy target. Second, to assess the companion diagnostic potential of exosomal P4HA3, prospective clinical cohorts are required. Plasma should be serially collected from GC patients before and during first-line or targeted therapies. Correlating baseline and on-treatment changes in exosomal P4HA3 levels with objective response rates and survival would determine its utility in predicting and monitoring therapeutic efficacy.

Tissue P4HA3 has been established as a frequently overexpressed oncoprotein across multiple tumor types, work by Niu et al. indicate that the expression of P4HA3 is related to the immune infiltration—particularly of macrophages and NK cells and is positively associated with immune checkpoint genes expression (36), suggesting its potential as an immunotherapeutic target in GC. However, the relationship between P4HA3 and exosomes remains poorly documented. Our findings demonstrate that P4HA3 is upregulated in the plasma exosomes from GC patients and is associated with immunotherapy-relevant pathways. Furthermore, our experiments confirm that tumor cells themselves are the direct source of exosomal P4HA3, and its secretion level is positively correlated with its intracellular expression. This provides mechanistic support for the hypothesis that elevated plasma exosomal P4HA3 directly reflects tumor activity. These insights offer a new strategy avenues for comprehensive GC treatment.

Nevertheless, several limitations of our study should be acknowledged. First, the sample size of our clinical cohort for exosomal analysis, though sufficient for initial discovery, is relatively modest. Future large-scale prospective cohort studies are warranted to validate the diagnostic and prognostic utility of exosomal P4HA3. In addition, the limited size of our clinical cohort constrained the statistical power of multivariable analyses. To mitigate this concern, we conducted a complementary analysis using the large, independent TCGA-STAD dataset, which confirmed that high tissue P4HA3 expression is independently associated with advanced clinical stage. Nonetheless, future prospective studies with larger cohorts and serial measurements are needed to definitively establish its independent prognostic value of exosomal P4HA3. Second, we acknowledge a constraint in our in vivo functional validation. While the knockdown experiment robustly established the necessity of P4HA3 for tumor growth, a complementary overexpression experiment in vivo was not performed. This decision was based on the consideration that the HGC-27 cell line used already exhibits a relatively high basal level of P4HA3 expression, potentially representing an oncogenically activated state where further overexpression might not yield additional growth phenotypes due to pathway saturation. Nevertheless, our in vitro gain-of-function data confirmed that forced P4HA3 expression is sufficient to drive pro-metastatic molecular changes, such as EMT. Future studies utilizing models with low endogenous P4HA3 would be valuable to formally test its sufficiency for tumorigenesis in vivo. Third, it is important to consider the potential influence of anesthetic drugs and the sample size of the in vivo xenograft model. Anesthetics like isoflurane may have longer-term effects on gene expression. While the identical protocol administered for all animals ensures the validity of our comparative analyses, the absolute effects of anesthesia warrant further exploration. Additionally, while the sample size (n=3 per group) is common for preliminary xenograft studies and provided clear phenotypic differences, future studies with larger cohort will strengthen the statistical power and generalizability of these in vivo findings. Finally, future studies should also explore the specific mechanisms by which exosomal P4HA3 is internalized and modulates recipient cell behavior, particularly in the context of immune cell communication.

Conclusions

In conclusion, this study identifies exosomal P4HA3 as a contributing factor in GC progression, demonstrating its role in promoting proliferation, migration and EMT-associated molecular changes. The consistent overexpression of P4HA3 in GC tissues and exosomes is strongly associated with advanced disease stage and unfavorable patient prognosis. Experimentally, modulation of P4HA3 altered EMT markers and metastatic potential, consistent with its putative role in relevant signaling pathways. Silencing P4HA3 robustly suppresses tumor growth in both cellular and animal models. These findings not only deepen the understanding of GC pathogenesis but also highlight exosomal P4HA3 as a promising diagnostic biomarker and a compelling therapeutic target. The insights gained pave the way for developing P4HA3-based diagnostic tools and novel targeted therapeutic strategies for future GC management.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the MDAR and ARRIVE reporting checklists. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0194/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0194/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0194/prf

Funding: This study was supported by the National Natural Science Foundation of China (grant No. 82060561), the Natural Science Foundation of Guangxi Province (grant No. 2018GXNSFBA050047) and Innovation Project of Guangxi Graduate Education (grant No. YCSW2022379).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2026-1-0194/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 protocol was approval by the Ethics Committee of Guilin People’s Hospital (Approval No. 2022-146KY). All procedures were performed in accordance with the ethical standards of the Declaration of Helsinki and its subsequent amendments. Written informed consent was obtained from all individual participants or their legal representatives. For the retrospective analysis of archived tissue specimens, the requirement for individual informed consent was waived by the same ethics committee. All animal experiments were approved by the Experimental Animal Ethics Committee of Guilin Medical College (approval No. GLMC202405186). All procedures were performed in accordance with the institutional guidelines for the care and use of laboratory animals.

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. Chen Y, Jia K, Xie Y, et al. The current landscape of gastric cancer and gastroesophageal junction cancer diagnosis and treatment in China: a comprehensive nationwide cohort analysis. J Hematol Oncol 2025;18:42. [Crossref] [PubMed]
2. Khalili-Tanha G, Khalili-Tanha N, Rouzbahani AK, et al. Diagnostic, prognostic, and predictive biomarkers in gastric cancer: from conventional to novel biomarkers. Transl Res 2024;274:35-48. [Crossref] [PubMed]
3. Zhao L, Chen Z, Zhang F, et al. Changing landscape in stage, treatment and survival of gastric cancer in China 1998-2022: Insights of 40,158 patients from the National Gastric Cancer Cohort. Cancer Lett 2026;642:218261. [Crossref] [PubMed]
4. Huang Y, Huo Y, Huang L, et al. Super-enhancers: Implications in gastric cancer. Mutat Res Rev Mutat Res 2024;793:108489. [Crossref] [PubMed]
5. Kuijper SC, Pape M, Vissers PAJ, et al. Trends in best-case, typical and worst-case survival scenarios of patients with non-metastatic esophagogastric cancer between 2006 and 2020: A population-based study. Int J Cancer 2023;153:33-43. [Crossref] [PubMed]
6. Izumi D, Zhu Z, Chen Y, et al. Assessment of the Diagnostic Efficiency of a Liquid Biopsy Assay for Early Detection of Gastric Cancer. JAMA Netw Open 2021;4:e2121129. [Crossref] [PubMed]
7. Hu C, Xia Y, Zheng Z, et al. AI-based large-scale screening of gastric cancer from noncontrast CT imaging. Nat Med 2025;31:3011-9. [Crossref] [PubMed]
8. Joshi SS, Badgwell BD. Current treatment and recent progress in gastric cancer. CA Cancer J Clin 2021;71:264-79. [Crossref] [PubMed]
9. Cai ZR, Zheng YQ, Hu Y, et al. Construction of exosome non-coding RNA feature for non-invasive, early detection of gastric cancer patients by machine learning: a multi-cohort study. Gut 2025;74:884-93. [Crossref] [PubMed]
10. Shibata C, Nakano T, Yasumoto A, et al. Comparison of CEA and CA19-9 as a predictive factor for recurrence after curative gastrectomy in gastric cancer. BMC Surg 2022;22:213. [Crossref] [PubMed]
11. Guerra-Andrés M, Fernández ÁF, Fontanil T. Exosomes, autophagy, and cancer: A complex triad. Int J Cancer 2025;157:405-15. [Crossref] [PubMed]
12. Hánělová K, Raudenská M, Masařík M, et al. Protein cargo in extracellular vesicles as the key mediator in the progression of cancer. Cell Commun Signal 2024;22:25. [Crossref] [PubMed]
13. Panda SS, Sahoo RK, Patra SK, et al. Molecular insights to therapeutic in cancer: role of exosomes in tumor microenvironment, metastatic progression and drug resistance. Drug Discov Today 2024;29:104061. [Crossref] [PubMed]
14. Saad SH, Kashanchi A, Zadeh MA, et al. Exosome-Mediated Crosstalk Between Cancer Cells and Tumor Microenvironment. Cells 2025;14:1750. [Crossref] [PubMed]
15. Chen C, Luo Y, He W, et al. Exosomal long noncoding RNA LNMAT2 promotes lymphatic metastasis in bladder cancer. J Clin Invest 2020;130:404-21. [Crossref] [PubMed]
16. Wang FW, Cao CH, Han K, et al. APC-activated long noncoding RNA inhibits colorectal carcinoma pathogenesis through reduction of exosome production. J Clin Invest 2019;129:727-43. [Crossref] [PubMed]
17. Xiang Z, Jiang B, Li W, et al. The diagnostic and prognostic value of serum exosome-derived carbamoyl phosphate synthase 1 in HEV-related acute liver failure patients. J Med Virol 2022;94:5015-25. [Crossref] [PubMed]
18. Li A, Hong J, Ma X, et al. Cancer-Derived Exosomal LINC01615 Induces M2 Polarization of Tumor-Associated Macrophages via RBMX-EZH2 Axis to Promote Colorectal Cancer Progression. Int J Nanomedicine 2025;20:7343-58. [Crossref] [PubMed]
19. Zhou H, Huang S, Shao C, et al. miR-1266-3p Suppresses Epithelial-Mesenchymal Transition in Colon Cancer by Targeting P4HA3. Anal Cell Pathol (Amst) 2022;2022:1542117. [Crossref] [PubMed]
20. Yusuf M, Gaskins J, Trawick E, et al. Effects of adjuvant radiation therapy on survival for patients with resected primary tracheal carcinoma: an analysis of the National Cancer Database. Jpn J Clin Oncol 2019;49:628-38. [Crossref] [PubMed]
21. Wang Q, Meng D, Shen S, et al. P4HA3 promotes head and neck squamous cell carcinoma progression via the WNT/β-catenin signaling pathway. Pathol Res Pract 2024;260:155481. [Crossref] [PubMed]
22. Zhang J, Liu S, Zhang J, et al. Helicobacter pylori induced miR-362-5p upregulation drives gastric cancer progression and links hepatocellular carcinoma through an exosome-dependent pathway. Front Cell Infect Microbiol 2025;15:1582131. [Crossref] [PubMed]
23. Nakasuka F, Tabata S, Sakamoto T, et al. TGF-β-dependent reprogramming of amino acid metabolism induces epithelial-mesenchymal transition in non-small cell lung cancers. Commun Biol 2021;4:782. [Crossref] [PubMed]
24. Zhou H, Zou J, Shao C, et al. Prolyl 4-hydroxylase subunit alpha 3 facilitates human colon cancer growth and metastasis through the TGF-β/Smad signaling pathway. Pathol Res Pract 2022;230:153749. [Crossref] [PubMed]
25. Wu Y, Zhang B, Nong J, et al. Systematic pan-cancer analysis of the potential tumor diagnosis and prognosis biomarker P4HA3. Front Genet 2023;14:1045061. [Crossref] [PubMed]
26. Shen DD, Pang JR, Bi YP, et al. LSD1 deletion decreases exosomal PD-L1 and restores T-cell response in gastric cancer. Mol Cancer 2022;21:75. [Crossref] [PubMed]
27. Chen Y, Wang B, Zhao Y, et al. Metabolomic machine learning predictor for diagnosis and prognosis of gastric cancer. Nat Commun 2024;15:1657. [Crossref] [PubMed]
28. Manoochehri H, La'ah AS, Babaeizad A, et al. Extracellular vesicles in cancer immunotherapy: Therapeutic, challenges and clinical progress. Asian J Pharm Sci 2025;20:101065. [Crossref] [PubMed]
29. Guo T, Tang XH, Gao XY, et al. A liquid biopsy signature of circulating exosome-derived mRNAs, miRNAs and lncRNAs predict therapeutic efficacy to neoadjuvant chemotherapy in patients with advanced gastric cancer. Mol Cancer 2022;21:216. [Crossref] [PubMed]
30. Zhang Z, Zhang Y, Zhang R. P4HA3 promotes clear cell renal cell carcinoma progression via the PI3K/AKT/GSK3β pathway. Med Oncol 2023;40:70. [Crossref] [PubMed]
31. Long R, Liu Z, Li J, et al. COL6A6 interacted with P4HA3 to suppress the growth and metastasis of pituitary adenoma via blocking PI3K-Akt pathway. Aging (Albany NY) 2019;11:8845-59. [Crossref] [PubMed]
32. Xin L, Wu Y, Liu C, et al. Exosome-mediated transfer of lncRNA HCG18 promotes M2 macrophage polarization in gastric cancer. Mol Immunol 2021;140:196-205. [Crossref] [PubMed]
33. Derynck R, Turley SJ, Akhurst RJ. TGFβ biology in cancer progression and immunotherapy. Nat Rev Clin Oncol 2021;18:9-34. [Crossref] [PubMed]
34. Li R, Han D, Shi J, et al. Choosing tumor mutational burden wisely for immunotherapy: A hard road to explore. Biochim Biophys Acta Rev Cancer 2020;1874:188420. [Crossref] [PubMed]
35. Yu Y, Luo K, Liu M, et al. Comprehensive analysis reveals that P4HA3 is a prognostic and diagnostic gastric cancer biomarker that can predict immunotherapy efficacy. Sci Rep 2024;14:22959. [Crossref] [PubMed]
36. Niu X, Ren L, Wang S, et al. High Prolyl 4-Hydroxylase Subunit Alpha 3 Expression as an Independent Prognostic Biomarker and Correlated With Immune Infiltration in Gastric Cancer. Front Genet 2022;13:952335. [Crossref] [PubMed]