---
title: "TMED3 and a Disulfidptosis-Linked Diagnostic Signature in Intrahepatic Cholangiocarcinoma"
id: "frontiers-in-immunology-5-a-tmed3-governed-disulfidptosis-related-diagnostic-signature-reveals-tumor"
canonical_url: "https://medichelpline.com/clinical-feed/frontiers-in-immunology-5-a-tmed3-governed-disulfidptosis-related-diagnostic-signature-reveals-tumor"
content_type: "clinical_feed_article"
specialty: "Oncology"
source_name: "Frontiers in Immunology"
source_url: "https://www.frontiersin.org/articles/10.3389/fimmu.2026.1858117"
published_at: "2026-07-29T00:00:00.000Z"
evidence_level: "Journal Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# TMED3 and a Disulfidptosis-Linked Diagnostic Signature in Intrahepatic Cholangiocarcinoma
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/frontiers-in-immunology-5-a-tmed3-governed-disulfidptosis-related-diagnostic-signature-reveals-tumor
- **Specialty:** [Oncology](https://medichelpline.com/clinical-feed/oncology.md)
- **Primary Source:** Frontiers in Immunology
- **Source URL:** [Original Journal Publication](https://www.frontiersin.org/articles/10.3389/fimmu.2026.1858117)
- **Published At:** 2026-07-29T00:00:00.000Z
- **Evidence Rating:** Journal Feed
## Executive GIST (TL;DR)
- The provided source text did not include the article body or study data; only site navigation and metadata were available. - The article title indicates a focus on **TMED3**, a **disulfidptosis**-related diagnostic signature, and remodeling of the **tumor microenvironment** in **intrahepatic cholangiocarcinoma** (iCCA). - No methodological details, patient cohorts, datasets, bioinformatic pipelines, or experimental results were reported in the provided source content. - No statistical outcomes, performance metrics for any diagnostic signature, or validation steps were available in the supplied text. - No mechanistic data or functional experiments on **TMED3** or the biological role of **disulfidptosis** in iCCA were included in the source extract. - The source did not report implications for prognosis, therapy selection, immunotherapy response, or clinical utility of the proposed signature. - Because the article content was not present in the provided source, specific recommendations, numeric results, or study limitations cannot be summarized from this input. - Readers should consult the original Frontiers in Immunology article for complete methods, results, figures, and authors' conclusions; those details were not available here.
## Clinical Analysis & Structured Key Points
Frontiers | A TMED3-governed disulfidptosis-related diagnostic signature reveals tumor microenvironment remodeling in intrahepatic cholangiocarcinoma ORIGINAL RESEARCH article Front. Immunol. , 29 July 2026 Sec. Cancer Immunity and Immunotherapy Volume 17 - 2026 | https://doi.org/10.3389/fimmu.2026.1858117 Published in Frontiers in Immunology Cancer Immunity and Immunotherapy 7 impact factor 11.3 citescore Part of a Research Topic Community series in integrating molecular mechanisms, immunotherapy, and drug sensitivity in cancer immunology and oncology: Volume II Submission open 31k views 17 articles Editor & Reviewers Edited by Q Z Qi Zhang Reviewed by L Y Longkuan Yin X L Xiaolong Liu Outline Figures and Tables Figure 1 View in article Figure 2 View in article Figure 3 View in article Figure 4 View in article Figure 5 View in article Figure 6 View in article Figure 7 View in article Figure 8 View in article Figure 9 View in article Figure 10 View in article Figure 11 View in article Table 1 Candidate therapeutic agents with significant proximity to the TMED3-correlated gene set. View in article ORIGINAL RESEARCH article Front. Immunol. , 29 July 2026 Sec. Cancer Immunity and Immunotherapy Volume 17 - 2026 | https://doi.org/10.3389/fimmu.2026.1858117 A TMED3-governed disulfidptosis-related diagnostic signature reveals tumor microenvironment remodeling in intrahepatic cholangiocarcinoma W Q Wanjia Qiao 1 Y H Yixiang He 1 J L Jing Li 1 X L Xiaohan Liu 1,2 L Z Lingfang Zhang 1,3 X B Xin Bai 1 Y W Yeying Wang 1,2 * J T Jianming Tang 1,2 * 1. The First Clinical Medical College of Lanzhou University, Lanzhou, China 2. The First Hospital of Lanzhou University, Lanzhou, China 3. Gansu Provincial Hospital, Lanzhou, China See more Article metrics View details Abstract Background: Intrahepatic cholangiocarcinoma (ICC) is an aggressive malignancy with poor prognosis and limited treatment options. Disulfidptosis, a novel cell death pathway driven by disulfide bond accumulation, has emerged as a potential mechanism in cancer biology; however, its role in ICC remains unclear. Methods: We integrated single−cell RNA sequencing (GSE138709) with bulk transcriptomic datasets (TCGA−CHOL, GSE107943, GSE32225) to systematically characterize the ICC cellular landscape. Analyses included CNV inference, stemness scoring, disulfidptosis activity assessment, and cell−cell communication profiling. A diagnostic model was constructed using LASSO−logistic regression with 10−fold cross−validation and validated in independent cohorts. TME characterization, survival analysis, and drug−target screening were also performed. Experimental validation included HPA immunohistochemistry, qRT−PCR, and functional assays following TMED3 knockdown. Results: Seven major cell types were identified, with malignant cholangiocytes exhibiting high aneuploidy (74%), elevated stemness, upregulated disulfidptosis activity, and extensive communication via SPP1−CD44 and IGFBP3−TMEM219 networks. A five−gene signature (TMED3, TMEM184B, MAPK13, MFSD10, GRB7) demonstrated robust diagnostic performance. Survival analysis showed borderline prognostic value for TMED3 (adjusted HR = 2.37, P = 0.073), while TMEM184B emerged as an independent prognostic factor (adjusted HR = 4.79, P = 0.028). PPI and co−expression analyses established links between signature genes and disulfidptosis regulators. Functional experiments confirmed that TMED3 knockdown suppressed ICC cell proliferation, migration, and enhanced sensitivity to glucose deprivation−induced disulfidptosis. Network−based drug screening identified eight high−priority candidates for therapeutic repurposing. Conclusion: This study provides a comprehensive single−cell atlas of ICC, identifies TMED3 as a key regulator of a disulfidptosis−related diagnostic signature, and demonstrates its functional role in promoting ICC malignancy. The five−gene signature shows diagnostic and prognostic promise, and the drug screening offers preliminary leads for therapeutic repurposing, providing a foundation for precision diagnosis and targeted therapy in ICC. 1 Introduction Cholangiocarcinoma (CCA), a malignant neoplasm originating from the biliary epithelium, is classified according to anatomical location into intrahepatic cholangiocarcinoma (iCCA), perihilar cholangiocarcinoma (pCCA), and distal cholangiocarcinoma (dCCA) ( 1 ). In recent years, the global incidence and mortality rates of iCCA have been on a continuous upward trend. As the second most prevalent primary liver cancer, iCCA constitutes approximately 10%–15% of all primary hepatic malignancies ( 2 – 4 ). Epidemiological data from 32 countries indicate a rising global mortality rate for iCCA ( 5 ). The prognosis remains extremely poor, with a 5-year survival rate of less than 20% ( 2 , 6 ). Due to nonspecific early symptoms and the absence of reliable screening biomarkers, the majority of patients are diagnosed at an advanced stage, precluding curative surgical resection. For these advanced cases, the standard first-line chemotherapy regimen, namely gemcitabine plus cisplatin, provides only limited clinical benefit ( 7 ). Although recent advances in combination immunotherapy, such as pembrolizumab with gemcitabine-cisplatin, have demonstrated modest improvements in outcomes ( 8 ), there is still a critical need to develop more effective therapeutic strategies to improve survival. The high heterogeneity of iCCA represents a major contributor to its therapeutic challenges. This heterogeneity manifests not only across patients (inter-tumoral heterogeneity) but also within individual tumor lesions (intra-tumoral heterogeneity) ( 9 , 10 ). Conventional bulk transcriptome sequencing yields only averaged gene expression profiles from heterogeneous cell populations, thereby limiting the ability to resolve the diverse cellular constituents and their dynamic interactions within the tumor microenvironment (TME) ( 11 ). In recent years, advances in single-cell RNA sequencing (scRNA-seq) technology have enabled the systematic dissection of the TME in iCCA at single-cell resolution, encompassing malignant and non-malignant epithelial cells, immune cells (including T cells, B cells, and macrophages), and stromal cells such as cancer-associated fibroblasts and endothelial cells ( 10 , 12 ). Notably, studies have identified six distinct fibroblast subsets in iCCA, revealing critical interactions between malignant cells and CD146-positive vascular cancer-associated fibroblasts mediated by IL-6 signaling ( 10 ). Moreover, recent research has delineated two molecular subtypes of iCCA, namely S100P + SPP1 − and S100P − SPP1 + , which are characterized by distinct tumor microenvironment compositions and clinical outcomes ( 13 ). The inference of somatic copy number variations (CNVs) from scRNA-seq data facilitates the direct identification of malignant cells at the single-cell level, offering novel insights into tumor clonal evolution. Within the TME of iCCA, malignant cholangiocytes engage in complex molecular crosstalk with the surrounding stromal and immune cells, and this collective interaction promotes tumor progression. Among these interactions, cellular metabolic reprogramming has emerged as a hallmark of cancer. Disulfidptosis is a recently identified form of regulated cell death. It is induced by the abnormal accumulation of intracellular disulfide bonds, which leads to the aberrant cross-linking of actin cytoskeletal proteins and subsequent cytoskeletal collapse ( 14 ). This process is particularly prevalent in cancer cells exhibiting high glycolytic activity and elevated expression of SLC7A11 (solute carrier family 7 member 11). These cells rely on SLC7A11 to import extracellular cystine, which is subsequently reduced to cysteine through the consumption of NADPH (nicotinamide adenine dinucleotide phosphate) ( 15 ). Under conditions of NADPH deficiency or impaired utilization, cystine accumulates intracellularly, driving excessive disulfide bond formation in actin-associated proteins, disrupting the actin network, and ultimately triggering disulfidptosis ( 16 ). However, the role of disulfidptosis in the pathogenesis and progression of iCCA, particularly its activation status within specific malignant cholangiocyte subpopulations and its interplay with components of the TME, remains poorly understood. In summary, it is of great significance to leverage scRNA-seq technology to comprehensively analyze the cellular composition, intercellular communication, and functional states of iCCA, with a focus on emerging biological processes such as disulfidptosis. This approach can contribute to elucidating the pathogenesis of iCCA and identifying novel therapeutic targets and diagnostic biomarkers. This study aims to comprehensively and systematically characterize the cellular landscape of iCCA through the integration of single-cell and bulk transcriptomic data. It also intends to identify key malignant cell populations, investigate their metabolic vulnerabilities, especially the susceptibility to disulfidptosis. Ultimately, it aims to develop molecular signatures with diagnostic and prognostic value, thereby providing a robust theoretical foundation and candidate targets for the precise diagnosis and treatment of iCCA. 2 Materials and methods 2.1 Data acquisition and preprocessing Bulk RNA-seq expression profiles and corresponding clinical data for cholangiocarcinoma (CHOL) patients were obtained from The Cancer Genome Atlas (TCGA) database ( https://portal.gdc.cancer.gov/ ). Only ICC samples and matched adjacent normal tissue samples were retained for downstream analysis. Additional gene expression datasets, GSE107943 and GSE32225, were downloaded from the Gene Expression Omnibus (GEO) database ( https://www.ncbi.nlm.nih.gov/geo/ ). For descriptive purposes and to facilitate candidate gene discovery, the three datasets (TCGA, GSE107943, and GSE32225) were integrated into a unified dataset after batch effect correction using the ComBat algorithm implemented in the sva R package (version 3.42.0), with sample status (tumor/adjacent normal) included as a covariate to preserve biological variation. This integrated dataset (referred to as the combined dataset) comprised 215 tumor samples and 42 adjacent normal tissue samples. Importantly, this dataset was used solely for descriptive analyses (e.g., ESTIMATE scoring, differential expression for candidate gene discovery) and was not used for model training or validation. To verify the effectiveness of batch correction, principal component analysis (PCA) was performed on the combined expression matrix before and after ComBat correction, with samples colored by dataset origin. As shown in Supplementary Figures 2A, B , samples from the three datasets exhibited clear separation prior to correction, indicating substantial batch effects, while they became well-intermixed after correction, confirming successful removal of batch effects. For model development, GSE107943 was used as the training set, while TCGA and GSE32225 served as two independent validation sets. No batch correction was applied across the training and validation sets to avoid data leakage. Each dataset was normalized separately prior to analysis: microarray datasets (GSE107943 and GSE32225) were log 2 -transformed and quantile-normalized, while TCGA RNA-seq raw counts were converted to log 2 (TPM + 1) to achieve a scale comparable to the microarray data. 2.2 Single-cell RNA sequencing data processing The single-cell RNA sequencing (scRNA-seq) dataset GSE138709, containing ICC samples, was retrieved from the GEO database. This dataset includes eight samples: five tumor specimens and three matched adjacent normal tissues. Raw data were processed using the Seurat R package (version 5.1.0) for quality control and preprocessing. Mitochondrial gene content was calculated using the Percentage FeatureSet function ( 17 ). Cells were filtered based on the following criteria: (1) detection of more than 250 expressed genes; (2) mitochondrial gene percentage below 10%; (3) unique molecular identifier (UMI) counts exceeding 250; and (4) log10(GenesPerUMI) > 0.8, where GenesPerUMI is defined as nFeature_RNA divided by nCount_RNA. To avoid potential artifacts from doublets—two or more cells captured within the same droplet—we performed doublet detection and removal using the DoubletFinder package (version 3.0) prior to downstream analysis. Briefly, artificial doublets were generated from the processed Seurat object, and the proportion of artificial nearest neighbors (pANN) was calculated for each real cell following PCA dimensionality reduction. The parameters were set as follows: pN = 0.25, the optimal pK was determined using the bimodality coefficient method, the expected doublet rate was estimated as 0.8% per 1,000 cells loaded according to 10x Genomics guidelines, and the homotypic doublet proportion was adjusted using the modelHomotypic function. Cells predicted as doublets were excluded from all subsequent analyses. Detailed doublet statistics for each sample, including raw cell numbers, expected doublet rates, detected doublets, and post-filtering cell counts, are provided in Supplementary Table 1 , and diagnostic plots (pANN score distributions and UMAP visualization of doublet vs. non-doublet cells) are shown in Supplementary Figure 1 . 2.3 Batch effect correction and dimensionality reduction Batch effects across samples were corrected using the Harmony integration method via the IntegrateLayers function in Seurat. Dimensionality reduction was performed through principal component analysis (PCA), with 30 principal components selected based on the elbow plot. Cell clustering was conducted using the FindNeighbors and FindClusters functions at a resolution of 0.2, yielding 13 distinct cell clusters. Visualization was carried out using Uniform Manifold Approximation and Projection (UMAP) and t-distributed Stochastic Neighbor Embedding (t-SNE). 2.4 Cell type annotation Cell type annotation was conducted based on canonical marker genes derived from previous studies ( 10 , 18 – 27 ). Seven major cell lineages were identified: cholangiocytes, T cells, NKT cells, B cells, myeloid cells, endothelial cells, and fibroblasts. NKT cells were defined as cells co−expressing CD3E (pan−T cell marker) and NCAM1 (CD56, a classical NK cell marker). Conventional T cells were identified as CD3E+NCAM1−, and NK cells as CD3E−NCAM1+.Finer sub−clusters (e.g., plasma B cells and myeloid sub−clusters shown in Figure 1C ) are nested within these seven major lineages and are presented only to illustrate detailed heterogeneity; for quantitative comparisons and model construction, the major−type annotation was used. Differential expression analysis to identify cluster−specific marker genes was performed using the FindAllMarkers function in Seurat with thresholds of fold change (FC) > 2 and false discovery rate (FDR) < 0.05. Functional enrichment analysis of the marker genes was performed using the clusterProfiler R package to characterize biological processes and pathways associated with each cell type. Figure 1 Single-cell transcriptomic analysis of the cellular composition and heterogeneity in the intrahepatic cholangiocarcinoma (iCCA) tumor microenvironment. (A) Schematic workflow illustrating quality control and integration of single-cell transcriptomic data from eight iCCA samples, including five tumor tissues and three paired adjacent non-tumor tissues. (B) UMAP visualization of single-cell clustering results, revealing 13 distinct transcriptional subpopulations. (C) Cell type annotation of the 13 subclusters based on canonical marker gene expression, classifying them into seven major cell types: cholangiocytes, T cells, B cells, macrophages, endothelial cells, fibroblasts, and other immune cells. Plasma B cells and myeloid sub−clusters are sub−types of the seven major cell lineages (see Methods). (D) Heatmap displaying the expression profiles of cell type–specific marker gene (rows: marker genes; columns: individual cells). (E) Bar graph depicting the proportional distribution of each cell type across the eight individual samples, highlighting inter-sample heterogeneity in cellular composition. (F) Comparative assessment of major cell type proportions between tumor and adjacent tissues, with notable alterations observed in cholangiocytes and NKT cells. (G) Schematic illustration of copy number variation (CNV) inference using inferCNV, with adjacent tissue cells serving as diploid reference controls. (H) Pie chart summarizing CNV status in annotated cholangiocytes, indicating that 74% exhibit aneuploid genomic profiles. (I) Quantitative comparison of aneuploidy frequency in cholangiocytes derived from tumor versus adjacent tissues. 2.5 Copy number variation analysis Copy number variation (CNV) inference was performed using the copyKAT algorithm to differentiate malignant from normal cells. Based on the analysis results, cells were classified as aneuploid (CNV-altered) or diploid (CNV-normal). The proportion of aneuploid cells was calculated for each cell type and compared between tumor and adjacent normal tissues. 2.6 Cell stemness analysis Cellular stemness was assessed using the CytoTRACE algorithm, which infers cellular differentiation states from single-cell gene expression profiles. Stemness scores were computed for each cell, with higher values indicating greater stemness and lower differentiation potential. Comparisons of stemness levels were performed across cell types and between tumor and adjacent normal tissue groups. 2.7 Disulfidptosis scoring analysis Disulfidptosis-related gene sets were curated from previously published literature ( 14 , 28 – 30 ). To evaluate disulfidptosis activity at the single-cell level, we calculated pathway activity scores using six independent gene set enrichment algorithms implemented in the irGSEA R package: AUCell, UCell, singscore, ssGSEA, JASMINE, and viper ( 31 , 32 ). Specifically, AUCell and UCell are ranking-based methods that assess pathway activity by evaluating gene expression ranks within each cell; singscore and ssGSEA are enrichment-based approaches that compute scores based on the relative expression levels of gene sets; and JASMINE and viper are network-based algorithms that infer pathway activity by integrating regulon information. These six algorithms were selected to provide a comprehensive and cross-validated assessment of disulfidptosis activity, each offering complementary strengths in scoring sensitivity and robustness. The scores obtained from each algorithm were then compared across different cell types to characterize cell-type-specific disulfidptosis activity patterns, as well as between tumor-derived and adjacent normal cholangiocytes to evaluate disulfidptosis activity changes associated with malignant transformation. Results from singscore, ssGSEA, and viper are presented in Supplementary Figures 3 – S5 , respectively. 2.8 Pathway scoring and transcription factor analysis Pathway activity scores for immune, metabolic, signaling, and proliferation-related pathways were calculated using six gene set enrichment algorithms implemented in the irGSEA R package: AUCell, UCell, singscore, ssGSEA, JASMINE, and viper. Specifically, AUCell and UCell are ranking-based methods that assess pathway activity by evaluating gene expression ranks within each cell; singscore and ssGSEA are enrichment-based approaches that compute scores based on the relative expression levels of gene sets; and JASMINE and viper are network-based algorithms that infer pathway activity by integrating regulon information. These six algorithms were selected to provide a comprehensive and cross-validated assessment of pathway activity, each offering complementary strengths in scoring sensitivity and robustness. For brevity, the main figures present results from three representative algori
## Related Clinical Research

- [PD-L1 expression and survival in unresectable or recurrent gastric cancer treated with first-line](https://medichelpline.com/clinical-feed/british-journal-of-cancer-0-pd-l1-expression-and-survival-in-unresectable-recurrent-gastric-cancer-treated.md)
- [CD39+CD103+CD8+ T cells (DP) improve neoadjuvant chemoimmunotherapy response in HNSCC](https://medichelpline.com/clinical-feed/british-journal-of-cancer-0-cd39-cd103-cd8-t-cells-enhance-neoadjuvant-chemoimmunotherapy-efficacy-in-head.md)
- [Precision trial design for FGFR2‑rearranged cholangiocarcinoma: lessons from FIGHT-302](https://medichelpline.com/clinical-feed/nature-reviews-clinical-oncology-0-precision-oncology-needs-precision-trial-design-lessons-from-fgfr2-rearranged.md)
- [Checkpoint immunotherapy in gastrointestinal cancer — editorial (Volume II) overview](https://medichelpline.com/clinical-feed/frontiers-in-immunology-17-editorial-checkpoint-immunotherapy-reshaping-the-landscape-of-gastrointestinal.md)
- [Perioperative enfortumab vedotin plus pembrolizumab improves outcomes in cisplatin-eligible MIBC](https://medichelpline.com/clinical-feed/nature-reviews-clinical-oncology-0-benefit-from-perioperative-enfortumab-vedotin-pembrolizumab-in-cisplatin.md)

## Navigation
- [← Back to Oncology Feed](https://medichelpline.com/clinical-feed/oncology.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.