Researchers have identified distinct yet partially overlapping transcriptional interactions between immune checkpoint and epithelial-mesenchymal transition pathways in esophageal squamous cell carcinoma. Published in the Egyptian Journal of Medical Human Genetics, this integrative bioinformatic study provides a systems-level characterization of how these biological processes intersect within the tumor microenvironment. The findings highlight the complexity of immune regulation and tumor progression, suggesting that single-parameter biomarkers may be insufficient to capture the full landscape of esophageal cancer biology.
The study focused on a predefined hypothesis-driven gene panel comprising five immune checkpoint genes and two epithelial-mesenchymal transition markers. The immune checkpoint genes analyzed were PDCD1, CD274, CTLA-4, HAVCR2, and LAG-3, while the EMT markers included TWIST1 and MMP13. This targeted selection strategy was designed to computationally extend clinically validated markers rather than to perform an exhaustive pathway-wide screen, allowing for a focused examination of specific molecular interactions in esophageal squamous cell carcinoma.
To analyze these interactions, the researchers utilized transcriptomic, somatic mutation, and clinical data from The Cancer Genome Atlas esophageal squamous cell carcinoma cohort, which included 95 tumors. The analytical framework combined these data with external cohort validation using an independent microarray dataset known as GSE53624. This approach allowed the team to assess the robustness of their findings across different measurement platforms and patient populations, ensuring that observed patterns were not artifacts of a single dataset.
Immune cell infiltration was estimated using multiple computational deconvolution algorithms, including CIBERSORT, TIMER, xCell, EPIC, and MCP-counter. These tools were applied to characterize the tumor immune microenvironment by estimating the abundance of various immune and stromal cell types from bulk RNA-seq data. The study also assessed tumor mutational burden, ESTIMATE scores, and TIDE scores as exploratory immune-related metrics to infer potential immune evasion patterns and tumor purity.
A key finding of the study was the distinctive behavior of HAVCR2, also known as TIM-3, which demonstrated the strongest and most consistent association with EMT-related signatures across both datasets. HAVCR2 showed statistically significant correlations with both TWIST1 and MMP13, indicating a strong link between this immune checkpoint gene and EMT processes. This observation aligns with independent evidence from other cancer types suggesting that HAVCR2 may be implicated in EMT activation and metastatic phenotypes, supporting models where immune and EMT pathways partially converge.
The analysis revealed that immune checkpoint-high tumors were characterized by features of an immunologically active yet regulatory-enriched microenvironment. These tumors exhibited higher levels of activated immune subsets, including CD8+ T cells, memory CD4+ T cells, and M1 macrophages, consistent with an immune-infiltrated phenotype. However, high immune checkpoint expression was also associated with increased infiltration of regulatory T cells, whose immunosuppressive roles are well documented in solid tumors. This simultaneous presence of effector and suppressor immune components may help explain why high immune checkpoint expression correlated with poorer overall survival despite apparent immune activation.
In contrast, EMT-related genes were linked to fibroblast- and stroma-dominant profiles. EMT-high tumors were enriched in naïve CD4+ T cells, resting memory CD4+ T cells, M0 macrophages, and resting mast cells. Cancer-associated fibroblast levels showed strong positive correlations with EMT-related gene expression, and HAVCR2 clustered with EMT genes in the context of fibroblast infiltration. Although higher fibroblast levels are generally associated with worse outcomes in esophageal cancer, this analysis suggested a more nuanced pattern where fibroblasts were unexpectedly linked to improved overall survival in this specific cohort, underscoring the functional heterogeneity of stromal populations.
Survival analysis indicated that elevated expression of most immune checkpoint genes and MMP13 was associated with poorer overall survival. PDCD1 demonstrated a similar trend in the same direction, although it did not reach conventional statistical significance. In contrast, TWIST1 exhibited dataset-dependent prognostic behavior; while higher TWIST1 expression was associated with improved survival in the TCGA cohort, it was associated with worse survival in the GSE53624 validation cohort. These discrepancies may reflect biological heterogeneity, differences in treatment exposure, and platform-specific measurement effects inherent to transcriptomic technologies.
Tumor mutational burden was found to be low overall in the esophageal squamous cell carcinoma cohort, with a mean of 3.1 mutations per megabase. TMB demonstrated a decreasing trend across pathological stages, from stage I to IV, and was not significantly associated with overall survival. This finding suggests that TMB may not be a reliable standalone predictor of immunotherapy response in esophageal squamous cell carcinoma, highlighting the need for broader immune and stromal signatures in biomarker development.
Advanced-stage tumors displayed concurrent immune activation and immune evasion signals, indicating a complex evolution of the tumor microenvironment. Stage-associated transcriptomic analyses revealed deregulation of pathways involved in cytoskeletal remodeling, actin filament organization, and thrombin-activated signaling in higher-stage tumors. These changes are central to tumor motility and EMT progression, suggesting that ESCC progression involves coordinated evolution of immune suppression, stromal activation, and transcriptional reprogramming.
The study acknowledges several limitations, including the exploratory nature of the associations and the reliance on computational analyses of publicly available datasets. The selected gene panel was focused on prior clinical relevance rather than comprehensive pathway representation, which limits pathway-level generalizability. Additionally, survival analyses were constrained by limited event numbers relative to covariates, restricting the stability of multivariable modeling. The researchers emphasize that all associations reported are hypothesis-generating and require mechanistic and prospective clinical validation prior to translational application.
These findings delineate the context-dependent transcriptional interactions between immune checkpoint and EMT pathways in esophageal squamous cell carcinoma. By highlighting the complexity of the tumor immune microenvironment, the study provides a structured framework for future mechanistic and translational studies. The results suggest that understanding the interplay between immune regulation and EMT programs may offer new insights into therapeutic strategies for esophageal cancer, particularly in earlier disease stages where immunotherapy options remain limited.
The study was approved by the Ethics Committee of Mashhad University of Medical Sciences in Mashhad, Iran, under ethical code IR.MUMS.FHMPM.REC.1401.192. The research team utilized RNA-seq expression values processed by RSEM-normalized expression values, which were log2-transformed for subsequent analyses. To ensure the robustness of the gene expression measurements, the researchers applied a rigorous quality-control process to the TCGA data. This included filtering cases based on histopathological annotations to retain only esophageal squamous cell carcinoma tumors while excluding esophageal adenocarcinoma samples. When multiple tumor samples were available for a single patient, duplicate specimens were removed to retain a single primary tumor sample per patient. Samples lacking essential RNA-seq expression or survival data were excluded prior to downstream analyses, resulting in the final cohort of 95 tumors used for the primary investigation.
For the validation cohort, the researchers analyzed the GSE53624 dataset, which comprised 119 paired tumor and adjacent normal esophageal squamous cell carcinoma tissues. Because explicit probe-to-gene annotation was not available in a sufficiently detailed format in the original Agilent annotation, the team extracted raw probe sequences and converted them to FASTA format. Only probes annotated as protein-coding mRNA with complete sequence information were retained, while those with missing, ambiguous, or non-protein-coding sequences were excluded. The retained probe sequences were aligned against human protein-coding transcript references using the HISAT2 tool within the Galaxy platform. This re-annotation procedure achieved a 99.25% overall alignment rate, with 93.61% of probes uniquely mapped, supporting the robustness of the data. Transcript identities were assigned using the GENCODE v19 human protein-coding transcript reference, and probes mapping to the same transcript were summarized at the gene level by averaging expression values.
Statistical analyses were conducted using R version 4.0.2, with data visualization performed using pheatmap, ggplot2, and ggcorrplot2. Normality of the data was assessed using the Shapiro-Wilk test, and depending on the distribution, Welch’s t-test or Wilcoxon rank-sum test was used for two-group comparisons. For multiple-group comparisons, ANOVA or Kruskal-Wallis tests were applied, while associations between categorical variables were evaluated using chi-square tests. Correlations between continuous variables were assessed using Spearman’s rank correlation coefficient. Survival analysis was performed using Kaplan-Meier methods, with expression cutoffs primarily determined using maximally selected rank statistics. To assess the robustness of dichotomization-based findings, Cox proportional hazards models incorporating continuous expression values were fitted as sensitivity analyses. Statistical significance was defined as P < 0.05, and to control for multiple testing, analyses were organized into predefined analytical families with FDR correction using the Benjamini-Hochberg method.
Differential gene expression analysis comparing stage II versus stage I tumors identified 11 significantly downregulated genes and 156 upregulated genes. A Venn diagram analysis revealed four commonly upregulated genes, including ALMS1, HLTF, CHD6, and STMN1, and nineteen commonly downregulated genes across stages II and III relative to stage I. Functional enrichment analysis indicated that upregulated genes were enriched in pathways related to actin cytoskeleton organization, thrombin-activated receptor signaling, and chromatin remodeling. Conversely, downregulated genes were enriched in pathways related to carbohydrate transport and absorption, lipid homeostasis, natural killer and T cell activity, angiogenesis, and protein unfolding. These transcriptomic changes suggest that higher-stage tumors undergo significant reprogramming of cellular processes, particularly those involved in structural integrity and immune signaling.
Exploratory gene-drug interaction analyses identified several compounds linked to stage-associated genes, including Paclitaxel, Tropisetron, Vinorelbine, and Atropine. These associations were derived from curated databases and represent exploratory computational predictions intended to generate hypotheses for future experimental investigation. The researchers emphasized that these gene-drug interactions do not constitute clinically actionable treatment recommendations, as no pharmacokinetic, pharmacodynamic, or tumor-context modeling was performed. The study also noted that patients in the GSE53624 cohort who received adjuvant therapy had significantly poorer overall survival compared to those who did not, with median survival times of 31.5 versus 61.0 months. This finding suggests that treatment-related variables may influence gene-survival relationships, and in patients who did not receive adjuvant therapy, gene-expression patterns more closely resembled the TCGA findings.
The study highlights the inherent limitations of bulk transcriptomics in resolving functional heterogeneity among phenotypically similar immune cells. Single-cell analyses in esophageal squamous cell carcinoma have previously revealed extensive T-cell exhaustion characterized by high PDCD1 expression within specific subsets, suggesting that increased immune checkpoint expression across tumor stages may reflect shifts in cellular states rather than simple changes in cell abundance. Spatial immune profiling studies have also demonstrated that cellular proximity and spatial organization may predict therapeutic response better than cell counts alone. These observations reinforce the importance of integrating single-cell and spatial technologies in future immune profiling efforts to better understand the complex interactions within the tumor microenvironment.
Subject of Research: Transcriptional interactions between immune checkpoint and EMT pathways in esophageal squamous cell carcinoma
Article Title: Context-dependent interactions between immune checkpoint and epithelial-mesenchymal transition programs shape the tumor microenvironment in esophageal squamous cell carcinoma
Article References: Montazer, M., Fardi Golyan, F., Ramian Yarasmi, M., Hossein, F., Gholamin, M., & Mahmoudian, R. A. (2026). Context-dependent interactions between immune checkpoint and epithelial-mesenchymal transition programs shape the tumor microenvironment in esophageal squamous cell carcinoma. Egyptian Journal of Medical Human Genetics, 27(1), Article 68. https://doi.org/10.1186/s43042-026-00897-7
Image Credits: AI Generated
DOI: 10.1186/s43042-026-00897-7
Keywords: esophageal squamous cell carcinoma, immune checkpoint, epithelial-mesenchymal transition, tumor microenvironment, bioinformatics, HAVCR2, TWIST1, tumor mutational burden, immune infiltration, cancer progression, Context-dependent, interactions
News Source: Nathaniel Bowman. (November 8, 2026). Immune Checkpoint and EMT Interactions Shape Esophageal Tumor Microenvironment. Scienmag.



