Identification of telomere-related genes in the progression of colon adenocarcinoma: a bioinformatics analysis

Introduction

According to GLOBOCAN 2020, colon adenocarcinoma (COAD) ranks second in mortality and third in morbidity following breast cancer and lung cancer globally (1). Due to rising morbidity and mortality in China, COAD poses an increasing threat to people’s health. In 2020, the incidence and mortality of COAD ranked second and fifth, respectively, among malignancies in China, with about 555,000 new cases and 286,000 deaths. Moreover, they differ between urban and rural areas, which are higher in the former (2), possibly due to several independent risk factors (higher intake of food of animal origin, dietary habits, and sedentary lifestyle). Therefore, raising public awareness of early screening for COAD and enhancing prevention are important for reducing the incidence of this disease (3). The TGF-β signaling pathway is important for keeping normal physiological functions of the intestine and facilitating tumor development, and its dysfunction is closely associated with the unfavorable prognosis of COAD (4). KRAS and NRAS gene mutations are most frequently observed in COAD, accounting for about 35–45% and 3–5%, respectively (5). The latest study revealed that LY3009120, an RAF dimer inhibitor, can effectively restrain the BRAF/CRAF heterodimer activity by occupying two promoters in the dimer, and it exhibits anti-proliferation activity in KRAS-mutated COAD cells and anti-growth activity in KRAS-mutated COAD xenograft models (6). Fecal testing, colonoscopy, and biomarker testing are the main techniques for COAD screening nowadays. However, fecal testing is less sensitive and prone to the influence of drugs and diet. As an invasive test, colonoscopy has certain risks in anesthesia and operation, and it involves high costs and requires high intestinal cleanliness, so subjects’ compliance is generally low. Therefore, an in-depth study of biomarkers for COAD will help establish new methods for COAD diagnosis and treatment.

Telomeres are protein-DNA complexes that are found at the chromosomal end and keep them safe from illicit ligation and resection, which are essential for chromosomal stability. Human telomeres consist of complexes between telomeric DNA and a six-protein complex called shelterin (7). Telomeres become progressively shorter during cell division and can eventually result in termination of cell division and death if they become too short. Telomerase reverse transcriptase (TERT) is the core component of telomerase. A mutation in the TERT promoter region is often observed in glioma, bladder cancer, and thyroid cancer, increasing the telomerase activity, which, in turn, helps maintain the cancer cells’ ability of infinite proliferation (8). Regulating shelterins may affect the length of telomeres and their tumorigenic ability. It is less likely for telomeric repeat-binding factor 1 (TRF1) to form tumors when not modified by AKT phosphorylation, suggesting that telomeres are possibly the most important intracellular target for tumorigenesis by the AKT pathway (9). A variety of telomere-related genes (TRGs) exert different effects on cancer development. For example, ATP1A1 gene mutations can cause aldosterone-producing adenoma (10). RFC2 can regulate the DNA replication and cell cycle, thus facilitating hepatocellular carcinoma (HCC) progression (11). By mediating APOA1 and ERK ubiquitination, TRIM15 enhances the development of non-small cell lung cancer (NSCLC), pancreatic cancer, and esophageal squamous cell carcinoma (12). However, the molecular mechanism of TRGs in COAD development is yet to be elucidated.

In this study, RNA sequencing (RNA-Seq) data and TRGs from The Cancer Genome Atlas (TCGA)-COAD and TelNet, respectively, were used. Then, differential expression, least absolute shrinkage and selection operator (LASSO) regression, univariate/multivariate Cox regression, and stepwise regression analyses were conducted to identify six differentially expressed TRGs, and their mechanism of action in COAD was further analyzed by enrichment, immune infiltration, and drug sensitivity analyses. A risk prediction model for COAD prognosis was constructed based on TRGs, and its potential effect on drug selection was evaluated, providing a reference for clinical therapy. We present this article in accordance with the TRIPOD+AI reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1075/rc).

Methods

Data collection

Using the “TCGAbiolinks” package, RNA-Seq data (41 normal samples and 481 tumor samples) were obtained from TCGA. After those samples were excluded due to missing clinical data and incomplete survival information, the remaining 481 tumor samples were left for later use, with the corresponding patient survival information from UCSC Xena (https://xena.ucsc.edu/). Moreover, the GSE17538 dataset containing 176 COAD samples downloaded from the Gene Expression Omnibus (GEO) was utilized for external validation; 2,093 TRGs were also obtained, and the intersection was taken using the “VennDiagram” package. Differentially expressed TRGs were identified [false discovery rate (FDR) <0.05 and |log₂ fold change (FC)| >1] using the “edgeR” package.

Risk prediction model construction

We randomly split 448 TCGA-COAD samples into a training (n=224) and a validation cohort (n=224). Univariate Cox regression analyses were first conducted on the samples with complete clinical information in the training cohort. Then the optimal penalty parameter was selected by LASSO regression analyses and 10-fold cross-validation. Based on the minimum value of Lambda of 0.0499, eight candidate prognostic genes underwent multivariate Cox and stepwise regression analyses. Finally, six prognostic genes were obtained, based on which a risk prediction model was established.

Riskcsore=∑n=1ncoefi∗xi[1]

where xi and coefi represent the RNA expression and its corresponding coefficient. According to the median risk score, low- and high-risk groups were set up in the training cohort. Kaplan-Meier (K-M) curves were drawn using the “survminer” and “survival” to compare the overall survival (OS). Moreover, the model’s predictive accuracy was evaluated by the area under the curve (AUC) for 1-, 2-, and 3-year OS with the “timeROC” package. The “ggpubr” and “ggplot” packages were adopted to create the box plot for prognostic gene expression, and the “ComplexHeatmap” package was used for plotting the complex heatmap. The diagrams for risk score in the training cohort and COAD survival time in the two groups were also generated.

Model validation

The GSE17538 dataset (176 COAD patients with complete survival information) was used for external validation. Following univariate Cox regression analyses, a risk prediction model was established based on the gene expressions and coefficients. According to the median risk score, low- and high-risk groups were set up. The “survminer” and “survival” packages were adopted to generate K-M survival curves for comparing the OS between the two groups. Moreover, the model’s predictive accuracy was evaluated by the AUC for OS with the “timeROC” package. The “ggpubr” and “ggplot” packages were adopted to create the box plot for prognostic gene expression, and the “ComplexHeatmap” package was used for plotting the complex heatmap. The diagrams for risk score and COAD survival time were also generated in both groups.

Construction of nomogram

Using the “rms” package, a nomogram was created for the OS prediction of COAD, which integrated prognostic factors [risk score, age, sex, and clinical tumor-node-metastasis (TNM) stage]. To assess its predictive accuracy and discrimination, calibration curves were plotted. Meanwhile, the risk scores and clinicopathologic signatures underwent univariate/multivariate Cox regression analyses to identify prognostic factors. Based on the clinical TNM stage, risk scores were calculated, and K-M curves were plotted to compare OS. Finally, the risk scores were compared under different survival status (alive, dead), sex (male, female), age (≤60, >60 years), T-stage (T1–T4), N-stage (N0–N2), and M-stage (M0/M1) (Figure S1A-S1F). Additionally, to further explore potential biological signaling pathways involved, we used the “C2: KEGG subset of CP” from the Molecular Signatures Database (MSigDB) as a gene reference set and performed gene set enrichment analysis (GSEA) by the R “ClusterProfiler”, “org.Hs.eg.db”, and “enrichplot” to generate an enrichment curve for each TRG, with P<0.05 as statistical significance.

Functional enrichment analyses

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional annotations were obtained from DAVID (https://david.ncifcrf.gov/). Using the “ggplot2” package, GO analysis was conducted on the 10 selected genes to identify the most enriched biological processes (BPs), molecular functions (MFs), and cellular components (CCs). KEGG analysis was also performed to explore their potential functions and signaling pathways, with adjusted P<0.05 as statistical significance. In addition, “C2: CP: REACTOME” from the MSigDB (https://www.example.com gsea/msigdb) was used as a reference set, and the heatmap was generated using the “msigdbr” and “pheatmap” packages to explore its enriched signaling pathways. The immune checkpoint gene expression was analyzed using the “ggpubr” and “ggplot” packages. Tumor-infiltrating immune cells underwent single-sample Gene Set Enrichment Analysis (ssGSEA) using the “GSVA” package. Additionally, to further explore potential signaling pathways and BPs involved, we used the “C2: KEGG subset of CP” and “C5: BP subset of GO” from MSigDB as gene reference sets and performed GSEA by the R “ClusterProfiler”, “org.Hs.eg.db”, and “enrichplot” to generate enrichment curves for six TRGs, with P<0.05 as statistical significance.

Immune infiltration analysis

Cell-type Identification By Estimating Relative Subsets Of RNA Transcripts (CIBERSORT) was utilized. CIBERSORT is an analytical tool for detecting the composition of complex tissues by their gene expression using gene expression data. We distinguished 22 human immune cell phenotypes by the CIBERSORTx and LM22. Downloaded from CIBERSORT (http://CIBERSORT.stanford.edu/), LM22 (leukocyte signature matrix) contains marker genes for 22 immune cells. CIBERSORTx, a new machine learning method, estimates the cell cluster abundance in bulk RNA-seq data. The association of risk scores with immune cells was analyzed using the “ggplot2” package and Pearson correlation analysis. The R “ggplot2” was utilized to generate the lollipop chart of the gene-immune cell correlation and the heatmap of immune cell-immune cell correlation, with P<0.05 as statistically significant.

Tumor Immune Dysfunction and Exclusion (TIDE) scores were calculated for COAD patients on TIDE. High TIDE scores suggest a high likelihood of tumor immune evasion, and poorer treatment response and prognosis. Differences in the TIDE/Dysfunction/Exclusion scores were analyzed in both groups using the “ggplot2” package plus Wilcoxon tests. Heatmaps of immune cell infiltration were generated using the “IOBR”.

Drug sensitivity

The half-maximal inhibitory concentration (IC₅₀) value of conventional chemotherapy or targeted drugs was predicted using the “oncoPredict” package. The dataset used for the proposed regression model was uploaded and updated by the “oncoPredict” team, including resources from the Genomics of Drug Sensitivity in Cancer 2 (GDSC2) (containing 198 common conventional chemotherapy or targeted drugs) and the Cancer Therapeutics Response Portal (CTRP) V2. Finally, this dataset was used to perform Pearson correlation analyses to explore the association of TRGs with drug sensitivity, and the most relevant gene-drug pairs were selected based on the P value. The R “ggplot2” was utilized to generate the scatter plot of gene-drug correlation.

Validation of prognostic genes

The Human Protein Atlas (HPA) is designed to display the distribution of 24,000 human proteins across tissues and cells and to show the results of immunohistochemical (IHC) staining of over 20 types of cancers. In this study, the IHC staining images of COAD vs. normal tissues were observed to compare RNA expression differences. The Tumor Immune Single-cell Hub (TISCH) is a single-cell database for immune infiltration analysis, which includes 79 datasets of 28 tumor types. Data stored in Gene Expression Profiling Interactive Analysis 2 (GEPIA2) are from TCGA and GTEx, and eight major analyses can be conducted with RNA-seq data to analyze the target gene expression and to clarify the differential gene expression in specific tumors. Tissue sections were collected from a cohort of 30 COAD patients at the Chinese PLA General Hospital, containing a representative distribution across different cancer stages: stage I (n=4), stage II (n=5), stage III (n=15), and stage IV (n=6). IHC staining was performed on normal and COAD tissues not treated with antitumor therapy. The antibodies used included EPHA6 (1:50, Proteintech, Wuhan, China, Cat#20211-1-AP), IP6K3 (1:100, Cusabio Biotech, Wuhan, China, Cat#CSB-PA003042), TRIM7 (1:200, Proteintech, Cat#26285-1-AP), USP2 (1:100, FineTest Biotech, Wuhan, China, Cat#FNab09316), CALML6 (1:50, FineTest Biotech, Cat#FNab01207), and COCH (1:100, FineTest Biotech, Cat#FNab01824). This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Chinese PLA General Hospital (No. S2024-523-01). The requirement for informed consent was waived by the Ethics Committee due to the retrospective nature of the study. Two experienced pathologists independently interpreted the results. The scores of the area of positive cells: 0, ≤5%; 1, 6–25%; 2, 26–50%; 3, 51–75%; 4, 76–100%. The scores of the staining intensity: 0, −; 1, +; 2, ++; 3, +++. Then the IHC score was calculated by multiplying the staining intensity (0–3) by the area of positive cells (0–4). Moreover, the differences in the expression of TRGs and IHC scores between COAD and normal tissues were assessed by the Wilcoxon signed-rank test. The total scores of 0–4 indicated low expression, while the total scores of 6–12 indicated high expression. Western blotting was also conducted on EPHA6 using the normal human colonic epithelial cell line NCM460 and COAD cell lines HCT116 and HT29 (ATCC, Manassas, VA, USA). Specifically, all cell lines were cultured at 37 ℃ in a humidified environment with 5% CO₂. They were harvested and lysed with RIPA lysis buffer containing protease inhibitors to extract total proteins. The protein concentration was measured using the bicinchoninic acid (BCA) method. An equal amount of protein (30 µg per well) was separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto a PVDF membrane. The membrane was blocked with 5% skim milk at room temperature for 2 h, and incubated overnight at 4 ℃ with primary antibodies against EPHA6 (1:50, ZenBio, Research Triangle Park, NC, USA, Cat#163533) and β-actin (1:1,000, Abcam, Cambridge, UK, Cat#ab8226). After washing with TBST, the membrane was incubated with HRP-labeled secondary antibody (1:3,000–1:5,000) at room temperature for 1–2 h. Finally, the blots were developed using an enhanced chemiluminescence kit, and images were captured using a chemiluminescence imaging system. All tests were independently repeated at least three times. Data were described by mean ± standard deviation. One-way analysis of variance (ANOVA) was performed for comparisons among groups.

Association of risk score with tumor mutation burden (TMB)

Downloaded from TCGA, the data on mutations in COAD patients were analyzed by the “maftools” package. The association of risk scores with TMB was estimated by Pearson correlation analysis using the “ggplot2” package.

Statistical analysis

R4.3.3 was utilized for statistical analyses. The “survival” and “survminer” were adopted for Cox proportional hazards regression and K-M survival analyses, and the log-rank test was also applied to the latter. The predictive accuracy was assessed using the AUC. The IC₅₀ of conventional chemotherapy or targeted drugs was compared by the Wilcoxon test. The risk scores were compared under different clinicopathologic signatures by the Wilcoxon test and one-way ANOVA using the “ggpubr” and “ggplot” packages. The correlation of immune cells with risk scores was analyzed by Pearson correlation analysis, and the “ggplot2” and “ggpubr” packages were adopted for visualization. The EPHA6 expression was compared in COAD vs. normal tissues by the Wilcoxon signed-rank test. The OS in high and low risk groups was assessed by K-M curves. P<0.05 suggested statistical significance if not stated otherwise.

Results

Differentially expressed TRGs

RNA-Seq data (41 normal samples and 481 COAD samples) were obtained from TCGA first, from which 22,002 genes were identified including 2,093 TRGs. Then 5,009 differentially expressed genes (DEGs) were identified (FDR <0.05 and |log₂FC| >1) between normal and tumor tissues, including 5,470 up-regulated ones and 2,297 down-regulated ones, as shown in Figure 1A. Finally, 457 differentially expressed TRGs were identified in the Venn diagram (Figure 1B).

Figure 1 Characterization of differentially expressed TRGs. (A) Volcano plot of DEGs in COAD. (B) Venn diagram of DEGs and TRGs. COAD, colon adenocarcinoma; DEGs, differentially expressed genes; FC, fold change; FDR, false discovery rate; TCGA, The Cancer Genome Atlas; TRG, telomere-related gene.

Construction and validation of a six-TRG risk prediction model

To test the prognostic value of these DEGs, we randomly split TCGA-COAD samples into training and validation cohorts. Following LASSO regression and univariate Cox regression analyses, candidate prognostic genes were selected, and the optimal penalty parameter was selected by 10-fold cross-validation. We finally identified eight DEGs (EPHA6, IP6K3, CALML6, UCHL1, COCH, PPM1N, USP2, and TRIM7) related to COAD prognosis (Figure 2A,2B), from which six TRGs (USP2, TRIM7, EPHA6, IP6K3, CALML6, and COCH) were selected as the independent risk factors (Figure 2C) by multivariate Cox and stepwise regression analyses. The risk score was calculated: 0.2514738 * EPHA6 − 0.2133231 * USP2 + 0.1981439 * COCH + 0.1244138 * TRIM7 − 0.3431239 * IP6K3 + 0.3100913 * CALML6. The high- and low-risk groups were set up in the training cohort based on the median risk score (0.9276243), each with 112 cases. In the training cohort, K-M curves showed shorter OS in the high-risk group (P<0.001; Figure 2D). The AUCs were 0.746, 0.750, and 0.726, respectively, for 1-, 2-, and 3-year OS (Figure 2E), suggesting the model’s good predictive performance. The high-risk group was prone to the outcome of death (Figure 2F). As shown in the box plot (Figure 2G,2H), the high-risk group had significant increases in CALML6, COCH, EPHA6, and TRIM7, and decreases in USP2 and IP6K3. The expression of each gene in the heat map was totally consistent with that in the box plot.

Figure 2 Training set for the construction of the risk model based on TRGs for COAD prognosis. (A) Univariate Cox regression analysis of TRGs. (B,C) LASSO regression analysis. (D) K-M curves of survival status and survival time between high and low risk groups in the training set. (E) ROC curves at 1, 2, and 3 years in the training set. (F) Distribution of risk scores and overall survival status of COAD patients in the training set. (G,H) Box plot and heat map of TRGs in the training set. **, P<0.01; ***, P<0.001; ****, P<0.0001. AUC, area under the curve; CI, confidence interval; COAD, colon adenocarcinoma; K-M, Kaplan-Meier; LASSO, least absolute shrinkage and selection operator; ROC, receiver operating characteristic; TRG, telomere-related gene.

Validation of risk model

To validate the risk prediction model, the high- and low-risk groups were set up in the validation cohort based on the median risk score (0.98579355), each with 112 cases. In the validation cohort, K-M curves showed shorter OS in the high-risk group (P=0.02; Figure S2A). The AUCs were 0.659, 0.650, and 0.656, respectively, for 1-, 2-, and 3-year OS (Figure S2B), suggesting the model’s good predictive performance. The high-risk group was prone to the outcome of death (Figure S2C). As shown in the box plot (Figure S2D,S2E), the high-risk group had significant increases in CALML6, COCH, EPHA6, and TRIM7, and decreases in USP2 and IP6K3. The expression of each gene in the heat map was totally consistent with that in the box plot.

Six-TRG prognostic signature as an independent predictor for COAD prognosis

In addition, univariate/multivariate Cox regression analyses were performed on clinical signatures. First, univariate Cox regression analyses showed that age (P=0.04), T-stage (P<0.001), N-stage (P<0.001), M-stage (P<0.001), and risk scores (P<0.001) were associated with the OS of COAD (Figure 3A). Multivariate Cox regression analyses revealed that the risk score of the six-TRG risk prediction model (P<0.001), age (P=0.02), T-stage (P=0.002), N-stage (P=0.01), and M-stage (P<0.001) could serve as independent factors for predicting the OS rate in patients with COAD (Figure 3B). In addition, a nomogram of age, sex, clinical TNM stage, and risk scores was created to separately predict the OS of COAD (Figure 3C). Based on the total score, the likelihood of survival (1–5-year OS rates) was calculated. Then to evaluate the model’s prognostic value, calibration curves were generated, and it was observed that the predicted 1–5-year OS was similar to the actual value (Figure 3D). In addition, Cox regression analyses were carried out on T-stage, N-stage, and M-stage, based on which high- and low-risk groups were set up. K-M survival analyses showed that the high-risk group had significantly shorter survival time (Figure 3E). Finally, the risk scores were compared under different survival status (alive, dead), sex (male, female), age (≤60, >60 years), T-stage (T1–T4), N-stage (N0–N2), and M-stage (M0/M1) (Figure S1A-S1F). We assessed the signaling pathways associated with TRGs using GSEA, and the top five pathways are shown in (Figure 3F). The results showed that multiple TRGs (USP2, TRIM7, IP6K3, EPHA6, COCH, and CALML6) were significantly enriched in two signaling pathways highly associated with the core function of telomere (KEGG_CELL_CYCLE and KEGG_PATHWAYS_IN_CANCER). Additionally, multiple TRGs were also significantly enriched in KEGG_MAPK_SIGNALING_PATHWAY, which plays a crucial role in cellular stress responses and growth regulation. Specifically, USP2, IP6K3, EPHA6, COCH, and CALML6 were significantly associated with KEGG_CELL_CYCLE, and USP2, TRIM7, IP6K3, EPHA6, COCH, and CALML6 were significantly associated with both KEGG_PATHWAYS_IN_CANCER and KEGG_MAPK_SIGNALING_PATHWAY. Meanwhile, a significant association was also present between TRIM7 and KEGG_LYSOSOME.

Figure 3 TRGs are independent prognostic factors for COAD. (A) Univariate Cox analysis. (B) Multivariate Cox analysis. (C) Nomogram based on TRGs and clinicopathological factors. Gender: 0= male, 1= female. (D) Generate calibration curves to predict 1–5 years of OS. (E) K-M survival analysis of the T/N/M stage on the OS of COAD patients in different risk groups. (F) KEGG enrichment curves of each telomere-related prognostic gene. CI, confidence interval; COAD, colon adenocarcinoma; KEGG, Kyoto Encyclopedia of Genes and Genomes; K-M, Kaplan-Meier; OS, overall survival; T/N/M, tumor/node/metastasis; TRG, telomere-related gene.

External validation for six-TRG prognostic signature

Downloaded from the GEO, the GSE17538 dataset (176 COAD patients with complete survival information) was utilized for external validation. K-M curves revealed far shorter OS in the high-risk group in the training cohort (P<0.001; Figure S3A). The AUCs were 0.628, 0.639, and 0.631, respectively, for 1-, 2-, and 3-year OS (Figure S3B), suggesting the model’s good predictive performance. The high-risk group was prone to the outcome of death (Figure S3C). As shown in the box plot (Figure S3D,S3E), the levels of CALML6, COCH, EPHA6, TRIM7, USP2, and IP6K3 were significantly lower in the high-risk group. The expression of genes in the heat map was completely consistent with that in the box map.

Functional enrichment analyses of differentially expressed TRGs

Furthermore, DEGs were detected for the differences in BP and pathways by GO and KEGG analyses (P<0.05). GO analyses revealed that differentially expressed TRGs were highly enriched in BP (DNA repair, cell cycle, and DNA replication), MF (protein binding, ATP binding, and DNA binding), and CC (nucleus, nucleoplasm, and cytosol) (Figure 4A). KEGG analyses revealed that they were highly enriched in cell-associated pathways (cell cycle, cellular senescence, and DNA replication) (Figure 4B). In addition, “C2: CP: REACTOME” from the MSigDB was used as a reference set. Ten gene sets (861 genes) were found to be greatly different between the two groups, including three up-regulated ones in the high-risk group (fceri_mediated_mapk_activation, cell_cycle, and negative_regulation_of_mapk_pathway) and seven up-regulated ones in the low-risk group (rhedgehog_ligand_biogenesis, activated_tak1_mediates_p38_mapk_activation, regulation_of_tp53_expression_and_degradation, pi3k_akt_signaling_in_cancer, beta_catenin_independent_wnt_signaling, mapk_family_signaling_cascades, and initiation_of_nuclear_envelope_ne_reformation) (Figure 4C). In addition, we assessed the signaling pathways and BPs associated with TRGs using GSEA, and the top five pathways are shown in (Figure 4D). The six TRGs were significantly associated with BPs including GOBP_DNA_REPLICATION_INITIATION, GOBP_FATTY_ACID_METABOLIC_PROCESS, GOBP_MONOCARBOXYLIC_ACID_METABOLIC_PROCESS, GOBP_NEGATIVE_REGULATION_OF_KINASE_ACTIVITY, and GOBP_REGULATION_OF_MITOTIC_CELL_CYCLE, and with signaling pathways, including KEGG_CELL_CYCLE, KEGG_DNA_REPLICATION, KEGG_PATHWAYS_IN_CANCER, KEGG_PURINE_METABOLISM, and KEGG_PYRIMIDINE_METABOLISM.

Figure 4 Functional enrichment analysis of differentially expressed TRGs in different risk groups. (A) GO enrichment analysis results. (B) KEGG enrichment analysis results of differentially expressed TRGs. (C) GSVA analysis for the REACTOME term analysis results of COAD patients in different risk groups. (D) KEGG and GO: BP enrichment curves of the six-TRG prognostic signature (GSEA). The plot displays the top 5 significantly enriched pathways. BP, biological process; COAD, colon adenocarcinoma; GO, Gene Ontology; GSVA, gene set variation analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; TRG, telomere-related gene.

Immune infiltration analysis and immunotherapy response analysis

The significant differences in immune cell levels were revealed by the CIBERSORT algorithm. A box plot for 22 immune cells in both groups was generated (Figure 5A), and the association of significantly different immune cells with risk scores was also plotted. The proportions of eosinophils, Dendritic.cells.activated, and T.cells.follicular.helper were positively associated with the risk score (P=0.02, 0.248, and 0.01), while the proportion of T. cells.CD8 had a negative association with the risk score (P=0.003) (Figure 5B). Additionally, the expression of TRGs also had an association with immune cells (Figure 5C). For example, the USP2 expression was negatively correlated with T.cells.follicular.helper and Mast.cells.resting, and positively correlated with Macrophages.M1. The TRIM7 expression was negatively correlated with T.cells.gamma.delta, T.cells.CD8, T.cells.CD4.naive, T.cells.CD4.memory.resting, and Macrophages.M1, and positively correlated with B.cells.memory, Dendritic.cells.resting, Macrophages.M0, Mast.cells.resting, neutrophils, NK.cells.resting, and T.cells.CD4.memory.activated. The expression of IP6K3 had a negative association with NK.cells.resting and positively correlated with Macrophages.M1. The expression of EPHA6 was negatively correlated with T.cells.CD4.memory.activated and Mast.cells.resting, and positively correlated with eosinophils and B.cells.naive. The expression of COCH was negatively correlated with T.cells.regulatory, Mast.cells.resting, and NK.cells.activated, and positively correlated with T.cells.gamma.delta, T.cells.CD8, and Macrophages.M1. The expression of CALML6 had a negative correlation with Dendritic.cells.resting and positively correlated with T.cells.CD4.memory.resting, T.cells.follicular.helper, and T.cells.CD4.naive. Moreover, strong correlations were also found among the 22 immune cells (Figure 5D). For example, T.cells.CD4.naive and T.cells.CD8 displayed strong positive correlations with each other (R=0.81, P<0.01). To sum up, TRGs are important in the function of infiltrating immune cells and may be potential targets for immunotherapy.

Figure 5 Immune infiltration analysis of COAD patients in risk groups. (A) Differences in immune infiltration of 22 immune cells in different risk groups. (B) Correlation between the abundance of four immune cells and risk scores. (C) Lollipop plot of the correlation between six telomere-related genes and immune cells. (D) Heatmap of immune cell inter-correlation based on CIBERSORT results. (E) Differences in expression levels of immune checkpoint genes in different risk groups. (F) Single-sample GSEA of immune infiltrating cells. ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001. COAD, colon adenocarcinoma; GSEA gene set enrichment analysis.

TIDE scores were calculated for COAD patients on TIDE. High TIDE scores suggest a high likelihood of tumor immune evasion, and poorer treatment response and prognosis. Differences in the TIDE/Dysfunction/Exclusion scores were analyzed (Figure S4A-S4C). The expressions of immune checkpoint genes were also compared (Figure 5E). Seven differential immune checkpoint genes were found to be highly expressed in the high-risk group, including many validated targets for immunotherapy (LAG3, CD276, TGFB1, HLA-DMA, and CD47). To sum up, the association of risk scores calculated with immune infiltration was assessed, which was verified by the results. ssGSEA revealed higher immune cell infiltration in the high-risk group. According to the correlation analysis, a few immune cells, especially Activated.CD8 and Memory.B.cells, had negative correlations with the risk score, suggesting a better prognosis and longer survival in the low-risk group (Figure 5F).

Drug sensitivity analysis

To further explore the differences in drug resistance, the estimated IC₅₀ of 198 conventional chemotherapy or targeted drugs was compared. As shown in Figure 6A-6G, seven representative drugs, dasatinib, docetaxel, erlotinib, and gefitinib were possible candidate drugs for the high-risk group, whereas irinotecan, cediranib, and IGF-1R/IR inhibitors might not be suitable for the high-risk group. Six TRGs (USP2, TRIM7, EPHA6, IP6K3, CALML6, and COCH) were associated with sensitivity to specific chemotherapy drugs (P<0.05). For example, an elevated USP2 expression was associated with increased sensitivity to dasatinib in CRC patients (R=0.81, P<0.05). Similarly, an elevated IP6K3 expression was associated with increased sensitivity to dasatinib (R=0.68, P<0.05). In contrast, elevated USP2 was associated with decreased sensitivity to docetaxel in CRC patients (R=−0.67, P<0.05), elevated COCH was associated with decreased sensitivity to docetaxel (R=−0.8, P<0.05) and erlotinib (R=−0.63, P<0.05), and elevated TRIM7 was associated with decreased sensitivity to gefitinib (R=−0.82, P<0.05) (Figure 6H). To sum up, six TRGs exhibited widespread correlations with a variety of drugs (Figure S5), suggesting complex interactions between TRGs and sensitivity to chemotherapy drugs for cancer.

Figure 6 Drug sensitivity analysis for COAD patients in different risk groups. (A) BMS-754807. (B) Cediranib. (C) Dasatinib. (D) Docetaxel. (E) Erlotinib. (F) Gefitinib. (G) Irinotecan. (H) Scatter plots visualizing correlations between multiple genes and drugs. ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001. COAD, colon adenocarcinoma.

Association of risk score with TMB

TMB is known as an important indicator for evaluating immunotherapy efficacy. In this study, the association of risk score with TMB was evaluated. The top 10 most frequently mutated genes (APC, TP53, TTN, KRAS, PIK3CA, FAT4, MUC16, SYNE1, OBSCN, and ZFHX4 in the low-risk group; APC, TP53, TTN, KRAS, PIK3CA, FAT4, MUC16, SYNE1, OBSCN, and ZFHX4 in the high-risk group) are shown in Figure S6A,S6B. The risk score had a positive correlation with TMB in COAD patients (Figure S6C, P=0.01, Cor =0.13).

Validation of prognostic genes

To validate the effect of TRGs on the COAD progression, the corresponding protein expression levels were compared by IHC staining using HPA. The results showed that both EPHA6 and USP2 were lowly/moderately or weakly expressed in tumor tissues, while TRIM7 was highly expressed in tumor tissues (Figure 7A). IHC staining results for IP6K3, CALML6, and COCH in COAD are not included in HPA. These findings demonstrated the importance of TRGs in COAD. The IHC staining revealed that the EPHA6 expression was weak in COAD tissues but strong in normal tissues (Figure 7B). The IHC score showed that the EPHA6 expression had a statistically significant difference between COAD and adjacent normal tissues (Figure 7B). However, no statistically significant differences were observed in the expression of the remaining five genes (USP2, TRIM7, IP6K3, CALML6, and COCH) between COAD and adjacent normal tissues (Figure S7A-S7E). Western blotting revealed that HCT116 and HT29 cells had decreased EPHA6 expression compared with NCM460 cells, with statistically significant differences (Figure 7C). These findings all aligned with bioinformatics predictions, confirming the association of EPHA6 with COAD. In addition, the expression of the six TRGs in COAD was detected using the single-cell RNA-seq dataset CRC-GSE166555, which includes 32 cell clusters and 13 major cell types, describing the cluster distribution and number (Figure 7D), and the proportions and expression levels of USP2, TRIM7, IP6K3, CALML6, and COCH in each cell type (Figure 7E,7F). These data suggest that the five DEGs may benefit the tumor microenvironment of COAD. Single-cell RNA-seq data of EPHA6 in COAD are not included in TISCH.

Figure 7 Validation of TRG prognostic genes. (A) Immunohistochemical images of EPHA6, USP2, and TRIM7 in COAD and normal tissues from the Human Protein Atlas (HPA). Image credit: Human Protein Atlas. Images available from https://www.proteinatlas.org/.(accessed on 12 June 2025). (B) Representative IHC images (left) and scores (right) of EPHA6 expression in COAD and adjacent normal tissues (magnification 20× and 100×). (C) Representative blots (upper) and densitometric quantification (lower) of EPHA6 and β-actin in NCM460, HCT116, and HT29 cells. Band gray values were normalized for quantitative analysis. Data are expressed as mean ± SD. Statistical comparisons were performed relative to the NCM460 group, with differences indicated by asterisks (***, P<0.001). (D) Classification of all cell types and the number of each cell type in CRC-GSE166555. (E,F) Proportion and expression of USP2, TRIM7, IP6K3, CALML6, and COCH. COAD, colon adenocarcinoma; HPA Human Protein Atlas; IHC, immunohistochemical; K-M, Kaplan-Meier; SD, standard deviation; TRG, telomere-related gene.

Discussion

We searched for 457 differentially expressed TRGs in TCGA, and established a six-TRG risk prediction model (USP2, TRIM7, EPHA6, IP6K3, CALML6, and COCH), which could become an independent predictor for COAD prognosis and exhibited good clinical utility. We explored the differences in BP and pathways of DEGs, and found that DEGs were highly enriched in cell cycle and DNA replication. ssGSEA revealed stronger immune cell infiltration in the high-risk group. In addition, Activated.CD8 and Memory.B.cells had negative correlations with the risk score, indicating a better prognosis and longer OS in the low-risk group. Dasatinib, docetaxel, erlotinib, and gefitinib were possible candidate drugs for the high-risk group, whereas irinotecan, cediranib, and IGF-1R/IR inhibitors might not be suitable. In summary, immune checkpoint inhibitors and targeted drugs are better options for high-risk COAD patients. Finally, reduced EPHA6 expression in COAD was confirmed by IHC staining, consistent with bioinformatics analysis.

Indeed, the associations of the above-mentioned six genes with malignancy development have been verified. USP2 is a multifunctional deubiquitinating enzyme encoded by the gene located on chromosome 11q23.3 (13), and its overexpression promotes cancer stem cell (CSC) expansion, cancer cell migration, and progression in triple-negative breast carcinoma (TNBC) (14). USP2 synergizes with HSP90 inhibitors by counteracting endocytic degradation, offering a novel treatment strategy for ErbB2-positive breast cancer (15). Moreover, USP2 stabilizes programmed death-ligand 1 (PD-L1) by specific interactions in prostate cancer cells, which may enhance anti-tumor immune responses by modulating PD-L1 and also make tumor cells prone to T cell-mediated killing (16). Deletion of USP2 inhibits the Twist (a key regulator of stem cell renewal) and Bmi1 expressions, thus affecting the tumor stem cells’ self-renewal capacity (14). TRIM7, as an E3 ligase, is a critical regulator in physiopathological processes. The expression of TRIM7 becomes imbalanced in a variety of malignancies, and it can promote or inhibit tumor cell growth (17). TRIM7 binds to SLC7A11 to promote its polyubiquitination and suppresses gastric cancer cell growth by suppressing the SLC7A11/glutathione peroxidase 4 (GPX4) axis, thereby triggering ferroptosis in gastric cancer cells (18). By direct interactions, TRIM7 induces Src polyubiquitination and decreases Src abundance in HCC cells, thereby inhibiting HCC progression (19). Moreover, TRIM7 mediates aberrant cell signaling in PI3K/AKT, TGF-β, JAK/STAT, and Wnt/β-catenin, and thus affects the cancer cell invasion, migration, and proliferation. The TRIM family proteins are also key in drug resistance and affect tumor responsiveness to therapy by regulating the epithelial-mesenchymal transition (EMT) and CSCs (20). EPHA6 is an Eph receptor family member, and Eph signaling is important in regulating tissue development and homeostasis including cell adhesion, differentiation, cytoplasmic division, survival, and apoptosis (21). By inducing prostate cancer angiogenesis and tumor cell metastasis, EPHA6 is involved in prostate cancer progression. EPHA6 has a persistent overexpression in metastatic prostate cancer cells, which has positive correlations with vascular/nerve invasion, TNM stage, and PSA level, and EPHA6 knockdown can reduce invasion in vitro and lymph node/pulmonary metastasis (22). Research suggests that glioblastoma (GBM) has a lower EPHA6 expression than normal tissues. In addition, the EPHA6 receptor can enhance the cell sensitivity to bone morphogenetic protein 2 (BMP-2)-induced apoptotic response, suggesting that EPHA6 may suppress GBM (21). EPHA6 is also significantly associated with HCC growth, invasion, and metastasis (23). In addition, IP6K3 belongs to the inositol phosphokinase (IPK) family and may be an important player in cellular metabolism, energy sensing, and lifespan regulation (24). A genome-wide association study (GWAS)-based study revealed that the IP6K3 gene is significantly related to renal cell carcinoma, and one of the genes significantly related to lung carcinoma, gastric carcinoma, and esophageal squamous cell carcinoma (25). CALML6, an EF-hand (helix-loop-helix structural motif) Ca²⁺-binding protein, can inhibit infection-induced innate immune response and pro-inflammatory response (26), and regulate EP4-induced migration of OSCC cells (CAMKK2 and EP4 correlate with progression and survival of breast cancer), indicating that CALML6 may work in breast cancer through the EP4/CAMKK2 signaling pathway. Moreover, CALML6 may influence lung cancer development and metastasis by affecting mitochondrial function and reactive oxygen species (ROS) production (27). Located at 14q12-13, COCH is highly conserved in multiple species, and it encodes a secreted protein cochlin. Low COCH expression corresponds to better OS and disease-free survival (DFS) (28). Bioinformatics analysis revealed an association of COCH with poor survival in TNBC patients (29). All the above studies demonstrate the potential and usability of the six TRGs as prognostic biomarkers for COAD. Among them, it has been reported that USP2 can stabilize PD-L1 by deubiquitination, contributing to tumor immune escape. This suggests that USP2 may indirectly influence clinical outcomes by affecting the tumor microenvironment (16). However, evidence regarding the mechanism of USP2 and the remaining five genes (TRIM7, EPHA6, IP6K3, CALML6, and COCH) in directly regulating telomeres in COAD is still lacking. Therefore, these six TRGs can be considered candidate genes derived from the telomere-associated gene set and with prognostic significance for COAD, and their exact roles in the telomere regulatory network remain to be validated by further experimental and network analyses.

Furthermore, the low-risk group had lower proportions of eosinophils (P=0.02) and T.cells.follicular.helper (P=0.01), and higher proportions of T.cells.CD8 (P=0.003) than the high-risk group. Research suggests that elevated absolute eosinophil counts may adversely affect the lung cancer prognosis (30). They may contribute to tumor metastasis and angiogenesis by secreting MMP9, VEGF, FGF, and PDGF, while IL-4 and IL-13 are generated to promote differentiation of tumor-associated macrophages into the M2 phenotype (31). In specific B-cell lymphomas such as DLBCL, the Tfh/B cell interaction may facilitate the tumor cell proliferation and survival, which is primarily associated with IL-10, i.e., higher IL-10 mRNA levels are correlated with increased proliferation and decreased apoptosis of DLBCL cells (32). In COAD, CD8⁺ T-cell infiltration is more frequent at the tumor margins than at the tumor center, showing a positive correlation. High CD8⁺ T-cell infiltration indicates better OS, with higher median OS rates, suggesting that high CD8⁺ T-cell infiltration at the tumor center may be an independent protector for OS (33). In this study, the high-risk group had a significant reduction of CD8⁺ T cells, so it was hypothesized that CD8⁺ T-cell function may be suppressed or attenuated in high-risk COAD. Previous studies showed that CRC exhibits different immune phenotypes (i.e., immune-active, immune-desert, stroma-rich) rather than a single immune state. The immune-desert phenotype is typically characterized by insufficient infiltration of adaptive immune cells (34). For example, research has clearly demonstrated that the infiltration level of effector CD8⁺ T cells is closely linked to prognosis. The internationally validated Immunoscore is based on the density of CD3⁺ and CD8⁺ T cells in the tumor core and invasion margin, which is used to assess the risk of recurrence and clinical outcomes in CRC patients (35). It is generally agreed that reduced CD8⁺ T cell infiltration typically indicates weakened antitumor immunity and is associated with unsatisfactory prognosis. Meanwhile, increasingly more studies have recently suggested that B-cell infiltration in CRC, particularly B-cell responses associated with tertiary lymphoid structure (TLS), is often associated with more active local adaptive immunity and better clinical outcomes (36). TLS can create a local microenvironment for B-cell maturation, antigen presentation, and T/B lymphocyte cooperation. In this context, activated CD8⁺ T cells and memory B cells were negatively correlated with risk scores in this study, generally consistent with the overall conclusion of previous CRC immunology research.

GO and KEGG analyses revealed that these DEGs are important in related pathways. Activation of ERK-MAPK reduced the expression of cell cycle-related genes, accelerated the G1/S phase transition, and promoted COAD proliferation. The Reactome enrichment analysis showed that the MAPK-related genes in the high-risk group were up-regulated. Downregulation of CD147 can weaken proliferation and enhance apoptosis of COAD cells, possibly related to the inhibition on the MAPK pathway (37). Activation of the high-affinity IgE receptor (FcεRI) may promote phosphorylation and activation of MAPK family members (ERK, JNK, or p38), thus affecting the COAD cell proliferation, apoptosis, and invasion.

To further explore the differences in drug resistance, the estimated IC₅₀ of 198 conventional chemotherapy or targeted drugs was compared. The high-risk group had higher predictive sensitivity to some drugs, suggesting their different treatment responses. (I) Dasatinib (ClinicalTrials.gov ID: NCT00504153; NCT, National Clinical Trial): this phase II trial intends to assess the efficacy of dasatinib for treating previously treated metastatic COAD. Dasatinib can suppress tumor cell growth by blocking enzymes required for tumor progression. Additionally, preclinical studies suggest that dasatinib possesses anti-proliferative and antitumor activity in CRC cell lines and xenograft models, which lays a basis for its further investigation in metastatic CRC (38). (II) Docetaxel (ClinicalTrials.gov ID: NCT01639131): this trial intends to assess the efficacy of gemcitabine plus docetaxel for treating metastatic COAD with checkpoint with forkhead and ring finger domains (CHFR) and/or microsatellite instability (MSI) phenotypes. The potential value of docetaxel for treating CRC may be primarily observed in populations selected by specific biomarkers, particularly those with CHFR methylation or MSI tumors (39). (III) As an EGFR inhibitor, erlotinib has been used for targeted therapy for COAD and is mentioned in the Chinese Expert Consensus on Targeted Therapy for Colorectal Cancer. Additionally, studies on maintenance therapy for metastatic CRC suggest that erlotinib plus bevacizumab may produce certain clinical benefits in specific scenarios (40). (IV) Gefitinib (ClinicalTrials.gov ID: NCT00025350): this randomized phase II trial compared the efficacy of two different doses of gefitinib for treating recurrent or metastatic COAD. Primarily due to the low EGFR mutation rate in COAD and the fact that EGFR inhibitors are less effective for treating COAD than NSCLC, the use of gefitinib for treating COAD is limited. Nevertheless, the phase II trial demonstrates that gefitinib combined intervention may enhance the antitumor efficacy of FOLFOX-4 in previously untreated metastatic CRC (41). (V) Irinotecan is commonly used in combination with 5-FU/leucovorin (FOLFIRI) as first- or second-line therapy for metastatic COAD, and its clinical benefits in metastatic CRC have been well documented in randomized trials. Irinotecan remains an important component of the standard therapeutic regimen (42). (VI) Cediranib (ClinicalTrials.gov ID: NCT00056446): this trial intends to assess the efficacy of 5-FU/oxaliplatin/folinic acid plus cediranib and 5-FU/oxaliplatin/folinic acid plus placebo for treating CRC with distant metastasis and progression following irinotecan treatment. The phase III trial suggests that cediranib exhibits antitumor activity in metastatic CRC, mainly manifested as limited progression-free survival benefits (43). (VII) IGF-1R/IR inhibitors (ClinicalTrials.gov ID: NCT01154335): this trial intends to identify the maximum tolerated dose of everolimus (an mTOR inhibitor) plus OSI-906 (a dual IGF-1R/IR tyrosine kinase inhibitor) for treating refractory metastatic COAD. Moreover, IGF-1R/IR-targeted inhibition has demonstrated potential for antitumor activity and reversal of drug resistance in preclinical CRC models, suggesting its further research value in CRC (44). Recently, research on CRC biomarkers has constantly advanced, which has gradually expanded from traditional pathological assessment to molecular subtyping, liquid biopsy, and non-invasive testing. In clinical practice, mismatch repair deficiency/MSI (MMR/MSI) status, and KRAS, NRAS, and BRAF mutations have become the most commonly used molecular markers for CRC stratification and treatment decision-making. For some metastatic cases, abnormalities such as human epidermal growth factor receptor 2 (HER2) amplification and neurotrophic tyrosine receptor kinase (NTRK) fusions also hold clear diagnostic significance (45). Meanwhile, carcinoembryonic antigen (CEA) remains one of the most widely used traditional markers in CRC follow-up monitoring. Circulating tumor DNA (ctDNA) and methylation markers in fecal or blood samples constitute important directions of recent research on CRC recurrence monitoring, risk assessment, and early screening (46). In this sense, the six-TRG prognostic signature created in this study tends to offer a new perspective on the biological heterogeneity of COAD besides the existing biomarker profiles. It is not a substitute for established molecular or clinical biomarkers but provides a supplementary clue for understanding prognostic differences only from telomere biology.

Some limitations in this study are worth noting. First, although the candidate genes were all derived from TelNet, no measurements of telomere length or activity corresponding to the same samples were made in the public cohorts (TCGA-COAD and GSE17538). As a result, the association of these genes with telomere length and activity could not be directly assessed. Nevertheless, the functional enrichment analysis revealed that these genes were involved in pathways closely related to telomere maintenance, such as cell cycle, and DNA replication. This suggests that the prognostic signature created reflects telomere biological characteristics to some extent. Moreover, the specific regulatory mechanisms of TRGs on COAD prognosis were not validated by in vivo and in vitro tests, which will be the focus of future studies. As pointed out in the guideline issued by the American College of Radiology, computed tomography colonography (CTC) is usually indicated for COAD screening in intermediate-risk groups, with follow-up every five years for those negative subjects. Additionally, it states that if a patient is in good health and has a life expectancy of more than ten years, screening should continue until the patient is 75 years old. However, the long-term screening follow-up recommended may carry a high economic burden (47), and it is impossible to validate our findings in a single trial. Second, the sample size was small, including only 522 TCGA-COAD samples as a test cohort and samples from the GEO for external validation. In addition, IHC staining results for IP6K3, CALML6, and COCH in COAD are not included in HPA, and single-cell RNA-seq data of EPHA6 in COAD are not included in TISCH. As the gene sequencing technique develops, inter-tumor and intra-tumor heterogeneity has received much attention (48). Different genetic phenotypes and pathways produce diverse genetic clones and changing tumor microenvironments (49), and clinical implications may vary across heterogeneity. In the future, we hope to predict such heterogeneity and personalize the COAD treatment.

Conclusions

In conclusion, a new six-TRG risk prediction model was constructed for COAD prognosis, and it may act as an independent predictor for COAD prognosis. The risk model’s correlation with the tumor immune microenvironment was preliminarily determined. In addition, a preliminary assessment was performed on chemotherapy drug sensitivity in different risk groups. Our findings provide some possibilities for predicting the COAD prognosis, and may help discover potential therapeutic targets and select the drug for COAD patients.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD+AI reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1075/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1075/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1075/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1075/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. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This study was approved by the Ethics Committee of Chinese PLA General Hospital (No. S2024-523-01) and the requirement for informed consent was waived.

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. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
2. Guo LW, Zhang XL, Cai L, et al. Current status of global colorectal cancer prevalence, prevention and control. Zhonghua Zhong Liu Za Zhi 2024;46:57-65. [Crossref] [PubMed]
3. Shi JF, Wang L, Ran JC, et al. Clinical characteristics, medical service utilization, and expenditure for colorectal cancer in China, 2005 to 2014: Overall design and results from a multicenter retrospective epidemiologic survey. Cancer 2021;127:1880-93. [Crossref] [PubMed]
4. Shan H, Tian G, Zhang Y, et al. Exploring the molecular mechanisms and therapeutic potential of SMAD4 in colorectal cancer. Cancer Biol Ther 2024;25:2392341. [Crossref] [PubMed]
5. El Agy F, El Bardai S, El Otmani I, et al. Mutation status and prognostic value of KRAS and NRAS mutations in Moroccan colon cancer patients: A first report. PLoS One 2021;16:e0248522. [Crossref] [PubMed]
6. Zhu G, Pei L, Xia H, et al. Role of oncogenic KRAS in the prognosis, diagnosis and treatment of colorectal cancer. Mol Cancer 2021;20:143. [Crossref] [PubMed]
7. Smith EM, Pendlebury DF, Nandakumar J. Structural biology of telomeres and telomerase. Cell Mol Life Sci 2020;77:61-79. [Crossref] [PubMed]
8. Liu M, Zhang Y, Jian Y, et al. The regulations of telomerase reverse transcriptase (TERT) in cancer. Cell Death Dis 2024;15:90. [Crossref] [PubMed]
9. Sánchez-Vázquez R, Martínez P, Blasco MA. AKT-dependent signaling of extracellular cues through telomeres impact on tumorigenesis. PLoS Genet 2021;17:e1009410. [Crossref] [PubMed]
10. Funder JW. Primary Aldosteronism. Hypertension 2019;74:458-66. [Crossref] [PubMed]
11. Ji Z, Li J, Wang J. Up-regulated RFC2 predicts unfavorable progression in hepatocellular carcinoma. Hereditas 2021;158:17. [Crossref] [PubMed]
12. Sun Y, Ren D, Yang C, et al. TRIM15 promotes the invasion and metastasis of pancreatic cancer cells by mediating APOA1 ubiquitination and degradation. Biochim Biophys Acta Mol Basis Dis 2021;1867:166213. [Crossref] [PubMed]
13. Mengying Z. Discovery of active compounds targeting the deubiquitinating enzyme USP 2 and its mechanism of antitumor action [Doctor]: University of Chinese Academy of Sciences (Shanghai Institute of Materia Medica, Chinese Academy of Sciences); 2022.
14. He J, Lee HJ, Saha S, et al. Inhibition of USP2 eliminates cancer stem cells and enhances TNBC responsiveness to chemotherapy. Cell Death Dis 2019;10:285. [Crossref] [PubMed]
15. Zhang J, Liu S, Li Q, et al. The deubiquitylase USP2 maintains ErbB2 abundance via counteracting endocytic degradation and represents a therapeutic target in ErbB2-positive breast cancer. Cell Death Differ 2020;27:2710-25. [Crossref] [PubMed]
16. Kuang Z, Liu X, Zhang N, et al. USP2 promotes tumor immune evasion via deubiquitination and stabilization of PD-L1. Cell Death Differ 2023;30:2249-64. [Crossref] [PubMed]
17. Li K, Chen B, Xu A, et al. TRIM7 modulates NCOA4-mediated ferritinophagy and ferroptosis in glioblastoma cells. Redox Biol 2022;56:102451. [Crossref] [PubMed]
18. Chen Q, Zhang T, Zeng R, et al. The E3 ligase TRIM7 suppresses the tumorigenesis of gastric cancer by targeting SLC7A11. Sci Rep 2024;14:6655. [Crossref] [PubMed]
19. Zhu L, Qin C, Li T, et al. The E3 ubiquitin ligase TRIM7 suppressed hepatocellular carcinoma progression by directly targeting Src protein. Cell Death Differ 2020;27:1819-31. [Crossref] [PubMed]
20. Huang N, Sun X, Li P, et al. TRIM family contribute to tumorigenesis, cancer development, and drug resistance. Exp Hematol Oncol 2022;11:75. [Crossref] [PubMed]
21. Raja E, Morikawa M, Nishida J, et al. Tyrosine kinase Eph receptor A6 sensitizes glioma-initiating cells towards bone morphogenetic protein-induced apoptosis. Cancer Sci 2019;110:3486-96. [Crossref] [PubMed]
22. Li S, Ma Y, Xie C, et al. EphA6 promotes angiogenesis and prostate cancer metastasis and is associated with human prostate cancer progression. Oncotarget 2015;6:22587-97. [Crossref] [PubMed]
23. Tong W, Liu W, Hu G. Role of erythropoietin-producing hepatocyte receptors in the pathogenesis of liver fibrosis and hepatocellular carcinoma. Journal of Clinical Hepatology 2021;37:2663-6.
24. Crocco P, Saiardi A, Wilson MS, et al. Contribution of polymorphic variation of inositol hexakisphosphate kinase 3 (IP6K3) gene promoter to the susceptibility to late onset Alzheimer's disease. Biochim Biophys Acta 2016;1862:1766-73. [Crossref] [PubMed]
25. Tan J, Yu CY, Wang ZH, et al. Genetic variants in the inositol phosphate metabolism pathway and risk of different types of cancer. Sci Rep 2015;5:8473. [Crossref] [PubMed]
26. Sheng C, Wang Z, Yao C, et al. CALML6 Controls TAK1 Ubiquitination and Confers Protection against Acute Inflammation. J Immunol 2020;204:3008-18. [Crossref] [PubMed]
27. Ishikawa S, Umemura M, Nakakaji R, et al. EP4-induced mitochondrial localization and cell migration mediated by CALML6 in human oral squamous cell carcinoma. Commun Biol 2024;7:567. [Crossref] [PubMed]
28. Wang C, Ding ZW, Zheng CG, et al. COCH predicts survival and adjuvant TACE response in patients with HCC. Oncol Lett 2021;21:275. [Crossref] [PubMed]
29. Hu Y, Zou D. Combined mRNAs and clinical factors model on predicting prognosis in patients with triple-negative breast cancer. PLoS One 2021;16:e0260811. [Crossref] [PubMed]
30. Yi F, Wang Y, Xu A. Evaluation Value of Peripheral Absolute Eosinophil Count for the Prognosis of Lung Cancer. Chinese General Practice 2024;27:4001-8.
31. Ge W, Wu W. Influencing Factors and Significance of Tumor-associated Macrophage Polarization in Tumor Microenvironment. Zhongguo Fei Ai Za Zhi 2023;26:228-37. [Crossref] [PubMed]
32. Wu J, Lu AD, Zhang LP, et al. Study of clinical outcome and prognosis in pediatric core binding factor-acute myeloid leukemia. Zhonghua Xue Ye Xue Za Zhi 2019;40:52-7. [Crossref] [PubMed]
33. Zou Q, Hu B, Yu HC, et al. Characteristics of CD8+ T cell infiltration in colorectal cancer and their correlation with prognosis. Zhonghua Wei Chang Wai Ke Za Zhi 2021;24:1086-92. [Crossref] [PubMed]
34. Mao Y, Xu Y, Chang J, et al. The immune phenotypes and different immune escape mechanisms in colorectal cancer. Front Immunol 2022;13:968089. [Crossref] [PubMed]
35. Pagès F, Mlecnik B, Marliot F, et al. International validation of the consensus Immunoscore for the classification of colon cancer: a prognostic and accuracy study. Lancet 2018;391:2128-39. [Crossref] [PubMed]
36. Teillaud JL, Houel A, Panouillot M, et al. Tertiary lymphoid structures in anticancer immunity. Nat Rev Cancer 2024;24:629-46. [Crossref] [PubMed]
37. Jin Y, Fu X, Chen M, et al. CD147 promotes cell proliferation and reduces apoptosis in colon cancer cells by regulating the MAPK signaling pathway. Chinese Journal of Clinicians 2024;18:474-80. (Electronic Edition).
38. Scott AJ, Song EK, Bagby S, et al. Evaluation of the efficacy of dasatinib, a Src/Abl inhibitor, in colorectal cancer cell lines and explant mouse model. PLoS One 2017;12:e0187173. [Crossref] [PubMed]
39. Baretti M, Karunasena E, Zahurak M, et al. A phase 2 trial of gemcitabine and docetaxel in patients with metastatic colorectal adenocarcinoma with methylated checkpoint with forkhead and ring finger domain promoter and/or microsatellite instability phenotype. Clin Transl Sci 2021;14:954-63. [Crossref] [PubMed]
40. Tournigand C, Chibaudel B, Samson B, et al. Bevacizumab with or without erlotinib as maintenance therapy in patients with metastatic colorectal cancer (GERCOR DREAM; OPTIMOX3): a randomised, open-label, phase 3 trial. Lancet Oncol 2015;16:1493-505. [Crossref] [PubMed]
41. Fisher GA, Kuo T, Ramsey M, et al. A phase II study of gefitinib, 5-fluorouracil, leucovorin, and oxaliplatin in previously untreated patients with metastatic colorectal cancer. Clin Cancer Res 2008;14:7074-9. [Crossref] [PubMed]
42. Chai Y, Liu JL, Zhang S, et al. The effective combination therapies with irinotecan for colorectal cancer. Front Pharmacol 2024;15:1356708. [Crossref] [PubMed]
43. Hoff PM, Hochhaus A, Pestalozzi BC, et al. Cediranib plus FOLFOX/CAPOX versus placebo plus FOLFOX/CAPOX in patients with previously untreated metastatic colorectal cancer: a randomized, double-blind, phase III study (HORIZON II). J Clin Oncol 2012;30:3596-603. [Crossref] [PubMed]
44. Kang B, Zhang X, Wang W, et al. The Novel IGF-1R Inhibitor PB-020 Acts Synergistically with Anti-PD-1 and Mebendazole against Colorectal Cancer. Cancers (Basel) 2022;14:5747. [Crossref] [PubMed]
45. Cervantes A, Adam R, Roselló S, et al. Metastatic colorectal cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol 2023;34:10-32. [Crossref] [PubMed]
46. Reinert T, Schøler LV, Thomsen R, et al. Analysis of circulating tumour DNA to monitor disease burden following colorectal cancer surgery. Gut 2016;65:625-34. [Crossref] [PubMed]
47. Guan Y, Wu S, Zhang X, et al. Recent Advances in Research on Cost-effectiveness of Colorectal Cancer Screening. Chinese General Practice 2021;24:4177-84.
48. Yang Q, Liu Y, Zeng Y, et al. Mechanism, detection and clinical implication of tumor heterogeneity. Journal of International Oncology 2017;922-5.
49. Sobral D, Martins M, Kaplan S, et al. Genetic and microenvironmental intra-tumor heterogeneity impacts colorectal cancer evolution and metastatic development. Commun Biol 2022;5:937. [Crossref] [PubMed]