Screening telomere-related genes to predict prognosis, immunotherapy response, and drug sensitivity in esophageal cancer using a machine learning approach.
Esophageal cancer (EC) is a common malignancy with poor prognosis. Telomeres, composed of repetitive DNA sequences and shelterin complexes, play important roles in tumor biology. However, the prognostic value of telomere-related genes (TRGs) in EC remains unclear.
TRGs were obtained from TelNet, and transcriptomic and clinical data were collected from The Cancer Genome Atlas (TCGA). Prognostic TRGs were identified using multivariate Cox regression analysis, Least Absolute Shrinkage and Selection Operator (LASSO), Random Forest (RF), and Support Vector Machine (SVM) algorithms to construct a risk model. Model performance was evaluated by Kaplan-Meier(K-M) and Receiver Operating Characteristic (ROC) analyses, and a nomogram integrating clinical variables was developed. Somatic mutations, immune infiltration, immunotherapy response, and drug sensitivity were compared between high- and low-risk groups. In addition, functional assays were performed to verify the biological role of the key gene PTGES3.
Six TRGs significantly associated with prognosis were identified to establish a risk model. High-risk patients had worse survival, higher TP53 and TTN mutation rates, altered immune infiltration, poorer predicted immunotherapy response, and distinct drug sensitivity profiles. Knockdown of PTGES3 significantly suppressed EC cell migration, invasion, and clonogenic ability, supporting its oncogenic role.
A TRGs-based prognostic model effectively predicts survival in EC and reveals associations with somatic mutations, immune infiltration, and drug sensitivity. Functional validation of PTGES3 further supports its potential as a therapeutic target.
TRGs were obtained from TelNet, and transcriptomic and clinical data were collected from The Cancer Genome Atlas (TCGA). Prognostic TRGs were identified using multivariate Cox regression analysis, Least Absolute Shrinkage and Selection Operator (LASSO), Random Forest (RF), and Support Vector Machine (SVM) algorithms to construct a risk model. Model performance was evaluated by Kaplan-Meier(K-M) and Receiver Operating Characteristic (ROC) analyses, and a nomogram integrating clinical variables was developed. Somatic mutations, immune infiltration, immunotherapy response, and drug sensitivity were compared between high- and low-risk groups. In addition, functional assays were performed to verify the biological role of the key gene PTGES3.
Six TRGs significantly associated with prognosis were identified to establish a risk model. High-risk patients had worse survival, higher TP53 and TTN mutation rates, altered immune infiltration, poorer predicted immunotherapy response, and distinct drug sensitivity profiles. Knockdown of PTGES3 significantly suppressed EC cell migration, invasion, and clonogenic ability, supporting its oncogenic role.
A TRGs-based prognostic model effectively predicts survival in EC and reveals associations with somatic mutations, immune infiltration, and drug sensitivity. Functional validation of PTGES3 further supports its potential as a therapeutic target.