A novel predictive model based on inflammatory response-related genes for predicting endometrial cancer prognosis and its experimental validation

Abstract

Inflammatory response is an important feature of most tumors. Local inflammation promotes tumor cell immune evasion and chemotherapeutic drug resistance. We aimed to build a prognostic model for endometrial cancer patients based on inflammatory response-related genes (IRGs). RNA sequencing and clinical data for uterine corpus endometrial cancer were obtained from TCGA datasets. LASSO-penalized Cox regression was used to obtain the risk formula of the model: the score = e^{sum(corresponding coefficient × each gene’s expression)}. The “ESTIMATE” and “pRRophetic” packages in R were used to evaluate the tumor microenvironment and the sensitivity of patients to chemotherapy drugs. Data sets from IMvigor210 were used to evaluate the efficacy of immunotherapy in cancer patients. For experimental verification, 37 endometrial cancer and 43 normal endometrial tissues samples were collected. The mRNA expression of the IRGs was measured using qRT-PCR. The effects of IRGs on the malignant biological behaviors of endometrial cancer were detected using CCK-8, colony formation, Transwell invasion, and apoptosis assays. We developed a novel prognostic signature comprising 13 IRGs, which is an independent prognostic marker for endometrial cancer. A nomogram was developed to predict patient survival accurately. Three key IRGs (LAMP3, MEP1A, and ROS1) were identified in this model. Furthermore, we verified the expression of the three key IRGs using qRT-PCR. Functional experiments also confirmed the influence of the three key IRGs on the malignant biological behavior of endometrial cancer. Thus, a characteristic model constructed using IRGs can predict the survival, chemotherapeutic drug sensitivity, and immunotherapy response in patients with endometrial cancer.