A Prognostic Model Based on Immune-Related Long Non-Coding RNAs for Patients With Cervical Cancer

Abstract

Objectives: The study is performed to analyze the relationship between immune-related long non-coding RNAs (lncRNAs) and the prognosis of cervical cancer patients. We constructed a prognostic model and explored the immune characteristics of different risk groups. Methods: We downloaded the gene expression profiles and clinical data of 227 patients from The Cancer Genome Atlas database and extracted immune-related lncRNAs. Cox regression analysis was used to pick out the predictive lncRNAs. The risk score of each patient was calculated based on the expression level of lncRNAs and regression coefficient (β), and a prognostic model was constructed. The overall survival (OS) of different risk groups was analyzed and compared by the Kaplan-Meier method. To analyze the distribution of immune-related genes in each group, principal component analysis and Gene set enrichment analysis were carried out. Estimation of STromal and Immune cells in MAlignant Tumors using Expression data was performed to explore the immune microenvironment. Results: Patients were divided into training set and validation set. Five immune-related lncRNAs (H1FX-AS1, AL441992.1, USP30-AS1, AP001527.2, and AL031123.2) were selected for the construction of the prognostic model. Patients in the training set were divided into high-risk group with longer OS and low-risk group with shorter OS (p = 0.004); meanwhile, similar result were found in validation set (p = 0.013), combination set (p < 0.001) and patients with different tumor stages. This model was further confirmed in 56 cervical cancer tissues by Q-PCR. The distribution of immune-related genes was significantly different in each group. In addition, the immune score and the programmed death-ligand 1 expression of the low-risk group was higher. Conclusions: The prognostic model based on immune-related lncRNAs could predict the prognosis and immune status of cervical cancer patients which is conducive to clinical prognosis judgment and individual treatment.

Keywords: Cervical cancer; Gene express; Immunology; Long non-coding RNA; Prognosis.

FIGURE 1

Construction of the five immune-related Long non-coding RNAs prognostic model of cervical cancer patients. Based on the median risk score, we divided patients from the training set into high-risk group with 84 individuals and low-risk group with 83 individuals. (A), Kaplan–Meier curve of the high-risk and low-risk groups of training set showed differences in survival rate. (B), Kaplan–Meier curve of the high-risk and low-risk groups of validation set also showed differences in survival rate.

FIGURE 2

Characteristics of the patients in the combination set based on the constructed model. (A), Kaplan–Meier curve of the high-risk and low-risk groups of combination set showed differences in survival rate. (B), Distribution of risk score, survival status of each patients, and heat map expression of five immune-related Long non-coding RNAs in high-risk and low-risk groups were presented.

FIGURE 3

Applicability and reliability of the prognostic model. (A,B) Kaplan–Meier curve of FIGO stages I and II of the high- and low-risk groups. (C), Receiver operating characteristic (ROC) curve of the risk score, age, grade, and stage. The area under the ROC curve (AUC) of the prediction model was 0.780, which was much better than that of age (0.505), grade (0.620) and FIGO stage (0.711). (D) Kaplan–Meier curve of 56 cervical cancer patients. 56 patients from our cancer center were selected and QRT-PCR was used to calculate the expression level of five Long non-coding RNAs, finally the prognostic model was validated using their accordingly clinical data.

FIGURE 4

Immune status of the high- and low-risk groups. (A), Principal component analysis (PCA) showed that whole gene expression profiles between the high- and low-risk groups were mixed up. (B), PCA indicated that the distribution of immune-related genes between the high- and low-risk groups were significantly different. (C), Gene set enrichment analysis implied that the low-risk group had adequate immune response and immune system process pathways. NES, Normalized Enrichment Score.

FIGURE 5

Immune microenvironment of the high- and low-risk groups. (A), Estimation of STromal and Immune cells in MAlignant Tumors using Expression data (ESTIMATE) analysis showed that the immune score of the low-risk group was higher than that of the high-risk group. (B), ESTIMATE analysis indicated that the low-risk group had more immune and stromal cells but lower tumor purity. (C), Programmed death-ligand 1 expression of the low-risk group was higher than that of the high-risk group.