Identification and Validation of an Autophagy-Related lncRNA Signature for Patients With Breast Cancer

Abstract

Background: Autophagy is a "self-feeding" phenomenon of cells, which is crucial in mammalian development. Long non-coding RNA (lncRNA) is a new regulatory factor for cell autophagy, which can regulate the process of autophagy to affect tumor progression. However, poor attention has been paid to the roles of autophagy-related lncRNAs in breast cancer.

Objective: This study aimed to construct an autophagy-related lncRNA signature that can effectively predict the prognosis of breast cancer patients and explore the potential functions of these lncRNAs.

Methods: The RNA sequencing (RNA-Seq) data of breast cancer patients was collected from The Cancer Genome Atlas (TCGA) database and the GSE20685 database. Multivariate Cox analysis was implemented to produce an autophagy-related lncRNA signature in the TCGA cohort. The signature was then validated in the GSE20685 cohort. The receiver operator characteristic (ROC) curve was performed to evaluate the predictive ability of the signature. Gene set enrichment analysis (GSEA) was used to explore the potential functions based on the signature. Finally, the study developed a nomogram and internal verification based on the autophagy-related lncRNAs.

Results: A signature composed of 9 autophagy-related lncRNAs was determined as a prognostic model, and 1,109 breast cancer patients were divided into high-risk group and low-risk group based on median risk score of the signature. Further analysis demonstrated that the over survival (OS) of breast cancer patients in the high-risk group was poorer than that in the low-risk group based on the prognostic signature. The area under the curve (AUC) of ROC curve verified the sensitivity and specificity of this signature. Additionally, we confirmed the signature is an independent factor and found it may be correlated to the progression of breast cancer. GSEA showed gene sets were notably enriched in carcinogenic activation pathways and autophagy-related pathways. The qRT-PCR identified 5 lncRNAs with significantly differential expression in breast cancer cells based on the 9 lncRNAs of the prognostic model, and the results were consistent with the tissues.

Conclusion: In summary, our signature has potential predictive value in the prognosis of breast cancer and these autophagy-related lncRNAs may play significant roles in the diagnosis and treatment of breast cancer.

Keywords: The Cancer Genome Atlas; autophagy; breast cancer; long non-coding RNA; prognostic signature.

Figure 1

Flowchart for identification and validation of an autophagy-related lncRNA signature in breast cancer patients. The study was carried out in TCGA and GSE20685 database. The TCGA training set was used to identify prognostic lncRNAs. The multivariate Cox hazard model analysis was performed to construct a prognostic signature based on the prognostic lncRNAs. The prognosis signature was validated in the GSE20685 testing set.

Figure 2

Identification of an autophagy-related lncRNA signature in breast cancer. (A) Venn diagram describes 180 shared lncRNAs from 1,270 autophagy-related lncRNAs and 1,146 lncRNAs of GSE20685. (B) The network of autophagy-genes and lncRNAs. The blue nodes indicate autophagy genes and the pink nodes indicate lncRNA. The co-expression network is performed by CYTOSCAPE 3.7.2. (C) Univariate Cox regression analysis was used to confirm the lncRNAs associated with autophagy were strongly with patients’ OS in training dataset. (D) Multivariate Cox regression analysis to construct the prognostic signature.

Figure 3

A depiction of the regulation network of the prognostic lncRNAs and autophagy-related genes in breast cancer. (A) The blue nodes indicate autophagy genes and the pink nodes indicate lncRNA. The co-expression network is performed by CYTOSCAPE 3.7.2. (B) Among the risk types, the lncRNAs linked to dark green is protective lncRNAs, and dark red lncRNAs represents risk lncRNAs.

Figure 4

Risk score analysis of the prognostic signature in the training dataset. (A) Kaplan-Meier survival analysis for high-risk group and low-risk group. (B) ROC curves for predicting OS based on risk score. (C) Risk score distribution, survival status, and expression heat map.

Figure 5

Risk score analysis of the prognostic signature in the testing dataset. (A) Kaplan-Meier survival for low-risk group and high-risk group. (B) ROC curves for predicting OS based on risk score. (C) Risk score distribution, survival status, and expression heat map.

Figure 6

Cox regression analysis of clinical characteristics related to OS in training dataset and testing dataset. (A, B) Univariate and multivariate Cox regression analysis of clinical characteristics related to OS in training dataset. (C, D) Univariate and multivariate Cox regression analysis of clinical characteristics related to OS in testing dataset.

Figure 7

Gene set enrichment analysis. (A–F) GSEA suggested notably enrichment of cancer-related pathways in the high-risk group based on training set. (G–I) GSEA suggested significant autophagy-related enrichment based on training set.

Figure 8

The nomogram establishing and the prognostic value validating base on 9 lncRNAs in TCGA dataset. (A) Establishment of a nomogram for 1-, 3-, and 5-year OS prediction in breast cancer. (B) Validation the prognostic value of these 9 autophagy-related lncRNAs in breast cancer by Kaplan Meier-plotter curve.

Figure 9

The expression of 9 lncRNAs in breast cancer tissues and cells. (A) Heat map of 5 lncRNAs with significantly differential expression in breast cancer tissues. (B–E) qRT-PCR results showed that USP30-AS1, TFAP2A-AS1, MAPT-AS1, and LINC01087 expression were higher in breast cancer cell lines than in the normal cell lines. *P<0.05, **P<0.01 (F) qRT-PCR results showed that HOXB-AS1 expression was lower in breast cancer cell lines than in the normal cell lines. ***P<0.001.