Comprehensive Analysis of Ferroptosis-Related LncRNAs in Breast Cancer Patients Reveals Prognostic Value and Relationship With Tumor Immune Microenvironment

Abstract

Background: Breast cancer (BC) is a heterogeneous malignant tumor, leading to the second major cause of female mortality. This study aimed to establish an in-depth relationship between ferroptosis-related LncRNA (FRlncRNA) and the prognosis as well as immune microenvironment of the patients with BC. Methods: We downloaded and integrated the gene expression data and the clinical information of the patients with BC from The Cancer Genome Atlas (TCGA) database. The co-expression network analysis and univariate Cox regression analysis were performed to screen out the FRlncRNAs related to prognosis. A cluster analysis was adopted to explore the difference of immune microenvironment between the clusters. Furthermore, we determined the optimal survival-related FRLncRNAs for final signature by LASSO Cox regression analysis. Afterward, we constructed and validated the prediction models, which were further tested in different subgroups. Results: A total of 31 FRLncRNAs were filtrated as prognostic biomarkers. Two clusters were determined, and C1 showed better prognosis and higher infiltration level of immune cells, such as B cells naive, plasma cells, T cells CD8, and T cells CD4 memory activated. However, there were no significantly different clinical characters between the clusters. Gene Set Enrichment Analysis (GSEA) revealed that some metabolism-related pathways and immune-associated pathways were exposed. In addition, 12 FRLncRNAs were determined by LASSO analysis and used to construct a prognostic signature. In both the training and testing sets, patients in the high-risk group had a worse survival than the low-risk patients. The area under the curves (AUCs) of receiver operator characteristic (ROC) curves were about 0.700, showing positive prognostic capacity. More notably, through the comprehensive analysis of heatmap, we regarded LINC01871, LINC02384, LIPE-AS1, and HSD11B1-AS1 as protective LncRNAs, while LINC00393, AC121247.2, AC010655.2, LINC01419, PTPRD-AS1, AC099329.2, OTUD6B-AS1, and LINC02266 were classified as risk LncRNAs. At the same time, the patients in the low-risk groups were more likely to be assigned to C1 and had a higher immune score, which were consistent with a better prognosis. Conclusion: Our research indicated that the ferroptosis-related prognostic signature could be used as novel biomarkers for predicting the prognosis of BC. The differences in the immune microenvironment exhibited by BC patients with different risks and clusters suggested that there may be a complementary synergistic effect between ferroptosis and immunotherapy.

Keywords: breast cancer; ferroptosis; immune microenvironment; lncRNA; prognosis.

Figure 1

Co-expression network analysis on target LncRNAs related to ferroptosis-related genes. The red node represents the ferroptosis-related genes and the blue node represents the LncRNA coexpressed with the ferroptosis-related genes.

Figure 2

Univariate Cox regression analysis of the FRLncRNAs related to prognosis. The red and green boxes represent risk factors or protective factors, respectively.

Figure 3

The heatmap shows the differential expression of the prognostic between normal and tumor tissues. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001.

Figure 4

Hierarchical consensus clustering based on the prognostic FRLncRNAs. (A) Consensus clustering analysis identification of two clusters (n = 896); (B) Cumulative distribution function (CDF) for k = 2–9; (C) Kaplan–Meier (K–M) curves for the 896 patients breast cancer (BC) stratified by cluster; (D) Heatmap on the prognostic FRLncRNAs ordered by clusters. The association with clusters, survival probability, and clinical information (age, stage, T stage, N stage, and M stage) were investigated.

Figure 5

Correlation analysis between PD-L1 gene and the prognostic FRLncRNAs. Blue is a negative correlation, red is a positive correlation, and if there is an asterisk in the grid between two genes, they are significantly correlated.

Figure 6

Evaluation of the correlation with immune features between clusters. (A) Immune score between clusters; (B) Stromal score between clusters; (C) The violin plot of comparison of 22 types of immune cells between clusters.

Figure 7

Gene Set Enrichment Analysis (GSEA) between the clusters (A–L).

Figure 8

Selection of the optimal survival-related LncRNAs by LASSO Cox regression. (A) LASSO coefficient profiles of the candidate survival-related LncRNAs. A coefficient profile plot was produced against the log λ sequence. (B) Dotted vertical lines were drawn at the optimal values using the minimum criteria.

Figure 9

The 12-LncRNA signature in the training set. (A) The distribution of risk score; (B) the survival time and status of patients; (C) the bottom shows the heatmap of 12-LncRNA expression profile. Colors from red to green indicate decreasing expression level from high to low; (D) the K–M curves for high- and low-risk groups. Purple color represents the low-risk group, whereas red color represents the high-risk group; (E) receiver operator characteristic (ROC) curves for patients with BC in testing set. AUC, area under the curve.

Figure 10

Testing for the 12-LncRNA signature. (A) The distribution of risk score; (B) the survival time and status of patients; (C) the bottom shows the heatmap of 12-LncRNA expression profile. Colors from red to green indicate decreasing the expression level from high to low; (D) the K–M curves for high- and low-risk groups. Purple color represents the low-risk group, whereas red color represents the high-risk group; (E) ROC curves for patients with BC in the testing set. (F) The K–M curves for the high- and low-risk groups in GSE69031 cohort. Purple color represents the low-risk group, whereas red color represents the high-risk group; (G) ROC curves for patients with BC in GSE69031 cohort. AUC, area under the curve.

Figure 11

Identification of the independence of risk score prognostic model by the Cox regression analyses. (A) The univariate and multivariate Cox regression analyses of the risk score in the training cohort. (B) The univariate and multivariate Cox regression analyses of the risk score in the test cohort. The green boxes represent risk factors of the univariate Cox regression analysis and the red boxes represent risk factors of the multivariate Cox regression analysis.

Figure 12

Validation of the risk score prognostic model among the different clinical groups. (A) Age ≤65; (B) age >65; (C) T1-2; (D) T3-4; (E) N0; (F) N1–3; (G) stage I–II; (H) stage III–IV. The blue lines represent the low-risk groups, the red lines represent the high-risk groups.

Figure 13

Comprehensive analysis of the differences between the high- and low-risk groups.