Construction of a prognostic model based on ferroptosis- and mitochondrial metabolism-related genes for patients with breast cancer

Abstract

Mitochondrial metabolism (MM)-mediated ferroptosis plays a critical role in breast cancer (BC). However, the potential targets based on ferroptosis and MM in BC remain poorly understood. This study aimed to explore the prognostic role of ferroptosis- and MM-related genes (FPMMs) in BC. Differentially expressed FPMMs were identified, and functional analyses were performed. Univariate Cox, LASSO, and multivariate Cox regression analyses were used to screen hub genes, and a prognostic risk model was then constructed and validated in external datasets. Gene set variation analysis was conducted to investigate their regulatory functions. Furthermore, immune infiltration analysis was performed using the "quantiseq" algorithm. We identified 206 differentially expressed FPMMs. A prognostic risk model consisting of 6 genes (BRD4, FLT3, SIAH2, CS, EMC2, and PI3KCA) was constructed, exhibiting good predictive capability and stability. These 6 prognostic genes were dysregulated in BC, with PI3KCA exhibiting the highest mutation frequency. Gene set variation analysis further revealed that the PI3K-AKT-mTOR signaling was suppressed in BC. In addition, the risk score based on the prognostic model was associated with immune infiltration, particularly with B cells, T cells, CD4, and dendritic cells. Our study highlights the potential of the prognostic model based on FPMMs as a valuable tool for BC prognosis prediction.

Keywords: breast cancer; ferroptosis; metabolism; mitochondrion; prognosis.

Figure 1.

Functional enrichment analysis of differentially expressed FPMMs in BC. (A) Volcano plot of DEGs in BC. Red dots are upregulated genes, and blue dots are downregulated genes. (B) The Venn diagram of DEGs, MM-related genes, and ferroptosis-related genes. (C–E) GO enrichment analysis. (F) KEGG pathway enrichment analysis. BC = breast cancer, BP = biological process, CC = cellular component, DEGs = differentially expressed genes, FPMMs = ferroptosis- and mitochondrial metabolism-related genes, GO = Gene Ontology, KEGG = Kyoto Encyclopedia of Genes and Genomes, MF = molecular function, MM = mitochondrial metabolism.

Figure 2.

Construction of FPMM-based prognostic risk model. (A) Univariate Cox regression analysis based on the differentially expressed FPMMs. (B–C) LASSO regression analysis removed redundant features. (D) Multivariate Cox regression analysis identified 6 FPMM-related prognostic genes. FPMMs = ferroptosis- and mitochondrial metabolism-related genes.

Figure 3.

Assessment of the FPMM-based prognostic risk model. (A) Kaplan–Meier curve of BC patients in the low-risk and high-risk groups based on the TCGA cohort. (B) ROC curve at 3-, 5-, and 10-year for BC patients in the TCGA cohort. (C) Kaplan–Meier curve of BC patients in the low-risk and high-risk groups based on the GSE20685 dataset. (D) ROC curve at 3-, 5-, and 10-year for BC patients in the GSE20685 dataset. BC = breast cancer, FPMMs = ferroptosis- and mitochondrial metabolism-related genes, ROC = receiver operator characteristic.

Figure 4.

Independence of the FPMM-based risk score from clinical parameters of BC. (A–F) Risk score differences in gender, age, pathology, T, N, and M stages. (G) Nomogram of clinical parameters. (H) Calibration curves for 3-, 5-, and 10-year.*P < .05, **P < .01, ***P < .001. BC = breast cancer, FPMMs = ferroptosis- and mitochondrial metabolism-related genes, ns = no significance, patho. = pathology, pb. = probability, yr. = year.

Figure 5.

Expression levels of FPMM-related prognostic genes. (A) Expression levels of 6 prognostic genes in low- and high-risk groups. (B) Expression levels of 6 prognostic genes in normal and tumor samples. (C) Expression levels of SLC7A11 and SLC3A2 in normal and tumor samples. (D) Pearson correlation of 6 prognostic genes with SLC7A11 and SLC3A2. *P < .05, **P < .01, ***P < .001, ****P < .0001. coef. = coefficient, corre. = correlation, FPMMs = ferroptosis- and mitochondrial metabolism-related genes.

Figure 6.

Mutation analysis. (A) Mutation frequency of prognostic genes in the low-risk group. (B) Mutation frequency of prognostic genes in the high-risk group.

Figure 7.

GSVA analysis of prognostic genes. GSVA = gene set variation analysis.

Figure 8.

Immune infiltration analysis. (A) Immune cell infiltration in high- and low-risk groups using “quantiseq” algorithm. (B) Correlation between immune cells and correlation of immune cells with 6 prognostic genes or risk scores. *P < .05, **P < .01, ***P < .001.