Identification of novel cuproptosis-related lncRNA signatures to predict the prognosis and immune microenvironment of breast cancer patients

Abstract

Background: Cuproptosis is a new modality of cell death regulation that is currently considered as a new cancer treatment strategy. Nevertheless, the prognostic predictive value of cuproptosis-related lncRNAs in breast cancer (BC) remains unknown. Using cuproptosis-related lncRNAs, this study aims to predict the immune microenvironment and prognosis of BC patients. and develop new therapeutic strategies that target the disease.

Methods: The Cancer Genome Atlas (TCGA) database provided the RNA-seq data along with the corresponding clinical and prognostic information. Univariate and multivariate Cox regression analyses were performed to acquire lncRNAs associated with cuproptosis to establish predictive features. The Kaplan-Meier method was used to calculate the overall survival rate (OS) in the high-risk and low-risk groups. High risk and low risk gene sets were enriched to explore functional discrepancies among risk teams. The mutation data were analyzed using the "MAFTools" r-package. The ties of predictive characteristics and immune status had been explored by single sample gene set enrichment analysis (ssGSEA). Last, the correlation between predictive features and treatment condition in patients with BC was analyzed. Based on prognostic risk models, we assessed associations between risk subgroups and immune scores and immune checkpoints. In addition, drug responses in at-risk populations were predicted.

Results: We identified a set of 11 Cuproptosis-Related lncRNAs (GORAB-AS1, AC 079922.2, AL 589765.4, AC 005696.4, Cytor, ZNF 197-AS1, AC 002398.1, AL 451085.3, YTH DF 3-AS1, AC 008771.1, LINC 02446), based on which to construct the risk model. In comparison to the high-risk group, the low-risk patients lived longer (p < 0.001). Moreover, cuproptosis-related lncRNA profiles can independently predict prognosis in BC patients. The AUC values for receiver operating characteristics (ROC) of 1-, 3-, and 5-year risk were 0.849, 0.779, and 0.794, respectively. Patients in the high-risk group had lower OS than those in the low-risk group when they were divided into groups based on various clinicopathological variables. The tumor burden mutations (TMB) correlation analysis showed that high TMB had a worse prognosis than low-TMB, and gene mutations were found to be different in high and low TMB groups, such as PIK3CA (36% versus 32%), SYNE1 (4% versus 6%). Gene enrichment analysis indicated that the differential genes were significantly concentrated in immune-related pathways. The predictive traits were significantly correlated with the immune status of BC patients, according to ssGSEA results. Finally, high-risk patients showed high sensitivity in anti-CD276 immunotherapy and conventional chemotherapeutic drugs such as imatinib, lapatinib, and pazopanib.

Conclusion: We successfully constructed of a cuproptosis-related lncRNA signature, which can independently predict the prognosis of BC patients and can be used to estimate OS and clinical treatment outcomes in BRCA patients. It will serve as a foundation for further research into the mechanism of cuproptosis-related lncRNAs in breast cancer, as well as for the development of new markers and therapeutic targets for the disease.

Keywords: breast cancer; cuproptosis; lncRNA; tumor microenvironment; tumor mutation burden.

Figure 1

Construction of a signature for predicting features of cuproptosis-related lncRNA using the Lasso method. (A) Cross validation diagram. (B) LASSO coefficients of prognostic genes. (C) Sankey diagram of prognostic cuproptosis-related lncRNAs.

Figure 2

Relationship between prognosis and predictive features of patients with BC. (A) OS rates for BC patients in the high-risk and low-risk teams according to a Kaplan-Meier analysis. (B) Forest map of univariate Cox regression analysis. (C) Forest map of multivariate Cox regression analysis. (D) Risk score of BRCA patients calculated according to the model and division of high- and low-risk teams. (E) Gene expression heat maps. (F) Survival status. (G) ROC curve of the predictive characteristic and AUC of 1-, 3-, and 5-year survival. OS, overall survival; ROC, receiver operating characteristics; AUC, area under curve; T, tumor; N, lymph nodes; M, Metastasis.

Figure 3

Construction and verification of nomograms. (A) The 1,3and 5 years OS of patients with BC is predicted by nomograms that combine clinical pathology variables with risk scores. (B) The calibration curve tested the agreement between the actual OS rate and the predicted 1-,3-, and 5-year survival rates.

Figure 4

Kaplan-Meier survival curves of high-risk and low-risk teams in the patients sorted by different clinical pathological variables. (A, B) age. (C, D) T staging. (E, F) N staging. (G, H) M staging. T, tumor; N, lymph node; M, distant transfer.

Figure 5

Internal verification of OS prediction signatures based on the TCGA dataset. (A) Kaplan-Meier survival curves in the test cohort. (B) Kaplan-Meier survival curves in the total cohort. (C) ROC curves and AUC for 1-, 3-, and 5-year survival in the test cohort. (D) ROC curves and AUC for 1-, 3-, and 5-year survival in the total cohort.

Figure 6

Association of risk score with TMB and gene mutations. Kaplan-Meier curves for high and low TMB teams (A). Kaplan-Meier curve for patients layered by TMB and risk score (B). OncoPrint was built with a low-risk score (C) and a high-risk score (D).

Figure 7

Clinically relevant heat map and GO/KEGG pathway enrichment analysis. (A) On the basis of risk characteristics associated with prognosis, a heat map of distribution of cuproptosis-related lncRNA and clinical pathology variables was plotted. The more intense the red, the more intense the expression. The more intense the blue, the more subdued the expression. **p < 0.01, ***p < 0.001. (B) The graph depicts the GO and KEGG analysis of differential genes with high and low risk..

Figure 8

Immune infiltration cell score and immune-related function in high-risk and low-risk population. (A) The ssGSEA algorithm was used to calculate the levels of infiltration of 16 immune cells in high-risk and low-risk populations. (B) Relationship between predictive features and 13 immune-related functions. *p< 0.05; **p< 0.01; ***p< 0.001; ns, not significant.

Figure 9

Immune checkpoint expression in BRCA patients from two different risk groups. Expression of two immune checkpoints (CD274, CD276, PDCD1 and CTLA4) in the TCGA cohort. ANOVA was applied as significance test, * P <0.05, ** P <0.01, *** P <0.001.

Figure 10

(A–L) IC50 for small molecule drugs in high and low risk populations.(A): Bosutinib, (B): Cisplatin, (C): Cytarabine, (D): Gefitinib, (E): Gemcitabine, (F): Imatinib, (G): Lapatinib, (H): Lenalidomide, (I): Nilotinib, (J): Paclitaxel, (K): Pazopanib, (L): Sunitinib. IC50, half maximum inhibitory concentration.