Identification of disulfidptosis-related clusters and construction of a disulfidptosis-related gene prognostic signature in triple-negative breast cancer

Abstract

Objective: This study aimed to explore disulfidptosis-related clusters of triple-negative breast cancer (TNBC) and build a reliable disulfidptosis-related gene signature for forecasting TNBC prognosis.

Methods: The disulfidptosis-related clusters of TNBC were identified based on public datasets, and a comparative analysis was conducted to assess their differences in the overall survival (OS) and immune cell infiltration. Morever, the differentially expressed genes (DEGs) between clusters were recognized. Then, the prognostic DEGs were then chosen. A prognostic signature was constructed by the prognostic DEGs, followed by nomogram construction, drug sensitivity, immune correlation, immunotherapy response prediction, and cluster association analyses.

Results: Two disulfidptosis-related clusters of TNBC were identified, which had different OS and macrophage infiltration. Moreover, 235 DEGs were identified between two clusters. A prognostic signature was then constructed by five prognostic DEGs including HLA-DQA2, CCL13, GBP1, LAMP3, and SLC7A11. This signature was highly valuable in predicting prognosis. A nomogram was built by risk score and AJCC stage, which could forecast OS accurately. Moreover, patients with high-risk scores exhibited greater sensitivity to chemotherapy drugs such as lapatinib and had a lower immunotherapy response.

Conclusions: Two TNBC clusters linked to disulfidptosis were identified, with different OS and immune cell infiltration. Moreover, a five-disulfidptosis-related gene signature may be a powerful prognostic biomarker for TNBC.

Keywords: Disulfidptosis; Gene prognostic signature; Immunotherapy; Nomogram.

Fig. 1

Analysis of disulfidptosis-related genes based on TCGA-TNBC data. A: The expression difference of 10 disulfidptosis-related genes between tumor and normal samples. B: The mutation frequency of disulfidptosis-related genes. C: The correlation of Based on TCGA-TNBC data genes in tumor samples.

Fig. 2

Identification of disulfidptosis-related clusters. A: Cumulative distribution function (CDF) distribution curve. B: Heatmap of disulfidptosis-related clusters. C: Delta area line graph. D: Principal component analysis (PCA) plot of sample distribution in the two clusters. E: Comparison of disulfidptosis score of two clusters. F: Kaplan–Meier survival curve of two clusters. G: Heat maps of disulfidptosis-related gene expression in two clusters and their correlation with different clinical information. Red indicates high expression and blue indicates low expression. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3

Comparison of immune microenvironment between two clusters. A: The results of immune cell infiltration between two clusters using CIBERSORT algorithm. B. The results of immune cell infiltration between two clusters using the ssGSEA algorithm. C. The results of ESTIMATE score, immune score, and stromal score between two clusters using the ESTIMATE analysis. D. The expression of immune checkpoint genes between two clusters. E: The expression of HLA family genes between two clusters. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 4

Gene set enrichment analysis (GSEA) and identification of differentially expressed genes (DEGs) between two clusters. A: GSEA showed the differentially enriched hallmark gene sets between two clusters. B: Volcano plot of DEGs between two clusters. The blue and red dots represent downregulated and upregulated genes, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 5

Construction and validation of the prognostic signature. A: Stepwise Cox regression analysis revealed the optimal gene combination for construction of the prognostic signature. B: Analysis of the predictive performance of the prognostic signature based on TCGA-TNBC training dataset. C: Analysis of the predictive performance of the prognostic signature based on GSE103091 dataset.

Fig. 6

Establishment and validation of a nomogram. A: Univariate and multivariate Cox regression analyses for screening independent prognostic factors. B: The constructed nomogram. C: The calibration curves of nomogram. D: The Kaplan-Meier survival curve showed the prognostic value of nomogram. E: receiver operating characteristic (ROC) curves showed the predictive performance of nomogram.

Fig. 7

Drug sensitivity, immune cell infiltration, immunotherapy response and cluster distribution of two risk groups. A: Drug sensitivity analysis showed the most significant drugs between two risk groups. B: The results of immune cell infiltration between two risk groups using the ssGSEA algorithm. C. The correlation between immune cell infiltration proportion and prognostic signature gene expression. D: Analysis of immunotherapy response of two risk groups, including TIDE, CYT and TLS scores. E: Analysis of the association between risk score and immunotherapy response using a validation dataset GSE103668. F: The association between cluster distribution and different risk groups. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.