Bioinformatics-based multi-omics and machine learning analysis identifies stemness-associated molecular subtypes and a prognostic index in breast cancer

基于生物信息学的多组学和机器学习分析可识别乳腺癌中与干细胞特性相关的分子亚型和预后指数

阅读:1

Abstract

BACKGROUND: Breast cancer (BC) recurrence and therapy resistance are driven by cancer stem cells. However, molecular subtypes specifically driven by tumor stemness remain poorly defined, limiting their clinical utility. This study aimed to determine the molecular types associated with BC stemness and to assess the prognostic value of associated risk models. METHODS: BC samples from The Cancer Genome Atlas (TCGA) were analyzed for mRNA stemness indices (mRNAsi) using the one-class logistic regression (OCLR) algorithm, and target gene modules highly correlated with mRNAsi were identified by weighted gene co-expression network analysis (WGCNA). BC stemness subtypes were constructed by screening selected prognostic genes using univariate Cox regression analysis. These subtypes had potential precision therapy applications in biological function, somatic mutation, status of tumor microenvironment and immunotherapy response prediction. In addition, an mRNAsi-related risk model was developed using least absolute shrinkage and selection operator (LASSO) regression analysis and validated in external datasets (GSE20685, METABRIC). Independent prognostic analyses were also performed on the mRNAsi-related risk model. A nomogram was subsequently developed by integrating this risk score with clinical factors, and its accuracy was assessed via calibration curves. RESULTS: In the TCGA-BRCA cohort (n=973), three gene modules were identified that showed different correlations with the mRNAsi. A two-subtype classification in BC patients was established based on 11 mRNAsi-related genes that exhibited prognostic value. Among them, C2 patients had higher stemness, poorer prognosis, greater tumor mutation burden, and lower levels of immune cell infiltration. A robust 7-gene prognostic signature (BEND5, TNN, PDLIM4, CD24, POP1, PRDX1, PGK1) was developed and validated in external cohorts (GSE20685, n=327; METABRIC, n=1403), demonstrating significant predictive value for patient stratification. Moreover, this signature is an independent prognostic factor for BC patients. A nomogram integrating this signature with key clinical factors (including age, tumor size, lymph node status, and distant metastasis) showed excellent calibration between predicted and observed survival outcomes. CONCLUSIONS: This study classified BC by mRNAsi-related genes and established corresponding risk models. The findings of the present study propose a novel classification tool based on stem cell characteristics, which has the potential to be employed for prognostic stratification in BC patients and to offer guidance for developing personalised treatment strategies.

特别声明

1、本页面内容包含部分的内容是基于公开信息的合理引用;引用内容仅为补充信息,不代表本站立场。

2、若认为本页面引用内容涉及侵权,请及时与本站联系,我们将第一时间处理。

3、其他媒体/个人如需使用本页面原创内容,需注明“来源:[生知库]”并获得授权;使用引用内容的,需自行联系原作者获得许可。

4、投稿及合作请联系:info@biocloudy.com。