Machine learning-based diagnostic and prognostic models for breast cancer: a new frontier on the clinical application of natural killer cell-related gene signatures in precision medicine DOI Creative Commons

Yutong Fang,

Rongji Zheng,

Yi Xiao

et al.

Frontiers in Immunology, Journal Year: 2025, Volume and Issue: 16

Published: May 27, 2025

Background Breast cancer (BC) remains a leading cause of cancer-related mortality among women worldwide. Natural killer (NK) cells play crucial role in the innate immune system and exhibit significant anti-tumor activity. However, NK cell-related genes (NRGs) BC diagnosis prognosis underexplored. With advent machine learning (ML) techniques, predictive modeling based on NRGs may offer new avenue for precision oncology. Methods We collected transcriptomic clinical data from The Cancer Genome Atlas (TCGA) Gene Expression Omnibus (GEO) databases. Differentially expressed (DEGs) were identified, key prognostic selected using univariate multivariate Cox regression analyses. constructed ML-based diagnostic models 12 algorithms evaluated their performance identifying optimal ML model. Additionally, risk model was developed LASSO-Cox regression, its validated independent cohorts. To explore potential mechanisms underlying differences between high-risk low-risk patient groups, as well drug treatment sensitivities, we conducted functional enrichment analysis, tumor microenvironment immunotherapy prediction, sensitivity mutation analysis. Results ULBP2, CCL5, PRDX1, IL21, NFATC2, CD2, VAV3 identified construction models. Among models, Random Forest (RF) demonstrated best performance, which robust distinguishing normal tissues both training validation In terms model, score effectively distinguished patients, with patients group exhibiting significantly poorer overall survival (OS) compared to those group, GEO Patients displayed increased proliferation, evasion, reduced cell infiltration, correlating lower response rates immunotherapy. Furthermore, analysis indicated that more sensitive Thapsigargin, Docetaxel, AKT inhibitor VIII, Pyrimethamine, Epothilone B, while showing higher resistance drugs such I-BET-762, PHA-665752, Belinostat. Conclusion This study provides comprehensive establishes reliable findings highlight relevance progression, regulation, therapy response, offering targets personalized strategies.

Language: Английский

Machine learning-based diagnostic and prognostic models for breast cancer: a new frontier on the clinical application of natural killer cell-related gene signatures in precision medicine DOI Creative Commons

Yutong Fang,

Rongji Zheng,

Yi Xiao

et al.

Frontiers in Immunology, Journal Year: 2025, Volume and Issue: 16

Published: May 27, 2025

Background Breast cancer (BC) remains a leading cause of cancer-related mortality among women worldwide. Natural killer (NK) cells play crucial role in the innate immune system and exhibit significant anti-tumor activity. However, NK cell-related genes (NRGs) BC diagnosis prognosis underexplored. With advent machine learning (ML) techniques, predictive modeling based on NRGs may offer new avenue for precision oncology. Methods We collected transcriptomic clinical data from The Cancer Genome Atlas (TCGA) Gene Expression Omnibus (GEO) databases. Differentially expressed (DEGs) were identified, key prognostic selected using univariate multivariate Cox regression analyses. constructed ML-based diagnostic models 12 algorithms evaluated their performance identifying optimal ML model. Additionally, risk model was developed LASSO-Cox regression, its validated independent cohorts. To explore potential mechanisms underlying differences between high-risk low-risk patient groups, as well drug treatment sensitivities, we conducted functional enrichment analysis, tumor microenvironment immunotherapy prediction, sensitivity mutation analysis. Results ULBP2, CCL5, PRDX1, IL21, NFATC2, CD2, VAV3 identified construction models. Among models, Random Forest (RF) demonstrated best performance, which robust distinguishing normal tissues both training validation In terms model, score effectively distinguished patients, with patients group exhibiting significantly poorer overall survival (OS) compared to those group, GEO Patients displayed increased proliferation, evasion, reduced cell infiltration, correlating lower response rates immunotherapy. Furthermore, analysis indicated that more sensitive Thapsigargin, Docetaxel, AKT inhibitor VIII, Pyrimethamine, Epothilone B, while showing higher resistance drugs such I-BET-762, PHA-665752, Belinostat. Conclusion This study provides comprehensive establishes reliable findings highlight relevance progression, regulation, therapy response, offering targets personalized strategies.

Language: Английский

Citations

0