Multimodal diagnostic models and subtype analysis for neoadjuvant therapy in breast cancer DOI Creative Commons
Zheng Ye, Jiaqi Yuan,

Deqing Hong

et al.

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

Published: March 18, 2025

Breast cancer, a heterogeneous malignancy, comprises multiple subtypes and poses substantial threat to women's health globally. Neoadjuvant therapy (NAT), administered prior surgery, is integral breast cancer treatment strategies. It aims downsize tumors, optimize surgical outcomes, evaluate tumor responsiveness treatment. However, accurately predicting NAT efficacy remains challenging due the disease's complexity diverse responses across different molecular subtypes. In this study, we harnessed multimodal data, including proteomic, genomic, MRI imaging, clinical information, sourced from cohorts such as I-SPY2, TCGA-BRCA, GSE161529, METABRIC. Post data preprocessing, Lasso regression was utilized for feature extraction selection. Five machine learning algorithms were employed construct diagnostic models, with pathological complete response (pCR) predictive endpoint. Our results revealed that multi-omics Ridge model achieved optimal performance in pCR, an AUC of 0.917. Through unsupervised clustering using R package MOVICS nine algorithms, identified four distinct associated NAT. These exhibited significant differences proteomic profiles, hallmark gene sets, pathway activities, immune microenvironments, transcription factor characteristics. For instance, CS1 subtype, predominantly ER-positive, had low pCR rate poor chemotherapy drugs, while CS4 characterized by high infiltration, showed better immunotherapy. At single-cell level, detected heterogeneity microenvironment among Malignant cells displayed copy number variations, differentiation levels, evolutionary trajectories. Cell-cell communication analysis further highlighted differential interaction patterns subtypes, implications progression response. subtype provide novel insights into cancer. findings hold promise guiding personalized Future research should focus on experimental validation, in-depth exploration underlying mechanisms, extension these methods other cancers modalities.

Language: Английский

Opportunities for predictive proteogenomic biomarkers of drug treatment sensitivity in epithelial ovarian cancer DOI Creative Commons
Trudy Janice Philips, Britt Erickson, Stefani N. Thomas

et al.

Frontiers in Oncology, Journal Year: 2025, Volume and Issue: 14

Published: Jan. 7, 2025

Genomic analysis has played a significant role in the identification of driver mutations that are linked to disease progression and response drug treatment ovarian cancer. A prominent example is stratification epithelial cancer (EOC) patients with homologous recombination deficiency (HRD) characterized by DNA damage repair genes such as BRCA1/2 for PARP inhibitors. However, recent studies have shown some tumors respond inhibitors irrespective their HRD or BRCA mutation status. An exclusive focus on genome overlooks insight can be gained from other biological analytes, including proteins, which carry out cellular functions. Proteogenomics integration genomics, transcriptomics, epigenomics proteomics data. This review paper provides novel into proteogenomics an analytical approach identify predictive biomarkers Proteogenomic facilitate response, consequently greatly improving EOC towards goal personalized medicine.

Language: Английский

Citations

0
Multimodal diagnostic models and subtype analysis for neoadjuvant therapy in breast cancer DOI Creative Commons
Zheng Ye, Jiaqi Yuan,

Deqing Hong

et al.

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

Published: March 18, 2025

Breast cancer, a heterogeneous malignancy, comprises multiple subtypes and poses substantial threat to women's health globally. Neoadjuvant therapy (NAT), administered prior surgery, is integral breast cancer treatment strategies. It aims downsize tumors, optimize surgical outcomes, evaluate tumor responsiveness treatment. However, accurately predicting NAT efficacy remains challenging due the disease's complexity diverse responses across different molecular subtypes. In this study, we harnessed multimodal data, including proteomic, genomic, MRI imaging, clinical information, sourced from cohorts such as I-SPY2, TCGA-BRCA, GSE161529, METABRIC. Post data preprocessing, Lasso regression was utilized for feature extraction selection. Five machine learning algorithms were employed construct diagnostic models, with pathological complete response (pCR) predictive endpoint. Our results revealed that multi-omics Ridge model achieved optimal performance in pCR, an AUC of 0.917. Through unsupervised clustering using R package MOVICS nine algorithms, identified four distinct associated NAT. These exhibited significant differences proteomic profiles, hallmark gene sets, pathway activities, immune microenvironments, transcription factor characteristics. For instance, CS1 subtype, predominantly ER-positive, had low pCR rate poor chemotherapy drugs, while CS4 characterized by high infiltration, showed better immunotherapy. At single-cell level, detected heterogeneity microenvironment among Malignant cells displayed copy number variations, differentiation levels, evolutionary trajectories. Cell-cell communication analysis further highlighted differential interaction patterns subtypes, implications progression response. subtype provide novel insights into cancer. findings hold promise guiding personalized Future research should focus on experimental validation, in-depth exploration underlying mechanisms, extension these methods other cancers modalities.

Language: Английский

Citations

0