Automated quantification predicts treatment response in breast cancer, indicating potential for personalized care.
Key Points
The aim is to evaluate if automated quantification of background parenchymal enhancement can serve as a biomarker for treatment outcomes in breast cancer.
Analyzed data from 922 patients with breast cancer and validated with 152 additional patients.
Performed automated fibroglandular tissue segmentation and BPE quantification on dynamic contrast-enhanced MRI.
Established optimal enhancement thresholds and evaluated their performance using AUC.
Utilized Cox proportional hazards models to predict overall survival and recurrence-free survival.
Peak-contrast BPE correlated strongly with radiologist-defined BPE categories (AUC 0.70–0.86).
BPE decreased after neoadjuvant chemotherapy.
Reduction in BPE grade predicted pathological complete response in both high and low baseline BPE groups.
Baseline BPE was linked to improved overall survival but not to recurrence-free survival.