Automated calculation of background parenchymal enhancement as a biomarker of treatment responses and recurrence-free survival in breast cancer

Background. Background parenchymal enhancement (BPE)—the degree of contrast uptake in fibroglandular tissue (FGT) from dynamic-contrast-enhanced MRI —has the potential to serve as a surrogate marker of clinical outcomes in breast cancer. Methods. We implemented automated FGT segmentation and BPE calculation using DCE-MRI, as well as optimized the signal enhancement thresholds for BPE quantification and assessed its predictive value for overall survival (OS), recurrence-free survival (RFS), and pathological complete response (pCR). The multi-institutional data consisted of 922 biopsy-confirmed invasive breast cancer patients from the Duke-Breast-Cancer-MRI dataset for training and testing, and 152 patients with whole-breast pre- (T0) and/or post (T3) DCE-MRI from the I-SPY2 dataset for external validation. The optimal enhancement threshold by volume-based method was established against four radiologist-defined BPE categories. The area-under-the-curve (AUC) were obtained for classification of BPE categories. Cox proportional hazards models were used to predict OS and RFS. Logistic regression was used to predict pCR. Results. The BPE calculated from peak contrast showed strong correlation with radiologist-defined BPE, with the best performance at a 55% signal enhancement threshold. AUCs for distinguishing BPE grades ranged from 0.70 to 0.86, and performance indices were relatively invariant across 20–80% enhancement thresholds. BPE from late post-contrast MRI yielded lower performance than BPE from peak post-contrast MRI. BPE decreased post neoadjuvant chemotherapy. A reduction in BPE grade after neoadjuvant chemotherapy was predictive of pCR (adjusted odds ratio = 5.88 [1.03, 33.33] for high baseline BPE group, and adjusted odds ratio = 6.54 [1.26, 33.94] for low baseline BPE group). Baseline BPE was independently associated with improved OS (adjusted hazard ratio 0.58 [0.34, 0.99]) but not associated with RFS. Conclusions. These findings highlight the potential role of automated determination of BPE in improving risk stratification and guiding treatment decisions in breast cancer care.