Prognostic value of ultrasound Breast Imaging Reporting and Data System for disease-free survival in early-stage breast cancer: a retrospective study (2011-2018).

Background: The Breast Imaging Reporting and Data System (BI-RADS) offers a standardized diagnostic risk stratification framework for assessing breast lesions. However, whether the imaging phenotype captured by the preoperative ultrasound (US) BI-RADS category is associated with long-term survival in early-stage breast cancer (BC) remains unclear. The aim of this study was to evaluate the prognostic value of the preoperative US BI-RADS category and to develop a nomogram integrating this imaging marker with clinicopathological factors to predict disease-free survival (DFS). Methods: We retrospectively analyzed 681 patients with early-stage BC (2011-2018). Survival outcomes across different US BI-RADS categories were compared using Kaplan-Meier curves and log-rank tests. Associations between candidate variables and survival outcomes were evaluated using univariate and multivariate Cox regression analyses. We developed a US-based nomogram with US BI-RADS category, lymph node metastasis (LNM), and Ki-67 to predict 5- and 10-year DFS. We also built a clinical model based on standard clinicopathological factors for comparison. We assessed both models using bootstrap-corrected concordance index (C-index) and area under the receiver operating characteristic (ROC) curve (AUC), and constructed calibration curves to compare predicted with observed DFS. Results: Patients with US BI-RADS categories 3-4B showed significantly superior DFS compared to those with categories 4C-5 (P=0.004). Multivariate Cox regression analysis identified US BI-RADS categories 4C-5 and Ki-67 index ≥20% as independent predictors of less favorable DFS. The US-based nomogram demonstrated good predictive performance, with a C-index of 0.701 [95% confidence interval (CI): 0.611-0.770] and well-fitted calibration curves. The 5- and 10-year time-dependent AUCs were 0.720 (95% CI: 0.624-0.801) and 0.647 (95% CI: 0.549-0.738), respectively. Compared with the clinical model, the US-based nomogram showed improved discrimination. Conclusions: We identified the preoperative US BI-RADS category as a promising prognostic imaging phenotype marker for early-stage BC. Based on this marker, we developed a nomogram that outperformed the clinical model based on traditional prognostic factors. However, these preliminary findings require validation in broader populations prior to clinical application.