A preoperative predictive model for margin status in breast-conserving surgery

Abstract

Background and purpose

Positive margins after breast-conserving surgery (BCS) not only frequently necessitate re-excision but also represent the most significant risk factor for local recurrence. This study aimed to identify preoperative predictors of positive margins in BCS and establish a predictive model.

Materials and methods

A retrospective analysis was conducted on 2837 patients with primary breast cancer (BC) who underwent BCS at Tianjin Medical University Cancer Institute & Hospital between June 2014 and June 2024. All patients underwent preoperative imaging evaluations, including ultrasonography (US), mammography (MG), and magnetic resonance imaging (MRI). Patients were randomly divided into a training cohort (n = 1,986, 70 %) and a validation cohort (n = 851, 30 %). A nomogram was developed in the training cohort using univariate and multivariate logistic regression to identify significant clinicopathological and imaging predictors. Discrimination was evaluated by calculating the C-index, while the Hosmer-Lemeshow goodness-of-fit test was applied to validate calibration performance.

Results

The positive margin rate in our cohort was 18.6 %. The predictive model incorporated seven variables: histological type; MRI parameters including maximum lesion size, fibroglandular tissue (FGT), background parenchymal enhancement (BPE), non-mass enhancement (NME), multifocality, and axillary lymph node metastasis (ALNM). C-indices were calculated of 0.782 (95 % CI: 0.757–0.807) and 0.761 (95 % CI: 0.719–0.803) for the modeling and the validation group, respectively. Hosmer-Lemeshow test:X-squared = 3.3163, df = 3, p-value = 0.3454.

Conclusion

We developed and validated a preoperative nomogram for predicting the risk of positive margins in BCS, integrating key clinicopathological and imaging parameters.