This study develops a preoperative prediction model for negative sentinel lymph node status (pN0) in primary breast cancer patients, combining radiological and clinical variables. The key findings include:
Overview:
- A clinical preoperative model was developed using only preoperatively available data to predict pN0 status in breast cancer patients
- Radiological tumor size emerged as the strongest predictor of pN0, potentially replacing pathological tumor size in preoperative models
- The model could support omitting sentinel lymph node biopsy (SLNB) in 21% of patients if a 10% false negative rate is accepted
Key details:
- The clinical preoperative model included radiological tumor size, ER status, age, mode of detection, histological type, and tumor localization
- It achieved an AUC of 0.68 in internal validation and 0.64 in external validation
- Mammographic features from AI analysis were associated with pN0 but did not improve overall model performance when added
- A nomogram was developed to visualize the model for potential clinical use
Significance:
This preoperative prediction model could support more personalized decision-making about axillary surgery in breast cancer patients. By potentially identifying patients who can safely avoid SLNB, it may help reduce unnecessary procedures and associated morbidity. The model’s use of only preoperatively available data makes it more clinically applicable than previous models relying on postoperative information. However, prospective validation is still needed before clinical implementation.
