A novel model for predicting the risk of non-sentinel lymph node metastasis after positive sentinel lymph node biopsy in Chinese women diagnosed with early breast cancer

Background: 30 to 70% of patients with positive sentinel lymph nodes (SLNs) in early breast cancer do not develop non-SLN metastases. They are exposed to the potential complications and sequelae of axillary lymph node dissection (ALND) without gaining additional therapeutic benefit. Therefore, a prediction model for non-SLN metastasis for Chinese breast cancer patients is needed. Methods: We enrolled 1717 patients with early breast cancer who underwent SLN biopsy, and 481 of these patients underwent further ALND. An additional 113 patients served as a validation cohort. A new predictive model was established using univariate and multivariate Logit regression. The Memorial Sloan Kettering Cancer Center (MSKCC) and Shanghai Cancer Hospital (SCH) models were used for comparison with our new model. Results: Multivariate regression analysis showed that tumor size, multifocality, lymphovascular invasion, extracapsular extension, number of negative SLNs, number of positive SLNs, size of the SLN metastasis, and metastatic SLN locationwere independent indicators for non-SLN metastasis. The nomogram established based on these eight variables was well applied in the training cohort (AUC: 0.830) and validation cohort (AUC: 0.785). Moreover, the diagnostic value of our model was superior to that of the MSKCC and SCH models (both P = 0.000). Decision curve analysis showed that the net benefit of our model surpasses that of both the MSKCC and SCH models for the same risk threshold, resulting in greater benefits for patients. With a guaranteed false-negative rate, our model could accurately predict up to 24.5% of patients suitable for exemption from ALND. Meanwhile, our model evaluated the non-SLN status of patients with 3 or more positive SLNs (AUC: 0.843). Conclusions: We developed a new model to predict non-SLN metastatic status in Chinese patients with early SLN-positive breast cancer. Our model showed good performance in both cohorts and significantly outperforms the MSKCC and SCH models.