A multi-dimensional nomogram to predict non-sentinel lymph node metastases in T1-2HR+ breast cancer

Explainable machine learning versus known nomogram for predicting non-sentinel lymph node metastases in breast cancer patients: A comparative study

INTRODUCTION

Axillary lymph node dissection (ALND) is the standard of care for breast cancer patients with positive sentinel lymph nodes (SLN), which are the first lymph nodes that drain the breast. However, many patients with positive SLNs may not have additional positive nodes, making the prediction of non-sentinel lymph node (NSLN) metastasis challenging. Reliable prognostic tools are essential for accurately assessing NSLN metastasis. The Memorial Sloan Kettering Cancer Center (MSKCC) nomogram has demonstrated effectiveness in this context, but it requires further evaluation within the Iranian breast cancer population. While ALND remains the gold standard, its unnecessary application in patients without evidence of additional positive nodes raises concerns due to potential complications such as lymphedema, nerve injury, and shoulder joint dysfunction. Furthermore, integrating Artificial Intelligence (AI) and Machine Learning (ML) techniques presents an opportunity to enhance the precision of NSLN metastasis predictions.

METHOD

This study conducts an extensive comparative analysis between the MSKCC nomogram and various ML models to predict NSLN metastasis, utilizing a dataset of Iranian breast cancer patients. Employing eXplainable Artificial Intelligence (XAI) methodologies, we analyzed 16 clinical features across a cohort of 183 patients. Our methodology includes rigorous statistical evaluations and the training and validation of ML models to assess the precision and robustness of these models compared to the MSKCC nomogram.

RESULTS

Our analysis revealed that the Random Forest (RF) model outperformed the MSKCC nomogram, achieving an accuracy of 72.2 % and an AUC of 0.77, compared to the nomogram's AUC of 0.73. Logistic Regression (LR) also demonstrated competitive performance with an accuracy of 65 % and an AUC of 0.73. The RF model exhibited high sensitivity (75 %) and precision (73 %), effectively identifying critical predictors of NSLN metastasis, including the presence of ductal carcinoma in situ (DCIS) and tumor characteristics such as type and grade. Explainable AI techniques, particularly SHAP values, provided insights into feature importance, enhancing model interpretability.

CONCLUSION

Our study offers a comprehensive comparison between ML models and the MSKCC nomogram for predicting NSLN metastasis among Iranian breast cancer patients. These findings contribute valuable insights to the discourse on personalized treatment approaches, emphasizing the need for tailored prognostic tools across diverse populations. The implications of this research extend to clinical decision-making, potentially improving the accuracy and efficacy of breast cancer management within the Iranian healthcare landscape.

The prevalence of non-sentinel lymph node metastasis among breast cancer patients with sentinel lymph node involvement and its impact on clinical decision-making: a single-centred retrospective study

Background

Sentinel lymph node biopsy (SLNB) has become standard procedure for early breast cancer patients with clinically node negative disease. The patients with SLN metastasis normally underwent axillary lymph node dissection (ALND). However, the metastatic status of non-sentinel Lymph nodes (non-SLNs) varied significantly in different reports. Here, we evaluated the prevalence of non-SLNs metastasis among breast cancer patients with sentinel lymph node metastasis and its impact on clinical decision-making.

Materials and Methods

We identified 892 female patients with operable cT1-3N0 invasive breast cancer who underwent ALND in our center due to SLN metastasis from 2017 to 2023, retrospectively. The prevalence of non-SLN metastasis among different clinicopathological traits and its correlation with the number of positive SLNs were analyzed. The optimal clinical decision-making was generalized.

Results

The median number of SLN+, SLN, non-SLN+ and non-SLN was 2, 4, 1 and 14 among the enrolled 892 female patients, respectively. 504 (56.50%) patients with SLN + had at least one metastatic lymph node in the harvested non-SLNs. Among the enrolled 892 female patients, 435 (48.77%) patients with 1 positive SLN, of which 180 (41.38%) had at least one additional metastatic non-SLNs. 242 (27.13%) patients with 2 positive SLNs, of which 146 (60.33%) had at least one metastatic non-SLNs. For the rest 215 (24.10%) patients with at least 3 metastatic SLNs, 178 (82.79%) had at least one metastatic non-SLNs. In the univariate analysis, the non-SLNs metastatic status was correlated with the number of SLNs+, tumor size, tumor grade, lymphovascular invasion (LVI) and molecular subtypes, but not histopathologic type. In the multivariate analysis, the risk of additional non-SLNs metastasis correlated with the number of SLNs+, SLNs, non-SLNs and LVI.

Conclusion

Omiting ALND in patients with higher non-SLNs + rate outside the American College of Surgeons Oncology Group (ACSOG) Z0011 and the European Organization for Research and Treatment of Cancer (EORTC) 10,981-22023 AMAROS criteria should be considered with caution in clinical decision-making. To evaluate whether axillary radiotherapy and ALND provides equivalent regional control in breast cancer patients with obvious residual metastatic lymph nodes undesected in the axilla, a well-matched prospective randomized controlled trial is an urgent need.

A new prediction nomogram of non-sentinel lymph node metastasis in cT1-2 breast cancer patients with positive sentinel lymph nodes

We aimed to analyze the risk factors and construct a new nomogram to predict non-sentinel lymph node (NSLN) metastasis for cT1-2 breast cancer patients with positivity after sentinel lymph node biopsy (SLNB). A total of 830 breast cancer patients who underwent surgery between 2016 and 2021 at multi-center were included in the retrospective analysis. Patients were divided into training (n = 410), internal validation (n = 298), and external validation cohorts (n = 122) based on periods and centers. A nomogram-based prediction model for the risk of NSLN metastasis was constructed by incorporating independent predictors of NSLN metastasis identified through univariate and multivariate logistic regression analyses in the training cohort and then validated by validation cohorts. The multivariate logistic regression analysis revealed that the number of positive sentinel lymph nodes (SLNs) (P < 0.001), the proportion of positive SLNs (P = 0.029), lymph-vascular invasion (P = 0.029), perineural invasion (P = 0.023), and estrogen receptor (ER) status (P = 0.034) were independent risk factors for NSLN metastasis. The area under the receiver operating characteristics curve (AUC) value of this model was 0.730 (95% CI 0.676-0.785) for the training, 0.701 (95% CI 0.630-0.773) for internal validation, and 0.813 (95% CI 0.734-0.891) for external validation cohorts. Decision curve analysis also showed that the model could be effectively applied in clinical practice. The proposed nomogram estimated the likelihood of positive NSLNs and assisted the surgeon in deciding whether to perform further axillary lymph node dissection (ALND) and avoid non-essential ALND as well as postoperative complications.