Related Experiment Video
Updated: May 14, 2026
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
Assessment of an existing and modified model for predicting non sentinel lymph node metastasis in breast cancer
M Al-Masri1, G Darwazeh, M El-Ghanem
1Chairman, Department of Surgery, King Hussein Cancer Center, Amman, Jordan. Box 1269, Amman 11941, Jordan, Fax + (962 6) 535-3001;
Summary:
The Memorial Sloan Kettering Cancer Center (MSKCC) breast nomogram has been validated in different populations. In this study, the nomogram was validated for the first time in a Middle East population sample. Although our sample was found to have significant differences from the dataset from which the model was derived, the nomogram proved to be accurate in predicting non sentinel axillary lymph node metastasis. An attempt to use the proportions of involved sentinel lymph nodes instead of absolute numbers of positive and negative sentinel lymph nodes, yet using the same online calculator to predict the probability of non sentinel axillary lymph node metastasis, improved the accuracy, specificity, negative predictive value, and false negative rate.
Background:
Axillary clearance is the standard of care in patients with invasive breast cancer and positive sentinel lymph node biopsy. However, in 40-60% of patients, the sentinel lymph nodes are the only involved lymph nodes in the axilla. The Memorial Sloan Kettering Cancer Center (MSKCC) breast nomogram serves to identify a subgroup of patients with low risk of non sentinel lymph node (NSLN) metastasis, in whom axillary lymph node dissection (ALND) could be spared, and thereby, preventing the unwarranted associated morbidity.
Methods:
The MSKCC nomogram was applied on 91 patients who met the criteria. A modified predictive model was developed by substituting proportions of positive and negative SLN for their absolute numbers. The accuracy was assessed by calculating the area under the receiver-operator characteristic (ROC) curve.
Results:
The MSKCC nomogram achieved an area under the ROC curve of 0.76. The area under the curve for the modified predictive model was 0.81. The specificity, negative predictive value, and false negative were 30%, 71%, 20% (MSKCC model) and 55%, 84%, 17% (modified model) at 20% predicted probability cut-off values.
Conclusion:
Although differences existed in characteristics of our breast cancer population, and in the methods of sentinel lymph node metastasis detection, the MSKCC model proved to be accurate. An attempt to replace the number of positive and negative SLNs with proportions in the MSKCC model raised the accuracy but did not achieve statistical significance (p = 0.09).
Keywords:
Breast cancer, Sentinel lymph node, Non sentinel lymph node, Axillary clearance, Predictive model.
