Lymph node positivity in different early breast carcinoma phenotypes: a predictive model

Area of Science:

• Oncology
• Surgical Pathology
• Biostatistics

Background:

• Breast cancer (BC) molecular subtypes correlate with axillary lymph node (LN) status.
• Accurate prediction of LN positivity is essential for guiding treatment strategies.

Purpose of the Study:

• To enhance multivariable models for predicting LN metastases.
• To develop nomograms using logistic regression with clinical and pathological variables, from surgical results or biopsy data.

Main Methods:

• A retrospective cohort of 12,572 early BC patients was divided into training and validation sets.
• Multivariable logistic regression was used to build predictive nomograms for LN metastases risk.
• Model performance was evaluated using discrimination (AUC) and calibration on both sets.

Main Results:

• Pathologic and pre-operative models achieved AUCs of 0.780 and 0.717 (training) and 0.796 and 0.725 (validation), respectively.
• Age, tumor size, lymphovascular invasion (LVI), and molecular subtype were significant predictors of LN metastases.
• The models demonstrated good calibration accuracy.

Conclusions:

• Proposed nomograms effectively predict the risk of sentinel lymph node (SLN) and non-sentinel node (NSN) involvement.
• These models can inform decisions on axillary LN staging and the need for axillary lymph node dissection (ALND).
• They also assist in planning immediate breast reconstruction for patients without LN metastasis requiring radiotherapy.