2729 - A Simplified Predictive Model for Residual Axillary Disease on Completion Dissection After Neoadjuvant Chemotherapy with Positive Sentinel Node Biopsy Using Clinical Nodal Stage and Nodal Residual Cancer Burden

Purpose/Objective(s):

Optimal management of the axilla in breast cancer patients with positive sentinel nodes after neoadjuvant chemotherapy (NAC) remains uncertain, pending results of the Alliance A011202 trial. Quantitative and clinical factors may improve risk stratification. We sought to develop a multivariate model incorporating clinicopathologic variables to predict residual axillary disease on completion lymph node dissection (cLND).

Materials/Methods:

Retrospective analysis was performed on 66 patients with positive sentinel lymph node biopsy (SLNB) following NAC who underwent cLND (2014–2022) at a single institution. Candidate predictors included race, age, tumor laterality and quadrant, clinical (cT, cN) and posttreatment (ypT) stage, receptor phenotype, TAD, and the nodal component of the MD Anderson Residual Cancer Burden index (N-RCB). Logistic regression with L2-penalization and 5-fold cross-validation was used to identify key predictors of residual nodal disease.

Results:

The full 10 variable model (AUC = 0.68) identified clinical nodal stage (cN) and N-RCB (range 0.7 – 2.0) as the most predictive variables. A simplified 2-factor model using only cN and N-RCB achieved higher discrimination (cross-validated AUC = 0.76 ± 0.10). Using the Youden-optimal N-RCB threshold (1.6), risk stratification based on these two variables showed stepwise increases in residual nodal disease: low-risk (low RCB and cN0) predicted ~20 % (vs actual 0%), intermediate-risk (either high RCB or cN+) predicted ~40 % (vs actual 38%), and high-risk (high RCB and cN+) predicted 70 % (vs actual 85%). Using clinically more conservative threshold (1.2) increased sensitivity and lowered the predicted residual risk to ~10%, ~25%, and ~55% respectively, though with fewer number of candidate patients in the lower risk groups.

Conclusion:

A simple 2-factor model using initial clinical nodal status and quantitative sentinel nodal residual cancer burden from pathology accurately estimates further residual axillary disease after NAC and positive sentinel node biopsy. This model outperforms higher-order multivariate approaches and provides a practical framework for discussion of selective omission of completion dissection in appropriately low-risk patients. External validation in larger datasets is warranted before clinical adoption.