Baseline FDG PET/CT-Defined Axillary Burden Predicts Axillary Pathologic Complete Response After Neoadjuvant Therapy in Breast Cancer.

Abstract

Aim: Predicting axillary pathologic complete response (pCR) after neoadjuvant systemic therapy (NST) may enable risk-adapted axillary management in clinically node-positive breast cancer (BC), but practical baseline PET-derived predictors are limited.

Methods: Single-center retrospective cohort of consecutive cN+, M0 patients with baseline [¹⁸F]FDG PET/CT-reported axillary nodal metastasis treated with NST and surgery (complete-case n=115). Axillary pCR was defined as no residual invasive carcinoma in axillary lymph nodes; isolated tumor cells and/or micrometastases were classified as non-pCR. Two prespecified ridge-penalized logistic models combined routine clinical variables with baseline PET-defined axillary burden as (i) hypermetabolic ALN count (Model A) or (ii) limited (1-3) versus advanced (≥4) burden group (Model B). Stratified 5-fold cross-validation generated out-of-fold (OOF) predictions; AUC, Brier score, and logistic recalibration slope/intercept were bootstrapped on OOF predictions (2000 resamples).

Results: Axillary pCR occurred in 59/115 (51.3%). Discrimination was good: AUC 0.80 (95% CI 0.72-0.88) for Model A and 0.79 (95% CI 0.71-0.87) for Model B. Calibration was acceptable (bootstrap-mean slope/intercept ≈1.10/0.04 and 1.08/0.03, respectively). Greater baseline PET-defined axillary extent was associated with lower odds of pCR; molecular subtype and neoadjuvant anti-HER2 therapy remained strongly associated with response.

Conclusion: Prespecified ridge models combining routine clinical variables with baseline [¹⁸F]FDG PET/CT-defined axillary burden showed good internal performance for predicting axillary pCR. External validation is required before response-adapted clinical use.