Abstract: Background and purpose: Neoadjuvant chemotherapy (NAC) has become the standard treatment method for locally advanced breast cancer. In 2018, researchers established a multivariate predictive model for predicting the probability of axillary complete response (ypN₀) after NAC in patients with clinical axillary lymph node positive (cN₊) disease. This study aimed to evaluate the related factors of ypN₀ after NAC, and to verify and modify the Olga Kantor predictive model for breast cancer patients. Methods: A total of 350 consecutive node-positive breast cancer patients who received axillary lymph node dissection (ALND) after NAC in Shandong Cancer Hospital Affiliated to Shandong University were retrospectively analyzed. As the pathological evaluation of the primary breast tumor response in the Olga Kantor model could not be used for preoperative prediction of ypN₀, preoperative image after NAC was incorporated in our study as a surrogate to modify this model. Independent predictive factors were analyzed both in the validation model and the modification model, and the predictive accuracy was assessed by the area under receiver operating characteristic (ROC) curve (AUC) in the two models. Results: Age, molecular subtype, Ki-67 and pathological extent of breast tumor response were independent predictors of ypN₀ in the validation model, while age, molecular subtype and clinical extent of breast tumor response were independent predictors in the modification model (P<0.05, respectively). The validation and modification models achieved the AUC of 0.788 and 0.782, respectively (P>0.05). In the modification model, patients with predictive score ≤3, 4-7 and ≥8 reached ypN0 rates of 2.5% (1/40), 22.4% (51/228) and 68.3% (56/82), respectively. Conclusion: Olga Kantor model could accurately predict ypN₀ after NAC, and our modification model could reach the same predictive power but was more in line with clinical practice, which could provide a reasonable support for patient selection for sentinel lymph node biopsy (SLNB) after NAC. ALND is suggested in patients with score ≤3, SLNB is suitable for patients with score 4-7, and SLNB is recommended to patients with score ≥8.

Key words: Breast cancer, Neoadjuvant chemotherapy, Axillary lymph node pathologic complete response, Prediction, Sentinel lymph node biopsy