Abstract
Background/aim
Human epidermal growth factor receptor 2 (HER2)-low breast cancer shows consistently low pathological complete response (pCR) rates after neoadjuvant chemotherapy (NAC), yet no preoperative tool exists to identify non-responders for whom alternative strategies - including upfront surgery or early trastuzumab deruxtecan (T-DXd) - may be considered. We investigated whether a three-factor model using routinely available pretreatment data could identify this chemoresistant subgroup.Patients and methods
We retrospectively analyzed 124 patients who underwent surgery after neoadjuvant chemotherapy (2020-2024). Multivariable logistic regression was performed in the overall cohort and HER2-low subgroup. A three-factor scoring model (non-mass enhancement, histologic grade 1-2, and clinical T stage ≥3; 1 point each; total 0-3) was evaluated by receiver operating characteristic curve analysis.Results
HER2-low tumors showed low pCR rates regardless of hormone receptor (HR) status (HR+: 10.0%; HR-: 10.3%). In multivariable analysis, HER2-low status [odds ratio (OR)=0.176, p=0.002], histologic grade (OR=2.55, p=0.046), and clinical T stage (OR=0.38, p=0.017) were independently associated with pCR; among HER2-low patients, histologic grade was the only independent predictor (OR=8.40, p=0.023). The scoring model yielded pCR rates of 36.4%, 7.7%, 0%, and 0% for scores 0-3 [p for trend=0.005; area under the curve (AUC)=0.836, 95% confidence interval (CI)=0.701-0.953]. No pCR occurred in the 22 patients with a score ≥2 [positive predictive value (PPV)=100%, 95%CI=84.6-100%].Conclusion
A three-factor scoring model identified a HER2-low subgroup (score ≥2, n=22) in whom no patient achieved pCR (PPV=100%, 95%CI=84.6-100%). If validated externally, this score may inform preoperative decisions regarding upfront surgery or early-line trastuzumab deruxtecan. These findings are hypothesis-generating and require prospective multicenter validation.