Early On-Treatment Ultrasound Tumor Shrinkage and HER2-Positive Residual Invasive Disease After Neoadjuvant Therapy: A Two-Center Study

Background Residual invasive disease after neoadjuvant therapy may differ in human epidermal growth factor receptor 2 (HER2) status from the pretreatment tumor. We evaluated whether early ultrasound response adds information for estimating HER2-positive residual invasive disease at surgery beyond baseline factors. Patients and Methods This retrospective two-center study included women with nonmetastatic HER2-positive breast cancer treated with neoadjuvant therapy and surgery. HER2-positive residual invasive disease was the primary endpoint; pathological complete response (pCR) and HER2-negative residual invasive disease formed the comparison group. The clinical model included nodal stage, estrogen receptor expression, pretreatment tumor size, baseline HER2 immunohistochemistry, treatment intensity, and timing of HER2-directed therapy. The dynamic model additionally included tumor shrinkage after two cycles. Performance was assessed by bootstrap correction, repeated cross-validation, and an external supportive cohort. Results Of 815 development-cohort patients, 224 had HER2-positive residual invasive disease; 26 of 83 external-cohort patients had this outcome. Median shrinkage was 40.8% for pCR, 27.8% for HER2-negative residual disease, and 18.5% for HER2-positive residual disease. Each 10-percentage-point greater shrinkage was associated with lower odds of HER2-positive residual invasive disease (odds ratio, 0.763; 95% confidence interval, 0.708–0.823). Repeated cross-validated area under the curve increased from 0.667 to 0.736; the external dynamic-model area under the curve was 0.874. Within residual disease, the increase was modest (0.634 to 0.656). Conclusion Early ultrasound tumor shrinkage provided incremental risk information, but phenotype-specific discrimination was modest. Prospective evaluation is required before model estimates guide management. MicroAbstract In 815 development-cohort and 83 external-supportive-cohort patients with HER2-positive breast cancer, early ultrasound tumor shrinkage after two neoadjuvant cycles added risk information for HER2-positive residual invasive disease beyond baseline factors. Because the primary comparator combined pCR with HER2-negative residual disease, phenotype-specific discrimination within residual disease was modest. External findings were supportive rather than definitive; the model should be considered for risk stratification rather than treatment selection pending prospective validation.