Assessing residual cancer cells using MRI and US after preoperative chemotherapy in primary breast cancer to omit surgery

Background Enhanced magnetic resonance imaging (MRI) and ultrasonography (US) are used to assess residual lesions after preoperative chemotherapy before surgery. However, residual lesion assessments based on preoperative imaging often differ from postoperative pathologic diagnoses. We retrospectively reviewed the accuracy of preoperative residual lesion assessments, including ductal carcinoma in situ (DCIS) cases to find criteria for cases in which surgery can be omitted. Methods We reviewed 201 patients who received preoperative chemotherapy and surgery in our hospital from January 2013 to November 2016. Presurgical evaluations regarding the possible existence of residual lesions, and clinical Complete Response (cCR) or non-cCR, were compared with postoperative pathological diagnoses. Results Of the 201 patients, 52 were diagnosed with cCR, and 39 with pathological complete response (pCR). Predictions for residual lesions were 86.4% sensitive, 76.9% specific, and 84.6% accurate. When patients were divided into 4 groups by estrogen receptor (ER) and HER2 status, sensitivity in each group was ER + /HER2 − : 91.4%; ER − /HER2 − : 94.1%; ER + /HER2 + : 78.6%; and ER − /HER2 + : 78.5%. Of the 22 patients preoperatively assessed with cCR, but diagnosed with non-pCR, the median invasive residual tumor size was 2 mm (range 0–46 mm); 5 patients (22.7%) had only DCIS. Conclusions Predicting residual lesions after preoperative chemotherapy by using MRI and US is a reasonable strategy. However, current methods are inadequate for identifying patients who can omit surgery; therefore, a new strategy for detecting small tumors in these patients is needed.