Predictive Value of Routine Imaging for the Diagnosis of Pathologic Complete Response After Neoadjuvant Chemotherapy in Breast Cancer.

BACKGROUND: After neoadjuvant chemotherapy (NACT) of breast cancer, an accurate response assessment is essential for the de-escalation of surgical therapy. This study evaluated whether breast imaging procedures can reliably predict a pathologic complete response (pCR).

METHODS: This retrospective, monocenter study included 388 cases of invasive breast cancer treated by NACT. Most of the patients received mammography (MG) and ultrasound (US). In some cases, magnetic resonance imaging (MRI) was performed for special indications. The absence of residual tumor on imaging was defined as clinical complete response (cCR). Minimal signs of residual tumor were categorized as near-cCR. The primary endpoints were negative predictive values (NPVs) and false-negative rates (FNRs) of all imaging methods, combined and separately, for the whole cohort and subgroups of triple-negative (TN), human epidermal growth factor 2-positive (HER2+), and hormone receptor-positive (HR+)/HER2- breast cancer.

RESULTS: The overall pCR rate was 40.7 %, with most pCR cases having triple-negative breast cancer (TNBC) (n = 85, 57.8 %). The highest NPV (71.6 %) and the lowest FNR (11.0 %) in the overall cohort were achieved by the combination of ultrasound and mammography. In the TNBC subgroup, NPVs ranged from 78.1 % to 83.7 %, compared with lower NPVs in the HR+ subgroup (34.2-50.0 %). If near-cCR cases were defined together with cCR as a "negative" test result, the predictive accuracy of imaging decreased in the whole cohort (FNR, 34.8 %; NPV, 56.1 %).

CONCLUSION: Standard imaging is insufficient to predict pCR reliably in an unselected cohort. Strict patient selection and additional diagnostics are necessary for safe confirmation of pCR.