ABSTRACT
BACKGROUND: Magnetic resonance imaging (MRI) accuracy after neoadjuvant systemic therapy (NST) for breast cancer varies according to hormone receptor (HR), human epidermal growth factor receptor type-2 (HER2) subtype and Ki-67 proliferation index. Whether MRI accuracy varies by genomic signatures is unknown. We examined the accuracy of MRI in the NEONAB trial (Clinicaltrials.gov #: NCT01830244). AIM: To examine the accuracy of MRI to predict pathological response to neoadjuvant therapy for breast cancer in the NEONAB trial. METHODS: Patients with stages II-III breast cancer received sequential epirubicin, cyclophosphamide and nab-paclitaxel and trastuzumab if they were HER2+. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated to assess the utility of preoperative MRI to predict pathological complete response (pCR). Bland-Altman plots were used to assess agreement between MRI and pathological assessment of residual disease. RESULTS: MRI correctly predicted pCR in 64.1% of the cohort. Sensitivity and specificity were 52% and 78%, respectively; PPV 73% and NPV 58%. MRI predicted pCR most accurately in HER2-positive patients; sensitivity 58%, specificity 100%, PPV 100% and NPV 38%. MRI had higher PPV and NPV in tumours with Ki-67 ≥ 15% than tumours with Ki-67 < 15%, 75% versus 50% and 57.5% versus 50%, respectively. In this study, MRI underestimated residual tumour size by 1.65 mm (limits of agreement: 43.07-39.77 mm). CONCLUSIONS: MRI appears more accurate for predicting pCR in HER2+ disease than other subtypes and in cancers with Ki-67 ≥ 15% compared to those with Ki-67 < 15%. Accuracy of MRI in our HR+, RS ≥ 25 cohort is comparable to previous reports of unselected HR+ disease. MRI post-NST should be interpreted in conjunction with HER2 status and Ki-67 index of the primary.
