The aim of this study was to establish an magnetic resonance imaging (MRI)-based interpretation model to facilitate the selection of breast-conserving surgery (BCS) after neoadjuvant chemotherapy (NAC).
Although MRI is the most reliable method to assess tumor size after NAC, criteria for the correct selection of surgery remain unclear.
In 208 patients, dynamic contrast-enhanced MRI was performed before and after NAC. Imaging was correlated with pathology. Differences <20 mm in tumor extent were considered to accurately indicate disease extent. Multivariate analysis with cross-validation was performed to analyze features affecting the potential of MRI to correctly indicate BCS (ie, residual tumor size <30 mm on pathology).
The accuracy of MRI to detect residual disease was 76% (158/208). The positive and negative predictive value of MRI were 90% (130/144) and 44% (28/64), respectively. In 35 patients (17%), MRI underestimated the tumor size by >20 mm and in 27 patients (13%) this would have lead to incorrect indication of BCS. The features most predictive of indicating feasibility of BCS in tumors <30 mm on preoperative MRI were the largest diameter at the baseline MRI, the reduction in diameter and the tumor subtype based on hormone-, and human epidermal growth factor receptor 2-status (area under the curve: 0.78).
Optimal selection of patients for BCS after NAC based on MRI should take into account (1) the tumor size at baseline (2) the reduction in tumor size, and (3) the subtype based on hormone-, and human epidermal growth factor receptor 2-status.
After neoadjuvant chemotherapy, MRI underestimates the residual tumor size in 17% (35/208) of breast cancer patients, especially when a complete remission is anticipated. We created a model based on 3 features that helps in interpreting preoperative MRIs and guides surgeons in choosing a suitable surgical treatment.
From the Departments of *Surgical Oncology, †Radiology, and ‡Pathology, The Netherlands Cancer Institute, Antoni van Leeuwenhoek Hospital, Amsterdam, The Netherlands.
The authors did not receive any financial support.
Presented at the San Antonio Breast Cancer Symposium 2008, San Antonio, TX.
The authors have no conflicts of interest to declare.
Reprints: Kenneth G. A. Gilhuijs, PhD, Department of Radiology, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands. E-mail: firstname.lastname@example.org.