The assessment of breast cancer response to neoadjuvant chemotherapy: comparison of magnetic resonance imaging and 18F-fluorodeoxyglucose positron emission tomography

Abstract

Neoadjuvant chemotherapy for locally advanced breast cancer is a widely accepted treatment. For assessment of the tumor response after chemotherapy, both magnetic resonance imaging (MRI) and (18)F-fluorodeoxyglucose positron emission tomography (PET) are promising methods. To retrospectively compare MRI and PET in the assessment of tumor response to neoadjuvant chemotherapy for primary breast cancer with the pathologic response as the reference standard. Between August 2006 and May 2008, 32 women with breast cancer underwent concurrent MRI and PET before and after neoadjuvant chemotherapy. For response assessment, we calculated the changes in the maximum diameters of the tumor (ΔD(max)) on MRI, and the changes in the standard uptake values (ΔSUV) on PET. The correlation between the ΔD(max) and ΔSUV was analyzed using Pearson's correlation coefficient. The correspondence rates between each imaging modality and pathologic assessment were calculated. For prediction of the pathologic complete response (pCR), the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were analyzed using the McNemar test. The pathologic assessment of tumor response to neoadjuvant chemotherapy identified eight complete responses (25.0%), 10 partial responses (31.2%), and 14 non-responses (43.8%). The change in size on MRI was moderately correlated with the change in SUV on PET (r=0.574, p=0.001). The correspondence rate of response assessment was 75.0% (24/32) between MRI and pathologic response and 53.1% (17/32) between PET and pathologic response. For the pCR, specificity (95.8% vs. 62.5%) and PPV (83.3% vs. 47.1%) were statistically higher on MRI than PET (p < 0.05), while sensitivity (100.0% vs. 62.5%) and NPV (100.0% vs. 88.5%) on PET tended to be higher than MRI. Before and after neoadjuvant chemotherapy for breast cancer, the ΔD(max) of MRI correlated moderately with the ΔSUV on PET. For prediction of the pCR, MRI proved to be a more specific modality than PET.