Predictive value of [18F]FDG PET for pathological response of breast cancer to neo-adjuvant chemotherapy

Abstract

The aim of this prospective study was to evaluate the predictive value of [18F]fluorodeoxyglucose positron emission tomography (FDG PET) for the pathological response of breast cancer after completion of neo-adjuvant chemotherapy. Fifty patients with newly diagnosed, non-inflammatory, large or locally advanced breast cancer undergoing neo-adjuvant chemotherapy were eligible for this study. Clinical assessment was accomplished by comparing initial tumor size with preoperative tumor size. Pathological responses were classified into three groups: pathological non-response (pNR), pathological partial response (pPR) and pathological complete response (pCR). To determine the effect of reduction rate (RR) of peak standardized uptake values for tumor responses, logistic regression analyses were performed. To identify an optimal threshold value of RR for the prediction of pathological response, receiver operating characteristic analysis was performed. Eight per cent (four of 50) of the patients had pCR and 46% had pPR. Ten per cent of patients had clinical CR and 52% had clinical PR. In clinical response, the RRs (+/- SD) of CR (-83.4 +/- 12), PR (-81.8 +/- 22.7) and NR (-79.7 +/- 31.9) showed no statistical differences (P > 0.05). However, for pathological responses, the RR of CR (-96.5 +/- 3.4) had a lower value than those of PR (-87.9 +/- 15.1) and NR (-56.2 +/- 29.6) (P = 0.0006; CR versus PR, P < 0.05; CR versus NR, P < 0.05; PR versus NR, P < 0.01). When -88% of RR was used as threshold value for differentiation between pCR and pPR, the area under the curve (AUC) was 0.788 [standard error (SE) 0.106; 95% confidence interval (CI) 0.589-0.920]. The sensitivity and specificity were 100% and 56.5%, respectively. When -79% of RR was used as threshold value for differentiation between pathological responders and non-responders, the AUC was 0.838 (SE 0.059; 95% CI 0.707-0.927). The sensitivity and specificity were 85.2% and 82.6%, respectively. Despite some limitations, this study suggests a possible predictive value of FDG PET for the assessment of the pathological response of primary breast cancer after neo-adjuvant chemotherapy. However, these findings deserve further investigation on a larger number of patients, and more frequent and earlier PET scans in each patient need to be performed to allow a better validation of the differentiation between the responder and non-responder groups.