Comparison between 18F-FDG PET image-derived indices for early prediction of response to neoadjuvant chemotherapy in breast cancer.

UNLABELLED The goal of this study was to determine the best predictive factor among image-derived parameters extracted from sequential (18)F-FDG PET scans for early tumor response prediction after 2 cycles of neoadjuvant chemotherapy in breast cancer. METHODS 51 breast cancer patients were included. Responder and nonresponder status was determined by histopathologic examination according to the tumor and node Sataloff scale. PET indices (maximum and mean standardized uptake value [SUV], metabolically active tumor volume, and total lesion glycolysis [TLG]), at baseline and their variation (Δ) after 2 cycles of neoadjuvant chemotherapy were extracted from the PET images. Their predictive value was investigated using Mann-Whitney U tests and receiver-operating-characteristic analysis. Subgroup analysis was also performed by considering estrogen receptor (ER)-positive/human epidermal growth factor receptor 2 (HER2)-negative, triple-negative, and HER2-positive tumors separately. The impact of partial-volume correction was also investigated using an iterative deconvolution algorithm. RESULTS There were 24 pathologic nonresponders and 27 responders. None of the baseline PET parameters was correlated with response. After 2 neoadjuvant chemotherapy cycles, the reduction of each parameter was significantly associated with response, the best prediction of response being obtained with ΔTLG (96% sensitivity, 92% specificity, and 94% accuracy), which had a significantly higher area under the curve (0.91 vs. 0.82, P = 0.01) than did ΔSUVmax (63% sensitivity, 92% specificity, and 77% accuracy). Subgroup analysis confirmed a significantly higher accuracy for ΔTLG than ΔSUV for ER-positive/HER-negative but not for triple-negative and HER2-positive tumors. Partial-volume correction had no impact on the predictive value of any of the PET image-derived parameters despite significant changes in their absolute values. CONCLUSION Our results suggest that the reduction after 2 neoadjuvant chemotherapy cycles of the metabolically active volume of primary tumor measurements such as ΔTLG predicts histopathologic tumor response with higher accuracy than does ΔSUV measurements, especially for ER-positive/HER2-negative breast cancer. These results should be confirmed in a larger group of patients as they may potentially increase the clinical value and efficiency of (18)F-FDG PET for early prediction of response to neoadjuvant chemotherapy.