Pattern of breast cancer blood flow and metabolism, assessed using dual-acquisition 18FDG PET: Correlation with tumor phenotypic features and pathological response to neoadjuvant chemotherapy

Abstract

Introduction In breast cancer, early changes in tumor glucose metabolism and blood flow (BF) have been evaluated separately and are proposed to monitor tumor response to neoadjuvant chemotherapy (NAC). This study used a single 18FDG dual-acquisitions PET exam to simultaneously assess these imaging features and to answer two questions: (i) Do tumor blood flow (BF) and tumor metabolism correlated with the same pre-therapy tumor phenotypic features? (ii) Are early changes in tumor BF and metabolism in response to NAC comparable or complementary in the ability to predict the final pathological response (pCR)? Methods 150 women with breast cancer and an indication for NAC were prospectively included. Women had a baseline PET exam with a 2-min chest-centered dynamic acquisition, started at the time of 18F-FDG injection, followed by a delayed static PET acquisition performed at 90 min. Tumor BF was calculated from the dynamic image using a validated first-pass model, and tumor glucose metabolism (SUVmax) was calculated on the delayed acquisition. This dual-PET acquisition was repeated after the first cycle of NAC to measure early changes in tumor BF (?BF) and SUVmax (?SUVmax). Results A weak correlation was found between baseline tumor SUVmax and BF ( r = 0.22; P = 0.006). A higher baseline SUVmax was associated with all biological markers of tumor aggressiveness, including the Triple Negative Breast Cancer (TNBC) subtype ( P P = 0.002). Mean?SUVmax was −44.6 ± 26.9% and varied significantly depending on the SBR grade, the overexpression of HER2+ and the lack of hormonal receptor expression ( P = 0.04, P P = 0.01, respectively). Mean?BF was −26.9 ± 54.3% and a drastic reduction was only observed in HER2-positive subtypes (−58.7 ± 30.0%), supporting the anti-angiogenic effect of Trastuzumab. Changes in tumor glucose metabolism outperformed changes in BF to predict pCR in all tumor subtypes: the AUC of?SUVmax (ROC curves analyses) were 0.82, 0.65 and 0.90 in the TNBC, HER2-positive and Luminal subtypes, respectively. Conclusion Of the two biological hallmarks of cancer evaluated in this study, the reduction in tumor metabolism was more accurate than the reduction in BF to predict pCR in the different subtypes of breast cancer.