RT Journal Article
SR Electronic
T1 Evaluation of primary breast tumors using dedicated breast PET and whole body PET
JF Journal of Nuclear Medicine
JO J Nucl Med
FD Society of Nuclear Medicine
SP 610
OP 610
VO 61
IS supplement 1
A1 Ella Jones
A1 Deep Hathi
A1 Wen Li
A1 Youngho Seo
A1 Robert Flavell
A1 Benjamin Franc
A1 Bonnie Joe
A1 Laura Esserman
A1 Nola Hylton
YR 2020
UL http://jnm.snmjournals.org/content/61/supplement_1/610.abstract
AB 610Objectives: Metabolic imaging of the primary breast tumor with 18F-fluorodeoxyglucose (FDG) whole-body PET (wbPET) may provide predictive information for stratifying treatment response in the neoadjuvant chemotherapy (NAC) setting. However, wbPET is limited in quantitative and precise imaging of primary breast tumors due to significant partial volume effect caused by limited spatial resolution. Dedicated breast PET (dbPET) is a high-resolution modality that has demonstrated ability in highlighting intra-tumor heterogeneity and identifying small lesions in the breast volume. In this study, we characterized the uptake of 18F-FDG in both dbPET and wbPET to highlight the similarities and differences in distribution of PET signal in a cohort of breast cancer patients prior to treatment. Methods: 13 patients with biopsy confirmed, locally advanced breast cancer were enrolled in this study and received bilateral dbPET (MAMMI, General Equipment and Medical Imaging SA (OncoVision), Valencia, Spain) following administration of 186 MBq 18F-FDG. A subset of 11 patients also received 307 MBq 18F-FDG wbPET on a separate day. Tumor volumes were segmented in the dbPET and wbPET images and converted to lean-body mass corrected standardized uptake values (SULs). Uptake metrics (max, mean, peak), tumor-background ratios, metabolic tumor volume, and total lesion glycolysis were compared between both modalities. 19 radiomic features based on morphology, tumor intensity, and texture were also calculated on dbPET and wbPET tumor volumes. To prevent overfitting and to minimize number of redundant radiomic features, feature selection was performed with the least absolute shrinkage and selection operator (LASSO) regression model with three-fold cross-validation. Results: 18F-FDG uptake metrics were significantly higher for dbPET compared to wbPET. DbPET exhibited a 4-fold increase in SULpeak (mean ± SD) (dbPET: 7.70 ± 3.36 g/mL; wbPET: 2.03 ± 1.25 g/mL; p < 0.001) and a 1.5-fold increase in tumor-background ratios relative to wbPET using the contralateral breast as background. Following scaling to unit mean and variance, radiomic features were fit using the LASSO regression model to identify the highest weighted features differentiating dbPET and wbPET (r2 = 0.92). Of the 19 radiomic features, the three highest weighted features were the neighborhood grey-tone difference matrix strength (p = 0.10) and correlation (p = 3.1e-5) and normalized inverse difference (p = 0.002) calculated from the grey-level co-occurrence matrix. Finally, tumor intensity features reflecting entropy and uniformity had the strongest correlation, while morphologic features such as tumor volume and surface area were negatively correlated between the two modalities. Conclusions: The higher 18F-FDG uptake metrics measured by dbPET highlighted the breast-specific modality’s sensitivity. This characteristic, when combined with the higher spatial resolution, may enable dbPET to detect treatment response in the primary tumor during NAC. Results from this study support dbPET as a complementary method for assessing the primary tumor in breast cancers, and future studies with a larger cohort are warranted.