OBJECTIVE. The purpose of this study was to assess the diagnostic value of F-18-FDG PET/CT and MRI in predicting the clinicopathologic subtypes of breast cancer.
MATERIALS AND METHODS. The cases of 89 patients with mass-type invasive breast cancer who underwent FDG PET/CT and MRI before therapy were retrospectively analyzed. Eight imaging variables-maximum standardized uptake value (SUVmax), apparent diffusion coefficient, size, shape, margin, intratumoral enhancement, dynamic kinetics, and high intratumoral signal intensity on T2-weighted images-were compared with results for the pathologic markers Ki-67 antibody, estrogen receptor (ER), progesterone receptor (PR), and ERBB2 (formerly HER2 or HER2/neu). The diagnostic performance of the imaging variables for subtyping was evaluated, and the predictors of the subtypes were elucidated.
RESULTS. Higher SUVmax was significantly associated with a high Ki-67 index (p < 0.0001), ER-negative status (p = 0.0001), and PR-negative status (p = 0.047). Significant correlation was also found between size and ER status (p = 0.002) and between shape and PR status (p = 0.044). The AUC exceeded 0.7 only in identification of the luminal A subtype by application of cutoff values for SUVmax (AUC, 0.751). When smaller tumors were excluded, AUC increased (AUC, 0.803 for tumors > 16 mm). Multivariate analysis showed that SUVmax was the sole independent predictor of luminal A subtype (odds ratio per SD, 0.291; p < 0.0001). SUVmax was significantly lower for luminal A (4.4 +/- 2.2) than non-luminal A (8.1 +/- 4.4; p < 0.0001) tumors. A cutoff value of 5.4 yielded 79% sensitivity and 68% specificity for prediction that a tumor was the luminal A subtype.
CONCLUSION. FDG PET/CT findings may contribute to differentiation of the luminal A and non-luminal A subtypes of invasive breast cancer.