Synergic role of preoperative 18F-fluorodeoxyglucose PET and MRI parameters in predicting histopathological features of endometrial cancer

Background The aim of the present study is to explore the correlation between PET and MRI parameters of primary tumour and clinicopathological features and to determine their synergic predictive role in patients with endometrial cancer candidate to surgery. Methods Retrospective study including 27 patients with endometrial cancer and preoperative 18F-fluorodeoxyglucose (18F-FDG)-PET and MRI scan. The following parameters, calculated on the primary tumour, were used for analysis: maximum standardized uptake value (SUVmax), SUVmean, metabolic tumour volume (MTV) and total lesion glycolysis (TLG) for PET scans; mean apparent diffusion coefficient (ADCmean) and volume index for MRI scans. FIGO stage, grade, histotype, lymphovascular space invasion (LVSI) and myometrial invasion were the considered clinicopathological features. Results MRI volume index was a good predictor for deep myometrial invasion [area under the curve (AUC) = 0.85; P = 0.003] and for LVSI (AUC = 0.74; P = 0.039). A cutoff value of 9.555 for MRI volume index was predictive for deep myometrial invasion (sensitivity = 84.6%; specificity = 76.9%); a cutoff of 12.165 was predictive for LVSI (sensitivity = 69.2%; specificity = 83.3%). A TLG cutoff value of 26.03 was predictive for deep myometrial invasion (sensitivity = 84.6%; specificity = 76.9%). A high-direct correlation was found with MRI volume index (rho = 0.722; P < 0.001); low-direct correlation with SUVmax (rho = 0.484; P = 0.012), SUVmean (rho = 0.47; P = 0.015) and TLG (rho = 0.482; P = 0.013) were identified. The SUVmax/ADCmean ratio showed a low-direct correlation with percentage of myometrial invasion (rho = 0.467; P = 0.016). Conclusion Volume index, TLG and SUVmax/ADCmean ratio are associated with deep myometrial invasion. As myometrial invasion is the index used to predict lymph node involvement in endometrial cancer, the synergic use of these imaging parameters may be suggested to predict lymphnodal metastases.