ABSTRACT
The aim of this prospective study was to investigate the accuracy and inter-observer reliability of MRI in detection of local recurrence (LR) of pancreatic adenocarcinoma (PAC) after surgery, which was proved by PET-CT and access correlation between functional MRI and PET parameters. Forty-five patients who underwent PET-CT and MRI for follow-up purposes after radical operation of PAC were included. Twenty-three were PET positive (study group) and 22 negative for LR (control group). MR examination was performed within one month after PET-CT and three readers who were blind for PET-CT findings searched LR in T2W, 3D-dynamic post-contrast T1W-FS and DWI sequences, respectively. Sensitivity and specificity were calculated while inter-reader agreement was estimated by Cronbach's Alpha reliability coefficient (CARC). Apparent diffusion coefficient (ADC) of LR was correlated with the size (maximal diameter) and functional PET-CT parameters: mean and maximum standardized uptake values (SUVmean, SUVmax), metabolic tumor volume (MTV) and total lesion glycolysis (TLG), using Spearman's correlation coefficient (rS). Sensitivity and specificity among three readers in detecting the LR were 70% and 77-84% in T2W (CARC 0.806), 91-100% and 100% in 3D post-contrast T1W-FS (CARC 0.980), and both 100% in DWI sequences (CARC 1.000). Moderate inverse correlation was found between the ADC and SUVmean (rS = - 0.484), MTV (rS = - 0.494), TLG (rS = - 0.519) and lesion size (rS = - 0.567). MRI with DWI shows high diagnostic accuracy in detecting the LR of PAC in comparison to PET-CT as reference standard. ADC significantly inversely correlates with standard and advanced PET parameters and size of LR.
