Liver Standardized Uptake Value Corrected for Lean Body Mass at FDG PET/CT: Effect of FDG Uptake Time

OBJECTIVEThe objective of this study is to establish the magnitude change and interreader reliability of the liver standardized uptake value corrected for lean body mass (SULmean) in dual-time-point imaging at 1 and 2 hours and 1 and 4 hours. PATIENTS AND METHODSEarly and delayed FDG PET/CT scans were included for 28 patients (13 men and 15 women) who had normal liver by CT or ultrasound. The average uptake time between the early and delayed scans were 55 minutes (range, 44–69 minutes) for pancreatic adenocarcinoma patients (n = 19) and 184 minutes (range, 140–197 minutes) for neurofibromatosis patients (n = 9). A 30-mm-diameter spherical volume of interest was placed within the right lobe of the liver above, below, and at the level of the main portal vein by 2 independent readers. Correlation coefficients, analysis of variance, intraclass correlation coefficient, and Bland-Altman analysis were performed. RESULTSThe mean liver SULmean was between 1.39 and 1.42 and between 1.28 and 1.3 in early and delayed images, respectively (P = 0.001). There is time-dependent reduction in the mean liver SULmean at 2-hour (7%–8%) and 4-hour uptake time (15%–21%) compared with 1-hour uptake time. The correlation coefficient between delayed uptake time and liver SULmean reduction is 0.39 to 0.41 at the upper aspect of the liver. The intraclass correlation coefficient for 2 readers varied between 0.997 and 0.998 and between 0.995 and 0.999 in early and delayed images, respectively (P = 0.001). CONCLUSIONSThere is time-dependent reduction of mean liver SULmean, about 7% to 8% within the clinically relevant FDG uptake time, in the same patient with excellent interreader agreement in early and delayed images within the right lobe of the liver. Therefore, liver SULmean could represent a useful reference parameter in quantitative analysis of dual-phase FDG PET/CT in malignancy or atypical infection/inflammatory disease. Furthermore, it may be suitable as a normalization factor in currently available formulae quantifying therapy response on PET imaging.