The effect of new formulas for lean body mass on lean- body-mass-normalized SUV in oncologic 18F-FDG PET/CT

Abstract

Because of better precision and intercompatibility, the use of lean body mass (LBM) as a mass estimate in the calculation of SUV (SUL) has become more common in research and clinical studies today. Thus, the equations deciding this quantity must be those that best represent the actual body composition. Methods: LBM was calculated for 44 patients examined with 18F-FDG PET/CT scans by means of the sex-specific predictive equations of James and Janmahasatians, and the results were validated using a CT-based method that makes use of the eyes-to-thighs CT component of the PET/CT aquisition and segments the voxels according to Hounsfield units. Intraclass correlation coefficients and Bland-Altman plots were used to assess agreement between the various methods. Results: A mean difference of 6.3 kg (limits of agreement, -15.1 to 2.5 kg) between LBMjames and LBMCT1 was found. This difference was higher than the 3.8-kg difference observed between LBMjan and LBMCT1 (limits of agreement, -12.5 to 4.9 kg). In addition, LBMjan had a higher intraclass correlation coefficient with LBMCT1 (0.87; 95%confidence interval, 0.60-0.94) than with LBMjames (0.77; 95% confidence interval, 0.11-0.91). Thus, we obtained better agreement between LBMjan and LBMCT1. Although there were exceptions, the overall effect on SUL was that SULjames was greater than SULjan. Conclusion: We have verified the reliability of the suggested LBMjan formulas with a CT-derived reference standard. Compared with the more traditional and available set of LBMjames equations, the LBMjan formulas tend to yield better agreement.