Abstract
Standardized uptake value normalized to lean body mass (SUL) is commonly calculated using predictive equations; however, these equations may not accurately reflect lean body mass (LBM) across diverse populations. This study compared SUL values derived from two conventional predictive equations (Pieterman and Janmahasatian) with those obtained using a body composition analyzer (BCA)-based method in a Japanese population and evaluated their clinical utility. Mean SUVs were measured in the liver using a 1-cm diameter region of interest on PET images from 85 subjects (43 males, 42 females) undergoing PET screening. Four subjects (three males and one female) diagnosed with fatty liver were excluded from the analysis. LBM and SUL were estimated using two equation-based methods and a BCA-based method. Pearson's product-moment correlation was used to assess the relationship between body weight and both SUV and SUL. Agreement between equation-based and BCA-based SUL was evaluated using Bland-Altman analysis (with limits defined by the 2.5th and 97.5th percentiles) and intraclass correlation coefficients (ICC, model 2,1). There were significant positive correlations between body weight and both SUV and SUL for all subjects (all P < 0.001). In males, the correlation for SUV exceeded that for SUL, whereas in females, the equation-based SUL values showed weak or negative correlations with body weight. Multiple comparisons revealed that, for the overall cohort, SUL from Janmahasatian's method did not differ significantly from that obtained using the BCA-based method, whereas in subgroup analyses by gender, BCA-based SULs were significantly different from both prediction formulas. Bland-Altman analysis indicated minimal median differences between methods: the median difference between BCA-based SUL and Pieterman's SUL was 0.05 g*mL-1 (-0.07 g*mL-1-0.25 g*mL-1), and between BCA-based SUL and Janmahasatian's SUL was 0.00 g*mL-1 (-0.12 g*mL-1-0.03 g*mL-1), suggesting negligible systematic bias. The ICC between Janmahasatian's SUL and the BCA-based method was 0.932 overall (0.945 in males and 0.840 in females), while for Pieterman's method it was 0.845 overall (0.893 in males and 0.658 in females). In this Japanese cohort, SUL calculated using the BCA-based method was overall comparable to that derived from Janmahasatian's predictive equation, supporting its viability as an alternative for LBM estimation and SUL calculation. However, subgroup analysis revealed that the BCA-based method performed less reliably in females, highlighting the importance of considering population-specific and gender-related differences in clinical PET studies.