The Utility of CT-derived Lean Body Mass Correction of Weight-based Standardized Uptake Values.

1897 Objectives 18F-FDG PET has widespread use in cancer patients to image and quantify glucose metabolism in vivo. Quantification is expressed as a “standardized uptake value” (SUV) - usually determined using patient weight, denoted SUVW. Diagnostic algorithms increasingly use SUV to monitor response to therapy or to identify progression. Body composition analysis is becoming a major area of cancer research and lean body mass (LBM) is increasingly seen as an important factor in predicting cancer outcomes. LBM is calculated by subtracting adipose tissue (AT) weight from total body weight. CT scans precisely quantify patient-specific LBM and total muscle cross sectional area at L3 is directly related to LBM (R2 = 0.94). 18F-FDG distributes almost exclusively in LBM. Changes in LBM/AT ratios will result in lesion SUV changes, incorrectly suggesting changes in glucose metabolism. Using SUV normalized by LBM rather than overall weight better reflects changes in actual glucose metabolism. Two approaches, James and Janmahasatian (Janma) equations, have previously been used to define LBM normalized SUV (SULJames and SULJanma). These equations, however, describe average LBM for healthy populations, not populations of cancer patients and neither equation estimates patient-specific changes in body composition from serial scanning time points. We propose that patient-specific LBM, derived from CT, be used to normalize SUV in patients, denoted SULps. Methods Quantitative analysis of serial CT images, ~ 3 months apart, was conducted to assess loss/gain of LBM and AT in a population-based cohort of patients with advanced cancer. We assessed variability between SUVw, SULJames, SULJanma, and SULps and evaluated if typical changes in body composition experienced by cancer patients caused these values to change by an amount considered clinically relevant. Results Of 1080 PET/CT scan intervals evaluated, 50.1% (n=546) had stable LBM ( Conclusions 18F-FDG distributes almost exclusively in LBM. Patients with advanced cancer, or on treatment, tend to lose muscle, and may gain fat. SUVw is highly dependent on the technique used for SUV normalization. We have shown that changes in body composition will significantly modify these results, independent of tissue glycolysis. If PET outcome measures, such as PERCIST, are to become routine tools in cancer management a more robust measure of SUVw is required. Both SULJames and SULJanma overestimate LBM for patients with advanced cancer. Using SULps for SUV normalization provides more accurate values and, when adopted, will lead to enhanced patient care.