Computed tomography-derived assessments of regional muscle volume: Validating their use as predictors of whole body muscle volume in cancer patients

OBJECTIVE:: Evaluate the accuracy of CT-derived regional skeletal muscle volume (SMV) measurements to predict whole body SMV in patients with melanoma. METHODS:: 148 patients with advanced melanoma who underwent whole body positron emission tomography/CT were studied. Whole body SMV was measured on CT and used as the reference standard. CT-derived regional measures of SMV were obtained in the thorax, abdomen, pelvis, and lower limbs. Models were developed on a discovery cohort (n-98), using linear regression to model whole body SMV as a function of each regional measure, and clinical factors. Predictive performance of the derived models was evaluated in a validation cohort (n = 50) by estimating the explained variation (R²) of each model. RESULTS:: In the discovery cohort, all regional SMV measurements were significantly associated with whole body SMV [β1 range: 0.673-1.153, all p < 0.001)]. The magnitude of association was greatest for pelvic regional measurements {β = 1.153, [95% confidence interval (0.989, 1.317)]}. Prediction algorithms incorporating clinical variables and regional SMVs were developed to estimate whole body SMV from regional assessments. Using the validation cohort to predict whole body SMV, the R² values for the pelvic, abdominal and thoracic regional measurements were 0.89, 0.86, 0.78. CONCLUSION:: Regional measures of SMV are strong predictors of whole body SMV in patients with advanced melanoma. ADVANCES IN KNOWLEDGE:: The first study utilizing whole body imaging as a reference standard validating the use of regional SMVs in cancer patients, including validating the use of regional SMVs outside of traditionally assessed areas.