Accuracy of Resting Energy Expenditure Predictive Equations in Patients With Cancer

Background Our purpose was to assess the accuracy of resting energy expenditure (REE) equations in patients with newly diagnosed stage I-IV non-small cell lung, rectal, colon, renal, or pancreatic cancer. Methods In this cross-sectional study, REE was measured using indirect calorimetry and compared with 23 equations. Agreement between measured and predicted REE was assessed via paired t-tests, Bland-Altman analysis, and percent of estimations <= 10% of measured values. Accuracy was measured among subgroups of body mass index (BMI), stage (I-III vs IV), and cancer type (lung, rectal, and colon) categories. Fat mass (FM) and fat-free mass (FFM) were assessed using dual x-ray absorptiometry. Results Among 125 patients, most had lung, colon, or rectal cancer (92%, BMI: 27.5 +/- 5.6 kg/m(2), age: 61 +/- 11 years, REE: 1629 +/- 321 kcal/d). Thirteen (56.5%) equations yielded REE values different than measured (P < 0.05). Limits of agreement were wide for all equations, with Mifflin-St. Jeor equation having the smallest limits of agreement, -21.7% to 11.3% (-394 to 203 kcal/d). Equations with FFM were not more accurate except for one equation (Huang with body composition; bias, limits of agreement: -0.3 +/- 11.3% vs without body composition: 2.3 +/- 10.1%, P < 0.001). Bias in body composition equations was consistently positively correlated with age and frequently negatively correlated with FM. Bias and limits of agreement were similar among subgroups of patients. Conclusion REE cannot be accurately predicted on an individual level, and bias relates to age and FM.