Evaluating the Serum Carbon Isotope Ratio as a Biomarker for Animal Protein Ratio in a Controlled Feeding Study of US Adults

Abstract

ObjectivesRecent studies have identified the serum natural abundance carbon isotope ratio (CIR) as a candidate biomarker of animal protein intake in postmenopausal women. Such a biomarker would help clarify the contribution of dietary protein source (animal vs. vegetable) to chronic disease risk. Here we evaluate biomarker performance and develop a biomarker calibration equation in a mixed-age and – gender cohort. MethodsWe conducted a 15-d feeding study of 100 adults (age 18–70, 55% women) in Phoenix, AZ. Participants were provided individualized diets that approximated habitual food intakes. Total CIR and nitrogen isotope ratio (NIR) were measured in sera collected at the end of the feeding period. We expressed animal protein as a ratio of total protein intake (APratio). We evaluated a model of serum CIR based on APratio, the serum NIR, gender, age and body weight, and the resulting regression equation was inverted to develop an equation for the APratio that we call the calibrated biomarker. We evaluated the association of the calibrated biomarker with actual APratio using Pearson correlation and 5-fold cross validation. ResultsAnimal protein intake in this study was 73 ± 30 g/d (mean ± SD) and the APratio was 0.63 ± 0.13. Our model explained a large proportion of the variation in serum CIR (R2 = 0.77) and APratio was the only significant model effect (coefficient = 6.22, SE = 0.44, P < 0.0001). Inverting that model generated the following biomarker calibration equation: APratio = (CIR – 26.35 – 0.06 (gender) + 0.068 * In age – 0.215 * In body weight – 0.204 * serum NIR)/6.22, where gender = 1,0 (male, female). There was a strong correlation between model-predicted and actual APratio (rP = 0.85, P < 0.0001), with the mean model-predicted APratio differing from mean actual APratio by 0.0015 (SE = 0.0077). The standard deviation of the prediction error was 0.076. The 5-fold cross validation procedure produced very similar model R2, effects, and prediction errors. ConclusionsThese data suggest that the serum CIR has potential as a predictive biomarker of APratio, providing a useful tool for objectively assessing dietary protein intake patterns. Such a tool could help resolve the contribution of dietary patterns favoring animal protein intake to chronic disease risk. Funding SourcesThis work was funded by NIH U01 CA197902.