Individual Variability Is More Important Than Analytical Methods When Calculating Relative Speed of Beverage Bioavailability

Abstract

Deuterium oxide (D2O) appearance in blood is a marker of fluid bioavailability. However, whether biomarker robustness (e.g., relative fluid delivery speed) is consistent across analytical methods (e.g., cavity ring-down spectroscopy) remains unclear. Fourteen men ingested fluid (6 ml/kg body mass) containing 0.15 g/kg D2O followed by 45 min blood sampling. Plasma (D2O) was detected (n = 8) by the following: isotope-ratio mass spectrometry after vapor equilibration (IRMS-equilibrated water) or distillation (IRMS-plasma) and cavity ring-down spectroscopy. Two models calculated D2O halftime to peak (t1/2max): sigmoid curve fit versus asymmetric triangle (TRI). Background (D2O) differed (p < .001, η2 = .98) among IRMS-equilibrated water, IRMS-plasma, and cavity ring-down spectroscopy (152.2 ± 0.8, 147.2 ± 1.5, and 137.7 ± 2.2 ppm), but did not influence (p > .05) D2O appearance (Δppm), time to peak, or t1/2max. Stratifying participants based on mean t1/2max (12 min) into “slow” versus “fast” subgroups resulted in a 5.8 min difference (p < .001, η2 = .73). Significant t1/2max model (p = .01, η2 = .44) and Model × Speed Subgroup interaction (p = .005, η2 = .50) effects were observed. Bias between TRI and sigmoid curve fit increased with t1/2max speed: no difference (p = .75) for fast (9.0 min vs. 9.2 min, respectively) but greater t1/2max (p = .001) with TRI for the slow subgroup (16.1 min vs. 13.7 min). Fluid bioavailability markers are less influenced by which laboratory method is used to measure D2O as compared with the individual variability effects that influence models for calculating t1/2max. Thus, TRI model may not be appropriate for individuals with slow fluid delivery speeds.