Characterization of a high-resolution breath acetone meter for ketosis monitoring

Background The ketone bodies beta-hydroxybutyrate (BHB) and acetone are endogenous products of fatty acid metabolism. Although ketone levels can be monitored by measuring either blood BHB or breath acetone, determining the precise correlation between these two measurement methods has been challenging. The purpose of this study is to characterize the performance of a novel portable breath acetone meter (PBAM) developed by Readout, Inc., to compare single versus multiple daily ketone measurements, and to compare breath acetone (BrAce) and blood BHB measurements. Methods We conducted a 14-day prospective observational cohort study of 21 subjects attempting to follow either a low-carbohydrate/ketogenic or a standard diet. Subjects were asked to concurrently measure both blood BHB and BrAce five times per day and report the results using an online data entry system. We evaluated the utility of multiple daily measurements by calculating the coefficient of variation (CV) for each daily group of measurements. We calculated the correlation between coincident BrAce and blood BHB measurements using linear ordinary least squares regression analysis. We assessed the ability of the BrAce measurement to accurately predict blood BHB states using receiver operating characteristic (ROC) analysis. Finally, we calculated a daily ketone exposure (DKE) using the area under the curve (AUC) of a ketone concentration versus time graph and compared the DKE of BrAce and blood BHB using linear ordinary least squares regression. Results BrAce and blood BHB varied throughout the day by an average of 44% and 46%, respectively. The BrAce measurement accurately predicted whether blood BHB was greater than or less than the following thresholds: 0.3 mM (AUC = 0.898), 0.5 mM (AUC = 0.854), 1.0 mM (AUC = 0.887), and 1.5 mM (AUC = 0.935). Coincident BrAce and blood BHB measurements were moderately correlated with R2 = 0.57 (P < 0.0001), similar to literature reported values. However, daily ketone exposures, or areas under the curve, for BrAce and blood BHB were highly correlated with R2 = 0.80 (P < 0.0001). Conclusions The results validated the performance of the PBAM. The BrAce/BHB correlation was similar to literature values where BrAce was measured using highly accurate lab instruments. Additionally, BrAce measurements using the PBAM can be used to predict blood BHB states. The relatively high daily variability of ketone levels indicate that single blood or breath ketone measurements are often not sufficient to assess daily ketone exposure for most users. Finally, although single coincident blood and breath ketone measurements show only a moderate correlation, possibly due to the temporal lag between BrAce and blood BHB, daily ketone exposures for blood and breath are highly correlated.