Sources of plasma glucose by automated Bayesian analysis of 2H NMR spectra.

Sources of blood glucose can be determined after oral ingestion of (2)H(2)O followed by isolation of plasma glucose and measurement of the relative (2)H enrichments in select positions within the glucose molecule. Typically, (2)H enrichments are obtained by mass spectrometry but (2)H NMR offers an alternative. Here it is demonstrated that the entire analysis may be automated by Bayesian analysis of a (2)H free induction decay signal of monoacetone glucose to obtain a direct readout of the relative contributions of glycogenolysis, glycerol, and phosphoenol pyruvate to plasma glucose production. Furthermore, Markov Chain Monte Carlo (MCMC) simulations of the posterior probability density provide uncertainties in all metabolic parameters from a single patient, thereby allowing comparisons in glucose metabolism from one individual to another. The combined MCMC Bayesian methodology is operationally simple and requires little intervention from the operator.