Glucose absorption and production following oral glucose: comparison of compartmental and arteriovenous-difference methods.

Both whole-body models and the regional arteriovenous (AV)-difference method have been used to calculate systemic glucose rates of appearance (Ras) under non-steady-state conditions. Although whole-body models have been experimentally validated in the dog, direct comparison of the whole-body and regional-balance approach has not been made in man. We reanalyzed published data obtained by combining the double-tracer technique ([3H]glucose infusion with ingestion of a [14C]glucose-labeled load) with hepatic vein catheterization. Steele's monocompartmental model underestimated the Ra of oral glucose ([RaO] by 12%, P < .0001) and overestimated the Ra of endogenous glucose ([RaE] by 19%, P < .0001) in comparison to a two-compartment (2-c) model calculation. Splanchnic balance data were used to compute the total glucose Ra (RaT) with either steady-state or non-steady-state equations (the latter by estimating splanchnic transit times). Except for one early time point (15 minutes), the two calculations agreed well with one another. The glucose RaT by the balance method was well correlated with that calculated by the 2-c whole-body model (r = .72, P < .0001 on all data points); however, the former significantly underestimated the latter by 0.62 mg.min-1.kg-1 (or 13%, P < .01) on average throughout the absorptive period. This bias was small in comparison to the estimated random error affecting the calculation of glucose RaT by the model, which averaged 1.4 mg.min-1.kg-1 (corresponding to a variation coefficient of approximately 30%).(ABSTRACT TRUNCATED AT 250 WORDS)