A comparison in neonatal piglets of model-dependent and model-independent methods for measuring glucose turnover

Steady state glucose turnover was measured in 53 conscious, unrestrained neonatal piglets by two different methods using 3H-labelled glucose as the tracer. In the first, model-dependent method, the plasma tracer concentrations following a spike injection of tracer were fitted in various different ways and the turnover was calculated from the area under the fitted curve extrapolated to infinity. Both exponential curves and negative power functions of time were fitted by non-linear least squares. In the second, model-independent method, tracer was infused continuously and turnover calculated from its equilibrium concentration. Conditions during 34 of the 53 experiments failed various tests of the steady state and the data were therefore discarded. In the remaining 19, glucose turnover measured by the different methods agreed fairly well. Although there was little difference between the results obtained by a two-exponential curve fit, by those obtained when the fast component was ignored, by a single exponential curve fit, and by the results from fitting a negative power function of time, the conventional two-exponential curve is to be preferred. There was a suggestion that at high turnover rates, the injection result underestimated the infusion result, assuming the latter to represent the reference method. This might have been due to extrapolation errors in the injection technique, either in the first few minutes following injection, or in the interval after the last sample. However, the notional volumes of distribution of the tracer suggest that the terminal phase of tracer elimination is in fact slightly faster than that predicted on the basis of samples taken during the first 120 min post-injection.