Allopurinol kinetics in humans as a means to assess liver function: design of a loading test

A six-compartment model of allopurinol and oxipurinol kinetics, after intravenous allopurinol injection in the human, is studied further to improve the blood and urine specimen collection schedule for clinical use. The effects of various error sources are also investigated by simple techniques like real data set truncation and adding normally distributed random errors to data obtained from simulation of allopurinol and oxipurinol plasma curves with preset parameters. All parameters estimation is performed with the NONLIN parameter estimation program. Main interest was focused on estimation of the fractional rate constant of transport from the central "extracellular" compartment to the metabolically active compartment. This parameter is regarded as a lumped measure of liver perfusion and liver cell membrane transport. The blood sampling schedule can be reduced to six specimens collected over 60 min, without affecting the accuracy and precision of estimated clinical parameters. The maximum allowable coefficient of variation for preanalytical errors and the analytical within-run and between-run errors are around 5, 4, and 5%, respectively. Analytical between-run bias up to 20% does not affect the estimate of the principal parameter, when both allopurinol and oxipurinol are biased in the same direction. Collection and analysis of urine samples was shown to be unnecessary.