Uncertainty analysis of (99m)Tc-HEPIDA liver clearance determination.

BACKGROUND

The aim of the study was to obtain information on the accuracy and precision of (99m)Tc-HEPIDA hepatic (Cl(Hp)) and plasma (Cl(Pl)) clearances and selection of an appropriate estimator of the measurement uncertainty of a single determination of these quantities.

MATERIAL AND METHODS

In a simulation (Monte Carlo) experiment, it was assumed that the recorded results of plasma and hepatic clearances, as obtained from 185 patients, provided authentic information about (99m)Tc-HEPIDA behaviour in the body over a wide range of the clearances studied. The time course (99m)Tc-HEPIDA concentration in blood plasma has been described by means of biexponential function with parameter values derived for each patient. For each patient, using these data and urinary excretion data, there had been 5000 simulations performed; in each of the latter, the directly measured numbers have been substituted by simulated ones, obtained by means of varying the real ones, using random generated values. These reflected errors of plasma and radioactive standard pipetting (from 1 to 5%) and stochasticity of counting radioactive decay (1%). The time of blood sampling and urine voiding was also varied, assuming realistic uncertainty. The varied values were then used for computation of the simulated clearances. From the 5,000 calculated clearances for each patient, mean-values were calculated, as well as mean standard errors, standard deviations and mean uncertainty of measurements using a widely accepted rule of partial error propagation, and, in addition, a modified rule of the latter. Accuracy of clearance (Cl(Pl), Cl(Hp), Cl(Ur)) determination was assessed on the basis of comparison of mean values from simulations with those from directly recorded values. Precision was identified with standard deviation of each of the 5000 simulations. The uncertainty thus obtained was compared with results of calculated traditional and modified uncertainty. There was good agreement between standard deviation of the simulations with results of the modified calculation of total differential. Therefore, a coefficient of variation from simulation computations and a modified means of calculation of the propagated errors was accepted as a measure of uncertainty of a single determination.

RESULTS/CONCLUSIONS: There was a very high correlation between the mean values from simulations and those from direct determinations (r > 0.98 in each case). The regression lines practically corresponded to the lines of identity. These correlations were not affected by the assumed range of pipetting uncertainty. In conclusion, the methods of (99m)Tc-HEPIDA clearance determination are satisfactory. Precision of clearance measurements depends substantially upon uncertainty of pipetting. For plasma clearance, the coefficients of variations at Cl(Pl) > 350 ml/min and at ab.80 ml/min amounted to 2 and 11% respectively, at pipetting uncertainty of 2%. Similarly, for hepatic clearances of (99m)Tc-HEPIDA of 300 ml/min and 30 ml/min, CV was 2.5 and 25%, respectively (at the same uncertainty of pipetting).