Analysis of the variability of biokinetic model parameters due to inter-individual variation.

Biokinetic models for the assessment of individual internal doses represent the physiological processes of a standardized human, which affect the internal distribution of the radionuclide of interest. The flow from one compartment into another is specified by transfer rates, which may vary from person to person. The rate constants are then representing the mean values for the population as a whole, around which individual parameters fluctuate according to given probability densities. Analytical distribution propagation can be calculated only for very simple models. The influence of inter-individual variation can be studied with Monte Carlo simulations, which compare the simulated compartment content distributions using different initial distributions for the parameters. The simulations indicate that the form of test distributions affect the distributions of compartment contents only for very simple models in the early stages. Later on, the distributions converge to a lognormal shape. The coefficients of variation of the initial distributions can be adjusted so that the resulting distributions resemble each other. Due to the lack of significant differences, lognormal distributions--which are found in most measurable body parameters--were used for further studies. The range of inter-subject variability can be estimated by comparing data generated with Monte Carlo simulations with observed data. For the plutonium model, data retrieved long times post intake are most suitable for this purpose when the redistribution of the radionuclide in the compartments is in a state of quasi-equilibrium and the ratio of plutonium in different compartments is nearly constant. For the estimation of inter-individual variability, the ratios of the main excretion paths and the organs of main burden can be used. The comparison of observed and simulated standard deviations indicates a value of 0.6 for the coefficient of variation for all transfer rates. The generated distributions show good agreement with the available data and thus confirm that the simulations can represent the inter-individual variation in the biokinetic plutonium model.