Parameter identifiability and Extended Multiple Studies Analysis of a compartmental model for human vitamin A...

In the existing compartmental models of human vitamin A metabolism, parameters related to the absorption of the isotopic oral dose have not been well identified. We hypothesised that fixing some poorly identified parameters related to vitamin A absorption would improve parameter identifiability and add statistical certainty to such models. In the present study, data for serum vitamin A kinetics in nine subjects given [2H8]retinyl acetate orally and a model with absorption fixed at 75 % were used to test this hypothesis. In addition to absorption efficiency, we fixed two other fractional transfer coefficients: one representing the initial processing of the ingested dose and the other representing the direct secretion of retinol bound to retinol-binding protein (RBP) from enterocytes into the plasma. The Windows version of Simulation, Analysis and Modeling software (WinSAAM) was used to fit serum tracer data v. time for each subject. Then, a population model was generated by WinSAAM's Extended Multiple Studies Analysis. All the parameters had fractional standard deviations < 0·5, and none of the pairs of parameters had a correlation coefficient >0·8 (accepted criteria for well-identified parameters). Similar to the values predicted by the original model, total traced mass for retinol was 1160 (sd 468) μmol, and the time for retinol to appear in the plasma bound to RBP was 31·3 (sd 4·4) h. In conclusion, we suggest that this approach holds promise for advancing compartmental modelling of vitamin A kinetics in humans when the dose must be administered orally.