Development and Initial Validation of a Simplified Approach for Compartmental Modeling of Vitamin A Kinetics in...

Compartmental modeling can provide unique descriptive and quantitative information about the whole-body metabolism of nutrients. Successful application of modeling requires extensive experience with software such as Simulation, Analysis and Modeling (SAAM), which has limited its use in nutrition.

Using vitamin A (VA) as an example, we tested the hypothesis that one could successfully model a new set of data using a "master deck" (model input file) developed for another dataset by an experienced modeler.

We generated retinol kinetic parameters and state variables for 20 theoretical adults with a wide range of VA total body stores (TBS) and simulated plasma retinol tracer response after an oral isotope dose. Using SAAM, we created a master deck to model group mean tracer response and obtained estimates of model parameters, including TBS. Then we replaced the mean data with that for 4 of the theoretical individuals or for a group of previously studied Ghanaian women, adjusted key parameters, if needed, and determined model parameters for comparison with assigned values (theoretical subjects) or published results (Ghanaian women).

Ratios of model-predicted/reference values for TBS ranged from 1.00 to 1.02 for the 4 theoretical subjects and were 1.00 for the Ghanaian women; fitting required only minor adjustments, if any, in the model parameter related to retinol transfer from plasma to liver stores. Using the theoretical data, we also confirmed that, for adults with a wide range of TBS, the best time to apply retinol isotope dilution was at 21-28 d postdosing.

Results indicate that, using a previously published compartmental model and the associated modeling deck, researchers may be able to analyze a new VA kinetics dataset, making minor adjustments in key parameters as needed to allow nonlinear regression analysis. This approach may make compartmental analysis more accessible for VA researchers.