Model-based compartmental analysis of the kinetics of retinol and beta-carotine in humans: Statistical consideration in designing and building models for retinol plus expanded models for beta-carotene

Vitamin A is an essential nutrient. Kinetic studies have added to understanding of whole-body vitamin A metabolism. In particular, model-based compartmental analysis of retinol kinetics has deepened our knowledge about vitamin A metabolism, homeostasis, and dynamics. Here I discuss statistical considerations in designing and building improved compartment models for vitamin A kinetics in humans as well as the application of compartmental analysis to investigate the bioconversion and kinetics of beta-carotene metabolism.;In chapter 1, I review the literature including 1) previous studies that applied mathematical modeling to studies of vitamin A metabolism and kinetics; 2) beta-carotene metabolism and early studies of beta-carotene kinetics; and 3) statistical considerations for compartmental analysis, robustness testing, and the experimental design of vitamin A kinetic studies.;In chapter 2, I discuss fixing fractional transfer coefficients in a compartmental model for vitamin A kinetics. Fractional standard deviations (FSDs) related to the early stage of absorption of vitamin A showed high inter-individual variability in the compartmental model was applied to data collected from nine American subjects after an oral dose of [2H8]retinyl acetate. Then, I modified the original model so that transfer coefficients for absorption were fixed and analyzed the same kinetic data. The fixed parameter model provided a good data-model fit and none of the values for kinetic parameters were significantly different from those based on the original model. A population correlation coefficient matrix showed that no pairs of correlation coefficients were greater than 0.8, a threshold for numerical identifiability. I conclude that the fixed parameter model for vitamin A kinetics, with just five adjustable parameters, is simpler and more useful for studying vitamin A kinetics in human subjects without significantly affecting the accuracy of fitting the data.;In chapter 3, I discuss blood sampling times in vitamin A kinetic studies. One of the most important issues for an experiment using compartmental analysis in human subjects is determining the optimal times for blood collection. Sensitivity analysis was conducted to determine the most sensitive data point(s) for each parameter and eliminated the less sensitive data points. I decreased the number of blood samplings in original study (20 to 22 samples) to nine based on sensitivity analysis. Three samples were needed for the absorption stage before reaching the peak, one sample to define the peak of the curve, two to define vitamin A turnover, and three to describe the terminal slope. Similar to the results for the theoretical analysis, modeling the reduced data set did not affect model-predicted fractional transfer coefficients or other kinetic parameters compared to the full data set model.;In chapter 4, I study the bioconversion of beta-carotene to retinol in human subjects. A compartmental model was developed for retinol derived from two different sources of beta-carotene: [2H]beta-carotene contained in Golden Rice and pure [2H8]beta-carotene. After testing the initial model for each individual.s data (n = 12), there was both intestinal- and post-absorptive conversion in some subjects, whereas others did not show discernible post-absorptive bioconversion. The absorption efficiency for a single dose of beta-carotene ranged from 7.0 to 47.8 % in the subjects who received pure beta-carotene and 22.7 to 55.9 % in the subjects. High between-subject variability was observed in the efficiency of bioconversion, ranging from 6.9 to 54.5 %.;In chapter 5, to describe the kinetics of intact beta-carotene, the model includes one compartment for chylomicrons (remnant) beta-carotene, one for plasma non-chylomicron lipoprotein beta- carotene, and one for beta-carotene in the liver. Irreversible loss of beta-carotene from the system occurs directly from plasma in the proposed model. For retinol derived from beta-carotene in the enterocytes, the model includes one compartment for liver retinol, one for plasma retinol bound to retinol-binding protein, and one extravascular vitamin A pool that exchanges with plasma. The model predicted that 43 and 28 % of the beta-carotene dose was absorbed and that 82 and 66 % of the absorbed beta-carotene dose was taken up as intact beta-carotene by subjects 1 and 2, respectively. Total bioconversion for subjects 1 and 2 was 12.8 and 15.8 % and approximately 60 % of conversion occurred in the intestine for both participants. Only 10 % of extravascular â-carotene was converted to liver retinol via post-absorptive conversion for both subjects.;Overall this work contributes to the whole-body kinetics of retinol and beta-carotene metabolism in humans. The application of model-based compartmental analysis to nutritionally-interesting compounds allows investigators to generate unique information and thus advance knowledge.