Joanne Balmer Green

Use of Model‐Based Compartmental Analysis to Study Vitamin A Kinetics and Metabolism

We discuss the use of mathematical modeling, and specifically model-based compartmental analysis, to analyze vitamin A kinetic data obtained in rat and human studies over the past 25 years. Following an overview of whole-body vitamin A metabolism, a review of early kinetic studies, and an introduction to the approach and terminology of compartmental analysis, we summarize studies done in this laboratory to develop models of whole-body vitamin A metabolism in rats at varying levels of vitamin A status. Highlights of the results of these studies include the extensive recycling of vitamin A among plasma and tissues before irreversible utilization and the existence of significant extrahepatic pools of the vitamin. Our studies also document important differences in vitamin A kinetics as a function of vitamin A status and the importance of plasma retinol pool size in vitamin A utilization rate. Later we describe vitamin A kinetics and models developed for specific organs including the liver, eyes, kidneys, small intestine, lungs, testes, adrenals, and remaining carcass, and we discuss the effects of various exogenous factors (e.g., 4-HPR, dioxin, iron deficiency, dietary retinoic acid, and inflammation) on vitamin A dynamics. We also briefly review the retrospective application of model-based compartmental analysis to human vitamin A kinetic data. Overall, we conclude that the application of model-based compartmental analysis to vitamin A kinetic data provides unique insights into both quantitative and descriptive aspects of vitamin A metabolism and homeostasis in the intact animal.

Influence of dietary fat saturation on lipid absorption in the rat

Flux (mumole/h) of triglycerides, phospholipids and cholesterol into the thoracic duct lymph was measured in rats receiving a constant intraduodenal infusion of a cholesterol-free oil rich in either polyunsaturated (P/S = 4.8) or saturated (P/S = 0.2) fatty acids. Rats had ad libitum access to a fat-free semi-synthetic diet throughout the experiment. Oils were infused at a rate equivalent to a 10% (w/w) fat-containing diet for at least 18 h prior to collection of lymph; both oils were compared in each animal. Although absorption of the infused oils approximated 100%, triglyceride flux was significantly lower during infusion of the saturated compared to the polyunsaturated oil. Phospholipid and total cholesterol fluxes were not significantly affected by the type of oil, but the percent of lymph total cholesterol which was esterified was slightly but significantly lower during infusion of the unsaturated oil. Using the molar phospholipid/triglyceride ratio as a index of lymph lipoprotein size, it was found that absorption of the oil rich in polyunsaturated fatty acids resulted in an increase in the mean size of lymph lipoproteins. The potential significance of an influence of dietary fat saturation on lymph lipoprotein size and cholesterol esterification for the ultimate metabolic fate of absorptive lipoprotein constituents is discussed.

Development of a Compartmental Model Describing the Dynamics of Vitamin A Metabolism in Men

Model-based compartmental analysis was used with the Simulation, Analysis and Modeling (SAAM) computer programs to analyze data on plasma retinoid kinetics in adult male subjects for 7 d after a single oral dose of 105 mumol of [8,9,19-13C]retinyl palmitate. We present here the data for one subject and discuss in detail the steps taken to develop a physiologically-based compartmental model that describes the dynamic behavior of plasma retinyl esters, [12C]retinol, [8,9,19-13C]retinol, and the sum of [12C] and [13C]retinol. First an absorption model was developed to fit data on the plasma appearance and disappearance of retinyl esters; this was used as input in development of models for labeled and unlabeled retinol. The large oral load of labeled vitamin A perturbed the unlabeled tracee system, and thus parallel models for tracer and tracee were developed; and a time-variant fractional transfer coefficient was incorporated into the tracee model. Following the absorption model, four-compartment models were developed to describe the dynamics of both labeled and unlabeled retinol. These models predict that, in spite of the large vitamin A load, the absorption efficiency was 34%; hepatic (presumably parenchymal cell) processing of the absorbed dose was essentially complete by 24 h; and, by 7 days, approximately 80% of the absorbed dose was in a compartment that presumably represents stored liver retinyl esters. The model also predicts that approximately 50 mumol of retinol passed through the plasma each day, compared to an estimated utilization rate of 4 mumol/day. This project provides unique and important information about whole-body vitamin A dynamic in humans, and presents approaches to specific modeling issues that may be encountered by others.

Chapter 1 – Quantitative and Conceptual Contributions of Mathematical Modeling to Current Views on Vitamin a Metabolism, Biochemistry, and Nutrition

This chapter aims to demonstrate that the application of a variety of modeling techniques to the vitamin A system that resulted in different levels of insight in the metabolism, biochemistry, and nutrition of this essential nutrient. The chapter integrates the ideas generated by the application of mathematical modeling to vitamin A kinetic data to illustrate how the approach has advanced and complicated the understanding of retinol metabolism. An empirical compartmental analysis is used to develop a general three-compartment model for vitamin A kinetics in rats. To develop a whole-body model for vitamin A metabolism that incorporates dynamics in many tissues, a large number of data points from plasma, urine, feces, and individual organs must be modeled simultaneously. Triglyceride-depleted, vitamin A-containing chylomicron remnants were thought to be quantitatively cleared by the liver; there the vitamin was processed and secreted into plasma bound to its specific plasma transport protein, retinol-binding protein (RBP). Plasma retinol is the major determinant of vitamin A disposal rate and hypothesize that degradation may be driven by an intracellular pool of vitamin A in equilibrium with plasma retinol.

Plasma Retinol Kinetics and β-Carotene Bioefficacy Are Quantified by Model-Based Compartmental Analysis in Healthy Young Adults with Low Vitamin A Stores

Background: Model-based compartmental analysis of data on plasma retinol kinetics after administration of labeled retinol provides unique information about whole-body vitamin A metabolism. If labeled β-carotene is coadministered, its bioefficacy relative to the retinol reference dose can also be estimated. Objectives: The objectives were to model plasma retinol kinetics after administration of labeled preformed vitamin A and provitamin A β-carotene and to determine relative β-carotene bioefficacy. Methods: We used the Simulation, Analysis and Modeling software (WinSAAM version 3.0.8; http://www.WinSAAM.org) to analyze previously collected data on plasma [13C10]- and [13C5]retinol kinetics for 14 d after oral administration of 1 mg [13C10]retinyl acetate and 2 mg [13C10]β-carotene in oil to 30 healthy young adults of European ancestry [13 men, 17 women; mean ± SD age: 24.5 ± 4.2 y; mean ± SD body weight: 65.2 ± 10 kg; mean ± SD body mass index (in kg/m2): 22.5 ± 1.9] with moderate vitamin A intakes. Results: A 6-component model provided the best fit to the data, including compartments for initial metabolism of vitamin A, plasma retinol, and extravascular vitamin A storage. The disposal rate was 6.7 ± 3.1 μmol/d, fractional catabolic rate was 6.0% ± 2.3%/d, and vitamin A stores were 123 ± 71 μmol. Relative β-carotene bioefficacy, based on the ratio of the areas under the fraction of dose curves calculated by WinSAAM, averaged 13.5% ± 6.02% (retinol activity equivalents = 7.7:1.0 μg). Interindividual variation in relative β-carotene bioefficacy was high (CV: 44%). Conclusions: Vitamin A kinetics in these young adults were best described by essentially the same model that had been previously developed by using data for older adults with higher vitamin A stores; differences in parameter values reflected differences in vitamin A status. Estimated β-carotene bioefficacy was relatively low but similar to previously reported estimates obtained by graphical methods. This trial was registered at the UK Clinical Research Network as UKCRN 7413.

A Retinol Isotope Dilution Equation Predicts Both Group and Individual Total Body Vitamin A Stores in Adults Based on Data from an Early Postdosing Blood Sample

Background: Retinol isotope dilution (RID) is used to determine vitamin A total body stores (TBS) after an oral dose of a vitamin A stable isotope. The generally accepted prediction equation proposed by Olson’s group in 1989 (Furr et al. Am J Clin Nutr 1989;49:713–6) includes factors related to dose absorption and retention, isotope equilibration in plasma compared with stores, catabolism during the mixing period, and the optimal time for measuring plasma isotope enrichment. Objectives: The objectives were 1) to develop a modified RID equation and identify an earlier sampling time for predicting TBS and 2) to improve prediction in individuals as well as groups. Methods: To develop a modified RID equation, we used results of model-based compartmental analysis [the Simulation, Analysis and Modeling software (WinSAAM version 3.0.8; http://www.WinSAAM.org)] of plasma [13C10]retinol kinetic data from 32 previously studied, healthy young adults of European ancestry who had moderate vitamin A intakes and who ingested 2.95 μmol [13C10]retinyl acetate. Results: We examined the time dependence of factors in the prediction equation related to absorption/retention (Fa) and isotope equilibration (S) and determined that 4 or 5 d postdosing was the optimal sampling time. TBS calculated by the equation TBS = Fa x S x (1/SAp), where SAp is plasma retinol specific activity (fraction of dose/μmol), were highly correlated with model-predicted TBS (r = 0.95 and 0.96 for 4 and 5 d, respectively; P < 0.001); predictions for individuals were also highly correlated (Rs = 0.94 and 0.94; P < 0.001). Conclusion: The equation TBS ≈ 0.5 × (1/SAp) accurately predicted vitamin A TBS in this group of 32 healthy young adults and its individual members with the use of data from 1 blood sample taken 4 d after isotope administration.

Effects of dietary fat saturation and triglyceride and cholesterol load on lipid absorption in the rat

Lipid absorption was studied in non-fasting thoracic lymph duct-cannulated rats receiving continuous intraduodenal infusions of oils varying in triglyceride fatty acid saturation (P/S = 4.6 or 0.2), triglyceride load (66 or 165 mumoles/h) and cholesterol load (0 and 3.5 mumoles/h at the low oil infusion rate, and 0 and 11 mumoles/h at the high infusion rate). Triglyceride absorption averaged 95% for rats receiving the low and high infusion rates of the polyunsaturated oil (groups defined as 10P and 25P, respectively), and 77% for those receiving the saturated oil (10S and 25S). Estimated average relative lymph lipoprotein size was significantly larger during infusion of the high vs. low triglyceride load, and of the P vs. S oil (25P greater than 25S greater than 10P greater than 10S). Lymph cholesterol flux during infusion of oil only averaged 3.2 mumoles/h for all groups and was not significantly influenced by dietary fat saturation or triglyceride load. The initial rapid rise in lymph cholesterol after addition of cholesterol (mass + tracer) to the oils was due primarily to endogenous cholesterol. Thus, absorption of dietary cholesterol resulted in an initial displacement of endogenous cholesterol into the lymph. However, in samples collected 18--24 h after addition of cholesterol to the oil, 87--107% of the increased lymph cholesterol flux was exogenous (labeled). This increased flux was not significantly influenced by fat saturation of the infused oil and averaged 5.6 mumoles/h for rats in the low oil/lower cholesterol infusion groups, and 9.5 mumoles/h for those in the higher infusion groups. The percent esterified cholesterol in lymph of rats in the low oil infusion groups was significantly higher in the S vs. P animals during both infusion of oil only and of oil + cholesterol. The observed effects of fat saturation on lymph triglyceride flux, cholesterol esterification and lipoprotein size may have important consequences for subsequent metabolism of absorptive lipoproteins, and for their ultimate effects on plasma lipid levels.

The Use of Model-Based Compartmental Analysis to Study Vitamin A Metabolism in a Non-Steady State

Over the past 20 years, we have collaborated with several laboratories in using mathematical modeling to describe and quantitate whole-body Vitamin A Metabolism in the rat. Steady state models have been developed for animals at different levels of vitamin A status (Green et al., 1985; Lewis et al., 1990; Green and Green, 1994a; Kelley and Green, 1998) and in response to other variables (Kelley et al., 1998; Jang et al., 2000; Kelley et al., 2000). Limited experimental and mathematical evidence (Green and Green, 1994b; Novotny et al., 1995; v Reinersdorff et al., 1998) suggests that there are many similarities in Vitamin A Metabolism between rats and humans. As discussed at the 5th Conference on Mathematical Modeling in Experimental Nutrition in 1994 (Green and Green, 1996), modeling studies have uniquely contributed to current understanding of whole-body Vitamin A Metabolism. In particular, interpretation of kinetic data has revealed previously unrecognized complexities in vitamin A dynamics that facilitate homeostatic control of plasma (and probably tissue) vitamin A levels.

A Simple Plasma Retinol Isotope Ratio Method for Estimating β-Carotene Relative Bioefficacy in Humans: Validation with the Use of Model-Based Compartmental Analysis

Background: Provitamin A carotenoids are an important source of dietary vitamin A for many populations. Thus, accurate and simple methods for estimating carotenoid bioefficacy are needed to evaluate the vitamin A value of test solutions and plant sources. β-Carotene bioefficacy is often estimated from the ratio of the areas under plasma isotope response curves after subjects ingest labeled β-carotene and a labeled retinyl acetate reference dose [isotope reference method (IRM)], but to our knowledge, the method has not yet been evaluated for accuracy.Objectives: Our objectives were to develop and test a physiologically based compartmental model that includes both absorptive and postabsorptive β-carotene bioconversion and to use the model to evaluate the accuracy of the IRM and a simple plasma retinol isotope ratio [(RIR), labeled β-carotene-derived retinol/labeled reference-dose-derived retinol in one plasma sample] for estimating relative bioefficacy.Methods: We used model-based compartmental analysis (Simulation, Analysis and Modeling software) to develop and apply a model that provided known values for β-carotene bioefficacy. Theoretical data for 10 subjects were generated by the model and used to determine bioefficacy by RIR and IRM; predictions were compared with known values. We also applied RIR and IRM to previously published data.Results: Plasma RIR accurately predicted β-carotene relative bioefficacy at 14 d or later. IRM also accurately predicted bioefficacy by 14 d, except that, when there was substantial postabsorptive bioconversion, IRM underestimated bioefficacy. Based on our model, 1-d predictions of relative bioefficacy include absorptive plus a portion of early postabsorptive conversion.Conclusion: The plasma RIR is a simple tracer method that accurately predicts β-carotene relative bioefficacy based on analysis of one blood sample obtained at ≥14 d after co-ingestion of labeled β-carotene and retinyl acetate. The method also provides information about the contributions of absorptive and postabsorptive conversion to total bioefficacy if an additional sample is taken at 1 d.

Effects of native and fractionated vegetable oils on lipid absorption in the rat

Abstract We investigated the hypothesis that the hypocholesterolemic activity of corn oil and other unsaturated vegetable oils is related to effects of their nontriglyceride components on intestinal lipid absorption. To test this hypothesis, corn, sunflowerseed (sun) and soybean oils (soy) were separated chromatographically into their pure triglycerides (TGs) and 4 nonTG fractions. For each native oil, a 7-membered series was prepared: native oil, pure TGs, pure TGs plus each of the 4 nonTG fractions, and pure TGs plus all 4 nonTG fractions. Three groups of lymph duct-cannulated rats received continuous (9–10 d) intraduodenal infusions of oils with added cholesterol (equivalent to a diet containing 20% fat +0.4% cholesterol) in a randomized Latin Square design. Each oil was infused for 18 h before lymph was sampled. There were no significant differences in chylomicron TG, phospholipid (PL) and cholesterol (total, unesterified or esterified) transfer rates (μmole/h) or in relative chylomicron size (molar ratio of PL/TG) within any of the 3 series. Recovery of TG and cholesterol in chylomicrons, relative to the rates of infusion, averaged 91 and 99% (corn), 77 and 72% (sun) and 90 and 79% (soy). Our results indicate that, at the level present in these 3 oils, the nonTG components do not significantly influence chylomicron lipid composition or size, or the rate of intestinal secretion of lipid components.