Beruflich Dokumente
Kultur Dokumente
1
Posgrado en Ciencias de la Salud (PCS), Universidad de Sonora, Hermosillo, Sonora, Mexico; 2 Department of Nutritional Sciences, The
Pennsylvania State University, University Park, PA, USA; 3 Department of Nutrition, Research Center for Food and Development,
ABSTRACT
Background: Retinol isotope dilution (RID) and model-based compartmental analysis are recognized techniques for
assessing vitamin A (VA) status. Recent studies have shown that RID predictions of VA total body stores (TBS) can be
improved by using modeling and that VA kinetics and TBS in children can be effectively studied by applying population
modeling (“super-child” approach) to a composite data set.
Objectives: The objectives were to model whole-body retinol kinetics and predict VA TBS in a group of Mexican
preschoolers using the super-child approach and to use model predictions of RID coefficients to estimate TBS by RID in
individuals.
Methods: Twenty-four healthy Mexican children (aged 3–6 y) received an oral dose (2.96 μmol) of [13 C10 ]retinyl acetate
in corn oil. Blood samples were collected from 8 h to 21 d after dosing, with each child sampled at 4 d and at 1 other
time. Composite data for plasma labeled retinol compared with time were analyzed using a 6-component model to obtain
group retinol kinetic parameters and pool sizes. Model-predicted TBS was compared with mean RID predictions at 4 d;
RID estimates at 4 d were compared with those calculated at 7–21 d.
Results: Model-predicted TBS was 1097 μmol, equivalent to ∼2.4 y-worth of VA; using model-derived coefficients,
group mean RID-predicted TBS was 1096 μmol (IQR: 836–1492 μmol). TBS at 4 d compared with a later time was
similar (P = 0.33). The model predicted that retinol spent 1.5 h in plasma during each transit and recycled to plasma 13
times before utilization.
Conclusions: The super-child modeling approach provides information on whole-body VA kinetics and can be used
with RID to estimate TBS at any time between 4 and 21 d postdose. The high TBS predicted for these children suggests
positive VA balance, likely due to large-dose VA supplements, and warrants further investigation. J Nutr 2020;150:1644–
1651.
Keywords: children, mathematical modeling, model-based compartmental analysis, retinol isotope dilution, vitamin
A kinetics, vitamin A metabolism, WinSAAM
Copyright C The Author(s) on behalf of the American Society for Nutrition 2020.
Manuscript received November 23, 2019. Initial review completed January 8, 2020. Revision accepted February 12, 2020.
1644 First published online March 5, 2020; doi: https://doi.org/10.1093/jn/nxaa048.
Because it is generally not feasible to collect serial blood comparisons of TBS at 4 d when all children were sampled to
samples in children, a “super-child” approach has been values predicted at other times.
developed (10, 11) that facilitates retinol kinetic studies in this
age group. For this technique, 2 blood samples are collected
from each child and data from all subjects are pooled into
a composite data set for modeling, providing information on Methods
retinol kinetics, TBS, and the RID coefficients for the group 13
C-labeled VA and oral dose
(11). This approach has been applied in infants from Mexico All-trans-8,9,10,11,12,13,14,15,19,20-[13 C10 ]retinyl acetate (RAc)
(10); results showed that mean RID predictions calculated at was synthesized by Cambridge Isotope Laboratories to a chemical
3–6 d agreed well with model-predicted TBS. Recently, an purity ≥98%. For the oral dose, aliquots of the isotope (2.96 μmol)
improved super-child design, in which all children were sampled were dissolved in 200 μL corn oil and stored at −80◦ C until
at a common time (4 d), was applied in infants and children administered to participating children. The daily VA intake
recommendation for this age group is 300–400 μg retinol activity
from Bangladesh, Guatemala, and the Philippines (12); group
equivalents (RAE) per day (see below), and although the VA dose
predictions of TBS by RID are expected to agree with the model. provided here is larger than that (∼2 times), it is considered a small,
In addition, model-predicted group-specific RID coefficients can physiological oral dose of stable isotope-labeled VA that does not
be used to estimate TBS in individual subjects (8), whereas
FDp = [enrichment × retinol concentration (μmol/L)]
×plasma volume(L) /dose(μmol) (1)
where plasma volume was estimated for each child using the average
sex-specific hematocrit for this age group and an estimate of blood
volume (liters) calculated using the nomograms of Linderkamp et al.
(30) based on age, height, and body weight.
To generate a composite (super-child) data set, we calculated the
geometric mean FDp at each time. Data for geometric mean FDp were
plotted over time (8 h to 21 d), and model-based compartmental analysis
was applied using the Windows version of the Simulation, Analysis and
Modeling software (WinSAAM version 3.3.0) (31) and a previously
published (12) 6-component model describing whole-body VA kinetics
in children (Figure 1A; see figure legend for a description of the model
components). Model parameters include fractional transfer coefficients
[L(I,J)s, or the fraction of retinol in compartment J transferred to FIGURE 1 Conceptual compartmental model for retinol kinetics
compartment I each day] and delay times [DT(I), or the time (d) that in children (A), with results for Mexican preschoolers (B). Circles
retinol spends in compartment I] for components 3 and 8. Absorption represent compartments, the rectangles are delay components,
efficiency for preformed VA was fixed at 80% (2), and an estimate of and interconnectivities between compartments [L(I,J)s] are fractional
VA intake was included as weighted data (32) to improve confidence in transfer coefficients or the fraction of retinol transferred to com-
model predictions such as VA TBS and disposal rate (33). Because intake partment I from compartment J each day; DT(I)s are delay times
estimates included both preformed VA and provitamin A carotenoids (or the time spent in delay element I). DT(3) represents the site of
[as RAE based on the Institute of Medicine conversion factors (27)], we input of the orally administered [13 C10 ]retinyl acetate dose (∗) and
adjusted the contribution of provitamin A to its equivalent as preformed dietary VA [U(3)]. Component 3 and compartment 4 represent the
VA. Specifically, for each child, adjusted VA intake was calculated processes of VA digestion and absorption as well as chylomicron
as preformed VA + (provitamin A/0.8), assuming 80% absorption production and catabolism until clearance of chylomicron remnants by
efficiency for preformed VA; then the geometric mean of these values liver hepatocytes (compartment 4). Retinol in compartment 4 is bound
was used as a modeling input. For modeling, geometric mean FDp data to retinol-binding protein and secreted into plasma (compartment 5),
were weighted using a fractional SD (FSD) of 0.05 at all times except the site of sampling (triangle). Plasma retinol exchanges with VA in
4 d when the weight was increased to 0.025 in light of the larger 2 extravascular pools, 1 smaller (compartment 7) and 1 larger storage
number of children sampled at that time; the adjusted VA intake for pool (compartment 6), which is also a site of irreversible loss from
the group was weighted using an FSD of 0.05. The value for L(8,5) the system; the sum of VA in compartments 6 and 7 corresponds to
(Figure 1A) was calculated, as detailed in Ford et al. (12), to reflect TBS. A portion of the plasma retinol pool is also transferred irreversibly
50% of the disposal rate and fixed in the model; DT(8) was also to component 8, which represents uptake by tissues for immediate
fixed, at 75 min. Data were fit to the model shown in Figure 1A by utilization following a delay of 75 min (12). Panel B shows final model
systematically/manually adjusting the value for each parameter until results including L(I,J)s, R(I,J)s [i.e., transfer rates calculated as L(I,J) ×
the model predictions agreed with the observed FDp mop data. Then, M(J)], and DT(I)s; compartment masses [M(I)]s are indicated inside the
weighted nonlinear regression analysis was performed in WinSAAM to corresponding compartment. Absorption efficiency was fixed at 80%
obtain final parameter values and their statistical uncertainties. Finally, and the adjusted dietary VA intake (1.59 μmol RAE/d) was included
plasma retinol pool size [M(5), μmol; Figure 1A] was calculated for each in the model as weighted data. VA disposal rate = R(10,6) + R(8,5).
sample as plasma retinol concentration (μmol per liter) × estimated All 7 adjustable parameters were well identified (FSD < 0.5) (31); the
plasma volume (liters); the geometric mean was calculated and used in mean FSD was 0.187 (range: 0.0460–0.356). FSD; fractional standard
the steady state model solution in WinSAAM to predict the mass of deviation; RAc, retinyl acetate; RAE, retinol activity equivalent; TBS,
VA in other compartments [M(I), μmol], including TBS [M(6) + M(7)], total body stores; VA, vitamin A.
and VA transfer rates [R(I,J)s, or μmol in compartment J transferred to
compartment I each day], including VA disposal rate and intake [U(3)].
Characteristic Value
Age, y 4.80 ± 0.67
Weight, kg 19.98 ± 6.04
Height, cm 109 ± 6.22
Weight-for-age z-score 0.55 ± 1.52
Height-for-age z-score 0.04 ± 0.95
BMI-for-age z-score 0.75 ± 1.73
Whole blood hemoglobin,2 g/L 138 ± 12.2
Serum C-reactive protein, mg/L <6
Serum retinol, μmol/L 1.34 ± 0.34
1
Values are means ± SD for Mexican children (n = 24).
2
n = 23.