The Journal of Nutrition Vol. 128 No. 10 October 1998, pp. 1767-1773

Plasma Retinol Is a Major Determinant of Vitamin A Utilization in Rats^1,2,3

Sean K. Kelley and Michael H. Green⁴

Nutrition Department and Graduate Physiology Program, The Pennsylvania State University, University Park, PA 16802

	ABSTRACT

Abstract Introduction Methods Results & Discussion References

We studied relationships among vitamin A intake, liver levels of vitamin A, plasma retinol concentrations and the irreversible utilization of vitamin A. To supplement existing data, we first used model-based compartmental analysis to determine vitamin A utilization and other kinetic parameters in male Sprague-Dawley rats that had adequate liver vitamin A stores (~9000 nmol) and were fed a diet containing low levels of vitamin A. Plasma retinol kinetics were monitored for 43 d after administration of [³H]retinol-labeled plasma to rats consuming ~23 (Group 1, n = 6) or ~4.2 (Group 2, n = 6) nmol vitamin A/d. Data for plasma tracer vs. time and for tracer lost irreversibly by the end of the experiment, were fit to a three-compartment model in which plasma retinol exchanges with vitamin A in two kinetically distinct extravascular compartments. Irreversible utilization rates (~41 nmol/d) were similar to those for rats that are in vitamin A balance, suggesting that, when liver vitamin A stores are adequate, utilization rate is not decreased to compensate for a low vitamin A intake. Multiple linear regression analysis was then used to relate these and previously collected data (total, 62 rats) on vitamin A intake (4.2-49 nmol/d), plasma retinol concentration (1.4-2.5 µmol/L) and liver vitamin A level (1.2-11,000 nmol) to vitamin A utilization (disposal rate, 4.2-68 nmol/d). A significant relationship (R²_(adj) = 0.93) was found for the equation [disposal rate (nmol/d) = 0.720 (nmol/d) + 0.844 (d¹)·(plasma retinol; nmol) + 0.00139 (d¹)·(liver vitamin A; nmol) + 0.220·(vitamin A intake; nmol/d)]. Plasma retinol accounted for 92% of the variability in disposal rate (vs. 5% for liver vitamin A and 3% for vitamin A intake). We conclude that plasma retinol is a main determinant of the irreversible utilization of vitamin A in rats with low to moderate vitamin A intake.

	INTRODUCTION

Abstract Introduction Methods Results & Discussion References

Because the rate of utilization of vitamin A influences dietary requirements for the vitamin, it is important to define and quantify factors that influence vitamin A utilization. In an early study, Varma and Beaton (1972) orally administered radioactive vitamin A and measured the appearance of label in urine and feces. They showed that, as vitamin A intake and liver levels increased, so did the irreversible utilization of vitamin A. Later, Hicks et al. (1984) demonstrated that, when radioactive retinyl acetate was administered orally, the amount of radioactive metabolites found in bile was constant when liver vitamin A ranged from 21 to 1610 nmol, but increased when hepatic vitamin A mass was higher. More recently, we showed (Green and Green 1994) that, in rats with low to marginal liver vitamin A levels (<10 to ~500 nmol), irreversible utilization of vitamin A was influenced not only by liver vitamin A levels, but also by plasma retinol concentration and dietary vitamin A intake (25 or 50 nmol/d). At lower levels of vitamin A intake (7 nmol/d) and liver vitamin A (<3.5 nmol), vitamin A utilization was dramatically lower (Lewis et al. 1990).

To quantify the influence of plasma retinol levels, liver vitamin A content and vitamin A intake on vitamin A utilization, we first used model-based compartmental analysis to determine vitamin A utilization rate and other kinetic parameters of vitamin A metabolism in rats with liver vitamin A levels that were higher than those in previously studied rats (~9000 nmol) during a period of low vitamin A intake (4.2 or 23 nmol/d). Our hypothesis was that, if liver vitamin A can maintain normal plasma retinol levels, irreversible utilization will be maintained at normal levels in spite of depleting liver stores of vitamin A. Then we applied regression analysis to these and previously collected data to relate information on vitamin A intake, plasma retinol, liver vitamin A levels and vitamin A utilization rate in rats.

	MATERIALS AND METHODS

Abstract Introduction Methods Results & Discussion References

Animals and diets.  Weanling male Sprague-Dawley rats were obtained from Harlan Sprague Dawley, Indianapolis, IN. Rats were housed in hanging, stainless steel cages in an environmentally controlled room (22-23°C, 60% relative humidity) with a 12-h light cycle (lights on from 0700 to 1900 h). Tap water and nutritionally adequate purified diets (Duncan et al. 1993) were available at all times. The diets contained (g/kg) casein (200), dextrose (300), cornstarch (300), corn oil (100), cellulose (50), AIN 76 mineral mix (35; catalogue no. 170915, Teklad, Madison, WI), vitamin A-free vitamin mix (10; catalogue no. TD86502, Teklad), choline bitartrate (2.0), DL-methionine (3.0), ethoxyquin (0.01) and variable amounts of vitamin A as retinyl palmitate (see below). Body weights and food intake were recorded regularly. Animal procedures were approved by The Pennsylvania State University's Institutional Animal Care and Use Committee.

To increase liver vitamin A levels to ~9000 nmol (~1000 nmol/g), weanling rats (n = 18) were fed vitamin A-containing diets for 44 d until growth rates were low. On the basis of body weight, rats were then assigned to one of two groups in preparation for an in vivo kinetic study. Because we were interested in the effects of low vitamin A intakes on vitamin A utilization in rats with adequate liver vitamin A content, we chose dietary vitamin A levels that were substantially lower than the 50 nmol/d that we had found maintained vitamin A balance in vitamin A-sufficient rats studied previously (Green and Green 1994). Thus rats were provided diets containing either 0.86 nmol vitamin A/g (Group 1) or 0.17 nmol/g (Group 2) for 9 d before initiation of kinetic studies and then during the subsequent 43-d study. Based on the amount of vitamin A added to the diets and the estimated food consumption (g food offered  g food left in food cup), rats in Group 1 consumed 22.8 ± 1.6 nmol vitamin A/d and those in Group 2, 4.2 ± 0.54 nmol/d.

Preparation of retinol-labeled plasma.  All procedures involving vitamin A were carried out under shaded natural light or yellow light. Weanling rats to be used as donors of [³H]retinol-labeled plasma (n = 2) were fed a vitamin A-free purified diet (Duncan et al. 1993) for 44 d to deplete liver vitamin A reserves. As described in detail elsewhere (Green and Green 1990a), a dispersion (~0.5 mL containing 0.1 TBq) of [11,12(n)-³H]retinol (Amersham, Arlington Heights, IL; specific radioactivity, 1.47 TBq/mmol) in Tween 40 (Sigma Chemical, St. Louis, MO) was administered intravenously. Approximately 100 min later, blood containing [³H]retinol in its normal physiologic transport complex (Green et al. 1985) was harvested from the abdominal aorta into syringes containing Na₂EDTA. Plasma was separated, pooled in a sterile vial and refrigerated at 4°C under an atmosphere of nitrogen. During each of the next 2 d, half of the [³H]retinol-labeled plasma was gently warmed to ~37°C and injected into three recipients from each dietary group (see below).

In vivo kinetic study.  The biokinetic behavior of vitamin A was studied as previously described (Adams et al. 1995, Green and Green 1990a). Briefly, recipient rats (n = 12) were anesthetized with methoxyflurane (Pitman-Moore, Washington Crossing, NJ) between 0800 and 0900 h when the animals were in a postprandial state. Approximately 0.55 g of [³H]retinol-labeled plasma containing ~219 kBq of ³H was injected into an exposed external jugular vein and anesthesia was removed. Serial blood samples (~0.25 mL/sample; n = 24 per rat) were collected from a caudal vein into microcentrifuge tubes containing Na₂EDTA from 10 min to 43 d after dose administration. Plasma aliquots were frozen under a nitrogen atmosphere at 16°C for later analysis of radioactivity and, for some samples (n = 8 per rat), retinol.

Forty-three days after dose administration, rats were anesthetized with ketamine HCl/xylazine (100:20 mg/kg body weight; Aveco, Fort Dodge, IA and Mobay, Shawnee, KS, respectively). The whole body was perfused from the left cardiac ventricle to the right auricle with ~200 mL of Hanks' balanced salt solution (pH 7.4). Livers were excised, weighed, frozen under nitrogen, lyophilized and stored at 16°C under nitrogen for later analysis of tritium and vitamin A. Carcasses were also weighed and frozen for subsequent analysis of tritium.

Three days after administration of labeled plasma to the first group of recipient rats, the remaining 3 rats/group were killed; livers were obtained and analyzed for vitamin A. These data were used to estimate initial liver vitamin A levels for rats in the kinetic study.

Plasma and tissue analyses.  For analysis of tritium, aliquots of the injected dose and serial plasma samples were extracted using a modification (Green and Green 1990a) of the procedure of Thompson et al. (1971), aliquots of freeze-dried livers were saponified and extracted (Green et al., 1985, Thompson et al. 1971) and carcasses were ground and extracted using a modification (Adams et al. 1995) of the method of Hara and Radin (1978). Aliquots of these lipid extracts were analyzed by liquid scintillation spectrometry (Model 3801; Beckman Instruments, Irvine, CA) using Ecoscint O (National Diagnostics, Manville, NJ) as scintillation solution. Samples were counted twice to a final 2-sigma error of 1.0%; sample counts were automatically corrected for background and converted to becquerels by using an external standard method.

To determine vitamin A in selected plasma samples and in livers, plasma samples and extracts from saponified livers (see above) were spiked with retinyl acetate (internal standard). As described earlier (Duncan et al. 1993), retinol and retinyl acetate were separated by reverse-phase HPLC using a 5-µm C-18 Resolve column (Millipore, Milford, MA) and a mobile phase of methanol/water (90:10, v/v) at a flow rate of 1.0 mL/min. Retinoids were detected at 340 nm and quantified by using a Chromatopac C-R3A integrator (Shimadzu, Kyoto, Japan) and an internal standard method.

Kinetic analyses.  To obtain a value for the fraction of the injected dose remaining in plasma at each time, tracer concentration (Bq/mL) in each plasma sample was normalized to the plasma tracer concentration at time zero [Bq injected/estimated mean plasma volume during the turnover study, where plasma volume was calculated as body weight (g)·0.038 mL of plasma/g body weight (Wang 1959)]. We also calculated fraction of the injected dose present in the liver and carcass at the conclusion of the turnover study, and the fraction of the dose that was irreversibly lost over the course of the study [1  (fraction of dose in liver + carcass + plasma at the time of killing)]. A fractional standard deviation of 0.05 was assigned as a weighting factor to each datum.

Then, using an approach similar to that described by Green and Green (1994) and Adams et al. (1995), we used model-based compartmental analysis (Green and Green 1990a and 1990b) to characterize whole-body vitamin A dynamics as viewed from the plasma space. In this approach, processes with similar kinetics are lumped in the same compartment. In trying to determine the simplest model that would provide an adequate fit to data for plasma and irreversible loss, multiple models were tested (see Results and Discussion). Model complexity was increased only when it resulted in a significant improvement in the weighted sums of squares as determined by an F-statistic (Landaw and DiStefano 1984).

The three-compartment, mammillary model shown in Figure 1 provided a good fit to the observed tracer data for plasma and irreversible loss. In the proposed model, dietary input [U(1)] was modeled into compartment 1 (plasma), which was also the site of tracer introduction to the system. The model indicates that retinol in compartment 1 exchanges with vitamin A in two extravascular compartments; one of these (compartment 3) is larger and more slowly turning over than the other. We hypothesize that compartment 3 corresponds mainly to vitamin A stored in the liver and extrahepatic tissues, whereas compartment 2, the site of irreversible loss, may be retinol in interstitial fluid, retinol filtered by the kidneys and more quickly turning over intracellular retinol pools. The interconnectivities indicated in the model [L(I,J)⁵; see below] were determined for each rat by weighted, nonlinear regression analysis using the conversational version (CONSAM 31; Berman et al. 1983) of the Simulation, Analysis and Modeling computer program (SAAM; Berman and Weiss, 1978).

View larger version (8K): [in this window] [in a new window]   Fig 1. Proposed compartmental model for vitamin A turnover. Compartments are represented by circles. Shown with each vector is its associated fractional transfer coefficient [L(I,J)]. The asterisk denotes the site of introduction of [3H]retinol-labeled plasma; the triangle denotes the site of sampling. U(1) corresponds to input of dietary vitamin A.

The following model parameters were calculated for each rat; see Green and Green (1990a and 1990b) for more details on nomenclature and model-based calculations. Instantaneous fractional transfer coefficients [L(I,J); d¹] are defined as the fraction of vitamin A in compartment J transferred to compartment I each day. Based on the model-predicated L(I,J), we estimated the plasma fractional catabolic rate [FCR(I,J) or FCR_p; d¹], defined as the fraction of vitamin A in compartment I that leaves irreversibly each day after entering the system via compartment J. Mean transit time [(I); d] is the average length of time that a molecule of retinol that reaches compartment I remains in compartment I before it leaves that compartment reversibly or irreversibly. The mean residence time [(I,J); d] is defined as the mean of the distribution of times that an average molecule of retinol spends in compartment I before leaving compartment I irreversibly after entering the system via compartment J. (SYS), or the system mean residence time, is the average time that a molecule of retinol entering the system via compartment J spends in the system before irreversible loss and is computed as the sum of (I,J). Recycling number [(I)] is the average number of times a molecule of retinol recycles to compartment I before leaving compartment I irreversibly; it is calculated as [(I,J)/(I)]  1. Recycling time [(1); d], the average length of time it takes for a molecule of retinol to return to compartment 1 after it leaves compartment 1, is computed as [(SYS)  (1,1)]/(1).

In addition, we used the estimated plasma retinol pool size [M_p; nmol] in a steady-state solution to calculate transfer rates [R(I,J); nmol/d], defined as the rate of transfer of material to compartment I from compartment J. Using this solution, the system disposal rate (DR) in the model presented here (Fig. 1) is equal to M(2)·L(0,2).

Statistics.  Descriptive data, fractional transfer coefficients, kinetic parameters and model-predicted compartment masses are presented as the group arithmetic means ± SD. Means for kinetic parameters and compartment masses were calculated from individual animal models. Data were compared statistically by using Student's unpaired t test (Ryan et al. 1985) with an  level of 0.05.

To develop a predictive equation for disposal rate, multiple linear regression analysis (Ryan et al. 1985) was applied to data on vitamin A intake, liver vitamin A levels, plasma retinol pool size (predictor variables) and vitamin A utilization rates (response variable). The contribution of each of the predictor variables was determined by independent regression vs. utilization rate. Then, predictor variables, in order of their ability to predict disposal rate, were sequentially included in the multiple regression analysis. Data were given equal weight.

	RESULTS AND DISCUSSION

Abstract Introduction Methods Results & Discussion References

Animal outcome.  Body weights of rats in Groups 1 and 2 averaged 346 ± 5 g (mean ± SD, n = 6) and 345 ± 8 g, respectively, at the time of administration of [³H]retinol-labeled plasma (P > 0.05). Forty-three days later, body weights were 426 ± 8 and 430 ± 11 g (P > 0.05). During the turnover study, plasma retinol concentrations were not significantly different between groups, ranging from 1.82 to 2.22 µmol/L (mean, 2.06 ± 0.057 µmol/L) in Group 1 and from 1.95 to 2.32 µmol/L (mean, 2.15 ± 0.060 µmol/L) in Group 2 (P > 0.05).

At the beginning of the turnover study, three rats from each dietary group were killed to obtain estimates of initial liver vitamin A levels. Levels were not significantly different between groups, and averaged 9249 ± 505 nmol (Group 1) and 9011 ± 321 nmol (Group 2) (P > 0.05). There was also no significant difference in liver vitamin A content of rats killed at the end of the study [9700 ± 341 nmol (Group 1) vs. 8853 ± 395 nmol (Group 2); n = 6/group] (P > 0.05). We were surprised by the high mean terminal value obtained for Group 1 rats because we had chosen a vitamin A intake during the 43-d kinetic study (~23 nmol/d) that we assumed would induce negative vitamin A balance. In spite of the large individual variation in liver vitamin A levels, we conclude that all rats were in a negative vitamin A balance (~23 nmol/d, Group 1; ~36 nmol/d, Group 2) during the kinetic study on the basis of model-predicted disposal rates (see below) and estimated vitamin A intakes.

Kinetic data.  Plasma tracer response curves and data on irreversible loss for one representative animal from each group are shown in Figure 2. The curves for rats in both groups were characterized by a steep initial slope followed by a sharp bend in the curve beginning 3-4 d after dose administration. Subsequently, plasma tracer response curves entered a shallow terminal slope beginning at d 5-7. Based on previous work (Green and Green 1994), we hypothesize that the three distinguishing features of the plasma tracer disappearance curves (the steepness of the initial slope, the time and shape of the bend and the magnitude of the terminal slope) are related to the following: 1) the transfer and exchange of plasma [³H]retinol with vitamin A in both the quickly and slowly turning over extravascular vitamin A pools; 2) the exchange of vitamin A between plasma and the more slowly turning over extravascular vitamin A pool, i.e., the larger these extravascular pools, the sharper the bend; and 3) the fractional rate of loss of vitamin A from the system. The shapes of the plasma disappearance curves for rats in this experiment are similar to those seen in a previous kinetic study in which rats had much lower liver vitamin A levels (~500 nmol) and consumed either 50 (positive vitamin A balance) or 25 nmol vitamin A/d (negative vitamin A balance) (Green and Green, 1994). However, in this study, the fraction of the injected dose in plasma ~5 d after dose administration was 80% lower than in the previously studied rats. This down-shifting in the curves can be attributed to a dilution of label in the larger, very slowly turning over extravascular vitamin A pools of rats in this study. There was no observable difference in the terminal slopes for rats in Group 1 vs. 2. 

View larger version (17K): [in this window] [in a new window]   Fig 2. Plasma [3H]retinol kinetics in rats fed ~23 nmol vitamin A/d (Group 1) or ~4.2 nmol/d (Group 2). Shown are observed data (symbols) and model-predicted values (lines) for fraction of administered dose remaining in plasma or fraction of administered dose irreversibly lost by the end of the study vs. time after injection of [3H]retinol-labeled plasma for one representative rat from each group.

Compartmental model and model-derived kinetic parameters.  As illustrated by the model-predicted fit to the data (solid and dashed lines, Fig. 2), the proposed model (Fig. 1) provided a good fit for the current data. Based on minimizing the residual weighted sum of squares and on the fractional standard deviations associated with the model parameters, a good fit to tracer data for plasma and irreversible loss was obtained when input to the system (dietary vitamin A) was modeled into compartment 1 and output was from compartment 2. Although several previous studies using this approach have required model output from the large slowly turning over compartment 3 (Adams et al. 1995, Green and Green 1994), here output from compartment 2 was needed to fit the terminal portion of the plasma disappearance curve and to predict irreversible loss data accurately. We speculate that the need for different sites of output when using a simple compartmental model may be due to differences in vitamin A status in the various experiments. When using a simple model to describe whole-body vitamin A metabolism, the kinetic behavior of vitamin A in the more slowly turning over vitamin A pools (e.g., tissue retinyl esters) reflects the "lumping" of vitamin A from several body sites into one or two compartments. As vitamin A status and tissue retinyl ester stores increase, the kinetic behavior of vitamin A in compartment 3 is increasingly dominated by the large amounts of very slowly turning over retinyl esters. Consequently, the turnover of moderately dynamic vitamin A (perhaps the site of vitamin A output) becomes lumped with that of the most rapidly turning over extravascular pools of vitamin A (vitamin A in interstitial fluid and vitamin A filtered by the kidney) in compartment 2. More extensive tissue analysis makes it possible to identify distinct kinetic pools of vitamin A within tissues as was done previously in other experiments in this laboratory (Green et al. 1993, Lewis et al. 1990).

Regarding compartment 1 as the site of dietary vitamin A input, previous kinetic experiments suggest that after hepatic processing, the vast majority of dietary vitamin A circulates through the plasma (compartment 1) at least once bound to retinol-binding protein before being stored or degraded (Adams et al. 1995, Green and Green 1994). For the current data, when input was modeled into compartment 1, the model-predicted vitamin A mass in compartment 3 was 50% less than the analytically determined values for liver vitamin A (9700 nmol in Group 1 and 8853 nmol in Group 2). We speculate that this underprediction by the model may be due to a sequestering of some of the vitamin in extremely slowly turning over storage pools. Vitamin A in such pools might not be sufficiently kinetically active to equilibrate with plasma retinol during the time frame of this study and would thus not be kinetically "visible" from the plasma compartment. The majority of this sequestered vitamin A is most likely located in the liver stellate cells.

To further support the compartmental model's underprediction of vitamin A mass in compartment 3, we used an empirical model to estimate the total traced vitamin A mass as viewed from the plasma. First, we fit plasma tracer data to a three-component exponential equation of the form FD_t = I_ie^g_it, where FD_t is the fraction of the injected dose in plasma at a particular time (t), I_i are exponential constants and g_i are exponential coefficients. Then we calculated the total traced mass as the product of DR_p and MST_p, where MST_p is the mean sojourn time and, in this approach, is the same as system residence time; see Green and Green (1990a) for further details on the calculation of MST. Similar to our compartmental model, the empirical model also underpredicted liver vitamin A mass.

Using the group average model-predicted fractional transfer coefficients (Table 1), we calculated the kinetic parameters presented in Table 2. The model predicts that the average vitamin A molecule will spend 4.5-5.5 mo in these rats [(SYS)] before irreversible utilization. Compartment transit times for compartment 1 [(1)] and compartment 2 [(2)] were similar between groups, and were slightly higher than those calculated in earlier studies of rats with lower vitamin A status (Adams et al. 1995, Green and Green, 1994). However, in this study, the mean transit time for vitamin A in compartment 3 [(3)] was 6-10 times longer than in the previous studies; this is most likely related to the presence of, and the delivery of the label to large, very slowly turning over tissue vitamin A pools.

Model-predicted compartment masses and transfer rates are presented in Table 3. The estimated plasma retinol pool size (compartment 1) was ~30 nmol for rats in both groups. This is similar to what has been observed in previous studies using rats with lower vitamin A status (Adams et al. 1995, Green and Green 1994). Vitamin A mass in compartment 2, the rapidly turning over extravascular pool, was also similar to values seen in other experiments and was predicted to be <1% of the mass in compartment 3. These results further support the idea that both the vitamin A mass in compartment 1 (plasma) and the model-predicted mass of vitamin A in compartment 2 remain relatively constant over a wide range of vitamin A status (i.e., as long as there is either sufficient vitamin A intake or liver vitamin A stores to maintain these pools).

The model-predicted transfer rates [R(I,J)s; Table 3] were not significantly different between groups. Estimated vitamin A utilization or disposal rates (DR) and the plasma fractional catabolic rate (FCR_p) were similar to our earlier experiments with vitamin A-sufficient rats (liver vitamin A, 540-749 nmol; DR 29-42 nmol/d; FCR_p, 1.0-1.33 d¹) (Adams et al. 1995, Green and Green, 1994). However, a higher FCR_p (~2.0 d¹) was calculated by others (Lewis et al. 1994) in an experiment using rats with liver vitamin A content between ~800 and 1300 nmol. Although rats in this study were consuming amounts of vitamin A that would presumably result in negative vitamin A balance, they maintained an irreversible utilization rate that was similar to rates we have previously observed for rats with much lower vitamin A stores that were in a slightly positive vitamin A balance (Green and Green 1994). Assuming an absorption efficiency of 75% (Allen et al. 1994), then Group 1 rats were in a negative vitamin A balance of 23 nmol/d and those in Group 2, 36 nmol/d. It is interesting to note that the difference in estimated vitamin A balance between Groups 1 and 2 (13 nmol/d) is similar to the difference in vitamin A intake between groups (14 nmol/d), suggesting a vitamin A disposal rate that was unaffected by intake in those rats.

Taken together, our results suggest that there is an obligatory rate of vitamin A utilization in vitamin A-sufficient rats. As discussed previously (Green et al. 1994, Hicks et al. 1984, Varma and Beaton, 1972), vitamin A utilization (irreversible loss) is likely due to both functional and nonfunctional utilization of vitamin A (degradation). In our previous experiments, we found that rats require vitamin A intakes of ~50 nmol/d (of which ~75% is absorbed) to maintain vitamin A balance. However, rats that have very low vitamin A status (intakes, ~7 nmol/d; liver vitamin A, ~2 nmol) use only ~5.6 nmol/d, appear to grow normally and do not exhibit any signs of vitamin A deficiency (Lewis et al. 1990). Therefore we hypothesize that rats require 5.6 nmol vitamin A/d or less for functional purposes (maintenance of processes supported by vitamin A), and that the remaining vitamin A utilization is due to nonfunctional degradation. Of course, the amount of vitamin A required for functional purposes may increase at other times due to alterations in physiologic state (e.g., stress, infection or reproduction).

In summary, these data support the hypothesis that vitamin A utilization is not decreased to prevent depletion of vitamin A in response to a low vitamin A intake by rats with high vitamin A status. Our results indicate that, even in a state of low vitamin A intake, if liver and plasma vitamin A levels are sufficient, a normal vitamin A disposal rate will be maintained. It would be interesting in future experiments to further increase dietary input for rats with initial liver A levels similar to those used here and to determine whether vitamin A disposal would increase.

Determinants of vitamin A disposal rate.  To further evaluate effects of vitamin A intake, liver vitamin A levels and plasma retinol concentrations on vitamin A disposal rate, we examined data from this study together with similar data from 50 rats studied previously (Adams et al. 1995, Green et al. 1987, Green and Green 1994, Lewis et al. 1990). Collectively, these data from male Sprague-Dawley rats (n = 62) include a wide range of vitamin A intakes (4.2-49 nmol/d), plasma retinol concentrations (1.4-2.47 µmol/L), liver vitamin A levels (1.2-10,929 nmol), and vitamin A disposal rates (4.2-68.5 nmol/d). It should be noted that the estimates of disposal rate were obtained from models that varied in both structure and complexity. Using the available data, we applied multiple linear regression analysis to obtain the following equation: A significant relationship was observed between disposal rate and the three predictor variables (R²_(adj) = 0.93). Plasma retinol was the best predictor of disposal rate and accounted for 92% of the variability in DR. Liver vitamin A and vitamin A intake accounted for 5 and 3% of the variability, respectively. Regression plots presented in Figures 3A-C show the contribution of each of three variables in predicting disposal rate. Figure 3D compares disposal rates predicted by the regression equation with our model-predicted disposal rates. For liver (Fig. 3B), the line was generated by nonlinear regression analysis.

View larger version (15K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   View larger version (15K): [in this window] [in a new window]   View larger version (17K): [in this window] [in a new window]   Fig 3. Results from multiple linear regression analysis of model-predicted disposal rates (nmol/d) for 62 rats with varying plasma vitamin A (nmol), liver vitamin A (nmol) and vitamin A intakes (nmol/d). (Panel A) Plasma retinol vs. model-predicted disposal rate (R2 = 0.857); (panel B) liver vitamin A vs. model-predicted disposal rate (R2 = 0.325); (panel C) vitamin A intake vs. model-predicted disposal rate (R2 = 0.257); (panel D) vitamin A disposal rate predicted by the regression equation vs. model-predicted disposal rate (R2 = 0.932). Linear regression analysis was used to generate lines presented in panels A, C and D, whereas nonlinear least-squares regression analysis was used to generate the line in panel B.

The importance of plasma retinol in the predictive equation suggests that, as plasma retinol increases, so does vitamin A utilization. This is also supported by the linear relationship between vitamin A disposal rate and increased plasma vitamin A (Fig. 3A). The plot for liver vitamin A and disposal rate (Fig. 3B) revealed a steady increase in the rate of vitamin A utilization as liver vitamin A increases from ~2 nmol to ~500 nmol. Then, disposal rate remained relatively constant as liver vitamin A increased from ~500 to ~11,000 nmol, indicating that in these rats, vitamin A utilization rate was not directly related to changes in hepatic stores of vitamin A. Concurrently, this suggests that as vitamin A status decreases, vitamin A utilization will remain high until liver vitamin A stores are depleted to a point at which normal plasma retinol levels cannot be maintained, resulting in a decrease in disposal rate. At present, we are not able to determine whether it is an elevation in plasma retinol that directly drives the increased disposal rate or if plasma retinol equilibrates with an extravascular vitamin A pool, which is the substrate for vitamin A utilization.

	FOOTNOTES

Manuscript received 2 March 1998. Initial reviews completed 8 April 1998. Revision accepted 5 June 1998.

	ACKNOWLEDGMENT

We thank Joanne Balmer Green for assistance with surgical procedures and manuscript preparation.

	LITERATURE CITED

Abstract Introduction Methods Results & Discussion References

• Adams W. R., Smith J. E., Green M. H. Effects of N-(4-hydroxyphenyl)retinamide on vitamin A metabolism in rats. Proc. Soc. Exp. Biol. Med. 1995; 208:178-185[Abstract]
• Allen L. E., Green M. H., Green J. B. Correspondence re: S.E. Dew et al. Effects of pharmacological retinoids on several vitamin A-metabolizing enzymes [Cancer Res., 53: 2965-2969, 1993]. Cancer Res. 1994; 54:3319-3320[Free Full Text]
• Berman, M., Beltz, W. F., Greif, P. C., Chabay, R., & Boston, R. C. (1983) CONSAM User's Guide. PHS Publ. no. 1983-421-132:3279. U.S. Government Printing Office, Washington, DC.
• Berman, M. & Weiss, M. F. (1978) SAAM Manual. DHEW Publication no. 78-180. U.S. Government Printing Office, Washington, DC.
• Duncan T. E., Green J. B., Green M. H. Liver vitamin A levels in rats are predicted by a modified isotope dilution technique. J. Nutr. 1993; 123:933-939
• Green M. H., Green J. B. Experimental and kinetic methods for studying vitamin A dynamics in vivo. Methods Enzymol. 1990a; 190:304-317[Medline]
• Green M. H., Green J. B. The application of compartmental analysis to research in nutrition. Annu. Rev. Nutr. 1990b; 10:41-61[Medline]
• Green M. H., Green J. B. Vitamin A intake and status influence retinol balance, utilization and dynamics in rats. J. Nutr. 1994; 124:2477-2485
• Green M.H., Green J. B., Berg T., Norum K. R., Blomhoff R. . Vitamin A metabolism in rat liver: a kinetic model. Am. J. Physiol. 1993; 264:G509-G521[Abstract/Free Full Text]
• Green M. H., Green J. B., Lewis K. C. Variation in retinol utilization rate with vitamin A status in the rat. J. Nutr. 1987; 117:694-703
• Green M. H., Uhl L., Green J. B. A multicompartmental model of vitamin A kinetics in rats with marginal liver vitamin A stores. J. Lipid Res. 1985; 26:806-818[Abstract]
• Hara A., Radin N. S. Lipid extraction of tissues with a low-toxicity solvent. Anal. Biochem. 1978; 90:420-426[Medline]
• Hicks V. A., Gunning D. B., Olson J. A. Metabolism, plasma transport and biliary excretion of radioactive vitamin A and its metabolites as a function of liver reserves of vitamin A in the rat. J. Nutr. 1984; 114:1327-1333
• Landaw E., DiStefano J. III Multiexponential, multicompartmental, and noncompartmental modeling. II. Data analysis and statistical considerations. Am. J. Physiol. 1984; 246:R665-R677
• Lewis K. C., Green M. H., Green J. B., Zech L. A. Retinol metabolism in rats with low vitamin A status: a compartmental model. J. Lipid Res. 1990; 31:1535-1548[Abstract]
• Lewis K., Zech L., Phang J. Effects of N-(4-hydroxyphenyl)retinamide supplementation on vitamin A metabolism. Cancer Res. 1994; 54:4112-4117[Abstract/Free Full Text]
• Ryan, B. F., Joiner, B. L. & Ryan, T. A., Jr. (1985) Minitab Handbook, 2nd ed. PWS Publishers, Boston, MA.
• Thompson J. N., Erdody P., Brien R., Murray T. K. Fluorometric determination of vitamin A in human blood and liver. Biochem. Med. 1971; 5:67-89[Medline]
• Varma R., Beaton G. Quantitative aspects of the urinary and fecal excretion of radioactive metabolites of vitamin A in the rat. Can. J. Physiol. Pharmacol. 1972; 50:1026-1037[Medline]
• Wang L. Plasma volume, cell volume, total blood volume and F(cells) factor in the normal and splenectomized Sherman rat. Am. J. Physiol. 1959; 196:188-192

0022-3166/98 $3.00 ©1998 American Society for Nutritional Sciences