A compartmental model of magnesium metabolism in healthy men based on two stable isotope tracers

Transcription

1 Am J Physiol Regul Integr Comp Physiol 285: R656 R663, First published May 29, 2003; /ajpregu A compartmental model of magnesium metabolism in healthy men based on two stable isotope tracers Magalie Sabatier, 1 Frédéric Pont, 2 Maurice J. Arnaud, 1 and Judith R. Turnlund 3 1 Nestle Water Institute, Vittel; 2 Mass Spectrometry Unit, Institut National de la Santé et de la Recherche Médicale, 31024, Toulouse, France; and 3 Western Human Nutrition Research Center, United States Department of Agriculture, University of California, Davis, California Submitted 9 December 2002; accepted in final form 22 May 2003 Sabatier, Magalie, Frédéric Pont, Maurice J. Arnaud, and Judith R. Turnlund. A compartmental model of magnesium metabolism in healthy men based on two stable isotope tracers. Am J Physiol Regul Integr Comp Physiol 285: R656 R663, First published May 29, 2003; / ajpregu The aim of this study was to build a compartmental model of magnesium (Mg) kinetics by using data collected from six healthy adult men after oral administration of 26 Mg and intravenous administration of 25 Mg. Blood, urine, and feces were collected for 12 days after administration of the isotopes. Isotopic ratios were determined by inductively coupled plasma-mass spectrometry. Data were analyzed for each subject using SAAMII. We began with a compartmental model previously proposed (Avioli LV and Berman M. J Appl Physiol 21: , 1966) and developed an alternative approach to resolve the discrepancy between model-predicted curves and experimental data. This analysis enables the exploration of 25% of total body Mg that exchanges rapidly from plasma compartment with two extraplasma pools. One of the extraplasma compartments contains 80% of the exchangeable Mg with a transport rate of mg/h. The second exchanges mg of Mg/h. The model permitted estimation of kinetic parameters as well as fractional Mg absorption and fecal endogenous excretion. magnesium absorption; magnesium fecal endogenous excretion; inductively coupled plasma-mass spectrometry MAGNESIUM (Mg) is an essential mineral for humans, with a recommended dietary allowance (RDA) of 420 mg/day for adult men (17a). Its physiological role is mainly related to enzyme activity (27), and Mg deficiency has been associated with several pathologies. Despite its importance, minimal information is available regarding the distribution and turnover of exchangeable Mg in humans. The lack of studies of Mg metabolism is due primarily to limitations of the tracers that can be used. The Mg radioisotope has a short half-life (21.8 h), and the precise measurement of the enrichment of relatively abundant stable isotopes presented a challenge until the 1990s for nutrition research. This is accentuated by the fact that few tissues are available for analysis in humans in contrast to animals. These limitations can be overcome in part Address for reprint requests and other correspondence: M. J. Arnaud, Nestle Water Institute, BP 101, 88804, Vittel Cedex, France. R656 with compartmental analysis of Mg tracer kinetics, which can be used to obtain quantitative or predictive information about the dynamics of a system. This mathematical description of mineral distribution provides valuable insight into the metabolism of many nutrients in humans (41). Compartmental models have been developed for several essential minerals on the basis of stable isotope studies in humans, including calcium (37, 42, 43), zinc (13, 21, 31, 38 and 39), copper (6, 30), and molybdenum (16, 33, 34). In addition, parameters deduced from the model can be used as markers of trace element nutritional status. It has been demonstrated for selenium (8, 19) and Zn (17) that, depending on the ingested amount of a mineral, pool size could vary and could be used for mineral status evaluation. Models can also be used to investigate mineral interactions (31, 32). In addition, the bioavailability of a mineral can be deduced from a compartmental model. In a study comparing methods for Zn absorption determination, the result of a compartmental model was used as the reference method (22). Authors assumed that the model provided the most reliable estimation because more information was used to derive the rate of absorption than was used for any other methods. However, for the reasons described above and despite the advantages of such a tool, there are only two compartmental models of Mg in humans using isotopic tracers and plasma, urine, and fecal kinetics (5, 32). The study of Avioli and Berman (5) was carried out in men with the use of 28 Mg, a radioisotope. On the basis of this model, the second was developed in adolescent girls with the use of stable isotopes. To our knowledge, no Mg compartmental model has been built using stable isotopes in healthy men. The aim of this study was to develop a model using data collected in healthy men. Plasma, urine, and fecal kinetics data were obtained in a study also designed to compare isotopic methods for the determination of Mg absorption (29). The rate of Mg absorption and that of fecal endogenous excretion were deduced from the model and are presented hereafter. The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

2 COMPARTMENTAL MODEL OF MAGNESIUM IN HEALTHY MEN R657 SUBJECTS AND METHODS Subjects. Six healthy men aged yr from the San Francisco, CA, area were recruited for the study. Mg balance and absorption were calculated in all six subjects. A comparison of isotopic methods to determine Mg absorption was carried out and is reported elsewhere along with a detailed description of the study (29). Anthropometric characteristics of the subjects are summarized in Table 1. Subjects were nonsmokers. Written informed consent was obtained from all subjects. The institutional review board of the University of California, Davis, and the United States Department of Agriculture (USDA) Human Studies Review Committee approved this protocol. Protocol. All participants were studied in the metabolic research unit (MRU) of the USDA/Agricultural Research Service Western Human Nutrition Research Center (WHNRC) in San Francisco, CA. The study was carried out over an 18-day period of time, and stable isotopes were administered after 6 days of adjustment to the diet. The diet was a 3-day rotating menu with an energy level set to meet the energy requirement of the smallest subject. A liquid formula drink with additional minerals to meet the RDAs of all nutrients and energy to maintain the body weight of each subject was added to the diet, divided into three portions, and given with each meal. The average Mg content in the diet was mg/day. After inclusion of the isotopes administered, the average daily Mg intake was mg/day. Tracer administration. On the morning of the first day of the kinetic study, subjects were given an oral dose of 70 mg of 26 Mg (98.82 atom% 26 Mg) in 200 ml of water (Arrowhead) with breakfast. Fifteen minutes later, 30 mg of 25 Mg (99.58 atom% 25 Mg) were infused into the arm vein. The isotopically enriched Mg was obtained as enriched magnesium oxide powder from the Oak Ridge National Laboratory (Oak Ridge, TN). Sampling. Blood was drawn at the following times relative to the intravenous dose: 0, 5, 15, and 30 min and 1, 2, 4, 6, 11, 16, 24, 48, 72, 96, 120, 144, and 168 h. Complete urine and fecal collections were obtained throughout the study. Urine was pooled by 8-h periods for 3 days after isotope administration and by 24-h periods. Fecal samples were pooled by 3-day periods. Analysis. The plasma volume for each subject was calculated on the basis of age and body mass by use of a nomogram (20a). Mg concentrations in plasma, urine, and digested feces and diet composites were determined by flame atomic absorption spectrophotometry (18) (model 5100; Perkin-Elmer, Norwalk, CT). Isotopic ratios were determined with an inductively coupled plasma-mass spectrometer (ICP-MS; ELAN 6000, Perkin-Elmer Instruments, Norwalk, CT) equipped with an ultrasonic nebulizer (U-6000AT ; Cetac Technologies, Omaha, NE). Internal precision for 25 Mg/ 24 Mg and 26 Mg/ 24 Mg measurements was below 1%. External precision for both ratios was below 0.5% in urine, plasma, and fecal samples. Isotopic ratios were converted into milligrams of tracers excreted or appearing in matrices (35, 36). Kinetic modeling. Subjects were assumed to be in a steady state because they were consuming a diet with a constant Mg intake. Urinary, plasma, and fecal 25 Mg and 26 Mg kinetics were analyzed with the compartmental module of the SAAMII software (Simulation Analysis and Modeling; SAAM Institute, Seattle, WA). In the tracer model, plasma data were expressed as tracer-to-tracee ratios for oral and intravenous tracers. Urinary and fecal data were expressed as excreted cumulative dose (mg). Individual values of dietary unlabeled Mg intake were directed into compartment 6 in the corresponding tracee model. A data-based weighting structure was used for parameter estimation. A constant standard deviation of 0.1 was applied to the data because manipulation and measurement errors were the same for all samples. It resulted in the best data fit, contrary to a fractional standard deviation that privileged the points with a low value and did not improve the fit. The tracee model was built from a tracer model (21) derived from that of Avioli and Berman (5). The circles indicate compartments representing kinetically distinct pools of Mg. The arrows represent transfers between compartments. Fractional transfer coefficients are symbolized by k (i,j) (fraction/h) and are defined as the fraction of compartment j mass moving into compartment i per unit time. Transport rate K (i,j) (mg/h) is the mass of tracee transfer per unit time. K (i,j) was calculated as the product of the fractional transfer k (i,j) and the compartment mass of tracee Q j (mg): K (i,j) k (i,j) Q j. The model (Fig. 1) consists of a unidirectional chain of three compartments describing the gastrointestinal tract, in which a delay (compartment 9) was inserted, represented by a rectangle. Compartment 6 corresponds to the stomach; compartment 7 symbolizes the part of the digestive tract from which Mg is absorbed. Compartment 1 corresponds to plasma, which exchanges in a bidirectional manner with compartment 7 (absorption and endogenous excretion) and with two other extraplasma compartments (compartments 2 and 3). A unidirectional transfer is described between the plasma and urine compartment (compartment 4), which represents the urinary excretion. Tracers were introduced orally and intravenously into compartments 6 and 1. Tracer-to-tracee ratios were measured into plasma (compartment 1), urine (compartment 4), and fecal (compartment 8) compartments. Fractional absorption was calculated as the fraction of Mg entering the circulation from the intestinal site of absorption (compartment 7; Fig. 1) (14, 38) as follows fractional absorption k 1,7 / k 1,7 k 9,7 Table 1. Anthropometric characteristics and plasma Mg concentrations of subjects Subject No. Age, yr Weight, kg Height, cm BMI, kg/m 2 [Mg], mg/l Mean SD BMI, body mass index; [Mg], concentration of Mg.

3 R658 COMPARTMENTAL MODEL OF MAGNESIUM IN HEALTHY MEN mathematical prediction to the urinary observed values are reported in Fig. 2. The simulation revealed a systematic error appearing in the early part of the cumulative urinary excretion of the tracer that was administered intravenously when the fractional transfer coefficient between both compartments was constant [k (4,1) constant]. In other words, predicted values for the urinary excretion of the tracer were lower than the data, as illustrated in Fig. 2A. Changes introduced in parameters could not eliminate this discrepancy. Fig. 1. Tracer model of Mg metabolism in 6 men; 2 tracer inputs (large arrows) and 6 measured variables are indicated. E and F, Measurements of both tracers; circles with nos., compartments; arrows, Mg movement between compartments. Constant k (i,j) (fraction/h) is the fractional transfer coefficient into compartment i from compartment j. Compartments 2 and 3 are extraplasmatic pools exchanging rapidly with plasma compartment 1. The unidirectional chain of compartments 6 8, in which a delay (rectangle; compartment 9) is included, represents the gastrointestinal tract. k (1,7) illustrates the absorption, and k (7,1) is the fecal endogenous excretion. Irreversible losses are represented by compartments 8 and 4, which are feces and urine, respectively. Fecal endogenous excretion was calculated as the plasma Mg fraction entering the intestine without being reabsorbed as follows (38) endogenous fecal excretion k 7,1 (1 fractional absorption) Statistics. Individual parameter estimate values are reported for each subject as well as the mean SD. The Mg absorption rate derived from the model was compared with the values calculated with the fecal monitoring method reported elsewhere for the same subjects (29). Student s paired t-test was used for statistical comparison of the data. Means were considered significantly different when P RESULTS Plasma Mg concentration. Plasma Mg concentrations (mg/l) are reported in Table 1. Individual values were included in the normal range (17 23 mg/l) (4). The average was mg/l. Kinetics analysis. The fitting kinetics and the simultaneous elaboration of the model (shown in Fig. 1) were done following the chronology proposed by Avioli and Berman (5). First, we tried to fit the data with constant fractional transfer coefficients. A good fit of urinary data could not be obtained under these conditions. Because the urinary Mg is directly derived from the plasma Mg, urinary and plasma Mg kinetics relationships were studied separately from the rest of the model by use of a forcing function. The forcing function associated with the plasma kinetics forces the system to predict the urinary curve from the plasma data set (13, 24). Thus the shape of the urinary curve of the injected label was calculated from the measured plasma curve. This function and the adjustment of the Fig. 2. Utilization of a plasma forcing function (see Kinetics analysis) for the calculation of urinary kinetic excretion of the intravenously administered tracer (dotted line). It was calculated from observed ( ) and fitted (solid line) plasma kinetics. E, Observed values in urine. A: urinary kinetics calculated by SAAMII software (dotted line) when k (4,1) was a constant (cnst). A discrepancy between urinary calculated and observed values is visible in the early part of the curve. Cum dose, cumulative dose. B: urinary kinetics calculated by SAAMII software when k (4,1) was nonconstant. C: semilog plot of fractional transfer coefficient k (4,1) into compartment 4 (urine) from compartment 1 (plasma). Perturbation is temporary, with k (4,1) returning to its basal level rapidly after the intravenous tracer administration. Consequently, k (4,1) as a nonconstant value was used to build the model.

4 COMPARTMENTAL MODEL OF MAGNESIUM IN HEALTHY MEN R659 It was then concluded that the experiments did not solve a very rapid phase of excretion of the intravenous tracer during the first 15 h after the isotope injection. It has been suggested that after the intravenous tracer administration, a small fraction was rapidly excreted from the plasma compartment into the urine. In the model of Avioli and Berman (5), this discrepancy was also observed. An operational unit was thus introduced into the model to account for the early appearance of small amounts of the intravenous tracer in the urine. This operational unit did not correspond to a physiological entity but was required for the model to be solved. In the present study, instead of the operational unit, the fractional transfer coefficient between both compartments was transformed into a nonconstant function as follows k 4,1 q plasma where and are constant, and q plasma is the mass of plasma compartment. When leaving the k (4,1) value and thus K (4,1) (the urinary excretion rate) to modulate in relation to the amount of Mg appearing in the plasma compartment, observed kinetics could be fitted without any further changes in the model. As shown in Fig. 2B, this solution gave a very good fit of the urinary data. The forcing function was then turned off, and the data were fitted with all the k (i,j) adjustable plus the and values of k (4,1). The final result was a very good fit of the urinary curves. An example of the adjustment achieved between observed and calculated values for one subject is presented in Fig. 3. A lower standard deviation for most parameters and lower statistical indexes associated with the models indicated unambiguously that the improvement was statistically significant compared with the constant k (4,1) model. The important point is that the forcing function was only used as a study and simulation tool; it was not used in the final model. The relationship between time and the variations of the fractional transfer coefficient k (4,1) between urine and plasma is reported in Fig. 2C. It is shown that k (4,1) was increased temporarily after the injection and returned to its basal steady state in a few hours. The k (4,1) steady-state value was used to calculate the urinary excretion rate. Model and parameters. The multicompartmental Mg model in healthy men is illustrated in Fig. 1. It was tested in six men. All subjects had one central plasma compartment (compartment 1); the mean mass of this pool was mg (range mg). This compartment exchanges with two extraplasma compartments (compartments 2 and 3). Their mean masses were 1, mg (range 512 2,063 mg) and 5,350 1,420 mg (range 3,542 7,133 mg), respectively. Their transport rates were mg/h (range mg/h) and mg/h (range mg/h), respectively. Individual parameters are reported in Table 2. The combined mass of the three compartments (compartments 1 3) is 6,620 1,285 mg (range 4,706 7,946 Fig. 3. Relationship with time (T) between observed and fitted values for a subject. Symbols, measured values; dotted and solid lines, calculated values by the software SAAMII. A: generated curves after oral tracer administration in plasma ( ; tracer/tracee) and urine (E; cumulative dose, mg). B: generated curves after intravenous tracer administration in plasma ( ; tracer/tracee) and urine (E; cumulative dose, mg). C: cumulative appearance of oral (E) and intravenous ( ) tracers in feces. mg). The Mg content of an adult man (70 kg) is 24 g, i.e., 340 mg/kg body wt. Thus this technique allows the exploration of 25 3% of total body Mg, meaning that 25% of the total body Mg exchanges in 12 days. The pool masses differ considerably between and among themselves. Compartment 3 contains 80 9% of rapidly exchangeable Mg; 3 1% of the rapidly exchangeable Mg is contained in the plasma pool, which is 1% of total body Mg. For three subjects (subjects 3 5), a better fit was obtained with an additional extraplasma pool (pool 5),

5 R660 COMPARTMENTAL MODEL OF MAGNESIUM IN HEALTHY MEN Table 2. Summary of calculated multicompartmental model parameters of 6 subjects Mean SD Fractional transfer coefficient, fraction/h k (1,2) k (1,3) k (1,7) k (2,1) k (3,1) k (7,1) k (7,6) k (9,7) *k (4,1) ε Mg transport rate, mg/h K (1,2) K( 1,3) K (4,1) K (9,7) Intestinal delay, h Compartment size, mg Q Q 2 2,063 1, , , Q 3 4,691 4,159 7,133 6,168 6,406 3,542 5,350 1,420 Sum of compartment mass, mg 6,999 5,461 7,946 6,849 7,761 4,706 6,620 1,285 Fractional absorption Fecal endogenous excretion, mg/d Nos. along top represent subjects 1 6. * Steady-state value. which is not represented in Fig. 1. This compartment exchanged in a bidirectional manner with the plasma pool. It had an average mass of mg (range mg). This corresponds to only 2% of the combined mass of pools 1 3 and 5. Its transport rate was extremely rapid: mg/h (range mg/h). The addition of this pool for the three subjects changed the mean value of all parameters. Values are reported Table 3. Summary of calculated multicompartmental model parameters of 6 subjects, with a supplementary compartment 5 for subjects Mean SD Fractional transfer coefficient, fraction/h k (1,2) k (1,3) k (1,7) k (5,1) k (2,1) k (3,1) k (1,5) k (7,1) k (7,6) k (9,7) k (4,1) ε Mg transport rate, mg/h K (1,3) K (1,2) K (5,1) K (4,1) Intestinal delay, h Compartment size, mg Q Q Q 2 2,063 1, , , Q 3 4,691 4,159 6,624 5,908 6,101 3,542 5,171 1,219 Sum of compartment mass, mg 6,999 5,461 7,487 6,931 7,342 4,706 6,488 1,133 Fractional absorption Fecal endogenous excretion, mg/d Nos. along top represent subjects 1 6.

6 COMPARTMENTAL MODEL OF MAGNESIUM IN HEALTHY MEN R661 in Table 3. Those changes were not significantly different from the mean value obtained with the same number of compartments for all six subjects. However, for most of the parameters related to compartments 2 and 3, the coefficient of variation was lower when compartment 5 was added. In this case, the mean sizes of compartments 2 and 3 were 1, mg (range 642 2,063 mg) and 5,171 1,219 mg (range 3,542 6,624 mg), respectively. Their transport rates were mg/h (range mg/h) and 45 9 mg/h (range mg/h), respectively. No changes were observed in the intestinal delay, the absorption and excretion rate, or the fecal endogenous excretion. Because the compartment was small and not detectable in all subjects, it was not used in further analysis. Absorption and fecal endogenous excretion. Individual values of the true Mg absorption (MgA) and fecal endogenous excretion for the simpler model are reported in Table 2. The mean fractional absorption deduced from the model was (range ). Fractional absorption, using the fecal monitoring method, was calculated previously (29) and was (range ) for the apparent Mg absorption (MgAA) and (range ) when corrected for the fecal endogenous excretion, i.e., MgA. Statistical comparison of means showed no difference between MgAA and values predicted using the model. However, MgA calculated with the fecal monitoring was significantly higher than the value predicted by the model (P 0.0, ). The total fecal endogenous excretion calculated from MgA, total unabsorbed dietary Mg, and Mg excreted in feces (29) was estimated to be mg Mg/day (range 2 63 mg/day). The mean value for endogenous excretion into the gastrointestinal tract predicted by the model was mg Mg/day (range mg/day) and was not significantly different from the calculated value. The mean value for the urinary excretion rate is 5 2 mg/h (range 2 7 mg/h). The time for gastrointestinal transit after the end of Mg absorption represented by a delay component was on average h (range h). DISCUSSION The aim of this work was to study Mg metabolism in healthy adult men by use of a compartmental model and to compare the deduced absorption value with the result obtained with the fecal monitoring approach. The problem of the adjustment of mathematical curves to the experimental values for the intravenous tracer excreted in urine, emphasized during the preliminary study of kinetics, was observed in two other Mg models described in literature (5, 32). Avioli and Berman (5) tackled this difficulty and proposed a solution to solve the model. In the work of Sojka et al. (32), the disagreement between model-predicted curves and experimental points can be seen (see Fig. 2 in Ref. 32, whereby the fitting of the mean values obtained for urinary excretion is shown). Mg excretion in the urine is underestimated in the early part of the curve. The effect of such approximation on the final results of whole parameters is not known. However, a correction applied by Avioli and Berman or that proposed herein allows an appreciable amelioration of the fitting. The discrepancy between calculated and observed values could be explained by the intravenous amount of administered isotope. Because of the high natural abundance of Mg isotopes, it is necessary to administer large amounts of tracer to get measurable enrichment in samples. In our study, 30 mg 25 Mg was injected into the Mg plasma compartment (Q mg), i.e., 15% of the total amount of Mg. In the study of adolescent girls aged yr, Sojka et al. injected 20 mg into a pool of 275 mg. Avioli and Berman injected mg 28 Mg into a pool of mg in adult men, i.e., 7.2 and 7.6% of the total amount of the compartment, respectively. Those amounts could be responsible for a transitory hypermagnesemia, resulting in an increase in Mg urinary excretion in the early part of the experiment. It has been shown that Mg homeostasis is regulated by the kidney (26). Thus the fractional transfer coefficient from the plasma pool into the urinary compartment or the urinary excretion rate depends on the plasma Mg concentration. Consequently, it can be defined as a nonconstant function. This solution has been chosen in the present study to fit the kinetics. Inspection of the time course of the fractional transfer coefficient between plasma and urine showed that this state is transitory and lasted only a few hours. The steady state among different components of the model was then reached and maintained. The plasma kinetics of the oral tracer and the preliminary study of kinetics demonstrated that it was unlikely that the oral tracer was responsible for this perturbation, although the oral dose increased intake on the day of administration. This is probably because the oral tracer was given with breakfast, only 0.44 was absorbed, and it was absorbed over a period of time as it passed through the gastrointestinal tract. The utilization of a multicompartmental model enabled the study of 25% of total body Mg after 12 days of sampling. The combined mass of the defined compartments corresponds to mg Mg/kg body wt in adult men. Because of the short half-life of the radioisotope used in the experiment of Avioli and Berman (5), the study lasted only 6 days. Consequently, only 15% of total body Mg could be explored, and the size of the rapidly exchangeable Mg pool (expressed per kg body wt) is lower (43 mg/kg) than the estimated value in the present study. Sojka et al. (32) reported values similar to ours after a 14-day study using stable isotopes in adolescent girls (90 mg/kg). No individual values or standard deviations were reported in that study. A close correlation has been observed between the size of the exchangeable pool of Mg and the rate at which this pool exchanges with the longer-term storage pool and the fat-free mass (1), making a comparison of data difficult when expressed per kilogram of total body weight. Because only 25% of total body Mg can be explored in 12 days, 75% either exchanges more slowly or is not exchangeable. The study of this aspect of Mg kinetics

7 R662 COMPARTMENTAL MODEL OF MAGNESIUM IN HEALTHY MEN would require a longer study. Avioli and Berman (5) estimated the half-life of Mg in the body to be 1,000 h. Mg distribution is equally divided between the skeleton and soft tissues. About 60% of Mg are present in bone tissues, 20% in muscles, and 20% in other soft tissues (3). A fraction of bone Mg exchanges extremely slowly and could account for this unexplored compartment. However, it has been suggested that a part of bone Mg is in a readily exchangeable labile form (23). Consequently, it is difficult to make an association between the described compartment and anatomic entities. Eighty percent of rapidly exchangeable Mg is found in one compartment. Its rate of turnover is two times less important than the second extraplasma compartment present in all six subjects. The plasma compartment contains 1% of total body Mg. These data are in agreement with values found in the literature (5, 9). In the present experiment, the Mg in the plasma compartment had a mass of mg in adult men (mean SD). Sojka et al. (32) found the same compartment had a mass of 275 mg but gave no individual values or standard deviations, making comparison hazardous. However, as already indicated, the combined mass of the defined compartments was similar in adult men and adolescent women. This could suggest a difference in Mg distribution between the compartments due to age. Some authors suggested that variability of Mg metabolism according to age could be due to the rapid growth of bone and muscle tissue during early adolescence (11). The results showed that for three subjects the plasma pool could exchange with a third extraplasma compartment (5). This compartment was suppressed because data from three of the six subjects did not support a supplementary pool. This could be due to lack of accuracy in the measurements. When it exists, the pool size is small, and its transport rate is extremely rapid. A firm justification of its presence is difficult because of the absence of data enabling the evaluation of Mg status, stress, or pathological state. Although plasma Mg concentration did not indicate Mg deficiency, the lack of reliability of this marker for the evaluation of Mg status does not make it possible to establish a link between this supplementary pool and deficiency (10, 12, 25). However, the fit and associated statistics were better in 50% of our subjects when this pool was introduced. Its presence is due to the sharp shape of the curve at the early time points and could be also linked to the hypermagnesemia after isotope administration. The overestimation of the true Mg absorption calculated with the fecal monitoring compared with the predicted value by a model has also been observed for Zn (22, 40). In the case of Mg, the absolute disagreement of the fraction absorbed between both values is low (0.04). The fecal monitoring method was corrected with the tracer fecal endogenous excretion (tfee) to get the true absorption. On average, the tfee was 2.4% of tracer/day. Average total endogenous excretion into the gastrointestinal tract derived from the model was mg/day, with a Mg dietary intake of mg, and was in accordance with an estimate of the total fecal endogenous excretion based on fecal and dietary Mg and calculated true absorption. Similar values have been reported (20, 28). However, in the Avioli and Berman study (5), fecal endogenous excretion was mg/day. The use of various calculations and a different period of time to estimate this value could explain this difference. In adolescent girls, fecal endogenous excretion was also lower ( mg/day) (32). However, no close relationship between Mg intake and endogenous fecal excretion was found (2). Perspectives This study has established parameters for a Mg multicompartmental model in healthy men. The model was based on plasma, urinary, and fecal kinetics after oral and intravenous tracer administration. An approach was proposed to solve the discrepancy between model-predicted curves and experimental points. Therefore, the model can be used to evaluate parameters under other circumstances or in parallel with other tests to establish a relationship between the rapidly exchanging Mg pool and Mg status. We thank Drs. D. M. Foster and P. Vicini of the Resource Facility for Kinetic Analysis, Center for Bioengineering, Univ. of Washington, Seattle, WA, for training in use of the SAAM software and for advice on modeling procedures. We are also grateful to W. R. Keyes (WHNRC, USDA, Davis, CA) for help with mass spectrometry methods and Dr. V. Gildengorin (WHNRC) for assistance with statistical analysis. Part of the results was presented as a poster at the 11 th International Symposium on Trace Elements in Man and Animals, Berkeley, CA, June 2 6, DISCLOSURES This work was supported by the Nestle Water Institute and the USDA. SAAM software training was supported by National Center for Research Resources Grant RR REFERENCES 1. Abrams SA and Ellis KJ. Multicompartmental analysis of magnesium and calcium kinetics during growth: relationships with body composition. Magnes Res 11: , Abrams SA and Wen JP. Methodologies for using stable isotopes to assess magnesium absorption and secretion in children. J Am Coll Nutr 18: 30 35, Aikawa JK. Distribution of magnesium in human body. In: The Role of Magnesium in Biologic Processes. Springfield, IL: Thomas, 1963, p Altura BM. Importance of Mg in physiology and medicine and the need for ion selective electrodes. Scand Clin Lab Invest 54, Suppl 217: 5 9, Avioli LV and Berman M. Mg 28 kinetics in man. J Appl Physiol 21: , Buckley WT. Application of compartmental modeling to determination of trace element requirements in humans. J Nutr 126, Suppl 9: 2312S 2319S, Ducros V, Faure P, Ferry M, Couzy F, Biajoux I, and Favier A. The sizes of the exchangeable pools of selenium in elderly women and their relation to institutionalization. Br J Nutr 78: , Elin RJ. Assessment of magnesium status. Clin Chem 33: , Elin RJ. Status of mononuclear blood cell magnesium assay. Am Coll Nutr 6: , 1987.

8 COMPARTMENTAL MODEL OF MAGNESIUM IN HEALTHY MEN R Feillet-Coudray C, Coudray C, Tressol JC, Pepin D, Mazur A, Abrams SA, and Rayssiguier Y. Exchangeable magnesium pool masses in healthy women: effects of magnesium supplementation. Am J Clin Nutr 75: 72 78, Flink EB. Magnesium deficiency in alcoholism. Alcohol Clin Exp Res 10: , Foster DM, Aamodt RL, Henkin RI, and Berman M. Zinc metabolism in humans: a kinetic model. Am J Physiol Regul Integr Comp Physiol 237: R340 R349, Friel JK, Naake VL Jr, Miller LV, Fennessey PV, and Hambidge KM. The analysis of stable isotopes in urine to determine the fractional absorption of zinc. Am J Clin Nutr 55: , Giussani A, Cantone MC, de Bartolo D, Roth P, and Werner E. A revised model of molybdenum biokinetics in humans for application in radiation protection. Health Phys 75: , Hambidge M, Miller L, Naake V, Zhaohi H, Crebs NF, Wescott J, and Fennessey P. Zinc that exchanges with zinc in plasma within two days: variation with dietary zinc. In: Proceedings of Mathematical Modeling in Experimental Nutrition IV: Trace Element/Mineral Metabolism During Development, edited by Wastney ME and Subramanian KNS. Boca Raton, FL: CRC, 1995, p a.Institute of Medicine. Dietary Reference Intakes for Calcium, Phosphorus, Magnesium, Vitamin D, and Fluoride. Washington, DC: National Academy, Institute of Organic Chemistry. Handbook on Metals in Clinical and Analytical Chemistry, edited by Seiler HG, Sigel A, and Sigel H. Basel: Dekker, Janghorbani M, Martin RF, Kasper LJ, Sun XF, and Young VR. The selenite-exchangeable metabolic pool in humans: a new concept for the assessment of selenium status. Am J Clin Nutr 51: , Knudsen E, Sandstrom B, and Solgaard P. Zinc, copper, and magnesium absorption from a fibre-rich diet. J Trace Elem Med Biol 10: 68 76, a.Lentner C. Geigy Scientific Tables, Vol. 3. West Caldwell, NJ: Ciba Geigy, Lowe NM, Shames DM, Woodhouse LR, Matel JS, Roehl R, Saccomani MP, Toffolo G, Cobelli C, and King JC. A compartmental model of zinc metabolism in healthy women using oral and intravenous stable isotope tracers. Am J Clin Nutr 65: , Lowe NM, Woodhouse LR, Matel JS, and King JC. Comparison of estimates of zinc absorption in humans by using 4 stable isotopic tracer methods and compartmental analysis. Am J Clin Nutr 71: , MacIntyre I, Hanna S, Booth CC, and Read AE. Intracellular magnesium deficiency in man. Clin Sci (Lond) 20: , Pont F, Duvillard L, Verges B, and Gambert P. Development of compartmental models in stable-isotope experiments: application to lipid metabolism. Arterioscler Thromb Vasc Biol 18: , Quamme GA. Laboratory evaluation of magnesium status. Renal function and free intracellular magnesium concentration. Clin Lab Med 13: , Quamme GA. Renal magnesium handling: new insights in understanding old problems. Kidney Int 52: , Rude RK. Magnesium deficiency: a cause of heterogeneous disease in humans. J Bone Miner Res 13: , Schwartz R, Herta S, and Wentworth R. Measurement of magnesium absorption in man using stable 26 Mg as a tracer. Clin Chim Acta 87: , Sabatier M, Keyes WR, Pont F, Arnaud MJ, and Turnlund JR. Comparison of stable isotope methods for the determination of magnesium absorption in humans. Am J Clin Nutr 77: , Scott KC and Turnlund JR. Compartmental model of copper metabolism in adult men. J Nutr Biochem 5: , Scott KC and Turnlund JR. A compartmental model of zinc metabolism in adult men used to study effects of three levels of dietary copper. Am J Physiol Endocrinol Metab 267: E165 E173, Sojka J, Wastney M, Abrams S, Lewis SF, Martin B, Weaver C, and Peacock M. Magnesium kinetics in adolescent girls determined using stable isotopes: effects of high and low calcium intake. Am J Physiol Regul Integr Comp Physiol 273: R710 R715, Thompson KH, Scott KC, and Turnlund JR. Molybdenum metabolism in men with increasing molybdenum intakes: changes in kinetic parameters. J Appl Physiol 81: , Thompson KH and Turnlund JR. Kinetic model of molybdenum metabolism developed from dual stable isotope excretion in men consuming a low molybdenum diet. J Nutr 126: , Turnlund JR and Keyes WR. Automated analysis of stable isotopes of zinc, copper, iron, calcium, and magnesium by thermal ionization mass spectrometry using double isotope dilution for tracer studies in humans. J Micronutr Anal 7: , Turnlund JR, Michel MC, Keyes WR, King JC, and Margen S. Use of enriched isotopes to determine zinc and iron absorption in elderly men. Am J Clin Nutr 35: , Wajchenberg BL, Leme PR, Ferreira MN, Filho JM, Pieroni RR, and Berman M. Analysis of 47 Ca kinetics in normal subjects by means of a compartmental model with a non-exchangeable plasma calcium fraction. Clin Sci (Lond) 56: , Wastney ME, Angelus P, Barnes RM, and Subramanian KN. Zinc kinetics in preterm infants: a compartmental model based on stable isotope data. Am J Physiol Regul Integr Comp Physiol 271: R1452 R1459, Wastney M, Gokmen IG, Aamodt RL, Rumble WF, Gordon DT, and Henkin RI. Kinetic analysis of zinc metabolism in humans after simultaneous administration of 65 Zn and 70 Zn. Am J Physiol Regul Integr Comp Physiol 260: R134 R141, Wastney ME and Henkin RI. Calculation of zinc absorption in humans using tracers by fecal monitoring and a compartmental approach. J Nutr 119: , Wastney ME and Subramanian KNS (Editors). Kinetic Models of Trace Element and Mineral Metabolism During Development. Boca Raton, FL: CRC, Weiss GH, Goans RE, Gitterman M, Abrams SA, Vieira NE, and Yergey AL. A non-markovian model for calcium kinetics in the body. J Pharmacokinet Biopharm 22: , Yergey AL, Abrams SA, and Vieira NE. Recent studies of human calcium metabolism using stable isotopic tracers. Can J Physiol Pharmacol 68: , 1990.