1206 Am J Clin Nutr 2003;77:120612. Printed in USA.

2003 American Society for Clinical Nutrition

Comparison of stable-isotope-tracer methods for the determination
of magnesium absorption in humans
14
Magalie Sabatier, William R Keyes, Frdric Pont, Maurice J Arnaud, and Judith R Turnlund
ABSTRACT
Background: The double-labeling (DL) method for determining
magnesium absorption is less cumbersome than is the fecal mon-
itoring method, which has been used most often, but it has not
been validated.
Objective: The aim of this study was to compare methods and
several sampling protocols for determining magnesium absorption
to establish a simple and reliable alternative to the fecal monitor-
ing approach. Fecal monitoring was used as the standard against
which the DL methods based on urine data (DLU), plasma data
(DLP), and plasma kinetics with the use of a deconvolution analy-
sis (DP) were compared.
Design: Six healthy adult men received 70 mg
26
Mg orally and
30 mg
25
Mg intravenously. Multiple blood samples and complete
urine and fecal samples were collected over 12 d. Stable-isotope
ratios were determined by inductively coupled plasma mass
spectrometry.
Results: Results from DLU were not significantly different from
the fecal monitoring reference value (0.48 0.05; x

SD) when
based on 3-d urine pools from 72 to 144 h (0.54 0.04) and when
based on the 24-h urine pools from 48 to 72 h (0.49 0.06), 72 to
96 h (0.51 0.11), and 96 to 120 h (0.50 0.06). Results with the
DLP method 72 h after isotope administration also compared well
with those with the fecal monitoring method (0.54 0.09). Mag-
nesium absorption was 0.47 0.06 with the DP method, which
also agreed with the fecal monitoring value.
Conclusions: The DL methods are an alternative to fecal moni-
toring when applied within the appropriate time intervals. There-
fore, DLUthe simplest and least invasive approachis recom-
mended for determining magnesium absorption. Am J Clin
Nutr 2003;77:120612.
KEY WORDS Magnesium absorption, stable isotopes, mass
spectrometry, double-labeling methods, dysprosium, men
INTRODUCTION
In humans, > 300 enzyme systems are dependent on the pres-
ence of magnesium. It has been suggested that aging, stress, and
various diseases may increase magnesium requirements (1) and
that inadequate intake and impaired magnesium absorption (MgA)
may contribute to many pathologic conditions (2). Therefore, reli-
able methods for determining MgA are essential for investigating
magnesium metabolism and homeostasis.
Few studies of MgA determination with stable isotopes have
been reported, despite their safety and accuracy. One limitation of
1
From the Nestl Water Institute, Vittel, France (MS and MJA); the West-
ern Human Nutrition Research Center, US Department of Agriculture/Agri-
cultural Research Service, University of California, Davis, CA (WRK and JRT);
and the Mass Spectrometry Unit, INSERM, Toulouse, France (FP).
2
Portions of the results were presented as an oral communication at the 9th
Magnesium International Symposium, Mag 2000, Vichy, France, 1025 Sep-
tember 2000.
3
Supported by the Nestl Water Institute and the US Department of Agriculture.
4
Address reprint requests to MJ Arnaud, Nestl Water Institute, BP 101,
Vittel Cedex 04, France. E-mail: maurice.arnaud@waters.nestle.com.
Received June 26, 2002.
Accepted for publication September 26, 2002.
their use is that the 3 magnesium isotopes are relatively high in
abundance (3). Consequently, large amounts of isotope must be
administered for reliable isotopic measurements, generating a high
cost for such experiments. New generations of mass spectrome-
ters with improved accuracy and requiring smaller doses of iso-
topes to be administered have been developed, but sample prepa-
ration remained time consuming (46). The development of
inductively coupled plasma mass spectrometry (ICP-MS) brought
about higher sensitivity, rapid throughput, and less time-consum-
ing sample preparation, simplifying magnesium isotopic ratio
measurements (7).
This instrumentation made the investigation of MgA simpler,
allowing use of the double-labeling (DL) method. This method
requires the administration of 2 stable-isotope tracers, one orally
and one intravenously. For calculations, it is assumed that the oral
label, once absorbed, is indistinguishable from the intravenous
label. Both tracers are assumed to mix rapidly and completely in
the plasma portion of the total-body magnesium pool and are
excreted in urine in a way that reflects their abundance in the cir-
culation. Thus, evaluation of MgA can be based on urine (DLU)
or plasma (DLP) collections (8). The kinetics of the labels can also
be analyzed by deconvolution (DP) (9), another alternative for
estimating MgA. In the simplest protocol, MgA could be evalu-
ated from a single blood draw or a short urine collection. Fecal
monitoring and the oral administration of one isotope have been
used most frequently to determine MgA (1013). MgA is esti-
mated by determining the disappearance of the label during intes-
tinal passage, which is calculated as the difference between intake
and fecal content. However because fecal monitoring does not take
into account fecal endogenous excretion (FEE), MgA is underes-
timated and the result corresponds to the apparent MgA (MgAA).
FEE can be estimated with a second isotope administered intra-
venously and then used to correct MgAA. Thus, fecal monitoring

a
t

U
n
i
v

d
e

S
a
o

P
a
u
l
o

U
S
P

o
n

D
e
c
e
m
b
e
r

2
3
,

2
0
0
8

w
w
w
.
a
j
c
n
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

METHODS FOR DETERMINING MAGNESIUM ABSORPTION 1207
TABLE 1
Anthropometric characteristics of the subjects
Subject Age Weight Height BMI
y kg cm kg/m
2
1 41 93.0 185 27.1
2 36 57.7 160 22.2
3 38 88.0 175 28.6
4 40 84.5 180 26.2
5 27 76.4 171 26.0
6 26 66.8 173 22.4
x

SD 34.7 6.6 77.7 13.5 174.0 8.5 25.4 2.6

does not provide a direct measurement of MgA. The collection of
fecal samples associated with this technique is cumbersome com-
pared with the sampling of urine and blood with the DL tech-
niques. To our knowledge, except in rats (14), no comparisons
have been done between these various isotopic techniques. The
aim of this human study was to investigate the feasibility of study-
ing MgA with the DL methods and to compare the results with the
result from fecal monitoring corrected for FEE. Several sampling
protocols were tested to identify the simpler and the most reliable
alternative to the fecal monitoring method.
SUBJECTS AND METHODS
Subjects
Six healthy men were recruited for the study. These subjects
were a subset of 11 men participating in a copper supplementa-
tion study. The anthropometric characteristics of the subjects are
summarized in Table 1. Written informed consent was obtained
from all subjects. The Institutional Review Board of the Univer-
sity of California, Davis, and the US Department of Agriculture
Human Studies Review Committee approved the protocol.
Protocol
All subjects were studied in the Metabolic Research Unit of the
US Department of Agriculture/Agricultural Research Service
Western Human Nutrition Research Center in San Francisco. The
study was performed over 18 d. After 6 d of adjustment to the diet,
subjects received the isotopes. The basic diet was a 3-d rotating
menu with an energy level set to meet the energy requirement of
the subject with the smallest body weight. A liquid formula drink
containing additional minerals to meet the recommended dietary
allowances of all nutrients and additional energy to maintain the
body weight of each subject was added to the basic diet. The drink
was divided into 3 portions and given with each meal. During the
day, the subjects had free access to deionized water.
Isotope preparation
The
25
Mg (99.58 atom%) and
26
Mg (98.82 atom%) stable-isotope
solutions were prepared from enriched magnesium oxide powders
(Oak Ridge National Laboratory, Oak Ridge, TN). The magne-
sium oxide powders (500 mg
25
MgO and 800 mg
26
MgO) were
weighed into polytetrafluoroethylene beakers that had been
washed overnight in hydrochloric acid (1:1, by vol), rinsed with
deionized water, refluxed with ultrapure nitric acid (1:1, by vol;
Seastar Chemicals, Sidney, Canada), and rinsed again with deion-
ized water. Inside a laminar flow hood, ultrapure concentrated
hydrochloric acid (Seastar) was added and the beakers, covered
with acid-washed watch glasses, were warmed on a hot plate and
gently swirled from time to time until all powder was dissolved.
The solutions were quantitatively transferred to acid-washed poly-
propylene containers. Sufficient sterile 2 N sodium hydroxide was
added to adjust the pH to 2. Deionized water was added to achieve
the desired concentrations, and weighed aliquots were transferred
into acid-washed polypropylene bottles. Aliquots of the oral solu-
tion, containing 70 mg
26
Mg, were weighed into acid-washed
polypropylene test tubes, capped, and stored frozen until the iso-
tope feeding day.
The
25
Mg infusion solution was prepared using sterile water
for infusion, and all containers and pipette tips were autoclaved.
The infusion solution was filtered through a 0.22-m filter, the
filter was rinsed with sterile water, and the solution was trans-
ferred to vials for injection. An aliquot was tested for endotoxins.
The exact concentrations of the enriched solutions were deter-
mined by isotope dilution with the use of certified magnesium
standard solutions.
Tracer administration
On the morning of the seventh day, the subjects received the
isotopes intravenously through a catheter inserted into an arm
vein. A fasting blood sample was drawn. The oral dose of 70 mg
26
Mg was given in 200 mL water (Arrowhead; Nestl Waters
North America Inc, Greenwich, CT) with breakfast. The breakfast
comprised a 70-g bagel, 25 g jam, and 250 mL cranberry apple
juice. A dysprosium fecal marker (2 mg Dy as DyCl
3
) was admin-
istered along with the stable-isotope tracers to check the com-
pleteness of fecal collections. Fifteen minutes later, 30 mg
25
Mg
was infused into the arm vein.
Sample collection and preparation
A schema of the experiment and sampling is reported in
Figure 1. Blood samples were obtained via the catheter at the fol-
lowing times after administration of the intravenous dose: 5, 15,
and 30 min and 1, 2, 4, 6, 11, 16, 24, 48, 72, 96, 120, 144, and 168 h.
After centrifugation at 4000 rpm (1860 g) for 10 min at 4 C,
plasma was removed and stored at 20 C until analyzed. Complete
urine and fecal collections were obtained throughout the study.
Urine was pooled by 24-h and 3-d periods throughout the study and
by 8-h periods for 3 d after isotope administration. Urine samples
were weighed and acidified with 1 mL concentrated ultrapure
hydrochloric acid (Seastar Chemicals) per 200 mL urine before
storage at 20 C. Fecal samples were pooled and homogenized by
3-d periods. Duplicates of all food items were weighed for each day
and homogenized with weighed amounts of ultrapure water in a lab-
oratory blender. Diet and fecal samples were lyophilized, crushed,
and stored in plastic containers in desiccators. Before the magne-
sium analysis, sample aliquots were digested in duplicate in a
microwave digestion oven (MDS-2000; CEMCorp, Mathews, NC)
by using the appropriate programfor each. The digestion was done
on a 0.2-g sample and with 5 mL concentrated ultrapure HNO
3
added. Samples were diluted with ultrapure water for analysis.
Analysis
The plasma volume for each subject was calculated on the basis
of age and body mass by using a nomogram (15). The magnesium
content of the breakfast was analyzed by using the Nutrition Data
System for Research (NDS-R, version 4.01; Nutrition Coordinat-
ing Center, University of Minnesota, Minneapolis, 1998).

a
t

U
n
i
v

d
e

S
a
o

P
a
u
l
o

U
S
P

o
n

D
e
c
e
m
b
e
r

2
3
,

2
0
0
8

w
w
w
.
a
j
c
n
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1208 SABATIER ET AL
FIGURE 1. Study design. Oral and intravenous tracers were administered to 6 healthy men on the same day. The feces were pooled by 3-d periods.
Urine was pooled by 8-h, 24-h and 3-d periods. Blood was drawn at 5, 15, and 30 min and 1, 2, 4, 6, 11, 16, 24, 48, 72, 96, 120, 144, and 168 h. During
the experiment, the diet consumed by the subjects was a 3-d rotating menu.
Plasma, urine, and digested feces and diet composites were
diluted with 0.5% lanthanum chloride to a magnesium concentra-
tion of 0.10.4 ppm (16). Magnesium concentrations were deter-
mined by flame atomic absorption spectrophotometry (model
5100; Perkin-Elmer, Norwalk, CT). A certified bovine liver stan-
dard was analyzed with each set of samples (SRM 1577; National
Institute of Standards and Technology, Gaithersburg, MD).
Urine and digested fecal samples were diluted to 100 ppb
with ultrapure 1% HNO
3
for magnesium isotopic ratio determi-
nations. Plasma samples were diluted to 50 ppb to decrease the
matrix effect.
Dysprosium excreted in the feces was determined by ICP-MS
with the use of external calibration with standards of 0.5, 1, 1.5,
2.5, and 5 ppb Dy. The fecal digest was evaporated to dryness, dis-
solved again, and diluted to give an approximate concentration of
15 ppb Dy in 1% HNO
3
. Rhodium was used as internal standard
at a concentration of 1 ppb. The ICP-MS parameters used for the
dysprosium and magnesium analyses are shown in Table 2.
All acids and other chemicals used in sample preparation and
dilution were ultrapure. All flasks used for sample collection and
manipulations were acid washed in 1 mol HNO
3
/L for 24 h, fol-
lowed by rinsing in ultrapure deionized water.
Isotope ratio determinations
Isotopic ratios were determined with an ICP-MS (ELAN 6000;
Perkin-Elmer Instruments) equipped with an ultrasonic nebulizer
(U-6000AT+; Cetac Technologies Inc, Omaha). The instrument
parameters used for magnesium isotope ratio analysis are shown
in Table 2. Instrument bias over time was corrected for by meas-
uring an unenriched reference sample appropriate to each kind of
TABLE 2
Parameters of inductively coupled plasma mass spectrometry used for
dysprosium and magnesium analyses
Parameter Dysprosium Magnesium
RF power (W) 1000 1000
Sample intake (mL/min) 1.2 1.2
Detector mode Analog Analog
Scanning mode Peak hopping Peak hopping
Isotopes
163
Dy
24
Mg,
25
Mg, and
26
Mg
Dwell times (ms) 20 5, 36, and 34
Sweeps 50 100
Replicates 5 4

a
t

U
n
i
v

d
e

S
a
o

P
a
u
l
o

U
S
P

o
n

D
e
c
e
m
b
e
r

2
3
,

2
0
0
8

w
w
w
.
a
j
c
n
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

METHODS FOR DETERMINING MAGNESIUM ABSORPTION 1209
TABLE 3
Dysprosium recoveries in the fecal samples of the 6 subjects
Subject Dysprosium recovery
% of dose
1 104.2
2 89.3
3 104.6
4 102.2
5 100.5
6 102.4
x

SD 100.5 5.7
matrix with the same concentration. The ratio correction factor
was checked every 10 samples.
Within-run precision (5 replicates) was <1% for
25
Mg/
24
Mg and
26
Mg/
24
Mg ratios. Repeatability, determined by measuring base-
line samples several times over 5 h on the same day, was <0.3% for
both ratios in feces, urine, and plasma. Limits of detection (LODs)
of
25
Mg and
26
Mg enrichments in urine and feces were obtained
from the same measurements and were calculated by using the def-
inition that the LOD for measuring a change in an isotope ratio is
3 times the SD of the baseline (17). LODs for
25
Mg/
24
Mg and
26
Mg/
24
Mg were 0.62% and 0.62% in urine, 0.89% and 0.92% in fecal
samples, and 0.84% and 0.94% in plasma samples, respectively.
Calculations
Magnesium balance
Magnesium balance was calculated for each subject by sub-
tracting the magnesium excreted in urine (U) and feces (F) from
the amount ingested (I) as determined by atomic absorption
spectrophotometry:
Balance = I (U + F) (1)
Magnesium absorption
The isotopic ratios of the samples, determined by ICP-MS, and
their magnesium concentrations, determined by atomic absorption
spectrophotometry, were used to determine the masses of both
tracers that were excreted, as previously described (6, 13). The
fecal monitoring method uses only the amount of nonabsorbed
26
Mg oral tracer excreted in the fecal samples for the calculation
of the apparent absorption of magnesium (MgAA):
MgAA = (D M)/D (2)
where D is the amount of enriched
26
Mg oral tracer ingested, and
M is the amount of enriched
26
Mg oral tracer excreted in feces.
Apparent absorption is less than true absorption because a frac-
tion of the
26
Mg oral tracer that was absorbed is excreted into the
feces during the collection period, which increases M. This frac-
tion is called the FEE. MgAA is related to the true absorption
(MgA) as follows:
MgAA = MgA FEE (3)
The calculation of FEE assumes that the fraction of the infused
25
Mg tracer excreted in feces is the same as the fraction of the
absorbed
26
Mg oral tracer excreted in the feces. The determina-
tion was done over the same period used for the calculation of
absorption. The formula for FEE is as follows:
FEE = MgA (M
i
/D
i
) (4)
where M
i
is the amount of infused
25
Mg tracer excreted in the
feces during the collection period, and D
i
is the dose of
25
Mg
tracer infused.
Combining Equation 4 and Equation 3 and solving for MgA
yields the equation for absorption for the corrected fecal moni-
toring method as follows:
MgA = (1 M/D)/(1 M
i
/D
i
) (5)
For DLU and DLP, the determination of absorption uses the
fractions of oral and infused isotopes excreted in urine or appear-
ing in the plasma as follows:
MgA = (M/D)/(M
i
/D
i
) (6)
MgA was evaluated with Equation 6 from the 8-h, 24-h, and 3-d
urine pools and from each blood sample.
Deconvolution analysis was also used to determine MgA. With
this method, the absorption rate is determined from the plasma
kinetics of the oral and intravenous tracers as described previously
for estimating calcium (18) or zinc (19) absorption. MgA is eval-
uated by using the following equation:
MgA = [AUC
0t
M/D]/[AUC
0t
M
i
/D
i
] (7)
where AUC is the area under the curve of a plot of the plasma
tracer concentrations expressed as a percentage of the dose per
liter for the oral and intravenous isotopic tracers. SAAMII soft-
ware (SAAM Institute, Seattle) was used to calculate both inte-
grals from 0 to 168 h. It was assumed that this period of time was
great enough for the absorption process to be completed.
Statistical analysis
Statistical analysis was performed with the personal computer
version 6.12 of the Statistical Analysis System (20). Students
paired t test was used for all statistical comparisons of the data.
Means were considered significantly different when P < 0.05. All
values are expressed as means SDs.
RESULTS
The magnesium content in the diet was 252.9 26.9 mg/d
for the 3-d rotating menu. The liquid formula drink added
4.0 1.6 mg Mg/d per subject. Inclusion of the isotopes adminis-
tered yielded an average total daily intake of 265.3 1.6 mg Mg/d.
The endpoint of
26
Mg tracer excretion in feces and the com-
pleteness of fecal collection were checked by determining dys-
prosium recoveries in feces. Individual dysprosium recoveries are
presented in Table 3. The mean recovery was 100.5 5.7%,
ranging from 89.3% to 104.6% of the administered dysprosium
dose, suggesting that collections were complete. The data indi-
cated that the excretion of orally administered isotope was com-
plete after 69 d.
Magnesium intakes and fecal and urinary excretion during the
12 d after isotope administration are shown in Table 4. For 4 sub-
jects with an average magnesium intake of 265 mg/d, the sum of
fecal and urinary excretion exceeded the magnesiumintake, result-
ing in negative balance. However, the average magnesium balance
was not significantly different from zero (16 15 mg/d).
Individual FEE and magnesium absorption calculated by the
fecal monitoring and DP methods are summarized in Table 5. The
average FEE expressed as a percentage of the absorbed dose of
isotopic tracer excreted per day was 2.4 0.6%. The average value
calculated by fecal monitoring was 0.46 0.05% for MgAA and

a
t

U
n
i
v

d
e

S
a
o

P
a
u
l
o

U
S
P

o
n

D
e
c
e
m
b
e
r

2
3
,

2
0
0
8

w
w
w
.
a
j
c
n
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1210 SABATIER ET AL
TABLE 7
Fractional magnesium absorption determined in plasma samples with the
double-labeling method
1
Time (h) Value
4 0.44 0.06
6 0.61 0.10
2
11 0.63 0.11
2
16 0.59 0.10
2
24 0.56 0.09
2
48 0.55 0.09
2
72 0.54 0.09
96 0.53 0.09
120 0.49 0.09
144 0.53 0.13
168 0.57 0.12
1
x

SD; n = 6 men.
2
Signicantly different from the mean value determined with the fecal
monitoring method (0.48 0.05; see Table 5), P < 0.05.
TABLE 4
Magnesium intake, fecal and urinary excretion of magnesium, and
magnesium balance in 6 men during the last 12 of 18 d of a controlled diet
Fecal Urinary
Magnesium magnesium magnesium Magnesium
Subject intake excretion
1
excretion
1
balance
mg/d
1 267.0 138.6 33.2 120.6 16.1 7.8
2 262.7 202.2 19.7 85.3 29.0 24.8
3 266.5 195.9 27.4 102.0 10.6 31.4
4 265.9 152.1 13.7 134.7 11.6 20.9
5 265.4 162.0 68.1 132.7 4.6 29.3
6 264.3 163.2 37.6 98.6 6.7 2.5
x

SD 265.3 1.6 169.0 22.8 112.3 18.3 16.0 15.4

1
x

SD.
TABLE 6
Fractional magnesium absorption determined with the double-labeling
method based on urine data
1
Time (h) 8-h pools 24-h pools 3-d pools
08 0.14 0.04
2
816 0.61 0.12
2
1624 0.57 0.09
2
0.27 0.56
2
2432 0.51 0.12
2
3240 0.59 0.14
2
4048 0.62 0.17
2
0.56 0.10
4856 0.43 0.09
5664 0.54 0.10
6472 0.57 0.07
2
0.49 0.06 0.34 0.06
2
7296 0.51 0.11
96120 0.50 0.06
120144 0.45 0.07 0.54 0.04
144168 0.41 0.08
2
168192 0.43 0.06
2
192216 0.46 0.07 0.52 0.09
216240 0.45 0.08
240264 0.43 0.06
2
264288 0.39 0.12 0.46 0.07
1
x

SD; n = 6 men. Urine was pooled by 8-h, 24-h, and 3-d periods to
evaluate the best procedure for determining magnesium absorption.
2
Signicantly different from the mean value determined with the fecal
monitoring method (0.48 0.05; see Table 5), P < 0.05.
0.48 0.05 for MgA after correction for FEE. The results of the
DL method were compared with those of the fecal monitoring
method after correction for FEE. With the use of the DP calcula-
tion from plasma kinetics, the average MgA value was 0.47 0.06,
which was not significantly different from the value calculated by
fecal monitoring.
MgA values determined with the DLU method are reported in
Table 6. To evaluate the effect of the interval time of collection
on MgA values, urine was pooled by 8-h (first 72 h), 24-h, and
3-d periods of time. In the 8-h pools, statistical analysis showed
that absorption was significantly different from that determined
by fecal monitoring through 48 h after isotope administration.
Only values determined from the two 8-h urine pools between 48
and 64 h (0.43 0.09 and 0.54 0.10) were not significantly dif-
ferent from the reference value. The 6472-h mean (0.57 0.07)
was significantly different from the mean determined by fecal
monitoring. Except for the 08-h (0.14 0.04) and 4856-h
(0.43 0.09) pools, the 8-h means were higher than the fecal
monitoring mean.
For the 24-h urine pools, there were no significant differences
between the DLU means and the fecal monitoring means
24144 h, 192240 h, and 264288 h after isotope administra-
tion. The 24-h DLU mean was higher than the fecal monitoring
mean from 24 to 120 h and then lower than the fecal monitoring
mean thereafter. The closest agreements were between 4872 h
(0.49 0.06), 7296 h (0.51 0.11), and 96120 h (0.50 0.06).
No significant differences were observed between the 3-d urine
pools and fecal monitoring after the first 72 h after isotope admin-
istration; the closest agreement was during the 216288-h period
(0.46 0.07).
MgA values calculated with the DLP method are shown in
Table 7. As for the means in the 8-h urine pools, the means for all
time points through 48 h after isotope administration were signi-
cantly different from the fecal monitoring means, except at the 4-h
time point. No significant differences from the fecal monitoring
mean were observed at any time points from 72 to 168 h after iso-
tope administration, although the DLP mean was consistently higher
than the fecal monitoring mean. Beginning at the 4-h time point, no
signicant differences were observed when DLP-determined values
were compared with DLU-determined values from the 8-h, 24-h,
and 3-d urine pools after the first pool of each.
TABLE 5
Fractional apparent (MgAA) and true (MgA) magnesium absorption
values determined in the 6 subjects by the fecal monitoring method and
by deconvolution analysis (DP) and fecal endogenous excretion (FEE)
Fecal monitoring
Subject MgAA MgA MgA by DP FEE
%/d
1
1 0.47 0.49 0.49 2.0
2 0.40 0.42 0.41 1.9
3 0.48 0.50 0.41 2.5
4 0.47 0.51 0.49 3.4
5 0.52 0.55 0.55 1.9
6 0.39 0.41 0.44 2.6
x

SD 0.46 0.05 0.48 0.05 0.47 0.06 2.4 0.6

1
Percentage of the absorbed dose of isotopic tracer excreted daily.

a
t

U
n
i
v

d
e

S
a
o

P
a
u
l
o

U
S
P

o
n

D
e
c
e
m
b
e
r

2
3
,

2
0
0
8

w
w
w
.
a
j
c
n
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

METHODS FOR DETERMINING MAGNESIUM ABSORPTION 1211
DISCUSSION
After a 6-d adaptation period to the diet, magnesium balance
was negative in 2 of 3 of the subjects studied. However, on aver-
age, the magnesium balance of the group was not different from
zero. The dietary intake of magnesium was below the new dietary
reference intake of 400420 mg Mg/d for men (21). One inter-
pretation of the balance data is that balance does not determine
the requirement for a mineral element but is the intake required to
maintain the existing pool (22). Similar results were obtained in a
previous study after 7 d of adaptation to the experimental diet fol-
lowed by 14 d of sample collection. The investigators concluded
that their study was probably too short (23). It has been suggested
that 4 wk are needed to achieve a reliable estimate of magne-
sium balance (24).
The primary goal of the current study was to compare different
techniques for determining MgA to identify a simpler alternative
to fecal monitoring. We chose the results of the fecal monitoring
method as the reference because until now it was the most com-
monly used method (1013). This method evaluates MgAA, or
MgA after correction for FEE. The correction requires that a sec-
ond stable isotope be administered intravenously, and a complete
collection of fecal samples for 5 d is required. This method of
sampling is time consuming. In our study, DL and several intervals
of sample collection were tested. MgA calculated from the first
urine pool of 72 h was significantly lower than the value obtained
with the reference fecal monitoring method. This finding agrees
with previously reported results, showing that MgA was 80%
complete at 24 h and 95% complete at 72 h. This finding suggests
that absorption was underestimated when based on the 72-h urine
collection after isotope administration (25). Nevertheless, investi-
gators concluded that MgA might be measured with the DL
method and a complete 72-h urine collection without comparison
with another method or different time intervals. When compared
with the results of the reference method, our results showed that
it was necessary to wait for the second collection of 72 h (between
72 and 144 h) to closely evaluate MgA when urine was pooled by
3-d periods of time.
The analysis of the 24-h urine pools showed that the results
agreed with those of the reference method 24 h after isotope
administration. Results from the 8-h urine pools showed that only
2 time intervals (4856 and 5664 h) were adequate for the MgA
determination. A similar study carried out in rats found lower
MgA with the DL method than with a method based on fecal data
(14). The explanation for this difference is that it is assumed that
the 2 tracers behave the same once they are administered; how-
ever, the 2 tracers enter the circulation by different routes and may
not have behaved the same. Because plasma magnesium home-
ostasis is controlled by the kidney (26), the intravenous dose may
lead to transient hypermagnesemia, which is responsible for a
rapid initial urinary excretion of the tracer. This hypothesis would
explain an underestimation of MgA when compared with the fecal
data. The disagreement between urinary and fecal values obtained
in rats could also be linked to the protocol of isotope administra-
tion or to the administered amounts (14). A period of 2 h was
observed between the oral and the intravenous administration in
the rat study compared with an interval of 15 min in the current
study. The ratio between the intravenously administered dose and
the volume of distribution of magnesium was higher in the rat
study than the one calculated for this human study: 21% com-
pared with 15%. In both cases the volume of distribution was cal-
culated by using a compartmental model (M Sabatier, F Pont,
MJ Arnaud, JR Turnlund, unpublished observations, 2002; 27).
Thus, the injected dose could explain the difference between the
results of the fecal monitoring and DL methods in rats. Further-
more, the dose administered orally in rats was 5.4 times greater
than the amount administered intravenously. The ratio of admin-
istered amounts of tracers in this study was 2.3, as recommended
previously (25). Those observations tend to show that the results
are method dependent. The unusable initial fractions of time for
MgA determination represent the equilibration period required for
both tracers to behave in a similar way. Furthermore, diurnal vari-
ation of magnesium excretion (28) could have contributed to
invalid results of urine pooled by 8 h during the first 48-h period
of time. Three-day urine pools and early 24-h pools are usable for
the MgA determination when collected after the equilibration
period. Good agreement between the DL and fecal monitoring
methods was observed with the 24-h urine pools collected 24 h
after isotope administration. The closest agreement in the current
study was obtained between 48 and 144 h.
MgA values determined with the DLP method were not signi-
ficantly different from the values determined with the DLU
method or with the values determined by fecal monitoring 72 h
after isotope administration. Therefore, MgA can be determined
from a simple blood draw with the DL method. This method has
the disadvantage of being more invasive than urine collections.
The DP method is also invasive and, in addition, requires multi-
ple blood draws for kinetics modeling of the evolution of both
tracer concentrations. Even though the results obtained from the
DP analysis were not significantly different from the fecal data,
this method is cumbersome and necessitates the use of special
software. However, the result of the DP analysis could be more
valuable in studies that compare populations. When using the
DL method, magnesium absorption is calculated from a short
urine collection or blood sample. Consequently, the calculation
is based on the assumption that the rate of absorption is invari-
ant between persons with respect to physiologic state and diet
history and the assumption that the distribution of the intra-
venous labels is invariant with age or physiologic state. This
method cannot, however, be excluded for use in studies of
changes of response within similar groups. Another study, which
compared methods for determining zinc absorption, recom-
mended the use of urine collected 2 d after tracer administration
when the DLU method is used (19).
The rate of MgAA, when administered in water and consumed
with a breakfast containing 101 mg Mg (isotope included), was
0.46 0.05 in men. An experiment studying the MgAA from
water showed fractional absorption of 0.52 0.04 in women when
water was consumed with a breakfast containing 87 mg Mg (iso-
topes included) (11). The absolute amount absorbed in both cases
was the same. The results from studies that compared MgAA from
water are similar. The composition of the breakfasts, although dif-
ferent in both studies, did not seem to have an effect on MgAA.
In summary, the DL method was successful in measuring
MgAas it has been for calcium (29, 30) and zinc (19, 31). The
DLU and DLP methods were validated by comparing their
results with those of the fecal monitoring method. The simplest
and least invasive procedure was the DLU method. The most
similar results were obtained from the 24-h urine pools collected
between 48 and 144 h after isotope administration. An additional
advantage of the DLU method over the fecal monitoring method
is that the measurement of magnesium isotopic ratios in urine by
ICP-MS is less cumbersome than is that in fecal samples; these

a
t

U
n
i
v

d
e

S
a
o

P
a
u
l
o

U
S
P

o
n

D
e
c
e
m
b
e
r

2
3
,

2
0
0
8

w
w
w
.
a
j
c
n
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m

1212 SABATIER ET AL
results are valid under the conditions of this study. The DP
method is probably more appropriate for comparison of absorp-
tion across populations. To decrease the cost of such experi-
ments, additional studies should be done to optimize the doses of
oral and intravenous tracers and to bring the amounts adminis-
tered closer to tracer doses.
MS contributed to the data collection, sample and data analysis, and writ-
ing of the manuscript. WRK contributed to the sample and data collection and
analytic methods. FP contributed to the deconvolution analysis. MJA con-
tributed to the study design and writing of the manuscript. JRT contributed to
the experimental design, data collection and analysis, and writing of the man-
uscript. MS was supported by the Nestl Water Institute. MJA is the director of
the Nestl Water Institute. FP is employed by INSERM (France). JRT and WRK
are employed by the US Department of Agriculture.
REFERENCES
1. Shills ME. Magnesium in health and disease. Annu Rev Nutr 1988;
8:42960.
2. Rude RK. Magnesium deficiency: a cause of heterogeneous disease
in humans. J Bone Miner Res 1998;13:74958.
3. Rosman KJR, Taylor PDP. Isotopic composition of elements. Pure
Appl Chem 1998;70:21735.
4. Currie VE, Lengemann FW, Wentworth RA, Schwartz R. Stable
26
Mg
as an in vivo tracer in investigation of magnesium utilization. Int J
Nuclear Med Biol 1975;2:15964.
5. Schwartz R, Giesecke CC. Mass spectrometry of a volatile Mg
chelate in the measurement of stable
26
Mg when used as a tracer. Clin
Chim Acta 1979;97:18.
6. Turnlund JR, 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 1990;7:11745.
7. Schuette S, Vereault D, Ting BTG, Janghorbani M. Accurate meas-
urement of stable isotopes of magnesium in biological materials with
inductively coupled plasma mass spectrometry. Analyst 1988;113:
183742.
8. Cary EE, Wood RJ, Schwartz R. Stable Mg isotopes as tracers using
ICP-MS. J Micronutr Anal 1990;8:1322.
9. Benech H, Pruvost A, Batel A, Bourguignon M, Thomas JL, Grognet JM.
Use of stable isotopes technique to evaluate the bioavailability of a phar-
maceutical form of magnesium in man. Pharm Res 1998;15:34751.
10. Wauben IP, Atkinson SA, Bradley C, Halton JM, Barr RD. Magne-
sium absorption using stable isotope tracers in healthy children and
children treated for leukemia. Nutrition 2001;17:2214.
11. Sabatier M, Arnaud MJ, Kastenmayer P, Rytz A, Barclay DV. Meal
effect on magnesium bioavailability from mineral water in healthy
women using stable isotopes. Am J Clin Nutr 2002;75:6571.
12. Schuette SA, Lashner BA, Janghorbani M. Bioavailability of magne-
sium diglycinate vs magnesium oxide in patients with ileal resection.
JPEN J Parenter Enteral Nutr 1994;18:4305.
13. Turnlund JR, Michel MC, Keyes WR, King JC, Margen S. Use of
enriched stable isotopes to determine zinc and iron absorption in eld-
erly men. Am J Clin Nutr 1982;35:103340.
14. Coudray C, Pepin D, Tressol JC, Bellanger J, Rayssiguier Y. Study
of magnesium bioavailability using stable isotopes and the induc-
tively-coupled plasma mass spectrometry technique in the rat: single
and double labeling approaches. Br J Nutr 1997;77:95770.
15. Lentner C, Lentner C, Wink A, eds. Geigy scientific tables. West
Caldwell, NJ: 1984;3:66.
16. Anderson KA, Talcott PA. Magnesium. In: Seiler HG, Sigel A, Sigel H, eds.
Handbook on metals in clinical and analytical chemistry. New York:
Marcel Dekker, Inc, 1994:45366.
17. Miller JC, Miller JN. Statistics for analytical chemistry. 2nd ed. New
York: Ellis Horwood, 1988.
18. Yergey AL, Abrams SA, Vieira NE, Aldroubi A, Marini J, Sidbury JB.
Determination of fractional absorption of dietary calcium in humans.
J Nutr 1994;124:67482.
19. Lowe NM, Woodhouse LR, Matel JS, King JC. Comparison of esti-
mates of zinc absorption in humans by using 4 stable isotopic
tracer methods and compartmental analysis. Am J Clin Nutr 200;
71:5239.
20. SAS Institute, Inc. SAS/STAT users guide, version 6. Cary, NC: SAS
Institute, Inc, 1989.
21. Institute of Medicine. Dietary reference intakes for calcium, phos-
phorus, magnesium, vitamin D and uoride. Washington, DC: National
Academy Press, 1997.
22. Mertz W. Use and misuse of balance studies. J Nutr 1987;117:18113.
23. Knudsen E, Sandstrom B, Solgaard P. Zinc, copper and magne-
sium absorption from a fibre-rich diet. J Trace Elem Med Biol
1996;10:6876.
24. Schwartz R, Apgar BJ, Wien EM. Apparent absorption and retention
of Ca, Cu, Mg, Mn, and Zn from a diet containing bran. Am J Clin
Nutr 1986;43:44455.
25. Abrams SA, Wen JP. Methodologies for using stable isotopes to
assess magnesium absorption and secretion in children. J Am Coll
Nutr 1999;18:305.
26. Quamme GA. Renal magnesium handling: new insights in under-
standing old problems. Kidney Int 1997;52:118095.
27. Feuillet-Coudray C, Coudray C, Gueux E, et al. Compartmental
analysis of magnesium kinetics in Mg-sufficient and Mg-deficient
rats. Metabolism 2000;49:13269.
28. Darlu P. Proceedings: characteristics and reproducibility of circadian
rhythms of urinary excretion in man. J Physiol (Paris) 1973;67:338A
(abstr).
29. Yergey AL, Vieira NE, Covell DG. Direct measurement of dietary
fractional absorption using calcium isotopic tracers. Biomed Environ
Mass Spectrom 1987;14:6037.
30. Smith DL, Atkin C, Westenfelder C. Stable isotopes of calcium as
tracers: methodology. Clin Chim Acta 1985;146:97101.
31. Friel JK, Naake VL Jr, Miller LV, Fennessey PV, Hambidge KM. The
analysis of stable isotopes in urine to determine the fractional absorp-
tion of zinc. Am J Clin Nutr 1992;55:4737.

a
t

U
n
i
v

d
e

S
a
o

P
a
u
l
o

U
S
P

o
n

D
e
c
e
m
b
e
r

2
3
,

2
0
0
8

w
w
w
.
a
j
c
n
.
o
r
g
D
o
w
n
l
o
a
d
e
d

f
r
o
m