Você está na página 1de 3

Nutritional MethodologyResearch Communication

(Manuscript received 14 November 2002. Initial review completed


20 December 2002. Revision accepted 8 January 2003.)
Robert P. Heaney2
Osteoporosis Research Center, Creighton University, Omaha, NE

ABSTRACT Calcium absorption was measured simultaneously by tracer and pharmacokinetic methods in 12 men.
The purpose of the study was to calibrate the area under
the curve (AUC) for absorptive calcemia to permit comparison or pooling of pharmacokinetic data with tracer-derived
estimates. Each subject was studied twice during intake of
500 mg calcium loads (as the carbonate salt), once without
excipients, and once with a binder that impeded dissolution
of the salt, reducing its absorbability and thereby providing
a broad range of absorption values over which to calibrate
the method. Various time periods were evaluated, with the
best prediction being given by the AUC for the increment in
serum calcium calculated over 9 h (AUC9). For net absorbed
calcium (mmol) the relationship was given by 4.358 AUC9
0.820. The empirical regression coefficient was significant and the error of the estimate (0.820 mmol) acceptably
small. J. Nutr. 133: 1224 1226, 2003.

SUBJECTS AND METHODS


Subjects. Participants were 12 healthy men, 35.9 5.1 y old,
each studied twice using sources with differing absorbability (see
below). The protocol was approved by the Creighton University
Institutional Review Board and each participant gave written consent.
Protocol. The test calcium source was ingested in the morning,
midway through a low calcium breakfast providing 1.88 MJ and 1.4
mmol calcium. Blood was drawn just before feeding and then at
frequent intervals throughout the day (0.5, 2, 3, 5, 7, 9, 12 and 24 h).
A low calcium lunch (2.3 mmol) was consumed after the 5-h
sample was taken. The median interval between duplicate tests in any
given subject was 14 d. Except for control of calcium intake during
the test day, there were no other dietary restrictions. The test calcium
load was 12.5 mmol (500 mg), given as calcium carbonate formulated
either as the pure salt loosely packed into gelatin capsules or as
calcium tablets containing a resin binder that interfered with calcium
carbonate disintegration/dissolution, thereby lowering its absorbability. The calcium carbonate for both sources was intrinsically labeled
by the addition of 45CaCl2 to a solution of calcium chloride and then
by precipitating the salt with a slight stoichiometric excess of sodium
carbonate. The resulting powder was captured on a fritted glass filter,
washed with distilled water and dried at 100C before encapsulation
or tableting. Tracer doses were on the order of 5 Ci (0.185 MBq) per
test.
Measurements. Serum stable calcium was measured in duplicate
by atomic absorption spectrophotometry [AAnalyst 100, PerkinElmer, Norwalk, CT (within-assay precision: 0.73%)] and serum 45Ca
by liquid scintillation counting (LS-3150T counter, Beckman Instruments, Fullerton, CA). The serum 45Ca values were expressed as
specific radioactivity [i.e., as fraction of the administered dose (fxdose) of tracer/mmol calcium)]. True fractional calcium absorption
was measured from the 5-h serum 45Ca concentration using established methods (3,4). Areas under the curves at various times after
dosing (AUCt) for both 45Ca and stable calcium were calculated
using the trapezoidal method. For both stable and radioactive calcium, we plotted the increment above baseline after source ingestion
as the basis for calculating AUC. The dimensions of the AUC are
(mmol h)/L for stable calcium and (fxdose h)/mmol for 45Ca.
Data analysis. Because the sampling unit in this investigation
was the individual test, not the participant, we used both tests in each
subject to produce a total sample size of 24 tests, with a nearly
continuous range of fractional absorption values varying from 0.131

KEY WORDS: calcium absorption

calcium absorption measurement pharmacokinetic methods

area under the curve

True calcium absorption, i.e., unidirectional flux of calcium


from gut lumen into blood, is most accurately measured using
an oral calcium source intrinsically labeled with a suitable
calcium isotope and quantifying the tracer that appears in
blood, urine or body compartments after absorption is complete. The basic theory for this approach was devised by
Bronner (1), an applied method described by deGrazia et al.
(2) and various short-cut approximations developed by
Heaney, Recker and others (35). But tracer methods are not
applicable to marketed products nor to many food sources of
calcium because they often cannot be easily labeled.
Although extracellular fluid calcium concentration is
tightly regulated, calcium absorption nevertheless does produce a measurable, if small, degree of calcemia that can be
captured by pharmacokinetic methods [specifically, measure-

Supported by Creighton University research funds.


To whom correspondence should be addressed.
E-mail: rheaney@creighton.edu.
2

3
Abbreviations used: AbsFx, absorption fraction; AUC, area under the curve;
fxdose, fraction of the administered dose.

0022-3166/03 $3.00 2003 American Society for Nutritional Sciences.

1224

Downloaded from jn.nutrition.org at WAGENINGEN UNIV AND RESEARCH CENTRE on June 3, 2013

ment of the area under the curve (AUC)].3 The relationship


between the quantity of calcium absorbed and such pharmacokinetic variables as AUC is more complex than usually
encountered when measuring, for example, drug absorption.
This is the case because the calcemia of absorption evokes
physiologic responses that reduce calcium input into the blood
from bone, thereby damping the calcemia of absorption. Nevertheless, the degree of calcemia that does occur would be
expected to reflect the amount of calcium absorbed. Our goal
in this investigation was to determine how well that relationship holds and to derive suitable parameters for estimating the
absorption fraction from calcium pharmacokinetic data.

Quantifying Human Calcium


Absorption Using Pharmacokinetic
Methods1

QUANTIFYING CALCIUM ABSORPTION

1225

to 0.439. AUC were calculated for different time intervals, and


standard Pearsonian regression was used to derive parameters for the
relationship between AUC and the true absorption fraction, as well
as to test for the AUC interval that produced the best fit.

RESULTS

AbsFx 0.3126 AUC9 0.0697 0.0633

FIGURE 1 Plot of the incremental calcemia in men after ingestion


of a 12.5 mmol Ca load (A) for the 12 tests employing plain calcium
carbonate and (B) for the incremental Ca specific activity (SpecAct)
after ingestion of a 45Ca-labeled 12.5 mmol load. The error bars represent 1 SEM. The dimension for SpecAct is fraction of dose of tracer per
mmol Ca. (Copyright Robert P. Heaney, 2003. Used with permission.)

FIGURE 2 Plot of absorption fraction (AbsFx) calculated from the


5-h specific activity value against area under the curve to 24 h (AUC24)
for serum specific radioactivity for all 24 tests in 12 men. (Copyright
Robert P. Heaney, 2003. Used with permission.)

The coefficient of correlation was 0.66 (P 0.001).


Additionally, AUC9 provided the lowest estimate for the
y-axis intercept of any of the AUC intervals tested. This
intercept should, in theory, be zero. In fact, for AUC9, the
calculated intercept was not different from zero; thus the
equation could be recomputed by forcing the line through the
origin, resulting in the following equation:
AbsFx 0.4088 AUC9 0.0643
The resulting degradation in the coefficient of correlation was
minor (from 0.66 to 0.63).
If all calcium sources were tested at a 12.5 mmol load size,
Equation 2 would suffice, because it produces results directly
commensurate with the tracer methods at that load. However,
the calcemia after absorption is produced by the absolute
quantity absorbed, and not by the fraction absorbed. To make
provision for different load sizes, it is necessary to apply a
scaling factor having a dimension of mass equivalents (mmol
or g) to the factors in Equation 2. Here a distinction between
gross absorption (unidirectional) and net absorption (net difference between bidirectional fluxes) is necessary. The former
is what is measured by the tracer methods, whereas the latter
(if 0) is what elevates serum calcium. Because digestive
juices contain appreciable quantities of calcium, the enteral
load is always higher than the ingested load. We have shown

FIGURE 3 Plot of absorption fraction (AbsFx) calculated from the


5-h calcium specific activity value against area under the curve to 9 h
(AUC9) calculated for the rise in total serum calcium for all 24 tests in 12
men. (Copyright Robert P. Heaney, 2003. Used with permission.)

Downloaded from jn.nutrition.org at WAGENINGEN UNIV AND RESEARCH CENTRE on June 3, 2013

Serum calcium began to rise by 1 h, peaked at 5 h, and had


returned to within 1.5% of the starting baseline value by 12 h
(Fig. 1A), when the calcemia was no longer significantly
different from baseline. At its peak, the calcemia averaged
only slightly 0.125 mmol/L (0.5 mg/dL), or a rise of 5%.
The serum tracer pattern was similar except that residual
tracer remained easily detectable after the total calcium had
returned to baseline (Fig. 1B).
When the 5-h calcium specific activity value was plotted
against the 24-h AUC for the tracer specific activity (Fig. 2),
r2 was 0.92, and the residual variation around the regression
line was within the precision error of the 45Ca measurement at
the tracer dose used. This very close fit indicates that the 5-h
value captured essentially all of the information available from
a full computation of the AUC. Such validation is useful in
this context because it is the 5-h method that has been
calibrated (3,4) against the double tracer method of deGrazia
et al. (2), rather than the AUC.
AUC for serum stable calcium was calculated over periods
of 5, 7, 9 and 12 h, by which time, any residual calcemia had
all but completely dissipated (Fig. 1A). The true absorption
fraction (AbsFx), calculated in each individual from the 5-h
specific activity value, was plotted against the AUC calculated
for the four time intervals. Although all correlations were
significant, the best fit was given for AUC at 9 h (Fig. 3). The
corresponding linear regression equation is as follows:

HEANEY

1226

NetAbsCa mmol 4.358 AUC9 0.820


DISCUSSION
The results presented here provide a means for quantifying
calcium absorption from sources that are not easily labeled,
and for producing values that can be compared or integrated
with those derived using tracer methods. This approach may
be particularly applicable to commercially marketed products,
such as supplements or heavily fortified foods, as well as to
natural food sources, none of which may be readily susceptible
to labeling.
Although in this study we utilized AUC calculated out to
24 h, it was clear that 9 h suffices, and that the curve of
absorptive calcemia can be captured by approximately six
blood samples (e.g., 0, 1, 3, 5, 7 and 9 h). At the same time it
must be noted that this pharmacokinetic approach is inherently more time consuming and expensive than the simplified
tracer methods, which require only a single blood sample at
5 h (the method applied in this study), or a urine specimen
pooled over a 48-h period [the method used by Yergey et al.
(5)]. On the plus side, measurement of absorptive calcemia is
unaffected by the inverse relationship between load size and
absorption fraction described previously (7,8), which is a characteristic feature of the tracer-based estimates.
Although the load employed in this study was constant at
12.5 mmol, the amount absorbed varied from as little as 0.95
to as much as 5.03 mmol (the extreme values on the y-axis in
Fig. 4). This range was a result of differences intrinsic to both
the source and the subjects. Because it is the amount absorbed
that produces the calcemia, the approach described in this
paper should be applicable to virtually any size load producing
calcemia within this range, so long as it produces a measurable
degree of calcemia. However, although applicable in theory to
loads of any size, it is apparent that smaller loads will inevitably produce less calcemia, and hence will require correspondingly larger sample sizes to produce detectable signals
with sufficiently narrow ranges of uncertainty to permit useful

FIGURE 4 Plot of net absorbed calcium (NetAbsCa) against area


under the curve to 9 h (AUC9) for the rise in total serum calcium for all
24 tests in 12 men. (Copyright Robert P. Heaney, 2003. Used with
permission.)

analysis. Even at the 12.5 mmol load that we employed, AUC9


varied from a low of 1.41 (mmol h)/L (theoretical minimum
AUC: 0.0) to a high of 4.14 (mmol h)/L. An AUC9 of 1.41
(mmol h)/L means an average calcemia over the 9 h after
ingestion of only 0.04 mmol/dL (0.16 mg/dL), or just 1.7%
above baseline. With a signal this small and inevitably some
fluctuation in baseline, low AUC values carry a relatively
much broader uncertainty range. The degree of dispersion of
the values around the regression line is shown graphically in
Figures 3 and 4. Note how much larger this dispersion is,
relative to the corresponding dispersion for the tracer data in
Figure 2. The reasons for the contrast are first that the signalto-noise ratio for the detection of the isotope is very much
higher than for the detection of the absolute calcemia, and
second because the body does not regulate calcium isotope
concentration as such. Hence the rise in tracer concentration
is due mainly to absorption and is less influenced by offsetting
effects from intrabody calcium sources, as is the case with
carrier calcium.
Finally, because the 5-h absorption method incorporates an
empirical correction for body size and has itself been validated
only for individuals with body mass indices mainly in the range
from 17 to 32 kg/m2, it follows that the calibration described
here is itself valid only within that same range.
LITERATURE CITED
1. Bronner, F. (1962) Experimental studies of calcium absorption in man.
Bibl. Nutr. Dieta. 3: 2231.
2. DeGrazia, J. A., Ivanovich, P., Fellows, H. & Rich, C. (1965) A doubleisotope method for measurement of intestinal absorption of calcium in man.
J. Lab. Clin. Med. 66: 822 829.
3. Heaney, R. P. & Recker, R. R. (1985) Estimation of true calcium
absorption. Ann. Intern. Med. 103: 516 521.
4. Heaney, R. P. & Recker, R. R. (1988) Estimating true fractional calcium
absorption. Ann. Intern. Med. 108: 905906.
5. Yergey, A. L., Abrams, S. A., Vieira, N. E., Aldroubi, A., Marini, J. &
Sidbury, J. B. (1994) Determination of fractional absorption of dietary calcium
in humans. J. Nutr. 124: 674 682.
6. Heaney, R. P. & Recker, R. R. (1994) Determinants of endogenous
fecal calcium in healthy women. J. Bone Miner. Res. 9: 16211627.
7. Heaney, R. P., Weaver, C. M. & Fitzsimmons, M. L. (1990) The influence of calcium load on absorption fraction. J. Bone Miner. Res. 11: 11351138.
8. Blanchard, J. & Aeschlimann, J. M. (1989) Calcium absorption in
mansome dosing recommendations. J. Pharmacokinet. Biopharm. 17: 641
644.

Downloaded from jn.nutrition.org at WAGENINGEN UNIV AND RESEARCH CENTRE on June 3, 2013

elsewhere that total flux into the gut from endogenous calcium
pools varies with height (m) and is 2.16 mmol/(m d) (6),
or, for the participants in this study with a mean height of
1.81 m, 3.91 mmol/d. On the assumption that digestive juice
calcium flux into the lumen is proportionate to food intake,
and further assuming that the breakfast (at 1.88 MJ) was
20% of total daily energy intake for these men, 20% of the
daily digestive juice calcium (i.e., 0.78 mmol) would have
entered the gut from the blood, for a total enteral load of 12.5
0.78 mmol, or 13.28 mmol. In these experiments, with a
mean true absorption fraction of 0.283, transfer from lumen
into blood would be given by 0.283 13.28, or 3.76 mmol.
The offsetting digestive juice flux (0.78 mmol) reduces the
mean net inward flux to 2.98 mmol. It is this net absorbed
quantity that produces the calcemia measured as AUC.
The foregoing calculations were performed for each test in
each participant, using the subjects actual height and true
absorption fraction.
The corresponding plot of net absorbed calcium (NetAbsCa,
given as AbsFx load) is presented as Figure 4. The equation
is as follows: