special communication
Rapid and accurate 13CO2 isotopic measurement
in whole blood: comparison with expired gas
MARTIAL DANGIN,1,2 JEAN CLAUDE DESPORT,1 PIERRE GACHON,1
AND BERNARD BEAUFRÈRE1
1Laboratoire de Nutrition Humaine, Université Clermont Auvergne, Centre de Recherche
en Nutrition Humaine, 63009 Clermont-Ferrand, Cedex 1, France;
and 2Nestlé Research Center, CH-100 Lausanne 26, Switzerland
carbon dioxide; bicarbonate; isotopic fractionation; mass spectrometry
THE MEASUREMENTS OF 13CO2 enrichment are required in
numerous studies to estimate either nutrient oxidation
rates (17), CO2 production (9), or assimilation of various 13C-labeled substrates (7; i.e., breath tests). For
these purposes, 13CO2 atom percent (AP) is usually
determined by isotope ratio mass spectrometry on
breath samples collected at sequential time points (17).
However, voluntary breath sampling can be difficult or
impossible to perform in various situations such as
ventilatory abnormalities, artificially ventilated paThe 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.
E212
tients, or in specific populations such as newborns or
patients with neurological disorders (8, 14). Thus two
alternatives have been proposed to solve this problem.
The first one is to sample gas from the ventilated hood
of an indirect calorimeter (8). This method is not widely
used, in particular because large volumes of gas are
needed for isotope ratio mass spectrometry analysis
due to atmospheric air dilution. In addition, this technique is limited by the mandatory utilization of the
indirect calorimetry system, which is not required for
breath test or for the estimation of CO2 production by
an isotopic dilution method. The second method (4, 6,
10, 14, 15) analyzes the gaseous CO2 released from
blood by acid addition. Compared with the breath
method, this method is more demanding due to operator manipulations. For diagnosis purposes, it is important that blood samples can be processed rapidly,
reliably, and with minimal operator manipulations.
Considering that 1) blood CO2 is spontaneously released in the air and 2) the sensitivity of the isotope
ratio mass spectrometer has dramatically increased
over the past few years, it should be possible to improve
the latter method by the measurement of 13CO2 AP
spontaneously released from blood (e.g., without acid
addition). Thus the aim of the present work was to
determine whether CO2 released from the blood without acid addition can be used to measure 13CO2 AP. The
technical aspects of this methodology were validated,
and simultaneous comparisons for breath, blood with
acid, and blood without acid were made at each sampling point.
SUBJECTS AND METHODS
Subjects and experimental protocols. Blood and gas samples
were obtained from nine young healthy subjects (age 30.6 6
11.7 yr, weight 63.9 6 5.4 kg, height 171 6 7 cm, mean 6 SD)
who participated in two different protocols that were previously described in detail (2, 11). The protocols were approved
by the ethical committee of Clermont-Ferrand. Briefly, five
subjects received (11) a primed continuous constant intravenous infusion of L-[1-13C]leucine for 10 h after a prime dose of
[13C]bicarbonate, and samples were collected before the tracer
infusion and when the 13CO2 AP had reached a plateau (510,
540, 580, and 600 min). The four other volunteers (2) received
a single oral load of 30 g of whey protein added with free
L-[1-13C]leucine (13 mmol/kg). Samples were taken before and
0193-1849/99 $5.00 Copyright r 1999 the American Physiological Society
Downloaded from http://ajpendo.physiology.org/ by 10.220.32.247 on July 31, 2017
Dangin, Martial, Jean Claude Desport, Pierre
Gachon, and Bernard Beaufrère. Rapid and accurate
13CO isotopic measurement in whole blood: comparison with
2
expired gas. Am. J. Physiol. 276 (Endocrinol. Metab. 39):
E212–E216, 1999.—Determination of 13CO2 enrichment on
the CO2 released from blood by acid has been used in
situations in which breath sampling is difficult. This method
can be improved by measuring this enrichment on the CO2
spontaneously released from blood. Therefore, simultaneous
comparisons of 13CO2 content between breath and arterialized blood added with or without acid were performed in 51
samples from human studies, using the statistical method of
Bland and Altman (J. M. Bland and D. G. Altman. Lancet 1:
307–310, 1986). Strong relationships exist between the methods (r . 0.99) expressed in atom percent excess (APE).
Compared with breath, the acid method overestimates the
13CO enrichment (0.318 6 0.632 APE 3 1,000, P , 0.001).
2
The acid-free method shows similar enrichments to breath
(0.003 6 0.522 APE 3 1,000, P 5 0.97) with good precision
and degree of agreement (95% confidence interval 0.15 APE 3
1,000). The analysis can be performed up to 5 days after
sampling with a good reproducibility. In conclusion, measuring 13CO2 enrichments on the CO2 spontaneously released
from blood is feasible, gives identical results to the breath
method, and lessens operator manipulations. It allows study
of situations in which the breath sampling method is not
feasible.
13CO
2
DETERMINATION IN WHOLE BLOOD
d‰ vs. REF 5
RSAM 2 RREF
RREF
3 1,000
In this study, the reference gas had a d13C of 231.00 relative
to the international standard reference Pee Dee Belemnite
(RPDB 5 0.0112372). This calibration was obtained by periodic
comparison with other laboratories.
The results obtained in these conditions were corrected to
express the values in d‰ vs. PDB. The values calculated were
then converted in AP using the following formula
AP 5
5
d‰
11,000 1 12 3 R
d‰
311,000 1 12 3 R 4 1 1
PDB
PDB
6
3 100
Finally, the 13CO2 atom percent excess (APE) was obtained by
subtracting the background AP (before tracer administration)
from the sample AP (during or after tracer administration).
Results obtained were expressed in d‰ vs. PDB, AP, and in
APE 3 1,000.
Statistical analysis. Comparison between the methods was
tested using the statistical method described by Bland and
Altman (1). The statistics (Student’s t-test, ANOVA) were
performed using Statwiew 4.02 from Abacus Concepts (Berkeley, CA). Results were expressed as means 6 SD.
RESULTS
Interassay and intra-assay reproducibilities and storage. Interassay and intra-assay reproducibilities of the
measurement on Breath, Blood, and Blood1LA were
tested as shown in Table 1. The area of the CO2 peak
derived from arterialized venous blood decreased by
60% compared with Breath or Blood1LA. Despite this
difference, the SD of Blood in the interassay replicate
(n 5 5 subjects) was similar to Breath, whereas the SD
of Blood1LA was slightly higher. Intra-assay reproducibility was good for all three methods, indicating that
up to five replicates can be performed in the same tube.
There was no statistical difference between the Breath
and the Blood methods. In contrast, Blood1LA was
significantly higher than Breath (P ,0.001, t-test).
When 1.5 ml of lactic acid were analyzed, traces of CO2
were detected, but the very low intensity of the peak did
not allow any measurement of the 13C isotopic content.
The storage of Blood samples at 220°C for 5 days was
also tested in another batch of blood samples. No
significant change in the 13CO2 isotopic ratio was found
(P 5 0.69, t-test) between Blood immediately treated
(220.28 6 0.08 d‰ vs. PDB) and Blood treated after 5
days of storage (220.30 6 0.08 d‰ vs. PDB). In
addition, the storage did not modify the reproducibility
of the isotopic analysis (SD 5 0.08 d‰ vs. PDB in each
case, n 5 5 subjects).
Relationship, precision, and agreement between the
methods for 13CO2 isotopic content. The 13C-to-12C isotopic ratio tested in Breath ranged from 225.66 to 26.05
d‰ vs. PDB, in Blood from 226.75 to 27.49 d‰, and in
Blood1LA from 221.5 to 22.79 d‰ vs. PDB.
As shown in Fig. 1, there were strong positive
relationships between Breath and Blood and between
Breath and Blood1LA expressed in AP (r . 0.99). In
the range of our values, Blood1LA was systematically
higher than Breath and Blood. Moreover, in both cases,
y-intercepts were not significantly different from zero
(0.30 , P , 0.49, ANOVA).
With the use of the Bland and Altman (1) plot
(difference between the 2 methods of measurement vs.
their mean at each time point), there was good precision and agreement between the AP of Breath and AP of
Blood 13CO2 isotopic ratios, as shown in Fig. 2. Indeed,
there was no statistical difference between the two
methods (0.10 6 0.71 AP 3 1,000, P 5 0.31, t-test for
Table 1. Interassay and intra-assay reproducibilities
Reproducibility
Breath
Blood 1 LA
Blood
Interassay
Intra-assay
224.66 6 0.06
219.30 6 0.13*
224.76 6 0.08
224.62 6 0.06
219.52 6 0.03
224.70 6 0.07
Results are expressed as means 6 SD. Shown is intra-assay and
interassay reproducibilities (5 replicates) of gas chromatography
mass spectrometry analysis [d‰ vs. Pee Dee Belemnite (PDB)]
performed in exhaled air (Breath), gaseous CO2 released from blood
by lactic acid (Blood 1 LA), and gaseous CO2 spontaneously released
from blood (Blood). * Significant difference (P , 0.05, t-test) between
interassay Breath and the other interassay methods.
Downloaded from http://ajpendo.physiology.org/ by 10.220.32.247 on July 31, 2017
after (at 60, 120, 180, 240, and 300 min) the administration of
oral tracer and were utilized for the present study. We chose
this latter study because the obtained 13CO2 AP both covered
and exceeded the normal range usually observed in classic
oxidation studies at steady state. A total number of 51 time
points were obtained. For each of these, expired gas and arterialized blood samples were collected simultaneously. Breath samples
were collected as previously described (6). Before blood sampling,
the hand was warmed for 15 min at 60°C in a heated
ventilated box as previously described (2). Immediately after
sampling, 1 ml of arterialized venous blood was injected in a
10-ml tube (Vacutainer; Becton-Dickinson, Meylan, France)
previously flushed with helium, and 1 ml was injected in a
10-ml tube also previously flushed and containing 0.5 ml of
0.2 M lactic acid (Merck, Darmstadt, Germany). An additional set of experiments was carried out, with lactic acid
alone, to determine whether lactic acid in the absence of blood
generates a signal. The tubes were then immediately agitated
(1 rotation/min) during 2 h at room temperature before the
isotope analysis. Measurements were performed on the following three sets of samples: breath, CO2 spontaneously released
by blood, and CO2 released from blood by lactic acid. Each
method is further referred to as ‘‘Breath,’’ ‘‘Blood,’’ and
‘‘Blood1LA,’’ respectively.
Analytical methods and calculations. The 13CO2 analyses
were performed by gas chromatography isotope ratio mass
spectrometry (µGas System; Fisons Instruments, VG Isotech,
Middlewich, UK). Briefly, a sample (40 µl) of the gaseous
content of each tube analyzed was automatically injected
(Gilson automatic sampler) in the gas chromatograph (Hewlett Packard 5890; Palo Alto, CA) in which the CO2 was
separated from the other components using a packed column
(Chrompack column, type HAYSEP Q, 2.5 m 3 1/8, 60/80
mesh; Les Ulis, France) at 100°C. The CO2 was then introduced in the isotope ratio mass spectrometer to measure the
13C-to-12C isotopic ratio of the sample (R
SAM ) versus a reference gas (RREF ). The 13C-to-12C ratio of the samples was
expressed as delta per mil (d‰) versus RREF after Craigs
corrections (REF; see Ref. 3)
E213
E214
13CO
2
DETERMINATION IN WHOLE BLOOD
difference) and no relationship between the difference
and the mean (P 5 0.56, ANOVA). However, the scatter
of the difference increased as 13CO2 enrichment increased. The 95% confidence interval ranged from
20.099 to 0.300 AP 3 1,000.
In contrast, the difference between AP of Blood1LA
and AP of Breath was always positive (Fig. 2) and was
statistically different from zero (5.86 6 0.66 AP 3
1,000, P , 0.001). In addition, despite the absence of a
relationship between the difference against their mean
(P 5 0.16), the scatter of the difference also increased as
the 13CO2 abundance increased. The 95% confidence
Fig. 2. Difference between Blood (s) or Blood1LA (p) and Breath
expressed in AP vs. their means. Dotted lines show means of the
difference for Blood and Breath (0.10 AP 3 1,000) and the difference
between Blood1LA and Breath (5.86 AP 3 1,000).
DISCUSSION
Our intent was to develop an automated technique
for measuring 13CO2 APE in conditions in which the
Fig. 3. Relationships between Blood and Breath (s) or Blood1LA
and Breath (p) expressed as 13CO2 AP. Regression lines were as
follows: Blood 13CO2 atom percent excess (APE) 5 1.003x 1 0.015,
r 5 0.996; Blood1LA 13CO2 APE 5 1.0297x 1 0.131, r 5 0.995.
Downloaded from http://ajpendo.physiology.org/ by 10.220.32.247 on July 31, 2017
Fig. 1. Relationships between CO2 spontaneously released by blood
(Blood) and breath (Breath) (s) or CO2 released from blood by lactic
acid (Blood1LA) and Breath (p) expressed as 13CO2 atom percent
(AP). Regression lines were as follows: Blood 13CO2 AP 5 0.982x 1
0.019, r 5 0.993; Blood1LA 13CO2 AP 5 1.017x 1 0.019, r 5 0.994.
interval was always positive despite the precision
(range from 5.678 to 6.049 AP 3 1,000).
Relationship, precision, and agreement between the
methods for 13CO2 isotopic enrichments. Regression line
equations for APE of Breath vs. APE of Blood and APE
of Breath versus APE of Blood1LA are given in Fig. 3.
The slope for APE of Breath versus APE of Blood
(1.0003 6 0.013, 0.2 , P , 0.1) and the y-intercept
(20.015 6 0.110 APE 3 1,000, P 5 0.89), statistically,
did not differ from the median slope and zero, respectively. In contrast, for APE of Breath versus APE of
Blood1LA, the slope was different from the median
slope (1.029 6 0.015, P , 0.001), whereas the yintercept was not different from zero (0.131 6 0.129
APE 3 1,000, P 5 0.31).
The difference between APE of Blood and APE of
Breath (Fig. 4A, 0.003 6 0.522 APE 3 1,000) did not
appear to be biased (P 5 0.97, t-test of the difference),
and there was no relationship between the difference
vs. their mean (P 5 0.60). Despite this observation, the
difference increased as the 13CO2 enrichments increased. However, the 95% confidence interval (20.144
to 0.150 APE 3 1,000) was small enough to ensure that
the method based on the free CO2 released from blood
can be used with an acceptable precision and degree of
agreement.
In contrast (Fig. 4B), the difference between APE of
Blood1LA and APE of Breath (0.318 6 0.632 APE 3
1,000, P , 0.001; 95% confidence interval 5 0.141 to
0.496 APE 3 1,000) is significantly different from zero.
In addition, there was a positive statistical relationship
between the difference (APE of Blood1LA 2 APE of
Breath) vs. their mean (P 5 0.02), indicating a slight
overestimation of the enrichments at higher values.
13CO
2
DETERMINATION IN WHOLE BLOOD
breath method is difficult or impossible to perform. For
this purpose, we have compared the results obtained by
our method with the analysis performed in expired gas
(considered as the reference method) and in gaseous
CO2 released from blood by acid addition (4, 6, 10,
14, 15).
Comparison between exhaled air and gaseous CO2
released from blood by acid revealed higher 13CO2 AP in
whole blood, which is in close agreement with previously published works (10, 14). Because lactic acid
solution per se is not a significant provider of CO2, this
difference cannot be due to lactic acid itself. Therefore,
it implies that there is an isotopic fractionation during
the reaction leading to CO2 production. This reaction is
the following
CO2(gas) ` CO2(dissolved)
CO2(dissolved) 1 H2O ` H1 1 HCO2
3
the latter step being catalyzed in vivo by carbonic
anhydrase. When lactic acid is added to a test tube
containing blood, the equilibrium of the reaction is
modified to produce gaseous CO2, and the relative
contribution of carbonic anhydrase is strongly reduced.
Therefore, the 13C content measured on the released
CO2 primarily reflects the 13C content of bicarbonate.
An isotopic effect was demonstrated when acid was
added to dissolved bicarbonate, without carbonic anhydrase (5, 12, 13). The authors found that the 13C content
of bicarbonate was higher compared with CO2, which is
consistent with our results. A second isotopic effect was
also shown at the carbonic anhydrase step (13). Indeed,
in an in vitro system, enzymatic dehydration of bicarbonate dissolved in water results in an additive lower
13C content in the evolved CO compared with the
2
bicarbonate, which is also in agreement with our
findings. Therefore, the difference of 13C content that
we observed between expired CO2 and CO2 derived
from bicarbonate is likely to be due to these two factors.
Our data do not allow evaluation of their relative
contribution. In this respect, it would be of interest to
compare 13C content in expired CO2, whole blood, and
plasma (i.e., without carbonic anhydrase) or to use
carbonic anhydrase inhibitors.
When the 13CO2 content of exhaled air and the CO2
provided by acid addition were corrected by subtraction
of their respective basal value, we found a modest but
persistent overestimation of the enrichment, which
increases as the 13CO2 content increases, in blood
degassed by acid. Such results were also reported by
Read et al. (14) but not by Denne and Kalhan (4) and
El-Khoury et al. (6). However, in these studies, the
statistical analyses were only performed on a limited
number of samples. This uncertainty thus restricts the
use of this method because such a technique has to give
similar results to the measurement in breath, considered as the reference method.
In contrast, the measurements of the 13CO2 isotopic
content spontaneously released by blood gave good
precision and agreement with similar results to exhaled air whether it was expressed in AP or in APE. An
isotopic fractionation may also exist at the alveolar
blood-breath interface (i.e., in vivo) or at the water-air
interface (i.e., in the test tube; see Ref. 16). The absence
of difference in the carbon isotope analysis between the
expired CO2 and the CO2 spontaneously released by
blood suggests that the fractionation is negligible or
similar under these two circumstances.
A potential analytical drawback of the method developed is the small quantity of CO2 dissolved in the blood
and therefore evolved in the tube compared with the
two other methods (Breath and Blood1LA). This could
limit the number of replicates of isotope analysis in the
same tube. Nevertheless, in our conditions, up to six
replicates can be done with an excellent intra-assay
reproducibility. In practice, duplicate or triplicate measurements give reliable results. In addition, the gas
chromatography isotope ratio mass spectrometry analysis can be performed up to 5 days after blood sampling.
Compared with the acid method, our technique is also
superior for manipulator ease of performance and
analysis. It simply requires 1 ml of arterialized whole
blood introduced in a tube previously flushed with
helium and agitated (1 rotation/min) during 2 h at room
temperature before the isotope analysis.
Such a technique can be employed to estimate whole
body nutrient oxidation rate by measuring separately
13CO APE in whole blood and CO production rate by
2
2
indirect calorimetry in the situations mentioned in the
Introduction. Furthermore, when indirect calorimetry
Downloaded from http://ajpendo.physiology.org/ by 10.220.32.247 on July 31, 2017
Fig. 4. Difference between Blood and Breath expressed in APE 3
1,000 (e.g., by subtracting the basal value) vs. their means (A) and
between Blood1LA and Breath (B). Dotted line in B indicates the
relationship between the difference for Blood1LA vs. Breath and the
mean (y 5 0.034x 1 0.095, P 5 0.02)
E215
E216
13CO
2
DETERMINATION IN WHOLE BLOOD
is unavailable, it should be possible to assess whole
blood flow and blood CO2 concentrations and 13C enrichments. Finally, applications also exist to measure nutrient oxidation rate in specific organs and to quantify
CO2 production rate.
We thank C. McCormack and Y. Boirie for help and advice. The
technical and nursing assistance of P. Rousset and L. Morin is also
gratefully acknowledged.
Address for reprint requests: B. Beaufrère, Laboratoire de Nutrition Humaine, B.P. 321–58, rue Montalembert, 63009 ClermontFerrand Cedex 1, France.
Received 19 May 1998; accepted in final form 2 September 1998.
REFERENCES
Downloaded from http://ajpendo.physiology.org/ by 10.220.32.247 on July 31, 2017
1. Bland, J. M., and D. G. Altman. Statistical methods for
assessing agreement between two methods of clinical measurement. Lancet 1: 307–310, 1986.
2. Boirie, Y., P. Gachon, S. Corny, J. Fauquant, J. L. Maubois,
and B. Beaufrère. Acute postprandial changes in leucine
metabolism as assessed with an intrinsically labeled
milk protein. Am. J. Physiol. 271 (Endocrinol. Metab. 34):
E1083–E1091, 1996.
3. Craig, H. Isotopic standards for carbon and oxygen and correction factors for mass spectrometry analysis of carbon dioxide.
Geochim. Cosmochim. Acta 12: 133–149, 1957.
4. Denne, S. C., and S. C. Kalhan. Glucose carbon recycling and
oxidation in human newborns. Am. J. Physiol. 251 (Endocrinol.
Metab. 14): E71–E77, 1986.
5. Deuser, W. G., and E. T. Degens. Carbon isotope fractionation
in the system CO2(gas)-CO2(aqueous)-HCO3-(aqueous). Nature
215: 1033–1035, 1967.
6. El-Khoury, A. E., N. K. Fukagawa, M. Sanchez, R. H. Tsay,
R. E. Gleason, T. E. Chapman, and V. R. Young. Validation of
the tracer-balance concept with reference to leucine: 24-h intravenous tracer studies with L-[1-13C]leucine and [15N-15N]urea.
Am. J. Clin. Nutr. 59: 1000–1011, 1994.
7. Hiele, M., Y. Ghoos, P. Rutgeerts, and G. Vantrappen.
Starch digestion in normal subjects and patients with pancreatic
disease, using a 13CO2 breath test. Gastroenterology 96: 503–509,
1989.
8. Just, B., D. Darmaun, J. Koziet, M. Rongier, and B. Messing. Use of a ventilated canopy for assessment of [13C]leucine
oxidation in patients receiving total parenteral nutrition. J.
Parenter. Enteral Nutr. 15: 65–70, 1991.
9. Kien, C. L., and R. E. McClead. Estimation of CO2 production
in enterally fed preterm infants using an isotope dilution stable
tracer technique. J. Parenter. Enteral Nutr. 20: 389–393, 1996.
10. Lyon, T. D. B., and M. S. Baxter. Stable carbon isotopes in
human tissues. Nature 273: 750–751, 1978.
11. Mansoor, O., M. Cayol, P. Gachon, Y. Boirie, P. Schoeffler, C.
Obled, and B. Beaufrère. Albumin and fibrinogen synthesis
increase while muscle protein synthesis decreases in headinjured patients. Am. J. Physiol. 273 (Endocrinol. Metab. 36):
E898–E902, 1997.
12. Mook, W. G., J. C. Bommerson, and W. H. Staverman.
Carbon isotope fractionation between dissolved bicarbonate and
gaseous carbone dioxide. Earth Planet Sci. Lett. 22: 169–176,
1974.
13. Paneth, P., and M. H. O’Leary. Carbon isotope effect of
dehydration of bicarbonate ion catalyzed by carbonic anhydrase.
Biochemistry 24: 5143–5147, 1985.
14. Read, W. W., M. A. Read, M. J. Rennie, and R. C. Griggs.
Preparation of CO2 from blood and protein-bound amino acid
carboxyl groups for quantification and 13C-isotope measurements. Biomed. Mass Spectrom. 11: 348–352, 1984.
15. Scrimgeour, C. M., and M. J. Rennie. Automated measurement of the concentration and 13C-enrichment of carbon dioxide
in breath and blood samples using the finnigan MAT breath gas
analysis system. Biomed. Environ. Mass Spectrom. 15: 365–367,
1988.
16. Vogel, J. C., P. M. Grootes, and W. G. Mook. Isotopic fractionation between gaseous and dissolved carbon dioxide (Abstract).
Z. Phys. 230: 225, 1970.
17. Wolfe, R. R. Radioactive and stable isotope tracers in biomedecine. In: Principles and Practice of Kinetic Analysis. New York:
Wiley, 1992, p. 23–36.