Clinical Chemistry 45:7
1077–1081 (1999)
Nutrition
The Use of Infrared Spectrophotometry for
Measuring Body Water Spaces
Graham Jennings,1 Leslie Bluck,2 Antony Wright2 and Marinos Elia1*
Background: The conventional method of measuring
total body water by the deuterium isotope dilution
method uses gas isotope ratio mass spectrometry
(IRMS), which is both expensive and time-consuming.
We investigated an alternative method, using Fourier
transform infrared spectrophotometry (FTIR), which
uses less expensive instrumentation and requires little
sample preparation.
Method: Total body water measurements in human
subjects were made by obtaining plasma, saliva, and
urine samples before and after oral dosing with 1.5 mol
of deuterium oxide. The enrichments of the body
fluids were determined from the FTIR spectra in the
range 1800 –2800 cm21, using a novel algorithm for
estimation of instrumental response, and by IRMS for
comparison.
Results: The CV (n 5 5) for repeat determinations of
deuterium oxide in biological fluids and calibrator solutions (400 –1000 mmol/mol) was found to be in the
range 0.1– 0.9%. The use of the novel algorithm instead
of the integration routines supplied with the instrument
gave at least a threefold increase in precision, and there
was no significant difference between the results obtained with FTIR and those obtained with IRMS.
Conclusion: This improved infrared method for measuring deuterium enrichment in plasma and saliva requires no sample preparation, is rapid, and has potential
value to the clinician.
© 1999 American Association for Clinical Chemistry
Total body water has been used to assess human body
composition since the demonstration that a relatively
fixed fraction of lean body tissue is occupied by water (1 ).
The deuterated water (2H2O) isotope dilution method for
measuring total body water has now replaced the earlier
1
MRC Dunn Clinical Nutrition Centre, Hills Road, Cambridge CB2 2DH,
United Kingdom.
2
MRC Human Nutrition Research, Downhams Lane, Milton Road, Cambridge CB4 1XJ, United Kingdom.
* Author for correspondence. Fax 44 1223 426617.
Received February 18, 1999; accepted April 15, 1999.
radioactive method (3H) (2 ). However, the use of gas
isotope ratio mass spectrometry to measure 2H2O is
expensive, requires specialist knowledge for operation
and maintenance, and is somewhat tedious because the
water is first transformed to hydrogen (3, 4 ) or equilibrated with hydrogen (5, 6 ).
Other methods for the determination of deuterium (2H)
in water have been proposed (7–10 ), including infrared
spectrophotometry (11–16 ). This latter technique has received particular attention because of its relatively high
potential sensitivity, which suggests that only a small
quantity of deuterium need be administered. Unfortunately, the absorption band centered at ;2500 cm21,
attributable to excitation of the D–O bond, is present as a
shoulder on the band at ;2130 cm21, which is attributable
to the H–O bond. If a single wavelength is monitored (as
is usually done with a dispersive instrument), this leads to
uncertainty in the baseline reading and consequently
limits the accuracy of the technique at low enrichments. In
recent years, there have been considerable advances in
infrared instrumentation through the application of Fourier transform infrared (FTIR) techniques. This allows
rapid monitoring of the whole of the absorption region
(17–19 ). However, these do not appear to have been used
to measure enrichment in human plasma or other body
fluids. Therefore, the aim of this study was to assess the
precision of the new methodology and to compare results
obtained from several physiological fluids with those
obtained with mass spectrometry.
Subjects and Methods
Five subjects took part in this study (mean weight, 75.75 6
4.0 kg; mean body mass index, 24.2 6 1.70 kg/m2). Each
subject arrived at the laboratory after an overnight fast
(12 6 2 h) and received an oral dose of 1.5 mol (30 g) of
D2O, 99.9% (Sigma Chemicals) followed by ;200 g of tap
water [to produce an enrichment of 450 –750 mmol/mol
(ppm)]. The straw and dosing bottle used for the administration of the D2O were retained and weighed to obtain
the exact amount of dosing solution consumed. Samples of
blood (5- to 10-mL samples collected in heparin-containing stoppered tubes) and saliva (4-mL samples collected
1077
1078
Jennings et al.: Measuring Deuterium with Infrared Spectroscopy
into clean, dry tubes) were obtained immediately before
and at 4, 5, and 6 h after dosing. Spot urine samples were
also collected before and at 4, 5, and 6 h after dosing. The
subjects were asked to void their bladders 30 min before
the collection time and to refrain from drinking during the
30 min prior to each saliva collection. Otherwise, they
were allowed to drink small amounts of water ad lib
(,200 mL during the study).
treatment of samples
The saliva and plasma samples were centrifuged immediately after collection, and the supernatants were retained. All samples were stored frozen at 220 °C. Prior to
analysis, the samples were thawed, vortexed, and centrifuged. Preliminary work showed that the predose urine
samples were not suitable for use as background samples
for estimation of deuterium by infrared spectroscopy
because of the large variations in composition of sequential urine samples. To overcome these variations, we
freeze-dried weighed aliquots of the postdose urine samples and then reconstituted them gravimetrically with
local tap water containing naturally occurring deuterium
to provide samples of the same constituents, but without
deuterium enrichment, to act as background samples for
the postdose urine samples. Local tap water was also
analyzed [mass spectrometry value, 147.7 mmol/mol
(ppm; this value was essentially identical to values obtained from background samples of physiological fluids,
e.g., saliva and plasma)].
infrared spectroscopy
The CVs for the new method of analysis were assessed
using 8, 16, 32, and 64 scans for solutions containing 200,
400, 800, and 1000 mmol/mol deuterium. Five readings
were averaged to produce a single analytical result and
repeat analyses for each sample. The same procedure was
used to establish the CVs in plasma, saliva, and urine,
using 8 or 64 scans per sample. The temperature of the
sample compartment of the instrument was found to be
stable (32 6 1 °C) without thermostatic control. When not
in use, the sample holders were kept in the instrument
with the instrument switched on to maintain them at
operating temperature.
The precision of the standard procedure for measuring
areas under the curve, which is incorporated into the
software of FTIR spectrophotometer, and a modified
procedure (20 ) was determined using the techniques
described below. The infrared spectra of aqueous samples
were measured in the range 1800 –2800 cm21 on a Mattson
Genesis series FTIR spectrophotometer (Mattson Instruments), equipped with an automated sample shuttle and
a pair of matched calcium fluoride sample cells with a
0.1-mm pathlength. The specifications of the instrument
were as follows: infrared ceramic emitter; detector,
LaTaO3 (uncooled); beam splitter, KBr; spectral range,
5500 –225 cm21; and resolution, 1 cm21. When the
matched cells were switched, the error was less than 61
mmol/mol (SD). A sample volume of 0.2 mL is adequate
for the total procedure.
Urine samples were measured using a single-sample
system and a cell near the end of its useful life because
ammonia and phosphate are both known to react with
calcium fluoride, causing the crystal to become opaque.
Postdose (or calibrator) samples were read at the same
time as predose (background) samples under the same
ambient conditions. Before the spectrum of each sample
was recorded, that of a background sample with natural
deuterium abundance was measured. An automatic sample shuttle made it possible, without disturbing the spectrometer, to mimic a double-beam system. After correction for natural abundance, the digitized spectra were
exported to a spreadsheet for subsequent processing.
Calibration was made after measurements of eight samples. The calibration procedure involved preparation of
2
H2O calibrators by gravimetric dilution of 2H2O (99.9
APE; Sigma) with local tap water. The enrichment of these
calibrations was confirmed by mass spectrometry (see Fig.
1 for a typical calibration curve).
ftir data handling
The infrared spectrum of D2O in water obtained from the
FTIR instrument is shown in Fig. 2. The signal from the
deuterium-oxygen bond appears as a weak shoulder on
the intense nondeuterated water band. Baseline subtraction can be performed with unenriched water as a reference, using the sample shuttle, but it is still difficult to
determine the zero enrichment background with any
great precision. This is demonstrated in Fig. 3, in which
the zero background enrichment is subtracted from the
postdose spectrum. The line at 2500 cm21 shows the
single frequency at which most previous workers have
made measurements (13, 16 ).
To overcome this uncertainty in baseline positioning, a
similar approach to that developed in this laboratory for
gas chromatography/mass spectrometry isotope mass
Fig. 1. Typical calibration curve (r2 5 0.999).
Clinical Chemistry 45, No. 7, 1999
Fig. 2. The infrared spectrum of 1000 mmol/mol D2O in water.
The D–O bond vibration signal appears as a shoulder on the band produced
by the H–O bond. The sharp doublet at ;2300 cm21 is attributable to atmospheric CO2.
ratio determination was used (20 ). The spectra were
digitized in the region 2665–2415 cm21. The sample
spectrum was then compared with the sum of the reference spectrum and its first two derivatives, generated
numerically. The inclusion of the derivative functions
allowed correction for slight spectral energy shift. Baseline subtraction was achieved by the inclusion of a linear
function of light energy in the fitting procedure. This
method compared the spectrophotometer outputs for
sample and a reference across the whole of the relevant
peak, separating the responses into that which is common
to the absorption peak itself and that which is solely a
property of the linear baselines beneath the peaks. Admixture of the first, second, and subsequent derivatives of
the observed lineshape for the reference material allowed
for nonsuperimposibility of the spectra because of instrumental drift. The equation used is:
S~ y# ! 5
1
rd
~ d 1 1! R~ y# 1 1! 1 r ~1 2 d 2 ! R~ y# !
2
rd
~ d 2 1! R~ y# 2 1! 1 A y# 1 B
2
where y# is the wavenumber, S(y# ) is the sample spectrum,
R(y# ) is the reference spectrum, r is the ratio between the
peak intensities, d is the instrumental shift along the
wavenumber axis between the sample and reference
spectra, and A and B are constants related to the slopes
and intercepts of the backgrounds of the spectra. In practice,
the n points of the digitized spectra were exported from the
spectrophotometer software directly into a spreadsheet and
the following matrix expression was solved:
P 5 ~R 1 R! 2 1 R 1 S
where the (n 2 2) row vector S is made up of the digitized
data for the sample, omitting the first and last points. R is
a matrix of five columns with (n 2 2) rows. The first
column contains the spectrum for the reference, omitting
the last two readings. The second row contains the same
1079
Fig. 3. The infrared spectrum of 1000 mmol/mol D2O in water with the
spectrum of natural abundance water subtracted.
The vertical line intersecting the x-axis at 2500 cm21 indicates the single
frequency used most commonly by previous workers with dispersive instruments (17 ).
data, but now starts at the second reading, omitting the
last point; the third column contains the same data again,
but omitting the first two points. The fourth column is the
wavenumber, and each element of the fifth column is
equal to unity. P is a vector of the best estimates of the
parameters of the fit. Most importantly, the ratio of
spectral intensities, r, can be obtained by summing the
first three elements of P. The relative amounts of deuterium in the sample can then be estimated by assuming the
Beer-Lambert law to hold. In calculating the enrichment
of deuterium in salivary and plasma water, we assumed
(21 ) (and confirmed experimentally) that the hydration
fraction of saliva was 99.5% 6 0.2% (in kg/L), and that of
plasma, 94% 6 0.4%. The hydration fraction of the urine
samples was calculated from the freeze-dried samples.
mass spectrometry
Excess sample was isotopically equilibrated with hydrogen gas for 3 days, using platinum on alumina powder as
a catalyst (6 ). The deuterium content of the resultant gas
was compared with that of known calibrators, using a Sira
10 (VG Instruments) equipped with a dual inlet, and the
results were converted to the SMOW/SLAP scale.
preparation of calibrators
Samples enriched in deuterium by known amounts were
prepared by the gravimetric addition of D2O (99.9%;
Sigma-Aldrich) to an aliquot of natural abundance water
(Cambridge tap water) and then adjustment to a known
volume with tap water. A portion of this enriched sample
was further diluted in Cambridge tap water to give
samples enriched above natural abundance by 1000, 800,
600, 400, and 200 mmol/mol.
statistical analysis
Results were expressed as means and SD. Comparisons of
results obtained with the two methods and between
1080
Jennings et al.: Measuring Deuterium with Infrared Spectroscopy
Table 1. Comparison of the CV (n 5 5 separate samples)
obtained using two different methods for measuring the
area under the D2O peak.a
CV, %
D2O calibrator,
mmol/mol
Standard integrated method
New methodb
1000
800
400
200
3.7
6.2
3.0
5.0
0.35
0.39
0.87
1.4
a
b
Eight scans per reading, and five measurements per sample.
Improved algorithm used for measuring area under the curve (see text).
different physiological fluids collected at the same time
were made using the Bland and Altman (22 ) method.
ethics approval
The study was approved by the local ethics committee,
and informed written consent was obtained from all the
participants.
Results
At all levels of enrichment, the method of Bluck and
Coward (20 ) for measuring spectral intensity was found
to be more precise than the traditional integration method
used by the instrument (Table 1). There was little or no
significant difference between results obtained using 8,
16, 32, and 64 scans per reading (Table 2). The time taken
to obtain 64 scans was 2.5 min. The precision for measurement of deuterium enrichment plasma and saliva was
very similar to that of calibrator solutions (Table 3).
The results in Table 4 show that there is close agreement between both methods for mass spectrometry and
infrared deuterium analysis when measurements were
made in plasma, saliva, and urine (none of the comparisons were significantly different from each other; paired
t-test). There also was close agreement in the enrichment
between the different types of fluid analyzed (Table 5)
and no significant increase in the difference between
methods as deuterium enrichment increased. It was possible to analyze samples within 30 min of receiving them
(including time for centrifugation).
Discussion
This study demonstrates that infrared spectroscopy,
which incorporates an improved analytical algorithm for
Table 2. CV (%) for the infrared method using 8, 16, 32,
and 64 scans per reading.a
Table 3. CV (%) for infrared measurements of enrichment of
deuterium in saliva, plasma, and urine.a
Enrichment of
deuterium,
mmol/mol
Saliva
Plasma
Urine
1000
800
400
200
0.30
0.22
0.41
1.7
0.20
0.24
0.31
1.4
0.50
0.54
0.76
2.0
CV, %
a
Sixty-four scans per reading and five measurements per sample on 15
separate samples of each fluid. Single-cell analysis for urine and dual matchedcell analysis for saliva and plasma, using improved algorithm for measuring area
under the curve (see text).
measuring spectral intensity, can be used to measure the
enrichment of D2O in plasma and saliva rapidly and precisely (CV, 0.1–0.9% for samples with deuterium enrichment
of 400 mmol/mol or more). The technique has a precision
(CV, 0.1–0.9%) that is several fold better than the traditional
method for measuring D2O enrichment with FTIR (CV,
3–6% over the same range of enrichment; see text).
This study also demonstrates good agreement between
the infrared and mass spectrometry measurements of
deuterium enrichment in saliva, plasma, and urine. Furthermore, the differences between plasma and salivary
measurements obtained with the infrared technique were
similar to those obtained with mass spectrometry. However, the dose of deuterium required to achieve a CV ,1%
for repeated measurements of the same sample is several
fold greater with the infrared method than with mass
spectrometry. On the other hand, changes in background
enrichment, which may occur after the administration of
intravenous fluids or the drinking of distilled water, will
influence the estimation of total body water by the infrared method to a lesser extent than mass spectrometry,
which usually involves measurement of smaller increments in deuterium enrichment. Furthermore, the operation of mass spectrometers requires more expertise than
the operation of the infrared spectrophotometer, which
can yield results shortly after the samples are obtained
(plasma and saliva). A particular problem with urine
analysis is the presence of variable amounts of interfering
urinary substances, which makes predose urine samples
Table 4. Postdose comparison of deuterium enrichment
measurements (mmol/mol) in plasma, saliva, and urine
obtained with mass spectrometry and the infrared method.a
Plasma,
mmol/mol
CV, %
D2O calibrator,
mmol/mol
8 scans
16 scans
32 scans
64 scans
1000
800
400
200
0.35
0.39
0.87
1.4
0.49
0.23
0.56
1.3
0.11
0.37
0.75
1.4
0.20
0.25
0.30
1.4
a
Five measurements per scan of calibrator solutions of D2O. Analysis used an
improved algorithm for measuring area under the curve (see text) (17).
Mean of methods (mass
spectrometry and infrared)
Difference between methods
(mass spectrometry 2
infrared)
Saliva,
mmol/mol
Urine,
mmol/mol
610 6 71 614 6 69 617 6 67
6 6 11
461
22 6 12
a
Separate samples (n 5 15) were used for each fluid, with 64 scans per
reading. The improved algorithm was used for measuring the area under the
curve.
Clinical Chemistry 45, No. 7, 1999
Table 5. Postdose enrichments and differences in
enrichments (bias 6 SD) between physiological fluids,
according to method used.a
3.
mmol/mol
Enrichment
Plasma
Saliva
Urine
Differences in enrichment
Plasma 2 saliva
Plasma 2 urine
Saliva 2 urine
Mass spectrometry
Infrared
613 6 68
616 6 70
616 6 69
607 6 74
612 6 68
619 6 65
23 6 15
23 6 15
163
25 6 13
211 6 13
25 6 10
a
Separate samples (n 5 15) of each fluid were used, with 64 scans per
reading.
less suitable for background measurements than the corresponding plasma and saliva samples. Although appropriate predose urine samples can be prepared by removing
the enriched water from the urine (freeze-drying) and replacing it with tap water, the process is time-consuming and
unsuitable for bedside analysis. Another problem with urine
analysis is that the standard calcium fluoride cells are not
suitable for prolonged and repeated use, although our
preliminary studies with zinc selenide cells suggest that they
are a suitable alternative because they are not “attacked” by
urine. Sapphire cells may also be appropriate.
Finally, there have been reports that the use of a
single-cell system to measure enrichment of deuterium in
urine (instead of the matched-cell system for plasma and
saliva measurements) could explain the higher CV for the
urinary measurements (Table 3) and the greater discrepancy between urinary and alternative fluid measurements
when the infrared technique is used, compared with the
mass spectrometry method (Table 5). There have been
reports of temperature dependence of the infrared spectrum of both H2O and D2O in the shorter wavelength
5000 –16 000 cm21 region (23 ); therefore, some workers
have used thermostatically controlled cells (13, 16 ). However, because the operating temperature was stable and
the primary purpose of this work was to assess the
technique as a simple, rapid, and inexpensive method for
determination of total body water, no attempt was made
to thermoregulate the sample.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
In conclusion, this report describes an improved infrared
analytical method for measuring D2O enrichment in physiological fluids. It is simple, precise, rapid, and of potential value to the clinician who monitors and adjusts fluid
therapy on a daily basis.
20.
References
22.
1. Pace N, Rathburn E. Studies on body composition. III. The body
water and chemically combined nitrogen content in relation to fat
content. J Biol Chem 1945;158:685–91.
2. Panaretto B. Estimation of body composition by the dilution of
21.
23.
1081
hydrogen isotopes. Body Composition in animals and man. Washington DC: National Academy of Sciences-National Research
Council, 1968:200 –17.
Coleman ML, Shepherd TJ, Durham JJ, Rouse JE, Moore GA.
Reduction of water with zinc for hydrogen isotope analysis. Anal
Chem 1982;54:993–5.
Barrie A, Coward WA. A rapid analytical technique for the determination of energy expenditure by the doubly labeled water method.
Biomed Mass Spectrom 1985;12:535– 41.
Coplen T, Wildman JD, Chen J. Improvements in the gaseoushydrogen water equilibration technique for hydrogen isotope ratio
analysis. Anal Chem 1991;63:910 –2.
Scrimgeour CM, Rollo MM, Mudambo SMKT, Handley LL, Prosser
SJ. A simplified method for deuterium hydrogen isotope ratio
measurements on water samples of biological origin. Biol Mass
Spectrom 1993;22:383–7.
Schloerb R, Friis-Hansen J, Edelman IS, Sheldon DB, Moore SD.
The measurement of deuterium oxide in body fluids by the falling
drop method. J Lab Clin Med 1951;37:652– 62.
Reaser P. Determination of deuterium oxide in water by measurement of freezing point. Science 1958;128:415– 6.
Mendez J, Prokop E, Picon-Reategui E, Akers R, Buskirk ER. Total
body water by D2O dilution using saliva samples and gas chromatography. J Appl Physiol 1970;28:354 –7.
Nielsen WC, Krzywicki HJ, Johnson HL, Consolazio CF. Use and
evaluation of gas chromatography for determination of deuterium
in body fluids. J Appl Physiol 1971;31:957– 61.
Thornton V, Condon FE. Infrared spectrometric determination of
deuterium oxide in water. Anal Chem 1950;22:690 –1.
Turner MD, Neely WA, Hardy JD. Rapid determination of deuterium
oxide in biological fluids. J Appl Physiol 1960;15:309 –10.
Byers F. Extraction and measurement of deuterium oxide at tracer
levels in biological fluids. Anal Biochem 1979;98:208 –31.
Haschke F. Body composition of adolescent males. Acta Paediatr
Scand 1983;(Suppl 307):13–23.
Wang J, Pierson R, Kelly WG. A rapid method for the determination
of deuterium oxide in urine: application to the measurement of
total body water. J Lab Clin Med 1973;82:170 – 8.
Lukaski H, Johnson P. A simple, inexpensive method of determining total body water using a tracer dose of D2O and infrared
absorption of body fluids. Am J Clin Nutr 1985;41:363–70.
Pang KS, Xu N, Goresky CA. D2O as a substitute for 3H2O, as a
reference indicator in liver multiple-indicator dilution studies. Am J
Physiol 1991;261:G929 –36.
Fusch C, Spririg N, Moeller H. Fourier transform infrared spectroscopy measures 1H/2H ratios of native water with a precision
comparable to that of isotope ratio mass spectrometry. Eur J Clin
Chem Clin Biochem 1993;31:639 – 44.
Veryaminor SY, Prendergast FG. Water (H2O and D2O) molar
absorptivity in the 1000 – 4000 cm21 range and quantitative
infrared spectroscopy of aqueous solutions. Anal Biochem 1997;
248:234 – 45.
Bluck LJC, Coward WA. Peak measurement in gas chromatographic mass spectrometric isotope studies. J Mass Spectrom
1997;32:1212– 8.
Snyder WS, Cook MJ, Nasset ES, Karhausen LR, Howells GP,
Tipton IH. Report of the Task Group of Committee 2 of the
International Commission on Radiological Protection. Oxford, UK:
Pergamon Press, 1976:335– 467.
Bland JM, Altman DG. Statistical methods of assessing agreement between two methods of clinical measurement. Lancet
1986;i:307–10.
Waggener W. Absorbance of liquid water and deuterium oxide
between 0.6 and 1.8 microns. Anal Chem 1958;30:1569 –70.