Contents - Previous - Next

Chapter 1: Introduction

1.1 Significance
1.2 Principle of the doubly-labelled water method
1.3 Origin of the method
1.4 Development
1.5 Small animal studies
1.6 Human studies
1.7 Human validation studies
1.8 Purpose of the Cambridge workshop
1.9 References

Contributor: Kenneth A. Naqy

1.1 Significance

The rate at which a human uses metabolic energy can reveal a great: deal about their health and welfare. Energy is required to fuel the basic life processes, for work and activity, to meet challenges to survival such as disease, drought and famine, and to resist environmental stresses such as cold. Although metabolic rates can be measured routinely in laboratory settings either directly by assessing heat loss using whole-body calorimeters, or indirectly by measuring oxygen consumption and/or carbon dioxide production, measurements of energy metabolism by unrestrained humans in their normal surroundings had to await discovery of the doubly-labelled water (DLW) method.

1.2 Principle of the doubly-labelled water method

The technique involves enriching the body water of a subject with an isotope of hydrogen (*H) and an isotope of oxygen (*O), and then determining the washout kinetics of both isotopes as their concentrations decline exponentially toward natural abundance levels (Figures 1.1 & 1.2).

The concentration of the hydrogen isotope, nearly all of which remains associated with water molecules, decreases as a result of dilution of body water by new, unlabelled water (consumed as food and drink and produced during oxidation of foodstuffs), coupled with the simultaneous loss of labelled water via evaporation from lungs and skin, and via excretions and secretions. The rate constant for *H is derived as the slope of loan *H enrichment against time (Figure 1.2) and is a measure of the rate of water movement through the subject.

Most of the *O in a labelled subject is lost as water, but some is also lost as carbon dioxide because CO2 in body fluids is in isotopic equilibrium with body water due to the action of carbonic anhydrase present in red blood cells and elsewhere. Thus the slope of the washout line representing *O is steeper than the line for *H (Figure 1.2), and the difference between slopes represents CO2 production. This indirect measure of metabolic rate may then be converted to units of heat production by incorporating knowledge, or estimates, of the chemical composition of the foodstuffs being oxidised since this influences the energy equivalence of each litre of CO2 produced.

Determination of the two rate constants requires a minimum of two post-dose samples of body fluid, over a time period of several days to several weeks, depending on the subject's age and rate of water consumption. The isotopes of choice in human studies are deuterium (2H) and oxygen-18 (18O) since these avoid the need to use any radioactivity and can be safely used in any subjects.

Figure 1.1: Principle of the doubly-labelled water method

k = experimentally-determined rate constant (see Fig 1.2)

r = production rate


The carbonic anhydrase reaction in red blood cells and in the lung catalyses the following equilibria:

H2*O + CO2 carbonic anhydrase (r) H2C*O3

H2*O + C*O2 carbonic anhydrase (r) H2C*O3

Figure 1.2: Examples of isotope disappearance curves

a) Untransformed data

b) Log transformed data

The initial values have been normalised on the y axis for clarity.

1.3 Origin of the method

Professor Nathan Lifson and his colleagues at the University of Minnesota invented the doubly-labelled water method in the late 1940's and early 50's 1. Their discovery that the oxygen in respiratory carbon dioxide is in isotopic equilibrium with the oxygen in body water 2 provided the key information upon which the new method is based. The simultaneous development of sensitive and accurate mass spectrometers by A.O. Nier, also at the University of Minnesota, was an essential ingredient in the invention process.

1.4 Development

Lifson and his colleagues performed subsequent studies on laboratory rats and mice to validate and evaluate the potential errors in the DLW method 1, 3-7, In 1966 they published a review paper which listed the assumptions inherent in the model which forms the basis of the technique, and considered the likely effects on accuracy of any deviations from this model 8. The paper concluded that the method was relatively robust to the sorts of deviations which might be encountered in reality, and emphasised the potential value of DLW tin studies of free-ranging animals.

Aside from a study of the energetic cost of flight in pigeons by a student at the University of Minnesota 9, many years passed before researchers began using the DLW method. The reasons for this include the very high cost of the isotopes needed to enrich the body water of subjects (about US$ 1500 per kilogram of body mass at prevailing prices and recommended doses in early 1960), and the technical difficulty and relative unavailability of isotope measurements.

1.5 Small animal studies

The substitution of tritium, which is much more easily measured than deuterium, and the development of the proton activation method for measuring oxygen-18 in the early 1970's 10, 11 facilitated implementation of DLW studies on small animals. Large doses of 18O were still required for the proton activation analysis to yield accurate results, so isotope costs were still very high, despite reductions in prices (clown to about US$ 200 per kilogram of subject). Thus, studies on large animals and humans remained impractical. However, research on small animals mushroomed, with more than 12 species of reptiles, 23 species of eutherian mammals, 13 species of marsupial mammals and 25 species of birds having been studied in their natural habitats up until the mid-1980's 12, 13. Validation studies have compared metabolic rates determined using DLW with those determined using direct CO2 or O2 analysis or with other independent measures (Table 1.1). These yielded average errors of within 6% in 10 studies on small mammals and 7 studies on small birds, but indicated that errors may be larger in reptiles (8%) and especially in arthropods (37%) due to their specialised water-conservation mechanisms.

1.6 Human studies

Research on humans with the DLW method had to await a reduction of isotope purchase costs. This was made possible in the late 1970's, not by further reduction in isotope prices, but by the increased sensitivity and accuracy of 18O measurements which permitted use of much lower doses. Lifson et al published an analysis showing that with 18O analyses performed on an isotope ratio mass spectrometer, which provides much more accurate measurements than other types of mass spectrometers, results having errors of less than 10% could be obtained from adult human subjects with doses of 18O costing only US$ 75-250 per subject 14. Shortly thereafter, gas isotope ratio mass spectrometers became available commercially. In 1982, Dale Schoeller and Edzard van Santen published the first study to show that DLW can yield an economical and accurate measurement of energy expenditure in humans 15, The first field applications of DLW in human subjects were reported in 1985 16 and there has been a rapid expansion of the literature since then.

Table 1.1. Cross-validation studies in animals


% error in DLW method




Mouse (Mus)


(+20, -21)

Mouse (Mus)


(+8, -12)

Mouse (Perognathus)


(+6, -9)

Squirrel (Ammospermophilus)


(+17, -12)

Chipmunk (Tamias)


(+8, +1)

Chipmunk (Tamias)


(+18, -19)

Rat (Rattus)


(+10, -2)

Rat (Rattus)


(+6, -9)

Rat (Rattus)


(+12, -13)

Gopher (Thomomys)


(+15, -9)


Pigeon (Columba)


(+17, -12)

Martin (Delichon)


Sparrow (Passerculus)


(+11, -0.2)

Starling (Sturnus)


(+16, -15)

Sparrow (Zonotrichia)


(+13, -4)

Parakeet (Melopsittacus)


(+6, -5)

Quail (Callipepla)


(+8, -17)


Lizard (Sceloporus)


(+18, -6)

Lizard (Uta)


(+12, -22)

Tortoise (Gopherus)


(+25, -26)


Locust (Locusta)


(+60, -24)

Scorpion (Hadrurus)


(+71, +11)

Beetle (Eleodes)


Beetle (Cryptoglossa)


See reference 31 for original citations.

* significantly different from zero.

Table 1.2. Cross-validation studies in humans


% error (SD)

Ref. Method



Adults, n=4

-0.4 (5.6)


S, 2 point


Adult, n=1



L, multipoint


Adults, n=5

+1.5 (7.6)


S, 2 point


Adults, n=4

+1.9 (2.0)


C, multipoint


Exercising adults, n=2

-2.5 (4.9)


L, 2 point


Premature infants, n=4

-0.3 (2.6)


C, multipoint


Adults on TPN, n=5

+3.3 (5.9)


S, 2 point


Adults, n=9

+1.4 (7.7)


S, 2 point


Post-surgical infants, n=9

-0.9 (6.2)


S, 2 point


Infants, changing diet, n=8

-8.7 (12.9)


S, 2 point


Adults, n=5

+1.4 (3.9)


S, 2 point


Exercising adults, n=4x2

-1.0 (7.0)


S, 2 point


Soldiers in the field, n=16



S, 2 point


I/B = Intake corrected for change in body stores.
RGE = Respiratory gas exchange.
S = Method of Schoeller 28.
L = Method of Lifson 8.
C = Method of Coward 29.

1.7 Human validation studies

The doubly-labelled water method has been validated in 13 separate studies by 4 independent research groups with excellent results (Table 1.2). Three mathematical models have been employed in these validations as detailed in later chapters. In general, the Lifson model tends to underestimate carbon dioxide production and hence energy expenditure by several percent in adults and up to 13% in infants or other subjects with high water turnover rates 28. Because of this the Lifson model is not recommended for human studies. The remaining models of Coward 29 and Schoeller 28 have been found to be valid. Accuracy is generally in the order of 1 to 3% and precision 2 to 8%, with the Coward model using multi-point regression analysis of isotope elimination rates seeming to offer 2 to 3% better precision than the two-point method. The two major exceptions have both been studies in which the subjects changed their source of water during the isotope elimination period. These studies involved infants who were being weaned from total parenteral nutrition to oral nutrition 25, and soldiers who were transported from their barracks to a field exercise 27. The issues of changing water sources are discussed in Chapter 8, and are expected to lead to small loses of accuracy and precision. It should be noted, however, that the use of the intake/balance method based on self-monitored intake by the soldiers may have been more prone to error than the DLW method itself and is likely to be the cause of the large difference in this validation.

Perhaps the most important issue in the validations is that they have encompassed a range of circumstances. Studies in healthy adults have predominated, but they have included a wide range of subjects including: sedentary people; subjects exercising to exhaustion; subjects who were in energy balance; subjects who were underfed by 300 to 1500 kcal/day; and patients who were receiving total parenteral nutrition in excess of energy requirements. The other validations were performed in premature rapidly-growing infants, and in post-surgical infants. Thus the DLW method has been validated under a wide range of human conditions.

1.8 Purpose of the Cambridge workshop

The workshop was arranged to discuss different DLW techniques and procedures used by various laboratories, and to recommend standard procedures for use in further studies. This should render results comparable between laboratories, and benefit our goal of understanding human energetics.

The DLW method involves several assumptions about the behaviour of the isotopes, the body water pool and the exchange rates within that pool in the labelled animal 8, 30, These assumptions are:

(1) The volume of the body water pool remains constant throughout the measurement period.

(2) The rates of water influx, and water and CO2 efflux are constant throughout the measurement period.

(3) The isotopes label only the H2O and CO2 in the body.

(4) The isotopes leave the body only in the form of H2O and CO2.

(5) The concentrations of the isotopes in H2O and CO2 leaving the body are the same as those in body water at that time (i.e. there is no isotopic fractionation).

(6) No H2O or CO2 that has left the body re-enters the body.

(7) The natural abundance, or "background" levels of the isotopes remain constant during the measurement interval.

All of these assumptions are invalid to some degree in any DLW study, but a variety of corrections can be applied to completely or partially account for the resulting errors. New users of DLW are faced with a confusing array of technical decisions that must be made as part of this technique. Fortunately, the DLW method is sufficiently robust that making an inappropriate decision will, in most cases, cause less than a 10% error in the calculated rate of energy metabolism, provided that isotope concentration measurements (the largest potential source of error) are accurate.

The workshop participants pooled their knowledge, experience and different perspectives on the problems to generate the recommendations presented in this document. The recommendations are based on a variety of criteria including; (a) which procedure among several is theoretically correct in a given application; (b) which procedure is simplest and least prone to methodological errors; and (c) which procedure yields the lowest error in validation studies. We hope that this synthesis will clarify the many complex issues involved and hence encourage more researchers to use this exciting method to explore new areas of human biology and medicine.

Contents - Previous - Next