457
Clinical Science (1 98 1) 61,457-462
Total body nitrogen and its relation to body potassium and
fat-free mass in healthy subjects
L. B U R K I N S H A W ' , D . B. M O R G A N ' , N . P . S I L V E R T O N 3
R . D. T H O M A S 3 *
AND
Departments of lMedical Physics, 2Chemical Pathology and 'Cardiovascular Studies, University of Leeds,
The General Infirmary, Leeds, U.K.
(Received 4 June 1980120 February 1981; accepted 8 April 1981)
Summary
1. The amount of lean tissue in the body can
be assessed by measuring total body nitrogen,
total body potassium or fat-free mass. To
compare these techniques we have measured total
body nitrogen, total body potassium and fat-free
mass in 91 healthy subjects (62 males, 29
females).
2. Total body nitrogen in the women and
civilian men agreed closely with the few values
reported previously and was closely related to
total body potassium and fat-free mass.
3. The simplest estimate of total body nitrogen
in a subject whose body content has not been
measured is the mean value for healthy people of
the same sex. The standard deviation of individual values about this mean is 253 g. The
precision of the estimate can be improved
considerably by predicting body nitrogen from
fat-free mass (156 g) and somewhat more by
predicting it from body potassium (1 15 g). The
error of measuring total body nitrogen directly is
approximately 76 g.
4. When an individual's total body potassium
is measured in a search for potassium depletion,
the observed value must be compared with the
value expected if the subject were healthy. The
standard deviation of the healthy values about the
group means is 408 mmol. The precision of the
estimate can be improved by predicting total
body potassium from fat-free mass (SD 237
* Present address: Royal United Hospital, Combe
Park, Bath BA1 3NG, U.K.
Correspondence: Dr L. Burkinshaw, Department of
Medical Physics, General Infirmary, Leeds LS1 3EX,
U.K.
mmol), and rather more by predicting it from
total body nitrogen (SD 186 mmol). If gross body
composition is normal, measurement of total
body nitrogen has little advantage over measurement of fat-free mass by the anthropometric
technique.
5. These results suggest that the simpler
measure of fat-free mass from body weight and
skinfold thickness has a major role in the
assessment of total body nitrogen, and thus lean
body tissue, in the individual.
Key words: activation analysis, body composition, whole-body counting.
Introduction
A reliable measure of the amount of cellular or
lean tissue in the body is often required in studies
of growth and development, nutritional status, the
effects of physical exercise or of disease processes. A measure which is frequently used is the
fat-free mass, calculated by subtracting body fat,
estimated from skinfold thickness, from body
weight [l]. An alternative is to measure total
body potassium. More than 90% of the
potassium is within cells, so that, if its concentration in cells is normal, total body potassium is
also a measure of the cellular or lean tissue mass.
Total body potassium is also measured in
patients in a search for potassium depletion, but a
low value may result from tissue loss or wasting
alone. Therefore, to detect potassium depletion,
total body potassium must be related to some
index of the amount of lean tissue. The ratio of
total body potassium to body weight or fat-free
mass has been used. An alternative is to calculate
0143-5221/81/l00457~6$01.50/1@ 1981 The Biochemical Society and the Medical Research Society
458
L . Burkinshaw et al.
the linear regression of total body potassium on
age, height, weight and skinfold thickness for
healthy subjects of each sex and use this to
calculate the predicted total body potassium in
the individual. The observed and predicted values
are then compared. Although these approaches
are well established they have the disadvantage
that disease may be associated with changes in
the body content of fat or water, in which case
body weight or fat-free mass will not provide
valid estimates of the amount of lean tissue.
Total body nitrogen is largely determined by
the amount of tissue protein in the body, and
therefore it may be a better guide to the cellular
mass than total body potassium or fat-free mass.
Total body nitrogen might also be expected to
play an important role as a reference for total
body potassium in the detection of potassium
depletion, as it should be an index of the amount
of lean tissue which is unaffected by changes in
body water and fat. However, the measurement
of total body nitrogen is not widely available
since it requires a complex technique of neutron
activation and whole-body counting 12, 31 or the
detection of y-ray emission during neutron
irradiation [4-61. As a first step in assessing the
magnitude of the advantages of total body
nitrogen in practice we have measured total body
nitrogen, total body potassium and fat-free mass
in a group of healthy subjects.
Subjects and methods
Subjects
The volunteer subjects were hospital staff, staff
from the Local Authority Services and members
of the West Yorkshire Metropolitan Police.
Forty-four men aged between 2 2 and 68 years
were staff from the hospital or local authority,
and 18 were policemen aged between 44 and 54
years. Twenty-nine women aged between 21 and
62 years came from all three sources. The
volunteer subjects agreed to take part after the
procedure had been fully described and explained
to them. The study was approved by the local
Research Ethics Committee and by the Neutron
Panel of the Medical Research Council.
Methods
Each subject’s height and weight were
measured and the observed weight was corrected
to nude weight. Skinfold thicknesses were
measured with Harpenden calipers over the
biceps, triceps, subscapular and supra-iliac sites
on the subject’s left-hand side [71. Body fat was
calculated from body weight and the sum of the
four skinfold thicknesses with the formulae given
by Durnin & Womersley [ 11. Fat-free mass was
calculated by subtracting body fat from body
weight.
Total body nitrogen was measured by neutron
activation analysis in uiuo [2]. Each subject,
wearing only a cotton gown, was irradiated
bilaterally to a nominal dose-equivalent of 50
mrem (0.5 mSv) with 14 MeV neutrons and then
counted for 30 min in a whole-body radiation
counter. The resulting complex y-ray spectrum
was analysed to give the activities of the radionuclides induced in the body by the neutrons, and
the activity of the natural radioisotope of
potassium, 40K. The activities of the induced
radionuclides were corrected for decay and
normalized to the standard neutron dose.
Total body nitrogen was estimated from the
activity of 13N, produced by the reaction 14N(n,2n)13N, but the observed activity had first to
be corrected for a contribution of about 20% of
the total, generated by the interaction of recoiling
hydrogen nuclei with oxygen [160(p,u)13N].The
size of this contribution per unit mass of oxygen
was determined by irradiating and measuring
anthropomorphic phantoms filled with water t81.
When correcting for oxygen in our subjects, it
was assumed that 85% of the body’s oxygen is in
water, and body water was estimated from weight
and height with the formulae given by Hume &
Weyers 191.
Total body nitrogen and potassium were
estimated by comparing the corrected activity of
13N and the measured activity of 40K with the
corresponding activities observed when a phantom weighing 70 kg, containing physiological
quantities of the major body elements, was
irradiated and measured. However, these initial
estimates required further correction, since the
activity of 13N produced, and the response of the
counter to emitted prays depend on body size.
The necessary correction factors were determined
by irradiating and measuring anthropomorphic
phantoms containing physiological amounts of
the major body elements; they were found to lie
in the range 0.7-1.1 for phantom weights
between 40 and 80 kg [S]. If these corrections are
applied to measurements made on similar phantoms irradiated to a dose-equivalent of 50 mrem,
the measured amounts of nitrogen and potassium
will be distributed about the true values with
coefficients of variation of 3.6 and 2%
respectively.
Human measurements may be less accurate. In
the case of nitrogen the random errors of
estimating the contribution of oxygen to the
activity of 13N increase the standard error from
3.6 to 4.2%. Other errors will accrue if the
,
Total body nitrogen in healthy subjects
assumption that 85% of body oxygen is in water
is incorrect, or if the corrections for body weight,
determined with phantoms, are inaccurate when
applied to human measurements. These errors are
difficult to assess, but a detailed study suggests
that measured body oxygen is unlikely to be in
error by more than about 6% [ 101.
The correction factors for potassium were
compared indirectly with values determined by
another method, believed to give unbiased estimates, in which 42Kis given to human subjects as
an internal standard [ l l ] . McCarthy [121
originally determined the factors for potassium
for the whole-body counter configuration used in
this work, by measuring 20 volunteer subjects
with this configuration, and with a configuration
which had been calibrated by the 42Kmethod
1131; he assumed the values given by the latter
configuration to be correct. The factors he found
differed by no more than 3% from the more
recent values used in this work, determined as
described above [81.
Standard statistical calculations were performed with version 5 of the Statistical Package
for the Social Sciences [141 on the ICL 1906A
computer of the University of Leeds. In calculating multiple linear regressions variables were
introduced in sequence and only those which
significantly reduced the residual variance (P<
0.05) were retained [ 151. Regression equations
for different groups of subjects were compared,
and parallel regressions were adopted or groups
of subjects combined, if the residual variance was
not thereby increased significantly (P< 0.05).
Results
The 18 policemen in this study may be less
representative of the general population than the
44 civilian men, because men admitted to the
Force must be at least 1.75 m tall, and it
45 9
therefore seemed prudent to regard them initially
as a separate group.
Table 1 shows the mean values and standard
deviations of the measured and derived quantities
for the three groups of subjects. To test whether
these relatively small groups were representative
of the general population, the individual values
of body weight, fat-free mass and total body
potassium were compared with the values expected in healthy subjects of the same sex, age and
height. The expected values were calculated from
equations derived from an earlier unpublished
survey of 242 healthy people, made in our own
laboratory. The equations are given in the
Appendix. There were no significant differences
between observed and expected values for the
civilian men and the only difference in the women
was that their potassium was 13 1 mmol less than
expected. These two groups are therefore taken to
be representative samples of the larger population
measured previously. However, the policemen
were on average 6.45 kg heavier than expected
for height and age (P < 0.01), the extra weight
being made up of 2.45 kg of fat and 4.0 kg of
fat-free tissue. We have therefore retained the
policemen as a separate group.
Total body nitrogen was significantly correlated (P< 0.05)with total body potassium and
fat-free mass in all three groups. Table 2 gives
regression equations for the prediction of total
body nitrogen from body potassium, fat-free
mass and combinations of age, height, weight and
skinfold thickness. The standard deviations of the
measured values around the regression were all of
a similar order (range 115-217 g). These values
should be compared with the pooled standard
deviation of the measured total body nitrogen
within the groups which was 253 g.
Table 3 gives equations for predicting total
body potassium from body nitrogen, fat-free
mass and combinations of age, height, weight and
TABLE1. Means and standard deviations of the measured and derived quantities
Group
Civilian men
Mean
SD
Policemen
Mean
SD
Women
Mean
SD
Age
Height
Weight
Sum offour
Body fat
Fat-free
Total body
Total body
(years)
(cm)
(kg)
skinfold
thicknesses
(mm)
(kg)
mass
potassium
nitrogen
(kd
(mmol)
(g)
39.7
f11.7
176.1
f7.1
73.9
f8.4
34.8
f9.1
13.2
f4.6
60.7
f6.3
3643
f449
1974
f266
49.3
f3.1
180.6
f5.2
85.5
k8.3
41.7
5 10.6
19.3
k4.3
66.3
f5.8
3941
f405
2221
5-219
43.8
f11.7
162.8
f6.6
61.3
f11.4
55.3
ir26.0
19.5
f7.3
41.9
f5.2
2334
f331
1310
f213
L . Burkinshaw et al.
460
TABLE2. Linear-regression equations and standard deviations for predicting total
body nitrogen
Total body nitrogen (g, TBN) is predicted from age (years, A), height (cm, H ) , body
weight (kg, W), sum of biceps, triceps, subscapular and suprailiac skinfold thicknesses (mm, S), fat-free mass (kg) calculated from body weight and skinfold
thicknesses (FFM) and total body potassium (mmol, TBK).
Group
Civilian men
Policemen
Women
Civilian men +
poI i c e me n
Women
Civilian men +
policemen
Women
Civilian men +
policemen +
women
Civilian men
Policemen
Women
Equation
T B N = -1608
TBN = -1452
T B N = -2000
T B N = -397
T B N = -733
TBN = 113
1
1
SD from
regression
-0.144
+ 20.388
217
-4.104
+ 6 . 6 3 8 + 18.65W
165
-4.404
+ 1 , 9 3 8 + 25.23W-
4.71s
155
TBN = -99
TBN
+ 35.70 FFM
-180.50
TBN =-41.63
TBN = 39.58
T B N = 18.43
1
+ 0.553 TBK
156
115
TABLE3. Linear-regression equations for predicting total body potassium
The symbols are as defined in the legend of Table 2.
Group
Civilian men
Policemen
Women
Civilian men +
policemen
Women
Civilian men +
policemen
Women
Civilian men +
women +
policemen
Civilan men +
policemen
Women
Equation
TBK = -2158
TBK = -1970
TBK = -2989
TBK = -504
I
-5.494
1
TBK
= 221
TBK
= -329
332
Q
+ 15.62H + 24.95W
272
-11.634 + 4 , 6 6 H + 4 0 , 3 2 W - 11.00s
238
-10,934
TBK = -1259
TBK = 686
+ 34.198
SD from
regression
+ 64.77 FFM
23 7
TBK = 847
+ 1.41 T B N
TBK
186
= 488
skinfold thickness. The standard deviations from
regression range from 186 to 332 mmol. The
pooled standard deviation of the measured total
body potassium values within the groups is 408
mmol.
Discussion
There have been few previous reports of total
body nitrogen measurements in healthy subjects.
Vartsky et al. 161 measured 14 athletic young
men aged 20-3 1 years; their mean values (204 1 g
of N, 3899 mmol of K) agree with our values for
men aged 20-30 (1975 g of N, 3725 mmol of K).
McNeill et al. [161 measured 15 men and eight
women of unspecified ages; their mean values for
men (1970 g of N, 3969 mmol of K) and women
(1380 g of N, 2458 mmol of K) agree with ours
(Table 1).
The civilian subjects we have studied are
representative of a larger group of healthy people
whom we measured previously, suggesting that
the equations we-give for the estimation of total
body nitrogen (Table 2) can be applied to the
healthy general population.
Total body nitrogen is the most direct available
measure of total body protein, and hence of
cellular or lean tissue mass. However, the
technique is elaborate and not widely available.
Total body nitrogen in healthy subjects
Therefore it is important to examine how well
body nitrogen can be estimated when it cannot be
measured.
The simplest estimate of the nitrogen content
of an individual is the mean value for the group to
which he or she belongs. The uncertainty in this
estimate is given by the pooled standard deviation
of the values within the groups, which is 253 g. If
fat-free mass has been measured, then body
nitrogen can be estimated from it, and the
standard error of the estimate will be much less
(156 g). Alternatively, total body nitrogen can be
estimated from total body potassium, with a
standard error of 115 g. The smallest possible
standard error, that of a single direct measurement, is 76 g (4.2% of the grand mean for our
subjects). Thus the measurement of body
potassium gives only a modest improvement in
accuracy of estimation, compared with the much
simpler measurement of fat-free mass.
An alternative role for total body nitrogen is to
act as a reference for total body potassium when
testing for potassium depletion. The simplest
estimate of an individual’s healthy potassium
content is the mean value for the appropriate
group; the standard error of this estimate (the
standard deviation of the individual values within
the groups) is 408 mmol. If healthy potassium is
predicted from fat-free mass, the standard error
of the estimate is 237 mmol; if it is estimated
from total body nitrogen, the standard error is
reduced to 186 mmol. Again, the more elaborate
technique gives only a small improvement in
accuracy of prediction.
It appears, therefore, that measurements of
body weight and skinfold thickness, or of total
body potassium, together with the equations
given in Tables 2 and 3, can be used to estimate
body nitrogen in healthy people with only
moderate loss of accuracy. The same may be true
in disease. For instance, in a group of surgical
patients with various degrees of weight loss, not
only was the relationship between total body
nitrogen and fat-free mass similar to that found
here, but the accuracy of estimation was also
similar [17]. However, in other conditions, the
indirect approach is unsatisfactory. For instance,
patients receiving intravenous nutrition may
retain water without increasing body protein [181,
and in gross obesity skinfold thicknesses may be
difficult or impossible to measure.
Acknowledgments
We thank Miss D. W. Krupowicz and Mr K.
Brooks for the total body nitrogen and potassium
measurements, Miss R. A. Siwek for help with
the calculations, members of the City of Leeds
46 1
Leisure Services, the Chief Constable and
Officers of the West Yorkshire Metropolitan
Police Force and members of the hospital staff
who volunteered as control subjects. Development and application of neutron activation
analysis in vivo are supported by a programme
grant from the Medical Research Council.
References
I 1 I DURNIN,
J.V.G.A. & WOMERSLEY,
J. (1974) Body fat assessed
from total body density and its estimation from skinfold
thickness: measurements on 481 men and women aged from
16 to 72 years. British Journal of Nutrition, 32,77-97.
121 OXBY,C.B., APPLEBY,D.B., BROOKS,K., BURKINSHAW,
L.,
KRUPOWICZ,
D.W., MCCARTHY,I.D., OLDROYD,B., ELLIS,
R.E., COLLINS,J.P. & HILL, G.L. (1978) A technique for
measuring total-body nitrogen in clinical investigations using
the “N(n,2n)l1N reaction. International Journal of Applied
Radiation and Isotopes, 29,205-2 I I.
131 WILLIAMS,
E.D., BODDY,K., HARVEY,
I. & HAYWOOD,
J.K.
(1978) Calibration and evaluation of a system for total body in
vivo activation analysis using 14 MeV neutrons. Physics in
Medicine and Biology, 23,405-415.
141 HARVEY,
T.C., JAIN,S., DYKES,P.W., JAMES,H., CHEN,N.S.,
CHETTLE,D.R., ETTINGER,K.V., FREMLIN,J.H. & THOMAS,
B.J. (1973) Measurement of whole-body nitrogen by neutronactivation analysis. Lancet, ii. 395-399.
151 MERNAGH,
J.R., HARRISON,J.E. & MCNEILL,K.G. (1971) In
vivo determination of nitrogen using Pu-Be sources. Physics in
Medicine and Biology, 22,83 1-835.
161 VARTSKY, D., ELLIS, K.J. & COHN, S.H. (1979) In uivo
measurement of body nitrogen by analysis of prompt gammas
from neutron capture. Journal of Nuclear Medicine, 20,
1158-1 165.
171 WEINER,J.S. & LOURIE,
J.A. (1969) Human Biology: A Guide
to Field Methods. I.B.P. Handbook No. 9. Blackwell Scientific
Publications, Oxford.
181 SHARAFI,A.T.M. (1979) Multi-element analysis of the living
human body by neutron activation analysis. Thesis. University
of Leeds.
191 HUME,R. & WEYERS,E. (1971) Relationship between total
body water and surface area in normal and obese subjects.
Journal of Clinical Pathology, 24,234-238.
1101 MCCARTHY,
I.D., SHARAFI,A,, OXBY,C.B. & BURKINSHAW,
L. (1980) The accuracy of total body nitrogen determined by
neutron activation analysis. Physics in Medicine and Biology,
25,849-863.
1111 BURCH,P.R.J. & SPIERS,F.W. (1953) Measurement of the
y-radiation from the human body. Nature (Londort), 172,
5 19-52 1.
1121 MCCARTHY,I.D. (1978) Investigations using a 14 MeV
neutron facility for in viuo activation analysis to determine
total-body nitrogen. Thesis. University of Leeds.
I131 BURKINSHAW,
L. (1978) Sex-dependent calibration factor of a
whole-body radiation counter. International Journal of Applied Radiation and Isotopes, 29,387-390.
1141 NIE, N.H., BENT, D.H., HULL, C.H. (1970) Statistical
Package for the Social Sciences. McGraw-Hill, New York.
1 I5 1 ARMITAGE,P. (197 1) Statistical Methods in Medical
Research, 1st edn, p. 312. Blackwell Scientific Publications,
Oxford.
1161 MCNEILL, K.G., MERNAGH,J.R., JEEJEEBHOY,K.N.,
WOLMAN,S.L. & HARRISON,
J.E. (1979) In uiuo measurements of body protein based on the determination of nitrogen
by prompt y analysis. American Journal of Clinical Nutrition,
32,1955-1961.
1171 HILL, G.L., BRADLEY,J.A., COLLINS,J.P., MCCARTHY,I.,
OXSY, C.B. & BURKINSHAW,
L. (1978) Fat-free body mass
from skinfold thickness: a close relationship with total body
nitrogen. British Journal of Nutrition, 39,403-405.
1181 HILL, G.L., BRADLEY,J.A., SMITH, R.C., SMITH, A.H.,
I.D.,OXBY,C.B., BURKINSHAW,
L. & MORGAN,
MCCARTHY,
D.B. (1979) Changes in body weight and body protein with
intravenous nutrition. Journal of Parenteral and Enleral
Nutrition, 3,2 15-2 18.
462
L. Burkinshaw et al.
APPENDIX
Equations for estimating the values of weight, fat,
fat-free mass and total body potassium to be
expected in a healthy subject were derived from
the results of an earlier survey of 134 men aged
18-77 years and 108 women aged 18-58 years,
carried out in this laboratory; the equations (+SD
from regression) are shown below.
In these equations W is body weight (kg), A is
age (years), H is height (cm), FAT is body fat
calculated from W and S by the method
described by Durnin & Womersley [ l ] , FFM is
weight - fat and TBK is total body potassium
(mmol).
Men: W = -62.40 + 0.24A + 0.72H f 7.3
FAT = -1 1.1 1 + 0.29A + 0.08H f 3.6
FFM = -51.29 - 0.05A + 0.64H 5.0
TBK = -1462 - 7.644 + 30.578 f 416
TBK = 814 - 16.274 + 4 . 3 4 8 + 36-47W 322
Women: W = -44.12 + 0.23A + 0.58H f 6.5
FAT=-13*27 + 0.26A + 0.13H k 4.4
FFM = -30.85 - 0.03A + 0.45H _+ 3.2
TBK = -2149 + 1.10A + 28.058 f 240
TBK=-l339-3.19A+
1 7 * 3 1 H +18.36W f 2 0 9