Rapid Measurementof Serum Water to ... - Clinical Chemistry

CLIN. CHEM. 32/6, 983-986 (1986)
Rapid Measurementof Serum Water to AssessPseudohyponatremia
Sherry Faye1 and R. B. Payne2
Pseudohyponatremia is caused by an increased serum protein
or lipid concentration producing a “space-occupying
lesion” in serum water. Its presence and magnitude must be
assessed in hyponatremic patients with, for example, paraproteinemia
or hyperlipemic diabetic coma. In the absence of
a direct-reading ion-selective electrode system, a method for
measuring the water content of serum is required. We
describe two rapid methods for measunng the diffusible
water of serum: osmometry beforeand afterdilutionand
chloride measurement before and after ultrafiltration. Either
of these methods allows the true sodium statusof a patient’s
serum to be determined.
AdditIonal Keyphrases: electrolytes osmometry ultrafiltration
paraproteinemia . hyperlipemia ion-selective electrodes
In health, water accounts for 93% of serum volume, only
7% being occupied by dissolved solids. Serum water may be
considerablydecreased in paraproteunemia or severe hyperlipeima,
so that the concentration of a solute such as sodium
is decreased per unit volume of serum but may be normal
per unit volume of serum water. For example, if serum
water isdecreased to 80%, a low sodium concentrationof
120 mmol/L measured by a flame photometer or an indirect
ion-selective electrode USE) system represents a concentration
of 150 mmol/L in serum water and thus a normal
concentration of 140 minol/L m serum with a normal water
content of 93%. Unnecessary treatment of pseudohyponatremia
has caused morbidity and contributed to death (1). A
decrease m serum water content does not always cause
hyponatremia; it may, though less often, mask true hypernatremia
and produce “pseudonormonatremia” (2, 3).
The use of a direct-reading ISE system to measure sodium
concentration in serum water in such patients has been
advocated (4-6) but has some disadvantages that will be
discussed later. The alternative approach is to measure the
diffusible water content of serum sothat the sodium concentration
in serum water and thus normal whole serum can
be calculated from the measured sodium concentration.
Methods based on the measurement of the concentrationin
serum of either protein alone (7-10) or lipid alone (11)
cannot be used for all patients. We have compared three
methods for measuring serum water that can be applied in
both hyperproteinemia and hyperlipeniia.
Methods
Analytical Methods
Total proteinconcentration was determined by a manual
biuret method with a within-batch analytical CV of 3.4%.
Department of Chemical Pathology,St. James’sUniversity Hospital,
LeedsLS9 7FF, U.K.
‘Present address:University Department of Chemical Pathology,
Old Medical School, Leeds LS2 9NL, U.K.
2To whom correspondenceshouldbe addressed.
Received February 18, 1986; acceptedMarch 18, 1986.
Total lipid concentration was measured with a Mercotest
total lipidskitwith a within-batch analytical CV of 5.8%.
Amicon Centriflo membrane cones, type CF25, were used
for ultrafiltration. Before use, the cones were soaked in
distilled water forat leastan hour and then blotted and
centrifuged to remove excesswater. They were capped with
plastic film during centrifugation at room temperature to
minimize evaporation,and the first few milliliters of ultraifitrate
were discarded to avoid dilution with residual cone
water. In later experiments the Amicon Centrifree MicropartitionSystem
was used for ultrafiltration. Chloride concentration
was measured with a Corning-Eel 120 Chloride
Meter or, in later experiments, a Monitor Parallel discretionary
analyzer. The within-batch analytical CVs for single
measurements were 0.85% for the chloride meter and 1.53%
for the Parallel analyzer at normal serum concentrations.
Osmolality was measured with a Precision Instruments
Osmette-S osmometer, an Advanced Instruments Model
64131 osmometer, or a Roebling osmometer. Values measured
on the three instruments did not differ significantly
and the within-batch analytical CVs for single measurements
ranged from 0.25% to 0.65%. The pH of serum was
determined with an IL 313 blood gas analyzer.
Calculations
Serum water was calculated from the total lipid and total
protein concentrationsin serum by using the equationof
Waugh (12):
Serum water (%) = 99.1 - (1.03 x lipid concn) - (0.73 X
protein concn)
where total lipid and total protein are both expressed in
g/dL. This method is based on the average specific volumes
occupied by known weights of serum proteins as determined
by drying and weighing, and of lipids by derivation from
densitymeasurements. The method had a mean withinbatch
analytical CV of only 0.2%;it was small in spite of the
greater analytical imprecisions of the two measurements
because there is a large constant in the equation.
We calculated serum water from the means of replicate
chloride measurements in serum and its ultrafiltrate by
using the equation:
Serum water (%) = (chloride concn in serum/chloride concn
in ultrafiltrate) x 100
This method is based on a principle described by Little and
Payne (13): because chloride binding to serum proteins is
negligible at pH values >7.0 (14), the serum ultrafiltrate
chloride concentration is the same as the chloride concentration
in the diffusible water phase of serum. The means of
triplicate measurements had a mean within-batch analytical
CV of 0.9%.
We measured serum water by osmometry, using the
method of Rawles (15). We preferred this method to an
earlier method based on the same principle requiring osmolality
measurements before and after the addition of an
exact amount of anhydrous sodium chloride (16) because of
CLINICALCHEMISTRY, Vol. 32, No. 6, 1986 983

its technical simplicity. Serum was diluted with an equal
volume ofwater and replicate measurements oftheosmolality
of the original and diluted serum were made to determine
mean values. The calculation is:
Serum water (%) = osmolality of diluted seruml(osmolality
- diluted osmolality) x 100
The basis of this method is that, when a solution containing
a “space-occupying lesion” caused by proteins and lipidsis
diluted with an equal volume ofwater,the reductioninthe
concentration of crystalloids dissolved in the aqueous phase,
and therefore the reduction in the measured osmolality, is
greater than that predicted from the volumes mixed so that
the true volumes of aqueous phase mixed, and therefore
serum water, can be determined. With use of the means of
duplicate measurements, the method had a mean withinbatch
analytical CV of 1.3%.
Results
Preliminary Investigation of the Chloride Method
A pooled serum was ultrafiltered in a Centriflo cone and
the total protein and chloride concentrations of successive
aliquots of the ultrafiltrate and retentate were measured.
The concentrations of chloride in successivefiltrates did not
differ significantly, and they were protein-free. There was a
closenegative linear correlation between the concentrations
in the retentate of protein, which increased from 60 to 140
g/L, and of chloride (r = -0.993). The Deming regression
equation for the retentate values was:
Chloride (mmol/L) = 114.3 - (total protein (in g/L) x 0.178)
The intercept value agreed well with the ultrafiltrate chloride
concentration of 114.1 mmol/L. The constancy offiltrate
chloride concentration was confirmed by prolonged ultrafiltration
of a further six sera (Table 1). The pH values for all
pools of laboratory sera examined exceeded 7.72. We concluded
that there is negligible binding of chloride to protein
in separated sera at the pH reached after exposure to room
air, as predicted from the observations of Carr (14).
Recovery Experiments
Pooled serum from blood donors was centrifuged in a
number of Centriflo cones and the ultrafiltrates and retentates
were pooled. The pooled retentate had nearly twice the
protein concentration of the original pool and therefore had
a low water content. Retentate was mixed with ultrafiltrate
in the proportions 2:1, 1:1, 2:3, and 1:2, and the serum water
concentrations of the retentate and its dilutions with ultraifitrate
were determined eight to 10 times by each of the
Table 1. Chloride Concentrations in Retentate and
Filtrate during Prolonged Ultrafiltration of Six
Laboratory Sera
Chloride concn, mmol/L
15-mmcentrlfugatlon 45-mmcentrltugatlon
Retentate Filtrate Retentate Filtrate
97
97
98
95
96
95
114
114
113
114
114
114
68
71
71
67
68
71
984 CLINICAL CHEMISTRY, Vol. 32, No. 6, 1986
114
114
114
114
114
114
three methods. Water accounted for was calculated as the
difference between the mean percentage water in the concentrated
serum and the product of the mean percentage
water in the dilution and the relative volume of ultrafiltrate
added. Least-squares regression analysis of serum water
recovered vs serum water (ultrafiltrate) added gave correlationcoefficients
>0.9997 in each case and the regression
equations were:
Water recovered (Waugh) =
Water recovered (chloride) =
Water recovered (Rawles) =
(0.9901 x
0.0580
(0.9967 x
1.0293
(1.0293 x
0.1420
water added) -
water added) -
water added) -
The slope, and therefore recovery, by Waugh’s protein/lipid
equation was significantly less than 1.0 (p

‘0 0
w E
0
0
E
L
.l
E
98
96
94
92
90
88
86 -
I I I I
3:7 4:6 5:5 6:4 7:3
Relative volumes of added water and serum
Fig. 1. Serum water (%) by the osmoiality method in six sera as a
function of relativevolumesof added water and serum
triplicate chlorideconcentrationsmeasured in a Monitor
Parallel analyzer in the original sera and in their ultrafiltrates.
Serum water was calculated by the method of Rawles
(15) and by our modification of Rawles’ method from the
mean of duplicate osmolality measurements. The calculated
serum water concentrations did not differ significantly between
the chloride method and our modified osmolality
method (p = 0.90; paired t-test), but were significantly
greater (p =0.0016) by Rawles’ osmolality method(Table 2).
Can Ultrafiltration Be Used to Measure Serum Water
Sodium?
Binding of sodium to serum proteins at pH 7.5 has been
reported (18) and would be expected to increase with increasing
pH. The pH values of our separated serum specimens
all exceeded 7.72.We ultrafiltered eight laboratory
sera, using Centriflo systems as above. While the chloride
concentrationsaveraged 8.1 mmol/L higher in the ultrafiltrates
than in the whole sera (p

5. Swaminathan R, Morgan DB. Pseudohyponatraemia [Letter].
Lancet 1981i:96.
6. Jamieson MJ. Hyponatraemia. Br Med J 1985;290:1723-8.
7. Darrow DC, Hopper EB, Cary MK. Plasmaphoresis edema II.
The effect of reduction of serum protein on the electrolyte pattern
and calcium concentration. J Clin Invest 1932;11:701-15.
& Gilligan DR. Volk MC, Blumgart HL. Observation on the
chemical and physical relationship between the bloodserum and
body fluids: 1. The nature of edema fluids and the evidence
regarding the mechanism of edema formation. J Clin Invest
1934;13:365-81.
9. McLean FC, Hastings AB. The state of calcium in the fluids of
the body. 1. The conditions affecting the ionization ofcalcium. J Biol
Chem 1935;108:285-322.
10. Eisenxnan AJ, McKenzie LB, Peters JP. Protein and water of
serumand cellsofhuman blood,with a noteonthe measurementof
red blood cell volume.J Biol Chem 1936;116:35-45.
11. Steffes MW, Freier EF. A simple and precise method of determining
true sodium, potassium, and chloride concentrations in
hyperlipemia. J Lab Clin Med 1976;88:683-8.
12. Waugh WH. Utility of expressing serum sodium per unit of
water in assessinghyponatremia. Metabolism 1969;18:706-12.
13. Little AJ, Payne RB. The measurement of protein-binding
using dried polyacrylamide. Protides Biol Fluids Proc Colloq
1973;21:643-7.
14. CaIT CW. Studies on the binding of small ions in protein
solutions with the useof membraneelectrodes.ifi. Thebinding of
chloride ionsin solutionsofvarious proteins. Arch BiochemBiophys
1953;46:417-23.
15. Rawles JM. Severe hyponatraemia in hyperlipaemic diabetic
ketosis. Br Med J 1973;i:357.
16. Albrink MJ, Hald PM,Man EB, PetersJP. The displacement of
serum water by the lipidsofhyperlipemic serum. A newmethodfor
966 CLINICAL CHEMISTRY, Vol. 32, No. 6, 1986
the rapid determination of serum water. J Clin Invest
1955;34:1483-8.
17. Terepka AR, Chen PS, Toribara TY. Ultrafiltration: A conceptual
model and a study of sodium,potassium, chloride and water
distribution in normal serum. PhysiolChem Phys 1970;2:59-78.
18. Carr CW. Studies on the binding of small ions in protein
solutions with the use of membraneelectrodes.VI. The binding of
sodium and potassium ions in solutionsof various proteins. Arch
Biochem Biophys 1956;62:476-84.
19 Payne RB, Levell MJ. Redefinition of the normal range for
serum sodium. Clin Chem 1968;14:172-8.
20. Gill GV, Flear CTG. Hyponatraemia.RecentAdv ChinBiochem
1985;3:149-76.
21. Cowell DC, McGrady PM. Direct-measurement ion-selective
electrodes: analytical error in hyponatremia. Clin Chem
1985;31:2009-12.
22. Czaban JD, Cormier AD, Legg KD. Establishing the directpotentiometric“normal”
range for Na/K residual liquid junction
potential and activity coefficient effects. Clin Chem 1982;28:1936-
45.
23. Maas AHJ, Siggaard-Andersen 0, Weisberg HF, Zijlstra WG.
Ion-selective electrodes for sodium and potassium: a problem of
what is measured and what should be reported. Clin Chem
1985;31:482-5.
2& BuckleyBM, BroughtonPMG, RussellLI, Carter TJN. New
ways with old ions. Ann Chin Biochem 1984;21:75-7.
25. Broughton PMG, Smith SCH, Buckley BM. Calibration of
direct ion-selective electrodes for plasma Na to allow for the
influence of protein concentration [Letter]. Clin Chem
1985;31:1765-6.
26. Burgess ED, Schnurr LP, Wilkes P. Ion-selective electrodes and
inaccurate serum sodium measurements [Letter]. Ann Intern Med
1985;103:806.