Evidencefor Formationof
BicarbonateComplexeswith Na+
and K underPhysiological
CondItIons
To the Editor:
z
A previous communication
(1) hypothesized
that values for serum sodium
and potassium
ion concentrations
as
measured
electrode
di-
rect potentiometry
(ISE/DP)
should
6 to 7% higher than those measured
be
by
indirect
by ion-selective
measurements
Q
21
C
C,
z
z
C,
such as flame
photometry
or diluting
electrode
methods.
ion-selective
We agree that the theoretical
differwater
displacement
is considered,
but this
simple model ignores the following: (a)
ence would be 6-7% if only plasma
some serum sodium and potassium
ions
are bound by bicarbonate
and protein in
normal plasma or serum; (b) ion-selective electrodes respond only to free ions
in solution,
while flame photometry
measures total sodium and potassium
concentrations,
including
complexes;
and (c) the 6-7% difference from plasma
water
displacement
applies
only to
serum or plasma samples with normal
concentrations
of protein and lipids. To
suggest that a single value can be used to
interconvert
values obtained
by flame
photometry
and direct potentiometry
for electrolytes
is grossly misleading.
In
flame
photometry,
no distinction
is
made between total plasma volume and
plasma water volume.
Figure 1 illustrates the difference in
results for sodium and potassium
when
solutions with increasing bicarbonate
are measured with both the NOVA 1,
which is an ISE/DP instrument, and the
IL-343, which is a flame photometer.
The solutions prepared were of constant
ionic strength; total sodium and potas-
sium were kept constant at 140 and 4
mmol/L;
and chloride was decreased
from 144 to 104 mmol/L as bicarbonate
was increased from 0 to 40 mmol/L.
Clearly, increasing concentrations
of
bicarbonatecause
a decrease
rather
1938
data shown in Figure 1; they are approximately 1.0 L/mol and 1.3 L/mol,
respectively. Thus, when this measurable binding by bicarbonate
is considered, the actual difference between results by flame photometry
and direct
potentiometry
is 2-3%, as we claimed
earlier (2, 3).
We also point out the need to establish reference intervals for free sodium
and potassium ions in plasma. Our experience indicates that simple salt solutions such as those used by Mohan and
Bates (4) are the most straightforward
and accurate for this purpose. Because
reliable measurement
is now possible
with direct ion-selective electrodes such
as NOVA 1, we urge that conventional
scale reference solutions be developed
by the National Bureau of Standards in
order to standardize
instrument
calibration and accurate measurement
of
electrolytes.
Our main concern, however, is not the
extent of the difference between values
observed by the two methods for normal
the volume was increased by 0.5% and
the pH of the resulting solution still exceeded 7.0 (manuscript in preparation).
At a bicarbonate concentration of 40
mmol/L, the loss of bicarbonate ion as
CO2 gas caused the concentrations of
sodium and potassium ions to increase
from 133.9 to 139.8 mmolfL and from
3.84 to 3.98 mmol/L, respectively,
as
measured by the NOVA 1, but values for
sodium
and potassium
as measured
with
the IL-343 were unaffected.
Czaban and Cormier (1) doubt the
extent of bicarbonate binding to sodium
ions because their calculations showed
that about 20% of the total bicarbonate
would be bound at normal concentrations of sodium, potassium, and bicarbonate in plasma. Indeed, the data in
Table 1 show that their estimate agrees
reasonably well with our data. More
than 15% of the total bicarbonate is
bound at normal serum concentrations
of bicarbonate.
We calculated the stability constants of the sodium and p0tassiuin bicarbonate
complexes from the
Table 1. Expected vs Observed Concentrations of Bicarbonate,
Sodium, and Potassium Ions
Total
HCO
Na
Expected
K
HC03
Na
Conc.
10
20
30
40
than free ion
concentration, the values for sodium and
potassium ion are unaffected by the bi.
carbonate.
This bicarbonate binding can be reversed by adding a calculated amount of
HC1, so it is not an artifact. In so doing,
rnrnol/L)
Fig. 1. Percent increase in bound sodium and potassium with Increasing bicarbonate
concentration
in appar-
ent sodium and potassium ion concentrations asmeasured with the NOVA 1:
2.7% of the total sodium and 3.1% of the
total potassium are bound at 25 mmol of
bicarbonate
per liter. In contrast, because the flame photometer
measures
total concentration
8ICARBONATE
140
140
140
140
4
4
4
4
8.5
17.1
25.6
34.2
Observed
K
NCO,
Na
K
3.95
3.89
3.84
3.79
9.1
16.0
25.0
33.5
138.5
136.9
135.4
4.01
3.90
3.66
3.84
mmoi/L
138.6
137.2
135.8
134.4
Expected concentrations (mmol/L} were calculated by using the determined stability
constants for the sodium and potassium bicarbonate complexes.
CLINICAL CHEMISTRY, Vol. 27, No. 11, 1981
133.9
samples.
Because
clinical
chemistry
laboratories
must measure and interpret
results
for pathological
samples,
we
emphasize
that the inherent
errors of
flame photometry
and diluting ISE’s are
much affected by abnormalities
in protein and (or) lipid concentrations
in
plasma. Therefore,
the model of Czaban
and Cormier, which suggests the incorporation
of a constant
bias or a formula
that includes determined values for total
protein
and lipids for interconverting
values from flame photometry
and direct potentiometry
and of electrolytes
could be grossly misleading.
It circumvents the utility of electrolyte
analysis
by ISE/DP,
which bases its measurement on concentration
per unit volume
of plasma
water. To illustrate,
an instance in which an erroneous
diagnosis
of hyponatremia may have resulted in a
fatality was recently discussed (5). In
this case, the difference between electrolyte measurements
made by direct
and indirect (diluting ISE or flame
photometry) techniques
was 48%.
References
1. Czaban, J. D., and Cormier, A. D.,More on
direct
potentiometry-the
ion-selective
electrode
vs flame photometry.
Clin. Chem.
26, 1923-1924
(1980). Letter.
2. Coleman,
R. L., More on Na+ determination by ion-selective
electrodes
vs flame
photometry.
Clin. Chem. 25, 1865 (1979).
Letter.
3. Coleman,
R. L., Young, C. C., and Sidoni,
L., More on direct potentiometry-the
ionselective electrode vs flame photometry.
Curt.
Chern. 26, 1922-1923 (1980). Letter.
4. Mohan, M. S., and Bates, R. G., Calibration of ion-selective
electrodes
for use in biological
fluids.
Clin. Chem. 21, 864-872
(1975).
5. Dangerous pseudohyponatraemia. Lancet
ii, 1121 (1980).
Robert
L. Coleman
C. C. Young
NOVA Biomedical
Corp.
20 Ossipee Rd.
Newton,
MA 02164
Analytical Goalsfor Quantitative
Urine Analysis:A ClinicalView
We have previously reported studies
on the state of the art achieved in a regional survey (2) and on the short- and
long-term biological variation in urine
analytes in 10 apparently healthy young
men (3). We derived analytical goals for
10 quantitative
urine analytes in the
present studies, using the theoretical
postulates of Harris (4) and an empirical
approach.
All strategies for the delineation of
analytical goals are alleged to have disadvantages (1). We therefore considered
it advantageous
to derive goals in as
many ways as possible. This report details goals for 10 urine analytes, derived
from a small survey of clinical opinion in
our institution.
All clinicians at the Flinders Medical
Centre, a 500-bed university tertiarycare community hospital, were sent a
questionnaire.
This questionnaire
contained an exploratory preamble and, for
each of 10 analytes, at two concentrations, a result and five options regarding
the widest allowable variation from that
result were detailed. A representative
section of the questionnaire
is shown in
Table 1.
Because it has been suggested that
clinicians probably include pre-analysis
variation in their considerations
of the
variability of laboratory results (5), the
questionnaire
was specifically designed
to allow definition of the views of clinicians on both analytical
and biological
variation.
The questionnaire
contained
the following instructions:
A spot urine sample is collected from a
Normal (that is: non-diseased) subject, sent
to the laboratory,
and the First Result is obtained.
The following
day, a second urine
sample
is collected
from the same normal
subject, who has maintained his posture, diet,
fluid intake and is not on any drugs (that is:
pre-analytical variation has been minimized).
A Second
Result
is obtained.
This result
probably
will differ from the first result because of analytical
error and because of the
inherent
Acceptable
standards
of analytical
performance,
often termed “analytical
goals,” have been proposed for those
analytes commonly assayed in plasma or
serum by adoption of a number of theoretical and empirical approaches (1),
but there have been few studies on other
biological fluids such as urine.
from day to day that is a
Because a significant difference in two
consecutive
laboratory
results
is 2.8
standard
deviations
(SD) at p
0.05
(6), the mode of the Laboratory
responses was divided by 2.8 to obtain the
clinically
desirable
and analytical
imprecision.
By analysis
of variance and subsequent division by 2.8, the clinical view of
biological variation was dissected from
the mode of Total responses.
We obtainWreplies,
a 23.6% return
from
the
population
surveyed. The
limited response to this survey could be
due to a number of factors. Some clinicians may have had difficulty in understanding what was required because
there undoubtedly are semantic problems in communication between laboratory and clinical staffs. Some may not
have realized the value of knowing about
the imprecision of laboratory methods.
The survey paper may have been viewed
as an examination
of knowledge and
thus seen as a threat. The design of the
paper may not have been correct; in this
regard, it must be emphasized that the
range of responses from which the clinicians could select the widest allowable
variation was limited to the numerical
values we set.
The analytical goals for laboratory
imprecision derived from the mode of
the responses of the clinicians are shown
in Table 2; goals previously derived from
the state of art (2) and biological variation (3) are also detailed, for comparison. The imprecision
implied to be required by the clinicians for each analyte
for urinary
calcium, glucose,
and urea and proteins at high concentrations, smaller than that derived from
the state of art. It is generally considered
that analytical goals are currently best
derived from biological variation (4) but,
with the exception of urine glucose and
calcium, the goals inferred to be required by the clinicians are more strict
is, except
than
those
derived
from biological
variation.
Our laboratory
can achieve
Table 1. Representative Section of Questionnaire a
To the Editor:
All laboratory measurements have
inherent
imprecision,
but numerical
definition of the degree of imprecision
that is clinically acceptable
is difficult.
variation
perfectly
natural biological phenomenon.
Firstly, please indicate, beside Laboratory,
the variation that you, as a clinician, consider
the maximum
that the second result should
vary from the first for a significant
analytical
change to have occurred.
Secondly,
please
indicate beside Total, the variation that is the
maximum
that the second result should vary
from the first for a significant
change in the
patient
to have occurred;
this change would
be due to laboratory
error
biological
variation.
First
Second result-tick
result
Sodium lOmmol/L
widest
allowable variation
±1
±2
±4
±6
±8
±1
±2
±5
±10
±20
Laboratory
Total
Sodium 100 mmol/L
Laboratory
Total
8 Foreachanalyte, the options from which the clinicians could select their responsewere multiples of
the between-day Imprecision achievable in our laboratory as derived fromreplicate
analysis of qualIty-control
material.
CLINICAL CHEMISTRY, Vol.27,No. 11,1981
1939