tration, 6 g/L; f is the dilution factor,
0.02; and n is the number of dilutions.
To measure hemolysis, we pipetted
0.5 mL of test solution and 10 tL of the
threefold-diluted
blood into an 0.8-mL
disposable
polypropylene
sample
Chase AM. The osmotic resistance (fragility) of human red cells. J Clin Invest 26,
636-640 (1947).
lot
Y. S. Huang
75
K. Jenkins
cup
M. S. Manku
supplied for the Cobas Bio micro-centrifugal analyzer (Hoffmann-La Roche
Ltd., Vaudreuil, Quebec). For the complete-hemolysis standard we added 10
L of diluted blood to 0.5 mL of dis-
J. Davignon
50
Efamol Res. Inst.
P.O. Box 818
Kentville,
Nova Scotia
Canada B4N 4H8
and
Clin. Res. Inst. of Montreal
0
a
tilled water. Each cup was capped, and
the contents
were mixed
gently
and
allowed to stand at room temperature
for 30 mm. The cup was then centrifuged in a micro-scale centrifuge at
8000 x g for 2 mm, to pack unhemolyzed cells and membrane debris into
the tip of each cup. The cups were then
placed, without disturbance and in order of increasing hypotonicity,
into the
sample disc of a Cobas Bio analyzer.
Supernate from the completely hemolyzed erythrocytes (100% hemolysis),
in triplicate,
was
pipetted
into
25
I
50
analyzer
were
as follows:
tained with this automated method
Table 1. Reproducibility of the
OsmoticFragilityCurve for Blood
0 5; start
1 (water blank); printout
mode, 1.
The samples and the water blank
are automatically pipetted into the disposable cuvette rotor supplied for the
Cobas Bio and mixed by centrifugal
force, whereupon the measurement
is
then made and the reading printed out
automatically and directly as the percentage of hemolysis in each sample.
The osmotic fragility curve is then
constructedby plottingthe percentage
of hemolysis vs the concentrations of
the test solutions. A typical curve thus
obtainedisshown in Figure 1.
To assess our method, we repeatedly
measured
the values of Hso (the concentration
of test solution
at 50% hemolysis) and H2575 (the difference be-
tween the concentration of the test
solution at 25 and 75% hemolysis) in
14 preparations of normal blood. The
results (Table 1) indicate excellent reproducibility.
The CVs ranged from
0.22% to 0.75% for H and from 2.84%
to 8.82% for H25_75.The mean of H50
(4.460 g/L) for 14 subjects studied
agrees well with the mean reported by
Maeda et al. (2), 4.408 (SD 0.091) g/L,
who used the multiple-tube method (3).
With our procedure, the percentages
of hemolysis
of all samples
tested
are
Rawal et al. (Clin Chem 29: 586,
1983) recently concluded that the fluorescence polarizationimmunoassay
(FPIA) of serum digoxin (from Abbott
H
4.423
4.637
4.513
4.370
4.335
4.493
4.587
H25..75
± 0.018
± 0.016
±
0.022
0.023
0.010
0.018
0.023
4.343 0.008
4.571 ± 0.013
4.455 ± 0.010
4.457 ± 0.012
4.355 ± 0.021
4.492 ± 0.023
4.413 ± 0.033
±
±
±
±
±
0.228 ± 0.010
0.238 ± 0.021
0.272 ± 0.010
0.267 ± 0.008
0.302 ± 0.012
0.297 ± 0.010
0.377 ± 0.012
0.268 ± 0.017
0.352 ± 0.010
0.225 ± 0.013
0.253
0.022
±
0.342 ± 0.008
0.375 ± 0.012
0.395 ± 0.015
4.460 ± 0.095 0.299 ± 0.059
measured directly and simultaneously
vs both the completely hemolyzed sample (100%) and water (0% hemolysis);
hence, no further calculation
is required. In each run, a sample in 24
different
hypotonic
test solutions can
be measured simultaneously
within a
few minutes. Several sets of assays can
be preparedat the same time, and the
entire procedure, including blood preparation,
incubation,
h or longer for the multiple-tube
od.
meth-
References
1. Kitazima K, Shibata S. Coil planet
trifugationand itsapplicationto the
cen-
obser-
altered membrane properties of
erythrocytes in hepatobiliary disorders. J
Lab Clin Med 85, 855-864 (1975).
2. Maeda N, Aono K, Sekiya M, et a!. A
computerized method for the determination
of the osmotic
fragility curve of erythrocytes.AnalBiochem
83, 149-161 (1977).
3. Parpart AK, Lorenz PB, Parpart ER,
of
Laboratories,
Diagnostics
Division,
North Chicago, IL 60064) offers many
advantages, mainly reagent stability
and assay speed, as compared with
radioimmunoassay
(RIA). They compared the FPIA and the “Amerlex”
(Amersham Corp., Arlington Heights,
IL 60005) RIA for determining digoxin
in 223 patients’ samples, and the correlation was satisfactory.
We assayed samples for digoxin with
the kit we use routinely, the Phadebas
digoxin RIA (Pharmacia Diagnostics
AB, Uppsala 1, Sweden) and the Abbott FPIA, and were surprised to find a
poor correlation for 28 patients’ samples (Figure 1). Although the correlation coefficient (r) was 0.975, there was
a consistent and appreciable bias between the two techniques,with values
being lower for the FPLA than for the
RIA.
Common
quality
criteria
(repro-
4
and assay, takes
less than an hour, as compared with 2
vation
FluorescencePolarization
Immunoassay
To the Editor:
6
6
6
6
6
6
6
6
8
4
6
6
6
6
reagent volume, 0 pL; time of first
reading, 0.5 5; time interval, 30 s;
number of readings, 2; blanking mode,
Effect of Deprotelnization on
Determination of Serum Dlgoxin by
Measured by the Described
Procedure
Soin concn, mean ± SD, 9/1
tests
540 nm; sample volume,50 tL; diluent
volume (water), 25 .tL; reagent voltime,
112W 1R7
from 14 NormalPersons,as
No. of
100; standard 3 concentration,
100;
limit, 100; temperature
25 #{176}C;
type of
analysis, 1 (fixed point); wavelength,
ume, 0 zL; incubation
Canada
4,0
Fig. 1. A typical osmotic fragility curve ob-
the
reaction direction, +; units, own; calculation factor, 0; standard 1 concentration, 100; standard 2 concentration,
4.5
NoCI,
standard cups 1, 2, and 3 in the disposable plastic reagent tray.
The settings for the Cobas Bio microcentrifuge
ilOPine Ave. W.
Montreal, Qu#{233}bec
U-
z
0
C,
0
DIGOXIN
RIA
Fig.
1. Correlation between results by the
FPIA and RIA for serum digoxin (nmol/L):y =
0.856 x -0.36 (r = 0.975, n = 28)
CLINICAL CHEMISTRY, Vol. 30, No. 2, 1984
337
ducibility, linearity) and specificity of
antibodies cannot explain this discrepancy. After investigating several variables, we finally noticed the difference
in protein concentration in the serum
standards in the two assays: 49 g/L for
FPIA and 66 g/L the Pharmacia RIA.
We did not investigate the protein concentration of the Amerlex RIA standards.
For nine of the patients, whose serum protein concentration ranged from
62 to 75 g/L, digoxin concentrations
measured by FPIA were about 20 to
30% less than those measured by the
Phadebas RIA. However, when the
lid results unless one is certain that
the RIA method is reliable in monitor-
>+3O
ing the narrow range between adequate therapy and toxicity. Obviously,
if the individual laboratory has established the therapeutic range by using a
520
40
60
PROTEINS
80
iOo
(alL)
Fig. 2. Analytical recovery of serum digoxin,
as a function of protein concentration (4.1
nmol/L)
We conclude that the FPIA of digox-
as in control sera, precise values were
in does not give accurate results when
the protein concentration of the sample
is within the normal range (60-80 g/
L). This interference must be obviated,
obtained whichever assay was used
(Table 1). Interference by protein can
to avoid underestimation
of digoxin,
especially because this drug has a low
protein concentration
was low enough,
therapeutic
index.
RIA
FPIA
FPIA
Apparent digoxin concn,
nmoi/L
g/L
Patients’ samples
75
2.95
1.90
2.55
73
2.40
1.65
2.75
71
0.65
0.30
0.75
71
0.95
1.10
0.65
69
1.40
0.75
1.20
66
3.25
2.25
3.0
66
2.0
1.55
2.10
3.05
66
4.0
3.80
1.80
62
2.35
2.65
FPIA control sera’
49
4.20
4.35
4.30
49
1.0
0.90
0.95
49
2.05
1.80
1.75
Mean
2.27
1.74
2.25
After twofold dilution with saline (NaCI, 9 g/L).
#{176}FPIA
digoxin controls contain 0.96, 1.92, 4.48
nmol of digoxin per liter in normal human serum.
J. M. Scherrmann
R. Bourdon
Universit#{233}
Paris V
Lab. Biochim.
Toxicol.
HOpital Fernand
Widal
200, Rue du Faubourg
Saint Denis
75010 Paris, Cedex 10 France
The
authors
We further confirmed this protein effect by investigating
seven sera, all
containing the same concentration of
digoxin, 4.1 nmolIL, but having different protein concentrations, 5 to 100 g/
L. The higher the protein concentration, the lower the apparent digoxin
concentration,
by 10 to 30% (Figure 2);
we observed the same range of difference for the patients’ samples.
Thus protein at concentrations within the normal range can substantially
interfere with the FPIA procedure for
digoxin. This can explain the recent
finding of Erickson et al. (Clin Chem
29: 1239, 1983) that digoxin concentrations measured with the FPIA were
statistically lower than the mean RIA
values. Samples should therefore be
deproteinized
with trichloroacetic
acid
and the supernate assayed.
338
FPIA method, however.
References
1. Rawal N,
Leung
FY, Henderson AR.
polarization immunoassay
of
serum digoxin: Comparison of the Abbott
Fluoresence
immunoassay with the Amerlex radioClin Chem 29, 586 (1983).
of the letter
in question
Letter.
2. Holtzmann JL, Shafer RB, Erickson RR.
Methodological causes of discrepancies
in
RIA for digoxin in human serum. Clin
Chem 20, 1194-1 198 (1974).
3. Voshal! DL, Hunter L, Grady HJ. Effect
of albumin on serum
digoxin radioimmunoassays. Clin Chem 21, 402-406 (1975).
4. Goldstein J. Albumin and digoxin. Radioassay News 66, August 1975.
Naresh Rawal
respond:
F. Y. Leung
A. R. Henderson
To the Editor:
It is not uncommon to find unsatisfactory correlations
between assays,
includingtwo different commercially
available
radioimmunoassays
(RLA)
for digoxin. Our comparison between
the Abbott Laboratories digoxin fluorescence polarization
immunoassay
(FPIA) and the Amerlex RIA showed a
satisfactory
be suppressed by diluting serum samples with an equal volume of isotonic
NaCL. When this was done, we found
the mean digoxin concentration measured by the two assays was identical.
RIA method and wishes to
maintain the same reporting ranges
when using another technique, it may
be necessary to make adjustments in
the new assay, e.g., by diluting samples for FPIA with saline to obtain
comparable results. This practice may
not be satisfactory for all users of the
immunoassay.
Table 1. Digoxinas Measuredby
FPIA and RIA in Samples with
Various Protein Concentrations
Protein
concn,
particular
correlation
of results
(1).
We have no such experience with any
other commercially available R1A digoxin assays.
Variations in serum protein concentrations also affect the results of digoxin by RLA (2-4). The manufacturers of
the Amerlex
Digoxin RIA kit have
shown it to be insensitive to changes in
serum protein concentrations (package
insert). Our correlation was therefore
with a RIA method that was already
Dept. of Clin. Biochem.
University
Hosp.
London,
Ontario, Canada
RapidQuantitativeIsolationand
Esterification of Urinary Porphyrins
for Chromatographic Analysis
To the Editor:
Determination
nary porphyrin
of the pattern of uriexcretion
is useful for
the differential
diagnosis of disorders of
porphyrin
metabolism
(1). Conversion
of porphyrins
present
in biological
samples to the corresponding methyl
esters yields compounds that can be
easily separated for individual quanti-
fication,
either
by inexpensive
thin-
tested for and found unaffected by
changes in serum protein concentrations. Drs. Scherrmann and Bourdon
layer chromatography
(2) or by isocratic “high-performance”
liquid chromatography (1, 3, 4). Current techniques
did not include any data on the effects
of changes in serum protein concentrations on the results by their in-house
RIA for digoxin. It would be interesting
for isolation and esterification of urinary porphyrins (2) are, however, cumbersome and time-consuming,
and
to know what effect twofold dilution of
patients’ samples with isotonic saline
would have on their RIA results.
Before one can accept the validity of
porphyrin
a comparison
method, it must be
shown to be accurate and precise. Modifying the FPIA procedure to approxi-
mate one’s RIA results could give inva-
CLINICAL CHEMISTRY, Vol. 30, No. 2, 1984
they can entail
We
substantial
losses of
materials.
now
report
a procedure
that
yields quantitative
isolation and esterification of porphyrins from urine
samples in about a third the time required with previous methods, involving fewer manipulations and less laboratory equipment.