CLIN. CHEM. 36/3, 565-567 (1990)
a1-Acid Glycoprotein Decreases Recovery of Total Protein in Urine When Trichloroacetic
Acid Is Used to Precipitate the Proteins
J. P. Bellby and B. A. O’Leary
Total urine protein was measured in 132 samples by an
manual
automated
This difference was found in urines that contained concentrations of a1-acid glycoprotein (AAG) > 0.10 g/L. Here we
describe the effect of AAG on measurement
of urinary
proteins by the PS,TCA method.
benzethonium
chloride
method and the Ponceau-
S/trichloroacetic acid (PS/TCA) method. Of these, 27% gave
a result 0.1 g/L or more higher by the benzethonium chloride
method. Of this 27%, most contained an abnormally high
concentration of the acute-phase reactant, a1-acid glycoprotein. By assaying urine containing added a1-acid glycoprotein and albumin, we found that a1-acid glycoprotein causes
the PS/TCA method to underestimate the total urine protein
concentration,
whereas the benzethonium chloride method is
unaffected. Not allurinaryalbuminwas precipitated
by TCA
when a1-acid glycoprotein was present. Therefore, protein
methods in which trichioroacetic acid is used as a concentrating step before the assay will underestimate total urine
protein when the concentration of a1-acid glycoprotein is
acid (PS/TCA)
method.
MaterIals and Methods
Apparatus
We used a Cobas-Bio centrifugal
analyzer (Roche Diagnostics, Sydney, Australia). For electrophoresis we used the
Paragon System (Beckman Instruments,
Sydney, Australia), and before electrophoresis
urine was concentrated
with Minicon B15 concentrators (Amicon Scientific Australia, Victoria, Australia).
Reagents
high.
Additlenal Keyphrases: analytical error
nde method compared
benzethonium chlo-
Total urine protein has been quantified
by many different techniques,
including turbidimetry, colorimetry, and
dye-binding methods. In the 1988 Australasian
Urine
Quality Assurance Programme,
55% of laboratories used
methods with either trichloroacetic acid (TCA) or su]fosalicylic acid in their reagents, 30% used a Coomassie Brilliant Blue dye-binding method, and 12% used the alkalinized benzethonium chloride (BZC) method (1).1 Comparisons of the use of sulfosalicylic
acid and TCA show a
systematic bias because of the variable reactivity between
these two methods with different proteins (2). The BZC
method reportedly has a positive bias when compared with
a sulfosalicylic
suggested that
interfere with
et al. (3) found
determination.
Ponceau-S/trichloroacetic
acidlsodium sulfate method, and it has been
there may be compounds in the urine that
the BZC method. Because of this, Flachaire
the BZC method unsuitable for urine protein
The BZC method as proposed by Iwata and Nishikaze (4)
relies on the formation of turbidity with proteins at an
alkaline pH. McDowell (5) automated the BZC method for
use in a centrifugal analyzer, the Cobas-Bio, finding the
standard curve to be stable for 30 days without drift and
giving excellent precision, with CVs ranging from 1.5% to
3.7%. Results correlated well with those by a manual TCA
urine protein method; however, Iwata and Nishikaze (4)
noted that several urines gave widely divergent results, the
BZC results always being higher.
We also noted such discrepancies when comparing a BZC
method for urine proteins as used in the Cobas-Bio with a
Department of Clinical Biochemistry, Queen Elizabeth H Medical Centre, Nedlands, Western Australia, 6009, Australia.
1 Nonstandard
abbreviations: TCA, trichloroacetic acid; BZC,
benzethonium chloride; PSITCA, Ponceau-S/trichloroacetic acid;
and AAG, a1-acid glycoprotein.
Received April 12, 1989; acceptedDecember 20, 1989.
Human AAG, AAG antibody, and BZC were from Sigma
Chemical Co., St. Louis, MO. Antiserum to the TammHorsfall mucoprotein
was from Organon
Teknika,
West
Chester, PA. All other reagents were “Analar” grade from
BDH Chemicals Ltd., Victoria, Australia. Dade Human
Protein
standard
(cat. no. B5158) was from American
Monitor Corp., Aquada, Puerto Rico. “Humitrol
Normal,”
used as quality-control
material, was supplied by Commonwealth Serum Laboratory, Melbourne, Australia. AgaroseM was from Pharmacia, Sydney, Australia. Standards and
controls
for radial
Diagnostics
immunodiffusion
Australia,
were from Behring
Sydney, Australia.
Procedures
All urines were centrifuged
for 10 mm at 3000 x g before
assay.
PS/TCA method. Urine proteins were determined by a
method based on the Pesce and Strande PS/TCA method
(6), which can be described as follows. Pipet 200 L of
urine, quality control, standards, or water blank to each
glass tube. Add 2 mL of the PSITCA reagent to each tube,
and vortex-mix. Centrifuge (3000 x g, 10 mm), drain the
supernate from each tube, and wipe the inside of the tube
with a cotton swab to remove excess dye. Add 2 mL of 0.2
moIJL sodium hydroxide reagent, vortex-mix,
and measure
the absorbance at 560 nm, setting the blank to zero; then
calculate the protein concentration.
Benzethonium
chloride method. Total urine protein concentration was determined by the BZC method in a CobasBio containing a “dens” plotting program (4, 5). The BZC
was made up to 10 g/L with de-ionized water, and the
alkaline diluent contained 20 g of sodium hydroxide and
13.8g of tetrasodium
EDTA per liter. The Cobas-Bio was
programmed as follows:
1
2
3
Units
Calculation factor
Standard 1 concn
2
3 ,,
,,
g/L
1300
0.0
0.2
0.4
CLINICAL CHEMISTRY, Vol. 36, No. 3, 1990
565
Standard
6
7
8
9
10
11
4 concn
5
6
0.8
II
1.6
Limit
Temp, #{176}C
Type of analysis
0
37.0
7.6
450
Wavelength,
nm
Sample vol, L
Diluent vol, L
8
12
13
14
Reagent vol, L
Incubation time, s
Start reagent vol, pL
15
Time of first reading,
Time interval, s
Number of readings
16
17
18
19
35
320
60
16
and quality
controls.
5
6
7
S 9 10
(-)
(lanes 1-5) and AAG immunofixation (lanes
6-10) of urineswith discrepantproteinresults
1
-ig.
.5
s
Dade Human
2
1.
3
4
iectropnoresis
Samples in lanes 1-5 are in the same orderin lanes 6-10
20
21
0
5
Blanking mode
Printout mode
Standards
(+)
1.2
region.
Protein
Standard,
a lyophiuized
preparation
of human albumin,
was reconstituted
in de-ionized water and diluted with
isotonic saline to give a protein concentration of 1.60 g/L.
Humitrol
Normal (diluted 100-fold with isotonic saline)
was used for quality control.
Dialysis
crperiment
to detect TCA -soluble proteins. Add
120 mL of TCA (30 g/L) to 6 mL of urine, mix, and
centrifuge (3000 x g, 10 mm). Dialyze the supernate
against phosphate buffer (0.1 molJL, pH 7.4) for 24 h to
remove the TCA. Concentrate the solution 100-fold in a
Minicon B15 concentrator and subject this to electrophoresis in agarose gel, using the Beckman
Paragon system.
Immunofixation.
After the concentrated
urines were electrophoresed,
the gels were exposed to a solution of antibodies-either
anti-AAG
or anti-TammHorsfall
protein-at
room temperature for 2 h in a humid chamber, washed
several times in isotonic saline, dried, and stained with
Paragon blue.
Radial immunodiffusion.
Radial immunodiffusion
plates
were made according to the method of Mancini et al. (7).
Anti-AAG, 2.5 mL, was mixed into 100 mL of a 12.5 g/L
solution of Agarose-M.
Urinary albumin. Urinary albumin was assayed in the
Cobas-Bio by an immunoturbidimetric
method.
Addition of proteins to urine. Various amounts of AAG,
transferrin, IgG, and human serum albumin were added to
a urine containing no detectable protein by either method.
The urine protein was measured by each of the protein
methods.
Statistics. Results for urinary protein were analyzed by
Deming’s analysis.
Immunofixation
proved the major
protein
in the
band to be AAG (Figure 1).
Several urines that gave discrepant
results were investigated by precipitating
the proteins with TCA (at the same
concentration
as used in the PS/TCA method) and resuspending the precipitate
in sodium hydroxide. When the
precipitated protein was measured by the BZC and PS/TCA
methods, the results were similar, demonstrating
that the
TCA was precipitating
only some of the proteins and that it
was the precipitation
step that was causing the results to
differ. After removing the TCA by dialysis and concentration, we examined the supernate by electrophoresis. The
protein patterns of the supernate were similar to those of
the untreated urine samples, demonstrating
that many
proteins were not being precipitated by TCA. As shown by
immunofixation,
the most abundant of these proteins were
AAG and albumin. Gradient polyacrylamide
gel electrophoresis revealed the presence of at least eight other
proteins, with
90000 Da.
We studied
results by the
AAG, as much
molecular masses ranging from 14000 to
the effect of various protein mixtures on
two protein assays by adding albumin or
as 1.0 g/L, to a urine with an undetectable
protein concentration and assaying. By both methods 95%
to 105% of added albumin was accounted for. However,
when only AAG was added, it was not detected by the
PS/TCA method, and only 10% of it was accounted for by
the BZC method.
When we assayed urines containing various concentrations of AAG and albumin, the two methods gave significantly different results: the BZC method continued to give
results reflecting a 100% recovery of albumin and a 10%
recovery of AAG, but recovery by the PS,TCA method was
suppressed by up to 82% with increasing amounts of AAG
in the urine (Table 1). Similar suppression was noted when
AAG was added to urine containing IgG or transferrin
Results
Assay of 139 urine specimens for urine protein by the
BZC and the PS/TCA methods yielded a wide scatter of
results, with a correlation coefficient of 0.857, a regression
equation
of y(BZC) = 0.94x(PSITCA) + 0.08 g/L, and a
standard error of the estimate ofy = 0.136 g/L. Of the urine
samples, seven (5.0%) gave results more than 0.1 g/L lower
by the BZC method. These urine samples had been collected in an acid preservative (see below), causing the BZC
method to give falsely low results. On the other hand, for 38
(27.3%) of the urine samples, the protein concentration was
0.1 g/L or more higher when assayed by the BZC method;
for one urine the difference was 1.22 g/L. Electrophoresis of
these samples showed a significant band in the a1-globulmn
566 CLINICAL CHEMISTRY, Vol. 36, No. 3, 1990
Table 1. Effect of Addition of AAG and AlbumIn on
UrInary ProteIn Measured by the PSITCA Method
Albumin added, g/L
AAG added to
urine, g/l.
0
0.05
0.10
0.25
0.50
1.00
0
<0.02k
<0.02
<0.02
<0.02
<0.02
<0.02
0.05
0.06
0.04
0.03
0.02
<0.02
<0.02
0.1
0.25
0.50
1.0
0.11
0.26
0.29
0.28
0.18
0.09
0.04
0.52
0.52
0.62
0.54
0.32
0.17
0.97
0.95
0.98
1.07
1.03
0.66
0.11
0.08
0.04
0.02
<0.02
Protein measured,g/L (mean of triplicate determinations).
Table 2. ProteIn Measured In Urine from Selected
Patients by BZC and PS/TCA
ProteIn, g/L
PSITCA
BZC
0.02
0.76
AAG, g/L
0.25
2
0.03
0.29
0.15
3
0.07
0.74
0.28
0.65
1.44
0.96
0.58
0.26
0.29
0.24
0.27
0.50
Specimen
1
4
5
6
7
8
9
10
ND, notdetected.
0.05
0.22
0.08
0.27
0.05
<0.02
0.07
0.17
0.18
ND
ND
ND
before assay.
Table 2 lists the concentrations of AAG and total protein
as measured by the PS/TCA and BZC methods in patients’
urines. Differences
between results by the two methods
were largest when the concentration of AAG was the
highest. Also, some urine samples gave discrepant
results
by the two protein methods even when AAG was undetectable. We attribute
this finding to the different mix of
proteins in these patients’ urine, or perhaps the presence of
other proteins in the urine causes suppression similar to
AAG in the PS/TCA method.
Discussion
Our findings have significant consequences for urine
protein measurements when the PS/TCA method is used.
We have demonstrated that AAG is not assayed by the
PS/TCA or the BZC method. The presence of AAG also
suppresses the recovery of albumin, transferrin, IgG, and
probably other proteins in urine when the PS/TCA method
is used.
McDowell (5) reported a number of divergent results for
urinary protein when the BZC method was compared with
a manual TCA assay, but offered no explanation for these
differences. Shiba et al. (8) reported that discrepant results
between a Coomassie Brilliant Blue method and a sulfosalicylic acid method were due to some urine proteins not
precipitating
in sulfosalicylic acid that did react with the
Coomassie Brilliant
Blue method. They (8) found these
differences in urines from patients with malignant tumors,
pneumonia, rheumatoid arthritis, and renal disorders and
in patients after major surgery, in whom AAG was the
major urine protein. In our study, most urines containing
AAG were from patients with renal disease, tumors, and
severe infections.
AAG is an acute-phase protein synthesized in the liver
and leukocytes. It has a molecular mass of4l 000 Da and is
about 40% carbohydrate (9). AAG is excreted in urine from
pregnant women and patients with inflammation
or cancer. These conditions may be accompanied by increased
glomerular permeability and reduced tubular reabsorption
of AAG. The reference interval for AAG excretion in the
urine of healthy adults has been reported as 0.29 to 0.68
mg/d (10). We have noted urine concentrations of AAG at
approximately 1000 times this reference interval.
Apparently, no protein assay measures all urine proteins
well. To overcome these problems, clinical chemists should
replace urine protein
specific urine proteins,
insight into the type of
of urine quality-control
assays with the measurement
of
which would also provide greater
proteinuria
present. Moreover, use
material
that contains albumin as
the only protein will not detect differences between urine
protein methods for different protein mixes.
In summary,
methods in which TCA is used to precipitate proteins as a concentration step before assay will
underestimate
total urine protein concentration when the
AAG concentration is increased, a common finding
in
patients with an acute-phase reaction.
We thank Mr. J. Blenahasset from the Department of Biochemistry, Royal Perth Hospital, Perth, Western Australia, for the BZC
methodology.
References
1. Penberthy LA. Results from cycle 8, 1988 Australasian Urine
Quality Assurance Programme. Clin Biochem Rev 1989;10:169.
2. Glenn CC, Hathaway TK. Urinary chemistry-a new CAP
program. Am J Clin Pathol 1977;68:153-.8.
3. Flachaire E, Dainour 0, BienvenuJ, Aouiti T, Later R. Assessment of the benzethonium chloride method for routine determination of protein in cerebrospinal fluid and urine.Clin Chem
1983;29:343-5.
4. Iwata J, Nishikaze 0. New micro-turbidimetric method for
determination of protein in cerebrospinal fluid and urine. Clin
Chem 1979;25:1317-9.
5. McDowell T. Benzethonium chloride method for proteins
adapted to centrifugal analysis. Clin Chem 1985;31:864-6.
6. PeseeMA, Strande CS. A new micromethod of determination of
protein in CSF and urine. Clin Chem 1973;19:1265-7.
7. Mancini G, Carbonara AO, Heremans JF. Immunochemical
quantitation
of antigens
by single radial iminunodiffusion. Immunochemistry 1965;2:235-54.
8. Shiba KS, Kanamori K, Harada T, eta!. A cause of discrepancy
between values for urinary protein as assayed by the Coomassie
Brillant Blue G-250 method and the sulfosalicylic acid method.
Clin Chem 1985;31:1215-8.
9. Schmid K. a1-Acid glycoprotein. In: Putnam FW, ed. The
plasma proteins: structure, function, and genetic control, 2nd ed.
New York: Academic Press, 1975:183-228.
10. Tietz N. Textbook of clinical chemistry. Philadelphia: WB
Saunders Co., 1986:1811.
CLINICALCHEMISTRY, Vol. 36, No. 3, 1990 567