231
Clinica Chimica Acta, 90 (1978) 231-239 0 Elsevier/North-Holland Biomedical Press
CCA 9776
AUTOMATED
MEASUREMENT
OF URINARY
PROTEINS
ANDREW MUIR * and W.J. HENSLEY Department (Received
of Biochemistry,
Royal J’rince Alfred Hospital, Sydney N.S. W. 2050 (Australia)
May 29th, 1978)
Summary We have investigated the automation of the trichloroacetic acid turbidometric method for the estimation of urinary proteins using a standard centrifugal analyser. Conditions of temperature, trichloroacetic acid concentration, sample and reagent volumes have been examined. The method requires 10 ~1 of urine sample and results have been shown to be linear to 2.0 g per litre with undiluted specimens; higher levels of protein require dilution of the urine samples to obtain greater accuracy. The coefficient of variation at a level of 0.5 g per litre is 3.5% and at 1.0 g per litre is 1.3%; recoveries ranged from 98.9 to 108%. The estimation of urinary protein in 29 samples may be carried out in 5 min.
The measurement of urine protein is complicated by several factors. Urinary protein estimations are regarded by most workers in routine laboratories as one of the least interesting of the analytical techniques carried out and for this reason accuracy and precision may suffer. Variability of pigments present due to hydration and dehydration affects the basic colour of the urine; also, the presence of interfering drugs necessitates the use of individual blank estimations. Blank estimations are usually made by simple dilution of the urine with distilled water in place of reagent, when in fact the reagent may initially, or during the course of the reaction, react with the interfering substance. For this reason it is essential to carry out blank reactions in the presence of reagent where possible. The wide range of protein concentrations found in urine specimens, both normal and pathological (0 to 6 g per litre), has posed problems; some methods e.g. turbidometric ones, have been shown to be sensitive in the range 0 to 1 g * Correspondence should be addressed to: Dr. A. Muir. Department of Biochemistry. Hospital, Missenden Road, Camperdown. N.S.W. 2060. Australia.
Royal Prince Alfred
232
per litre, whilst other methods e.g. using the Biuret reaction, are less sensitive in this range but more satisfactory in the higher protein concentration range. This difference in sensitivity by methods is reflected in the normal ranges quoted in the literature. Turbidometric measurements have been shown to be more sensitive in that range which divides normal from abnormal proteinuria; we have therefore examined the possibility of an automated trichloroacetic acid (TCA) turbidometric method with a view to overcoming some of the problems associated with the measurement of urinary protein. Reference
standard
Trichloroacetic acid (TCA) was introduced in 1921 by Mestrezat [l] for the determination of protein in spinal fluid. Since then TCA has gained popularity as a turbidometric reagent because the discrepancies between the turbidities produced by albumin and globulins are less than with sulphosalicyclic acid. Henry et al. [2] state that in their hands gamma globulin produced 20% higher turbidity than albumin and recommended the use of dilute human serum for a reference standard. Rice [3] and Rice and Loftis [4] used recrystallised bovine albumin as a source of standard in the TCA turbidometric method. Pure human albumin is readily available and may be weighed accurately and thus used as a primary standard. It would therefore be a satisfactory standard provided conditions can be controlled such that albumin and globulins give the same turbidity per weight of protein. Mixing of reactants
With all turbidometric and precipitation methods, the order of addition of reactants, the concentration of reactants and the method and degree of mixing, control the particle size of the precipitate formed. The method of mixing is also responsible for the quantity and size of entrapped air bubbles which may be in the final suspension. The rate of formation and flocculation of the precipitate is variable with both type and concentration of the protein. The manual addition of reactants, with inevitable variations in timing and in degree of mixing poses a problem with turbidometric methods. Temperature
It has been reported that turbidity varies with temperature when using TCA; [3,5] between 20°C and 25°C variation of 25% can occur. At a temperature greater than 25°C greater turbidity is obtained with albumin than with globulins [ 51. Temperature control is therefore essential. Time
The time interval between the mixing of reactants absorbance has been found to be important [3].
and the measurement
of
Materials and method Reagents
Trichloroacetic
acid (TCA) from Ajax Chemicals
(Sydney),
analytical
grade,
233
used. In the initial experiments concentrations of 3, 5,10, and 20% (w/v) were used. Immediately before use the reagent was filtered through a Millipore “Millex” 0.22 pm membrane.
was
Standard protein Pure human albumin from Behring (Hoechst, Australia) was used as a primary standard. Diluted Technicon S.M.A.C. reference serum was used for comparison. The instrument The instrument trifugal analyser .
chosen
for automation
was a Centrifichem
l.lz
1.10
LOS
1.06
0.04
L%T:IN
CCh’lCENTR4TION
Fig. 1. Effect of TCA concentration
( ffih.
)
(in 5%)on the standard albumin calibration graph.
300 series cen-
234
Experimental
results
Wavelength Preliminary analysis showed that absorbances read at a wavelength of 340 nm were 2 to 2.5 higher than absorbances read at 405 nm which is the wavelength usually used for turbidometric analysis. All readings were therefore made at 340 nm. Concentration of trichloroace tic acid Fig. 1 shows the effect of TCA concentration calibration graph, absorbance of protein turbidity concentration of TCA. Effect of temperature We examined the kinetics
BSORBANCE
of albumin
on the standard albumin increasing with increasing
and serum protein
precipitation
%ONM.
i
0.06
.-•
.A”_ SERUM
--.
97.5
MG/DLa
./
_0,05
.
./
0.04
Q-03
.-•-. -.
./-’
OS02
ALBUMIN
I.00
MG/DL.
\.
./
Fig. 2. Kinetics
of protein
TIME (MINUTES ) precipitation for human albumin
and semm protein
at 25’C.
at 25,
235
30, and 37°C with time. The results are shown in Figs. 2, 3 and 4. It can be seen that at 25 and 30°C serum protein yields higher absorbance for the same protein concentration than does albumin over a lo-min period. At 37°C however (Fig. 4), there is a crossover point between 1 and 2 min where both yield the same absorbance value. Further experiments at varying protein concentrations have confirmed that albumin and globulins yield the same turbidity with 5% TCA at 37°C when the absorbance is measured between 1 and 2 min after mixing of the reactants. An example of some of these experiments is shown in Fig. 5. This kinetic pattern is similar to the reactions we have observed with immuniprecipitation reactions. A comparison between albumin and serum protein standards was made with both 5 and 10% TCA at 37°C and the absorbance read at 1.5 min reaction time. When 5% TCA was used, a reasonable equivalence between the Albumin and serum protein standards was obtained (Fig. 6). When 10% TCA was used however, albumin gave higher absorbance values than did serum protein.
ABSORBANCE
%oNM
1
1.07
.-
.A’-‘-’
.H
\ SERUM
./
97.5
MdDL. ‘1.
3.06 /
.-.-.
/
.I.
ALBUMIN .’
100
/
.-.
1.
MG/DL.
/
./
TIME
(
MINUTES)
Fig. 3. Kinetics of protein precipitation for human albumin and serum protein at 30°C.
236
Fig. 4. Kinetics of protein precipitation for human albumin and serum protein at 37’C.
Correlation with the existing manual method The manual method in use was that proposed by Meulemans [6] for the estimation of total protein in spinal fluid, modified by using 0.5 ml of urine in place of 1.0 ml of spinal fluid. 100 specimens were measured for protein by both methods and the correlation shown graphically in Fig. 7. The correlation coefficient was 0.9947 with a slope of 0.9485. Two specimens not included in Fig. 7, from a patient with myeloma, gave values of 9.76 and 6.72 with the manual method and 9.10 and 6.55 per litre with the automated method. Precision The results of precision
TABLE
and recovery
experiments
are shown in Table I.
I
PRECISION
AND
RECOVERY
EXPERIMENTS
Within batch
Batch to batch
50 mg/dI standard
100 mg/dl standard
50 mg/dI standard
100 mg/dI standard
n Mean S.D. C.V.%
10 51.3 1.8 3.5
10 100.6 2.14 4.2
20 50.6 2.14 4.2
20 100.5 1.26 1.3
Recovery 108%.
obtained
from the addition of 50 mg standard albumin to urine samples ranged from 98.9 to
237
1.10
1.00
0.90
0.80
il.70
Clinica Chimica Acta, 90 (1978) 231-239 0 Elsevier/North-Holland Biomedical Press
CCA 9776
AUTOMATED
MEASUREMENT
OF URINARY
PROTEINS
ANDREW MUIR * and W.J. HENSLEY Department (Received
of Biochemistry,
Royal J’rince Alfred Hospital, Sydney N.S. W. 2050 (Australia)
May 29th, 1978)
Summary We have investigated the automation of the trichloroacetic acid turbidometric method for the estimation of urinary proteins using a standard centrifugal analyser. Conditions of temperature, trichloroacetic acid concentration, sample and reagent volumes have been examined. The method requires 10 ~1 of urine sample and results have been shown to be linear to 2.0 g per litre with undiluted specimens; higher levels of protein require dilution of the urine samples to obtain greater accuracy. The coefficient of variation at a level of 0.5 g per litre is 3.5% and at 1.0 g per litre is 1.3%; recoveries ranged from 98.9 to 108%. The estimation of urinary protein in 29 samples may be carried out in 5 min.
The measurement of urine protein is complicated by several factors. Urinary protein estimations are regarded by most workers in routine laboratories as one of the least interesting of the analytical techniques carried out and for this reason accuracy and precision may suffer. Variability of pigments present due to hydration and dehydration affects the basic colour of the urine; also, the presence of interfering drugs necessitates the use of individual blank estimations. Blank estimations are usually made by simple dilution of the urine with distilled water in place of reagent, when in fact the reagent may initially, or during the course of the reaction, react with the interfering substance. For this reason it is essential to carry out blank reactions in the presence of reagent where possible. The wide range of protein concentrations found in urine specimens, both normal and pathological (0 to 6 g per litre), has posed problems; some methods e.g. turbidometric ones, have been shown to be sensitive in the range 0 to 1 g * Correspondence should be addressed to: Dr. A. Muir. Department of Biochemistry. Hospital, Missenden Road, Camperdown. N.S.W. 2060. Australia.
Royal Prince Alfred
232
per litre, whilst other methods e.g. using the Biuret reaction, are less sensitive in this range but more satisfactory in the higher protein concentration range. This difference in sensitivity by methods is reflected in the normal ranges quoted in the literature. Turbidometric measurements have been shown to be more sensitive in that range which divides normal from abnormal proteinuria; we have therefore examined the possibility of an automated trichloroacetic acid (TCA) turbidometric method with a view to overcoming some of the problems associated with the measurement of urinary protein. Reference
standard
Trichloroacetic acid (TCA) was introduced in 1921 by Mestrezat [l] for the determination of protein in spinal fluid. Since then TCA has gained popularity as a turbidometric reagent because the discrepancies between the turbidities produced by albumin and globulins are less than with sulphosalicyclic acid. Henry et al. [2] state that in their hands gamma globulin produced 20% higher turbidity than albumin and recommended the use of dilute human serum for a reference standard. Rice [3] and Rice and Loftis [4] used recrystallised bovine albumin as a source of standard in the TCA turbidometric method. Pure human albumin is readily available and may be weighed accurately and thus used as a primary standard. It would therefore be a satisfactory standard provided conditions can be controlled such that albumin and globulins give the same turbidity per weight of protein. Mixing of reactants
With all turbidometric and precipitation methods, the order of addition of reactants, the concentration of reactants and the method and degree of mixing, control the particle size of the precipitate formed. The method of mixing is also responsible for the quantity and size of entrapped air bubbles which may be in the final suspension. The rate of formation and flocculation of the precipitate is variable with both type and concentration of the protein. The manual addition of reactants, with inevitable variations in timing and in degree of mixing poses a problem with turbidometric methods. Temperature
It has been reported that turbidity varies with temperature when using TCA; [3,5] between 20°C and 25°C variation of 25% can occur. At a temperature greater than 25°C greater turbidity is obtained with albumin than with globulins [ 51. Temperature control is therefore essential. Time
The time interval between the mixing of reactants absorbance has been found to be important [3].
and the measurement
of
Materials and method Reagents
Trichloroacetic
acid (TCA) from Ajax Chemicals
(Sydney),
analytical
grade,
233
used. In the initial experiments concentrations of 3, 5,10, and 20% (w/v) were used. Immediately before use the reagent was filtered through a Millipore “Millex” 0.22 pm membrane.
was
Standard protein Pure human albumin from Behring (Hoechst, Australia) was used as a primary standard. Diluted Technicon S.M.A.C. reference serum was used for comparison. The instrument The instrument trifugal analyser .
chosen
for automation
was a Centrifichem
l.lz
1.10
LOS
1.06
0.04
L%T:IN
CCh’lCENTR4TION
Fig. 1. Effect of TCA concentration
( ffih.
)
(in 5%)on the standard albumin calibration graph.
300 series cen-
234
Experimental
results
Wavelength Preliminary analysis showed that absorbances read at a wavelength of 340 nm were 2 to 2.5 higher than absorbances read at 405 nm which is the wavelength usually used for turbidometric analysis. All readings were therefore made at 340 nm. Concentration of trichloroace tic acid Fig. 1 shows the effect of TCA concentration calibration graph, absorbance of protein turbidity concentration of TCA. Effect of temperature We examined the kinetics
BSORBANCE
of albumin
on the standard albumin increasing with increasing
and serum protein
precipitation
%ONM.
i
0.06
.-•
.A”_ SERUM
--.
97.5
MG/DLa
./
_0,05
.
./
0.04
Q-03
.-•-. -.
./-’
OS02
ALBUMIN
I.00
MG/DL.
\.
./
Fig. 2. Kinetics
of protein
TIME (MINUTES ) precipitation for human albumin
and semm protein
at 25’C.
at 25,
235
30, and 37°C with time. The results are shown in Figs. 2, 3 and 4. It can be seen that at 25 and 30°C serum protein yields higher absorbance for the same protein concentration than does albumin over a lo-min period. At 37°C however (Fig. 4), there is a crossover point between 1 and 2 min where both yield the same absorbance value. Further experiments at varying protein concentrations have confirmed that albumin and globulins yield the same turbidity with 5% TCA at 37°C when the absorbance is measured between 1 and 2 min after mixing of the reactants. An example of some of these experiments is shown in Fig. 5. This kinetic pattern is similar to the reactions we have observed with immuniprecipitation reactions. A comparison between albumin and serum protein standards was made with both 5 and 10% TCA at 37°C and the absorbance read at 1.5 min reaction time. When 5% TCA was used, a reasonable equivalence between the Albumin and serum protein standards was obtained (Fig. 6). When 10% TCA was used however, albumin gave higher absorbance values than did serum protein.
ABSORBANCE
%oNM
1
1.07
.-
.A’-‘-’
.H
\ SERUM
./
97.5
MdDL. ‘1.
3.06 /
.-.-.
/
.I.
ALBUMIN .’
100
/
.-.
1.
MG/DL.
/
./
TIME
(
MINUTES)
Fig. 3. Kinetics of protein precipitation for human albumin and serum protein at 30°C.
236
Fig. 4. Kinetics of protein precipitation for human albumin and serum protein at 37’C.
Correlation with the existing manual method The manual method in use was that proposed by Meulemans [6] for the estimation of total protein in spinal fluid, modified by using 0.5 ml of urine in place of 1.0 ml of spinal fluid. 100 specimens were measured for protein by both methods and the correlation shown graphically in Fig. 7. The correlation coefficient was 0.9947 with a slope of 0.9485. Two specimens not included in Fig. 7, from a patient with myeloma, gave values of 9.76 and 6.72 with the manual method and 9.10 and 6.55 per litre with the automated method. Precision The results of precision
TABLE
and recovery
experiments
are shown in Table I.
I
PRECISION
AND
RECOVERY
EXPERIMENTS
Within batch
Batch to batch
50 mg/dI standard
100 mg/dl standard
50 mg/dI standard
100 mg/dI standard
n Mean S.D. C.V.%
10 51.3 1.8 3.5
10 100.6 2.14 4.2
20 50.6 2.14 4.2
20 100.5 1.26 1.3
Recovery 108%.
obtained
from the addition of 50 mg standard albumin to urine samples ranged from 98.9 to
237
1.10
1.00
0.90
0.80
il.70
