dIN.

CHEM. 21/12,

1754-1760

(1975)

Evaluation of Trinders Glucose Oxidase Method for Measuring Glucose in Serum and Urine
John A. Lott and Kathie Turner

Trinders method for glucose has nearly all the attributes of an ideal automated colorimetric glucose oxidase procedure. The chemicals used in the color reaction with peroxidase are readily available, the solutions are stable and can be prepared by the user, the method is highly specific and largely free of interferences, the sensitivity can. be adjusted by the user to cover a wide range of glucose concentrations, and the reagents are not hazardous. We found very good agreement between results by this method and by the hexokinase and Beckman Glucose Analyzer methods. The method has been modified and adapted to the AutoAnalyzer I and SMA 6/60 (Technicon) with manifolds that give very little interaction between specimens. A study of the method by the simplex technique revealed that the glucose oxidase activity in the reagent is the most critical variable.
AddItIonal Keyphrases: continuous-flow analysis #{149}glucose and reagent preservatives #{149} intermethod comparison #{149} optimum analytical conditions #{149}normal (reference) values #{149} sample stability

Peroxidase, Type II, No. P-8250, Sigma. Note that this peroxidase is as satisfactory as the much more expensive Type VI, No. P-8375 from Sigma. 4-Aminoantipyrine, No. A-4382, Sigma (also known as 4-aminophenazone). Peroxidase/buffer reagent, pH 7.0. Dissolve 8.5 g of anhydrous reagent-grade Na2HPO4 and 5.3 g of anhydrous reagent-grade KH2PO4 in about 800 ml of distilled water, and adjust the pH to 7.0 0.1 with 1 mol/liter HC1 or NaOH if needed. Add 4 mg of peroxidase and 300 mg of 4-aminoantipyrine and dissolve, and dilute to 1 liter with distilled water. The final reagent contains 0.1 mol of phosphate per liter, and is

stable

for as long

as four

weeks

when

stored

in

amber-colored glass bottles at 4 #{176}C. Glucose oxidase/peroxidase reagent. Add 12 ml of stock glucose oxidase (or 12 000 U) to 1 liter of the

peroxidase/buffer reagent. The complete reagent is stable for 1 week at 4 #{176}C. Phenol solution, 1.5 g/liter. Dissolve 1.5 g of phenol, analytical-reagent grade, in enough distilled

The 1.11.1.7) 1969 is glucose lutions

glucose oxidase/peroxidase (EC 1.1.3.4/EC method for glucose described by Trinder in very attractive (1). The method is specific for and the reagents are all readily available. Soof the reagents are stable. The Trinder re-

agents are less of an occupational hazard than reagents used in other methods-o-toluidine, o-dianisidine, or N,N-dimethylaniline, which are all quite toxic. Here, we describe automated methods for serum and urinary glucose with use of the AutoAnalyzer I or SMA 6/60 (Technicon Instruments Corp., Tarrytown, N. V. 10591) and of reagents that the user can prepare. We also describe optimization of the method with the simplex technique, method interferences, normal values, and other data. Materials and Methods Reagents and Standards
Stock glucose oxidase,

water to make 1 liter of solution. Stable for at least six months when stored in amber-colored glass bottles at room temperature. Saline, 9 g/liter. Dissolve 9 g of sodium chloride in enough distilled water to make 1 liter of solution.
Add 1 ml of Tween 20 surfactant liter just before use. Stock glucose standard, 10 g/liter. (Technicon) Dissolve per 10.000

g of primary-standard grade dextrose in enough distilled water that is saturated with benzoic acid (about 3 g/liter) to make 1 liter of solution. Working glucose standards. Prepare dilutions of the stock glucose standard with a saturated aqueous benzoic acid solution to give standards containing 250, 500, 1000, 1500, 2000, 2500, 3000, and 4000 mg of
glucose per liter,

Procedure The flow diagrams Type V, No. Co., St. Louis, Mo. 63178.
106 U/liter, 6/60 are shown

for the AutoAnalyzer

1 and 2. Freshly

and SMA
separated

in Figures

G-6500, Sigma Chemical

Division of Clinical Chemistry, Department of Pathology, Ohio State University, 410 W. 10th Ave., Columbus, Ohio 43210. Presented in part at the Ninth International Congress on Clinical Chemistry, Toronto, Ont., 1975 (Clin. Chem. 21, 978 (1975), abstract). Received July 21, 1975; accepted Aug. 18, 1975.

serum or urine can be analyzed directly. In the case of the AutoAnalyzer I, the standards are analyzed in the order listed above and a calibration curve is prepared on semilog paper. The glucose in patients samples and in controls is estimated from the curve. The standards should be run at the beginning of each

1754

CLINICAL CHEMISTRY, Vol. 21, No. 12, 1975

Tube Flow,
id.in.
rt/min I.

Tube Flow,
id.in. mI/mm

__________8-turn

6 Dialyzer

Samole
Saline/Twn

.020 0l

0.16 10

I Air
Air Gluc.Ox./Perox.

.035 0.42
I.0
0.42 0.42 0.32 .035
.035

[
2,9-1
turn

Sdine/Tween .051

LU1709
----

coils

Both Color-

37#{149}
Recorder

Phenol

.030

Flowcell Retum.05l Sampler IV

1.0

________ PhosingCoil
meter 505 nm ______

Fig. 2. Flow diagram for the SMA 6/60

Fig. 1. Flow diagram for the AutoAnalyzer I

Table 1. Analytical Recovery of Glucose from Bovine Albumin Solutions

Added
mg/liter

agent. In the case of either the AutoAnalyzer I or the SMA 6/60, replace the manifold tubing after no more
than 140 h of running time.

Founda

Recovery,

Resufts

500

1000

510 1000
2000

2000 3000

3040
4020

4000

102 100 100 101 101

Analytical Variables
Bovine albumin (No. 905-10, Sigma) contained no glucose detectable by the AutoAnalyzer I method described here or by the Calbiochem hexokinase procedure (No. 869204; Calbiochem, San Diego, Calif. 92112). Solutions were prepared to contain, per liter, 70 g of albumin and 500, 1000, 2000, 3000, and 4000 mg of glucose, and they were then assayed with the AutoAnalyzer I. Because the analytical recoveries (Table 1) were all within about 2% of the expected values, we concluded that aqueous standards can be used to calibrate the AutoAnalyzer I and that protein does not interfere in
Analytical recovery. the analysis Precision. for glucose.

SAIl results are means of three determinations.

third tray of 40 samples. At least two controls be analyzed on each tray.

should

The glucose concentration of the serum-based calibrating material used to set the SMA 6/60 should be

established by analysis with the method proposed here. The insert or label values must not be used
uncritically, because they may have been established by a less-specific method. Standards and at least two controls should be analyzed on every tray of 40 sam-

ples on the SMA 6/60. At the end of the day, 1 mol/liter NaOH is pumped through all lines, including the dialyzer, for 15-30 mm, followed by a 30-mm wash with distilled water.
This washing effectively prevents shifting baselines,

The method described has been in routine use here since October 1974. Between November 1, 1974 and March 31, 1975, we did about 44 000 serum glucose determinations. Blind controls (2)
from the same lot numbers were randomly ed between patients samples during that distributtime, and

drift, clogged tubing, etc. A sodium hypochlorite solution must not be used, because it is difficult to wash out completely, and the hypochlorite reacts to form a color with the glucose oxidase/peroxidase re-

the results are listed in Table 2. An equal number of blind controls were analyzed with the AutoAnalyzer I and SMA 6/60. We think that the precision of the
method over this time span is satisfactory. Comparison studies. We compared our results for this method to those obtained by the Calbiochem

Table 2. Summary of Quality-Control

Versatola
Month n
Mean

Versatol
CV n Mean

Data for Glucose in Samples Analyzed as Blinds

A0 Versatol A-AItarnate Mean

CV %

CV
%

Pooled serum Mean

CV
%

mg/liter

mg/liter

mg/liter

mg/liter

Nov. 74 Dec. 74 Jan. 75 Feb. 75

20 21 22 20

820 820 830 820

2.4 3.0 3.0 4.3

20 21 22 20

2020 1980 1990 1980

2,9 2.1 1.7 1.6

30 31 31 28

2990 2980 2990 2980

2.0 1.8 2,8 2.2

30 31 31 28

840

840
840 810

3.1 2.3 3.1 2.7

Mar. 75
a General

21

820
Morris Plains,

2.7
N.J. 07950.

21
Lot

1980
numbers, left

2.6
to right, were

31
2406103.

2940
2262043,

1.3

31

810

4.1

D iagnostics,

and 1176112.

CUNICAL CHEMISTRY,

Vol. 21, No. 12, 1975

1755

Table 3. Results by Three Methods for Serum Glucose Compared (Mean of Duplicate Results)
AutoMaterial Analyzer I

commonly
The

used oral hypoglycemic

of metabolites

concentrations

normal range and generally

agents interfered. are far above the exceed those seen for cre-

SMA 6/60 Hexokinase

mg/liter

glucose,

Versatola Versatol
Calibratea VersatoI Automated

760 1840
2960

780 1790
2820

770 1840
2790

Versatol A Alternate Lo

atinine, urea, and uric acid even in patients with severe azotemia. For all of these, we observed no interference. Uric acid was examined in more detail. Pooled serum was diluted with saline or a stock uric acid solution to give pools with 100, 250, and 500 mg of uric

1830 770
2350

1840 800
2310

1810 840
2300

Serum Referenceb Scale lb

860
3710

870
3550

870
3590

Scale Ilb Pool IC (lipemic)

Pool

IIC

Pool IlIC Pool lV Pool VC (lipemic) Pool VIC Pool VIIC

O

470 560 850 940 1100

1410
3000

470 550 850 940 1050

1380
2860

320 540 850

910

840
1380
2930

General Diagnostics. bTechnicon C Freshly pooled human

serum.

acid per liter. The addition of uric acid did not change the observed glucose concentration of 630 mg/liter as compared to the same pool diluted with saline (Table 4). Likewise, when uric acid was added to three other poois to give a concentration of 200 mg of uric acid/liter, the glucose concentration of 510, 970, and 1940 mg/liter were the same as was observed when saline was added to the poois. That the sugars listed in Table 4 do not interfere reflects the specificity of the method, maltose being an exception. The interference from maltose was due to the presence of maltase in the glucose oxidase. Hemoglobin did not interfere, as was also reported by Gochman et al. (4). Ascorbic acid produced dramatic decreases in the observed glucose value, in contrast to the findings of others (5) who observed no effect on results by Trinders method (1) of ascorbic acid, 1000
mg/liter.

hexokinase procedure for various lyophilized control sera and pooled fresh sera. Two serum pools were made up from lipemic samples. Agreement was good (Table 3) except for the lipemic pools (pools I and V), for which the hexokinase procedure gave somewhat lower results. We also assayed 54 freshly collected patients sera
containing 540 to 4760 mg of glucose per liter, the Glucose Analyzer (Beckman Instruments, with Inc.,

In vivo concentrations

of ascorbic

acid are too low

to significantly interfere. In a study by Schrauzer and Rhead, the maximum ascorbic acid concentration in plasma or erythrocytes never exceeded 27.5 6.5

(SD) and 15.1 3.6 mg/liter, respectively, in 17 volunteers who had taken 2 g of the drug daily for nine
days (6). A large fraction of ingested ascorbic acid is

Fullerton, Calif. 92634) and with the SMA 6/60. Results obtained with the two instruments agreed well. The means and standard deviations for the SMA and Beckman were 1460 930 and 1470 910 mg of glucose per liter, respectively, the correlation coefficient was 0.9994, and the slope and intercept were: Beckman = 0.972 (SMA) + 4.5 (3). To be certain that we had no bias between the AutoAnalyzer I and SMA 6/60, we assayed 73 fresh patients sera with both instruments. The range of values was 350 to 7890 mg/ liter, the means and SD were 1260 970 (AutoAnalyzer I), and 1260 960 mg/liter (SMA 6/60). The correlation coefficient was 0.9993, and the slope and intercept were: SMA = 0.98.4(AutoAnalyzer I) + 2.51.
Interferences. Various anticoagulants, drugs, metabolites, sugars and other compounds were tested for potential interferences with the method (Table 4). For the first group the same glucose concentration was observed when either saline or a solution of the compound was added to pooled fresh serum. The concentration of anticoagulants that we tested is much higher than is commonly used, but none interfered. The serum drug concentrations that we studied are much higher than would be expected after a therapeutic dose. It is significant that none of the
1758 CLINICAL CHEMISTRY, Vol. 21, No. 12, 1975

excreted unchanged in the urine (7), hence a potential interference exists in cases of renal failure. Gentisic acid, a metabolite of salicylic acid, interferes with the method. However, only a small fraction [of salicylic acid] is converted to this metabolite (8), so that, in vivo, gentisic acid is probably not a
source of interference.

Reduced glutathione is present in whole blood at a concentration of 280-340 mg/liter (9). When we added reduced glutathione to serum, we observed
falsely low glucose values with the method. But when

whole blood was intentionally hemolyzed to produce plasma with 20 g of hemoglobin per liter, we observed no interference. The intentional hemolysis may have destroyed the reduced glutathione so it remains a potential interferant in hemolyzed blood. Levodopa (L-DOPA) in the concentrations indicated in Table 4 seriously interferes. These concentrations of L-DOPA are much higher than have been observed in vivo. Muenter and Tyce (10) observed peak

concentrations of 0.40 to 7.3 mg/liter of plasma after 0.25- to 2-g doses of L-DOPA, in a study involving 26 patients. In another study with 15 patients (11), peak
concentrations of 1.0 to 4.0 mg/liter of plasma were

observed tabolites

after doses of 0.25 to 1.9 g. Whether the meof L-DOPA interfere is an open question; it

Table 4. Interferences Study

Substance Concns in pooled serum, mg/liter or as stated#{176} Glucose concn in poo1, mg/liter Substance Concns in pooled serum, mg/liter or as stated0 Glucose concn in pool, mg/liter

Compounds that do not change the obser,.ed glucose Anticoagulants

Heparin Sodium citrate 75000and 150000 1080 units/liter 5000, 10 000, and Ascorbic acid

Compounds that interferea

0 50 100 200

1300 1220
1110

1020 1070
Gentisic
acid

Sodium fluoride Sodium oxalate Drugs Acetohexamideb ChlorpropamideC Phenformind Tolbutamidee Tolazamidee

20 000 10 000, 20 000,

40000 2000, 4000,

500 0 100 200 400 800 0 250 500

790 0
1030

8000

990
1010

830

760
630 570

Sodium
salicylate Metabo/ites Bilirubin Creatinine Hemoglobin standard Urea Uric acid Uric acid Sugars Fructose Galactose Lactose Mannose Sucrose
d-Xylose

50, 100, 200 50, 100, 200 20, 40, 80 50, 100, 200, 400 50, 100, 200 100, 200, 500, 1000

1080 1010 1100

1080

Glutathione (reduced)

1240
1100 1000 840 1110 820 620 530
1070

1070 L4)QPAI

woo
0 100
200

50, 100, 200 500, 1000, 2000 5000, 10 000, 20 000

500, 1000, 2000

1160
1060 440 1090
Maltose

400 0 1000

100, 250, 500 200

1000 2000 5000

630
510, 970, 1940
1030 1020

2000
5000

1230 1380
1830

1000, 2000, 5000 1000, 2000, 5000 1000,2000, 5000 1000, 2000, 5000 1000, 2000, 5000

1010
1030

a Either saline or the indicated compound was added to a serum pool. For any compound, the actual concentration of glucose in the particular pool was constant. For the uric acid study, see text. bEli Lilly and Co., Indianapolis, md. 46206.

1070 1080

Pfizer Inc., New York, N.Y. 10017. dGeigy Pharmaceuticals, Ardsley, N.Y. 10502. e The Upjohn Co., Kalamazoo, Mich. 49001.
C

Eaton Laboratories,

Norwich,

N.Y. 13815.

is metabolized

by at least

three

major

pathways

(12).

The major components excreted in urine are the unchanged drug, dopamine, and homovanillic acid (13). Serum Normal Values Serum from 72 presumed healthy adult volunteers who had fasted for at least 12 h was examined for glucose by this method (Table 5). Our median and mean were both 840 mg of glucose per liter and the histogram was reasonably gaussian. For comparison, we have also listed normal values for several methods in which glucose oxidase but different chromophores were used. In his review, Free lists normal (reference)
values for glucose in serum as measured by glucose

2500, 5000, and 10000 mg/liter. The analytical recovery was 98-104% (average, 100%). The urinary glucose estimations were done with the AutoAnalyzer I. They can be done with the SMA 6/60, but then aqueous glucose solutions must be used to calibrate the instrument. As much as 1.6 g of uric acid in per liter of urine, or boric acid at concentrations of 0.4, 0.8, and 1.6 g/liter,
or a saturated aqueous of glucose solution in urine. of thymol do not in-

terfere

with the method.

Four patients urine samples were chosen for study on the basis that they were free of glucose and they contained more than 100 000 organisms per milliliter. Glucose was added
Stability to each urine to prepare samples of each to contain

oxidase methods

published in Urine

before 1962 (14).

Glucose Estimation

about

Trinders method is also suitable for quantitating glucose in urine. We added glucose to four different glucose-free urines to give glucose concentrations of

5000 and 10000 mg of glucose per liter. One gram of boric acid, or about 200 mg of thymol, or nothing was added to three 100-ml aliquots of each sample. These 24 samples were analyzed for glucose with the AutoAnalyzer I immediately after preparaCUNICAL CHEMSTRY, Vol.

21, No. 12,

1975

1757

Table 5. Normal (Fasting) Values for Serum

ance was measured

trophotometer
Ref.

at 505 nm in a Model
Instrument

2000 spec-

Glucose, by Various Glucose Oxidase Methods

Mean Mean mg/liter

(Gilford

Laboratories,

2S0
No. Comment

830
880 920

750-1080 700-1070 720-1160

32 32 151,age

AutoAnalyzer AutoAnalyzer M BTH-D MA0 SMA12/60,

I, II,

Inc., Oberlin, Ohio 44074) vs. a water blank. The optimum sought was the maximum color intensity. The concentrations of each of the reagents used in
the final reaction volume (9.6 ml) is given in Table 6

MBTH-D MAO 18 22 2 2

along with the progress of the simplex. The simplex has four dimensions and five vertices and was therefore treated by Longs calculation technique (15). The starting concentrations (vertex 1) were deliberately set far from the presumed optimum, and a step size of 80% of the starting concentrations was used. The simplex study was stopped at 18 experiments, even though we had not found the optimum. We did find that glucose oxidase activity is a primary deter-

20-49
1110 840-1280 185,age over 50

MBTH-DMA0
(fasting?)

SMA12/60, MBTH-DMA0
(fasting?)

890

670-1120

58

AutoAnalyzer
neocuproine

II,

23

glucose oxidase/

880
920

660-1110
700-1150

58
58 72b
to peroxidase

Glucose oxidase/
dianisidine Glucoseoxidase/ M BTH-D MA0 This method,
AutoAnalyzer I described in ref. 18.

23 23

840

670-1010
linked

minant of the final color intensity. Glucose oxidase activity plotted vs. absorbance gives a straight line with some scatter of the points (correlation coefficient, r = 0.96). The final color intensity is less sensitive to changes in the concentration of 4-aminoantipyrine (r = 0.43) and still less sensitive to changes in the peroxidase activity (r = 0.27) or the concentration of phenol (r = 0.14).

The concentrations
entering

of the reagents
(see Figures

in the solution
1 and 2) of the

a Chromophore b 61 women,

the 37 #{176}C bath

11 men; age range 20-57

yr (mean, 29).

tion and again after 1, 2, 3, 7, and 14 days of storage at room temperature. Boric acid is somewhat superior to thymol as a preservative, although neither is ideal. The maximum loss in glucose after 1, 2, 3, 7, and 14 days was 8, 14, 26, 27, and 34%, respectively, for boric acid and 14, 28, 28, 36, and 54%, respectively, for thymol. Unpreserved samples lost as much as 40% of their glucose in one day. From these limited data, we concluded that analysis for glucose in urine is invalid after the urine has stood for more than 1 day at room temperature, even when thymol or boric acid is present. Simplex Optimization of Analytical Conditions

I and SMA 6/60 are also listed in Table 6. The peroxidase activity and the concentrations of 4-aminoantipyrine and phenol are somewhat in excess of what is needed. The solutions with concentrations described at vertices 13, 14, and 15 (Table 6) give nearly the same absorbancies and were obtained by using about the same amount of glucose oxidase. A large variation in the peroxidase activity (vertex 13
vs. 15), the concentration of 4-aminoantipyrine (vertex 13 vs. 15) or of phenol (vertex 13 vs. 14) had practically no effect on the absorbance,

AutoAnalyzer

In another
was varied,

experiment,
and the

the glucose oxidase activity

of the other re-

concentrations

agents
and

were the same as described under Materials Methods. We found that the sensitivity of the

There

are many variables

in this method:

pH, type
1.2
Sc

of buffer, incubation temperature, concentrations of the reagents and sample in the final reaction mixture, sample-to-wash ratio, etc. The simplex method for

1.0 0.8
0 .0
.

0 ;

optimizing

analytical

conditions

has been described

.! =
0 0
.

,/mg/liter

elsewhere (15-17). We chose to use a 0.1 mol/liter phosphate buffer because it has been used successfully by others (18), but our decision was really arbitrary. A pH of 7.0 was chosen because it is close to both the pK2 of phosphoric acid (7.13) and the reported (19) optimal range for glucose oxidase (pH 4.0 to 6.5). The variables we investigated by using the simplex method were the glucose oxidase and peroxidase activity and the concentrations of 4-aminoantipyrine and phenol in the final reaction mixture. One milliliter each of solutions of glucose oxidase, peroxidase, and 4-aminoantipyrine were mixed and 6.5 ml of a phenol solution was added. The mixture was incubated at 37 #{176}C for 10 mm, and the absorb1758 CLINICAL CHEMISTRY, Vol. 21, No. 12. 75

/1
,

0
.0

0.6
0 0000

0.4 0.2

N 0

.//
,-

/
.-,

2000
$000

,/,, :-

0
Log Glucose

5
U/liter

Oxidose Activity,

Fig. 3. Absorbance of the 1000, 2000, and 4000 mg per liter glucose standards after reaction with glucose oxidase/peroxidase reagent containing increasing amounts of glucose oxidase

Table 6. Data for Simplex Optimization of Glucose Oxidase Concentrations in Final Reaction Mixture
Glucose oxidase
Peroxidase 4-aminoantipyrine mg/liter Phenol

Reaction.
Vertices retained from previous simplex
-

Absorbance
observed

Vertex no. 1 2 3 4

U/liter 417 750 583 583

0.596 0.596

26.0 26.0 26.0 43.0

169.2 169.2 169.2 169.2

0.092 0.138 0.113 0.104

1.009
0.734

5 6 7 8
9 10

583 833 792 896

1052 973

0.734 0.941 0.906 0.992

0.709 0.713

30.2 36.6 16.4 22.3

24.7 38.4

276.0 223.0 250.0 129.7

216.8 119.4

0.106 0.139 0.129 0.138

0.148 0.151

2, 3,4, 5 2, 3, 5, 6 2, 3, 6, 7
2, 6, 7, 8 2,6,8,9
-

20
6 11 12 13

750
833 1125 1097 1291

0.596
0.941 1.082 0.731 0.677

26.0
6.6 35.0 45.1 35.0

169.2
223.0 175.3 237.6 151.6

0.130
0.145 0.179 0.187 0.179

6a,8,9,10 6a, 9, 10, 11

9,10,11,12

14
90

1308
1052 1358 3200
1a

0.887
0.709 0.980 1.07 0.966
by proposed method.

31.5
24.7 48.6 80 72.4

271,3
216.8 201.1 267 276

0.192
0.155 0.185

9, 11, 12, 13
-

15

11, 12, 13, 14

AutoAnalyzer

SMA 6/600
0

2900
in final reaction mixtures

Concentrations

method could be altered by simply changing the glucose oxidase activity, as is illustrated in Figure 3. At a glucose oxidase activity of 48000 U/liter, the curve begins to flatten. For the 2000 mg/liter glucose standard, the absorbance increases by 0.142 when the glucose oxidase activity is doubled (from 12 000 to 24 000 U/liter). When the glucose oxidase activity is doubled again (from 24 000 to 48 000 U/liter), the absorbance increases by only 0.067 for the same standard. We chose to use 12000 U of glucose oxidase per liter of reagent because this gave us a linear curve to 4000 mg of glucose per liter and good sensitivity in the 0-1500 mg/liter region with a 0.2 ml serum sample.
Limits of the Method

Some Other Factors Affecting the Analysis

Choice of peroxidase. The cheaper peroxidase (No. 8250, Sigma, 5/mg), gave the same linearity and sensitivity at a concentration of 4 mg/liter, as did the more expensive peroxidase (No. 8375, Sigma, 55/ mg). At 2 mg/liter, there was also no observable difference between the two, but at 1 mg/liter, the reagent containing the cheaper peroxidase gave a nonlinear calibration curve. Choice of wetting agent. Polyoxyethylene ether of

lauryl

alcohol (Brij-35; Technicon) should not be used. With this wetting agent at a concentration of 0.5 mi/liter of saline, the results for fresh sera with the AutoAnalyzer I were all about 5% higher than the results from the SMA 6/60. When we used polyoxyethylene sorbitan monooleate (Tween 20, 1 ml/liter
of saline), this bias disappeared. Interaction and manifold. In the original method described by Trinder (1), only buffer, saline, and 4aminoantipyrine were pumped through the dialyzer.

Glucose concentration and absorbance are linearly related to at least 4000 mg/liter on the AutoAnalyzer I and to 5000 mg/liter of the SMA 6/60. The AutoAnalyzer I method can be altered to permit analysis of samples containing as much as 5000 mg of glucose per liter by reducing the sample line one size, to 0.10 ml/min, but some precision is lost in the low concentration range results. With either instrument the method is suitable for analysis of samples with glucose concentrations of <500 mg/liter. Pooled serum was diluted with saline to give glucose concentrations of 100, 200, 300, 400, and 500 mg/liter. For each pool and for both instruments, the value found on analysis was within 30 mg/liter of the expected value.

In the modification of Trinders method by Pennock et al. (5), 2 g of phenol per liter is present in both the diluent and the recipient streams in the dialyzer. Figure 4 shows a recorder tracing from the AutoAnaiyzer I. The percent interaction (20) between the 1000 and 4000 mg/liter standards (center of Figure 4) with our manifold was 2.1%. By contrast, we
found 5.8% interaction when we used the manifold of

Pennock et al. (5) with our reagents. Obviously, phenol should not be allowed to go through the dialyzer.
On the AutoAnalyzer I, the glucose oxidase/peroxi1759

CUNICAL CHEMISTRY,

Vol. 21, No. 12, 1975

free of sediment
I
I

after six months of storage at room temperature in amber-colored glass bottles, and could not be distinguished from a freshly prepared phenol solution when used in conjunction with the other reagents for glucose.
We thank Drs. H.-D. Gruemer and G. F. Grannis for helpful comments on the manuscript, and Rita Beal, B. W. Durham, Joan Mercier, Kathy Rieger, and Tim Walters for technical assistance.

$ $
I

$ S
$

r
iI
Stondords Interoction

References

II
SteodVStole

/
Low Control

1. Trinder, idase with

P., Determination

an alternative

of glucose in blood using glucose oxoxygen acceptor. Ann. Clin Biochem. 6,

test

ltgh

control

Fig. 4. Recorder tracing from the AutoAnalyzer I

Steady-State tracing obtained by constant sampling of the 4000 mg/liter standard for 5 mu; Standards are the eight aqueous glucose standards dosctlbed In the text; Interaction test, tracings for the 1000, 4000, 1000, and 1000 mg/liter standards; b ConVol and Low ConVol replIcate analyses of two commercial control set-a

24 (1969). 2. Allen, J. R., Earp, R., Farrell, E. C., Jr., and Gruemer, H.-D., Analytical bias in a quality control scheme. Clin. Chem. 15, 1039 (1969). 3. Moroney, M. J., Facts from Figures, Penguin Books, Baltimore, Md., 1951, p291. 4. Gochman, N., Ryan, W. T., Sterling, It E., and Widdowson, G. M., Interlaboratory comparison of enzymatic methods for serum glucose determination. Clin. Chem. 21,359 (1975). 5. Pennock, C. A., Murphy, D., Sellers, J., and Longdon, K. J., A
comparison of AutoAnalyser in blood. Clin. Chim. Acta

48, 193

methods for the estimation (1973).

of glucose

dase reagent should be added through the middle connection of the GO cactus (Figure 1), to minimize

interaction.
Reagent stability. We examined several different preservatives for the peroxidase/buffer reagent listed earlier. This reagent developed a fine sediment after one to two weeks at room temperature, and after

three to four weeks, sensitivity declined. The reagent is stable for at least four weeks at 4 #{176}C. We have found the stability of phosphate buffers to be quite capricious. Some appear to be stable for months at room temperature, while mold is growing in other lots soon after preparation. Apparently the stability of the buffer is determined by what spores, dust, etc., fall into the solution at the time of preparation. Trig (hydroxymethyl)aminomethane buffers also showed mold growth. We investigated three preservatives in some detail. Sodium azide (4 g/liter) was unsatisfactory; the peroxidase/buffer reagent became yellow after one week, and linearity and sensitivity deteriorated after two weeks. According to Bergmeyer et al. (21), azide inhibits peroxidase. Thimerosal (Merthiolate), 20 mg/
liter, is unsatisfactory. Results for glucose with the SMA 6/60 were lower when thimerosal was present in the final reagent (glucose oxidase/peroxidase reagent) vs. the reagent without thimerosal. This was

6. Schrauzer, G. N., and Rhead, W. J., Ascorbic acid abuse: Effects of long term ingestion of excessive amounts on blood levels and urinary excretion. mt. J. Vitam. Nutr. Res. 43, 201 (1973). 7. Mayersohn M., Ascorbic acid in man. Pharmo-kinetic implications. Eur. J. Pharmacol. 19, 140 (1972). 8. Goodman, L. S., and GiJman, A., The Pharmacologic Basis of Therapeutics, 4th ed., Macmillan, 1970, p 234. 9. Ibbott, F. A., Amino acids and related substances. In Clinical Chemistry: Principles and Technics, 2nd ed., R. J. Henry, D. C. Cannon, and J. W. Winkelman, Eds. Harper and Row, Hagerstown, Md., 1974, p 618. 10. Muenter, M. D., and Tyce, G. M., L-DOPA therapy of Parkinsons disease: Plasma L-DOPA concentration, therapeutic response, and side effects. Mayo Clin. Proc. 46, 231 (1971). 11. Tyce, G. M., Muenter, M. D., and Owen, C. A., Dihydroxyphenylalanine (DOPA) in plasma during DOPA treatment in patients with Parkinsons disease. Mayo Clin. Proc. 45,438, (1970). 12. Weiss, J. L., and Chase, T. N., Levodopa in parkinsonism. Drugs 2, 257 (1971). 13. Abrams, W. B., Coutinho, C. B., Leon, A. S., and Spiegel, H. E., Absorption and metabolism of levodopa. J. Am. Med. Assoc. 218, 1912 (1971). 14. Free, A. H., Enzymatic determination of glucose. Adu. Clin. Chem. 6,84 (1963).
15. Long, D. E., Simplex optimization of the response from chemical systems. Anal. Chim. Acta 46, 193 (1969). 16. Deming, S. N., and Morgan, S. L., Simplex optimization of variables in analytical chemistry. Anal. Chem. 45, 278A (1973).

17. Krause,
optimize

R. D., and Lott,

conditions

J. A., Use of the simplex method

in clinical chemistry.

analytical

to Clin. Chem.

20, 775 (1974).

18. Gochman, N., and Schmitz, J. M., Application of new peroxide

not true of the AutoAnalyzer with and without thimerosal same results.

I; identical sera assayed in the reagent gave the

indicator reaction to the specific, automated determination cose with glucose oxidase. Clin. Chem. 18,943 (1972). 19. Ref. 14, p 70.
20. Thiers, R. E., Cole R. R., and Kirsch, W. J., Kinetic ters of continuous flow analysis. Clin. Chem. 13,451(1967). 21. Bergmeyer, H. -U., et al., Biochemical reagents. Enzymatic Analysis, Section D. H.-U. Bergmeyer, Press, New York, N. Y., 1965, p 991.

of glu-

parame-

Cacodylic acid (dimethylarsinic acid) looked promising as a preservative because the peroxidase/buffer reagent containing only 10 mmol of cacodylate per

In Methods of Ed. Academic

liter was stable for five months at room temperature. Unfortunately, reagent with cacodylate gave consistently higher results on the AutoAnalyzer I with fresh

22. Carey, R. N., Feldbruegge, D., and Westgard, J. 0., Evaluation of the adaptation of the glucose oxidase/peroxidase 3-methyl-2-

patients sera than did the reagent without cacodylate. The 1.5 g/liter phenol solution was colorless and
1780 CLINICAL CHEMISTRY, Vol. 21, No. 12 75

benzothiazolinone hydrazone-N,N-dimethylaniline procedure to the Technicon SMA 12/60, and comparison with other automated methods for glucose. Clin. Chem. 20,595 (1974). 23. Romano, A. T., Automated glucose methods: Evaluation of a glucose oxidase-peroxidase procedure. Clin. Chem. 19, 1152
(1973).