Feb 26, 1975 - those obtained by the hexokinase and the glucose oxi- dase/peroxidase methods. A better agreement was found with the hexokinase method.
CIJN. CHEM. 21
/ 10, 1372-1377 (1975)
Kinetic Determination of Glucose with the GEMSAEC (ENI) Centrifugal Analyzer by the Glucose Dehydrogenase Reaction, and Comparison with Two Commonly Used Procedures Rudolf A. Lutz1’2 and JUrg FlUckiger2
A new glucose dehydrogenase
preparation
has been
used to determine the glucose concentration in serum or plasma with the GEMSAEC(ENI) analyzer. The reac-
tion is sufficiently linear to be suitable for a kinetic determination. A sample volume of 15 jl or less is needed, and 14 determinations can be done simultaneously within 160 s (the time needed for loading the samples and reagents into the distribution disc plus a reaction time of 70 s). The reaction is linear up to 300 mg of glucose per 100 ml, but with special computer software linearity could be extended to 1 g of glucose per 100 ml. Day-to-day and within-day precision have been tested with a number of different sera and standards. Accuracy has been checked by comparing the results with those obtained by the hexokinase and the glucose oxidase/peroxidase methods. A better agreement was found with the hexokinase method. The proposed procedure has the advantage of involving only one enzymatic step and of directly measuring reduced coenzyme formation at 340 nm. Addftlonal Keyphrases: dase/peroxidase
hexokinase
method compared
method and glucose ox!#{149}analytical system
Several enzymatic methods have been proposed for determining glucose in biological fluids, but none was very satisfactory for use with the GEMSAEC centrifugal analyzer. In the case of glucose oxidase, interfering substances have been reported (1-4), including uric acid, a natural constituent of plasma or serum. The determination with the hexokinase reaction on the other hand has the drawback that two enzymatic steps are required, the first of which is nonspecific: hexokinase also phosphorylates fructose and mannose (5). Although the next step is specific, enzymatic impurities, glucosephosphate isomerase (EC 5.3.1.9) or mannosephosphate isomerase (EC 5.3.1.8), may form additional glucose-6-phosphate, which is then included in the determination. Glucose-6-phosphate itself may cause some interference if it is present. Furthermore the reaction is not very suitable for kinetic measurements, and to our knowledge only a two-point kinetic method has been proposed 1 Department of Molecular Biology and Biophysics, Swiss Federal Institute of Technology, CH- 8049 Zurich, Switzerland. Dr. Otto Lutz AG, Graben 35, CH- 8400 Winterthur, Swit-
zerland. Address Received
reprint requests to R.A.L. at the address in footnote February 26, 1975; accepted May 23, 1975.
1372 CLINICALCHEMISTRY,
Vol. 21, No. 10, 1975
2.
to measure the glucose concentration with a commercial GEMSAEC analyzer (6). Because with centrifugal analyzers one is unable to take an absorbance reading at zero time, some authors extrapolate the initial absorbance back by cubic curve fitting (7). Preliminary experiments have shown that this is not possible for the GEMSAEC (ENI) analyzer, as the first reading is only made about 25 to 30 s after the reaction is started. Recently a glucose dehydrogenase has been isolated from Bacillus cereus (8). This enzyme seems very suitable for determination of glucose in body fluids, because only D-xylose (if present in amounts of more than 30 mg/100 ml) and mannose have been reported
to cause some interference among the carbohydrates in question. There is little if any interference by uric acid,
bilirubin,
glutathione, heparin, sodium fluoride, or ascorbic acid when measured The applicability of this enzyme preparation to determination of glucose in body fluids has been tested (8, 9), but to our knowledge no such study has been undertaken with the GEMSAEC (ENI) centrifugal analyzer. Our results are compared with those obtained by a modified hexokinase method and by the glucose-oxidase/peroxidase method.
sodium oxalate, kinetically (8).
Materials and Methods Apparatus The GEMSAEC centrifugal analyzer (ElectroNucleonics Inc., Fairfield, N. J. 07006) connected to a PDP 8/E computer with a core capacity of 8K words equipped with a DEC-Tape TD8-E and a DEC-Writer LC8-E (Digital Equipment Corp., Maynard, Mass. 01754) was used for the glucose dehydrogenase and the hexokinase methods. The glucose oxidase method was used with a continuous-flow system consisting of a distributor SD3, a flow photometer FF/3 (Carlo Erba Scientific Instruments Division, Milan, Italy), and a 25-channel ISMATEC pump MP-25 (Ismatec SA, CH-8031, ZUrich, Switzerland). A Varian strip-chart recorder G2500 (Varian Aerograph, Calif. 94598) was used to trace the output of the flow photometer. The Carlo Erba dialyzer (code no. 08.59.10060) was used. Statistical calculations were done with an HP-65
calculator
(Hewlett-Packard,
Cupertino,
Calif.
95014) and manufacturer.
plotted tories,
with
the “Stat The linear
with a System Inc, Tewksbury,
Pac 1” supplied by the regression analysis was 700 C/702/708 (Wang LaboraPa. 01876).
Table 1. Comparison of the Experimental Conditions and the Instrumental Settings of the GEMSAEC (ENI) Centrifugal Analyzer for the Glc-DH and the Hexokinase Methods GIc-DH
Reagents Glucose dehydrogenase (Glc-DH, -D-glucose: NAD-oxidoreductase, E.C. 1.1.1.47, kindly supplied by E. Merck, 61 Darmstadt 2, Germany), isolated from Bacillus cereus M 1020. The specific activity was 2.78 U/mg and the Km was 3.22 mmol/liter. Nicotinamide adenine dinucleotide (NAD, No. 24542; E. Merck) was used as supplied. Glucose standards were prepared by dissolving dried glucose (Glucosum hydratum; Pharmacop. Helvet. VI) in redistilled water containing 1 mg of sodium azide per milliliter. This solution was allowed to stand for at least 24 h, to obtain the equilibrium of mutarotation. Protein-containing standards were obtained by adding 5 g of bovine serum albumin (BSA) to 100 ml of the water before dissolving the glucose. These solutions are stable for at least a month if stored at 4 #{176}C. Glucose dehydrogenase reagent: Dissolve 86.5 mg of NAD and 6 mg of Glc-DH in 50 ml of tris buffer (200 mmol/liter, pH 7.8). Glucose oxidase/peroxidase reagent: Dissolve the contents of bottle 2 (GOD-Perid test kit 15756; Boehringer Mannheim, Germany) in 1 liter of redistilled water. Hexokinase reagent (test kit 15931; Boehringer): Mix 18 ml of solution 1, 1 ml of solution 2, 1 ml of solution 3, and 0.2 ml of solution 4. For determination of blank values, use 0.2 ml of glucose-6-phosphate dehydrogenase (Boehringer, No. 15 303) instead of solution 4. Blood samples (about 2 ml) containing 1.5 mg of ammonium heparinate and 3 mg of sodium iodoacetate were obtained from general practitioners. Some of these samples were strongly hemolytic, owing to hyperosmolality. They were centrifuged and analyzed
within 3 h. Quality control sera were purchased from Merz & Dade. “Monitrol I” (lot No. 124AB), “Qualtrol” (lot No. QT-30B) and “Monitrol II” (lot No. 33 AB) were used.
Procedures Glucose dehydrogenase method: The distribution disc is loaded in well C with 500 jl of glucose dehydrogenase reagent and in well B-if not stated otherwise-with 15 sl of sample, flushed out of the pipettor tip with 100 zl of physiological saline. The instrumental settings of the analyzer and the instructions for the computer are given in Tables 1 and 2. About 25 s is needed for the rotor to accelerate enough to transfer the samples and the reagents into the cuvettes and for the mixing. After a further 10 s (initial reading) a sequence of 4 absorbance reading is made at 20-s intervals (reading interval), In cuvette 1,
method
Hexokinase
method
Flush vol (0.9% NaCI) Reagent vol
100 p1
20pl (1:10 diluted) 40pl
500 p1
500 p1
Sample switch position
B
B
Blank switch Temperature
Water
Reagent
25#{176}C
25#{176}C
340 nm Auto, Rate
340 nm
lOs 20 s
300
15p1 or less
Sample vol
position
Wavelength Control Initial
module
settings
reading
Reading interval readings Total anal. timea button
until b A blank
2times300sb
70s
a Add about 35 s to obtain
s
60 s 1
4
No.
Endpoint
Auto,
the time elapsed
the last reading is obtained. run must be performed using
from pressing the
the start
reagents
blank
(see
Materials and Methods).
Table 2. Instructions Stored
for the Computer:
by the Header
Dehydrogenase
Method
IR 10 RI = 20 NR= 4 TF not used or 0.9675l KT = not used or 326’ Sc = 100 TC= 2or7a =
a Value
used for the alternative
mode
AD=
4
CD= SA =
0 2
HI
=120
LO
=
RM=
70 2
XX=
1
of calculation.
water is used as photometric reference sorbances of the cuvettes are compared sorbance of the standard in cuvette 2. otherwise, an aqueous standard containing cose per liter is used. The computer then lowing formula to calculate the glucose tion:
v
Variables
Tape, Glucose
and the abwith the abIf not stated 1 g of gluuses the folconcentra-
_G1#{149}SC E
()
where V = glucose concentration of the sample in the ith cuvette; G = delta absorbance of the ith cuvette; SC = standard concentration; and E = delta absorbance of the standard cuvette. For increasing the range of linearity, an alternative mode of calculation is proposed. With this approach the nonlinearity can be corrected by the formula: (G#{149}SC\ E
Vj=TF\ where KT computer.
and
TF
CLINICAL
=
correction
CHEMISTRY,
KT
(G.SC\2 E
factors
I
used
Vol. 21, No. 10, 1975
(2)
by the
1373
By straightforward FOCAL programming techniques, the program supplied by the manufacturers was changed in the following way: Another choice for the variable TC (= type of concentration) has been introduced that causes the computer to read a different block of instructions on the DEC- tape. In this block the factors KT and TF (normally used for international units calculation) are used as indicated in equation 2. To determine the values of these correction coefficients, we did at least five runs with seven standards, each containing 100 mg and 1 g glucose per 100 ml, with use of the original program. Through the three points corresponding to 0 mg, 100 mg, and 1 g of glucose per 100 ml a curve fitting was then applied according to the formula:
Fig. 1. Continuous flow system diagram of of glucose by the glucose oxidase method A single mlxkig coil has been used and the abbrevIations codes for glass fittings
0.5
y=ax+bx2
determination
the
are manufacturer’s
-
(3)
where y
=
expected
values
and x
mean
of the obtained
By comparison of equation it is evident that TF
3 with
=
E
values equations
C 0 C., 4
025.
1 and 2
0_0_0_
a
-0-0-0-0-0
,o_0
and KT
=
b 106
The hexokinase method was performed according to the manufacturers recommendations with the following exceptions: The composition of the reagents was slightly changed (see Table 1) and a blank run was introduced. In this run the hexokinase but not the glucose-6-phosphate dehydrogenase is omitted (see Materials and Methods), all the other conditions remained unchanged. In this way the serum blanks and the interferences due to glucose-6-phosphate and other substances were taken care of, except for the impurities introduced by the hexokinase enzyme preparation. A blank and an analysis run with a reading at 300 s are therefore carried out for each determination. For the glucose oxidase/peroxidase method the continuous-flow system was used according to Figure 1. The samples were aspirated for 20 s at a rate of 30/h. The glucose values were read manually from a standard curve plotted on semilogarithmic paper. Protein-containing standards with concentrations of 400, 200, 100, 50, and 25 mg of glucose per 100 ml were used.
Results and Discussion Figure 2 shows a comparison of the kinetic behaviors of the Glc-DH and the hexokinase reaction. Attempts to measure the glucose kinetically by the hexokinase reaction were unsuccessful with the GEMSAEC system. This suggested that an end-point pro1374
CLINICAL
CHEMISTRY,
Vol. 21, No.
100
0
10, 1975
200
Fig. 2. Comparison of the kinetic behaviors of the Glc-DH and the hexokinase
reaction
0. hexokinase method; #{149}, GIc-DH method. glucose per 100 ml was used
A standard
contak*ig
100 mg
cedure should be used. Another implication was that the samples had to be diluted 10-fold. The linearity of a standard curve when 5 pl or more was used was very bad by either the endpoint or the two-point kinetic method as described elsewhere (6). The deviation from linearity at concentrations of 300 mg of glucose per 100 ml was 14% or more with both methods. Furthermore a blank run had to be introduced, as some of our samples were strongly hemolytic. The closest agreement with the glucose oxidase/peroxidase method was obtained only when glucose-6-phosphate dehydrogenase was used for the blank determinations as described under procedures. There was also a significant difference (paired t -statistic, P = 0.005) when the blank value was read immediately, e.g., 10 s after mixing was completed as compared to when the blank value was read after 300 s. The glucose values were always higher in the latter case (mean difference, D = -6.64). A total reaction time of 600 s was therefore needed. On the other hand, the glucose dehydrogenase reaction can be run kinetically within only 70 s (see Table 1). The kinetics of the reaction for a wide range
Table 3. Coefficient of Variation and Deviation from Linearity as a Function of the Sample Volume (Glc-DH method, uncorrected) Sample vol. sl
15.0 10.0 7.5 a Calculated
100-mg standard glucose value ob. tamed (mg/100 ml)
Within-disc
Deviation
from
CV (%)
Iinearitya
(%)
99.12
0.45 0.50 1.01
99.49 98.03 according /expected
19.4 14.0 12.0
to the formula: -
obtained\ 1100
#{163}4expe,j
/
where ‘4expected and 4obtained are the absorbance differences of a 1000 mg glucose/i 00 ml standard as they were expected and obtained, respectively.
0.1
-‘
0
Fig. 3.
100 Kinetics
of the
200s
Glc-DH reaction
ent concentrations The arabic numerals on the right-hand centrations In mg/100 ml
as a function of differ-
side correspond
to the glucose
con-
of concentrations is shown in Figure 3. Although a perfect linearity was not obtained, it was sufficient for practical use with the GEMSAEC system, because the LR flag was never printed out by the curvesearch algorithm of the computer program. But the slight nonlinear relationship is demonstrated by the
appearance curves
of a higher-order
were submitted y
=
1.87
term when the standard
to second-order + 1.02x
(2.0
-
regression: 1Ox2)
#{149}
(4)
It has been stated that 2.5 g of bovine serum albumin should be added to the buffer per 100 ml of reagent if the samples were not deproteinized before determination.3 Therefore the effect of adding bovine serum albumin was also studied, and under the same conditions as above the following equation was obtained:
y
=
Comparison
-9.58
+
1.13x
of equations
-
(2.4
.
104x2)
(5)
4 and 5 shows that bovine
serum albumin slows down the reaction at low concentrations, but has no marked effect at high glucose concentrations.
Optimization
of the Method
Coefficient of variation: The reading interval and the number of readings have been optimized as variables in a simplex optimization procedure (10) with the goal of minimizing the within-run coefficient of variation. The initial reading was kept constant at 10 s. A reading interval of 20 s and four readings seemed to be optimal (coefficient of variation of a standard containing 100 mg glucose per 100 ml = 0.49%) 3Dr.
R. Helger,
E. Merck,
Darmstadt,
personal
communication.
(Table 2). Decreasing the reading interval from 20 to 15 s increased the CV by a factor of 2.0, and no improvement was observed by increasing the number of readings or the reading interval, or both. But these variables did not seem to be very critical and the coefficient of variation depends much more on other factors such as the cleanliness of the cuvettes and the state of the pipette units. Linearity: The deviation from linearity of the uncorrected glucose dehydrogenase method was 4% at 300 mg, 8% at 500 mg, and 19% at 1 g of glucose per 100 ml. To improve this linearity, we investigated the effect of various variables. Shortening the reading interval seemed to have no effect on the linearity. Using the two-point kinetic mode, where only one reading interval is considered (number of readings = 2), we found no improvement in linearity by shortening the interval from 60 to 40 or 20 s. Table 3 shows the effect of sample volume on the coefficient of variation and the linearity of the standard curve. Evidently decreasing the sample volume increases the linearity, but also increases the coefficient of variation. A sample volume of 10 pl seemed to be the best compromise. Another possibility for increasing linearity would be to increase the enzyme concentration. But there is a limitation: the ENI-GEMSAEC photometer is only linear up to an absorbance of 2.0 units. With the settings shown in Table 1 and a 15-jsl sample volume, the last absorbance reading was about 1.3 when a 1000-mg standard was used. Increasing the enzyme concentration threefold caused the standard to be out of the linear range of the photometer. The 500-mg standard had the absorbance of 1.7 in this case. With only a 10-p1 sample volume, this value could be decreased to 1.6 absorbance units, whereas the 1000 mg standard was still out of the linear range. With twice the enzyme concentration (24 mg/100 ml buffer) and a sample volume of 10 tl, the last absorbance reading of the 1000-mg standard was still within the linear range of the photometer with an absorbance of 1.8 units. But as these measurements were made with an aqueous standard the absorption attributable to the serum blank added up to CLINICAL
CHEMISTRY,
Vol. 21, No. 10. 1975
1375
all of these absorption values. These blank values were usually about 0.05 to 0.1 absorbance units. This blank value would therefore shift the absorbance of a sample containing 1000 mg of glucose per 100 ml close to the limit of linearity. By using a 10-p1 sample volume and 18 mg of enzyme per 100 ml of buffer, a last absorbance reading of 1.5 absorbance units was obtained for the 1000-mg standard. But this standard curve was still not linear, because there was a 7% deviation from linearity at 600 mg/100 ml. It therefore seems impossible to reach the linearity claimed by others (8) using other instrumentations. An alternative mode of calculation is therefore proposed: The introduction of the nonlinear term of equation 2 could increase the linearity. The numerical values of TF and KT were 0.9675 and 326, respectively. The standard curves used for the calculation of equation 4 were corrected according to equation 2 and were then again submitted to linear regression analysis; the following equation was obtained: y
a0 a1
=
a0
=
4.333. 0.999
±
r2
+
iO
a1x, where ±
1.235
Obtained mean
No.
SD (mgi 100 ml)
Sam-
mg/i 00 ml
plea
CV. %
Day-to-day precision of the GIc-DH method Monitrol
I
79(76)
74.1
33
3.1
4.17
Monitrol
II
203 (198)
204.6 103.2
20 75
6.9 3.7
3.37
59
29.6
Qual-trol
-
Standardb
1000
1024.6
3.58
2.89
Within-day precision of the Glc-DH method Monitrol
I
79 (76)
Qual-trol
-
Pathological
-
76.2
20
0.9
1.17
97.7 181,6
20 27
0.67 0.89
0.69
981.7
20
4.69
0.48
0.49
serum Standardb
1000
a The values
for the hexokinase-method
are given
first, the values
for the GOD-Perid method follow in parentheses. bA standard containing 1000mg glucose per 100 ml bovine um albumin solution was used (see Materials and Methods).
ser-
400
(S.E.)
and
S
0.999
With this correction the standard curves became linear with a very high degree of correlation and the deviation from linearity was about 0.8% at 1 g of glucose per 100 thl. This correction was therefore used throughout the rest of this study. If these software changes are not made, it would be best to use a 10-pl sample volume and an enzyme concentration of about 0.72 U/ml of buffer, corresponding to 24 mg/100 ml of buffer in the present case. Samples containing more than 400 mg of glucose per 100 ml would then have to be diluted, as the deviation from linearity was above 4% at this concentration. Precision and accuracy: The variation within series and from series to series (including from day to day) has been checked by submitting nine runs (a total of 126 determinations performed on four con-
secutive days) to analysis of variation. A mean value of 99.9 mg of glucose per 100 ml was obtained and the coefficients of variation were 0.78% and 2.54% withinrun and day-to-day, respectively. Day-to-day precision has routinely been checked over a period of two and a half months with different standards and quality-control sera. The results are summarized in Table 4. At low normal and normal glucose concentrations the coefficient of variation is about 4% and at high levels it is about 3% under normal routine working conditions. The accuracy is very good; the stated values of the Monitrol quality control sera agree closely with the obtained values. With the standard containing 1 g of glucose per 100 ml of BSA solution, the mean deviated by only 2.5% from the expected value during this long time. The accuracy was also 1376
Expected value(s)a
Standard or serum
(6)
2.426#{149}10- (S.E.), =
Table 4. Data on Precision
CLINICAL CHEMISTRY, Vol. 21, No. 10, 1975
300
I ‘S
200
100
0 100
200
300
400
Fig. 4. Comparison of 192 glucose values as obtained by the hexokinase
method (x-axis) and the GIc-DH method (y-axis)
The line obtaIned by linear regression Is Indicated. The data plotted are the same as used for the calculation In table 5 (Comparison of the hexoklnase vs. GIc-DH method)
checked by comparing the corrected values obtained by the Glc-DH method with those obtained by the hexokinase and the GOD-Perid method. The comparison with the hexokinase method is shown on Figure 4. The regression coefficients with their standard errors and the coefficients of determination of both comparisons are given in Table 5. The difference in slope of these comparisons is significant (P = 0.020), the GOD-Perid method yielding lower and the hex-
okinase
method
slightly
did the Glc-DH method. variation in the GOD-Perid two methods, which may
higher
glucose
values
than
There seems to be more method than in the other either be due to the fact
Table 5. Comparison of the GIc-DH Method with the Hexokinase and the GOD-Perid Methods by Linear Regression Analysisa Method Hexokinase
GOD-Perid a According
a0
s
a,
-2.678
0.997
-6.115
3.708
to the equation
and s, of the regression
y
r2
n
0.971
0.007
0.992
192
1.030
0.029
0.907
a0
+
5,
a1x, with the standard
129 errors
coefficients.
the values of the latter are read from a standard curve by eye or to the interfering substances mentioned earlier. Although slightly higher values are obtained with the hexokinase method than with the Glc-DH method, the agreement between these two methods can be considered very satisfactory for practical uses. Specificity: The specificity of this reaction has been extensively tested elsewhere (8) and some interference has been reported with xylose and mannose; 10 to 100 mg of D-xylose and D-mannose have therefore been added, in 10-mg increments, to a standard containing 100 mg of glucose per 100 ml. The mean increase in the glucose value was 12.5 pg and 1.9 pg per milligram of added xylose and mannose, respectively-values so small that they can be neglected for practical purposes. The Glc-DH is specific for the anomer of D-glucose only. This has to be taken into account when endpoint reactions are carried out (8), but it has no or little effect on kinetic measurements as the substrate is in excess. D-Glucose in solution or in the circulation is in an equilibrium consisting of about 64 and 36% of the fi and a anomers, respectively (11). This equilibrium is not appreciably disturbed as only about 0.5% of the 3-D-glucose (0.49% during 60 s under the given experimental conditions) is used up during the reaction period. Therefore no mutarotase has to be added. that
We conclude that the glucose dehydrogenase method seems to be a very suitable procedure for determination of glucose with centrifugal analyzers.
This reaction seems to be particularly suited for kinetic measurements. It is a very specific reaction, but even though it is specific for fl-D-glucose alone no addition of mutarotase is necessary when glucose is measured kinetically. The limited range of linearity can easily be expanded with a simple mathematical correction. The close agreement with the hexokinase method and the accuracy testing prove its applicability in the routine clinical laboratory.
We thank Drs. H. Lang and R. Helger (E. Merck, 61 Darmstadt 2, Postfach 4119, West Germany) for the generous supply of the enzyme preparation and Prof. H. Keller (Zentrallaboratorium, Kantonsspital St. Gallen, Switzerland) for the valuable information. The skillful technical assistance of Silvia Brogli and the secretarial help of Rosmarie Huber-Zehnder is gratefully acknowledged.
References 1. Meites, S., and Saniel-Banrey, K., Modified glucose oxidase method for determination of glucose in whole blood. Clin. Chem. 19, 308 (1973). 2. Genshaw, M. A., and Gunter, C. R., Optical bleaching anisidine glucose tests. Clin. C/tern. 19, 1227 (1973). 3. Chinh, N. H., Mechanism of interference cose oxidase/peroxidase method for serum 499 (1974).
in o-di-
by uric acid in the gluglucose. Clin Chem. 20,
4. Blaedel, W. J., and Uhl, J. M., Nature of materials in serum that interfere in the glucose oxidase-peroxidase-.o.dianisidine method for glucose, and their mode of action. Clin. Chem. 21, 119 (1975). 5. Sline, M. W., In Methoden der Enzymatischen Bergmeyer, Ed., Verlag Chemie, Weinheim/Bergstr., 6. Haeckel, tion of the GEMSAEC (1973).
Analyse, H. U. 1962, p 117.
R., Evaluation of an enzymatic two-point determinaglucose concentration in 40 zl blood samples with a Analyzer. Z. Kim. Chem. Kim. Biochem. 11, 243
7. Hasson, W., Penton, J. R., and Widdowson, G. M., Determination of glucose with a research model Aminco “Rotochem” by the hexokinase reaction. Clin. Chem. 20, 15 (1974). 8. Banauch, D., Br#{252}mmer,W., Ebeling, W., et al., Eine GlucoseDehydrogenase f#{252}r die Glucose-Bestimmung in Korperflussigkei. ten. Z. Kim. Chern. Kim. Biochem. 13, 101 (1975). 9. Keller, H., Faust, U., and Becker, J., Enzymatische GlucoseBestimmung durch Kurzzeitmessung der Reaktionskinetik. Chem. Rundsch. 26, 24 (1973). 10. Krause, R. D., and Lott, J. A., Use of the simplex method to optimize analytical conditions in clinical chemistry. Clin. Chem. 20, 775 (1974). 11. Lundsgaard, C., and Holboll, S. A., Studies in carbohydrate metabolism. II. Investigations into the mutarotation of fl-glucose under various conditions. J. Biol. Chern. 65, 305 (1925).
CLINICAL
CHEMISTRY,
Vol. 21, No. 10, 1975
1377
