- Email: [email protected]
The use of mass fragmentography in the evaluation of routine methods for glucose determination
353
Clinica Chimica Acta, 7 2 (1976) 353-362 @ Elsevier/North-Holland Biomedical Press. Amsterdam
-
Printed
in The Netherlands
CCA 8124
THE USE OF MASS FRAGMENTOGRAPHY IN THE EVALUATION ROUTINE METHODS FOR GLUCOSE DETERMINATION
INGEMAR
BJijRKHEM
Department (Sweden) (Received
*, ROLF
of Clinical Chemistry,
BLOMSTRAND,
Karolinska
OLLE
Institutet
FALK
and GtSSTA
at Huddinge
Hospital,
OF
OHMAN
Huddinge
June 3, 1976)
Summary A highly specific and accurate mass fragmentographic reference method for determination of serum glucose is described. A fixed amount of hepta-deuterated glucose is added to a fixed amount of serum. The mixture is lyophilized, converted into the penta-trimethylsilyl-methyloxime derivative and subjected to analysis with a combined gas chromatograph - mass spectrometer equipped with a MID-unit (multiple ion detector). The amount of unlabeled glucose was determined from the ratio between recordings at m/e 319 and 323. The two ions used correspond to the base peak in the mass spectrum of derivative of unlabeled and hepta-deuterium labeled glucose, respectively. The relative standard deviation of the method was 1.9%. The method was compared with different enzymatic methods based on the hexokinase reaction and the glucose oxidase reaction. Of the different methods tested, a glucose oxidase method with determination of maximal rate of consumption of oxygen gave results most close to the results of the reference method.
Introduction There is a great number of clinical chemical methods available for determination of glucose, and consequently there is an immense literature concerning comparisons between different methods [ 11. In the evaluation of new routine methods for glucose determination, comparisons are often made with methods which have the merit of being traditionally accepted as clinical chemical routine methods. These traditional methods may however not be the most accurate and specific methods available. Hitherto, enzymatic methods are consider* To whom correspondence
should be addressed.
354
ed to be more specific than other methods for determination of glucose. It must be assumed that the enzyme preparations are sufficiently pure and that the reaction products can be quantitated accurately. In the widely used glucose oxidase-peroxidase methods there may be a lack of specificity in the chromogen coupling step. It seems to be generally accepted that enzymatic methods based on the hexokinase reaction are more specific than most other enzymatic methods. According to our knowledge, however, hexokinase methods have not been tested with use of methods which may be considered to be still more accurate and specific. As part of a general project to develop reference methods of very high accuracy based on mass fragmentography and use of isotope labeled internal standards, we here describe a mass fragmentographic method for determination of glucose in serum. The results obtained with this reference method have been compared with results obtained with different routine methods. A mass fragmentographic analysis of glucose has previously been published by Sweeley et al. [2] with use of hexa-trimethylsilyl derivative of glucose and hepta-deuterated glucose. If high accuracy is wanted use of hexa-trimethylsilyl derivative of glucose is less suited. Thus the (Yand /3 forms of glucose separate on the column and it is difficult to obtain the same ratio between the o( and p form in the isotope labeled standard as in the unlabeled glucose. In the present work we have used the penta-trimethylsilyl-methyloxime derivative in order to eliminate the stereochemistry at C, (cf. ref. 3). Use of methyloxime derivatives gives syn and anti isomers which also often separate on a gas chromatographic column. This is however of little importance in a mass fragmentographic assay with use of deuterium labeled internal standard since the ratio between unlabeled molecules and labeled molecules must be the same in the syn and anti isomers. Materials
and methods
[‘H7]G1ucose was obtained from Merck, Sharp and Dohm (Canada Ltd., Montreal, Canada) as one component in a deuterated algal sugar mixture. The other deuterated sugars in the mixture did not interfere with the determination. Thus in the gas-liquid chromatography one major and one minor peak corresponding to the syn and anti forms respectively of methoxime-trimethylsilyl derivative of glucose were obtained and there were no other peaks with similar chromatographic properties. The material in the gas-chromatographic peak corresponding to methoxime-trimethylsilyl derivative of glucose consisted exclusively of hepta- and hexa-deuterated molecules (cf. Results).
Unlabeled glucose Glucose monohydrate was obtained from Merck, Darmstadt (GmbH). No contaminating compounds could be detected by gas-liquid chromatography of the methoxime-trimethylsilyl derivative or the hexa-trimethylsilyl derivative. When subjected to thin-layer chromatography according to Kudla and McVean [4] an unidentified spot appeared where the sample was applied. Densitometry with use of a Vitatron [5] showed that this compound apparently corresponded to less than 0.5% of the total mass. After drying in vacua over
355
Pz05 for 24 h at 6O”C, the material had lost 1.018 equivalents of water. Loss of 1.018 equivalents of water instead of 1.000 corresponds to an error with respect to weight of glucose of less than 0.2%. The dried material had a m.p. of 145-150°C (reported m.p. for an equilibrium mixture, 146-150°C [6]). Serum
Serum without visible hemolysis was collected from diabetic and kidneytransplanted patients. The fresh serum was frozen and kept at -20°C prior to analysis. Repeated analysis with the same method at different times after the freezing gave identical results. Each serum was analyzed within 4 h after the thawing. Routine
laboratory
procedures
Serum glucose was determined with five different methods. When not otherwise stated, the LKB-Ultrolab system was used. In order to obtain optimal accuracy evaluation was performed manually from a standard curve. 1. HK-NADPH. The hexokinase-glucose-6-phosphate dehydrogenase method with direct measurement of NADPH at 340 nm [ 71. Reagent kit from Boehringer Mannheim was used. 2. HK-IN?“. The hexokinase-glucose-6-phosphate method coupled with colorimetric determination of reduction of the chromogen 2-p-iodophenyl-3-p-nitrophenyl-5-phenyl-tetrazolium chloride (INT) [ 81. The reagent kit “Glucostrate/ Division of Warner-Lambert Company, improved” from General Diagnostic, Morris Plains, New Jersey was used. 3. GO-DC. The glucose oxidase-peroxidase method coupled with oxidation of the chromogen dicarboxidin (y,y’-(4,4’-diamino-3,3_diphenyldioxide dibutyric acid)). The reagent kit “Glox novum” from AB KABI, Stockholm, Sweden was used. 4. GO-ORM. The glucose oxidase oxygen rate method with use of a Beckman Glucose analyzer [ 91. 5. DRM. The Dipstick-Reflomat method [lo] using dipsticks “Reflotest-glucase” and a Reflomat from Boehringer Mannheim. Preparations
of samples for mass fragmentography
(cf. ref. 3)
The deuterated sugar mixture, 50 ~1, containing 300 nmoles of hepta-deuterated glucose, was lyophilized together with 50 ~1 of serum over-night. A solution of methoxylamine * HCl in dry pyridine, 0.5 ml, 10 mg/ml, was then added. The reaction was allowed to proceed for 2 h at 80°C. Bistrimethylsilyltrifluoroacetamide, 150 ~1, was then added. This was allowed to react for an additional 15 min at 80°C. Mass fragmen tography
The above solution was analyzed directly after cooling by combined gas chromatography - mass spectrometry, using the LKB 9000 instrument equipped with a 3% SE-30 column (on Chromosorb W, 80-100 mesh, 2 mm X 2.5 m). Carrier gas was helium and the flow rate was 30 ml/min. The column temperature was 200°C and the temperature of the flash heater and ion source both about 250°C. The electron multiplier sensitivity was set to 160. The first chan-
356
nel of the MID-unit was focused on the ion at m/e 319 and the second at m/e 323, corresponding to the base peak in the mass spectrum of derivative of unlabeled and labeled glucose, respectively. The amplification used for both channels was 10X. The filter settings were 0.5 Hz for both channels and the measuring time was 20 ms. The MID-recordings were made on UV-paper and the peak heights were measured.
Statistics Comparison between the MID-method and the routine methods lated with two independent variables (“line of best fit” [ll]).
was calcu-
Results The mass spectra of the penta-trimethylsilyl-D -glucose-methyloxime derivatives of unlabeled and hepta-deuterated glucose are shown in Fig. 1. The fragmentation pattern was in complete accord with that reported by Laine and
loo319
OCH3 i
F”
HCOTMS .-h
TMSacai .,, ..____............... HCOTMS
319 --‘.--m
HCOTMS
x --
103
H20TMS
m/e
loo-
323
TMSOW I,.....: DCOTMS _.,._
p3 ‘boa
DCOTMS ..,.. __ . CD20TMS
. . . . . . . ios
m/e Fig. trum)
1.
Mass and
spectrum
hepta-deuterated
of
penta-trimethylsilyl-methyloxime glucose
(lower
spectrum).
derivative
of
unlabeled
glucose
(upper
spec-
357
323
1 dentfon Fig.
2
3
4
retentmntime
tnme (mInutesI
2. MID-recording
of derivative
of unlabeled
glucose
(left)
(minutes)
and
hepta-deuterated
glucose
(right).
Sweeley [3]. The fragment with m/e 319 and m/e 323 corresponding to carbon atoms C3-C6 [3] was used in the mass fragmentographic analysis of the labeled and unlabeled derivative, respectively. It was calculated [12] that the fragment derived from the derivative of the deuterated glucose contained 93% tetradeuterated molecules and 7% trideuterated molecules. The MID-recordings of derivatives of labeled and unlabeled glucose are shown in Fig. 2. One major peak was obtained followed by a small accompanying peak. The two peaks cor-
15-
/ ,i./
1 o-
.1’1’ ,~i7 /’ N
05-
//’
mmol Fig.
3.
the
range
Standard O-30
curve mmol/l.
for
1 30
20
10
glucose determination
per
liter of
serum
glucose
with
the
mass
fragmentowaphic
method
in
25 I
A
.
.. .
f ,
5’
, t’
Y=O.98
-4 . ’
.
J I*
5
10 mmol
+049
r = 0 998
glucose
15
per
20
Idler
25 MID
serum
25 ;
y=lOlx-0
25
r=O 998
5
10
15
20
25
359
v=0.99x r :
mmol
Fig.
4.
(B),
the
Comparison GO-DC-method
glucose
per
between (C),
liter
the the
+ 086
0988
serunl
MID-method
GO-ORM-method
MID and (D)
the and
HK-NADPH-method the
DRM-method
(A).
the
HK-INT-method
(E).
respond to the syn and anti forms of the methyloxime derivative [ 31. The two isomers had identical mass spectra. Only the first major peak was used in the quantitation, and it was clearly shown that the presence of the minor peak did not influence on the results obtained. Thus it was considered to be unnecessary to use a long retention time in the gas-liquid chromatography in order to obtain a baseline separation between the two isomers. The height of the first peak was used in the quantitation and it was shown that this gave the same results as when using the area of the combined major and minor peak. The ratio between the peak heights of m/e 319 and m/e 323 obtained in MID-recordings from analysis of different standard mixtures together with 300 nmoles of [2H7]glucase was linear with the amount of glucose up to a concentration of 30 mmol/l (Fig. 3). In the serum analysis it was shown that the relative standard deviation
360
TABLE
I
COMPARISON ROUTINE Method
BETWEEN
THE
MASS
FRAGMENTOGRAPHIC
METHOD
THE
Slope
Intercept
Correlation
Relative
0.98
0.49
0.998
2.07
MID HK-N
AND
DIFFERENT
METHODS S.D.
(%)
1.93 ADPH
HK-INT
1.14
a.15
0.996
1.34
GO-DC
1.01
--0.26
0.998
2.92
GO-ORM
1.04
-3.09
0.998
2.49
DRM
0.99
0.988
3.01
0.86
of the method was 1.9% as calculated from 40 duplicate analyses. The accuracy of the method was tested by addition of 300 and 600 nmoles of glucose to a serum containing 211 nmoles of glucose. The result obtained with the MIDmethod was 510 nmoles (0% error) and 800 nmoles (1.2% error), respectively. The specificity of the method was tested by using the fragments at m/e 205 and m/e 208 in the analysis (cf. Fig. 1). Exactly the same results were obtained as when using the fragments at m/e 319 and m/e 323. Analysis of serum samples to which no deuterated glucose had been added gave the same ratio between the tracings at m/e 319 and m/e 323 as obtained in analysis of unlabeled glucose, indicating that no other serum constituents interfere significantly with the determination. Comparisons between the MID-method and five methods used in clinical chemical routine work is shown in Fig. 4. The constants of the regression lines as well as the relative standard deviation of the different methods are summarized in Table I. Discussion The very high specificity of the present mass-fragmentographic method for glucose determination in combination with the fact that the ratio between labeled and unlabeled molecules is determined with a high degree of accuracy, makes it likely that the most significant errors in the method are related to the errors in the pipettations. It should be pointed out that methoxime-trimethylsilyl derivatives of all other sugars which may be present in serum was separated from the derivative of glucose under the conditions employed. In view of the possibility that a hexokinase preparation may contain other enzymes it seems probable that the accuracy of the present method is higher than that of the hexokinase method, which previously has been used as reference method. The slope of the regression line in the comparison between the MID-method and the HK-NADPH method was very close to 1.00. A surprisingly high positive intercept was however found in repeated experiments and with use of different batches of the kit. In principle, part of the explanation could be that the enzyme preparation according to the specification contains trace amount of phosphohexosisomerase. In the presence of fructose, the activity of this enzyme should give higher values in the over-all determination. The concentration
361
of fructose in serum is however in general lower than the intercept obtained and addition of fructose to different glucose mixtures prior to the determination gave little or no increase in absorbance. The slope of the regression line in the comparison between the MID-method and the HK-INT method was significantly higher than 1.00. In addition a small negative intercept was found. This complex regression picture is difficult to explain. It should be pointed out that this method is the most complicated one from a chemical point of view with three coupled reactions. The slope of the regression line in the comparison between the MID-method and the GO-DC method was very close to 1.00. A small negative intercept was found. It is well known that hydrogen donors in serum may compete with the chromogen to reduce the hydrogen peroxide in glucose-oxidase chromogen methods with lower results as a consequence. It has been shown that 5-hydroxylated indoles may give rise to a constant as well as a systematic error in glucase-oxidase chromogen methods [13]. The negative intercept is in agreement with the results obtained by Lott and Turner [14] who compared a glucoseoxidase-chromogen method with GO-ORM. The slope of the regression line in the comparison between the MID-method and the GO-ORM method was very close to 1.00 and only a very small negative intercept was found. From this finding together with the finding of a correlation coefficient of 0.998 it can be concluded that the GO-ORM method is the method which gives results most similar to the results of the reference method in the range tested. The high specificity of the GO-ORM method must be related to the fact that the maximal reaction velocity is determined. In such a procedure a possible unspecificity of the enzyme preparation should influence the results to a smaller extent than in other procedures. Thus interfering compounds must react to a considerable extent within ten seconds in order to influence the results. In addition the method lacks a coupling chromogene which makes it less sensitive towards side-reactions. It may be expected that the DRM-method should be less accurate and specific than the other methods tested. This method is difficult to calibrate exactly and the accuracy as well as the precision is dependent upon the skill with which the laboratory procedure is performed. In spite of that the present determinations with the DRM-method were performed under optimal conditions, it is surprising that such high correlation with the reference method could be obtained. The slope of the regression line in the comparison with the MIDmethod was very close to 1.00. There was however a considerable positive intercept (0.86 mmol/l). To summarize, all the routine methods parallel well with the reference method and the small discrepancies found are of little or no importance from a clinical point of view under most conditions. In determinations of low glucose concentrations with the HK-NADPH and DRM-methods, the error in the methods may sometimes be significant. A consequence of the present study is that when a mass-fragmentographic method is not available, the GO-ORM method is more suitable as a reference method than the HK-NADPH method. Acknowledgments This work
was supported
by the Bank
of Sweden
Tercentenary
Fund.
362
References 1
Gochman.
2
Sweeley.
N.,
Ryan,
C.C.,
3
Laine,
R.A.
4
Kudla,
R.
5
Liljeqvist,
6
Handbook
7
Slein.
M.W.
(1965)
demic
Press.
New
and and L.,
Sweeley,
Giirtler.
Coburn.
H.J.
and
Stevens,
J.F.
(1971)
10
Kattermann. Strewhart, Biemann,
14
Lott,
New K. D.R.
d.A.
and
(1971)
Anal.
Tech.
and
Blomstrand,
and
Physics of
Widdowson,
Ryhage,
R.
G.M.
(1966)
Biochem.
Bull. R.
(1975)
Reg.
43. Med.
(1973) 56th
Enzymatic
edn.
38,
Chem.
p. C-311.
Analysis
Clin.
Chem.
38,
J.J.
(1973)
Chim.
Naroska,
(1931)
and
Acta
I. (1974)
Economic
in Mass Huggins.
Turner,
21.
356
1549
279 Sand.
CRC
(Bergmeyer,
27, Press, 11.-U.,
Clin. 32.
197 Cleveland. cd.),
pp.
Ohio 117-123,
Aca-
Chem.
19,
127
199
Dtsch. Control
Med. of
Wochenschr.
Quality
of
99,
2603
Manufactured
Y ark
(1962)
Chem.
533 Tech.
Acta
(1975)
Anal.
N.Y.
Clin.
and
and
I. and
(1968)
Methods
Carroll,
W.A.
Nostrand, Nelson,
R.
in York,
8
13
J.
R.E.
Fries,
C.C. D.
of Chemistry
11
Sterling,
W.H..
McVean,
9
12
W.T.,
Elliott,
K.
Spectromrtrv. A.K.
(1974)
(1975)
Clin.
p. 223. Anal. Chem.
McGraw-Hill,
Biochem. 21.
1754
59,
46
New
York
Products,
p.
104,
D.
van
Clinica Chimica Acta, 7 2 (1976) 353-362 @ Elsevier/North-Holland Biomedical Press. Amsterdam
-
Printed
in The Netherlands
CCA 8124
THE USE OF MASS FRAGMENTOGRAPHY IN THE EVALUATION ROUTINE METHODS FOR GLUCOSE DETERMINATION
INGEMAR
BJijRKHEM
Department (Sweden) (Received
*, ROLF
of Clinical Chemistry,
BLOMSTRAND,
Karolinska
OLLE
Institutet
FALK
and GtSSTA
at Huddinge
Hospital,
OF
OHMAN
Huddinge
June 3, 1976)
Summary A highly specific and accurate mass fragmentographic reference method for determination of serum glucose is described. A fixed amount of hepta-deuterated glucose is added to a fixed amount of serum. The mixture is lyophilized, converted into the penta-trimethylsilyl-methyloxime derivative and subjected to analysis with a combined gas chromatograph - mass spectrometer equipped with a MID-unit (multiple ion detector). The amount of unlabeled glucose was determined from the ratio between recordings at m/e 319 and 323. The two ions used correspond to the base peak in the mass spectrum of derivative of unlabeled and hepta-deuterium labeled glucose, respectively. The relative standard deviation of the method was 1.9%. The method was compared with different enzymatic methods based on the hexokinase reaction and the glucose oxidase reaction. Of the different methods tested, a glucose oxidase method with determination of maximal rate of consumption of oxygen gave results most close to the results of the reference method.
Introduction There is a great number of clinical chemical methods available for determination of glucose, and consequently there is an immense literature concerning comparisons between different methods [ 11. In the evaluation of new routine methods for glucose determination, comparisons are often made with methods which have the merit of being traditionally accepted as clinical chemical routine methods. These traditional methods may however not be the most accurate and specific methods available. Hitherto, enzymatic methods are consider* To whom correspondence
should be addressed.
354
ed to be more specific than other methods for determination of glucose. It must be assumed that the enzyme preparations are sufficiently pure and that the reaction products can be quantitated accurately. In the widely used glucose oxidase-peroxidase methods there may be a lack of specificity in the chromogen coupling step. It seems to be generally accepted that enzymatic methods based on the hexokinase reaction are more specific than most other enzymatic methods. According to our knowledge, however, hexokinase methods have not been tested with use of methods which may be considered to be still more accurate and specific. As part of a general project to develop reference methods of very high accuracy based on mass fragmentography and use of isotope labeled internal standards, we here describe a mass fragmentographic method for determination of glucose in serum. The results obtained with this reference method have been compared with results obtained with different routine methods. A mass fragmentographic analysis of glucose has previously been published by Sweeley et al. [2] with use of hexa-trimethylsilyl derivative of glucose and hepta-deuterated glucose. If high accuracy is wanted use of hexa-trimethylsilyl derivative of glucose is less suited. Thus the (Yand /3 forms of glucose separate on the column and it is difficult to obtain the same ratio between the o( and p form in the isotope labeled standard as in the unlabeled glucose. In the present work we have used the penta-trimethylsilyl-methyloxime derivative in order to eliminate the stereochemistry at C, (cf. ref. 3). Use of methyloxime derivatives gives syn and anti isomers which also often separate on a gas chromatographic column. This is however of little importance in a mass fragmentographic assay with use of deuterium labeled internal standard since the ratio between unlabeled molecules and labeled molecules must be the same in the syn and anti isomers. Materials
and methods
[‘H7]G1ucose was obtained from Merck, Sharp and Dohm (Canada Ltd., Montreal, Canada) as one component in a deuterated algal sugar mixture. The other deuterated sugars in the mixture did not interfere with the determination. Thus in the gas-liquid chromatography one major and one minor peak corresponding to the syn and anti forms respectively of methoxime-trimethylsilyl derivative of glucose were obtained and there were no other peaks with similar chromatographic properties. The material in the gas-chromatographic peak corresponding to methoxime-trimethylsilyl derivative of glucose consisted exclusively of hepta- and hexa-deuterated molecules (cf. Results).
Unlabeled glucose Glucose monohydrate was obtained from Merck, Darmstadt (GmbH). No contaminating compounds could be detected by gas-liquid chromatography of the methoxime-trimethylsilyl derivative or the hexa-trimethylsilyl derivative. When subjected to thin-layer chromatography according to Kudla and McVean [4] an unidentified spot appeared where the sample was applied. Densitometry with use of a Vitatron [5] showed that this compound apparently corresponded to less than 0.5% of the total mass. After drying in vacua over
355
Pz05 for 24 h at 6O”C, the material had lost 1.018 equivalents of water. Loss of 1.018 equivalents of water instead of 1.000 corresponds to an error with respect to weight of glucose of less than 0.2%. The dried material had a m.p. of 145-150°C (reported m.p. for an equilibrium mixture, 146-150°C [6]). Serum
Serum without visible hemolysis was collected from diabetic and kidneytransplanted patients. The fresh serum was frozen and kept at -20°C prior to analysis. Repeated analysis with the same method at different times after the freezing gave identical results. Each serum was analyzed within 4 h after the thawing. Routine
laboratory
procedures
Serum glucose was determined with five different methods. When not otherwise stated, the LKB-Ultrolab system was used. In order to obtain optimal accuracy evaluation was performed manually from a standard curve. 1. HK-NADPH. The hexokinase-glucose-6-phosphate dehydrogenase method with direct measurement of NADPH at 340 nm [ 71. Reagent kit from Boehringer Mannheim was used. 2. HK-IN?“. The hexokinase-glucose-6-phosphate method coupled with colorimetric determination of reduction of the chromogen 2-p-iodophenyl-3-p-nitrophenyl-5-phenyl-tetrazolium chloride (INT) [ 81. The reagent kit “Glucostrate/ Division of Warner-Lambert Company, improved” from General Diagnostic, Morris Plains, New Jersey was used. 3. GO-DC. The glucose oxidase-peroxidase method coupled with oxidation of the chromogen dicarboxidin (y,y’-(4,4’-diamino-3,3_diphenyldioxide dibutyric acid)). The reagent kit “Glox novum” from AB KABI, Stockholm, Sweden was used. 4. GO-ORM. The glucose oxidase oxygen rate method with use of a Beckman Glucose analyzer [ 91. 5. DRM. The Dipstick-Reflomat method [lo] using dipsticks “Reflotest-glucase” and a Reflomat from Boehringer Mannheim. Preparations
of samples for mass fragmentography
(cf. ref. 3)
The deuterated sugar mixture, 50 ~1, containing 300 nmoles of hepta-deuterated glucose, was lyophilized together with 50 ~1 of serum over-night. A solution of methoxylamine * HCl in dry pyridine, 0.5 ml, 10 mg/ml, was then added. The reaction was allowed to proceed for 2 h at 80°C. Bistrimethylsilyltrifluoroacetamide, 150 ~1, was then added. This was allowed to react for an additional 15 min at 80°C. Mass fragmen tography
The above solution was analyzed directly after cooling by combined gas chromatography - mass spectrometry, using the LKB 9000 instrument equipped with a 3% SE-30 column (on Chromosorb W, 80-100 mesh, 2 mm X 2.5 m). Carrier gas was helium and the flow rate was 30 ml/min. The column temperature was 200°C and the temperature of the flash heater and ion source both about 250°C. The electron multiplier sensitivity was set to 160. The first chan-
356
nel of the MID-unit was focused on the ion at m/e 319 and the second at m/e 323, corresponding to the base peak in the mass spectrum of derivative of unlabeled and labeled glucose, respectively. The amplification used for both channels was 10X. The filter settings were 0.5 Hz for both channels and the measuring time was 20 ms. The MID-recordings were made on UV-paper and the peak heights were measured.
Statistics Comparison between the MID-method and the routine methods lated with two independent variables (“line of best fit” [ll]).
was calcu-
Results The mass spectra of the penta-trimethylsilyl-D -glucose-methyloxime derivatives of unlabeled and hepta-deuterated glucose are shown in Fig. 1. The fragmentation pattern was in complete accord with that reported by Laine and
loo319
OCH3 i
F”
HCOTMS .-h
TMSacai .,, ..____............... HCOTMS
319 --‘.--m
HCOTMS
x --
103
H20TMS
m/e
loo-
323
TMSOW I,.....: DCOTMS _.,._
p3 ‘boa
DCOTMS ..,.. __ . CD20TMS
. . . . . . . ios
m/e Fig. trum)
1.
Mass and
spectrum
hepta-deuterated
of
penta-trimethylsilyl-methyloxime glucose
(lower
spectrum).
derivative
of
unlabeled
glucose
(upper
spec-
357
323
1 dentfon Fig.
2
3
4
retentmntime
tnme (mInutesI
2. MID-recording
of derivative
of unlabeled
glucose
(left)
(minutes)
and
hepta-deuterated
glucose
(right).
Sweeley [3]. The fragment with m/e 319 and m/e 323 corresponding to carbon atoms C3-C6 [3] was used in the mass fragmentographic analysis of the labeled and unlabeled derivative, respectively. It was calculated [12] that the fragment derived from the derivative of the deuterated glucose contained 93% tetradeuterated molecules and 7% trideuterated molecules. The MID-recordings of derivatives of labeled and unlabeled glucose are shown in Fig. 2. One major peak was obtained followed by a small accompanying peak. The two peaks cor-
15-
/ ,i./
1 o-
.1’1’ ,~i7 /’ N
05-
//’
mmol Fig.
3.
the
range
Standard O-30
curve mmol/l.
for
1 30
20
10
glucose determination
per
liter of
serum
glucose
with
the
mass
fragmentowaphic
method
in
25 I
A
.
.. .
f ,
5’
, t’
Y=O.98
-4 . ’
.
J I*
5
10 mmol
+049
r = 0 998
glucose
15
per
20
Idler
25 MID
serum
25 ;
y=lOlx-0
25
r=O 998
5
10
15
20
25
359
v=0.99x r :
mmol
Fig.
4.
(B),
the
Comparison GO-DC-method
glucose
per
between (C),
liter
the the
+ 086
0988
serunl
MID-method
GO-ORM-method
MID and (D)
the and
HK-NADPH-method the
DRM-method
(A).
the
HK-INT-method
(E).
respond to the syn and anti forms of the methyloxime derivative [ 31. The two isomers had identical mass spectra. Only the first major peak was used in the quantitation, and it was clearly shown that the presence of the minor peak did not influence on the results obtained. Thus it was considered to be unnecessary to use a long retention time in the gas-liquid chromatography in order to obtain a baseline separation between the two isomers. The height of the first peak was used in the quantitation and it was shown that this gave the same results as when using the area of the combined major and minor peak. The ratio between the peak heights of m/e 319 and m/e 323 obtained in MID-recordings from analysis of different standard mixtures together with 300 nmoles of [2H7]glucase was linear with the amount of glucose up to a concentration of 30 mmol/l (Fig. 3). In the serum analysis it was shown that the relative standard deviation
360
TABLE
I
COMPARISON ROUTINE Method
BETWEEN
THE
MASS
FRAGMENTOGRAPHIC
METHOD
THE
Slope
Intercept
Correlation
Relative
0.98
0.49
0.998
2.07
MID HK-N
AND
DIFFERENT
METHODS S.D.
(%)
1.93 ADPH
HK-INT
1.14
a.15
0.996
1.34
GO-DC
1.01
--0.26
0.998
2.92
GO-ORM
1.04
-3.09
0.998
2.49
DRM
0.99
0.988
3.01
0.86
of the method was 1.9% as calculated from 40 duplicate analyses. The accuracy of the method was tested by addition of 300 and 600 nmoles of glucose to a serum containing 211 nmoles of glucose. The result obtained with the MIDmethod was 510 nmoles (0% error) and 800 nmoles (1.2% error), respectively. The specificity of the method was tested by using the fragments at m/e 205 and m/e 208 in the analysis (cf. Fig. 1). Exactly the same results were obtained as when using the fragments at m/e 319 and m/e 323. Analysis of serum samples to which no deuterated glucose had been added gave the same ratio between the tracings at m/e 319 and m/e 323 as obtained in analysis of unlabeled glucose, indicating that no other serum constituents interfere significantly with the determination. Comparisons between the MID-method and five methods used in clinical chemical routine work is shown in Fig. 4. The constants of the regression lines as well as the relative standard deviation of the different methods are summarized in Table I. Discussion The very high specificity of the present mass-fragmentographic method for glucose determination in combination with the fact that the ratio between labeled and unlabeled molecules is determined with a high degree of accuracy, makes it likely that the most significant errors in the method are related to the errors in the pipettations. It should be pointed out that methoxime-trimethylsilyl derivatives of all other sugars which may be present in serum was separated from the derivative of glucose under the conditions employed. In view of the possibility that a hexokinase preparation may contain other enzymes it seems probable that the accuracy of the present method is higher than that of the hexokinase method, which previously has been used as reference method. The slope of the regression line in the comparison between the MID-method and the HK-NADPH method was very close to 1.00. A surprisingly high positive intercept was however found in repeated experiments and with use of different batches of the kit. In principle, part of the explanation could be that the enzyme preparation according to the specification contains trace amount of phosphohexosisomerase. In the presence of fructose, the activity of this enzyme should give higher values in the over-all determination. The concentration
361
of fructose in serum is however in general lower than the intercept obtained and addition of fructose to different glucose mixtures prior to the determination gave little or no increase in absorbance. The slope of the regression line in the comparison between the MID-method and the HK-INT method was significantly higher than 1.00. In addition a small negative intercept was found. This complex regression picture is difficult to explain. It should be pointed out that this method is the most complicated one from a chemical point of view with three coupled reactions. The slope of the regression line in the comparison between the MID-method and the GO-DC method was very close to 1.00. A small negative intercept was found. It is well known that hydrogen donors in serum may compete with the chromogen to reduce the hydrogen peroxide in glucose-oxidase chromogen methods with lower results as a consequence. It has been shown that 5-hydroxylated indoles may give rise to a constant as well as a systematic error in glucase-oxidase chromogen methods [13]. The negative intercept is in agreement with the results obtained by Lott and Turner [14] who compared a glucoseoxidase-chromogen method with GO-ORM. The slope of the regression line in the comparison between the MID-method and the GO-ORM method was very close to 1.00 and only a very small negative intercept was found. From this finding together with the finding of a correlation coefficient of 0.998 it can be concluded that the GO-ORM method is the method which gives results most similar to the results of the reference method in the range tested. The high specificity of the GO-ORM method must be related to the fact that the maximal reaction velocity is determined. In such a procedure a possible unspecificity of the enzyme preparation should influence the results to a smaller extent than in other procedures. Thus interfering compounds must react to a considerable extent within ten seconds in order to influence the results. In addition the method lacks a coupling chromogene which makes it less sensitive towards side-reactions. It may be expected that the DRM-method should be less accurate and specific than the other methods tested. This method is difficult to calibrate exactly and the accuracy as well as the precision is dependent upon the skill with which the laboratory procedure is performed. In spite of that the present determinations with the DRM-method were performed under optimal conditions, it is surprising that such high correlation with the reference method could be obtained. The slope of the regression line in the comparison with the MIDmethod was very close to 1.00. There was however a considerable positive intercept (0.86 mmol/l). To summarize, all the routine methods parallel well with the reference method and the small discrepancies found are of little or no importance from a clinical point of view under most conditions. In determinations of low glucose concentrations with the HK-NADPH and DRM-methods, the error in the methods may sometimes be significant. A consequence of the present study is that when a mass-fragmentographic method is not available, the GO-ORM method is more suitable as a reference method than the HK-NADPH method. Acknowledgments This work
was supported
by the Bank
of Sweden
Tercentenary
Fund.
362
References 1
Gochman.
2
Sweeley.
N.,
Ryan,
C.C.,
3
Laine,
R.A.
4
Kudla,
R.
5
Liljeqvist,
6
Handbook
7
Slein.
M.W.
(1965)
demic
Press.
New
and and L.,
Sweeley,
Giirtler.
Coburn.
H.J.
and
Stevens,
J.F.
(1971)
10
Kattermann. Strewhart, Biemann,
14
Lott,
New K. D.R.
d.A.
and
(1971)
Anal.
Tech.
and
Blomstrand,
and
Physics of
Widdowson,
Ryhage,
R.
G.M.
(1966)
Biochem.
Bull. R.
(1975)
Reg.
43. Med.
(1973) 56th
Enzymatic
edn.
38,
Chem.
p. C-311.
Analysis
Clin.
Chem.
38,
J.J.
(1973)
Chim.
Naroska,
(1931)
and
Acta
I. (1974)
Economic
in Mass Huggins.
Turner,
21.
356
1549
279 Sand.
CRC
(Bergmeyer,
27, Press, 11.-U.,
Clin. 32.
197 Cleveland. cd.),
pp.
Ohio 117-123,
Aca-
Chem.
19,
127
199
Dtsch. Control
Med. of
Wochenschr.
Quality
of
99,
2603
Manufactured
Y ark
(1962)
Chem.
533 Tech.
Acta
(1975)
Anal.
N.Y.
Clin.
and
and
I. and
(1968)
Methods
Carroll,
W.A.
Nostrand, Nelson,
R.
in York,
8
13
J.
R.E.
Fries,
C.C. D.
of Chemistry
11
Sterling,
W.H..
McVean,
9
12
W.T.,
Elliott,
K.
Spectromrtrv. A.K.
(1974)
(1975)
Clin.
p. 223. Anal. Chem.
McGraw-Hill,
Biochem. 21.
1754
59,
46
New
York
Products,
p.
104,
D.
van