Clinica Chimica
Acta, 98
@ Elsevier/North-Holland
(1979) 53-59 Biomedical Press
CCA 1130 A SEMI-AUTOMATED METHOD FOR THE DETERMINATION GLYCOSYLATED HAEMOGLOBIN
I.S. ROSS*
and
Department (Received
P.F.
GIBSON
of Chemical March
OF
26th,
Pathology,
University
of Aberdeen,
Foresterhill,
Aberdeen
(U.K.)
1979)
Summary A new routine, automated method for the determination of glycosylated haemoglobin is presented. The method offers improvements in analytical capacity, and precision, together with reduced running costs over existing methods, which should assist in making the determination of glycosylated haemoglobin more readily available to the clinician treating diabetes.
Introduction Much interest has been shown recently in determining glycosylated haemoglobin as a measure of long-term stability of a patient’s diabetic control. It is now generally accepted that this parameter is an accurate and consistent index of such control [l]. Glycosylated haemoglobins are a sub-grouping of haemoglobin A which appears as three fast moving fractions on separation of the haemoglobin by ion exchange chromatography. All these “fast” fractions (AlA, AlB, and AlC) have been noted to be elevated in diabetes. Only the structure of haemoglobin AlC is known in detail, being haemoglobin A with one terminal cu-aminoacid on the fl chain blocked by a hexose molecule. The hexose is bound initially by a Schiff base which mostly undergoes an amadori rearrangement to form a stable bond [ 21. Methods of measuring haemoglobin AlC itself or all fast haemoglobin fractions fall into several categories. By far the most commonly used method is the separation of haemoglobin AlC from the main haemoglobin A and if desired, haemoglobin AlA and AlB also, on ion exchange columns [3]. The initial method was very time consuming and many attempts have been made to simplify it. This is done either by * Correspondence sitv
of
Aberdeen.
should Foresterhill,
be
addressed Aberdeen.
to:
Dr. U.K.
Iain
S.
Ross,
Department
of
Chemical
Pathology,
Univrr-
miniaturising the columns used in the ion exchange syst,em [ 4) or by pressurising the system using the same ion exchange resin by either employing high pressure chromatography [ 51 or by centrifugation during the column elution step [ 6). Methods of measuring haemoglobin AlC by radioimmunoassay [7] and by electrofocussing [S] have also been reported. A fourth method shows a different approach in that haemoglobin AlC is quantitated by measuring the hexosc moiety. This method has previously been published as a manual procedure [ 9 1. the method being little modified from the original publication dealing with th(l detection of furfuraldehyde in milk products [lo], Ideally the method of choice in estimating the glycosylated haemoglohin fraction should be fast enough to supply the clinical need for this technique to be applied to diabetics attending on an out-patient basis, and yet it must be precise enough to be able to assess the stability of the patient’s control over a long period of time. In our hands neither the previously described chromatographic methods on ion exchange columns nor the chemical manual method [9] were particularly precise and most were very time consuming. An attempt has therefore been made by automation and miniaturisation to adapt the furfuraldehyde/thiobarbituric acid reaction to fulfil these stated criteria. The method is based upon the production of 5-hydroxymethyl/furfural (5HMF) from the l-amino, l-deoxy, fructose attached to the N-terminal valine of the fl chain of the haemoglobin AlC. 5HMF is produced by the digestion with 0.3 mol/l oxalic acid at 1OO’C for 2 h of the diluted blood sample. Furfural compounds react with 2-thiobarbituric acid to yield yellow pigments which absorb maximally at 440-445 nm. The 5HMF produced by digestion of the haemoglobin AlC is therefore reacted with thioharbituric acid and thr product read at 440 nm. The method described in this paper uses samples of red cells or haemolysate either obtained following capillary sampling for plasma glucose estimation or cells collected following centrifugation of venous specimens of blood preserved for plasma glucose assay in fluoride oxalate tubes. Materials and methods
Sample 100 ~1 of whole blood or equivalent of diluted red cells or haemolysate is mixed with 1.9 ml of 0.3 mol/l oxalic acid (B.D.H., Poole, Dorset). In order to assist work simplification and to maintain precision we used a diluter (Hook and Tucker Instruments, New Addington, England). The mixture is shaken. The test tube is capped tightly and placed in a 100°C water bath for 2 h. The samples when cool are loaded on to the autoanalyser.
Reagents for the autoanalyser thiobarbituric acid method globin determination by measurement of acid haematin
and for total haemo-
Poole, Dorset) reagent was prepared by (a) Thiobarbituric acid (B.D.H., adding 7 g thiobarbituric acid per litre of 0.3 mol/l oxalic acid and stirring at 35°C for 2 h. The reagent is then kept at 2’C overnight and filtered for use next
day. Before use the reagent has added to it 5 ml of 30% Brij 35 solution per litre. This reagent is stable for at least one week if kept at 2’C between autoanalyser runs. (b) Diluent for the acid haematin total haemoglobin determination is distilled water with 5 ml of 30% Brij 35 added per litre. Standardisation Standards are prepared by dissolving weighed amounts of haemoglobin powder (B.D.H. technical grade) in water. The haemoglobin dissolves slowly with stirring. The total haemoglobin is calculated from the initial weighing and haemoglobin is obtained by calibrating the the percentage of “glycosylated” prepared solution on an ion exchange column [3]. The standards are then divided into aliquots and stored at -18-C. Working standards are obtained each day by gently thawing out a set of the pre-prepared frozen serially diluted haemoglobin solutions of known total haemoglobin and haemoglobin AlC content. Standardisation in this way has been found to remain stable for at least one month. This dried haemoglobin preparation is of bovine origin, the “glycosylated” fraction elutes in the position of HbAlA and AlB on column chromatography. The digestion characteristics of this preparation are in our experience reproducible and are indistinguishable in the calorimetric method from serially diluted human diabetic blood samples. The preparation is not contaminated with fructose. This increased “glycosylated” fraction in dried bovine haemoglobin was found to be present in very similar amounts in six different batches of the product obtained from the manufacturer. Work is proceeding to identify the precise nature of this haemoglobin, but in the meantime we have found its use to be a practical approach to the problem of standardisation of the haemoglobin AlC method. An alternative approach could be by preparing, using column chromatography, fractions of human haemoglobin AlC in order to increase the percentage of this fraction in a human haemoglobin pool. We have as yet found no reason to doubt the use of dried bovine haemoglobin standard. Six working standards are prepared from the bovine haemoglobin pool and give a slightly curved calibration line. Human blood specimens, both normal and diabetic, when run as serial dilutions also demonstrate the non-linear colour production in a parallel fashion to that seen with the bovine standards. It has not proved possible to use simple fructose solutions to calibrate this method as the colour development of the fructose solution was found to be linear over the range of concentration required. Results
Automation Fig. 1 shows the flow diagram for the automated part of the method. pump and calorimeter used on the glycosylated haemoglobin manifold purchased from Technicon Instruments Ltd., Basingstoke, Hampshire. sampler, dialyser and calorimeter used on the total haemoglobin manifold
The were The were
purchased at a speed Measurement
from Chemlab Instruments of 45/h with a l/l sample
Ltd., Ilford, wash ratio.
Essex.
The sampler
was run
of total haemoglobin
Incubation of the blood sample with 0.3 mol/l oxalic acid converts the haemoglobin to acid haematin. Provided the concentration of haemoglobin in the digestion mixtures does not exceed 20 g/l, the resultant solution remains optically clear and the method can be used to provide an acid haematin total haemoglobin estimation using the haemoglobin standards described above. The manifold as shown in Fig. 1 is constructed simply to dilute the sample of acid haematin to a suitable optical density. The sample when aspirated is diluted in two steps in order to preserve the bubble pattern to ensure adequal~~ mixing in the one single coil and one double mixing coil. The total dilution involved is 1 in 32. The diluted brown acid haematin is measured in thr) calorimeter at 550 nm. Between-batch precision of the total haemoglobin part of the method, commencing with the dilution of the original sample, may IX> exrlressed as the standard deviation of the values obtained for a sample analysed daily for 20 working days [ 111. The mean value was 13.4 g Hb/l, S.D. 0.51. Carry over was 1.4% [ll]. Measurement
of glycosylated
haemoglobin
The dried haemoglobin used as a standard contains approximately 30% “fast” fraction. With the total haemoglobin standard or 20 g/l, the glycosylated haemoglobin top standard is therefore approximately 7 g/l when calibrated by ion exchange chromatography [ 41. The sample is aspirated (0.6 ml), air-segmented, and dialysed using a 24 inch dialyser with a ‘C’ type membrane, into the thiobarhituric acid/oxalic acid recipient stream. Colour is developed at room temperature in a double mixing coil and read at 440 nm in a S.C.I.C. calorimeter (Technicon Instruments). For the complete measurement of glycosylated haemoglobin including the
initial dilution of the whole blood, between-batch precision of a specimen on 20 working days gives a mean value of 3.41 g/l, S.D. 0.095. Carry over was 3.7% [ll]. Results are normally expressed as the percentage of the glycosylated haemoglobin in the total haemoglobin measured in that specimen. The overall between-batch precision on 20 working days with a mean value of 24.6% gives a S.D. = 0.47. Carry over was 4% 111. It is interesting to note that the results as a of total haemoglohin essentially having two parts, a manipulative step before and an automated reaction and colorimet,ry. imprecision occurs in the handling of the sample before the automated calorimetry. During this part of t,he method any variation in the handling, diluting etc. will tend to he compensated haemoglobin obtained on the same sample, and, consequently, expressing the results related to that total haemoglobin, the precision would be improved. The results of the carryover experiments improvement in when the results are expressed as a of total haemoglobin, as one would expect; a slight worsening is in fact seen.
Specificity
of the glycosylatecl
The chemical to be a problem.
haemoglohin
method
reactions used are not specific. Fig. 2 shows other carbohydrates
In practice, this was not found tested for interference; only
c
B
I
. . . .. .. . .
. . . . ::
.. . . :
-2 . .. :
I :::::::* .:;.
I Fig.
2. Tracing
Fig.
3.
diabetic;
of
autoanalyser
Percentage (R)
diabetics
of
chart
glycosylated treated
with
of
possible
interfering
haemoglobin oral
hypoglycaemic
found
substances in
three
agents;
and
in the
I thiobarbitunc
groups
of
(C)
diabetics
hospital
acid
in-patients:
treated
with
method. c.4)
insulin.
nnn-
.
. .
’
. .
fructose gives a measurable peak. This reaction does not appear to bias the results obtained on patients. Fructose solutions can, in fact. be used as a quality control material. Fig. 3 shows the results found in 123 hospital inpatients. both in diabetics on oral hypoglycaemic agents and insulin and in nondiabetic patients. admitted for reasons unrelated to metabolic upset.. This figure shows a pattern of results very similar to that found by methods based on ion exchange chromatography [ 51. Comparison
of two
methods
of measuring
glycosylated
haernoglohill
Specimens were obtained from patients attending the diabetic clinic in Aberdeen and also from healthy volunteers. The specimens were analysed both by the chemical method described herein, and by small commercial ion exchange columns, based on the Abraham method [5] (BioRad Laboratories. Watford, Hertfordshire). Results are presented in Fig. 3. Discussion This semi-automated chemical method for estimating glycosylated haemoglohin offers an alternative to the existing ion exchange method. Chemical costs amount to a few pence per sample as compared with 60~ to fl for each of the commercially available disposable ion exchange columns. This very large saving in cost makes it economically possible to offer glycosylated haemoglohin estimations to all patients at diabetic clinics. The equipment used is of a type normally available in most hospital routine laboratories and the method has, in our hands, proved trouble-free over many months. A technician can process up to 120 samples on an average day. As the blood sample appears to be stable
when stored at 2’C for a few days (it has been claimed that stability extends for some weeks) 141, it is possible to run the method with batches of stored specimens in situations which do not require results on the same day. The precision of the chemical method has, in our hands, proved to be acceptable. The patients results obtained by this method follow the same pattern as has been reported in studies using ion exchange chromatography, therefore any lack of chemical specificity does not appear to be a practical problem. Correlation between results obtained by ion exchange chromatography and by the semi-automated chemical method now described has also been shown to he satisfactory. This method with its small sample requirement, inexpensive reagent consumption and good precision, should, therefore, make the clinical use of glycosylated haemoglobin more generally available. Acknowledgements We are grateful to the staff and patients of the Diabetic Clinic, Aberdeen Royal Infirmary, for providing many of the blood specimens used in this study, and to Miss Janet Drinkwater, Miss Mabel Morrison and Mr. Alan Gauld, for their technical assistance. References 1
Ditzel,
J. and
2
Bunn.
H.F..
Kjaergaard. Haney,
J.J. (1978)
D.N.,
Brit.
Gabbay,
K.H.
Med. and
J. 1, 741-742 Gallop,
P.M.
(1975)
Biochcm.
Biophys.
Res.
Commun.
67.103-109 3
Trivelli.
4
Abraham.
L.A..
E.C..
Ranney.
Diabetes
27,931-937
5
Davis,
J.E.,
McDonald,
6
Davis.
R.E.
and
J., Pettis,
7
Javid,
8
Schoos.
9
Fliickiger,
R..
10
Keeney,
11
Broughton. 11.
K.,
and
T.A..
D.J.
Koenig.
R.J.
and
R.
(1959)
Engl.
J.B..
J. Med.
284,
Bransome,
Diabetes
27,
353-357
E.D.
and
Iluisman,
T.H.J.
(1978)
102-107
2,350-351
Cerami. (1976) J. Dairy
A.H.,
N.
L. (1978) A.
Lambotte,
K.H.
Gowenlock.
(1971) Wilson,
Lancet
S. and
Winterhalter.
Bassette,
P.M.G..
H.T. N.D.,
and Jarrett. (1978)
Schons-Barbctte.
M. and
Lai.
Cope.
J.M.
Nicol.
R. and
207-218
H.M.
Huff,
(1978)
Brit.
C. (1978) FEBS Sci.
McCormack.
Lett. 42.
J. Haematol.
Clin. 71,
Chim.
Acta
38. 86.
329-337 6145
365-360
945-960 J.J.
and
Ncill,
D.W.
(1974)
Ann.
Clin.
Biochcm.
Acta, 98
@ Elsevier/North-Holland
(1979) 53-59 Biomedical Press
CCA 1130 A SEMI-AUTOMATED METHOD FOR THE DETERMINATION GLYCOSYLATED HAEMOGLOBIN
I.S. ROSS*
and
Department (Received
P.F.
GIBSON
of Chemical March
OF
26th,
Pathology,
University
of Aberdeen,
Foresterhill,
Aberdeen
(U.K.)
1979)
Summary A new routine, automated method for the determination of glycosylated haemoglobin is presented. The method offers improvements in analytical capacity, and precision, together with reduced running costs over existing methods, which should assist in making the determination of glycosylated haemoglobin more readily available to the clinician treating diabetes.
Introduction Much interest has been shown recently in determining glycosylated haemoglobin as a measure of long-term stability of a patient’s diabetic control. It is now generally accepted that this parameter is an accurate and consistent index of such control [l]. Glycosylated haemoglobins are a sub-grouping of haemoglobin A which appears as three fast moving fractions on separation of the haemoglobin by ion exchange chromatography. All these “fast” fractions (AlA, AlB, and AlC) have been noted to be elevated in diabetes. Only the structure of haemoglobin AlC is known in detail, being haemoglobin A with one terminal cu-aminoacid on the fl chain blocked by a hexose molecule. The hexose is bound initially by a Schiff base which mostly undergoes an amadori rearrangement to form a stable bond [ 21. Methods of measuring haemoglobin AlC itself or all fast haemoglobin fractions fall into several categories. By far the most commonly used method is the separation of haemoglobin AlC from the main haemoglobin A and if desired, haemoglobin AlA and AlB also, on ion exchange columns [3]. The initial method was very time consuming and many attempts have been made to simplify it. This is done either by * Correspondence sitv
of
Aberdeen.
should Foresterhill,
be
addressed Aberdeen.
to:
Dr. U.K.
Iain
S.
Ross,
Department
of
Chemical
Pathology,
Univrr-
miniaturising the columns used in the ion exchange syst,em [ 4) or by pressurising the system using the same ion exchange resin by either employing high pressure chromatography [ 51 or by centrifugation during the column elution step [ 6). Methods of measuring haemoglobin AlC by radioimmunoassay [7] and by electrofocussing [S] have also been reported. A fourth method shows a different approach in that haemoglobin AlC is quantitated by measuring the hexosc moiety. This method has previously been published as a manual procedure [ 9 1. the method being little modified from the original publication dealing with th(l detection of furfuraldehyde in milk products [lo], Ideally the method of choice in estimating the glycosylated haemoglohin fraction should be fast enough to supply the clinical need for this technique to be applied to diabetics attending on an out-patient basis, and yet it must be precise enough to be able to assess the stability of the patient’s control over a long period of time. In our hands neither the previously described chromatographic methods on ion exchange columns nor the chemical manual method [9] were particularly precise and most were very time consuming. An attempt has therefore been made by automation and miniaturisation to adapt the furfuraldehyde/thiobarbituric acid reaction to fulfil these stated criteria. The method is based upon the production of 5-hydroxymethyl/furfural (5HMF) from the l-amino, l-deoxy, fructose attached to the N-terminal valine of the fl chain of the haemoglobin AlC. 5HMF is produced by the digestion with 0.3 mol/l oxalic acid at 1OO’C for 2 h of the diluted blood sample. Furfural compounds react with 2-thiobarbituric acid to yield yellow pigments which absorb maximally at 440-445 nm. The 5HMF produced by digestion of the haemoglobin AlC is therefore reacted with thioharbituric acid and thr product read at 440 nm. The method described in this paper uses samples of red cells or haemolysate either obtained following capillary sampling for plasma glucose estimation or cells collected following centrifugation of venous specimens of blood preserved for plasma glucose assay in fluoride oxalate tubes. Materials and methods
Sample 100 ~1 of whole blood or equivalent of diluted red cells or haemolysate is mixed with 1.9 ml of 0.3 mol/l oxalic acid (B.D.H., Poole, Dorset). In order to assist work simplification and to maintain precision we used a diluter (Hook and Tucker Instruments, New Addington, England). The mixture is shaken. The test tube is capped tightly and placed in a 100°C water bath for 2 h. The samples when cool are loaded on to the autoanalyser.
Reagents for the autoanalyser thiobarbituric acid method globin determination by measurement of acid haematin
and for total haemo-
Poole, Dorset) reagent was prepared by (a) Thiobarbituric acid (B.D.H., adding 7 g thiobarbituric acid per litre of 0.3 mol/l oxalic acid and stirring at 35°C for 2 h. The reagent is then kept at 2’C overnight and filtered for use next
day. Before use the reagent has added to it 5 ml of 30% Brij 35 solution per litre. This reagent is stable for at least one week if kept at 2’C between autoanalyser runs. (b) Diluent for the acid haematin total haemoglobin determination is distilled water with 5 ml of 30% Brij 35 added per litre. Standardisation Standards are prepared by dissolving weighed amounts of haemoglobin powder (B.D.H. technical grade) in water. The haemoglobin dissolves slowly with stirring. The total haemoglobin is calculated from the initial weighing and haemoglobin is obtained by calibrating the the percentage of “glycosylated” prepared solution on an ion exchange column [3]. The standards are then divided into aliquots and stored at -18-C. Working standards are obtained each day by gently thawing out a set of the pre-prepared frozen serially diluted haemoglobin solutions of known total haemoglobin and haemoglobin AlC content. Standardisation in this way has been found to remain stable for at least one month. This dried haemoglobin preparation is of bovine origin, the “glycosylated” fraction elutes in the position of HbAlA and AlB on column chromatography. The digestion characteristics of this preparation are in our experience reproducible and are indistinguishable in the calorimetric method from serially diluted human diabetic blood samples. The preparation is not contaminated with fructose. This increased “glycosylated” fraction in dried bovine haemoglobin was found to be present in very similar amounts in six different batches of the product obtained from the manufacturer. Work is proceeding to identify the precise nature of this haemoglobin, but in the meantime we have found its use to be a practical approach to the problem of standardisation of the haemoglobin AlC method. An alternative approach could be by preparing, using column chromatography, fractions of human haemoglobin AlC in order to increase the percentage of this fraction in a human haemoglobin pool. We have as yet found no reason to doubt the use of dried bovine haemoglobin standard. Six working standards are prepared from the bovine haemoglobin pool and give a slightly curved calibration line. Human blood specimens, both normal and diabetic, when run as serial dilutions also demonstrate the non-linear colour production in a parallel fashion to that seen with the bovine standards. It has not proved possible to use simple fructose solutions to calibrate this method as the colour development of the fructose solution was found to be linear over the range of concentration required. Results
Automation Fig. 1 shows the flow diagram for the automated part of the method. pump and calorimeter used on the glycosylated haemoglobin manifold purchased from Technicon Instruments Ltd., Basingstoke, Hampshire. sampler, dialyser and calorimeter used on the total haemoglobin manifold
The were The were
purchased at a speed Measurement
from Chemlab Instruments of 45/h with a l/l sample
Ltd., Ilford, wash ratio.
Essex.
The sampler
was run
of total haemoglobin
Incubation of the blood sample with 0.3 mol/l oxalic acid converts the haemoglobin to acid haematin. Provided the concentration of haemoglobin in the digestion mixtures does not exceed 20 g/l, the resultant solution remains optically clear and the method can be used to provide an acid haematin total haemoglobin estimation using the haemoglobin standards described above. The manifold as shown in Fig. 1 is constructed simply to dilute the sample of acid haematin to a suitable optical density. The sample when aspirated is diluted in two steps in order to preserve the bubble pattern to ensure adequal~~ mixing in the one single coil and one double mixing coil. The total dilution involved is 1 in 32. The diluted brown acid haematin is measured in thr) calorimeter at 550 nm. Between-batch precision of the total haemoglobin part of the method, commencing with the dilution of the original sample, may IX> exrlressed as the standard deviation of the values obtained for a sample analysed daily for 20 working days [ 111. The mean value was 13.4 g Hb/l, S.D. 0.51. Carry over was 1.4% [ll]. Measurement
of glycosylated
haemoglobin
The dried haemoglobin used as a standard contains approximately 30% “fast” fraction. With the total haemoglobin standard or 20 g/l, the glycosylated haemoglobin top standard is therefore approximately 7 g/l when calibrated by ion exchange chromatography [ 41. The sample is aspirated (0.6 ml), air-segmented, and dialysed using a 24 inch dialyser with a ‘C’ type membrane, into the thiobarhituric acid/oxalic acid recipient stream. Colour is developed at room temperature in a double mixing coil and read at 440 nm in a S.C.I.C. calorimeter (Technicon Instruments). For the complete measurement of glycosylated haemoglobin including the
initial dilution of the whole blood, between-batch precision of a specimen on 20 working days gives a mean value of 3.41 g/l, S.D. 0.095. Carry over was 3.7% [ll]. Results are normally expressed as the percentage of the glycosylated haemoglobin in the total haemoglobin measured in that specimen. The overall between-batch precision on 20 working days with a mean value of 24.6% gives a S.D. = 0.47. Carry over was 4% 111. It is interesting to note that the results as a of total haemoglohin essentially having two parts, a manipulative step before and an automated reaction and colorimet,ry. imprecision occurs in the handling of the sample before the automated calorimetry. During this part of t,he method any variation in the handling, diluting etc. will tend to he compensated haemoglobin obtained on the same sample, and, consequently, expressing the results related to that total haemoglobin, the precision would be improved. The results of the carryover experiments improvement in when the results are expressed as a of total haemoglobin, as one would expect; a slight worsening is in fact seen.
Specificity
of the glycosylatecl
The chemical to be a problem.
haemoglohin
method
reactions used are not specific. Fig. 2 shows other carbohydrates
In practice, this was not found tested for interference; only
c
B
I
. . . .. .. . .
. . . . ::
.. . . :
-2 . .. :
I :::::::* .:;.
I Fig.
2. Tracing
Fig.
3.
diabetic;
of
autoanalyser
Percentage (R)
diabetics
of
chart
glycosylated treated
with
of
possible
interfering
haemoglobin oral
hypoglycaemic
found
substances in
three
agents;
and
in the
I thiobarbitunc
groups
of
(C)
diabetics
hospital
acid
in-patients:
treated
with
method. c.4)
insulin.
nnn-
.
. .
’
. .
fructose gives a measurable peak. This reaction does not appear to bias the results obtained on patients. Fructose solutions can, in fact. be used as a quality control material. Fig. 3 shows the results found in 123 hospital inpatients. both in diabetics on oral hypoglycaemic agents and insulin and in nondiabetic patients. admitted for reasons unrelated to metabolic upset.. This figure shows a pattern of results very similar to that found by methods based on ion exchange chromatography [ 51. Comparison
of two
methods
of measuring
glycosylated
haernoglohill
Specimens were obtained from patients attending the diabetic clinic in Aberdeen and also from healthy volunteers. The specimens were analysed both by the chemical method described herein, and by small commercial ion exchange columns, based on the Abraham method [5] (BioRad Laboratories. Watford, Hertfordshire). Results are presented in Fig. 3. Discussion This semi-automated chemical method for estimating glycosylated haemoglohin offers an alternative to the existing ion exchange method. Chemical costs amount to a few pence per sample as compared with 60~ to fl for each of the commercially available disposable ion exchange columns. This very large saving in cost makes it economically possible to offer glycosylated haemoglohin estimations to all patients at diabetic clinics. The equipment used is of a type normally available in most hospital routine laboratories and the method has, in our hands, proved trouble-free over many months. A technician can process up to 120 samples on an average day. As the blood sample appears to be stable
when stored at 2’C for a few days (it has been claimed that stability extends for some weeks) 141, it is possible to run the method with batches of stored specimens in situations which do not require results on the same day. The precision of the chemical method has, in our hands, proved to be acceptable. The patients results obtained by this method follow the same pattern as has been reported in studies using ion exchange chromatography, therefore any lack of chemical specificity does not appear to be a practical problem. Correlation between results obtained by ion exchange chromatography and by the semi-automated chemical method now described has also been shown to he satisfactory. This method with its small sample requirement, inexpensive reagent consumption and good precision, should, therefore, make the clinical use of glycosylated haemoglobin more generally available. Acknowledgements We are grateful to the staff and patients of the Diabetic Clinic, Aberdeen Royal Infirmary, for providing many of the blood specimens used in this study, and to Miss Janet Drinkwater, Miss Mabel Morrison and Mr. Alan Gauld, for their technical assistance. References 1
Ditzel,
J. and
2
Bunn.
H.F..
Kjaergaard. Haney,
J.J. (1978)
D.N.,
Brit.
Gabbay,
K.H.
Med. and
J. 1, 741-742 Gallop,
P.M.
(1975)
Biochcm.
Biophys.
Res.
Commun.
67.103-109 3
Trivelli.
4
Abraham.
L.A..
E.C..
Ranney.
Diabetes
27,931-937
5
Davis,
J.E.,
McDonald,
6
Davis.
R.E.
and
J., Pettis,
7
Javid,
8
Schoos.
9
Fliickiger,
R..
10
Keeney,
11
Broughton. 11.
K.,
and
T.A..
D.J.
Koenig.
R.J.
and
R.
(1959)
Engl.
J.B..
J. Med.
284,
Bransome,
Diabetes
27,
353-357
E.D.
and
Iluisman,
T.H.J.
(1978)
102-107
2,350-351
Cerami. (1976) J. Dairy
A.H.,
N.
L. (1978) A.
Lambotte,
K.H.
Gowenlock.
(1971) Wilson,
Lancet
S. and
Winterhalter.
Bassette,
P.M.G..
H.T. N.D.,
and Jarrett. (1978)
Schons-Barbctte.
M. and
Lai.
Cope.
J.M.
Nicol.
R. and
207-218
H.M.
Huff,
(1978)
Brit.
C. (1978) FEBS Sci.
McCormack.
Lett. 42.
J. Haematol.
Clin. 71,
Chim.
Acta
38. 86.
329-337 6145
365-360
945-960 J.J.
and
Ncill,
D.W.
(1974)
Ann.
Clin.
Biochcm.
