Jul 31, 1984 - Boulton FE, Huntsman. RG. The detection of myoglobin in urine and its distinction from normal and variant haemoglobins.J Clin. Pat/wi 24 ...
and hemoglobin, then analyzed with the Centricon-30 microconcentrator.
References 1. Elek SD, Anderson HF. Paroxysmal paralytic myoglobinuria. Br Med J ii, 533-536 (1953).
2. Boulton FE, Huntsman RG. The detection of myoglobin in urine and its distinction from normal and variant haemoglobins. J Clin Pat/wi 24, 816-824 (1971). 3. Farmer TA, Hammack WJ, Frornmeyer WB. Idiopathic recurrent rhabdomyolysis associated with myoglobinuria. N Engi J Med 264, 60-66 (1961).
4. Fletcher UD, Prankerd TM. Paroxysmal myoglobinuria. Lancet i, 1072 (1955). 5. Prankerd TM. Electrophoretic properties of myoglobinand its character in sickle-cell diseaseand paroxysmal myoglobinuria. BrJ Haernatoi ii, 80-83 (1956). 6. Adams EC. Differentiation of myoglobin and hemoglobin in biological fluids. Ann C/in Lab Sci 1, 208-221 (1971). 7. Blondheim SH, Margoliash E, Shafrir EA. A simple test for myohemoglobinuria (myoglobinuria). J Am Med Assoc 167, 453454 (1958). 8. Theil GB. Separation and identification of myoglobin and hemoglobin. Am J Clin Pat/wi 49, 190-195 (1968). 9. Duma RJ, Trig JW, Hammack WJ. Primary myoglobinura. Ann Intern Med 56, 97-404 (1962).
CLIN. CHEM. 31/1, 114-117 (1985)
Effect of Temperature on Quantifying Glycated (Glycosylated) Hemoglobin by Cation-Exchange Chromatography Rudolf FiUckigerand Thomas Woodtii As a consequence of nonideal chromatographic conditions, temperature more than doubles precision and reduces imprecision (CVs) to well below 5% (3, 4), accuracy is still values for stable glycated hemoglobin (HbA1) determined by inadequate, and results for the same analyte determined in cation-exchange chromatography in a commercial minicodifferent assay systems are rarely directly comparable (9, Iumn system or by “high-performance” liquid chromatog10). This has hampered the interpretation of data, and in raphy (x) differ markedly, yielding the regression line y = some instances has even led to discontinuation of this 0.82x + 0.6. With use of the protocol specified by the clinically useful determination (11). manufacturer, 20% of the HbA1 peak is not collected in the In this investigation we clarify the reasons for the lack of HbAIC fraction. Increasing the ionic strength of the eluting buffer by increasing the operating temperature to 28 #{176}Caccuracy. When the protocol specified by the manufacturer is used, the elution profile is such that a substantial part of increases the rate of elution from the minicolumn, making the HbA1 is not collected in the HbA1 fraction, and so the results of the two methods more closely comparable (y = observed values are lower than they should be. Increasing 0.98x - 0.22). Because at a given pH the elution volume is the temperature is the simplest of several ways to increase determined primarily by the ionic strength, close limits on the the rate of elution. composition of the eluting buffer are set by the temperatureThe temperature-dependence of the ionic strength of the dependence of its ionic strength. At a specified temperature eluent and the elution time of the glycated hemoglobin set and pH the position of a peak can be judged to within a close limits for the composition of the buffer. volume of 1 mL if the conductivity of the eluent does not vary Materials and Methods by more than ± 0.05 mS.
(
Glycated hemoglobin (1) is an objective index of an individual’s glycemic control during the four to six weeks that preceded the blood sampling (2). The most widely used methodology for estimating glycated hemoglobin is measurement of the glycated hemoglobin components HbA1 and HbA1 by cation-exchange chromatography. Such determinations are often performed with commercial minicolumns because of their simplicity. However, these separations are extremely sensitive to variations in temperature, ionic strength, and pH of the eluent, and users of the minicolumn systems have been urged to implement strict temperature control (3-5) and use calibrators (6) or nomograms (7, 8). Although
operating
minicolumn
systems
at a constant
Department of Research and Internal Medicine, University Clinics, Kantonsspital, Basel, Switzerland. Address for correspondence: Zentrum f#{252}r Lehre und Forschung, Diabetologie, Hebelstrasse 20, CH-4031 Basel, Switzerland. Received July 31, 1984; accepted October 4, 1984.
114 CLINICALCHEMISTRY, Vol. 31, No. 1, 1985
Materials NaH2PO4 HO, Na2HPO4, KCN (all analytical grade), and Triton X-100 detergent were obtained from Merck, Darmstadt, F.R.G. The BioRex-70 resin (400 mesh) and the column test for the determination of HbA1 were products of Bio-Rad Labs, Richmond, CA 94804. Blood samples were obtained by venipuncture from normo- and hyperglycemic persons and anticoagulated with EDTA. Buffer solutions. We prepared various buffer solutions by mixing 0.2 mollL mono- and dibasic sodium phosphate stock solutions and diluting with distilled water. Use of the relative volumes listed in Table 1 makes it unnecessary to adjust pH or ionic strength. Results obtained with these buffers are identical to those previously reported. The mean proportion of stable HbAj measured for normoglycemic persons was 4.9% (SD 0.4%). .
Procedures “High-performance”
liquid
chromatography
(HPLC).
This
Table 1. Preparation of HPLC Buffers 2nd step: dilute let step: mix vol. of Buffer Hemolyzing Diluent Low-phosphate (pH 6.73; 5.2 mS) High-phosphate (pH 6.4; 11.2 mS)
Na,HPO4
NaH2PO4 1
3rd
Vol.
Vol.
step:
of
of
add
mixture
1
H20 1 mg/mL 3 Triton X-1 00
KCN
2 2
3 3
1 1
3 3
-
3
7
10
3
-
hemolysate with 10 mL of the second elutiorildeveloping reagent. We then determined the hemoglobin concentration spectrophotometrically by its absorbance at 415 nm, and calculated the percentage of HbAla+b and HbA1 in the eluted fractions. Data analysis. Precision was estimated from duplicate determinations by calculating the percent deviation of the mean value. Results of the method comparison were statistically evaluated according to the method of Passing and Bablock (14) by a computer program written in BASIC and run on an Apple H computer. This method involves no assumptions as to the distributional properties of the experimental data.
Results
was done as described previously (12) except that the system Figure 1 shows the temperature dependence of HbA1 was now fully automated, and we used a cyanide-free lowvalues as determined by HPLC and by a minicolumn phosphate eluent and a simplified procedure for sample system. The former values remained constant over the preparation. In brief, the automated HPLC system consisted temperature range from 20 to 34 #{176}C, whereas the latter of a Du Pont automatic sampler (Model 834), a Valco AH 45 values increased with temperature. At the “optimal” operatvalve with a 10-L sample loop, a three-way valve adapted ing temperature (24 #{176}C) of the minicolumns, HbA1 values for external control by Paul Bucher (Basel), a Waters M-45 were approximately 20% lower than the HPLC-values, solvent-delivery system, a Bioanalytical Systems column whereas results were identical at 28-30 #{176}C. When we deterheater, a Uvicon LCD 725 spectrophotometer (Kontron AG, mined elution profiles at temperature increments of 2 #{176}C, we Zurich, Switzerland), and a Chromatopac C-R2A data profound that the peak elution volume for HbA1 increased by cessor (Shimadzu Corp., Kyoto, Japan). The 4.5 x 125 mm 0.9 mLf#{176}C (Table 2). Figure 2 shows the elution profiles of a stainless-steel column, adapted by our technical service, is blood sample from a normo- and a hyperglycemic person at maintained at 28 #{176}C and used at a flow rate of 0.7 mL/min. the “optimal” operating temperature of 24 #{176}C, and at 28#{176}C. The effluent is monitored at 415 nm and the peak areas are At 24 #{176}C, collection of a 4-mL fraction for HbAla+b and a 10integrated on the basis of a horizontal baseline. About 10 zg mL fraction for HbA1 is clearly not appropriate for quantiof hemoglobin is injected, and the HbA1 components are tative recovery of HbA1. An amount corresponding to eluted with a low-phosphate eluent (see Table 1). The highapproximately 20% of the HbA1 peak is missed, and the phosphate eluent is applied after 6 mm and column recollected fraction for HbA1 includes 12% of HbAla+b mateequilibration for 10 mm is started 35 mm after sample rial. At 28 “C, the entire HbA1 peak is collected within the injection. manufacturer-specified fraction volumes. Sample preparation: Blood was drawn into 50-zL heparinized micro-hematocrit tubes (Clay Adams, Div. of Becton 10 Dickinson, Parsippany, NJ) and the erythrocytes were sedi#{149}_#{149} S merited by centrifugation for 5 mm in a hematocrit centrifuge. The segment of the capillary containing the erythro8 cytes was cut open with an ampoule opener and transferred . S into a 1.5-mL Eppendorf vial containing 200 iL of hemolyzS 6 ing solution (Table 1). These vials were then capped and vortex-mixed for 20 mm, after which the hemolysates were diluted with 1.3 mL of diluent, ifitered through 0.2-gm 4 (pore-size) ifiters (MF 1; Bioanalytical Systems, Inc., Lafayette, IN), and analyzed for HbA1. a, Minicolumn chromatography. We used minicolumns from HPLC (U Bio-Rad Inc. to determine stable HbA1, exactly following the manufacturer’s instructions. The columns were mountIll II III II Ill. ed in the temperature-controlled multi-column rack from S Bio-Rad, and the temperature was controlled to ±0.1 #{176}C with a circulating water bath equipped with a cooling aggregate (Digital Variostat; P. Huber, Offenburg, F.R.G.). We prepared hemolysate by mixing 100 iiL of blood with 500 iL of the commercial hemolysis reagent.’ We then applied 100 L of this to each column and let it enter the resin bed. After adding 4 mL of the first commercial 4 4.9’’dHPLC elutionldeveloping reagent-carefully, to avoid disturbing the resin-we collected the eluent containing the HbAl0+b. 2 We then eluted the HbA1,, fraction with 10 mL of the second Minicolumn commercial elutiorildeveloping reagent. The total amount of I I I I I I I I material applied was determined by diluting 20 L of 4(I,
20 ‘Removal of labile HbA,, by this hemolysis reagent (0.33% polyoxyethylene ether in a borate buffer, pH 4.7) has been proposed to occur by complex formation between borate and glucose. Note, however, that borate cannot form stable complexes in acid and that Schiff’s bases are rapidly dissociated by acid catalysis alone (12).
22
24
26
28
30
32
34
TEMPERATURE (#{176}C) Fig. 1. Temperaturedependence of stableHbA1CdeterminedbyHPLC and minicolumn chromatographyfor four differenthemolysates TheHPLC results are also shown for the twosamplesdetermined byminicolumn chromatography CLINICAL CHEMISTRY, Vol. 31, No.1, 1985
115
Table 2. Temperature Dependence of Conductivity and Elution Volume in the Mlnlcoiumn System ConductivIty.mS Buffer 1 Temp, C
(pH 6.83)
20 24 28 30
2.57 2.60 2.79 2.93
apH of Buffer1
and
Peak
Buffer 2 (pH 6.73) 5.57 5.79
elution vol, mL
6.14 6.36
7-8 6
15 10-11
2 measuredat 24#{176}C.
0.3
02
4
6
8
10
HbAic (stabte)- HPLC Fig. 3. Comparison of HbA1C determinedby chromatographyon minicolumns at 28#{176}C and by HPLC Top solid line, regression linefor data points shown;bottom solid line, line resultingwith use of minicolumns at 24#{176}C; line of identity
01
E C
(-
IJ
L&J U
z
03
0.2
0,1
0
2
4
6
8
10
12
14
16
18
FRACTION NUMBER Fig. 2. Elutionprofilesfrom minicolumnsat 24 and 28#{176}C for samples from a normoglycemic (-) and a hyperglycemic(- - -) subject The vertical lines indicatethe fractionscollectedandthe shadedareathe HbA1 not collected
A comparison of results obtained by minicolumn chromatography at 28 #{176}C and by HPLC indicated that the regression line (y = 0.98x 0.22; n = 28) is not substantially different from the line of identity (Figure 3). The Pearson correlation coefficient (r) for these data was 0.93. Performing the analyses at 28#{176}C did not alter the CV, which was 2.5% at 24#{176}C and 2.7% at 28 #{176}C. Furthermore, if minicolumn analyses were performed either at 28 “C or at 24#{176}C with use of buffers with a correspondingly higher ionic strength (Table 2), results were found to be similar. -
Discussion are commonly found in values for HbA1 as determined with different chromatographic
Discrepancies
and HbA1 116
CLINICAL CHEMISTRY, Vol. 31, No. 1, 1985
-
-)
systems. This has produced difficulties in the interpretation of this clinically important parameter. We recently defined a nomogram relating HbA1 and HbA1 values from different assay systems with the averaged blood glucose concentration (15). The most striking finding was that even if identical determinations were made with different assay systems, results differed. From our results it is clear that inappropriate “optimal” conditions for chromatography are the cause for the observed lower values in the minicolunin system used for the measurement of HbA1. With the protocol specified by the manufacturer, the HbA1 peak is only partly collected, resulting in an underestimate of its magnitude (Figure 2). The shift towards smaller elution volumes at higher temperatures explains the marked temperature-dependence of the minicolumn systems in which the HbA1 components are collected in a specified volume. In the HPLC-reference method, peak areas are integrated and elution time is not critical. In ion-exchange chromatography, the elution volume is determined by the pH and ionic strength of the eluent. Elution time/volume decreases with decreasing pH and decreases with increasing ionic strength. As indicated by our HPLC separations, the HbA1 components are satisfactorily resolved over a wide pH range if the ionic strength is adjusted accordingly. In the minmcolumn separations, the effect of temperature on elution seems to be mediated primarily by the changes in ionic strength of the buffer, because the pH of the phosphate buffers varies only slightly with temperature, 0.005 pH unit/#{176}C (8). This view is supported by the fact that similar results were obtained irrespective of whether the temperature was increased or the ionic strength was increased by increasing the salt concentration. One can more conveniently manipulate the elution proffle by varying the operating temperature. Our results show that reasonably precise estimates for stable HbA1 can be obtained with this minicolumn system at 28 “C. Some material that is not HbA10 is also collected under these conditions, as indicated by the shoulder at the trailing edge of the peak for HbA1 of the diabetic sample
and by the additional small but consistently observed peak in the controlsample. Increasing assay temperature has a practical advantage over increasing the ionic strength of the buffer, in that counter-cooling
by tap-water
is more
efficient
at higher
temperatures. In view of the marked temperature-dependence of results, temperature control is an absolute necessity for the determination of HbA1 or HbA1 with minicolumn systems. The optimal temperature should be determined for each minicolumn system by the procedure described above. The temperature-dependence of peak elution is also strongly influenced by column geometry. For example, in our HPLC system, a temperature increment of 10#{176}C decreases the elution volume by 0.5 mL (data not shown), while the same temperature increment in the minicolumn system decreases the elution volume by 9 mL. For quantitative peak collection, the elution volume corresponding to the peak absorbance should not vary by more than 1 mL, which means that a variation in conductivity of the buffer of ±0.05 mS can be tolerated if pH and temperature are held constant. The scatter in our data is partly a consequence of the collection of two separate fractions. In a protocol where HbAla±b is collected separately, HbAla+b can contaminate the HbA1 fraction and result in increased HbA1 values. Two previous reports comparing results of HbA1 determinations on minicolumns with an established reference method have also found a regression
line identical
to the proce-
line of identity, although with more time-consuming dures (16, 17). Our results show that a line of identity can be achieved with a shortened protocol. However, with shorter analysis time, buffer composition and control of temperature become more critical. In conclusion, the confusion of results from glycohemoglobin determinations can be avoided if the protocol is designed such that results match those of an accepted comparison method. The specifications for buffer composition and assay conditions can be easily determined by using the strategy described in this report, which must be evaluated separately for every minicolumn system.
We thank Mrs. D. Erne for help in preparing the manuscript, H. Passing and W. Bablock (Mannheim, F.R.G.) for the computer program for statistical evaluation of the method comparison, and R. Schnell (Bio-Rad, Zurich) for providing the minicolumns. This work was supported by a grant from the “Fonda zur FOrderung von Lehre sad Forschung der Universitat Basel” and a grant from the Frits Hoffmann LaRoche Stiftung.
References 1. Roth M. “Glycated hemoglobin,” not “glycosylated” or “glucosylated.” Clin Chem 29, 1991 (1983). Letter. 2. Nathan DM, Singer DE, Hurxthal K, Goodson JD. The clinical information value of the glycosylated hemoglobin assay. N Engi J Med 310, 341-346 (1984). 3. Worth RC, Ashworth LA, Burrin JM, et a). Column assay of haemoglobin A1: Critical effect of temperature. Clin Chim Ada 104, 401-404 (1980).
4. Mosca A, Carenini A, Samaja M, Saibene V. Temperature control in assay of glycosylated hemoglobins. Clin Chem 26, 11061107 (1980). Letter. 5. Johnson MW, Dobrea GM, Bendezu R, Wieland RG. Temperature dependence of the chromatographic assay of hemoglobin A, and application of a temperature controlled assay to clinical evaluation of diabetic control. Clin Chim Acta 104, 319-328 (1980). 6. Dix D, Cohen P, Kingsley S, eta). Evaluation of a glycohemoglobin kit. Clin Chem 24, 2073 (1978). Letter. 7. Hankins WD, Holladay L. A temperature conversion nomogram for glycosylated hemoglobin analysis. Clin Chim Acta 104,251-257 (1980). 8. Rosenthal MA. The effect of temperature on the fast hemoglobin test system. Hemoglobin 3, 215-217 (1979). 9. Hammons GT, Junger K, McDonald JM, Ladenson JH. Evaluation of three minicolumn procedures for measuring hemoglobin A,. Clin Chem 28, 1775-1778 (1982). 10. Rowe DJF, Goodland FC. Three chromatographic methods compared for measurement of hemoglobin A, and A,. Clin Chem
30, 156-157 (1984). Letter. 11. Lau HKY, Howard SF, Tomczak G, Tan TK. A lower reference interval for hemoglobin A,. Clin Chem 28, 1822-1823 (1982). Letter.
12. Bisse E, Berger W, Fluckiger R. Quantitation of glycosylated hemoglobin. Elimination of labile glycohemoglobin during sample hemolysis at pH 5. Diabetes 31, 630-633 (1982). 13. Bisse E, Berger W, Fluckiger R. Elimination von labilem Hamoglobin. Lab Med 8, 122-123 (1984). 14. Passing H, Bablock W. A new biometrical procedure for testing the equality of measurements from two different analytical methods: Application of linear regression procedures for method comparisonstudiesin clinical chemistry.J Clin Chem Clin Biochem 21, 709-720 (1983). 15. Berger W, Fluckiger R. Bestimmung der glycosylierten Hamoglobine HbAI mit kommerziellen Methoden-Folgerungen aus einem Methodenvergleich. Schw Aerztezeitung 65, 1620-1622
(1984). 16. JonesMB, Koler RD. JonesRT. Micro-column method for the determinationofhemoglobinminor fractions Ala+b and A,. Hemoglobin 2, 53-58 (1978). 17. Maquart FX, Gillery P, Bernard JF, et al. A method for specifically measuring haemoglobin A, with a disposablecommercial ion-exchange column. Clin Chim Ada 108, 329-332 (1980).
CLINICALCHEMISTRY, Vol. 31, No. 1, 1985 117
