Upsala J Med Sci 95: 291-297, 1990 Quality Control and Quality Requirements for the Measurement of Glycated Hemoglobin I. Penttila,' J. Gavert,' A . Julkunen' and T. Tantanen' 'Department of Clinical Chemistry, University of Kuopio, SF-70210 Kuopio, Finland, 2Labquality L t d , Aseniamiehenkatu 4 A , SF-00520 Helsinki, Finland In the follow-up of the treatment of diabetes mellitus the determination of glycated blood proteins, especially blood hemoglobin, has been proved to be very important in estimation of the long-term balance of the disease. Many methods have been developed, but its seems quite evident that the HPLC-methods which measure the proportion of glycated hemoglobin Alc (GHbAlc) or the affinity chromatography measuring all glycated forms of hemoglobin (GHb) are the most useful for the routine clinical laboratory work (1,2,3,5). There are many difficulties in organizing the quality control of the determination of glycated hemoglobin. First problem e.g. in Finland is that there are at least six commercial methods in use. In Table 1 there are presented the methods and their reference intervals for the determination of glycated hemoglobins. The second is the measurement temperature, which must be know when the values for glycated hemoglobins has been calculated by using minicolumn techniques. The best way is to perform the columnphase in a thermostated water bath. The third and perhaps the most difficult problem is to find the ideal control material which fulfills the criteria: the control material should be identical with patient samples, the concentration of the analyte should be at the desired level and the stability of the control material should be long enough. In the university central hospital district of Kuopio we have selected an automated HPLC-method and measured GHbAlc values for the whole district. Today we have two different instruments for the measurement of GHbAlc: the older one ( 3 ) is the FPLC-system (Pharmacia, Uppsala, Sweden) and the newer one an HPLC-instrument of Shimadzu (6) with an strong ion-exchange 291 column supplied by Beckman Instruments Inc. (Palo Alto, CA, U.S.A.).To follow-up the analytical level of our method we have used washed and frozen (-70°C) concentrated erythrocytes (1) . The typical variation coefficients within day and also between days have been between 1.5 and 2.0 %. Our method measures glycated HbAlc and acetyl HbAO (3,7), it is well linear between 3 . 0 and 20 % and is not influenced by HbF thus being suitable for analyses during pregnancy. In the meetings with our clinicians it has been stated that the difference of the GHbAlc value measured should not exceed 0.5 % from the right value in the normal reference interval of 4.0-6.0 %. This means that the analytical variation should be below 2.5 %. It seems that this can be reached when the method is carefully selected, calibrated and continuously followed. We believe that this these criteria will be obtained in most laboratories where automated HPLC-techniques for GHbAlc is used.From the clinical point of view it is almost important that on a district, where the diabetics visit to different physicians and thus to many laboratories, an agreement on the principle (one method) is made for clinical use. The more difficult question concerns the right value when performing the quality control surveys in Finland and in Nordic countries. The controls should contain both the stabile ketoamine form as well as the labile Schiff's base so that they correspond as closely as possible the situation in clinical laboratories. The first step in the measurement of GHbAl or GHbAlc is the elimination the labile Schiff's base, which it is not so marked when using affinity chromatography for GHb ( 4 ) . For this reason lyophilized hemolysates may give too good results when using minicolumns and ion-exchange technique. Secondly many commercial lyophilized preparations contain components or fractions which render the HPLC-techniques (1). In most surveys we have seen (Table 2, Fig. 1) that the HPLC-methods as a whole are not very much better in performance than the manual methods using either affinity or ionexchange minicolumn. It can, however, be calculated that EDTA-blood samples are better than the hemolysates (Fig. 1) and the smallest coefficient of variation is seen in the group of HPLC-methods especially when the variation coefficients are 292 Table 1. Different methods used by participating laboratories in the the surveys of Labquality Ltd. for the determination of glycated hemoglobins and they reference intervals Method Component Reference measured intervals Isoelectric focusing GHbAlc 4.0 - 6.5 % HPLCIFPLC ionex chromat. Acetyl HbAlc 4.0 - 6.1 % Mini-ionex chromat.1 ( HbF 1 3.8 - 6.1 % Biorad Electrophoresis GHbAla,b,c 5.5 - 9.0 % Mini ionex chromat. Acetyl HbAlc 5.6 - 8.6 % HbF (Other Hb:s) Mini affinity chromatography GHbAla,b,c GHbF GHbA2 GHBAO 4.6 - 8.2 % 293 Table 2. The quality control results from the surveys during the years 1987 to 1989 by Labquality Ltd. in Finland. The number of laboratories using mini-column techniques (GHbAl) has decreased, the users of affinity chromatography (GHb) correspondingly increased and those with HPLC-techniques (GHbAlc) clearly increased. Method Sample GHbAl EDTA-b10od/l/87 EDTA-blood/l/88 EDTA-blood/l/89 Hemolysate/l/87 Hemolysate/l/88 Hemolysate/l/89 GHb EDTA-b10od/l/87 EDTA-blood/l/88 EDTA-b10od/l/89 Hemolysate/l/87 Hemolysate/l/88 Hemolysate/l/89 GHbAlc EDTA-b100d/l/87 EDTA-b100d/l/88 EDTA-b100d/l/89 Hemolysate/l/87 Hemolysate/l/88 Hemolysate/l/89 No. Mean SD 21 6.1 % 0.45 20 10.5 % 0.77 14 6.5 % 0.89 21 7.5 % 0.89 19 12.8 % 0.49 14 5.3 % 0.74 8 4.9 % 0.51 8 13.7 % 1.07 13 9.95 % 0.52 8 5.0 % 0.64 8 13.6 % 1.57 13 5.3 % 0.94 5 4.7 % 0.49 6 9.8 % 0.35 22 4.8 % 1.04 5 5.1 % 0.11 6 10.5 % 0.45 22 5.1 % 0.77 cv 7.3 % 7.3 % 13.7 % 11.9 % 10.0 % 10.7 % 10.4 % 7.8 % 10.0 % 12.8 % 11.6 % 17.9 % 10.5 % 3.6 % 14.7 % 2.2 % 4.5 % 14.9 % 294 Fig. 1. The results from the quality control surveys of glycated hemoglobin by three most common methods in Finland. The results are expressed as the mean values in the survey 211989. CONTROL SURVEY 2/1989 OF LABQUALITY z m 2 (3 0 r 10 - W I n W 0 (3 t 5 EDTA-L EDTA-H HEMOL-L HEMOL-H GHbAl GHb GHbAlc 295 calculated separately for each Nordic country. As stated before it seems possible to maintain the intralaboratory variation good enough, eg. below 3.0 %. However, when the interlaboratory variation is much more higher, it seems that there must be troubles either in quality control samples or in the standardization. The great importance of the value of glycated hemoglobin in the treatment of diabetics and on the other hand possible technical difficulties in the measuring procedures, according to our opinion, is a good reason to start a Nordic project. REFERENCES: 1. 2. 3. 4. 5 . 6. Ellis G, Diamandis EP, Giesbrecht E, Daneman D, Allen LC. An automated "high-pressure" liquid-chromatographic assay for hemoglobin Alc. Clin Chem 1984; 30: 1746-1752. Goldstein DE, Little RR, Wiedmeyer H-M, England JD, McKenzie M. Glycated hemoglobin: Methodologies and clinicalapplications. Clin Chem 1986; 32: B65-B70. Jeppsson J-0, Jerntorp P, Sundkvist G, Englund H, Nylund V. Measurement of hemoglobin Alc by a new liquid-chomatographic assay: methodogy, clinical utility, and relation to glucose tolerance evaluated. Clin Chem 1986; 32: 1867-1872. John WG. Effect of Schiff base (labile fraction) on the measurement of glycated hemoglobin by affinity chromatography. Clin Chem 1984; 30: 1111-1112. Klenk DC, Hermanson GT, Krohn RI, Fujimoto EK, Mallia AK, Smith PK, England JK, Wiedmeyer H-M, Little RR, Goldstein DE. Determination of glycosylated hemoglobin by affinity chromatography: comparison with colorimetric and ion-exchange methods, and effects of common interferences. Clin Chem 1982; 28: 2088-2094. Parviainen M, Julkunen A, Penttila I. *'A rapid GHbAlc-analysis by using an HPLC-instrument of Beckman System Gold". Kliin. Lab 1990; 7: 71-73. 296 7. Suhonen L, Stenman U-H, Koivisto V, Teramo K. Correlation of HbAlc, glycated serum proteins and albumin, and fructosamine with the 24-h glucose profile in insulindependent pregnant diabetics. Clin Chem 1989; 35: 922-925. Correspondence: Professor I. Penttila, Department of Clinical Chemistry, University of Kuopio, SF.70210 Kuopio, Finland 297