ANNALS O F CLIN ICA L AND LABORATORY SCIEN CE, Vol. 25, No. 1 Copyright 1995, Institute for Clinical Science, Inc. A Hemoglobin A1C Immunoassay Method Not Affected By Carbamylated Hemoglobin* ANDREA M. ROSE,! CAROLYN TONGATE, and ROLAND VALDES Jr., P h.d.tt Departments of Pathologyf and Biochem istryt University o f Louisville, and University o f Louisville H ospital, Louisville, KY 40292 ABSTRACT Hem oglobin A1C (HbAlc ) methods based on charge separation of Hb species are subject to interference from carbamylated Hb (carb Hb). Carb Hb adducts are formed via interaction of terminal amino groups of HbA with isocyanic acid, after the spontaneous dissociation of urea to cyanate. It is hypothesized that a new immunoassay method, using a monoclonal antibody that recognizes the N-terminus of the Hb p-chain and its sugar moiety, should be refractory to cross-reactive interference from carb Hb. To test this hypothesis, Hb was carbamylated in vitro and co-migration of carb Hb assessed w ith HbAlc using an electrophoretic method. D ensitom etric scans post sodium cyanate incubation and electrophoretic separation show ed a 5 to 7 fold elevation of the HbAlc peak only, w hile HbAlc values obtained using im m unoassay w ere unaffected. Also assessed was carbamylation interference in vivo, and a positive proportional bias w ith the electrophoretic system (Y) was observed compared to the immunoassay system (X) (y = 1.2x 0.21 percent). Others have shown that carb Hb may cause a clinically significant false elevation in patient HbAlc values, when methods based on charge separation of Hb species are used. It is our conclusion, however, that while carb Hb may play a role, the differences observed in this study are largely due to calibration. Introduction betics, but can reach levels of 20 p ercen t......,. in diabetics m aintaining poor control of G lycated hem oglobins (hem oglobins blood glucqse Thg recendy com pleted w ith covalently bound sugars) constitute D iabetes C ontrol and C om plications 4 to 8 percent of the total Hb of nondia- Tdal (DCCT) confirmed a 3 5 to 7 0 per_ IT i ^ ^ T cent overall reduction in the risk o f * bend reprint requests to: Dr. Roland Valdes Jr.,, University of Louisville, D epartm ent of Pathology, diabetic complications, including nephro- Louisville, KY 40292. pathy, among patients maintaining ade- 13 0091-7370/95/0100-0013 $01.20 Institute for Clinical Science, Inc.
14 ROSE, TONGATE, AND VALDES quate control of blood glucose. To reduce complications, the DCCT recom mends frequent blood glucose m onitoring and periodic m easurem ent of HbA l c. 1 Glycated Hb, of which HbAlc is the best characterized and predom inant species, form nonenzymatically in red blood cells (RBC). The amount of glycated Hb form ed is proportional to the average blood glucose level over the 1 2 to 16 week life span of the RBC, and, thus, is a m easure of glycem ic control.2 H em oglob in A 1C differs from HbA, the n o n glycated form, in that it has a glucose residue covalently linked to the amino group of the N -term inal am ino acid of each beta-chain.3 Analogous to glycation, carbamylation of Hb also occurs nonenzym atically in patients w ith urem ia. U rea spontaneously d isso ciates in plasm a to form ammonium and cyanate ions. Terminal amino groups of the Hb beta-chains interact with the reactive isocyanic acid to irreversibly form a carbamyl derivative whose concentration is proportional to the time-averaged concentration of blood urea.4 Thus, like HbAlc, carb Hb accum ulates over the life of the RBC. Carbamylation of various proteins has been suggested to play a pathophysiologic role in the m echanism of urem ic toxicity.5 By altering the H b charge, carbamylation interferes with glycated H b methods based on charge separation of H b species, i.e., ion exchange (including high pressure liquid chromatography [HPLC]) and electrophoresis.6 Carb Hb typically migrates with HbAlc. The Miles DCA 2 0 0 0 immunoassay system* uses a monoclonal antibody that recognizes the sugar m oiety and the amino term inus of the (3-chain of H ba l c. T herefo re, unlike m ethods th a t se p arate H b b a sed on c h a rg e, th is m eth o d sh o u ld n o t be affected by carb Hb. To test this hypothesis, Hb was carbamylated in vitro, and the HbAlc results obtained by im m unoassay* were compared with the results obtained using e lectro p h o resis.t The effect of carbamylation in vivo was also assessed by comparing the HbAlc values obtained using both m ethods in diabetics w ith norm al renal function, diabetics with uremia, nondiabetics with uremia, and normal controls. Materials and M ethods S a m p l e P o p u l a t i o n In vitro carbamylation of Hb was perform ed on w h o le b lood e th y le n e d i- aminetetraacetic acid (EDTA)-preserved samples obtained by venipuncture from five normal, nondiabetic controls with blood urea nitrogen (reference range 6 to 23 mg/dl), Hb (reference range 12 to 18 g/dl), and hem atocrit (reference range 37 to 52 percent) values all w ithin their referen ce ranges. In vivo studies used whole blood ED TA -preserved samples obtained from: (1 ) diabetics w ithout uremia with blood urea nitrogen, Hb and hem atocrit values w ithin norm al reference range; and (2 ) nondiabetics and diabetics w ith urem ia w hose values for blood urea nitrogen ranged from 26 to 97 mg/dl (mean 48.2 mg/dl), H b ranged from 9.6 to 15.7 g/dl (mean 10.4 g/dl), and hematocrit ranged from 28.5 to 47.8 percent (mean 34.1 percent). INSTRUM ENTATION The Miles DCA 2000* immunoassay system is an autom ated, self-contained instrum ent that analyzes 1 jxl of whole blood. Reagents in the unitized cartridge * Miles Inc., Diagnostic D ivision, Elkhart, IN 46515. t Beckman Instrum ents, Inc., 200 S. Kraemer Blvd., Brea, CA 92621-6209.
HEMOGLOBIN A1C IMMUNOASSAY M ETHOD NOT A FFECTED BY CARBAMYLATED HEMOGLOBIN 15 lyse the RBCs, and the H b is oxidized to m ethem oglobin using potassium ferricyanide. M ethemoglobin complexes with thiocyanate to form thiocyanm ethem o- globin which is m easured spectrophotom etrically. Color developm ent is proportional to the concentration of total Hb in the sample. HbAlc is m easured using an inhibition of latex agglutination tech nique. An agglutinator (synthetic polym er containing m ultiple copies of the im m unoreactive portion of HbAlc, i.e., first three am ino acids at the amino term inus of the p-chain and the sugar) causes agglutination of latex coated with H ba lc -specific m ouse m onoclonal antibody. Agglutination increases light scattering, which is m easured as an increase in absorbance. Hemoglobin A1C in whole blood specimens competes for the lim ited num ber of antibody-latex bin d in g sites. In h ib i tion of agglutination, and decreased scattering is m easured as a decrease in absorban ce. T he H ba lc c o n c e n tra tio n is quantitated using a 1 2 point calibration curve of absorbance versus H ba lc concentration. The percent of HbAlc in the sam ple is then calculated as percent HbAlc = [HbAlc ]/[Total Hb] x 100. All m easurem ents and calculations are perform ed autom atically by the analyzer.7 The Beckman D iatract electrophoretic system performs a charge separation of Hb species. Hemolysate is applied to an agar-based support m edium provided by the m anufacturer and an electrical potential is in tro duced p er m anufacturer s instructions. H b bands w ere scanned with a Beckman Appraise densitom etert for quantitation. The percent HbAlc in the sam ple was calculated as percent H ba lc = [H ba l c ]/[H ba 0 + H ba lc ] x 1 0 0. I n V i t r o C a r b a m y l a t i o n R e a c t i o n In vitro carbamylation was performed as described previously.8 O ne m L of E D T A -p re se rv e d w h o le b lo o d w as mixed with 10 ml isotonic saline (0.85 percent). F ollow ing centrifugation at 1500 x g, for 10 min, the plasm a and buffy coat w ere rem oved. T he RBCs were washed again in 10 ml saline, then resuspended in 5 ml Suspension Buffer (5.14 mm Na2 H P 0 4 2H 2 0, 1.53 mm KH2 P 0 4, 145 mm NaCl, ph 7.2). Fifty xl of 0.5M sodium cyanatet was added to each tube, and the cells w ere in cu bated at 37 C for 1 and 2 hr. The cells were washed with 5 ml of isotonic saline as described previously, resuspended in 0.5 ml isotonic saline, and stored at 4 C prior to analysis by both m ethods the next day. Results An in vitro carbamylation reaction was performed using whole red blood cells from healthy nondiabetics to determ ine w hether or not the HbAlc immunoassay system is subject to interference from carb H b. U sin g an e le c tro p h o re tic m ethod that separates Hb species based on charge, the degree of H b carbam ylation was densitom etrically detected following one or two hours of incubation with sodium cyanate. The densitom etric scans visually dem onstrated that carb Hb co-migrates with HbAlc. A 5 to 7 fold increase in the HbAlc peak was observed follow ing sodium cyanate incubation using the electrophoretic system. A comparison of the HbAlc values obtained using the immunoassay system and the electrophoretic system pre- and postt Beckman Instrum ents, Inc., 200 S. Kraemer t Aldrich Chemical Co., P.O. Box 355, Milwau- Blvd., Brea, CA 92621-6209 kee, WI 53201.
16 ROSE, TONGATE, AND VALDES F igure 1. C om pari- 50 son of the percent HbAlc values obtained from five nondiabetics w ith normal 40 renal function, prior to or p o s t in c u b a tio n w ith sodium cy an ate, usin g Miles DCA 2000 immu- S? 30 noassay m ethod and the 0 Beckman Diatrac electro- < phoretic system. Each bar x 20 g ra p h r e p r e s e n ts th e m ean value for all five individuals. Bars represent 1 SD. A nalysis of each patient sample was perform ed in duplicate at each tim e p o in t. DCA 0 2000: M ean H ba lc values 0 1 2 of 3 tim e points, SD and CV for each nondiabetic C yanate Incubation (hr) control. (1) Mean 4.7 percent, SD 0.29, CV 6.2 percent; (2) Mean 4.0 percent, SD 0.30, CV 7.6 percent; (3) M ean 4.2 percent, SD 0.1, CV 2.4 percent; (4) Mean 4.3 percent, SD 0.1, CV 2.3 percent; and (5) 4.2 percent, SD 0.15, CV 3.6 percent. incubation is show n in figure 1. Statistical analysis of results obtained with the immunoassay m ethod (figure 1 legend) indicate that carb Hb does not interfere with the accurate assessment of HbAlc by this method. The current authors also w ished to determ ine w hether or not the HbAlc values observed in diabetics with urem ia could partially be attributed to H b carbamylation in vivo, and to what extent. Figure 2 com pares the HbAlc results using both m ethods for 29 individuals. The reference range established by the A m erican D ia b e te s A sso ciatio n for HbAlc percentages in the nondiabetic population is 4 to 6 percent. Both m ethods gave HbAlc results within this range for th e five norm al co n tro ls. W hen assessed by immunoassay, only one of the seven urem ic, nondiabetic individuals had a HbAlc result above 6 percent; how ever, w hen assessed by electrophoresis, five of these same uremic, nondiab etic in d iv id u als had H ba lc values above 6 percent. All HbAlc values for diabetics, with and without uremia, were elevated above 7 percen t by both m ethods, w ith clear overlap b e tw e en th e two groups, except for one d iab etic, urem ic individual. Figure 2 also shows a positive proportio n al bias w ith the e le c tro p h o re tic m ethod as compared to the immunoassay method, indicating that as the percentage of H ba lc in c re a se s, th e d iffe re n c e betw een the values obtained by both m ethods for each individual increases. However, a plot of blood urea nitrogen versus percent bias (figure 3) for all individuals has a slope which is essentially zero (Y = 0.005X + 20.75 percent), suggesting no correlation betw een the bias and blood urea nitrogen. Discussion Analysis of whole blood samples preand post-incubation with a carbamylating agent allowed the authors to determ ine that the immunoassay m ethod is not subjec t to interference from H b carbam ylation. R ecent studies,6 9 including our own, suggest that in vivo, carbamylation of Hb may increase the H ba lc value w hen assessed by m ethods b ased on
HEMOGLOBIN Alc IMMUNOASSAY METHOD NOT AFFECTED BY CARBAMYLATED HEMOGLOBIN 17 F ig u r e 2. C om parison of the percent HbAlc valu es o b ta in e d u sin g im m u n o a s s a y (M ile s DCA 2000) and electrophoresis (Beckman Diatrac) for norm al controls ( +, n = 5); d ia b e tic s w ithout urem ia (T, n = 6); diabetics w ith urem ia (A, n = 11); and urem ic individuals w ithout diabetes (O, n = 7). Y = 1.2X - 0.21 percent; r = 0.978. Immunoassay % HbAlc charge separation of H b species. This could explain the HbAlc percentages in excess of the 4 to 6 percent normal range observed for five of the seven uremic, nondiabetic individuals w hen evaluated by the electrophoretic method. Hyperu re m ic in d iv id u a ls o fte n e x h ib it decreased red cell survival, m aking accurate assessm ent of glycation or carbamylation problematic. Both of the uremic, nondiabetic individuals with electrophoretically determ ined HbAlc percentages w ithin the norm al range had decreased hem oglobin and hem atocrit levels. A nem ia could, therefore, account for the lo w e r v a lu e s o b s e r v e d in th e s e two individuals. W hile carb H b has been show n to interfere in m ethods based on charge separation of Hb species, it cannot solely acco u n t for th e d isp a rity in re su lts observed in this study (figure 2 ), betw een these two methods, or betw een glycated Hb methods in general. 10,11 For example, although p ercen t H ba lc for th e normal, nondiabetic, nonurem ic controls < BUN (mg/dl) 10 0 FIGURE 3. The difference (Beckm an D iatrac electro p h o retic m ethod m inus M iles DCA 2000 im m unoassay m eth o d ) b e tw e e n th e p e r c e n t HbAlc values divided by the value obtained using the immunoassay m ethod x 100 (percent bias) (Y) plotted against the blood u re a n itr o g e n v a lu e s (BUN) (X) for each individual. N orm al controls ( + ), d ia b e tic s w ith o u t u re m ia ( ), d ia b e tic s w ith uremia (A), and uremic individuals w ithout d i a b e t e s (O ). Y = 0.005X + 20.75 percent; r = 0.065.
18 ROSE, TONGATE, AND VALDES rem ained w ithin the 4 to 6 percent normal range using both m ethods, bias ranging from 8 to 25 percent was observed (figure 3). In addition, the bias observed betw een the two m ethods for all groups did not correlate w ith blood urea nitrogen levels, suggesting involvem ent of some other factor. The proportional bias, shown in figure 2, suggests that the percent HbAlc differences may be due to standardization, as m ight be observed w ith instrum ent calibration variability. G lycation and carb am y latio n both involve the nonenzymatic condensation of glucose or urea-derived cyanate with hemoglobin. The extent of glycation or carbamylation depends on the relative affinities of glucose and cyanate for hem oglobin over the life of the RBC. Assuming comparable affinities, HbAlc values for diabetics w ith and w ithout urem ia w ould be expected to be sim ilar w hen assessed by electrophoresis. However, uremic, diabetic individuals might be expected to have low er on average HbAlc levels than diabetics w ithout uremia w hen evaluated using an immunoassay m ethod that does not recognize carb H b. T his issue should b e addressed using a larger data set, keeping in mind that the presence of anem ia will also falsely decrease the HbAlc result. W hile the patient num ber is small, this study suggests that carbam ylation may falsely inflate the percentage of HbAlc when obtained using m ethods based on charge separation of hem oglobin. M oreover, the study confirms the need for individuals to be evaluated against their ow n p rev io u s H b A lc v a lu e s, u sin g the same m ethod, in order to obtain a true picture. These results reinforce the utility of standardization in the evaluation of glycated Hb. Obviously, the contribution of any non-hbalc com ponent with certain methodologies compounds the problem of lack of standardization in evaluation of HbA l c. 10,11 A recent quality assurance survey co n d u cted by th e C ollege of Am erican Pathologists show ed that 65 percent of laboratories reporting HbAlc values use m ethods based on charge separation of Hb species. 12 Weykamp et al have determ ined that HPLC, disposable ion-exchange, and electrophoresis, all of w hich identify H b species based on charge differences, are subject to interference by Hb derivatives such as carbamylated and acetylated H b.6,13 The prevalence of renal complications owing to diabetes significantly contributes to the num ber of kidney transplants performed yearly and to the cost of m edical care, which is approaching $ 2 billion p e r y e a r. 14 T he recen tly co m p le te d D C C T 1 has concluded that intensive therapy, which includes three or more insulin injections per day or the use of an insulin pum p, diet and exercise counseling, fre q u e n t glucose m onitoring, frequent interaction with members of the patient s m anagem ent team, and periodic monitoring of HbAlc, effectively delays the onset and slows the progression of diabetic retinopathy, nephropathy, and neuropathy. Therefore, lack of consensus on e ith e r a stan d ard or a re fe re n c e m ethod interferes w ith patien t m anagem ent and the assessm ent of glycem ic control and also hinders im plem entation of the recommendations of the American D iabetes Association. 14,15 The authors conclude that the DCA 2 0 0 0 immunoassay system is not subject to interference from carb Hb, and that the specificity achieved using this im m unochemical approach is ideal for assessing glycemic control. Acknowledgem ents Thanks are extended to Miles Inc., Elkhart, IN, and to Beckman Instrum ents, Inc., Brea, CA, for providing instrum entation and materials used in this study.
HEMOGLOBIN Alc IMMUNOASSAY M ETHOD NOT AFFECTED BY CARBAMYLATED HEMOGLOBIN 19 References 1. The D iabetes Control and Complication Trial Research Group. The effect of intensive treatm ent of diabetes on the developm ent and progression of long-term complications in insulindependent diabetes mellitus. N Engl J M ed 1993;329:977-86. 2. Bunn H F, H aney DN, Kamin S, Gabbay KH, Gallop PM. The biosynthesis of hum an hem o globin A1C. Slow glycosylation of hemoglobin in vivo. J Clin Invest 1976;57:1652-9. 3. Bunn H F, Gabbay KH, Gallop PM. The glycosylation of hemoglobin: relevance to diabetes m ellitus. Science 1978;200:21-7. 4. Fluckiger R, Harmon W, M eier W, Loo S, Gabbay KH. Hemoglobin carbamylation in uremia. N Engl J M ed 1981;304:823-7. 5. Smith WGJ, H olden M, Benton M, Brown CB. C arbam ylated haem oglobin in chronic renal failure. Clin Chim Acta 1988;178:297-304. 6. W eykamp CW, Penders TJ, Siebelder CWM, M uskiet FAJ, van der Slik W. Interference of carbam ylated and acetylated hem oglobins in assays of glycohemoglobin by HPLC, electrophoresis, affinity chromatography, and enzyme immunoassay. Clin Chem 1993;39:138-42. 7. Procedure M anual, DCA 2000 A nalyzer for HbA lc, Miles Inc. Elkhart, IN, October 1991. 8. E ngbaek F, C hristensen SE, Jesp ersen B. E nzym e im m unoassay o f H em oglobin A1C: Analytical characteristics and clinical performance for patients with diabetes m ellitus, w ith and w ithout uremia. Clin Chem 1989;35:93-7. 9. B runnekreeft JW I, E idhof HHM. Im proved rapid procedure for simultaneous determ inations of Hemoglobins Ala, Alb, Ale, F, C, and S, w ith indication for acétylation or carbamylation by cation-exchange liquid chrom atography. Clin Chem 1993;39:2514-8. 10. Bruns D E. Standardization, calibration, and the care of diabetic patients. Clin C hem 1992;38: 2363-4. 11. Weykamp CW, Penders TJ, M uskiet FAJ, van der Slik W. Effect of calibration on dispersion of glycohemoglobin values determ ined by 111 laboratories using 21 m ethods. C lin C hem 1994;40:138-44. 12. College of American Pathologists, Proficiency Testing Program, Glycohemoglobin (Percent), Survey Series EC-A, EC1-01, N orthfield, IL, 1993. 13. Weykamp CW, Penders TJ, M uskiet FAJ, van der Slik W. Influence of hemoglobin variants and derivatives on glycohemoglobin determ i nations as investigated by 102 laboratories using 16 methods. Clin Chem 1993;39:1717-23. 14. Santiago JV. Lessons from the D iabetes Control and C om plications Trial. D iabetes 1993;42: 1549-54. 15. B enjam in RJ, Sacks DB. G lycated p ro tein update: Implications of recent studies, including the D iabetes Control and Complications Trial. Clin Chem 1994;40:683-7.