Determination of glycated albumin by enzyme-linked boronate immunoassay (ELBIA)

¹ Department of Urology & Biochemistry, Kumamoto University School of Medicine, Honjo, 2-2-1, Kumamoto 860, Japan.

² The Research Institute, Nacalai Tesque, Kyoto, Japan.

³ Department of Clinical Pathology, Showa University School of Medicine, Tokyo, Japan.

⁴ Laboratory of Clinical Chemistry, National Cardiovascular Center Hospital, Suita, Japan.

	Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
A new affinity method for quantification of glycated albumin by an enzyme-linked boronate-immunoassay (ELBIA) has been established, based on the interaction between boronic acids and the cis-diols of glycated human serum albumin (HSA) trapped by anti-HSA antibody. To evaluate the ELBIA, we first examined the accuracy of the conventional boronate affinity chromatographic (BAC) method. In the BAC method, 8.1–18.9% of nonglycated albumin calibrator nonspecifically bound to the boronate affinity column, values that were regarded as the column blank. In the modified BAC method, therefore, we subtracted the column blank value from the measured glycated albumin value to obtain the true value. Because glycated albumin values measured by ELBIA were exactly the same as reported by the modified BAC method, we suggest that the ELBIA results reflect the real status of albumin glycation. We have also developed a fully automated ELBIA system, allowing multiple, rapid, and precise measurements of glycated albumin.

	Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
The plasma concentration of glycated albumin reflects short-term control of hyperglycemia in diabetes (1). Affinity-chromatographic methods based on specific interaction of boronic acids with the cis-diols of glycated proteins have been developed (2)(3)(4) and are widely used for measuring glycated proteins (5). The affinity-chromatographic method has also been applied to determine serum concentrations of glycated albumin (6). The shorter half-life in vivo of glycated albumin makes it potentially a more sensitive indicator of hyperglycemic control than glycohemoglobin is, and several recent studies have focused on the significance of the clinical applicability of glycated albumin (7)(8)(9)(10)(11)(12)(13).

In these previous reports (7)(8)(9)(10)(11)(12)(13), glycated albumin values were determined by high-performance boronate affinity chromatography (HP-BAC)¹ or conventional boronate affinity chromatography (BAC). Fujita et al. (14) recently compared the HP-BAC method with a modified BAC method to determine the amount of glycated albumin in samples. In the modified method, they subtracted the column blank value from the measured glycated albumin value to obtain the true glycated albumin value. They showed that the concentration measured by the first method was always higher than by the latter. Their experiments clearly showed that such error in estimation occurred because the peak of the adsorbed fraction of glycated albumin was not well separated from that of the nonadsorbed fraction in the HP-BAC method. To confirm this mechanism, they demonstrated that prolonging the period required for separation resulted in the two methods measuring similar concentrations of glycated albumin.

Miyake and Tanaka (15) have recently described a new enzyme-linked boronate immunoassay (ELBIA) affinity method for quantifying glycated albumin (Fig. 1 ). In this method, boronic acids interact with cis-diols of glycated human serum albumin (HSA) trapped by anti-HSA. In the present study, we compared the accuracy and sensitivity of the ELBIA with that of the modified BAC method for measuring the amount of glycated albumin in human serum samples from apparently healthy subjects and from diabetic patients.

View larger version (15K): [in this window] [in a new window]   Figure 1. Determination of glycated albumin by ELBIA, which is based on the interaction between boronic acids and the cis-diols of glycated human serum albumin (HSA) trapped by anti-HSA antibody coated onto microtiter plate wells.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
subjects
Venous blood samples were obtained from 581 fasting healthy or diabetic subjects. Of the 350 samples obtained from subjects who received a 75-g oral glucose tolerance test (OGTT) at Kumamoto Health Care Center Japanese Red Cross, samples from 112 subjects (ages 47.6 ± 8.7 years) were used to determine the reference interval for the glycated albumin concentration. The plasma glucose concentrations during OGTT in these subjects were 5.08 ± 0.36, 6.50 ± 1.34, and 5.56 ± 0.78 mmol/L (<110, 160, and 120 mg/dL) at 0, 60, and 120 min, respectively. A 75-g OGTT was performed in the morning after a 12-h overnight fast. To exclude liver diseases, we selected the subjects on the basis of normal results for liver function tests, including serum albumin, serum total protein, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, and lactate dehydrogenase. Serum was separated by centrifugation at 1500g at room temperature and stored at -80 °C before determination. Anticoagulated whole blood was collected into EDTA-Na₂-containing tubes to determine the proportion of stable Hb A_1c present.

procedures
Preparation of nonglycated albumin solution.
One gram of HSA (Fraction V; Sigma) dissolved in 300 mL of 0.5 mol/L glycine–NaOH containing 20 g/L MgCl₂ (pH 8.5) was incubated with stirring for 2 h at room temperature with 500 mL of phenylboronic acid resin (PBA-60; Amicon) equilibrated with the same buffer. The solution was passed through a glass filter. The filtrate was incubated again for 2 h at room temperature with another portion of the same resin, and was again filtered through a glass filter. The final solution obtained was concentrated to 30 g/L by an ultrafiltration system. The nonglycated albumin solution was assayed with the fructosamine assay kit from Boehringer Mannheim. Furosine, formed by acid hydrolysis of fructoselysine, was determined by amino acid analysis (16).

Preparation of glycated calibrator.
A portion of the nonglycated albumin solution was further concentrated to 80 g/L by ultrafiltration. We added 830 mg of D-glucose to 2.5 mL of the nonglycated albumin solution and incubated with stirring at 56 °C for up to 80 min. At specific intervals, portions were withdrawn, cooled on ice, and dialyzed against saline. The prepared solution (protein concentration of 30 g/L) was used as the glycated albumin calibrator in the present study.

Purification of albumin fraction from human serum.
After packing 4 mL of Blue Sepharose resin (CL-6B; Pharmacia) into a column and equilibrating this with 40 mL of 0.05 mol/L Tris-HCl buffer (pH 7.0) containing 0.1 mol/L KCl (buffer A), we loaded 500 µL of human serum onto the column and washed the column with 100 mL of buffer A. The adsorbed albumin fraction was then eluted with 10 mL of 0.05 mol/L Tris-HCl buffer (pH 7.0) containing 1.5 mol/L KCl. The eluate was dialyzed against saline, concentrated to 30 g/L, and used as a purified albumin fraction. The purity of the albumin fraction was >95%, as evaluated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis.

Determination of glycated albumin by the modified BAC method.
Glycated albumin was determined by affinity chromatography as described previously (15). Briefly, 3 mL of phenylboronic acid resin (PBA-60) was equilibrated with 0.5 mol/L glycine–NaOH (pH 8.5) containing 20 g/L MgCl₂ and 2 mol/L urea (buffer B) and was packed into the column. Nonglycated or glycated albumin calibrator solution (50 µL) or albumin purified from human serum as described above was loaded onto the column, followed by 200 µL of buffer B. After 10 min, the column was washed with 50 mL of buffer B to elute the passed through nonglycated albumin fraction. The adsorbed glycated albumin fraction was eluted with 25 mL of 0.5 mol/L glycine-HCl (pH 8.5) containing 20 g/L MgCl₂ and 0.1 mol/L sorbitol. Because the fluorescence intensity measured correlates with the amount of tryptophan, a constitutive amino acid of albumin, we determined the concentration of albumin in each fraction with a fluorescence spectrophotometer (F3010; Hitachi) at excitation/emission wavelengths of 285/340 nm. We also determined the nonspecific binding of albumin to boronic acid resin, the "column blank", representing the binding of nonglycated albumin to the resin. Finally, the concentration of glycated albumin was obtained by subtracting the column blank from the glycated albumin concentration measured by fluorescence spectrophotometry.

Determination of glycated albumin by ELBIA.
The amount of glycated albumin was determined with manual ELBIA (14) with a glycated albumin assay kit (Labofit Glycoalbumin; Nacalai Tesque, Kyoto) by a skilled technician. As shown in Fig. 1 , this method depends on the interaction of boronic acids and cis-diols of glycated HSA trapped by anti-HSA antibody coated onto a microtiter plate well. Assays were performed at room temperature with incubation on a shaker. To each well of a 96-well microtiter plate coated with anti-HSA antibody we added 50 µL of 10 mmol/L phosphate buffer (pH 7.4) containing 0.5 mL/L Tween 20 and then 20 µL of the test sample. After incubation for 20 min, the wells were washed three times with 300 µL of 5 mmol/L sodium carbonate buffer containing 20 g/L MgCl₂ (washing buffer) and incubated for 20 min with the boronate–horseradish peroxidase conjugate in 50 µL of 100 mmol/L glycine–NaOH (pH 9.0) containing 20 g/L MgCl₂, 3 g/L bovine hemoglobin, and 0.5 mL/L Tween 20. The wells were then washed five times with washing buffer and reacted for 20 min in the dark with 1 g/L o-phenylenediamine in 50 µL of 0.1 mol/L citrate buffer (pH 5.8) containing 4.4 mmol/L H₂O₂. The reaction was terminated by adding 50 µL of 2 mol/L sulfuric acid. The absorbance at 492 nm was measured with a microplate reader (MTP-120; Corona Electronic). The nonglycated albumin calibrator (NGA) and the glycated albumin calibrator, in which 40% of the albumin was glycated (GA40%), were used as to quantify the proportion of glycated albumin in each sample. The absorbance of NGA represented the blank value observed in the ELBIA procedure, i.e., nonspecific binding of nonglycated calibrator to boronate. The A_{492 nm} of NGA was ~0.4. The percentage of glycated albumin in a sample (%GA) was calculated according to the formula:

In a fully automated ELBIA system (AP-960 GA version; Kyowa Medex), all steps from the sampling to the calculation of glycated albumin values were executed automatically.

Determination of glycated albumin by the HP-BAC method.
The amount of glycated albumin was determined by a fully automated HP-BAC system (GAA-2000; Kyoto Daiichi Kagaku). The system consists of an anion-exchange column for isolation of albumin and a BAC for separation of glycated from nonglycated albumin. The eluate was monitored by fluorescence intensity (excitation wavelength 285 nm; emission wavelength 340 nm) to evaluate the yield of albumin.

Determination of stable Hb A_1c by automated analysis.
Stable Hb A_1c of the anticoagulated whole blood was determined by an automated analyzer (HLC-723GHbIII-S; Tosoh) for stable glycohemoglobin. The system is based on a high-performance ion-exchange liquid-chromatographic method.

statistical analysis
To determine the reference interval for glycated albumin, we considered 6 forms of the variable (log x, x^1/3, x^1/2, x, x, and x) and selected the most suitable distribution according to the maximum likelihood method (17)(18). In each transformation, the abnormal values (outliers) were excluded by Smirnov–Grubbs' test. Data were expressed as means ± SD.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
assay evaluation
Determination of glycated albumin by the modified BAC method.
When the glycated albumin was assayed by the modified BAC method, nonadsorbed fractions were eluted in the first peak; adsorbed fractions were eluted with sorbitol buffer in the second peak. The nonadsorbed fraction was completely separated from the adsorbed fraction.

To evaluate nonspecific binding of albumin molecules to boronic acid resin, referred to here as the column blank, we applied nonglycated albumin solution to the boronate column. The amount of fructosamine in the nonglycated albumin solution (concentration: 30 g/L) was negligible (10.9 µmol/L). The furosine content of the acid hydrolysate of the nonglycated albumin was below the detection limit (<0.01 mol/mol of protein) by amino acid analysis. As shown in Table 1 , 8.1% of the nonglycated albumin bound to lot A of boronate column. Rechromatography of the nonadsorbed fraction on the same column showed that 8.3% of the totally applied nonglycated albumin was again adsorbed to the column. On the other hand, 18.9% of the nonglycated albumin bound to lot B of boronate column. These results indicate that the column blank differs from one batch of boronate column to another. The percentage of the glycated albumin adsorbed was 48.5 on lot A and 58.6 on lot B. After subtracting the column blank value from the apparent value, the corrected glycated albumin values were practically identical, 40.4% and 39.7%, indicating that the column blank was the source of large errors in this assay procedure. Thus, subtracting the column blank from the apparent glycated albumin value provided accurate values for glycated albumin by the modified BAC method.

Correlation of glycated albumin results determined by ELBIA and by the modified BAC method.
Fig. 2 demonstrates excellent correlation between ELBIA and the modified BAC method in estimating the concentration of glycated albumin. The correlation coefficients between ELBIA and the modified BAC method in measuring glycated albumin (n = 22) and albumin fractions purified from human sera (n = 6) were 0.995 and 0.991, respectively. The regression line for measurements of both kinds of samples together was y = 0.996x 0.846 (n = 28, r = 0.993). Thus, the values determined by ELBIA reflect the true value for glycated albumin.

View larger version (23K): [in this window] [in a new window]   Figure 2. Correlation between glycated albumin values determined by ELBIA and the modified BAC method. Glycated albumin values of glycated albumin solutions (n = 22) or albumin fractions purified from human sera (n = 6) were determined by ELBIA and the modified BAC method as described in the text. The regression line was y = 0.996x + 0.846 (n = 28, r = 0.993, Syx = 1.532).

Correlation between glycated albumin measured in untreated serum samples and that in purified albumin fractions.
To examine whether it is necessary to isolate the albumin fraction from serum before assaying, we compared the amount of glycated albumin determined by ELBIA in 8 samples of untreated serum with that in the purified albumin fraction (Fig. 3 ). The results for glycated albumin in the purified albumin fraction correlated well (r = 0.976) with those in untreated serum, the mean difference being 1.1% (range 0.3–2.6%). Evidently, therefore, the presence of other serum proteins does not affect the measurement of glycated albumin by ELBIA, and purification of the albumin fraction from serum before the assay is not necessary.

View larger version (19K): [in this window] [in a new window]   Figure 3. Correlation between glycated albumin values of untreated serum and those of purified albumin fraction. Glycated albumin values in human sera (n = 8) were determined by ELBIA before and after purification of the albumin fraction as described in the text. y = 0.890x + 1.771 (r = 0.976; Syx = 1.301).

Precision studies of ELBIA.
In the fully automated ELBIA system, the intraassay CVs (n = 92) were 3.7% and 3.5% for sera with glycated albumin values of 11.5% and 21.4%, respectively. The interassay CV (n = 5) was 4.2% for serum samples with glycated albumin contents of 14.2% (Table 2 ). In the manual ELBIA, the intraassay CVs were 6.0%, 6.3%, and 3.0% for sera with glycated values of 6.2%, 13.3%, and 24.9%, respectively. The respective interassay CVs were 7.0%, 2.7%, and 2.6% (Table 3 ).

clinical studies
Figure 4 A shows the correlation between glycated albumin in serum samples as determined by manual and automated ELBIA. The mean interassay difference was 1.1% (range 0.0–2.7%). Thus, glycated albumin values measured by manual or automated ELBIA were indistinguishable from each other. We also examined the correlation between glycated albumin measured by the HP-BAC method and those determined by automated ELBIA (Fig. 4B ) and manual ELBIA (Fig. 4C ). For the regression shown in Fig. 4B , the mean difference between the two methods was 15.2% (range 10.9–20.5%) For the methods compared in Fig. 4C , the mean difference was 11.2% (range 2.3–17.1%). In either case, the glycated albumin values determined by the HP-BAC method were much higher than those determined by ELBIA.

View larger version (16K): [in this window] [in a new window]   Figure 4. Clinical studies showing the relation between glycated albumin values determined by ELBIA and the HP-BAC method: (A) manual vs automated ELBIA, y = 0.992x - 0.462 (n = 48, r = 0.973, Syx = 1.346); (B) HP-BAC method vs automated ELBIA, y = 0.798x - 8.164 (n = 51, r = 0.973, Syx = 1.351); (C) HP-BAC method vs manual ELBIA, y = 0.883x - 8.196 (n = 300, r = 0.972, Syx = 2.075).

To establish the reference interval for glycated albumin, we used ELBIA to measure glycated albumin in 112 samples from healthy subjects. The histogram of the glycated albumin values is shown in Fig. 5 . According to the maximum likelihood method, x^1/2-transformed data for glycated albumin were most suitably approximated by normal (guassian) distribution. Therefore, the type of distribution for glycated albumin in healthy subjects is most likely log-normal. After deletion of abnormal outliers according to Smirnov–Grubbs' test, the reference interval (mean ± SD) for glycated albumin was 5.26% ± 0.96% (n = 110).

View larger version (18K): [in this window] [in a new window]   Figure 5. Histogram of glycated albumin values of healthy subjects (n = 112, ages 47.6 ± 8.7 years) determined by ELBIA, showing a log-normal distribution. After deletion of abnormal values (solid columns) according to Smirnov–Grubbs' test, the mean (±SD) glycated albumin value (n = 110) was 5.26% ± 0.96%.

Figure 6 shows the correlation between the values for glycated albumin determined by automated ELBIA and the stable Hb A_1c values. The glycated albumin values ranged from 1.1% to 47.8%, whereas the range for stable Hb A_1c was 4.2% to 17.0%.

View larger version (18K): [in this window] [in a new window]   Figure 6. Correlation between stable Hb A1c and glycated albumin determined by automated ELBIA: y = 2.904x - 9.783 (n = 470, r = 0.876, Syx = 4.004).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

 
analytical considerations
We here demonstrate that a newly developed affinity method for quantification of glycated albumin by ELBIA is accurate, precise, and clinically useful. With respect to determination of glycated albumin by the affinity method, two practical problems should be resolved: first, the nonspecific binding of albumin molecules to boronate resin, and second, the overestimation of glycated albumin by the HP-BAC method, mainly because of insufficient separation of nonadsorbed from adsorbed fractions (14). These two issues were solved in the present study, as described below.

Nonspecific binding of albumin molecule to boronate anion.
As shown in Table 1 , nonglycated albumin was found to bind to boronate-packed columns in a nonspecific manner, the nonspecific binding being expressed as the column blank. The nonglycated albumin solution used in the present study, prepared by extensive adsorption of glycated albumin with repeated use of boronic acid resin, showed a negligible amount of glycation, both by fructosamine assay and amino acid analysis. The nonadsorbed fraction was completely separated from the adsorbed fraction by boronate column chromatography. Rechromatography of the nonadsorbed fraction showed that the size of the column blank depended on the column used, probably according to the amount of boronate conjugated to the resin (Table 1 ). The column blank may be derived, at least in part, from the ionic interaction between the cationic portion of the albumin molecule and the boronate anions present in the weak alkaline conditions used in the chromatographic procedure (2)(3). Therefore, the nonspecific binding of nonglycated albumin to boronic acid led to substantial experimental errors, as reflected by the glycated albumin values (Table 1 ). Thus, subtraction of the column blank from the measured glycated albumin is necessary for accurate estimation of glycated albumin in the sample.

In addition, nonspecific binding of nonglycated albumin to boronate may also produce experimental errors in ELBIA. Accordingly, nonspecific binding of boronate to nonglycated albumin is subtracted as a blank to estimate the proportion of glycated albumin present. Our results showed an excellent correlation between glycated albumin values measured by ELBIA with those determined by the modified BAC method (Fig. 2 ), indicating that ELBIA provides accurate measurement of glycated albumin. Furthermore, the correlation was linear up to a glycation proportion of 60%, all values that are within the practical range. Moreover, we also confirmed that serum samples could be used directly in ELBIA, without extraction or purification, since the presence of other serum proteins did not influence the measured value of glycated albumin (Fig. 3 ). Thus, our findings indicate that ELBIA provides an accurate measurement of glycated albumin in clinical samples.

Reproducibility of glycated albumin measurements by automated ELBIA.
Quick procedures, stable temperature, and operator skill are necessary for precise measurements by ELBIA. To overcome these limitations, a fully automated ELBIA system (AP-960 GA) has recently been developed that provides strict control of a variety of conditions such as time and temperature. All wells of the microtiter plate are maintained at one temperature by an incubator equipped with a thermoelectric cooling element. As shown in Fig. 4A , there was a good correlation between glycated albumin values measured by manual and automated ELBIA. Furthermore, the precision in the automated measurement was similar to that of the manual procedure (Tables 2, 3), but did not require skilled labor. Thus, precise, rapid, and multiple determinations of glycated albumin can be provided by the clinical laboratory using this method.

Overestimation of glycated albumin by HP-BAC.
Fujita et al. (14) have already pointed out that glycated albumin values for healthy subjects determined by the HP-BAC method were higher than those determined by other methods. The reference value for glycated albumin measured by the HP-BAC method has been reported as 20.2% (10) or 16.1% (11). On the other hand, the reference values determined by methods other than BAC were fairly low: 7.0% by ion-exchange chromatography and 8.3% by the thiobarbituric acid method (1). In the present study, the mean glycated albumin in apparently healthy subjects (n = 110) measured by ELBIA was 5.26% ± 0.96%, comparable with that obtained by ion-exchange chromatography or thiobarbituric acid. Assaying glycated albumin solution as a sample, Fujita et al. (14) demonstrated that the peaks of adsorbed and nonadsorbed fractions were not separated very well in the HP-BAC method and that prolonging the period required for separation of the two peaks diminished the estimated glycated albumin value. They concluded that insufficient separation—in the rapid chromatographic procedure of the HP-BAC method—led to overestimation of glycated albumin values. Determination of the glycated albumin values of diabetic patients as well as nondiabetic subjects by ELBIA and by HP-BAC (Figs. 4B , C) showed that the values determined by the HP-BAC method were noticeably higher than those determined by manual or automated ELBIA—indicating that serum concentrations of glycated albumin determined by the HP-BAC method are also overestimated.

clinical application of glycated albumin measurements
Glycated albumin is now widely used for clinical evaluation of hyperglycemia, in addition to stable Hb A_1c. Glycated albumin is expected to reflect short-term (a few weeks) control of hyperglycemia in diabetes, whereas stable Hb A_1c is expected to reflect a much longer control period (a few months) (7)(19). As shown in Fig. 6 , glycated albumin values ranged from 1.1% to 47.8% compared with stable Hb A_1c values in the range of 4.2–17.0%. The 3-fold wider dynamic range for glycated albumin than for stable Hb A_1c, as estimated by linear regression analysis (y = 2.904x - 9.783) for glycated albumin vs stable Hb A_1c, supports the contention that glycated albumin is more sensitive to changes in glycemia. With ELBIA, monitoring of glycated albumin is clinically feasible and should allow frequent evaluation of diabetic control. Because the relative extents of glycation of different plasma proteins are a complex function of integrated glucose concentrations over time and of the half-life and chemical characteristics of each protein (20), some amino residues of albumin may be more susceptible to binding glucose than are the amino residues of hemoglobin. Furthermore, environmental factors might also affect the extents of glycation of these two proteins; albumin is a plasma protein, whereas hemoglobin is an intracellular protein. Thus, the discrepancy in extents of glycation between albumin and hemoglobin appears to reflect differences in intrinsic susceptibility to glycation as well as differences in environmental factors.

The "labile" fraction, Schiff base adduct, has been shown to increase the results of Hb A_1c measurements by as much as 10–20% when ion-exchange chromatography is used, but has a negligible effect in boronate affinity chromatography (5). In the present study, Schiff base adducts on glycated albumin were removed by extensive dialysis, although its exact amount could not be measured. At this point, the exact amount of Amadori product in the glycated solutions cannot be easily determined. Furthermore, the proportions of glycation, i.e., the numbers of glycated residues in each albumin molecule, all differ, making it very difficult to match the glycation value between in vivo and in vitro samples. To estimate the extent of glycation, we incubated nonglycated albumin solution with radiolabeled glucose (D-[U-C]glucose; Amersham) in the same manner as the glycated solutions, then subjected the radiolabeled glycated albumin to BAC analysis. The amount of glucose incorporated into the adsorbed fraction was 0.98-1.18 mol/mol of protein in each glycated fraction. Furosine concentrations in the glycated albumin solutions were <1 mol/mol of protein. Thus, one glycated residue per albumin molecule might be enough for reaction with boronate and allow determination of the percent of glycation for the solutions glycated by the modified BAC method.

To examine the effect of increasing BAC assay modification, further glycated albumin solutions were prepared by incubating for longer times. As the extent of glycation increased to exceed a practical range (>80%), the percent glycation value of the albumin increased up to saturated value of 100%, as determined by modified BAC method. In contrast, the value determined by ELBIA or fructosamine kept increasing without saturation, suggesting that each glycated residue was effectively recognized by these two methods (data not shown). This was also observed at the practical range in the present study. As shown in Fig. 4C , >90% of the datapoints with albumin glycation values >45% clearly are situated on one side of the line. These data might suggest that the ELBIA method was effectively measuring more than two glycated residues.

As the Diabetes Control and Complications Trial Research Group concluded (21), intensive antidiabetes therapy effectively delays the onset and slows the progression of diabetic complications. To achieve a good control of the glycemic status, frequent monitoring of diabetes is essential. For this purpose, glycated albumin is more suitable as a sensitive indicator than stable Hb A_1c. Glycated albumin may also be valuable for monitoring unstable glycemic conditions such as diabetes of pregnancy or the induction of antidiabetes therapy (8)(9)(13). Serial measurements of glycated albumin in patients experiencing considerable changes in glycemic control, e.g., newly diagnosed patients undergoing initial treatment and diabetic patients undergoing changes in treatment regimen, would be needed to test this notion.

Because >90% of the datapoints with glycation values >25% are situated on one side of the regression line, a curvilinear regression would probably be more appropriate for reporting results. This suggests that as glycation increases, ELBIA continues to proportionately determine its value, whereas the fraction of Hb A_1c might be saturated, especially at a high glucose concentration (5). In ion-exchange separations, the variant of interest, namely Hb A_1c, is known to represent a particular glucose modification that occurs at the N-terminal of the ß-chain. This species elutes faster than the main A₀ fraction on cation-exchange columns, thereby providing the means for quantification. Given, however, the demonstration by Klenk et al. (5) that the adsorbed fraction of hemoglobin to boronate affinity columns contained a heterogeneous population of hemoglobin variants, the affinity method may more accurately represent the extent of glycation, especially in diabetic individuals.

In conclusion, we have demonstrated that glycated albumin values determined by ELBIA reflect the real values of glycated albumin and that ELBIA is useful for determination of glycated albumin. The assay is accurate, precise, and clinically useful and the fully automated system allows rapid and precise measurement.

	Acknowledgments

 
We are grateful to Wasaku Koyama, Kumamoto Health Care Center Japanese Red Cross, and Takayuki Higashi, Hiroyuki Sano, and Yoshiteru Jinnouchi of our laboratory for collaborative endeavors and helpful discussion during the course of this study. We also thank F.G. Issa, Department of Medicine, University of Sydney, Sydney, Australia, for careful reading and editing of the manuscript.

	Footnotes

 
² Author for correspondence. Fax +81 (96) 364-6940; e-mail horiuchi{at}gpo.kumamoto-u.ac.jp.

¹ Nonstandard abbreviations: ELBIA, enzyme-linked boronate-immunoassay; HSA, human serum albumin; BAC, boronate affinity chromatography; Hb A_1c, glycohemoglobin; HP-BAC, high-performance boronate affinity chromatography; and OGTT, oral glucose tolerance test.

	References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. Guthrow CE, Morris MA, Day JF, Thorpe SR, Baynes JW. Enhanced nonenzymatic glucosylation of human serum albumin in diabetes mellitus. Proc Natl Acad Sci U S A 1979;76:4258-4261. [Abstract/Free Full Text]
2. Weith HL, Wiebers JL, Gilham PT. Synthesis of cellulose derivatives containing the dihydroxyboryl group and a study of their capacity to form specific complexes with sugars and nucleic acid components. Biochemistry 1970;9:4396-4401. [Medline] [Order article via Infotrieve]
3. Okayama H, Ueda K, Hayashi O. Purification of ADP-ribosylated nuclear proteins by covalent chromatography on dihydroxyboryl polyacrylamide beads and their characterization. Proc Natl Acad Sci U S A 1978;75:1111-1115. [Abstract/Free Full Text]
4. Brownlee M, Vlassara H, Cerami A. Measurement of glycosylated amino acids and peptides from urine of diabetic patients using affinity chromatography. Diabetes 1980;29:1044-1047. [Abstract]
5. Klenk DC, Hermanson GT, Krohn RI, Fujimoto EK, Mallia AK, Smith PK, et al. 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. [Abstract/Free Full Text]
6. Reed P, Bhatnagar D, Dhar H, Winocour PH. Precise measurement of glycated serum albumin by column affinity chromatography and immunoturbidimetry. Clin Chim Acta 1986;161:191-199. [Web of Science][Medline] [Order article via Infotrieve]
7. Day JF, Ingebretsen CG, Ingebretsen WR, Baynes JW, Thorpe SR. Nonenzymatic glucosylation of serum proteins and hemoglobin: response to changes in blood glucose levels in diabetic rats. Diabetes 1980;29:524-527. [Abstract]
8. Nelson DM, Barrows HJ, Clapp DH, Ortman-Nabi J, Whitehurst RM. Glycosylated serum protein levels in diabetic and nondiabetic pregnant patients: an indicator of short-term glycemic control in the diabetic patient. Am J Obstet Gynecol 1985;151:1042-1047. [Web of Science][Medline] [Order article via Infotrieve]
9. Morris MA, Grandis AS, Litton J. The correlations of glycosylated serum protein and glycosylated hemoglobin concentrations with blood glucose in diabetic pregnancy. Am J Obstet Gynecol 1985;153:257-260. [Web of Science][Medline] [Order article via Infotrieve]
10. Shima K, Ito N, Hirota M, Yano M, Yamamoto Y, Uchida T, et al. High-performance liquid chromatographic assay of serum glycated albumin. Diabetologia 1988;31:627-631. [Web of Science][Medline] [Order article via Infotrieve]
11. Yasukawa K. High-performance affinity chromatography system for the rapid, efficient assay of glycated albumin. J Chromatogr 1992;597:271-275. [Web of Science][Medline] [Order article via Infotrieve]
12. Akazawa Y, Tsujii S, Tashiro S. Glycated serum albumin: an indicator of short-term glycemic control in diabetic patients. J Med Pharm Sci 1993;29:269-274.
13. Wörner W, Pfleiderer S, Rietbrock N. Selective determination of non-enzymatic glycosylated serum albumin as a medium term index of diabetic control. Int J Clin Pharmacol 1993;31:218-222.
14. Fujita S, Katayama Y, Horiuchi S. The difference of glycated albumin value between high-performance liquid affinity chromatography and enzyme immunoassay. J Anal Bio-Sci 1995;18:113-119.
15. Miyake T, Tanaka K. Enzyme reagent for detecting glycated proteins. J Med Pharm Sci 1992;28:871-875.
16. Ikeda K, Higashi T, Sano H, Jinnouchi Y, Yoshida M, Araki T, et al. N-(Carboxymethyl)lysine protein adduct is a major immunological epitope in proteins modified with advanced glycation end products of the Maillard reaction. Biochemistry 1996;35:8075-8083. [Medline] [Order article via Infotrieve]
17. Iizuka Y, Kume H, Kitamura M. A new procedure for determining the normal range. Jpn J Clin Pathol 1979;27:158-162.
18. Usui T. Evaluation of distribution type selection for normal range calculation. Jpn J Clin Chem 1983;12:201-207.
19. Day JF, Thornburg RW, Thorpe SR, Baynes JW. Nonenzymatic glucosylation of rat albumin: studies in vitro and in vivo. J Biol Chem 1979;254:9394-9400. [Free Full Text]
20. Austin GE, Mullins RH, Morin LG. Non-enzymic glycation of individual plasma proteins in normoglycemic and hyperglycemic patients. Clin Chem 1987;33:2220-2224. [Abstract/Free Full Text]
21. . The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes. N Engl J Med 1993;329:977-986. [Abstract/Free Full Text]