Measuring glycated proteins: clinical and methodological aspects

The recognition that nonenzymatic glycation, wherein free sugar condenses with certain reactive protein amino groups, is increased in hyperglycemic states has led to improved means for assessing diabetic control through the measurement of concentrations of glycated proteins in the circulation. This article reviews the chemistry, methods for measurement, and clinical relevance of circulating glycated proteins in the management of diabetes.

Serum fructosamine concentration as an index of glycemia in cats with diabetes mellitus and stress hyperglycemia

The purpose of this study was to evaluate fructosamine concentrations in clinically healthy cats, sick cats with stress hyperglycemia, and untreated diabetic cats to determine the usefulness of this test in diagnosing diabetes mellitus in cats, and in differentiating the disease from stress-induced hyperglycemia. In addition, we evaluated if the degree of glycemic control in cats treated for diabetes influenced their serum fructosamine concentrations. In the 14 sick cats with stress hyperglycemia, the median serum fructosamine concentration (269 mumol/L) was not significantly different from the median value in the 26 clinically normal cats (252 mumol/L). Two of the 14 cats with stress hyperglycemia (14.3%) had serum fructosamine concentrations above the upper limit of the reference range (175 to 400 mumol/L); on the basis of these results, the test specificity was calculated as 0.86. In 30 cats with untreated diabetes mellitus, the median serum fructosamine concentration was 624 mumol/L, markedly higher than the value in either the normal cats or the cats with stress hyperglycemia. All but 2 of the 30 untreated diabetic cats (6.7%) had serum fructosamine concentration above the upper limit of the reference range; on the basis of these results, the sensitivity of serum fructosamine concentration as a diagnostic test for diabetes mellitus was 0.93. When 30 diabetic cats receiving treatment were divided into 3 groups according to their response to treatment (i.e., poor, fair, and good), the 16 cats that had a good response to treatment had significantly lower serum concentrations of both glucose and fructosamine compared with cats that had either a fair or poor response to treatment. A significant correlation (rs = .70, n = 100, P < .001) was found between serum concentrations of glucose and fructosamine. Results of this study indicate that quantification of serum fructosamine concentration is a meaningful test for the diagnosis of diabetes, for differentiating diabetes from stress hyperglycemia; and for monitoring the metabolic control in treated diabetic cats.

Effect of aging on plasma 1,5-anhydroglucitol levels in humans and rats

Since normal reference values change with age in some clinical parameters, we measured the plasma levels of 1,5-anhydroglucitol (AG), a new marker of glycemic control in diabetes mellitus, in healthy subjects and in rats. Our results showed a significantly negative correlation of the marker with age in humans and that the plasma AG levels of older rats were markedly lower than those of younger counterparts. This remarkable reduction of AG in the older rat group can be partially explained by our finding that aged animals excreted AG more rapidly in the urine than younger ones, besides a decrease in food intake. We therefore suggest that normal clinical reference values for plasma AG levels should be modified according to age.

Influence of serum proteins on fructosamine concentration in multiple myeloma

Serum fructosamine levels in 36 subjects with various types of multiple myeloma and in 64 normal controls were evaluated by means of a Nitroblue tetrazolium colorimetric assay. Only the IgA myeloma group showed significantly raised serum fructosamine values (P less than 0.001). In the IgG myeloma group, which showed a higher mean serum protein concentration, serum fructosamine levels were not significantly different from controls. The study shows that elevated IgA levels do influence serum fructosamine and this effect should be taken into due consideration in order to avoid possible misinterpretations in evaluating this widely used index of glucose metabolism.

Comparison of reliability of plasma fructosamine and glycosylated hemoglobin assays for assessing glycemic control in diabetic patients on hemodialysis

To search for a reliable marker of medium-term integrated blood glucose level in diabetics on maintenance hemodialysis (HD), plasma fructosamine and glycosylated hemoglobin (Hb) levels were determined every week and blood glucose levels were determined four times every day over 3 weeks. The mean values of fructosamine (mol/L per 40 g of albumin) and of glycosylated Hb of other the study period correlated (r = .746, P less than .001) for combined materials of diabetic and nondiabetic subjects. However, plasma fructosamine values at the end of the study period did not correlate with the overall mean blood glucose values during the preceding 8 to 21 days (r = .372, NS). In contrast, glycosylated Hb values correlated closely with the same mean blood glucose values (r = .703, P less than .001). Fructosamine values significantly decreased during a HD, irrespective of the increases in albumin and total protein. In conclusion, glycosylated Hb was a reliable marker of long-term integrated blood glucose even in diabetics on HD. However, fructosamine was not a reliable marker of medium-term integrated blood glucose in these patients.

Correction of fructosamine value for serum albumin and globulin concentrations

Using the data of 131 patients with non-insulin dependent diabetes mellitus (NIDDM), the correction formulas of fructosamine value ([FRA]) were devised to standardize the uncorrected [FRA] to serum albumin concentrations ([ALB]) of 4 g/dl, globulin concentrations ([GLB]) of 3 g/dl, and total protein concentrations ([TP]) of 7 g/dl. The following formula was derived for its maximum correlation coefficient (r) between corrected [FRA] ([FRAc]) and fasting blood glucose concentration ([G]); [FRAc] = [FRA]x33.3/(7.6 [ALB]+[GLB]). In these 131 diabetic patients, r between uncorrected [FRA] and [G] at 2 weeks ago was 0.562. When corrected by [FRA]x33.3/(7.6 [ALB]+[GLB]), [FRA]x4/[ALB], [FRA]+30 (4-[ALB]), [FRA]+23 (4-[ALB]), [FRA]+30 (7-[TP]), [FRA]x7/[TP], and [FRA]x3/[GLB], r was, respectively, 0.616, 0.612, 0.595, 0.589, 0.582, 0.581 and 0.478. In 24 patients with NIDDM whose [ALB] is either above 4.5 g/dl or below 3.5 g/dl, r between uncorrected [FRA] and [G] was as low as 0.389 without positive correlation. By using our correction formulas of [FRA]x33.3/(7.6 [ALB]+[GLB]) or [FRA] x 4/[ALB], r was statistically increased, respectively, to 0.769 or 0.788 (P less than 0.05 in both cases) in contrast to no significant increase of r by other formulas being at 0.598, 0.556, 0.540, 0.562 and 0.121. Based on these analyses, it is concluded that our correction formula of [FRA] by [FRAc]=[FRA]x33.3/(7.6 [ALB]+[GLB]) accurately reflects [G] in NIDDM even with hypo- or hyper-albuminemia, and [FRAc]=[FRA]x4/[ALB] is useful for practical application for its simplicity.

Longitudinal changes in serum fructosamine do not parallel those in glycated haemoglobin in young adults with insulin-dependent diabetes

In 93 adolescent and young adult patients with type 1 diabetes (163 paired comparisons) changes in fructosamine concentration correlated poorly with changes in HbA1 (r = 0.53); this correlation was no better if fructosamine values were adjusted for serum albumin by calculating a fructosamine/albumin index, F.A.I. (155 pairs, r = 0.50). These correlation coefficients were lower than those for cross-sectional comparisons (HbA1 vs. fructosamine, r = 0.74; HbA1 vs. F.A.I., r = 0.80). One-way analysis of variance showed that values of HbA1, fructosamine and F.A.I. all tended to increase as glycaemic control, judged by the clinician, worsened (P less than 0.001). HbA1 correlated better with clinical assessment than did either uncorrected fructosamine or F.A.I. We conclude that changes in fructosamine from one clinic visit to the next do not provide a basis for changing therapy. Clinical assessments tend to agree with values for HbA1, which may be more reliable than fructosamine because of its longer biological half life.

Maternal serum fructosamine and maternofetal glucose and insulin homeostasis in normal pregnancy

Well-defined normal values are necessary to identify pregnancies complicated by gestational diabetes (GD) and thus further reduce perinatal morbidity and mortality from this condition. The present study defined the range for the oral glucose tolerance test (oGTT) in 2578 pregnancies. After exclusion of abnormal results 822 randomized patients were used to define normal values for fructosamine, HbA1c, insulin, glucose and C-peptide in the maternal serum; insulin, glucose and fructosamine in the amniotic fluid; and insulin, glucose, C-peptide and fructosamine in the cord blood.

Clinical utility of assays of glycosylated haemoglobin and serum fructosamine compared: use of data on biological variation

Assessment of the relative clinical usefulness of glycosylated haemoglobin (HbA1) and serum fructosamine is complicated by their markedly different half-lives. They have therefore been compared using a number of indices derived from data on biological variation, as applied in a study of blood glucose control in newly diagnosed diabetic patients. The within-subject (CVI) and between-subject (CVG) variation of fructosamine and HbA1 were assessed in 8 stable diabetic patients. A slightly smaller relative change, or critical difference, is required between serial HbA1 results than between fructosamine results before a significant change can be said to have occurred. The heterogeneity of within-subject variance is also less for HbA1, rendering the critical difference more generally applicable. HbA1 may therefore be more appropriate for long-term monitoring. Monitoring of blood glucose control of a group of 26 newly diagnosed diabetic patients before the commencement of therapy and 1, 2, and 3 months later by serum fructosamine or HbA1 provided very similar information. Fructosamine responds more rapidly with changes of blood glucose control. However, whether this confers any clinical advantage will depend upon the frequency of testing.

A comparison of direct measures of glycaemia and glycated blood proteins in insulin-dependent diabetes mellitus

We have studied associations between various direct measures of glycaemia and glycated blood proteins in 113 subjects with insulin-dependent diabetes mellitus (IDDM), and examined whether or not the 'fructosamine' assay results were affected by differing patient serum concentrations of lipids, albumin or C peptide. Serum fructosamine correlated less closely with HbA1 (r = 0.44) than did HbA1 with glycated serum albumin (GSA) (r = 0.68). Serum fructosamine and GSA also were poorly correlated (r = 0.48). Although fructosamine, HbA1 and GSA correlated to a similar degree with fasting blood glucose (r range 0.34 to 0.37), GSA was most closely related to mean blood glucose (r = 0.39 vs. 0.30-0.35) and the M value (a marker of diurnal glycaemic instability) (r = 0.42 vs. 0.33-0.35). The serum concentration of fructosamine was not significantly affected by a variation in serum cholesterol, but tended to be lower in subjects with moderate hypertriglyceridaemia (p = 0.05). The fructosamine assay may be altered by moderately lipaemic serum but is not affected by serum albumin concentration in normoalbuminaemic patients with IDDM. Our study indicates, however, that GSA is a more reliable marker of short-term glycaemic control in IDDM than fructosamine.

Technical and clinical evaluation of fructosamine determination in serum

We evaluated a serum fructosamine (glycated serum proteins) assay for efficacy in the diagnosis and follow-up of diabetic patients. A Roche reagent kit, based on nitroblue tetrazolium reduction in alkaline medium, was used in COBAS FARA centrifugal analyzer. We demonstrated that this method is precise, linear and unaffected by serum hemolysis. However, bilirubin affected the test positively and lipemia negatively. Fructosamine (F) correlated positively with total protein (P) (r = 0.809) and albumin (r = 0.746) in a group of 48 non-diabetic individuals. A good correlation was observed between F and glycated hemoglobin from the sera of 514 patients (r = 0.794). A better correlation (r = 0.838) was obtained when F was corrected for P concentration (F/P). Different F and F/P means were calculated only in patients with overt diabetes, compared to normals. Gestational diabetes was associated with a highly significant F increase. However, its low sensitivity (21%) precludes the use of F as an effective screening test for that condition. Nevertheless, because of its simplicity, low cost and rapidity in reflecting changes in the metabolic control of diabetes, F should be considered a valuable test to assess glycemic control in diabetic patients.

The effect of hypoalbuminaemia, hyperbilirubinaemia and renal failure on serum fructosamine concentration in non-diabetic individuals

We have investigated the effects of hypoalbuminaemia, hyperbilirubinaemia and renal failure on serum fructosamine concentration in 39 non-diabetic patients. All patients were hypoalbuminaemic (median serum albumin 25 g/l, range 12-34 g/l). Group 1 (n = 19) were patients with hypoalbuminaemia alone, group 2 (n = 7) with hypoalbuminaemia and impaired renal function (median serum creatinine 226 mumol/l, range 154-461 mumol/l) and group 3 (n = 13) were subjects with hypoalbuminaemia and hyperbilirubinaemia (median serum bilirubin 34 mumol/l, range 19-83 mumol/l). Serum fructosamine was significantly lower in all three groups compared to age-matched normoalbuminaemic controls, but there was no significant difference in fructosamine concentrations between the groups. There was a correlation between fructosamine concentration and serum albumin. (r = 0.82, p less than 0.001) in all three groups combined. Serum fructosamine did correlate with serum bilirubin in patients with normal renal function (r = 0.0, p less than 0.001). In patients with abnormal renal function there was no correlation between serum fructosamine and either urea (r = 0.22, ns) or creatinine (r = 0.31, ns). Albumin is the major factor affecting serum fructosamine concentrations. Moderate hyperbilirubinaemia does not affect fructosamine concentration. No difference in fructosamine concentration could be demonstrated in patients with renal failure.

[Critical evaluation of fructosamine as a control parameter in the assessment of diabetic metabolic regulation]

For some years the glycosylated plasma proteins, the so-called fructosamines, have been used for the evaluation of metabolic control in diabetic patients. We studied the usefulness of fructosamine as an intermediate parameter in the control of diabetes and we also corrected fructosamine for serum protein. Furthermore, the dependence on daily times and amounts of food intake were examined for fructosamine alone as well as for protein-corrected fructosamine. Both parameters showed a statistically significant correlation with HbA1C (r = 0.7; P less than 0.001). Changes in the metabolic state of the diabetic patients were reflected more rapidly by both these parameters than by HbA1C. While the correction of fructosamine for serum protein eliminated its postprandial increase in the diabetic patients, circadian variation remained unchanged.

Clinical usefulness of estimation of serum fructosamine concentration as screening test for gestational diabetes

Serum fructosamine levels and fructosamine/protein ratios were measured in 100 pregnant women who underwent glucose tolerance tests because of clinical risk. Compared with normal pregnant women, the 13 study participants with gestational diabetes had higher fructosamine/protein levels (39 +/- 3.9 mumol/gm versus 37 +/- 3.2 mumol/gm, p less than 0.05), fasting serum glucose levels (107 +/- 13.7 mg/dl versus 82 +/- 8.6 mg/dl, p less than 0.001), and area under curve of glucose tolerance test (36 +/- 5 gm x min x dl-1 versus 22 +/- 3.6 gm x min x dl-1, p less than 0.001). The serum fructosamine levels were not significantly different between the two groups of participants (2.3 +/- 0.26 mmol/L versus 2.2 +/- 0.17 mmol/L); 10 of the 13 women with diabetes had a fructosamine/protein ratio within 2 SD of the mean of the groups of normal pregnant women. Spontaneous caloric intakes (r = 0.72, p less than 0.005) and the hospital mean daily capillary glucose levels during diabetic diet (r = 0.72, p less than 0.005) correlated better with the fructosamine/protein ratio than with fasting serum glucose levels (r = 0.58, p less than 0.05) and area under curve (r = 0.57, p less than 0.05). Consequently, serum fructosamine and fructosamine/protein ratio levels should be considered insensitive as a screening test in pregnant patients with clinical risk of gestational diabetes.

Serum fructosamine in uraemia, myeloma and acute inflammatory disorders--relationship to serum glucose and albumin levels

There has been considerable interest in the serum fructosamine assay as a measure of glycated serum proteins. We have measured serum fructosamine in three groups of patients--those with uraemia; those with multiple myeloma; and those with acute inflammatory conditions--none of whom were known to have diabetes. Serum fructosamine was significantly higher in the uraemic group than in the other two, and also than in a control group. When allowance was made for prevailing serum albumin levels fructosamine was shown to be increased in the acute inflammatory group also. There was a significant correlation between random plasma glucose and serum fructosamine only when fructosamine was adjusted for prevailing albumin levels. In control and uraemic subjects there was a significant positive correlation between serum fructosamine and albumin levels, whereas in the myeloma group there was a negative correlation with serum protein. These data would suggest the need to take into account serum albumin levels and protein composition if serum fructosamine is accurately to reflect short-term integrated glycaemia.

Plasma fructosamine assay in children with insulin-dependent diabetes mellitus

Plasma concentrations of fructosamine, an indicator of glycated plasma proteins, were measured in non-diabetic children and children with insulin-dependent diabetes mellitus (IDDM) to see if they also correlate with glycemic control in children as well as in adults. Non-diabetic children aged less than 4 yr had significantly lower plasma fructosamine than non-diabetic children aged 4 or more. Total plasma protein in these children was slightly lower or close to that of older children. There was no difference in fructosamine between non-diabetic children aged 4 or more and healthy adult subjects. Plasma fructosamine in children with IDDM was twofold that of age-matched controls. In children with IDDM, correlations between fructosamine and HbAI (r = 0.799) or HbAIc (r = 0.841) were high. The measurement of plasma fructosamine, which is practical in children because of the small sample volume needed and no influence of HbF, is useful in the management of children with IDDM.

Clinical utility of serum fructosamine in diabetes mellitus compared with hemoglobin A1c

To evaluate the clinical utility of fructosamine as a mean of monitor glycaemic control, fructosamine and HbA1c were compared in 46 random out-patients visiting a Diabetic Clinic as well as in 25 inpatients admitted to a Diabetes Day Care Unit. In the out-patients, there were a significant correlations between fructosamine and fasting blood glucose (r = 0.75) as well as between fructosamine and HbA1c (r = 0.91). However, when the reference values were considered, interesting differences were found; only 4% of the out-patients showed normal HbA1c values while 39% showed normal fructosamine values. Accordingly, fructosamine and HbA1c evaluate different aspects of glycaemic control. During an admission of 7 days to the Diabetes Day Care Unit no statistical changes in mean blood glucose and fructosamine values occurred. On the other hand, two weeks after discharge from the Unit, not only fructosamine (3.58 +/- 0.16 mmol vs 3.09 +/- 0.08 mmol/l) but also HbA1c (9.52 +/- 0.38% vs 8.33 +/- 0.23%) had improved significantly. Thus HbA1c measures improvements in glycaemic control as early as 3 weeks after changes in treatment. At six weeks after discharge HbA1c (7.63 +/- 0.34%) but not fructosamine (3.02 + 0.14 mmol/l) had improved further. HbA1c is a reliable marker of glycaemic control while the value of fructosamine in clinical practice is unclear.

The effect of calibration on the between-laboratory variation of serum fructosamine

Human pooled serum from diabetic and non-diabetic subjects, and calibrators containing glycated protein or 1-deoxy-1-morpholinofructose (DMF) in a variety of matrices, were distributed to 10 laboratories. When they used their own assay conditions and calibrators, the inter-laboratory variation was unacceptably high. However, when the pool from diabetic patients was reassayed using a calibrator with an assigned value prepared from freeze-dried human serum, and containing no DMF, inter-laboratory variation was reduced significantly. Inter-laboratory agreement for the pool from non-diabetic subjects remained poor despite recalibration. Recalibration using either serum or albumin based solutions of DMF as calibrator failed to effect any significant reductions in inter-laboratory variation. Secondary calibrators based on a protein matrix with no added DMF are recommended for routine use.

A proficiency score for assessing the reliability of self blood glucose monitoring

The reliability of self blood glucose analysis was studied in 83 insulin-treated diabetic patients. Reliability was assessed by laboratory analysis of capillary blood samples collected simultaneously on filter paper. Patient and laboratory results were compared using a scoring system which produces an easily understood Proficiency Score. The score was designed to highlight unreliable performance in measuring glucose concentrations in the euglycaemic as well as hypo- and hyperglycaemic ranges. Only 17% of patients achieved acceptable Proficiency Scores by maintaining adequate performance on five occasions.

The effect of altered thyroid function on serum fructosamine concentrations

The effect of altered thyroid function on serum fructosamine concentrations was investigated in 31 untreated hyperthyroid patients, 18 short-term hypothyroid patients (i.e., 20 days after withdrawal of thyroid hormone suppressive therapy for thyroid cancer), 7 untreated long-term hypothyroid patients, and 25 age-matched normal controls. No differences in serum fructosamine concentrations were observed between hyperthyroid patients and normal controls; conversely, serum fructosamine concentrations were significantly higher both in short-term and in long-term hypothyroid patients than those found in normal controls. Furthermore, long-term hypothyroid patients showed significantly higher serum fructosamine concentrations than short-term hypothyroid patients. L-Thyroxine (L-T4), replacement therapy in two hypothyroid patients, resulted in a marked decrease in serum fructosamine concentrations. In seven hyperthyroid patients, the restoration of euthyroidism with antithyroid drug therapy was associated with no significant changes in serum fructosamine concentrations. The results of the present study indicate that hypothyroidism is associated with a marked increase in serum fructosamine concentrations. This alteration does not appear to be the consequence of gross abnormalities in plasma protein or glucose metabolism. The duration of hypothyroidism seems to be an important factor, even though the mechanism underlying this alteration remains at present unexplained. These results also suggest that caution must be used in the interpretation of elevated serum fructosamine concentrations as an index of the metabolic control of diabetes mellitus in the presence of hypothyroidism.

Plasma fructosamine in non-diabetic pregnancy

Plasma fructosamine and its relation to plasma total protein and albumin was examined in 40 non-diabetic pregnant women. Plasma fructosamine did not correlate with haemoglobin A1 or plasma albumin. The mean intra-individual ranges were plasma fructosamine 0.53 mmol/l (SD 0.17); haemoglobin A1 1.42% (SD 0.42); plasma fructosamine/g albumin 0.024 mmol (SD 0.005) and plasma fructosamine/g total protein 0.008 mmol (SD 0.003). A high degree of individuality for these variables was also observed.

The prevalence of diabetes mellitus and an assessment of methods of detection among a community of elderly Chinese in Hong Kong

This study provides information on the prevalence of diabetes mellitus in a group of elderly Chinese subjects aged 60 and above living in the community in Hong Kong, and investigates the sensitivity of the urine sugar, random blood glucose, glycosylated haemoglobin, and fructosamine measurements compared to a glucose tolerance test in screening for diabetes mellitus in this population. Four hundred twenty-seven subjects aged 60 and above were studied. The National Diabetes Data Group Criteria were used for the diagnosis of diabetes. Those with a random blood glucose exceeding 12.5 mmol/l were considered diabetic, and all of these patients had glycosuria together with elevated total glycosylated haemoglobin and fructosamine concentrations. A diagnostic 75 g oral glucose tolerance test was performed on patients with one or more of the following abnormalities: glycosuria, random plasma blood glucose 7.8 mmol/l to 12.5 mmol/l, glycosylated haemoglobin 8.5%, and fructosamine 2.20 mmol/l. By these criteria, the prevalence of diabetes in this community was found to be 9.8%. An elevated random glucose greater than 11.1 mmol/l proved to be the only specific method of screening and glycosuria was found to be at least as good as fructosamine and HbA1. However, HbA1 is more sensitive than random glucose, glycosuria or fructosamine in detecting impaired glucose tolerance. On the basis of this study, a higher reference range for glycosylated haemoglobin for the elderly alone is also suggested (5.74-9.34%).

Does the measurement of serum fructosamine accurately reflect levels of glycated albumin in insulin-dependent diabetes?

We have measured serum glycated albumin (GSA) by affinity chromatography and immunoturbidimetry, and serum fructosamine using a Cobas FARA analyser in blood samples from 37 type I diabetics and 21 healthy controls. Random blood glucose and glycated haemoglobin levels were also measured. Glycated haemoglobin (HbA1) correlated with glycated albumin and fructosamine in the diabetic group. A less clear relationship was found between glycated albumin and fructosamine. HbA1, GSA and fructosamine correlated poorly with random blood glucose levels. These data indicate that serum fructosamine levels do not accurately reflect those of glycated albumin, as has recently been suggested, in type I insulin-dependent diabetics where glycaemic control fluctuates more than in type II diabetics. It is postulated that the two methods reflect varying glycaemic levels to a different degree, thereby accounting for the disparity.

Glycosylated serum proteins and glycosylated haemoglobin in normal pregnancy

Glycosylated haemoglobin (HbA1) and glycosylated serum protein concentrations (fructosamine) were monitored in 162 normal pregnancies. The response to the standard 75 g oral glucose tolerance test (OGTT) was used to exclude gestational diabetes. A significant increase in glucose intolerance, measured from the area under the OGTT-curve (r = 0.29, P less than 0.0005) and a highly significant decrease in serum albumin concentration (r = -0.79, P less than 0.0005) due to the redistribution of body water, as occurs during pregnancy, resulted in a statistically significant decrease in serum fructosamine concentration (r = -0.23, P less than 0.01) when related to gestational age in weeks.

The measurement of glycosylated albumin by reduction of alkaline nitro-blue tetrazolium

A method is described for the measurement of glycosylated albumin in human plasma. Purification of albumin is achieved by a simple one-step chromatographic method, adapted for use as a batch analysis technique. The glycosylated albumin fraction is subsequently measured by a modification of the fructosamine assay (alkaline reduction of nitroblue tetrazolium). The method correlates with the plasma fructosamine assay (r = 0.85; p less than 0.001) and with an electrophoretic method for the measurement of HbA1 (r = 0.78; p less than 0.001).