The relative value of glycated albumin, hemoglobin A 1c and fructosamine when screening for diabetes mellitus

Abstract

We compared the usefulness of three glycated serum proteins, glycated albumin (GA), glycated hemoglobin (HbA 1c) and fructosamine (FA), for diabetic screening purposes. We measured these indices in 302 adults, most of whom underwent yearly physical examinations. We measured GA and HbA 1c with high precision using high-performance liquid chromatography (interassay coefficients of variation 4.9 and 4.0%, respectively) and FA using commercial reagents (interassay coefficient of variation 1.65%). All the individuals underwent a 75-g oral glucose tolerance test, which revealed significant correlations between the values of the three glycated proteins and the four plasma glucose concentrations measured as well as the sum of these glucose concentrations, ΣBS (GA, r = 0.80; HbA 1c, r = 0.80; FA, r = 0.65). On the basis of the test, 130 of the subjects were classified as normal (N), 123 as borderline and 49 as having diabetes mellitus (D) according to the criteria of the Japan Diabetes Society. Of the 123 borderline cases, 26 showed impaired glucose tolerance (IGT) according to the WHO criteria. The normal group values of GA, HbA 1c and FA were 17.8 ± 0.17% (mean ± SEM), 5.02 ± 0.03%, and 2.55 ± 0.02 mM/l, respectively. Borderline and IGT subjects had significantly more GA and HbA 1c than normal but not more FA ( P < 0.01). We divided the subjects into 10 groups on the basis of their ΣBS values; those with values higher than 671 ± 4.7 mg/dl had significantly more GA and HbA 1c than normal, while those with values higher than 1068 ± 40.9 mg/dl (the most extreme cases) had significantly more FA. The GA level was a moderately sensitive (81.6%) indication of diabetes, but not of borderline diabetes (13%) or IGT (42.3%). HbA 1c was a more sensitive (91.8%) indicator than GA, whereas FA was much less sensitive (53.1%). Only five (3.8%) of the 130 N subjects had GA levels above normal. The other two parameters were also highly specific. We conclude that the plasma levels of GA and HbA 1c, as measured by the methods used in this study, can identify most of the people who are at risk of diabetes, but that the level of FA is unsuitable for screening.