The glycation gap and estimated glomerular filtration rate in individuals without diabetes mellitus.

The glycation gap (GG), the difference between measured glycosylated hemoglobin (Hb A1c) and Hb A1c predicted from fructosamine, has been used to quantify nonglycemic differences in formation of Hb A1c between individuals, but the technique is controversial (1 ). An increased GG in patients with diabetes has been associated with nephropathy diagnosed primarily by proteinuria (2– 4 ). The GG can be calculated irrespective of glycemia, that is, in normoglycemic as well as hyperglycemic individuals. We therefore hypothesized that, in individuals without diabetes, chronic kidney disease (CKD) would be associated with an increased GG compared with those without CKD. We studied a nonprobability convenience sample, after applying exclusion criteria, of 949 patients in the community whose CKD stage was based on estimated glomerular filtration rate (eGFR) (5 ). Over a 4-month period, we analyzed samples from primary care mostly on receipt and all within 24 h. Exclusion criteria were diabetes mellitus, pregnancy, age 16 years, hemoglobinopathy, unknown ethnicity or not of white or South Asian ethnicity, hemoglobin 9.0 g/dL, and albumin 3.0 or 5.0 g/dL. Laboratory analyses were performed by routine methodology in an accredited laboratory with acceptable and stable performance throughout the study period. Hemoglobin (flow cytometry, Sysmex XN-10, Sysmex Corp.), Hb A1c (ion-exchange HPLC, National Glycohemoglobin Standardization Program certified, G7 HPLC analyzer, Tosoh Corp.), fasting glucose (hexokinase), creatinine (compensated kinetic Jaffe with rateblanking), albumin (bromocresol green), and fructosamine (nitrotetrazolium blue) were measured (Modular P analyzer, Roche Diagnostics). CKD stages 3, 3, 4, and 5 were based on eGFR levels of 60, 30 –59, 15–29, and 15 mL min 1 (1.73 m) , respectively. All patients were in CKD stage 3 except for 2 (0.2%) with CKD 4. Because of the small numbers of those in CKD stage 3, the 2 patients were removed, with further analysis examining only those in stage 3 and below. Permission for data to be published has been granted by the Royal Wolverhampton NHS Trust Caldicott Guardian. We calculated the GG by converting fructosamine into a standardized normal deviate by subtracting the mean fructosamine from each value and dividing the result by the SD of fructosamine (4 ). This value was then converted into an estimate of Hb A1c by multiplying it by the SD of Hb A1c and adding the mean Hb A1c (4 ). From the measured Hb A1c for each sample, we subtracted its respective estimated Hb A1c to calculate the GG (4 ), and the GG was compared between those with CKD stage 3 and those with CKD stage 3. We then tested for the association between GG and presence of CKD stage 3 (yes/no) using multivariable forced logistic regression modeling with the presence of CKD 3 (yes/ no) as the outcome measure (Table 1, model 1). Independent variables known to affect the GG (4 ) were controlled for in this model including age, sex, ethnicity (white or South Asian), hemoglobin, fasting glucose, and fructosamine (Table 1, model 1). To address concerns (1 ) with regard to GG calculation and interpretation, a second model (Table 1, model 2) was run with GG excluded and Hb A1c included to test the influence of Hb A1c on CKD with the influence of fructosamine and glucose statistically removed. Multicollinearity diagnostics of variance inflation factors confirmed reasonable independence before analysis. Of 947 patients, 793 (83.7%) had CKD stage 3 (58% female) and 154 (16.3%) had CKD stage 3 (55% female) (Table 1). Those with CKD stage 3, compared to those with CKD stage 3, were older, but despite higher glycemic markers, their GG was similar (Table 1). When controlled for independent variables and compared to CKD stage 3, the GG was not associated with CKD (model 1), and neither was Hb A1c (model 2). The only variable that significantly predicted CKD stage was age. These data suggest that the GG is not associated with CKD in nondiabetic individuals, defined according to eGFR.