Glycation Gap Is Associated With Macroproteinuria but Not With Other Complications in Patients With Type 2 Diabetes

The cohort included 925 adult inpatients with type 2 diabetes who had been referred to the Diabetes Department of Jean Verdier Hospital (Bondy, France) between 1994 and 2011. None had been admitted for an acute disease or had recently modified lifestyle or treatment. All of them had fructosamine, HbA_1c, and creatininemia measurements. To avoid any influence on fructosamine measurement (8), we excluded individuals with renal failure (creatinine clearance <60 mL/min) (9). The patients had no known hemoglobinopathy or erythrocyte disorder.

Nephropathy was defined as incipient nephropathy or overt macroproteinuria. Incipient nephropathy was defined as urinary albumin excretion rate between 30 and 299 mg over 24 h, and macroproteinuria was defined as urinary protein excretion rate ≥300 mg over 24 h on at least two measurements. Diabetic retinopathy was graded according to the Early Treatment of Diabetic Retinopathy Study severity scale and defined as absent or present. The diagnosis of peripheral neuropathy was based on the presence of any two or more of the following: neuropathic symptoms; decreased distal sensation; or decreased or absent ankle reflexes. Cardiac autonomic neuropathy was assessed in a subgroup of 512 patients using the following three tests as recommended (10): Valsalva; deep breathing; and lying to standing. Age was taken into account for test interpretation as previously described (11). Cardiac autonomic neuropathy was defined by at least two abnormal tests.

A subset of 506 patients was screened for silent myocardial ischemia. These patients had no cardiac symptoms, no history of coronary artery disease, a normal 12-lead resting ECG, and at least one of the following additional cardiovascular risk factors: dyslipidemia (total cholesterol >6.5 mmol/L or LDL cholesterol >4.1 mmol/L or both, HDL cholesterol <0.9 mmol/L, triglycerides >2.3 mmol/L or lipid-lowering medication or both); hypertension (systolic and diastolic blood pressure ≥140/90 mmHg or antihypertensive therapy); smoking; nephropathy; family history of premature coronary artery disease; and peripheral or carotid occlusive arterial disease. As previously reported (12–15), each patient underwent a ²⁰¹Tl myocardial scintigraphy after an ECG stress test or a pharmacological stress test (dipyridamole injection) or both. An ECG stress test was performed in the patients who could exercise on a bicycle ergometer and were expected to have an interpretable exercise ECG. When the patient was unable to exercise or when the ECG stress test result was indeterminate, a pharmacological stress test using dipyridamole was performed. Silent myocardial ischemia was defined as an abnormal ECG stress test or an abnormal myocardial scintigraphy imaging, i.e., defects in at least three out of 17 segmental regions, or both.

All the other measurements were performed on the second day of hospitalization, at fasting. Glucose was measured on venous plasma by the glucose oxidase method (colorimetry, Kone Optima; Thermolab System, Paris La Défense, France). HbA_1c measurement was based on a turbidimetric inhibition immunoassay principle and total hemoglobin was measured using a modified alkaline hematin reaction (Dimension Technology; Siemens Healthcare Diagnostics). The intra-assay and interassay coefficients of variation were, respectively, 1.3 and 1.2% for a normal blood sample (mean HbA_1c 5.3%) and 1.5 and 2.3% for an elevated HbA_1c level (mean HbA_1c 8.6%). Reliability was tested daily using internal controls (low and high levels of a Biorad control). The same HbA_1c measurement technology was performed throughout the period of inclusion (16). Fructosamine was measured by the nitroblue tetrazolium colorimetric procedure based on the reducing ability of fructosamine in alkaline solution (COBAS; Roche Diagnostics Gmbh, Penzberg, Germany) (17). The intra-assay and interassay coefficients of variation were 1.2 and 1.6%, respectively.

The following measurements also were performed: creatinine and serum albumin (colorimetric assay, COBAS; Roche Diagnostics); urinary albumin excretion rate (mean of two measurements, laser immunonephelometry, BN100; Dade-Behring); total cholesterol; HDL cholesterol; and triglycerides (enzymatic colorimetry, Hitachi 912; Roche Diagnostic, Meylan, France). LDL cholesterol was calculated using the Friedwald formula. The accuracy of all the methods used was evaluated biannually by national (external) quality-control surveys.

Continuous variables were expressed as means ± SD values and compared by one-way ANOVA or the Mann-Whitney U test as adequate. The significance of differences in proportions was tested with the χ² test. Logistic regression was used for multivariate analyses based on a model including the factors that were associated with a diabetes complication with a P < 0.10 in univariate analyses. Because HbA_1c and G-Gap were highly correlated, they could not be entered together in the logistic regression models. Therefore, we matched patients with or without each complication for HbA_1c to evaluate the potential association of this complication with G-Gap, independently of HbA_1c. Statistical analyses were performed using SPSS software (SPSS, Chicago, IL). P = 0.05 was considered for statistical significance.

The main clinical and biological characteristics of our population are reported in Table 1. The prevalence of retinopathy was 37.1%, prevalence of nephropathy was 28.2%, including macroproteinuria 8.5%, prevalence of peripheral neuropathy was 46.7%, and prevalence of cardiac autonomic neuropathy was 30.9%. Silent myocardial ischemia was found in 148 patients (29.2%).

G-Gap was calculated as the difference between measured HbA_1c and HbA_1c predicted from fructosamine based on the HbA_1c–fructosamine regression equation (4–6): predicted HbA_1c = 0.020 × fructosamine + 2.05 (r = 0.731; P < 0.0001) (Fig. 1). The patients then were separated by G-Gap tertiles (Fig. 1 and Table 1). Increasing G-Gap tertiles were positively associated with female sex, BMI, the presence of dyslipidemia, HbA_1c, fructosamine, and fasting and postprandial glucose values. Regarding diabetes complications, increasing G-Gap tertiles were associated with nephropathy, urinary albumin excretion rate, and creatinine clearance, whereas no association with either retinopathy or peripheral neuropathy or cardiac autonomic neuropathy or silent myocardial ischemia was found (Table 1). The association with nephropathy was driven by the association with macroproteinuria because there was no association between tertiles of G-Gap and incipient nephropathy was considered separately (prevalence of incipient nephropathy in the patients of the first, second, and third tertiles of G-Gap: 20.8, 23.2, and 25.0%, respectively; P = 0.521).

Macroproteinuria was associated with male sex, diabetes duration, hypertension, dyslipidemia, G-Gap, HbA_1c, and fructosamine (Table 2). Compared with the first tertile of G-Gap considered as reference (2.9%), the prevalence of macroproteinuria did not differ significantly in the second tertile (6.2%; P = 0.06) and was higher in the third tertile (11.0%; odds ratio [OR] 4.1 [95% CI 1.9–9.2]; P < 0.001). G-Gap was greater in the patients with macroproteinuria than in those without (1.06 ± 1.62% vs. 0.03 ± 1.30%; P < 0.0001). Urinary albumin excretion rate (log) was correlated with G-Gap (r = 0.141; P < 0.0001). HbA_1c and G-Gap could not be considered together in multivariate analyses because HbA_1c was used to determine G-Gap. In a multivariate analysis including male sex, diabetes duration, hypertension, dyslipidemia, and HbA_1c to explain macroproteinuria, all the parameters but diabetes duration and HbA_1c were independently associated with macroproteinuria. When male sex, diabetes duration, hypertension, dyslipidemia, and high G-Gap were considered, all but diabetes duration also were associated with macroproteinuria, including the third tertile of G-Gap (2.1 [1.4–3.0]; P < 0.001).

Nephropathy was associated with male sex, diabetes duration, hypertension, G-Gap, and HbA_1c (Table 2). Compared with the first tertile of G-Gap considered as reference (23.1%), the prevalence of nephropathy was similar in the second tertile (28.0%; P = 0.18) and higher in the third tertile (33.2%; OR 1.65 [95% CI 1.1–2.4]; P < 0.01). In multivariate analysis including male sex, diabetes duration, hypertension, BMI, and HbA_1c to explain nephropathy, all the parameters, including HbA_1c (1.1 [1.1–1.2]), were independently associated with nephropathy. When male sex, diabetes duration, hypertension, BMI, and a high G-Gap were considered, all were also associated with nephropathy, including the third tertile of G-Gap (1.6 [1.1–2.2]).

Because HbA_1c and G-Gap are highly linked together, the association between G-Gap and nephropathy might be attributable to an association between HbA_1c and nephropathy. We therefore matched for HbA_1c 54 patients with macroproteinuria with 200 patients without macroproteinuria (Table 3). HbA_1c levels were similar in both groups. Macroproteinuria was associated with male sex, hypertension, lower albumin and fructosamine levels, higher G-Gap, and the third tertile of G-Gap (Table 3). Because of the potential influence of macroproteinuria on lower levels of serum albumin and fructosamine and, subsequently, on higher G-Gap levels, we corrected fructosamine values for variations in the concentrations of serum albumin (18) according to the following formula by Lamb et al. (19): albumin-indexed fructosamine = fructosamine (µmol/L) × 100/albumin (g/L). We then defined albumin-indexed predicted HbA_1c based on the HbA_1c/albumin-indexed fructosamine regression equation: 0.008 × albumin-indexed fructosamine + 2.109 (r = 0.747; P < 0.0001). Albumin-indexed G-Gap and its tertiles were then calculated as the difference between measured HbA_1c and albumin-indexed HbA_1c. Table 3 shows that macroproteinuria was associated with albumin-indexed G-Gap and its third tertile. Compared with the first tertile of albumin-indexed G-Gap considered as reference (11.5%), the prevalence of macroproteinuria was higher in the second (24.1%; OR 2.4 [95% CI 1.02–5.8]; P < 0.05) and third (30.8%; 3.4 [1.5–7.9]; P < 0.01) tertiles. In a multivariate analysis including sex, diabetes duration, hypertension, dyslipidemia, and the third tertile of albumin-indexed G-Gap, male sex, hypertension, and albumin-indexed G-Gap were independently associated with macroproteinuria (Table 3). When the same analysis was performed with albumin-indexed G-Gap rather than with its third tertile, albumin-indexed G-Gap also was independently associated with macroproteinuria (1.6 [1.2–2.1]; P < 0.01).

We also matched for HbA_1c 216 patients with nephropathy with 216 patients free of nephropathy (HbA_1c 8.5 ± 1.9% in both groups). G-Gap was similar in those with or without nephropathy (third tertile compared with first or second tertiles of G-Gap: prevalence of nephropathy 38.0 vs. 34.7%; P = 0.548).

There was a trend for an association between G-Gap tertiles and retinopathy or peripheral neuropathy (Table 1). Both complications also were associated with higher HbA_1c levels (retinopathy HbA_1c 8.7 ± 2.0% compared with no retinopathy HbA_1c 8.1 ± 1.2%; P < 0.0001; and neuropathy HbA_1c 8.6 ± 2.1% compared with no neuropathy HbA_1c 8.1 ± 1.9%; P < 0.0001). Therefore, we also matched for HbA_1c 260 patients with retinopathy and 260 without (HbA_1c in both groups 8.4 ± 1.8%). G-Gap was not associated with retinopathy (third tertile versus first or second tertiles of G-Gap: prevalence of retinopathy 50.5 vs. 49.7%; P = 0.972). We finally matched for HbA_1c 190 patients with peripheral neuropathy and 190 without (HbA_1c in both groups 8.3 ± 1.7%). G-Gap was not associated with peripheral neuropathy (third tertile versus first or second tertiles of G-Gap: prevalence of neuropathy 48.8 vs. 50.6%; P = 0.827).

We have shown here for the first time in a large cohort of patients with type 2 diabetes that a high G-Gap was associated with macroproteinuria but not with other complications. This association appears to be independent of HbA_1c, albumin levels, and other confounders. Because G-Gap may reflect nonglycemic determinants of hemoglobin glycation, also called “glycability” (5), these results suggest that intracellular susceptibility for glycation in the glomerulus may be involved in the development of this diabetes complication.

McCarter et al. (20) previously have reported an association between hemoglobin glycation index and nephropathy among the type 1 diabetic patients of the Diabetes Control and Complications Trial. However, those data have been criticized because hemoglobin glycation index was by mathematical necessity linked with HbA_1c. Because a reanalysis of those data showed no significant influence of hemoglobin glycation index on the risk of nephropathy if HbA_1c was taken into account, this association was actually assumed to reflect an association between nephropathy and HbA_1c per se (21). Using G-Gap in a sample of 40 type 1 diabetic patients, Cohen et al. (4) reported an association between G-Gap and nephropathy. The regression between G-Gap and nephropathy was not altered when HbA_1c, fructosamine, and diabetes duration were included as factors in the analysis (4). In the current study, nephropathy was associated with HbA_1c and G-Gap, whereas HbA1c and G-Gap were positively and highly correlated. Therefore, we matched for HbA1c patients with or without macroproteinuria to avoid any HbA_1c effect. This procedure did not affect this association. Thus, these data demonstrate a correlation between a high G-Gap and prevalent macroproteinuria, independently of HbA_1c levels.

A high G-Gap may be a marker or predictor of nephropathy and also a consequence of nephropathy. Because other complications did not appear to be associated with G-Gap, specific mechanisms related to nephropathy might explain a high G-Gap. First, renal failure might explain between-patient differences in the mean age of circulating erythrocytes and, thus, variation in HbA_1c levels (22). Such a mechanism is unlikely to be involved in our study because we did not include patients with kidney failure. Furthermore, a similar prevalence of renal insufficiency was reported in the lowest and highest tertiles of G-Gap in a large cohort of type 2 diabetic patients (5). Second, glycated albumin, i.e., fructosamine, also may be less excreted than nonglycated albumin, especially in diabetic patients with macroproteinuria (23). With this hypothesis, fructosamine would be higher in case of proteinuria, which was not the case here. Furthermore, fructosamine was reported to remain unchanged with nephropathy stages in type 1 (4) and type 2 diabetes (5). Finally, the lower level of fructosamine reported here in the patients with macroproteinuria appeared to account for a higher G-Gap, but only partially, because the levels of HbA_1c were concomitantly higher. Third, urinary protein excretion may reduce serum albumin levels and, therefore, fructosamine levels (18). The association between G-Gap and nephropathy was driven in our study by macroproteinuria but not by incipient nephropathy, which could be in favor of this hypothesis. Such an effect was previously modeled considering body albumin daily synthesis and proteinuria, with a negative result (4). In our subpopulation with or without macroproteinuria matched for HbA_1c, we have shown that there was a 2.3-fold increased risk of macroproteinuria in the third tertile of albumin-indexed G-Gap in the multivariate analysis considering the other confounders, i.e., male sex, hypertension, dyslipidemia, and diabetes duration. Therefore, urinary protein loss may contribute to a lower fructosamine level and, therefore, a higher G-Gap; however, this is likely to be insufficient to explain by itself the association between macroproteinuria and G-Gap level.

The prospective study by Rodríguez-Segade et al. (5) provides arguments for a role played by G-Gap in the pathophysiology of nephropathy. They reported in a cohort of 2,314 type 2 diabetic patients with a 6.5-year follow-up that G-Gap predicted the progression of nephropathy, independently of fructosamine, even after adjustment of HbA_1c (5). In that study, Cox regression analyses failed to show if HbA_1c or G-Gap was the best predictor, and we think that the associations of both are important. Because hemoglobin is an intracellular protein, and because fructosamine reflects extracellular proteins, a high G-Gap is considered as a marker of intracellular susceptibility for glycation. Such a mechanism therefore could contribute to nephropathy in type 1 and type 2 diabetes. Besides pathophysiological interest, G-Gap in clinical practice may be considered as a marker of proteinuria susceptibility and may lead to more preventive efforts in the patients with the highest G-Gap. This implies measuring, at least once, fructosamine together with HbA_1c to calculate the G-Gap.

Whether this process involves in a similar way tissues other than the renal glomerulus is unknown. There are only few data about other diabetes complications, and these data were reported only for type 1 diabetes. Regarding to retinopathy, the association between retinopathy and hemoglobin glycation index reported in patients from the Diabetes Control and Complications Trial (20) appeared to be attributable to HbA_1c (21). In a substudy from the Wisconsin Diabetes Registry Study, the probability of developing retinopathy increased with increasing G-Gap (7). We definitely did not find any independent association between retinopathy and G-Gap in this series of type 2 diabetic patients. Finally, we searched, for the first time, for an association between G-Gap and peripheral or cardiac autonomic neuropathy and silent myocardial ischemia in diabetic patients and failed to find any significant association. G-Gap therefore appears to play an important role only for overt nephropathy.

There are some limitations. The study was cross-sectional. We included only inpatients; therefore, the results are not necessarily generalizable to the diabetic population. Furthermore, we did not include patients with renal failure to improve the reliability of fructosamine measurement. G-Gap depends on HbA_1c and fructosamine levels; therefore, G-Gap variations may be explained by the different half-lives of its components. For example, a patient with improved glycemic control during the 2 weeks preceding measurement would decrease fructosamine level more than the HbA_1c value. As a consequence, the G-Gap level would be higher. This was not applicable to our patients because they were free of any acute condition at admission as checked with hospitalization code, and they had led a stable lifestyle during the weeks before admission. Furthermore, intraindividual G-Gap value has been reported to be consistent over time (4–6), with a genetic component (24).

To conclude, we demonstrate here for the first time in type 2 diabetic patients that overt nephropathy is associated with a high G-Gap, independently of HbA_1c levels and other confounders. G-Gap may be partly attributable to macroproteinuria, but data favor its involvement in renal disorders. A high G-Gap may represent tissue susceptibility to the risk associated with protein glycation and the formation of advanced glycation end products (25). This phenomenon appears to be restricted to the kidneys in type 2 diabetes.

This research was made possible by a grant from Merck Lipha santé, Lyon, France. No other potential conflicts of interest relevant to this article were reported.

E.C. directed research and wrote the manuscript. I.B., C.C.-.P., S.C., Y.J., and N.C. researched data and contributed to discussion. Q.C. researched data. M.T.N. researched data and performed the statistic analyses. P.V. directed research, contributed to discussion, and reviewed and edited manuscript. P.V. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.