Although GA testing may initially be viewed as an adjunct to A1c in the management of diabetes, its immediate applications are apparent as monthly tool for diabetes management. GA offers a similar reference discrepancy as A1c with a half-life on the order of 3 weeks rather than the 3 to 4 months of A1c. It also allows physicians to achieve better glycemic control and finally results in a superior index to use for reducing diabetic micro- and macrovascular complications. Amidst the ongoing debates about the interpretation of A1c, there is no consensus on a suitable cutoff value for A1c across different racial or ethnic populations. In a recent study on racial differences in glycemic markers, Selvin et al. [32] indicated that differences between African-American and Caucasian population with respect to GA, FA, and 1,5-AG levels parallel the differences in A1c among both non-diabetic and diabetic subjects. This result might imply a similarity between GA and A1c on racial and ethnic differences.

GA is now gaining popularity as an index of glycemic control during intensive treatment [33]. Won et al. [34] recently showed a significant direct correlation between changes in GA over a 3-week period and those in A1c over a 3-month period after intensified substitution of anti-diabetic drugs (oral hypoglycemic agent-based [r=0.735, P<0.01] and insulin-based [r=0.778, P<0.01]). A prospective, randomized, controlled trial has shown that initiation of once-daily basal insulin or biphasic insulin analogue followed by an intensified twice-daily biphasic insulin analogue for non-respondent subjects with greater than 20% GA at 3 weeks after the initiation of therapy yielded a significantly higher A1c reduction with guidance of interim GA values [35]. Thus, the use of GA as an intermediate-term index contributed in a guiding role to the control of glycemia. In addition, if the proper management of gestational diabetes or type 1 or 2 diabetes during pregnancy was found to be monthly diabetes management, GA could act as a standard glycation index. Thus, well-designed clinical studies for subjects with gestational diabetes are needed to determine the optimal glucose monitoring interval benefiting pregnant subjects and their offspring.

Even after proof of concept that intermediate-term glycemic indicators are beneficial and effective in achieving better glycemic control, leading to reduced diabetic complications through guiding glucose control, questions could be raised about which kind of glycemic marker would best satisfy the aforementioned unmet needs. As described previously, FA has physiologic disadvantages in that its levels are strongly influenced by the concentration of serum proteins and low-molecular-weight substances coexisting in the plasma [6,7]. FA has been scrutinized by the Food and Drug Administration in the United States, and a few clinical studies with mixed results were reported [36]. These clinical and practical data have led to its flagging popularity recently. During pregnancy, 1,5-AG is not a suitable parameter for glycemia management due to unstable renal hemodynamics [36].

Several studies have reported discrepancies in A1c as a standard indicator for glycemic control in patients with diabetes-associated pathologies. The diseases and pathological conditions in which GA is a more accurate glycation index might be diabetic subjects with hemolytic anemia, end-stage renal disease (ESRD), and iron deficiency. GA may also be more accurate for diabetic subjects during pregnancy [6,10,33,36]. Of the numerous pathologic conditions, studies targeting diabetic patients with ESRD have been well-designed and frequently conducted. Because of an altered erythrocyte lifespan caused by renal anemia, A1c measurements are low with respect to glycemic status in patients with ESRD. These findings issued concerns that physicians may underestimate and under-treat hyperglycemia in patients with chronic renal failure [36]. There is, however, no relevant evidence established by randomized controlled studies demonstrating that GA is better in preventing morbidity and mortality than A1c in this high-risk diabetic population. Though there is a lack of studies of abnormal albumin homeostasis in patients with ESRD, it is not difficult to imagine that there are potential fluctuations in GA in this population. In these regards, Mehrotra et al. [37] stated that it might be more reasonable to solve this drawback by obtaining consensus on different cutoffs for A1c in subjects with end-stage renal disease.

In contrast to A1c, GA itself is strongly involved in the development of major diabetic complications, including arterial stiffening, peripheral vascular calcification, nephropathy, retinopathy, and Alzheimer's disease [10]. Notwithstanding these merits, two important questions need to be answered through further prospective studies for GA to be considered a standard glycation index in these populations. First, evidence that changes in GA show better association with time-dependent proportional morbidities and mortality than those of A1c is required. Second, quarterly GA measurements need to assist physicians in achieving and maintaining better glycemic control and result in reducing diabetic complications in these populations.

Despite a very strong relationship between A1c and GA, some large discrepancies between the two values exist under certain physiologic and pathologic conditions. Because A1c and GA levels are dependent on the lifespan of erythrocytes and half-life of serum albumin, respectively, variability in the lifespan of erythrocytes and half-life of albumin between individuals could account for some unexplained GA/A1c ratio [38]. Physiologically, the rate of the nonenzymatic-glycation process on protein in vivo is approximately nine times faster than that of human hemoglobin [39], resulting in a ten-fold greater glycation reaction speed in albumin than in hemoglobin [40]. The conventional conversion of GA levels to A1c levels is often performed using the relationship A1c=GA/3, which is empirical and has a statistical basis [38]. Clinical reports in Korea and Japan have shown a wide range of mean ratios of GA to A1c (from 2.0 to 4.0) [34,38,41-44]. These differences in mean ratios of GA to A1c may originate from the heterogeneity of the study population, especially for A1c levels and stable or unstable status of glucose excursions. The mean GA/A1c ratio of those who participated in a periodic health examination who were mostly free of diabetes was 2.68 [42]. However, in studies targeting type 2 diabetics, including those with unstably controlled A1c with >0.5% fluctuation in A1c across 6 months, serum GA levels increased by 3% to 4% for every 1% increase in serum A1c levels [33,38]. Therefore, that GA/A1c ratio increased along with higher A1c may be attributable to more marked increases in GA levels than A1c levels in subjects with poorly controlled diabetes. Even among similar A1c levels, GA better reflected postprandial hyperglycemia [6,33,45], which is mainly caused by inadequate or dysfunctional endogenous insulin secretion. Koga et al. [41] measured GA and A1c in type 1 and 2 diabetes and found that the GA/A1c ratio was significantly higher in type 1 diabetic patients, in which glucose levels fluctuate over wider ranges, than in type 2 diabetic patients. Furthermore, Kim et al. [44] demonstrated that insulin secretory functions, such as HOMA-β and insulinogenic index, but not insulin resistance, associates with the ratio of GA to A1c in type 2 diabetics. The ratio of GA to A1c is increased as insulin secretory function is diminished. The reason behind these findings may lie in that fact that postprandial insulin secretory dysfunction in response to meal loading causes postprandial hyperglycemia, which directly leads to an augmented GA/A1c ratio. In a clinical view of this phenomenon, a physician could interpret the higher GA/A1c ratio as a greater postprandial glucose fluctuation due to deteriorating insulin secretory function. By assessing the meaning of the GA/A1c ratio, physicians may clinically relate the relevance of the higher ratio in situations of similar mean glucose levels. Because elevated levels of GA induce irreversible vascular damage in diabetes mellitus, further studies on the clinical relevance of higher GA/A1c ratios on diabetic complications would satisfy the unmet need of biologic impact of GA on cell pathophysiology. Thus, serial results guide its usefulness for management of these patients in clinical settings.

However, some physiological variables, such as body mass index or age, could affect GA levels. Also, in certain pathologic conditions affecting albumin metabolism, such as thyroid dysfunction, nephrotic syndrome, or liver cirrhosis, the ratio of GA to A1c should be interpreted with caution [10,46]. We summarized the clinical implication of GA/A1c (Table 1).

Despite the potential benefits of GA, on the lack of normal reference data on GA might limit its use as a diagnostic tool for diabetes. In contrast to A1c, GA levels are not influenced by sex [42,47]. In 2006, only the Japan Diabetes Society reported the reference intervals of GA (12.3% to 16.9%) in a normal population [42]. Recently, Hiramatsu et al. [48] reported reference intervals of GA in 676 healthy Japanese pregnant women of 11.5% to 15.7%. Furusyo et al. [42] measured fasting plasma glucose, A1c, and GA in 1,575 mostly non-diabetic Japanese subjects. Using receiver operating characteristic curve (ROC) analysis, GA ≥15.5% was optimal for predicting diabetes with a sensitivity of 83.3% and a specificity of 83.3%. Another Chinese study enrolled 1,971 outpatient subjects who underwent a 75 g oral glucose tolerance test and GA measurement and found a GA value of 17.1% to be an optimal cutoff in Chinese subjects using ROC analysis with 76.82% sensitivity and 76.89% specificity for the diagnosis of diabetes [49].

As previously described, GA is a more accurate glycation index to control glucose in certain diabetes-associated conditions, and further studies should be directed to assess the optimal cutoff value and clinical relevance in diabetic subjects with hemolytic anemia, ESRD, iron deficiency, and pregnancy.

In understanding the biologic impact of GA on diabetes and its complications, one should bear in mind that GA is a precursor of AGEs [50]. As previously described, high levels of glucose induce non-enzymatic glycation reactions between glucose and derivatives with free amine groups of albumin. These reactions successively produce a reversible unstable Schiff base, a stable fructosamine residue (ketoamine), an irreversible Amadori compound, and ultimately irreversible conjugates, called AGEs. Elevated levels of GA and AGEs are known to be actively involved in the development of diabetic vascular complications, such as retinopathy, nephropathy, neuropathy, and coronary artery disease [50-53]. Because of the approximately 17-day half-life of GA, GA may not reach the total levels of AGEs. However, GA is directly associated with the production of ROS, resulting in cell injury. Experimental endothelial cell damage due to ROS showed high clinical relevance in severity and extent of coronary artery disease and arterial stiffening in hemodialysis patients with type 2 diabetes [50,54,55].

To accept that GA is a pathogenic protein, one must also keep in mind that albumin contributes a large proportion of the antioxidant activity of total serum [20]. The antioxidant effects of albumin are due to limiting ROS production and scavenging ROS, as described previously. Glycation-induced modifications have a considerable impact on albumin functional properties, which can be related to alterations of its conformation [10,56]. Moreover, the sole residues prone to glycation in albumin are lysine, arginine, and cysteine, which is attributed to their highly nucleophilic characteristics [10]. Thus, glycation may block the powerful antioxidant properties of the Cys-34 residue. In addition, glycation is usually associated with oxidative modifications, which affect early stage glycation products. The simultaneous or combined reaction of oxidation during glycation is termed glycoxidation [10,57]. This oxidative state has been revealed through both an increase in carbonylated protein levels and a decrease in the reduced state of Cys-34 [10,58,59]. Therefore, the glycation of albumin brings about impaired antioxidant capacities partly via losing the free sulphydryl group of Cys-34 and partly via the pathologic effects of glycoxidation itself on endothelial and mesangial cells and macrophages. Furthermore, GA suppresses glucose-induced insulin secretion, impairing glucose metabolism in rat pancreatic β-cells [60]. These findings support and explain the strong involvement of GA in the development of major diabetic complications and the identification of people at risk for microvascular conditions.