Relation Between Serum 3-Deoxyglucosone and Development of Diabetic Microangiopathy

We studied 110 diabetic patients, aged 20–80 years, who were referred to Kobe University Hospital (Kobe, Japan) and National Hyogo Chuo Hospital (Sanda, Japan) for the evaluation and treatment of diabetes. Nine type 1 diabetic patients and 101 type 2 diabetic patients were included. Subjects with elevated serum creatinine concentrations (>115 μmol/l) were excluded from this study to eliminate the possible effect of a deficiency in renal clearance function on the serum 3-DG level. Fasting blood samples for the measurement of the serum 3-DG level were obtained in conjunction with the collection of samples for other routine laboratory studies, according to approved institutional guidelines. The characteristics of the study subjects are shown in Table 1. Each diabetic patient was provided with medical care to achieve glycemic control; 29 patients were treated with dietary modification, 36 patients with oral hypoglycemic drugs, and 45 patients with insulin. Fifty-seven age-matched control subjects were also obtained with informed consent from nondiabetic patients and healthy volunteers at the Kobe University Hospital. None of the nondiabetic control subjects had a past history of renal, ocular, or neurological disease.

Urine analysis was performed in all diabetic patients. Microalbuminuria and proteinuria were defined as a urinary albumin excretion of ≥30 and ≥300 mg/day, respectively. Only the patients who gave their consent underwent ophthalmological and neurological examinations (99 and 48 patients, respectively). The extent of retinopathy was assessed by ophthalmologists using fundus photography, and it was classified as normal, nonproliferative retinopathy, or proliferative retinopathy. Patients with a history of laser treatment were included in the last category. However, patients who had taken laser treatments for maculopathy were excluded from entry into this study. Electrophysiological data were obtained by a single physician at the clinical laboratory of National Hyogo Chuo Hospital. Sensory nerve conduction velocity (SCV) and motor nerve conduction velocity (MCV) were determined in the sural nerve of 47 patients and peroneal nerve of 46 patients, respectively. F-wave conduction velocity (FCV) in the proximal segment of the peroneal nerve was simultaneously measured in 31 patients.

The serum 3-DG concentration was determined as previously described (28). Briefly, 2,3-pentanedione was added to serum as an internal standard and deprotenized with 6% perchloric acid, followed by the reaction with 2,3-diaminonaphthalene to derivatize 3-DG to a stable compound. Authentic 3-DG was kindly supplied by Fumitaka Hayase (Meiji University, Kawasaki, Japan) and derivatized simultaneously to be used as a standard. The derivatives in the reaction mixture were extracted using ethyl acetate, dried, and reconstituted with methanol to be subjected to HPLC. The quantitation of 3-DG was performed by calculating the peak height ratio of the 3-DG–derived peak to an internal standard peak.

The data are expressed as means ± SD. Comparison between groups was performed using one-way ANOVA followed by Benferroni/Dunn multiple comparison test. We used Student’s t test only in the case of comparison between two groups. A P value <0.05 was considered significant.

The serum 3-DG concentrations of the diabetic patients were significantly higher than those of the nondiabetic control subjects (P < 0.001) (Table 1). Principally, there was a tendency for the patients with higher HbA_1c levels to show higher 3-DG levels. However, we did not find statistically significant correlation between the HbA_1c and serum 3-DG levels in the diabetic patients (P = 0.140), since several patients showed a 3-DG level markedly deviated from the recent blood glucose control state indicated by the HbA_1c level (Fig. 1). For example, patient A in Fig. 1 showed a much higher serum 3-DG concentration compared with the value expected from her HbA_1c level and already suffered from microalbuminuria and some decrease in nerve conduction velocities, despite short duration (3 years) of diabetes. The other two patients (patients B and C in Fig. 1), whose serum 3-DG levels were >500 nmol/l despite an HbA_1c level <7.0%, also showed certain signs of microangiopathy. Patient B (whose diabetes duration was still 2 years) started to show microalbuminuria and nonproliferative retinopathy, while patient C had already developed overt proteinuria and proliferative retinopathy with a diabetes duration of 20 years. In contrast, patient D in Fig. 1, whose serum 3-DG remained at the lower level (166 nmol/l), regardless of the poor blood glucose control at the time of this study, showed no signs of neuropathy, retinopathy, or nephropathy with a diabetes duration of 7 years.

The diabetic group was subdivided according to the severity of complications, and the serum 3-DG values of each group are summarized in Table 1. A prominent elevation of serum 3-DG levels was observed even in the diabetic patients with normoalbuminuria compared with the control subjects (P < 0.001) (Fig. 2). Significant differences in serum 3-DG concentration were found between the normoalbuminuria and other diabetic groups, such as microalbuminuria (P = 0.027) and overt proteinuria (P < 0.001). In addition, we conducted a logistic regression analysis. As a result, the relative risk of diabetic nephropathy (microalbuminuria and overt proteinuria) was 1.94 (95% CI 1.27–2.99) per 100-nmol/l increase in serum 3-DG level.

The above-mentioned tendency was reproduced by the assessment of the relation between the serum 3-DG concentrations and the severity of retinopathy. The serum 3-DG concentrations were higher in the diabetic patients, even without any signs of retinopathy, as compared with control subjects (P < 0.001) (Fig. 2). The distribution of the serum 3-DG concentration shifted to the higher level in the more severe retinopathy group, so that the mean value of serum 3-DG significantly increased as retinopathy developed to the nonproliferative or proliferative state (P = 0.008 and P < 0.001 vs. the diabetic patients without retinopathy, respectively). The logistic regression analysis showed that relative risk of any diabetic retinopathy was 2.04 (95% CI 1.26–3.29, P = 0.036) per 100-nmol/l increase in serum 3-DG level.

Regarding neuropathy, we investigated the relation between the serum 3-DG levels and abnormalities in the nerve conduction, including the SCV, MCV, and FCV. We separated the patients by the median value of each conduction velocity value (45.1 m/s for SCV, 40.2 for MCV, and 47.6 for FCV) (Fig. 3). The patients with reduced nerve conduction velocities tended to show higher serum 3-DG levels in each conduction study: a significant difference (P = 0.013) was observed in the MCV. Further analysis revealed that the serum 3-DG concentrations of the patients with a reduction in two or all three of the conduction velocity values (n = 17, 381 ± 107 nmol/l) were significantly (P = 0.035) higher than those of the patients, without any reduction in conduction velocity (n = 9, 291 ± 81 nmol/l). We alternatively conducted a correlation analysis to evaluate overall correlation between NCV measurements and serum 3-DG levels, so that FCV measurement showed significant correlation with 3-DG level (r = 0.392, P = 0.029).

Hyperglycemia has been believed to cause diabetes complications through the glycation reaction, protein kinase C activation, and the polyol pathway, as well as overproduction of reactive oxygen species from mitochondria (33). In fact, these mechanisms are thought to be implicated to each other. Considering the possibility that 3-DG forms from fructose (23,24) via the polyol pathway in addition to glucose-derived glycated proteins, the measurement of this key compound may give more comprehensive information regarding the hyperglycemia-related mechanisms.

In the present investigation, a cause-and-effect relation between the 3-DG and diabetic microangiopathy might be postulated since the elevation of the serum 3-DG concentration exists before the onset of microangiopathy and the degree of severity of the microangiopathy is in part dependent on the serum 3-DG concentration. Furthermore, we found that the development of diabetes complications was accelerated in the patients with an extremely high level of serum 3-DG (i.e., >500 nmol/l) and that they tended to suffer multiple and severe complications. In addition to subjects A–C in Fig. 1, some of the patients with high 3-DG levels showed a tendency to have more severe complications for their duration of diabetes, even if their HbA_1c levels were not high (data not shown). On the contrary, patients with low 3-DG seemed to be relatively resistant to the development of the complications. However, even the subjects with modest elevation of 3-DG developed complications over a longer period of time, suggesting that duration of diabetes should also be taken into account when estimating the effect of 3-DG. Nevertheless, the present logistic regression analysis revealed that serum 3-DG level contributed to the development of both nephropathy and retinopathy more significantly than duration of diabetes. Thus, we had an impression that the serum 3-DG level may be a useful marker to predict the prognosis of microangiopathy, rather than a marker to reflect the current status of the complications. In other words, intensive prevention might be required to delay the progression of diabetes complications in the subjects with high 3-DG levels.

The potential mechanisms of 3-DG action in vivo presumably consist of both direct and indirect effects via AGE formation. Concerning the latter mechanism, 3-DG is well known to form AGEs much more efficiently than glucose. For example, the cross-linking potency of 3-DG in protein polymerization was ∼10 times higher than that of glucose (34). These findings indicate that AGE formation is exponentially accelerated when glucose is converted to 3-DG. Immunohistochemical studies also showed 3-DG-derived AGEs, such as pyrraline (6) and imidazolone compound (35), accumulated in diabetic angiopathy lesions. In addition, recent studies have revealed that 3-DG itself has direct effects on cell functions (25–27). For example, Che et al. (27) reported that highly reactive dicarbonyls, including 3-DG, might induce intracellular oxidative stress by attacking the active sites of antioxidant enzymes, leading to the expression of heparin-binding epidermal growth factor-like growth factor in the aortic smooth muscle cells of rats. Thus, a longer exposure of a high level of 3-DG to cells in vivo might be responsible for some of the disorders observed in diabetic subjects. This may be supported by the present finding that diabetic patients with relatively higher 3-DG levels were prone to suffer from more severe complications.

The present study did not reveal a statistically significant correlation between serum 3-DG and HbA_1c levels because of the presence of some patients with deviated 3-DG levels. This result is inconsistent with the previous report (32), showing a significant correlation between plasma 3-DG and HbA_1c levels in 27 diabetic patients. We also believed that the levels of Amadori products, such as HbA_1c, contribute to 3-DG level in part, since these compounds are potent precursors of 3-DG. However, 3-DG can even form through other pathways, including the polyol pathway. In this relation, Hamada et al. (36) reported the beneficial effect of epalrestat, an aldose reductase inhibitor, on the reduction of erythrocyte 3-DG content in diabetic patients, supporting the potential involvement of the polyol pathway in the formation of 3-DG in vivo. Therefore, activity of the enzymes in this pathway may also influence the 3-DG levels independent of glycemic level. Furthermore, it is also known that 3-DG is metabolized to a relatively inert species by several enzymes such as aldehyde reductase (37) and oxaldehyde dehydrogenase (38). Considering that several milligrams of 3-DG are formed in the body per day, as estimated by Baynes et al. (29), this detoxification system is important to regulate the 3-DG level. It therefore appears that the level of serum 3-DG is determined not only by its formation, but also by the ability to metabolize it in vivo. The result of our additional analysis for correlation between glucose and 3-DG concentrations in 43 subjects (r = 0.459, P = 0.002) supports that glycemic status partly predicts 3-DG level but not completely. Takahashi et al. (14) reported the possibility that this detoxification system itself might be deteriorated due to glycation of the enzymes. Consequently, a “vicious circle” is established as follows: the elevation of 3-DG levels induced by hyperglycemia in turn inactivates these enzymes by modifying them, resulting in the retention of a high level of 3-DG in diabetic subjects. In addition, Beisswenger et al. (39) recently indicated that a postprandial increase in the 3-DG level was correlated with postprandial glycemic excursions, but not with HbA_1c level. This implies that 3-DG metabolism differs among the diabetic patients with a similar HbA_1c level. The possible individual differences in the metabolism of 3-DG may in part account for the incomplete correlation between HbA_1c and serum 3-DG levels observed in the present study. Furthermore, we found that the patients whose 3-DG concentrations were relatively higher compared with the values expected from their HbA_1c levels were likely to show a faster progression of complications. This suggests that an elevation of 3-DG level is more closely related to the development of diabetic microangiopathy, as compared with blood glucose itself.

The 3-DG assay is thus thought to give more integrated information about hyperglycemia-related mechanisms in the development of diabetic microangiopathy. Further prospective studies may give more useful information on the availability of serum 3-DG level as a predictive marker for diabetes complications. Finally, regarding pharmacological targets to prevent the progression of diabetes complications, it may well be beneficial to inhibit dicarbonyl compounds such as 3-DG.

This study was supported in part by a grant for research on Maillard Reaction from Ono Pharmaceutical.

We thank Fujio Hayashi, Masahiko Mukai, and Shunichi Kumagai (Department of Clinical Laboratory, Kobe University Hospital) for their assistance with the preparation of the blood and urine samples; Hiroyuki Yamada (Division of Diabetes, Digestive and Kidney diseases, Department of Clinical Molecular Medicine, Kobe University Graduate School of Medicine) for his assistance in the setup of the HPLC; Youichi Tominaga, Kenji Jinnai, and Keiichi Takahashi (Department of Internal Medicine, National Hyogo Chuo Hospital) for their participation in the clinical study; and Casmir Wambura for his assistance in preparation of this manuscript.