Corneal Advanced Glycation End Products Increase in Patients With Proliferative Diabetic Retinopathy

Corneal autofluorescence was measured in 26 eyes of 26 patients with type 2 diabetes and 13 eyes of 13 healthy age-matched control subjects. The patients with type 2 diabetes were divided into three groups: patients without DR, patients without PDR, and patients with PDR (Table 1). Eyes that had undergone surgery were excluded from this study. The study protocol was reviewed by the ethics committee of our institution. All procedures adhered to the tenets of the Declaration of Helsinki.

Corneal autofluorescence was measured by scanning with a lens measurement system (SpectRx) (11) that had been adjusted for corneal measurement. The system has a xenon lamp as an excitation source and a fluorescence detector. The noninvasive corneal autofluorescence measurements were performed at various distances along the optical axis of the cornea with the assistance of a pupil-tracking system. The pupil-tracking system, an automatic alignment system, was incorporated into the lens measurement system and works as follows. An infrared charged-coupled–device camera images the iris and the pupil, and the image is digitized. The position data derived from the digitized charged-coupled–device image are fed into a two-axis servomotor positioning system to maintain the x and y positions of the sample volume to within ±100 μm of the programmed “lock” position (11). The sample volume in the optics of the lens measurement system was ∼300 μm long, 600 μm high, and 80 μm wide. We used color filters at wavelengths of 360–370 nm for excitation and 430–450 nm for emission, because 360–370 and 430–450 nm (excitation and emission, respectively) were reported to be characteristic of AGE fluorescence (i.e., crossline, etc.) in the tissue (1,7,8). We reported that there was a significant positive correlation between lens autofluorescence values and the AGE levels in the same extracted lenses in diabetic hamsters (10). We used the peak corneal autofluorescence level as the corneal autofluorescence value. The corneal autofluorescence values measured by our fluorophotometry are defined as corneal AGE fluorescence. The corneal AGE fluorescence values are expressed as arbitrary units.

We compared the corneal AGE fluorescence values in the four groups of patients and examined the relationship between these values and age, duration of disease, and HbA_1c levels in the patients with type 2 diabetes.

The results were expressed as means ± SD. The distribution of corneal AGE fluorescence values in these four groups was compared using a one-way of analysis variance and Scheffe’s test. A P value of 0.05 or lower was considered significant. The Pearson correlation coefficient (r) was calculated to determine whether there were significant relationships between variables.

The results of the corneal AGE fluorescence values in the four groups of patients are shown in Fig. 1. The corneal AGE fluorescence values in the control subjects, the patients without DR, the patients without PDR, and the patients with PDR were 6.9 ± 1.3, 7.4 ± 2.1, 6.9 ± 2.2, and 11.9 ± 3.9 arbitrary units (mean ± SD), respectively. A significant difference was noted in the corneal AGE fluorescence values between the four groups (analysis of variance, P = 0.0001); corneal AGE fluorescence values were markedly higher in the patients with PDR than in the control subjects, the patients without DR, or the patients without PDR (Scheffe’s test; P = 0.0006, P = 0.0053, and P = 0.0078, respectively). In the patients with type 2 diabetes, no significant correlation was found between corneal AGE fluorescence values and age (r = 0.063), duration of disease (r = 0.21) (Fig. 2), or HbA_1c level (r = 0.38) (Fig. 3).

In the present study, we measured corneal AGE fluorescence values using our fluorophotometry system, the wavelength of which was characteristic of AGE fluorescence (1,7,8). We previously reported that there was a correlation between lens autofluorescence values detected by our fluorophotometry system and AGE levels in the same lenses measured by an enzyme-linked immunosorbent assay in diabetic animals (10). The previous reported results showed a strong correlation between the value of lens autofluorescence measured with the 360- to 370-nm and 430- to 450-nm wavelengths (excitation and emission, respectively) and the AGE levels measured by enzyme-linked immunosorbent assay in the same extracted lenses (10). It also has been reported that oxidized LDLs (oxLDLs), which are believed to increase in patients with diabetes, generate fluorescent products with emission at 430 nm when excitation is performed at 360 nm (12–14). That wavelength is characteristic of oxLDL fluorescence and similar to that of AGE fluorescence. However, there is much less autofluorescence of oxLDL than of AGEs (14). Thus, it is possible that corneal autofluorescence induced by these wavelengths as AGE fluorescence measured by our fluorophotometry might mainly reflect corneal AGE fluorescence.

Sady et al. (15), using a two-step high-performance liquid chromatography assay, reported the presence of pentosidine in the cornea. They also reported that those pentosidine levels correlated with the collagen-bound fluorescence and pentosidine levels and that collagen-bound fluorescence was higher in human diabetic corneas than in control corneas. Those authors suggested that AGEs in the cornea increase in patients with diabetes. Kaji et al. (16) reported that N-(carboxymethyl) lysine (CML) immunoreactivity was observed at the site of the epithelial basement membrane in the corneas of patients with diabetes. AGEs, which are produced by the nonenzymatic reaction of glucose and other aldoses with protein via the Maillard reaction and have specific fluorescence characteristics, seem to accumulate in the corneas of patients with diabetes. In the present study, the corneal AGE fluorescence value in the group of patients with PDR was significantly higher than that in the control subjects, the patients without DR, and the patients without PDR. These results indicate that corneal AGEs increase in patients with PDR.

Using our fluorophotometry, corneal AGE fluorescence could not be broken into separate signals from the corneal epithelium, stroma, and endothelium. The pentosidine concentration in human corneal collagen is increased in patients with diabetes (15). These results suggest that pentosidine, which is an AGE, can cause increased fluorescence in the corneal stroma of patients with diabetes. However, CML, which is another AGE, was observed at the site of the epithelial basement membrane in corneas of patients with diabetes (16). The distribution of AGEs in the corneas of patients with diabetes is still unclear.

In the present study, there was no correlation between HbA_1c levels or duration of disease and corneal AGE fluorescence values in the patients with type 2 diabetes. However, corneal AGE fluorescence values increased only in the patients with PDR. The vitreous concentrations of AGEs and vascular endothelial growth factor (VEGF) were both elevated in the patients with PDR (5,17,18). VEGF plays a role in breakdown of the blood-retinal barrier (BRB) (19–21). Increased values of AGEs in the vitreous in patients with PDR seem to be related to breakdown of the BRB. Furthermore, some authors reported that permeability of the blood-aqueous barrier (BAB) increased in patients with PDR (22,23). We speculated that increased corneal AGEs in patients with PDR may be caused by an increase of AGEs in the aqueous humor and vitreous resulting from breakdown of the BRB and/or BAB.

Diabetic corneal epitheliopathy has been recognized clinically, especially after vitrectomy (24). Structural changes in the endothelium have been reported in patients with diabetes (25,26). In corneas of patients with diabetes, decreased sensation, neurotrophic ulceration, delayed wound healing, recurrent erosions, and epithelial edema have been found clinically. Histologically, thickening and multilayering of the epithelial basement membrane and morphologic changes of the corneal epithelial cells have been reported (27). These changes are assumed to be due to intracellular edema, which may result from activation of the sorbitol pathway. The aldose reductase inhibitor has been reported to be effective in treating corneal epithelial defects (28). However, the pathogenic mechanisms underlying those abnormalities are still unclear. In vitro, nonenzymatic glycation of laminin-attenuated adhesion and spreading of corneal epithelial cells and the presence of aminoguanidine, an inhibitor of AGE formation in the incubation mixture during glycation, inhibited CML formation and promoted adhesion and spreading of corneal epithelial cells (16). We hypothesized that one of the corneal complications of diabetes, especially in patients with PDR, could be related to the accumulation of AGEs in the cornea. We believe that this should be considered in the future.

In conclusion, these results demonstrate that corneal AGEs may increase in patients with PDR compared with control subjects, patients without DR, and patients without PDR. In patients with PDR, increased corneal AGEs may play a role in diabetic keratopathy.

We thank N. Moriyama and M. Nakagawa, Teijin Limited, for technical assistance in measuring corneal autofluorescence.