drugs documented to prevent the whole spectrum of abnormalities induced by diabetes in glial cells, neurons, and vascular cells of the retina. This may have translational importance, because human diabetic retinopathy has recently become known to include glial and neuronal abnormalities.

The efficacy of AR inhibitors (ARIs) has been, for the most part, disappointing in humans. The major reason for the discrepant results in clinical vs. preclinical studies is likely discrepant doses; for example, in recent studies the ARI sorbinil proved successful in preventing retinopathy in diabetic rats when given at a dose 20-fold larger than the dose used unsuccessfully in a past clinical trial. It has become clear that larger doses of ARIs ensure that metabolic flux through both steps of the pathway is inhibited – as opposed to merely reducing sorbitol accumulation. A current hypothesis posits that normalization of glucose flux through the pathway is required in order to prevent excessive turnover of pathway cofactors and oxidative stress; and that the latter is a critical, if not the main, determinant of the tissue consequences of excess polyol pathway activity. Testing this concept in humans will become possible when new drugs, capable of inhibiting aldose reductase with higher in vivo efficacy and safety than the older ARIs, become available. It is reasonable and important to advocate, and work toward, the discovery of such drugs because some features of diabetic retinopathy appear best or uniquely approached via inhibition of excess polyol pathway activity.

Key Words: Diabetic retinopathy; polyol pathway; aldose reductase; sorbitol dehydrogenase; aldose reductase polymorphisms; sorbitol; fructose; osmotic stress; oxidative stress; advanced glycation endproducts; pericytes; endothelial cells; Müller glial cells; apoptosis; inflammation; aldose reductase inhibitors.

INTRODUCTION

Retinopathy is the most severe of the ocular complications of diabetes (1). More than all other complications of diabetes, retinopathy may begin as a rather pure manifestation of glucose toxicity. This is suggested and supported by the demonstration that experimental galactosemia induces in animals a picture of retinal microangiopathy highly similar to that of nonproliferative diabetic retinopathy (2, 3), while not mimicking fully complications of diabetes in other tissues (4). Galactosemia induced by a galactose-rich diet results in elevated hexose levels in blood without the array of metabolic and hormonal changes characteristic of diabetes (3). Hence, the galactosemic model provides a rigorous argument for the discrete role of hyperhexosemia in causing retinopathy. The role of hyperhexosemia is also suggested by the fact that the levels of glycated hemoglobin (HbA1c) over time are the dominant predictor of retinopathy progression in diabetic patients (5). This observation is less compelling in its pathogenic implications than the first because, while HbA1c does reflect glycemic levels, glycemic levels reflect a multitude of metabolic and hormonal changes. However, the HbA1c data are critically important because they are consistent with a role for glucose toxicity in human diabetic retinopathy.

The galactosemic model also points to a biochemical mechanism for how elevated hexose levels may lead to retinopathy: activation of the polyol pathway. The longest known consequence of polyol pathway activation is a rise in tissue polyol levels. For example, when blood galactose levels increase, many tissues accumulate galactitol, which is produced from galactose via the action of aldose reductase (AR), the first and