by a variety of enzymes into many different prostaglandins and leukotrienes. COX 1 is a constitutively expressed enzyme while COX 2 is induced by inflammation. Using a rodent model for ROP, one group demonstrated that COX 2 was heavily expressed in both retinal ganglion cells and in newly formed blood vessels (Wilkinson-Berka et al. 2003). Treatment with the COX 2 selective inhibitor, rofecoxib, resulted in a 37% decrease in blood vessels in this study.

Using several models of ocular angiogenesis, another group examined the effects of inhibiting either COX 1 or COX 2 (Castro et al. 2004). CNV was induced in Brown Norway rats using an argon laser while Hartley guinea pigs were treated with VEGF to produce intradermal extravasation of Evans Blue Dye (EBD)-albumin. They found that no NSAID was capable of blocking either laser CNV or VEGFinduced neovascularization. However, in corneal vascularization models inhibition of COX 2 (but not COX 1) had a significant effect. These data underscore the complexity of how different tissues within the eye respond to a similar therapeutic agent. A more recent survey of a variety of clinically relevant NSAIDs extended these observations by comparing the relative efficacy of blocking VEGF-mediated angiogenesis as compared to FGF-mediated angiogenesis using a corneal neovascularization model (Pakneshan et al. 2008). The authors found a great variability in the efficacy of different NSAIDs to inhibit VEGF-mediated angiogenesis from 3 (rofecoxib) to 66% (indomethacin). In comparison, inhibition of FGF-mediated angiogenesis was somewhat greater on average. Again, indomethacin provided the greatest degree of antiangiogenic efficacy. In this regard it is interesting to note that indomethacin has the ability to inhibit polymorphonuclear leukocyte migration, which is unrelated to its COX inhibitory activity (Goodwin 1984).

21.3.2  Diabetic Retinopathy

It is estimated that as of this writing, 23 million people within the United States are diagnosed with diabetes and the number is increasing each year. Complications associated with diabetes are responsible for the majority of cases of blindness among working age populations in developed countries (Congdon et al. 2003), making the treatment of this disease of importance as a prophylactic measure to ensure retinal health. As the number of diabetes cases increases, so will ocular pathologies associated with this disease.

21.3.2.1  Pathophysiology

There exist two forms of diabetes termed type I and type 2, which represent very different diseases. Type 1 diabetes, which was once called juvenile onset diabetes, is an autoimmune disease in which cytotoxic T cells attack and destroy the insulin secreting beta cells of the pancreas. This accounts for approximately 10% of diabetes cases. Type 2 diabetes is a metabolic disease which has many contributing factors.

Obesity coupled with lack of exercise tops the list of risk factors; a disconcerting fact given the degree of obesity in the world today (Han et al. 2010.). Type 2 diabetes is preceded by a long period in which the ability of insulin to function is compromised; a process which is called impaired glucose tolerance or insulin resistance. Decreased insulin function is compensated for by increased insulin production. The increased work load has its consequences, however, and over time, the beta cells die. Eventually, the pancreas can no longer produce enough insulin and the patient passes from a state of insulin resistance to diabetes.

The cause of eye disease, along with most other complications associated with diabetes, is high blood glucose. Glucose levels are only modestly increased during the prediabetic period; however, this is believed to be sufficient to begin to damage ocular tissues (Nguyen et al. 2007). One of the many functions of insulin is to promote the expression of the glucose transporter, GLUT4, on the surface of skeletal muscle cells as well as other tissues. Blood glucose can then passively move into the cell where it is quickly converted to glucose-6-phosphate, which cannot pass back through the channel, thus allowing glucose to steadily move out of the blood into its storage depot. The retina primarily expresses GLUT1 and GLUT3, which are not regulated by insulin. For this reason, the retina is considered “insulin independent.” All of these glucose transporters are upregulated during hypoxia allowing increased glucose to enter the cell. Elevated glucose levels have several pathological consequences, some of which are elaborated below.

Glycation:  Glucose itself is reactive and can form glycation products with proteins. Initially the glycation, involving the formation of a Schiff base, is reversible. Over time, these modifications become permanent and are referred to as advanced glycation end-products (AGE). AGE have wide spread effects on both mechanical aspects of tissue properties as well as on signal transduction events necessary for tissue homeostasis. Within Bruch’s membrane, for example, AGEs bind to collagen. The receptor for AGE (RAGE) is engaged and promotes an inflammatory response (Yamagishi et al. 2005). Vitronection, discussed earlier in the context of angiogenesis, has also been shown to function as a target of glycation contributing to retinopathy (Hammes et al. 1996). Additionally, glycation has been shown to affect Ca++ channels on pericytes (Hughes et al. 2004). These support cells associated with retinal capillaries, upon glycation, become less sensitive to endothelin-1-mediated contraction signals.

Polyol accumulation:  The second mechanism involves the conversion of glucose to sorbitol via aldose reductase (sometimes referred to as the polyol pathway). Since glucose movement into retinal tissues is insulin independent, high blood glucose leads to high retinal glucose. Glucose is preferentially utilized as a substrate by hexokinase. However, if glucose levels within the cell increase beyond the point where hexokinase can function, then the excess glucose is utilized by aldose reductase. Sorbitol, produced by aldose reductase, plays a role as an osmotic regulator along with myo-inositol and taurine. When sorbitol levels increase (due to increased cellular glucose), myo-inositol and taurine levels decrease. Alterations in signal transduction, due to decreases in myo-inositol and taurine, have been implicated in several diabetic complications including retinal dysfunction (Lorenzi 2007).

Conversion of glucose to sorbitol requires the conversion of NADPH to NADP. As the regeneration of glutathione (the primary means of protection against redox damage) also requires NADPH, the conversion of glucose to sorbitol decreases the ability of retinal tissue to respond to oxidative stress. Perhaps the best evidence of a role for aldose reductase in diabetic retinopathy comes from the study of aldose reductase deficient mice (Cheung et al. 2005). The authors, having created aldose reductase null mice (Ho et al. 2000), backcrossed the knockout allele into the C57BL/KsJ db/m strain to create aldose reductase deficient mice which spontaneously developed diabetes. Control littermates developed classic signs of diabetic retinopathy including a breakdown of the blood–brain barrier, loss of pericytes, and neovascularization. These pathological changes were significantly less severe in the aldose reductase null mice. Inhibition of aldose reductase is a viable therapeutic strategy; however, inhibitors such as sorbinil have been found to have limited efficacy due to delivery issues as well as toxicities that limit its usefulness (Tsai and Burnakis 1993). New delivery technologies may propel a resurgence of this target, however. One group has investigated the use of PLGA encapsulation to deliver the aldose reductase inhibitor, N-4-(benzoylaminophenylsulfonyl glycine) (BAPSG), in a sustained release implant to diabetic rats (Aukunuru et al. 2002). The formulation provided some improvement over oral dosing, lending hope to this strategy.

Diabetic retinopathy:  Diabetic retinopathy can be divided into five pathophysiological events that occur at the level of the retinal capillary (Chew 2000). These include (1) the formation of microaneurysms, (2) an increase in vascular permeability, (3) the formation of vascular occlusions, (4) neovascularization and scarring, and (5) contraction of the scar and the vitreous. Visual impairment is most directly caused either by increased vascular permeability or by vascular occlusions. These in turn lead to macular edema and the formation of scar tissue (fibrovascular proliferation). The contraction of the scarred tissue can lead to distortions in vision or retinal detachment.

In the early stages of diabetic retinopathy, the basement membrane in discrete areas of the retina begins to thicken, helping to precipitate the ischemic stress that will ultimately produce neovascularization. Plasminogen activator inhibitor-1 (PAI-1) is believed to play a role in this process. Urokinase plasminogen activator (uPA) binds to its cell surface receptor (uPAR) to initiate the conversion of plasminogen to plasmin (Fig. 21.4). Plasmin, in turn, cleaves and activates MMPs which act to balance synthesis and degradation rates for the ECM. PAI-1 levels are greatly increased in tissues of patients with non-proliferative diabetic retinopathy (PDR) (Grant et al. 1996). Transgenic mice overexpressing PAI-1 showed thickened basement membranes around retinal capillaries (Grant et al. 2000). Interestingly, evidence suggests that neovascularization is promoted at a particular level of PAI-1 and that too much or too little disrupts this process. In a murine oxygen-induced retinopathy model using wild type and PAI-1 knockout mice it was found that lack of PAI-1 resulted in approximately a 50% decrease in neovascularization (Basu et al. 2009). Conversely, in a separate study PAI-1 was administered by intravitreal injection in a rat model of ROP and found to decrease neovascularization (Penn and Rajaratnam 2003). Consistent with this concept, researchers found that mice lacking PAI-1 showed

decreased levels of neovascularization (Lambert et al. 2003a). By treating these mice with 100 mg recombinant PAI-1 they restored neovascularization. However, when they treated wild type mice with the same dose of PAI-1 after input into the laser CNV model, neovascularization was inhibited. PAI-1 both inhibits plasmin formation and physically interacts with vitronectin. To determine which of these separate processes may be involved in neovascularization, this group made use of the PAI-1(Q123K) mutant, which is deficient in vitronectin binding but is still capable of inhibiting plasmin formation. Interestingly, this mutant was as capable as wild type PAI-1 in the restoration of disease, suggesting that it is the inhibition of plasmin formation that serves as a useful target.

PEDF, discussed previously, also appears to play a role in diabetic retinopathy. It is downregulated in neovascular ocular diseases and (as mentioned earlier) adenovirus expressed PEDF was shown to have some efficacy in the treatment of advanced AMD (Campochiaro et al. 2006). PEDF has also been found to decrease advanced glycosylation end-product (AGE) mediated angiogenesis in a cultured porcine retinal EC model (Sheikpranbabu et al. 2009). Intravitreal injections of PEDF were found to block the progression of early stages of diabetic retinopathy in the streptozotocin rat model (Yoshida et al. 2009).

VEGF, a potent inducer of neovascularization, is also an important mediator of vascular leakage. It causes this by increasing intracellular Ca++ levels and promoting the activation of PKC [reviewed in Dvorak et al. (1995)]. VEGF was actually first identified as a vascular permeability factor. It acts primarily on microvessels such as post capillary venules and has a potency that is 50,000 times greater than histamine. Permeability is increased through the opening of vesicular-vacuolar fenestrae, which span the length of the vascular endothelial cytoplasm. Numerous experiments have been performed examining the effects of modulating VEGF activity on DR.

Endogenous antiangiogenic proteins include the NC1 fragments of collagen: arrestin, canstatin, and tumstatin as described earlier. In addition, a fragment of plasminogen, termed angiostatin, has been shown to play a protective role in DR (Wahl et al. 2004). Angiostatin contains a number of triple disulfide bond-linked loops referred to as kringle domains which have powerful antiangiogenic properties (O’Reilly et al. 1994). Indeed, expression of the kringle 5 (K5) domain of plasminogen was able to ameliorate diabetes-induced retinal vascular leakage (Park et al. 2009).

Diabetic macular edema:  Increased vascular permeability is an early (nonproliferative) stage of diabetic retinopathy and often occurs near the macula. The definition of diabetic macular edema (DME) is a thickening of the retina due to edema within one disc diameter of the macula (Chew 2000). Edema is often accompanied by a hard exudate, comprised of lipoprotein deposits. While edema may come and go with no consequence to visual acuity, these exudates have been associated with retinal damage and permanent vision loss.

Proliferative diabetic retinopathy:  Data from animal models suggest that the blockage of capillaries might be due to the formation of micro-thrombi consisting of aggregates of leukocytes. Leukocyte interactions with blood vessels are altered in diabetic retinopathy through changes in the expression of integrins and their ligands.