Diabetic Retinopathy and Diabetic Macular Edema: Pathophysiology, screening, and novel therapies

Diabetic retinopathy (DR) and diabetic macular edema (DME) are common microvascular complications in patients with diabetes and may have a sudden and debilitating impact on visual acuity (VA), eventually leading to blindness. Advanced stages of DR are characterized by the growth of abnormal retinal blood vessels secondary to ischemia. These blood vessels grow in an attempt to supply oxygenated blood to the hypoxic retina. At any time during the progression of DR, patients with diabetes can also develop DME, which involves retinal thickening in the macular area. DME occurs after breakdown of the blood-retinal barrier because of leakage of dilated hyperpermeable capillaries and microaneurysms. The current management strategy for DR/DME requires early detection and optimal glycemic control to slow the progression of disease. Adherence to these recommendations is hampered by the fact that the condition is generally asymptomatic at early stages. Current treatments for DR/DME, such as laser photocoagulation, only target advanced stages of disease. Several pharmacological therapies are being developed to treat early stages of DR/DME, but will require a renewed emphasis on early detection. This review will focus on the current understanding of the epidemiology and pathophysiology of DR/DME, the updated clinical diagnostic grading system, screening and management, and the rationale behind the potential for pharmacological treatments.

Type 2 diabetes has reached epidemic proportions, fueled by an aging population and the rapid increase in obesity (1). DR is a major cause of vision loss in patients with diabetes. The longer patients have diabetes, the higher the prevalence of DR (2). In developed countries, DR is recognized as the leading cause of blindness in the working-age population (20–74 years old) and is responsible for 12% of new cases of blindness each year (3).

In patients with type 1 diabetes, the cumulative 14-year incidences of visual impairment (VA 20/40 or worse in the better eye), doubling of the visual angle, and blindness were 12.7, 14.2, and 2.4%, respectively (4). DME is a frequent manifestation of DR (5) and is a leading cause of legal blindness in patients with type 2 diabetes. Over a 10-year period, non–clinically significant DME and clinically significant DME will, respectively, develop in 14 and 10% of Americans with known diabetes (6). Approximately half of patients with DME will lose two or more lines of VA within 2 years (7).

Even in cases where retinopathy has not yet progressed to blindness, loss in VA because of diabetes is a major problem and may lead to significant reductions in functional status. DR is the third leading cause of severe visual impairment among inner-city adults ≥40 years of age (8).

Diabetes-related blindness and visual impairment places a significant burden on society. The federal budgetary cost of blindness was estimated to be $4.1 billion in the U.S. for the year 1990, and 97% of these costs were accounted for by the working-age adult group (9). Health care and economic burdens of DR are further compounded by the resulting decline in quality of life (10); thus, the true impact on society cannot be estimated on a monetary basis alone.

Many studies have demonstrated that chronic hyperglycemia, as well as hyperlipidemia and hypertension, contribute to the pathogenesis of DR (11–14). The exact mechanisms by which elevated glucose initiates the vascular disruption in retinopathy remain poorly defined, and, not surprisingly, several pathways have been implicated. The vascular disruptions of DR/DME are characterized by abnormal vascular flow, disruptions in permeability, and/or closure or nonperfusion of capillaries.

A hallmark of early DR is the change in the structure and cellular composition of the microvasculature (15–17). Endothelial cells are responsible for maintaining the blood-retinal barrier, and damage to them results in increased vascular permeability. In early stages of DME, breakdown of the inner blood-retinal barrier may occur, resulting in accumulation of extracellular fluid in the macula (7,18).

Pericytes are essential cellular components in the regulation of retinal capillary perfusion, and damage to these cells in diabetes leads to altered retinal hemodynamics, including abnormal autoregulation of retinal blood flow (19). Loss of retinal pericytes represents another early feature of DR (20–22) and correlates with microaneurysm formation (15,20,23). Another common feature of DR is the thickening of the capillary basement membrane and increased deposition of extracellular matrix components. This feature may contribute to the development of abnormal retinal hemodynamics (24–26), including abnormal autoregulation of retinal blood flow.

There is evidence that retinal leukostasis may also play an important role in the pathogenesis of DR. Leukocytes possess large cell volume, high cytoplasmic rigidity, a natural tendency to adhere to the vascular endothelium, and a capacity to generate toxic superoxide radicals and proteolytic enzymes (27). In diabetes, there is increased retinal leukostasis, which affects retinal endothelial function, retinal perfusion, angiogenesis, and vascular permeability. In particular, leukocytes in diabetes are less deformable, a higher proportion are activated, and they may be involved in capillary nonperfusion, endothelial cell damage, and vascular leakage in the retinal microcirculation (27). A recent study showed that diabetic vascular leakage and nonperfusion are temporally and spatially associated with retinal leukostasis in streptozotocin-induced diabetic rats (28). There are many capillary occlusions by leukocytes and capillary dropout or degeneration associated with leukocytes in the diabetic retina (27). Serial acridine orange leukocyte fluorography and fluorescein angiography (FA) show trapped leukocytes directly associated with areas of downstream nonperfusion in the diabetic retinal microcirculation (27). Whereas leukostasis probably plays a key role in the pathogenesis of DR, platelets and erythrocytes are also involved in this process.

As a result of occluded capillaries, retinal ischemia stimulates a pathologic neovascularization mediated by angiogenic factors, such as vascular endothelial growth factor (VEGF), which results in proliferative diabetic retinopathy (PDR) (29,30). This neovascularization is the predominant feature of PDR. Hemorrhaging of new vessels into the vitreous may also lead to tractional retinal detachment (5).

Understanding the diabetes-induced mechanisms that contribute to pericyte loss, endothelial cell proliferation, neovascularization, and alterations in basement membrane structure is therefore central to the design of pharmacological therapeutic strategies to treat and prevent early diabetes-related microvascular changes.

DR is a progressive disease that includes the following stages: no apparent DR, nonproliferative DR, and PDR. Nonproliferative DR is characterized by the presence of venous dilatation, microaneurysms, retinal hemorrhages, retinal edema, and hard exudates (5). Neovascularization in DR can originate from the optic disk or elsewhere (5). The gold standard for grading the severity of DR is stereoscopic fundus photography through dilated pupils, using seven standard fields (31–33), and grading guidelines for these photographs established by the Early Treatment Diabetic Retinopathy Study (ETDRS) group (34). Although this staging system is used in research studies, it is rarely used clinically, given its complexity. To facilitate communication between retina specialists and health care professionals, International Clinical Diabetic Retinopathy and Diabetic Macular Edema Disease Severity Scales have recently been proposed (35) that were derived from the ETDRS (34) and the Wisconsin Epidemiologic Study of Diabetic Retinopathy (14,36) studies. The International Clinical Diabetic Retinopathy Disease Severity Scale identifies five levels of DR that are clinically meaningful to practitioners and facilitate communication between practitioners (Table 1).

Some DR patients may develop vision loss from DME. Clinically significant macular edema (CSME) occurs if there is thickening of the retina involving the center of the retina (macula) or the area within 500 μm of it, if there are hard exudates at or within 500 μm of the center of the retina with thickening of the adjacent retina, or if there is a zone of retinal thickening one disk area or larger in size, any part of which is within one disk diameter of the center of the retina (37). This definition of CSME generally refers to the threshold level at which laser photocoagulation is carried out. However, it is important to appreciate that the majority of visual loss occurs when macular edema involves the center.

The International Clinical Diabetic Macular Edema Disease Severity Scale includes two major levels: absent and present. If DME is present, it is divided into mild (some retinal thickening or hard exudates in the posterior pole, but distant from the center of the macula), moderate (retinal thickening or hard exudates approaching the center of the macula but not the center), and severe (involving retinal thickening or hard exudates involving the center).

Early detection of retinal abnormalities is essential in preventing DR/DME and loss of vision. Treatments, such as photocoagulation, can decrease vision loss (33,37,38). However, it is generally not possible to restore VA once it has deteriorated. Because DR can progress to irreversible stages with relatively few symptoms (33), the optimal time for treatment is before VA is impaired. Studies have confirmed that the clinical outcome is better if patients are screened and treated early (39). The benefits of early management, such as intensive diabetes control, persist for years, even with subsequent hyperglycemia (40). Thus, regular screening and early treatment for DR/DME can potentially save years of vision and reduce societal costs (33).

Current methods of color fundus photography use either a stereoscopic or nonstereoscopic camera to take seven 30°, three wide-angle 60°, or nine overlapping 45° fields (41,42). Fundus photography can also be used to track the progression of disease or efficacy of treatments. The utility of fundus photography as a large-scale screening procedure is limited because of its cost and the requirement for special equipment and trained personnel (32). Pupil dilation is another inconvenient aspect and may reduce compliance even further. Several technologies that offer simple, low-cost, and more accessible photographic screening for DR/DME are currently being evaluated.

To make fundus photography easier and more widely accessible, investigators have evaluated the potential for the use of digital cameras to obtain fundus images through nondilated pupils. The sensitivity and specificity of these nonstereoscopic digital screening methods have most often been compared with the ETDRS seven standard field images. Overall, these methods where in substantial agreement with the ETDRS classification for the grading of DR and in fair to moderate agreement for the grading of DME (43,44).

Another advantage of digital technologies is the ability to transmit images to a centralized reading center for grading. E-mailed digital images from dilated pupils were found to be suboptimal for detecting DME, but sensitive and specific enough to detect sight-threatening DR (45). Although image quality is reduced compared with that of 35-mm photographs, telemedicine-transmitted images appear to be useful for the detection of sight-threatening DR in the hands of experienced graders (46). A recent study showed that four 45° field nonstereoscopic color digital fundus photographs of each eye could be transmitted over the Internet and be used with great accuracy for the screening of DR (47). These methods have also been shown to have good sensitivity and specificity when using a centrally based trained grader.

This ability to digitally photograph and transmit retinal images has led to efforts to develop complete telemedicine screening programs for DR. The Joslin Diabetes Center in Boston has recently developed the Joslin Vision Network, which includes a remote imaging system, a centralized grading center at the Joslin Diabetes Center, and a data storage system. One recent study validated the agreement between nonmydriatic Joslin Vision Network images and dilated ETDRS photographs and suggested that this digital technique may be an effective telemedicine tool for remotely determining the level of DR, suggesting timing of next retinal evaluation and identifying the need for prompt referral to ophthalmology specialists (44). There are also commercial efforts underway, including Inoveon, based in Oklahoma City, and EyeTel, based in Virginia. Eyetel uses a novel approach that places an imaging system, called the Digiscope, in primary care offices to transmit images via a modem to a reading center at the Wilmer Eye Institute, Johns Hopkins Hospital, Baltimore, Maryland.

In addition to these efforts, newly developed automated methods for detection of DR have sensitivity and specificity ranges of 77.5–88.5 and 88.7–99.7%, respectively (48). However, tele-ophthalmology should not be currently viewed as a substitute for comprehensive eye examinations. Photographic techniques can miss disease occurring outside the photographic fields and are less effective than angiography at detecting capillary closure and leakage (see below). Given the lower sensitivity of nonstereoscopic digital photography to detect DME, supplemental measurement of VA may also be warranted (45). A more appropriate use of these technologies is the identification of patients with retinal lesions that warrant further evaluation by an ophthalmologist (44).

Ophthalmoscopy is a useful screening procedure, easy to use and accessible to ophthalmologists and other physicians, and requires no specialized equipment. Compared with fundus photography, ophthalmoscopy by an experienced examiner was found to agree with grading by fundus photography >85% of the time (49). In the hands of primary care physicians, ophthalmoscopy was less sensitive than fundus photography in detecting both any DR and sight-threatening DR (sensitivity 63 vs. 79% and 66 vs. 87%, respectively) (50). With additional training, however, ophthalmologic screening by primary care physicians may be a clinically acceptable and cost-saving strategy to refer patients for evaluation by an ophthalmologist. Ophthalmoscopy compared favorably with fundus photography, with 100% sensitivity for referral or follow-up within 1 year (51).

Even in the hands of experienced ophthalmologists, ophthalmoscopy is less sensitive than photography for detecting some of the earliest lesions of retinopathy (52). Ophthalmoscopy was less sensitive than photography for detecting DR/DME in patients with only a few microaneurysms (53). Fundus photography and ophthalmoscopy can be viewed as complementary techniques and may be particularly informative when used adjunctively (31).

FA is generally used for treatment planning. It is a method in which sodium fluorescein is intravenously administered followed by rapid sequence photography of the retina to evaluate its circulation. A method using orally administered fluorescein has also been developed (54). Normally, fluorescein cannot pass through the tight junctions of retinal capillaries; however, in some disease states, such as DR and DME, dye leakage occurs. The method is useful in detecting early alterations of the blood-retinal barrier, capillary closure, and microaneurysm formation (55). The major advantage of FA over fundus photography is its ability to detect macular ischemia denoted by nonperfusion of the retinal capillaries and to detect subtle DME as evidenced by fluorescein leakage from the capillaries (55). An automated method of quantitating microaneurysms from digitized fluorescein angiograms was shown to reliably detect microaneurysms with a sensitivity of 82% (56). Further improvement and automation may increase the utility and accessibility of FA.

FA and fundus photography are comparable for the detection of no or mild and moderate DR (57). Similar results were reported for comparing digital color photography and oral FA (sensitivity for DR, 87% for both methods), although FA was more sensitive for detecting DME (sensitivity 48% for photography and 87% for FA; P < 0.01) (43). In the Diabetes Control and Complications Trial, FA was able to detect “preretinopathy” in 21 and 42% of adults with diabetes, deemed negative with photography or ophthalmoscopy, respectively (58,59).

Drawbacks to using FA as a screening procedure are its invasiveness, time constraints, expensive equipment, and adverse reactions. Allergic-type reactions to sodium fluorescein have been reported in patients undergoing FA, although the incidence of serious complications are rare (60). In general, the use of FA is limited to determining method and location of laser photocoagulation for DME and for assessing the extent of nonperfusion. It has limited value over photography as a diagnostic tool and is not recommended for routine use.

Recent data indicate that annual dilated eye examinations should be implemented from the initial diagnosis of both type 1 and type 2 diabetes (32,61). Referral to an experienced ophthalmologist is required if any level of DME, severe nonproliferative DR, or PDR is detected in the examination (32). Regular follow-up is also essential to ensure early detection, even if no DR was found initially; this is especially important for high-risk patients (32).

Adherence to American Diabetes Association guidelines for annual ophthalmic examination is poor, ranging only from 34 to 65% (62–65). Even among patients at high risk for vision loss (preexistent DR or long duration of diabetes), the rates of adherence were only 61 and 57%, respectively (62). These findings suggest a need for practitioner and patient education about DR and its consequences. A standardized method of referring patients with diabetic eye disease for further evaluation may also be of great value in improving early detection.

Control of the metabolic abnormalities of diabetes has a major effect on the development of diabetic microvascular complications (66). The Diabetes Control and Complications Trial and the U.K. Prospective Diabetes Study showed that optimal metabolic control could reduce the incidence and progression of DR (67,68). The benefits of intensive glycemic control persisted over an extended follow-up (40). Thus, optimal metabolic control should be an important treatment goal and should be implemented early and maintained for as long as it is safely possible (40). Rigid control of hypertension is also effective in reducing disease progression (69–71). Hyperlipidemia has been linked to the presence of retinal hard exudates in patients with DR (13,72), and some evidence suggests that lipid-lowering therapy may reduce hard exudates and microaneurysms (73).

The recommended values for HbA_1c, blood pressure, and LDL cholesterol are <6.5–7%, <130/< 85 mmHg, and <100 mg/dl, respectively (32,74). However, many patients fail to achieve or maintain these levels of metabolic control. In patients who do achieve a significant reduction in HbA_1c, there is an associated increased risk of severe hypoglycemia (67,68,75). Primary care physicians need to recognize correctable risk factors (i.e., hyperglycemia, hypertension, and/or hyperlipidemia) so that appropriate monitoring and referral for eye care can be implemented.

Once sight-threatening DR has been detected, the treatment options are limited. Laser photocoagulation therapy has proven effective in reducing DR progression, and vitrectomy can in many cases prevent severe vision loss in patients with advanced stages of DR. Unfortunately, both treatments carry a risk of additional vision loss, and neither is effective at reversing loss of VA.

Laser photocoagulation is used to treat both DR and DME. The goal of macular laser photocoagulation for DME is to limit vascular leakage through a series of focal laser burns at leaking microaneurysms or grid laser burns in regions of diffuse breakdown of the blood-retinal barrier. The rationale of panretinal photocoagulation for DR is to ablate ischemic areas of the peripheral retina and thereby reduce the induction of angiogenic growth factors. Results of the Diabetic Retinopathy Study demonstrated that panretinal photocoagulation effectively reduces the risk of vision loss in a majority (60%) of patients with PDR (38,76). The ETDRS compared outcomes in eyes assigned to either deferral of macular laser photocoagulation or immediate treatment for clinically significant DME (37). Results showed that macular laser photocoagulation reduced the risk of vision loss by 50% for patients with clinically significant DME (33,37,77,78). Macular focal and grid laser photocoagulation is indicated for clinically significant DME, and panretinal photocoagulation is indicated for high-risk PDR (37,38,79,80).

In more severe cases of DR, specifically those with tractional retinal detachment or severe nonclearing vitreous hemorrhage, vitrectomy is indicated to prevent blindness and/or severe visual loss (80–83). During vitrectomy, incisions are made at the pars plana, a portion of the sclera located posterior to the cornea and lens but anterior to the retina. The procedure may also be used to release vitreoretinal traction by excising membranes causing tractional detachments of the retina (38). In addition, panretinal photocoagulation can be applied during pars plana vitrectomy to treat the underlying PDR. This is typically performed with a fiber optic endolaser probe intraoperatively.

Vitrectomy is clearly beneficial for the treatment of advanced active PDR (84). Early vitrectomy increased the percentage of eyes with a VA of ≥10/20 to 44%, compared with 28% in a conventionally managed group (84). The use of early vitrectomy is also warranted for eyes with very severe PDR, but not for patients with less severe DR (84,85). However, recent advances in surgical techniques and technology since the Diabetic Retinopathy Vitrectomy Study have led to enhancement of the risk-to-benefit ratio for pars plana vitrectomy and widening indications for this procedure.

Given the risk of blindness without treatment, laser photocoagulation and/or vitrectomy will continue to have a major role in the management of DR/DME. Both laser photocoagulation and vitrectomy improve quality of life for patients with DR and are cost-effective (86,87). However, these interventions are indicated only when DR has progressed to a measurably advanced stage in which some VA may already be lost. Side effects, such as loss of peripheral, night, or color vision, are rarely noted by some photocoagulation-treated patients (77). Vitrectomy can accelerate cataract formation and includes risks of retinal detachment and endophthalmitis, which fortunately are rare (88). In some patients treated with photocoagulation, DR continues to progress and ongoing treatment is necessary. DME can also reoccur.

Because of the limitations of current treatments, new pharmacological therapies are being developed, targeting the underlying biochemical mechanisms that cause DR/DME. The rationale behind the use of these agents is the prevention of diabetes-induced damage to the retinal microvasculature. The mechanisms that contribute to cellular damage in the retina include increased flux through the polyol pathway leading to sorbitol accumulation, production of advanced glycation end products (AGEs), increased oxidative stress, and activation of the protein kinase C (PKC)-β pathway (Fig. 1). Each of these mechanisms has been targeted with specific inhibitory compounds, some of which may become viable therapies to treat DR/DME. Blood vessel formation plays a pivotal role in the development of PDR, and various anti-angiogenic agents are also under investigation as potential therapies for DR. Because there is considerable overlap among these and other pathways in the pathogenesis of DR (89,90), combinations of therapies may prove to be more effective in preventing DR.

The hyperglycemia of diabetes leads to an increased flux through the polyol pathway, resulting in elevated levels of sorbitol (91). The net effect is a buildup of intracellular sorbitol and fructose. The ensuing disruption of the osmotic balance in the cell is believed to result in cellular damage (91), which may be important in the loss of integrity of the blood-retinal barrier, among other complications. Loss of retinal pericytes in the earliest stages of DR may be due to their sensitivity to polyols (20,21,92). The “myo-inositol depletion hypothesis” has also been put forth to explain the physiological impairment of retinal pigment epithelial cells and pericytes by hyperglycemia (93–95). Retinal pigment epithelial cells grown under high glucose conditions show marked increases in sorbitol and decreases in myo-inositol content, which were prevented by sorbinil, an aldose reductase inhibitor (ARI) (93). Pericytes grown under high glucose conditions also demonstrate ARI- and myo-inositol–reversible alterations in inositol phospholipid metabolism and DNA synthesis (94). Reducing the intracellular load of sorbitol in the retina and other tissues susceptible to microvascular damage (e.g., nerves and kidney) is a goal of these experimental therapies.

Clinical trials of ARI (sorbinil, ponalrestat, and tolrestat) have been conducted for the treatment of DR (96–99). Unfortunately, ARIs have shown little therapeutic promise for DR thus far. Treatment effects, such as decreases in microaneurysm count (96) and fluorescein leakage (99), are observed in patients. However, there appears to be little significance of these effects on the progression of DR (97,98). One study found that ARI treatment normalized nerve conduction abnormalities in diabetic dogs but had no effect on the development of DR (99,100). Despite numerous attempts to target inhibition of the aldose reductase pathway alone, it appears that this is insufficient to impact diabetic microvascular complications.

Carbohydrates interact with protein side chains in a nonenzymatic fashion to form Amadori products, and these may subsequently form AGEs, especially in the presence of high glucose (101,102). Excessive formation of AGEs has been proposed as another biochemical link between diabetes and the development of microvascular complications. AGEs may affect such functions as enzyme activity, binding of regulatory molecules, and susceptibility of proteins to proteolysis (101). The chronic interaction of these products with at least one specific cell surface receptor for AGEs (AGE-specific receptor) may perpetuate a proinflammatory signaling process and a pro-atherosclerotic state in vascular tissues (103,104). In vitro, the AGE–AGE-specific receptor interaction has been associated with oxidative stress and the activation of nuclear factor-κB, which leads to hyperexpression of proinflammatory cytokines, lymphocyte adhesion molecules (e.g., V-CAM-1), vasoactive mediators, and pro-coagulant factors (104). These processes may result in disruptions of retinal hemodynamics and/or damage to vascular endothelial cells. Accordingly, strategies to reduce AGE formation in the absence of achieving euglycemia have been investigated as potential preventive therapies for diabetic microvascular complications (102).

The inhibition of AGE formation using compounds such as aminoguanidine has been investigated to prevent some of the diabetic vascular abnormalities. AGE accumulation in the retinal capillaries of diabetic rats can be blocked with the use of aminoguanidine (105). The reduction in AGE accumulation was also associated with a reduced number of acellular capillaries and pericyte loss (105). In another rodent model of DR, aminoguanidine reduced retinal oxidative stress and PKC activity caused by diabetes or galactosemia (106), suggesting that these pathways may also be involved in its beneficial effects on DR. In diabetic dogs, aminoguanidine effectively prevented the development of DR but did not have a significant effect on AGE formation (107). The utility of these compounds for the prevention of DR remains to be proven in humans (108).

Experimental studies have shown that PKC activity and levels of diacylglycerol (DAG), an activator of PKC, are increased after exposure of vascular tissues to elevated glucose (109,110). Diabetes-induced DAG may derive from hydrolysis of phosphatidylinositides, metabolism of phosphatidylcholine, or de novo synthesis of phosphatidic acid (26). PKC activity is also increased after exposure of vascular endothelial cells to oxidative stress, another mechanism implicated in the development and progression of diabetic microvascular complications (90,111). PKC-β and -δ have been identified as the predominant isoforms activated in vascular tissues in response to hyperglycemia (26,109). PKC-β has been shown to have an important role in regulating endothelial cell permeability (112) and is an important signaling component for VEGF (113). Transgenic animals overexpressing PKC-β in vascular tissues developed retinal hemodynamic abnormalities similar to those observed in human DR (114).

The role of PKC in many cellular processes suggests that inhibition of all PKC isoforms would cause unacceptable toxicity (115). Ruboxistaurin (LY333531), a specific inhibitor of PKC-β1 and -β2 (115), has been shown to prevent and reverse microvascular complications in animal models of diabetes (116), to block neovascularization associated with retinal ischemia (117), and to inhibit the effect of VEGF on retinal permeability and endothelial cell growth (118). In patients with minimal DR, ruboxistaurin reversed retinal blood flow abnormalities and was well tolerated (119). Additional trials are evaluating the utility of ruboxistaurin for the treatment and prevention of DR and DME.

PKC412 inhibits the α, β, and γ isoforms with similar half-maximal inhibition concentration (IC₅₀) values ranging from 22 to 31 nmol/l and may be useful in the treatment of DR and other disorders (120,121). The kinase domain of the human VEGF receptor-2 (KDR) and the platelet-derived growth factor receptor β are also inhibited by PKC412 at IC₅₀ values ranging from 20 to 100 nmol/l (120). In an animal model of neovascularization, PKC412 inhibited ischemia-induced angiogenesis as well as retinal vessel formation during development (120). However, early pharmacodynamic studies of PKC412 have resulted in some adverse outcomes, which may reflect PKC412’s relative lack of specificity (122). Further trials will be needed to determine whether this compound will be useful in the treatment of DR in humans.

Production of reactive oxygen species (ROS) has been implicated in the development of diabetic complications (123). Diabetes may cause ROS production through glucose auto-oxidation, increased flux through the polyol pathway, and increases in protein glycation (123). ROS may activate aldose reductase and PKC and increase AGE production and DAG formation (90).

The pervading role of ROS in the biochemical processes leading to microvascular damage has prompted an investigation of antioxidants as preventive therapy for diabetic complications (124). Inhibition of superoxide production can effectively block sorbitol accumulation, AGE formation, and PKC activation (125). These findings suggest that ROS production is associated with at least three mechanisms of diabetes-induced vascular damage. Antioxidants are effective inhibitors of pericyte loss secondary to diabetes in experimental models (21,126). Tocopherol also inhibits hyperglycemia-induced DAG production and PKC (124,127). Tocopherol prevents retinal hemodynamic abnormalities in diabetic rats (127). In patients with type 1 diabetes and little or no baseline retinopathy, retinal blood flow was significantly increased after 4 months of tocopherol therapy (P < 0.001) (128). Recent results of the Heart Outcomes Prevention Evaluation study and other trials showed a lack of effect of tocopherol in the prevention of cardiovascular risks, despite suggestive evidence to the contrary (129–131).

New blood vessel (angiogenesis) formation is central to the pathology of PDR and is stimulated by such factors as VEGF in response to retinal ischemia, which occurs because of capillary loss and/or microaneurysm formation (29). VEGF is a key mediator of angiogenesis in the retina (30). Clinical studies have shown that VEGF levels increase in patients as they progress from nonproliferative DR to active PDR (29,30). Successful panretinal photocoagulation has been found to reduce intraocular VEGF levels by 75% (P = 0.008) in patients treated for ocular neovascularization (29). This suggests that specific inhibition of VEGF activity may prevent retinal neovascularization and associated blood flow abnormalities. Inhibition of VEGF signaling using ruboxistaurin prevented VEGF-induced increases in vascular permeability (118).

The importance of angiogenesis in the pathology of DR has prompted the investigation of angiostatic therapies for the treatment of DR and DME (132). Early results in mice have been encouraging (133). The role of VEGF in retinal neovascularization has also prompted the development of VEGF-specific inhibitors such as antibodies to VEGF (134). These antibodies may be especially useful for the prevention of neovascularization during the very early stages of PDR. Another potent endogenous inhibitor of angiogenesis is pigment endothelium-derived factor (PEDF). PEDF inhibits angiogenesis induced by a wide variety of growth factors in addition to VEGF (135,136). In a mouse model of ischemia-induced retinopathy, systemic administration of recombinant PEDF completely inhibited the development of ischemia-provoked retinal vascular anomalies without affecting the development of normal retinal vessels (136). These findings suggest that PEDF may be useful as a primary intervention in the treatment of early DR.

Somatostatin activity is linked with the progression of DR (137,138), and hypophysectomy has been proposed as an intervention for severe treatment-resistant DR. Consequently, somatostatin has been evaluated for the treatment of DR (137–139). Early results in patients with PDR were encouraging, although some evidence for resistance to the drug was noted (137,139). In a recent trial of patients with severe nonproliferative DR or early PDR, therapy with octreotide (a somatostatin analog) decreased the need for retinal photocoagulation compared with conventional treatment (1/22 vs. 9/24 eyes) (138). However, the incidence of progression to severe PDR was not significantly different between the treatment arms (138). Additional controlled trials will determine whether somatostatin analog therapy is a viable therapeutic option for patients with more advanced stages of DR.

Experimental results indicate that high-dose aspirin suppresses diabetes-induced retinal TNF expression, nuclear factor-κB activity, and leukocyte cell adhesion molecule expression, which are implicated in endothelial cell injury and breakdown of the blood-retinal barrier (140). There is evidence that aspirin alone or in combination with dipyridamole could decrease the yearly increase in microaneurysms in patients with early DR (141). However, among patients with mild to severe nonproliferative DR, aspirin therapy (650 mg/day) was shown to have no significant effect on DR progression nor to confer any increased risk of vitreous hemorrhage (142). Thus, whereas aspirin prophylaxis may be useful in the early stages of DR and DME, this benefit appears to be lost in later stages of disease.

There is evidence suggesting that inhibitors of the renin-angiotensin system may have additional effects on DR that are independent of their hypotensive abilities (69,71,143). Lisinopril, an ACE inhibitor, reduced retinal neovascularization, VEGF, and VEGF type 2 receptor expression in a rat model of retinopathy of prematurity (144). Similarly, diabetes-induced retinal VEGF expression and hyperpermeability were also inhibited by a similar treatment (145). Whether these beneficial effects are independent of the hypotensive action of these compounds is not yet known; nonetheless, their inhibitory effects on VEGF expression may be especially important in the early stages of PDR. Candesartan, a potent angiotensin II receptor antagonist, decreases VEGF expression and ameliorates retinal abnormalities in diabetic rats (146).

The evaluation of pharmacological interventions for a given condition requires standardized clinically meaningful end points that can be assessed in a quantitative manner. In the case of DR, the scale most widely used is the ETDRS DR severity scale (34). As discussed above, this scale has been used to assess the severity of retinopathy in major intervention trials for diabetes control (67). Other trials evaluated the effects of diabetes control on the development and progression of DR, using such end points as the need for photocoagulation or the development of PDR or severe nonproliferative DR. The use of these clinical end points implies that some degree of sight-threatening DR already exists in a subject, and end points such as these are of limited value in discerning the effects of pharmacological therapies on early DR. A more efficient way to assess these earlier stages in clinical trials would be to assess stepwise progression on the ETDRS scale. Data from the Wisconsin Epidemiologic Study of Diabetic Retinopathy demonstrate that a progression of one or more steps on the ETDRS scale is predictive of the development of PDR or clinically significant DME (relative risk, 5.9 and 3.8, respectively) (36). The use of the ≥1 step ETDRS progression as an end point in clinical trials of new pharmacological agents will reduce the sample size and trial duration needed to assess efficacy (36). Although the ETDRS staging system is widely used in clinical trials, it is rarely used clinically, given its complexity. Consequently, pharmacological agents that alter the risk of progression along steps of the ETDRS scale may not be clinically meaningful to practitioners. It will be necessary to correlate changes in steps of ETDRS progression to the International Clinical Diabetic Retinopathy and Diabetic Macular Edema Disease Severity Scales.

As new treatments become available, alternative outcomes must also be devised that can accurately predict the progression of DR from its initial stages. For example, the detection and quantitation of microaneurysm formation and/or other retinal capillary abnormalities using FA may be a sufficiently sensitive means to evaluate the effect of pharmacological therapy on early changes in the blood-retinal barrier (55). The situation is greatly complicated by the nonlinear progression of DR. Clinical signs may spontaneously resolve and thus may mask true progression of the disease (55). There is no straightforward method to address this issue at present and more research is needed in this area. The use of multiple surrogate end points, such as a two-step ETDRS progression, measurement of retinal thickness, and assessment of fluorescein leakage, has been proposed as one means to overcome this obstacle (55). Such outcomes will need to be validated in clinical trials to determine whether they can accurately predict progression of DR in a clinically useful manner.

Diabetic eye disease severely affects quality of life for patients with diabetes by decreasing VA and increasing the risk of blindness. The DR condition results in loss of capillary integrity, microaneurysm formation, and ischemia, which in turn drive the progression of PDR. Accumulation of fluid in the retina secondary to capillary leakage and/or microaneurysms results in DME, which contributes to loss of vision in DR. There is substantial evidence that control over metabolic factors can effectively prevent the development and progression of DR/DME. However, many patients fail to achieve or maintain optimal levels of metabolic control. For such patients, early detection and timely treatment of DR remains the standard of care. Although they are effective, sight-saving interventions, laser photocoagulation therapy, and vitrectomy are invasive, associated with destructive side effects, and only treat the late stages of disease. A number of pharmacological agents that could slow the progression of DR/DME in earlier stages are now being tested. These therapies have derived from improved understanding of the complex and often overlapping pathways involved in diabetes-induced microvascular damage. It is likely that one or more of these pharmacological interventions, or possibly combinations thereof, will be effective in reducing the progression of DR and DME and the associated vision loss. With the introduction of these therapies in the coming years, there will be a need for improved screening. It is critical that health care providers interact with one another in managing patients to ensure that high-risk individuals are screened early. Technological advances are giving patients more access to proper screening and may make this a more achievable goal.

The authors thank Drs. Erin A. Kingshill and Peter D. Ho for their help with the manuscript.