Introduction
Successful management of diabetic retinopathy requires not only local treatment (laser therapy, intravitreal pharmacotherapy, and vitrectomy) but also systemic control of hyperglycemia, blood pressure, and lipids. If fundus examinations are initiated before the development of significant retinopathy and repeated periodically—and if the recommendations of the Early Treatment Diabetic Retinopathy Study (ETDRS) and recent anti-vascular endothelial growth factor (VEGF) clinical trials results are applied in the management of diabetic macular edema or neovascularization—the risk of severe visual loss is less than 5%. However, diabetic retinopathy remains the number one cause of new blindness in most industrialized countries because of delays in seeking treatment. The vast majority of diabetic individuals who lose vision do so not because of an inability to treat their disease but due to a delay in seeking medical attention. In addition, in many countries, the incidence of diabetes is increasing dramatically.
Epidemiology
Duration of retinopathy is most closely associated with the incidence of diabetic retinopathy and remains the best predictor of diabetic retinopathy. The first five years of type 1 diabetes has a very low risk of retinopathy. However, 27% of those who have had diabetes for 5–10 years and 71%–90% of those who have had diabetes for longer than 10 years have diabetic retinopathy. After 20–30 years, the incidence rises to 95%, and about 30%–50% of these patients have proliferative diabetic retinopathy (PDR).
Yanko et al. found that the prevalence of retinopathy 11–13 years after the onset of type 2 diabetes was 23%; after 16 or more years, it was 60%; and 11 or more years after the onset, 3% of the patients had PDR. Klein et al. reported that 10 years after the diagnosis of type 2 diabetes, 67% of patients had retinopathy and 10% had PDR.
The duration of diabetes after the onset of puberty appears to be most important. For example, the risk of retinopathy is roughly the same for two 25-year-old patients, one of whom developed diabetes at the age of 6 and the other at the age of 12 years. The risk of retinopathy in children diagnosed prior to the age of 2 years has a negligible risk of retinopathy for the first 10 years.
The Diabetes Control and Complications Trial (DCCT) showed that type 1 diabetics who closely monitored their blood glucose via tight control (four measurements per day = tight control) had a 76% reduction in the rate of development of any retinopathy (primary prevention cohort) and a 54% reduction in progression of established retinopathy (secondary intervention cohort) as compared with the conventional treatment group (one measurement per day). For advanced retinopathy, however, even the most rigorous control of blood glucose may not prevent progression. The DCCT was halted early after 6.5 years when the benefit of tight control was deemed unlikely to be reversed with time. Most of the participants were followed in the Epidemiology of Diabetes Interventions and Complications (EDIC) study. The EDIC study showed continued benefit for the former tight control group over the former conventional treatment group, despite normalization of glucose control even after 7 years of follow-up. The value of intensive treatment has also been demonstrated for type 2 diabetes. The United Kingdom Prospective Diabetes Study (UKPDS) revealed a 21% reduction in the 1-year rate of progression of retinopathy. The DCCT has shown that for every 1% decrease in the hemoglobin A1C (HbA1C) level, the incidence of diabetic retinopathy decreases 28%.
It is important to realize that many patients with diabetes are unaware of their diabetic retinopathy. A Joslin Diabetes Center study on self-awareness of diabetic retinopathy found that 83% of patients with diabetic retinopathy and 78% of those with vision-threatening disease were unaware they had diabetic retinopathy at their first visit. Communication and education of the patient are very important. The Diabetic Retinopathy Clinical Research Network (DRCR) evaluated the effect on retinopathy of educating the diabetic patient about their HbA1C and blood pressure levels at their regular retinal examination visits. Despite physician-initiated discussion of in-office measurements of HbA1C and blood pressure readings with patients at their scheduled visits, the HbA1C levels did not significantly change from baseline. This suggests that more frequent and/or intensive intervention is needed than checking of these levels and discussing the results during a routine retinal evaluation. Renal disease, as evidenced by proteinuria, elevated blood urea nitrogen levels, elevated blood creatinine levels, and even microalbuminuria, is an excellent predictor of the presence of coexisting retinopathy. Among patients with symptomatic retinopathy, 35% have proteinuria, elevated blood urea nitrogen values, or elevated creatinine levels. Systemic hypertension is another independent risk factor for diabetic retinopathy. The UKPDS demonstrated that tighter blood pressure control significantly reduced the progression of diabetic retinopathy.
In pregnant diabetic women without retinopathy, the risk of developing nonproliferative diabetic retinopathy (NPDR) is about 10%. In contrast, those with NPDR and systemic hypertension at the onset of pregnancy—or those who develop systemic hypertension—are likely to show progression, with increased hemorrhages, cotton–wool spots, and macular edema. Fortunately, there is usually some regression after delivery. About 4% of pregnant women with NPDR progress to PDR. Those with untreated PDR at the onset of pregnancy frequently do poorly unless they are treated with panretinal photocoagulation (PRP). Previously treated PDR usually does not worsen during pregnancy. Women who begin pregnancy with poorly controlled diabetes who are suddenly brought under strict control frequently have severe deterioration of their retinopathy and do not always recover after delivery. It is very important to monitor pregnant patients carefully and closely, with some clinicians advocating quarterly examinations. Anti-VEGF treatment has not been studied in pregnant women. Laser or other non–anti-VEGF treatment is recommended. In addition, fluorescein dye is known to cross the placenta and is also excreted in breast milk for up to 72 hours.
Some drugs have been implicated in worsening retinopathy and in particular diabetic macular edema (DME). Glitazones are associated with an increased risk of worsening of DME. A study of 170 000 patients in the Kaiser Permanente database showed that patients treated with glitazones had a significantly higher risk of developing DME (odds ratio 2.6, 95% confidence interval 2.4–3.0) in a univariate analysis. After adjusting for confounding factors, the odds ratio was 1.6, 95% confidence interval 1.4–1.8. Some clinical trials have shown a higher risk of DME in patients taking insulin and glitazones (Actos package insert. Deerfield IL: Takeda Pharmaceutical America, Inc, 2011). It is therefore important to review a patient’s drug history when evaluating diabetic patients with macular edema. The glitazone rosiglitazone, however, was shown to lower the risk of progression to PDR and to have similar rates of DME as controls at 3 years. In The Health Improvement Network (THIN) database, a large retrospective cohort study of 103 368 patients with type II diabetes mellitus and no DME at baseline, an increased risk of DME was found at both 1 and 10 years for users of thiazolidinediones compared to nonusers. The adjusted odds ratio was 2.3 at 1 year and the adjusted hazard ratio was 2.3 at 10 years.
Pathogenesis
The final metabolic pathway that causes diabetic retinopathy is unknown. There are several theories that are not mutually exclusive.
Aldose Reductase
Aldose reductase converts sugars into their alcohols. For example, glucose is converted to sorbitol and galactose is converted to galactitol. However, sorbitol and galactitol cannot easily diffuse out of cells, causing increased intracellular concentration. Osmotic forces then cause water to diffuse into the cell. The resultant damage to lens epithelial cells, which have a high concentration of aldose reductase, is responsible for the cataract seen in children. Because aldose reductase is also found in high concentration in retinal pericytes and Schwann cells, some investigators suggest that diabetic retinopathy and neuropathy may be caused by aldose reductase–mediated damage. Despite these theoretical benefits, clinical trials have thus far failed to show a reduction in the incidence of diabetic retinopathy or of neuropathy by aldose reductase inhibitors, possibly because an effective aldose reductase inhibitor with few systemic side effects has yet to be developed.
Vasoproliferative Factors
The retina and retinal pigment epithelium release vasoproliferative factors, such as VEGF, which induce neovascularization. VEGF has a direct role in the proliferative retinal vascular abnormalities that are found in diabetes. Animal models have demonstrated that VEGF expression correlates with the development and regression of neovascularization. The concentration of VEGF in aqueous and vitreous directly correlates with the severity of retinopathy. VEGF is a potent vasopermeability factor and is responsible for DME. Several randomized controlled clinical trials have shown efficacy of anti-VEGF treatments for DME. There are other vasoactive cytokines released in diabetic eyes. These include tissue growth factor beta and connective tissue growth factor. The inflammatory component results from macrophage and complement activation. Extensive, dense deposition of C5b-9 as well as vitronectin were found in the connective matrix of the choriocapillaris. It is believed that complement activation results in increased neutrophils, which then cause endothelial damage. Lipids and proteins leak out of the capillaries. Extracellular matrix deposition may be triggered by the complement cascade’s effects on neighboring cells and result in thickened choriocapillaris and Bruch’s membrane. Inflammation thus plays a role in macular edema and diabetic retinopathy. It is believed that long-standing DME may have more of an inflammatory component and be more responsive to corticosteroids, which are also antiangiogenic.
Platelets and Blood Viscosity
Diabetes is associated with abnormalities of platelet function. It has been postulated that platelet abnormalities or alterations in blood viscosity in diabetics may contribute to diabetic retinopathy by causing focal capillary occlusion and focal areas of ischemia in the retina.
Aldose Reductase
Aldose reductase converts sugars into their alcohols. For example, glucose is converted to sorbitol and galactose is converted to galactitol. However, sorbitol and galactitol cannot easily diffuse out of cells, causing increased intracellular concentration. Osmotic forces then cause water to diffuse into the cell. The resultant damage to lens epithelial cells, which have a high concentration of aldose reductase, is responsible for the cataract seen in children. Because aldose reductase is also found in high concentration in retinal pericytes and Schwann cells, some investigators suggest that diabetic retinopathy and neuropathy may be caused by aldose reductase–mediated damage. Despite these theoretical benefits, clinical trials have thus far failed to show a reduction in the incidence of diabetic retinopathy or of neuropathy by aldose reductase inhibitors, possibly because an effective aldose reductase inhibitor with few systemic side effects has yet to be developed.
Vasoproliferative Factors
The retina and retinal pigment epithelium release vasoproliferative factors, such as VEGF, which induce neovascularization. VEGF has a direct role in the proliferative retinal vascular abnormalities that are found in diabetes. Animal models have demonstrated that VEGF expression correlates with the development and regression of neovascularization. The concentration of VEGF in aqueous and vitreous directly correlates with the severity of retinopathy. VEGF is a potent vasopermeability factor and is responsible for DME. Several randomized controlled clinical trials have shown efficacy of anti-VEGF treatments for DME. There are other vasoactive cytokines released in diabetic eyes. These include tissue growth factor beta and connective tissue growth factor. The inflammatory component results from macrophage and complement activation. Extensive, dense deposition of C5b-9 as well as vitronectin were found in the connective matrix of the choriocapillaris. It is believed that complement activation results in increased neutrophils, which then cause endothelial damage. Lipids and proteins leak out of the capillaries. Extracellular matrix deposition may be triggered by the complement cascade’s effects on neighboring cells and result in thickened choriocapillaris and Bruch’s membrane. Inflammation thus plays a role in macular edema and diabetic retinopathy. It is believed that long-standing DME may have more of an inflammatory component and be more responsive to corticosteroids, which are also antiangiogenic.
Platelets and Blood Viscosity
Diabetes is associated with abnormalities of platelet function. It has been postulated that platelet abnormalities or alterations in blood viscosity in diabetics may contribute to diabetic retinopathy by causing focal capillary occlusion and focal areas of ischemia in the retina.
Ocular Manifestations
The earliest stage of diabetic retinopathy is NPDR. In some patients, there is progression to proliferative retinopathy PDR. The incidence of more advanced levels of retinopathy increases based on duration of disease and glycemic control.
Early Nonproliferative Diabetic Retinopathy
Microaneurysms are the first ophthalmoscopically detectable change in diabetic retinopathy and are considered the hallmark of NPDR ( Fig. 6.22.1A ). They are seen as small red dots in the middle retinal layers, typically in the macula. When the wall of a capillary or microaneurysm is weakened enough, it may rupture, giving rise to an intraretinal hemorrhage. If the hemorrhage is deep (i.e., in the inner nuclear layer or outer plexiform layer), it usually is round or oval (“dot or blot”) (see Fig. 6.22.1A ). It is very difficult to distinguish a small dot hemorrhage from a microaneurysm by ophthalmoscopy. Fluorescein angiography helps to distinguish patent (and not one filled with clotted blood) microaneurysms because they leak dye (see Fig. 6.22.1B ). If the hemorrhage is superficial, in the nerve fiber layer, it takes a flame or splinter shape indistinguishable from a hemorrhage seen in hypertensive retinopathy ( Figs. 6.22.2 and 6.22.3 ). Diabetics who have normal blood pressure may have multiple splinter hemorrhages. Nevertheless, the presence of numerous splinter hemorrhages in a diabetic patient should prompt a blood pressure check.
DME (see Fig. 6.22.1A ) represents the leading cause of legal blindness in diabetics. The intercellular fluid comes from leaking microaneurysms or from diffuse capillary incompetence. Clinically, DME is best detected by slit-lamp biomicroscopy with a contact macular lens, although noncontact macular lenses can be used. The edema causes separation of cells, resulting in scattering of light by the multiple interfaces. This decreases the retina’s translucency such that the normal retinal pigment epithelial and choroidal background pattern is blurred (see Fig. 6.22.1A ). Pockets of fluid in the outer plexiform layer, if large enough, can be seen as cystoid macular edema (CME). Usually CME is seen in eyes that have other signs of severe NPDR. In rare cases, CME is due to generalized diffuse leakage from the entire capillary network and can be seen in eyes with very few other signs of diabetic retinopathy.
If the leakage of fluid is severe enough, lipid may accumulate in the retina (see Fig. 6.22.1A ); again, the outer plexiform layer is first to be affected. In some cases, lipid is scattered through the macula. In others, it accumulates in a ring around a group of leaking microaneurysms or around microaneurysms surrounding an area of capillary nonperfusion. This pattern is called circinate retinopathy (see Fig. 6.22.1A ).
The application of optical coherence tomography (OCT) to management of DME has been very useful. The degree of DME and response to therapy can be quantified on OCT. Specifically, the central subfield thickness (CST) can be used to follow a patient’s response to treatment of DME. In addition, the OCT presents qualitative information such as the presence of cysts, hard exudates, and degree of inner or outer retinal or external limiting membrane disruption and subretinal fluid. These findings are useful in following a patient’s response to therapy.
In eyes treated with anti-VEGFs, the presence of disorganization of retinal inner layers (DRIL) has been associated with poorer visual acuity outcomes. The length of DRIL was associated with subsequent vision. The change in DRIL was associated with change in visual acuity (VA), with resolution of DRIL having the best VA. Early change in the extent of DRIL is inversely predictive of subsequent changes in visual acuity.
Advanced Nonproliferative Diabetic Retinopathy
In advanced NPDR, signs of increasing inner retinal hypoxia appear, such as multiple retinal hemorrhages, cotton–wool spots (see Fig. 6.22.3 ), venous beading and vascular loops ( Fig. 6.22.4 ), intraretinal microvascular abnormalities (IRMAs) (see Figs. 6.22.1A and 6.22.4 ), and large areas of capillary nonperfusion seen on fluorescein angiography.
Cotton–wool spots, also called soft exudates or nerve fiber infarcts, result from ischemia, not exudation. Local ischemia causes effective obstruction of axoplasmic flow in the normally transparent nerve fiber layer, and the subsequent swelling of the nerve fibers gives cotton–wool spots their characteristic white fluffy appearance. Fluorescein angiography shows lack of capillary perfusion in the area corresponding to a cotton–wool spot. Microaneurysms frequently surround the hypoxic area (see Fig. 6.22.3 ).
Venous beading (see Fig. 6.22.4 ) is an important sign of sluggish retinal circulation. Venous loops are nearly always adjacent to large areas of capillary nonperfusion. IRMAs are dilated capillaries that seem to function as collateral channels and are frequently difficult to differentiate from surface retinal neovascularization. Fluorescein dye, however, does not leak from IRMAs but leaks profusely from neovascularization. Capillary hypoperfusion often surrounds IRMA (see Fig. 6.22.4 ).
The ETDRS found that IRMAs, multiple retinal hemorrhages, venous beading and loops, widespread capillary nonperfusion, and widespread leakage on fluorescein angiography were all significant risk factors for the development of PDR. Interestingly, cotton–wool spots were not.
Proliferative Diabetic Retinopathy
Although the macular edema, exudates, and capillary occlusions seen in NPDR often cause legal blindness, affected patients usually maintain at least ambulatory vision. PDR, on the other hand, may result in severe vitreous hemorrhage or retinal detachment, with hand-movements vision or worse. Approximately 50% of patients with very severe NPDR progress to PDR within 1 year. Proliferative vessels usually arise from retinal veins and often begin as a collection of multiple fine vessels. When they arise on or within 1 disc diameter of the optic nerve they are referred to as NVD (neovascularization of the disc, Figs. 6.22.5 and 6.22.6 ). When they arise further than 1 disc diameter away, they are called NVE (neovascularization elsewhere) (see Fig. 6.22.6B ). Unlike normal retinal vessels, NVD and NVE leak fluorescein into the vitreous.
Once the stimulus for growth of new vessels is present, the path of subsequent growth taken by neovascularization is along the route of least resistance. For example, the absence of a true internal limiting membrane on the disc could explain the prevalence of new vessels at that location. Also, neovascularization seems to grow more easily on a preformed connective tissue framework. Thus, a shallowly detached posterior vitreous face is a frequent site of growth of new vessels.
The new vessels usually progress through a stage of further proliferation, with associated connective tissue formation. As PDR progresses, the fibrous component becomes more prominent, with the fibrotic tissue being either vascular or avascular. The fibrovascular variety usually is found in association with vessels that extend into the vitreous cavity or with abnormal new vessels on the surface of the retina or disc. The avascular variety usually results from organization or thickening of the posterior hyaloid face. Vitreous traction is transmitted to the retina along these proliferations and may lead to traction retinal detachment.
NVE nearly always grows toward and into zones of retinal ischemia until posterior vitreous detachment occurs (see Fig. 6.22.6 ). Then the vessels are lifted into the vitreous cavity. The end stage is characterized by regression of the vascular tissue. Sometimes there may be contraction of the connective tissue components, development of subhyaloid bands, thickening of the posterior vitreous face, and the appearance of retinoschisis, retinal detachment, or formation of retinal breaks.
Posterior vitreous detachment in diabetics is characterized by a slow, overall shrinkage of the entire formed vitreous rather than by the formation of cavities caused by vitreous destruction. Davis et al. have stressed the role of the contracting vitreous in the production of vitreous hemorrhage, retinal breaks, and retinal detachment. Neovascular vessels do not “grow” forward into the vitreous cavity but are pulled into the vitreous by the contracting vitreous to which they adhere. Confirmation of the importance of the vitreous in the development and progression of PDR comes from the long-term follow-up of eyes that have undergone successful vitrectomy in which neovascularization shrinks, fluorescein leakage decreases, and new areas of neovascularization rarely arise.
It has long been assumed that sudden vitreous contractions tear the fragile new vessels, causing vitreous hemorrhage. However, the majority of diabetic vitreous hemorrhages occur during sleep, possibly because of an increase in blood pressure secondary to early morning hypoglycemia or to rapid eye movement sleep. Because so few hemorrhages occur during exercise, it is not necessary to restrict the activity of patients with PDR. When a hemorrhage occurs, if the erythrocytes are behind the posterior vitreous face, they usually quickly settle to the bottom of the eye and are absorbed. However, when erythrocytes break into the vitreous body, they adhere to the gel, and clearing may take months or years.
A large superficial hemorrhage may separate the internal limiting membrane from the rest of the retina. Such hemorrhages usually are round or oval but also may be boat shaped. The blood may remain confined between the internal limiting membrane and the rest of the retina for weeks or months before breaking into the vitreous. Subinternal limiting membrane hemorrhages were formerly thought to occur between the internal limiting membrane and the cortical vitreous and were called subhyaloid or preretinal hemorrhages. It is now felt that true subhyaloid hemorrhages probably are quite rare. Tight subinternal limiting membrane hemorrhages are dangerous because they may progress rapidly to traction retinal detachment.
As the vitreous contracts, it may pull on the optic disc, causing traction striae involving the macular area or actually drag the macula itself, both of which contribute to decreased visual acuity.
Two types of diabetic retinal detachments occur, those that are caused by traction alone (nonrhegmatogenous) and those caused by retinal break formation (rhegmatogenous). Characteristics of nonrhegmatogenous (traction) detachment in PDR include:
- •
The detached retina usually is confined to the posterior fundus and infrequently extends more than two thirds of the distance to the equator.
- •
The detached retina has a taut and shiny surface.
- •
The detached retina is concave toward the pupil.
- •
No shifting of subretinal fluid occurs.
Occasionally, a spontaneous decrease in the extent of a traction detachment may occur, but this is the exception rather than the rule. Traction on the retina also may cause focal areas of retinoschisis, which may be difficult to distinguish from full-thickness retinal detachment. In retinoschisis the elevated layer is thinner and more translucent.
When a detachment is rhegmatogenous, the borders of the elevated retina usually extend to the ora serrata. The retinal surface is dull and grayish and undulates because of retinal mobility due to shifting of subretinal fluid. Retinal breaks are usually in the posterior pole near areas of fibrovascular change. The breaks are oval in shape and appear to be partly the result of tangential traction from the proliferative tissue, as well as being due to vitreous traction. Determination of the location of retinal holes may be complicated by many factors, particularly poor dilatation of the pupil, lens opacity, increased vitreous turbidity, vitreous hemorrhage, intraretinal hemorrhage, and obscuration of the breaks by overlying proliferative tissue.
Other Ocular Complications of Diabetes Mellitus
Cornea
Corneal sensitivity is decreased in proportion to both the duration of the disease and the severity of the retinopathy. Corneal abrasions are more common in people with diabetes, presumably because adhesion between the basement membrane of the corneal epithelium and the corneal stroma is not as firm as that found in normal corneas. Following vitrectomy, recurrent corneal erosion, striate keratopathy, and corneal edema are more common in diabetics than in nondiabetics.
Glaucoma
The relationship between diabetes and primary open-angle glaucoma is unclear. Some population-based studies have found an association but others have not.
Neovascularization of the iris (NVI) usually is seen only in diabetics who have PDR. PRP not only has protective value against NVI, it also is an effective treatment against established NVI. If the media are clear, PRP should be performed prior to any other treatment for NVI, even in advanced cases. If the media are too cloudy for PRP, transscleral laser or peripheral retinal cryoablation are alternative means of treatment (see later in this chapter). The presence of rubeotic glaucoma is a poor prognostic indicator for visual acuity and for life expectancy. Anti-VEGF therapy is also used as an adjunctive treatment in eyes with NVI.
Lens
The risk of cataract is 2–4 times greater in diabetics than in nondiabetics and may be 15–25 times greater in diabetics under 40 years old.
Patients with diabetes mellitus who have no retinopathy have excellent results from cataract surgery, with 90%–95% having a final visual acuity of 20/40 or better, but chronic CME is about 14 times more common in diabetics than in nondiabetics. The best-known predictor of postoperative success is the preoperative severity of retinopathy. It was hoped that modern surgery, which leaves an intact posterior capsule, would protect the eye from NVI by reducing the diffusion of vasoproliferative factors into the anterior chamber, but several studies have shown that it does not. Furthermore, a neodymium-aluminum-garnet (Nd:YAG) laser capsulotomy does not increase the risk. Other anterior segment complications that are more common in diabetics than in nondiabetics are pupillary block, posterior synechiae, pigmented precipitates on the implant, and severe iritis.
Posterior segment complications of cataract surgery include macular edema, PDR, vitreous hemorrhage, and traction retinal detachment. Unlike prior reports, recent reports suggest that modern uncomplicated cataract surgery may not accelerate progression of diabetic retinopathy in type 2 diabetics with NPDR. Caution should be observed when considering cataract surgery in patients who have diabetic retinopathy, but up to 70% of these patients can attain a final visual acuity of 20/40 or better. In a pilot observational study of eyes with baseline DME at the time of cataract surgery, the DRCR showed that 32% improved four lines and 10% worsened at least two lines by week 16. The study was limited by small enrollment and concluded that it was unlikely that an adequate sample could be recruited within a reasonable time to pursue an interventional trial for eyes with DME in the setting of cataract surgery. For patients without macular edema at the time of cataract surgery, the DRCR found that preoperative noncentral DME or a history of DME treatment may increase the risk of developing central-involved DME 16 weeks after cataract extraction. In addition, visual outcomes were good, with over 85% achieving 20/40 or better visual acuity. However, this was achieved less often in eyes that developed central-involved ME (67%), although these eyes had a lower mean baseline VA (Snellen equivalent 20/63) than eyes that did not develop central-involved ME (Snellen equivalent 20/50).
Cataract surgery in patients with active PDR often results in poorer postoperative visual outcome because of the high risk of both anterior and posterior segment complications. In one series, no patient with active PDR or preproliferative diabetic retinopathy achieved better than 20/80. Most experts recommend aggressive preoperative PRP.
Optic Neuropathy
As demonstrated by increased latency and decreased amplitude of the visual evoked potential, many diabetic patients without retinopathy have subclinical optic neuropathy. They have an increased risk for anterior ischemic optic neuropathy. In addition, diabetics are susceptible to diabetic papillopathy, which is characterized by acute disc edema without the pale swelling of anterior ischemic optic neuropathy. It is bilateral in one-half of cases and may not show an afferent pupillary defect. Macular edema is a common concurrent finding and is the most common cause of failure of visual recovery in these patients. Visual fields may be normal or show an enlarged blind spot or other nerve fiber defects. The prognosis is excellent, with most patients recovering to 20/50 or better.
Cranial Neuropathy
Extraocular muscle palsies may occur in diabetics secondary to neuropathy involving the third, fourth, or sixth cranial nerves. The mechanism is believed to be a localized demyelinization of the nerve secondary to focal ischemia. Pain may or may not be experienced, and not infrequently extraocular muscle palsy may be the initial clue to a latent diabetic condition. Recovery of extraocular muscle function in diabetic cranial nerve palsies generally takes place within 1–3 months. When the third cranial nerve is involved, pupillary function is usually normal. This pupillary sparing in diabetic third cranial nerve palsy is an important diagnostic feature, helping to distinguish it from an intracranial tumor or aneurysm.
Early Nonproliferative Diabetic Retinopathy
Microaneurysms are the first ophthalmoscopically detectable change in diabetic retinopathy and are considered the hallmark of NPDR ( Fig. 6.22.1A ). They are seen as small red dots in the middle retinal layers, typically in the macula. When the wall of a capillary or microaneurysm is weakened enough, it may rupture, giving rise to an intraretinal hemorrhage. If the hemorrhage is deep (i.e., in the inner nuclear layer or outer plexiform layer), it usually is round or oval (“dot or blot”) (see Fig. 6.22.1A ). It is very difficult to distinguish a small dot hemorrhage from a microaneurysm by ophthalmoscopy. Fluorescein angiography helps to distinguish patent (and not one filled with clotted blood) microaneurysms because they leak dye (see Fig. 6.22.1B ). If the hemorrhage is superficial, in the nerve fiber layer, it takes a flame or splinter shape indistinguishable from a hemorrhage seen in hypertensive retinopathy ( Figs. 6.22.2 and 6.22.3 ). Diabetics who have normal blood pressure may have multiple splinter hemorrhages. Nevertheless, the presence of numerous splinter hemorrhages in a diabetic patient should prompt a blood pressure check.
DME (see Fig. 6.22.1A ) represents the leading cause of legal blindness in diabetics. The intercellular fluid comes from leaking microaneurysms or from diffuse capillary incompetence. Clinically, DME is best detected by slit-lamp biomicroscopy with a contact macular lens, although noncontact macular lenses can be used. The edema causes separation of cells, resulting in scattering of light by the multiple interfaces. This decreases the retina’s translucency such that the normal retinal pigment epithelial and choroidal background pattern is blurred (see Fig. 6.22.1A ). Pockets of fluid in the outer plexiform layer, if large enough, can be seen as cystoid macular edema (CME). Usually CME is seen in eyes that have other signs of severe NPDR. In rare cases, CME is due to generalized diffuse leakage from the entire capillary network and can be seen in eyes with very few other signs of diabetic retinopathy.
If the leakage of fluid is severe enough, lipid may accumulate in the retina (see Fig. 6.22.1A ); again, the outer plexiform layer is first to be affected. In some cases, lipid is scattered through the macula. In others, it accumulates in a ring around a group of leaking microaneurysms or around microaneurysms surrounding an area of capillary nonperfusion. This pattern is called circinate retinopathy (see Fig. 6.22.1A ).
The application of optical coherence tomography (OCT) to management of DME has been very useful. The degree of DME and response to therapy can be quantified on OCT. Specifically, the central subfield thickness (CST) can be used to follow a patient’s response to treatment of DME. In addition, the OCT presents qualitative information such as the presence of cysts, hard exudates, and degree of inner or outer retinal or external limiting membrane disruption and subretinal fluid. These findings are useful in following a patient’s response to therapy.
In eyes treated with anti-VEGFs, the presence of disorganization of retinal inner layers (DRIL) has been associated with poorer visual acuity outcomes. The length of DRIL was associated with subsequent vision. The change in DRIL was associated with change in visual acuity (VA), with resolution of DRIL having the best VA. Early change in the extent of DRIL is inversely predictive of subsequent changes in visual acuity.
Advanced Nonproliferative Diabetic Retinopathy
In advanced NPDR, signs of increasing inner retinal hypoxia appear, such as multiple retinal hemorrhages, cotton–wool spots (see Fig. 6.22.3 ), venous beading and vascular loops ( Fig. 6.22.4 ), intraretinal microvascular abnormalities (IRMAs) (see Figs. 6.22.1A and 6.22.4 ), and large areas of capillary nonperfusion seen on fluorescein angiography.
Cotton–wool spots, also called soft exudates or nerve fiber infarcts, result from ischemia, not exudation. Local ischemia causes effective obstruction of axoplasmic flow in the normally transparent nerve fiber layer, and the subsequent swelling of the nerve fibers gives cotton–wool spots their characteristic white fluffy appearance. Fluorescein angiography shows lack of capillary perfusion in the area corresponding to a cotton–wool spot. Microaneurysms frequently surround the hypoxic area (see Fig. 6.22.3 ).
Venous beading (see Fig. 6.22.4 ) is an important sign of sluggish retinal circulation. Venous loops are nearly always adjacent to large areas of capillary nonperfusion. IRMAs are dilated capillaries that seem to function as collateral channels and are frequently difficult to differentiate from surface retinal neovascularization. Fluorescein dye, however, does not leak from IRMAs but leaks profusely from neovascularization. Capillary hypoperfusion often surrounds IRMA (see Fig. 6.22.4 ).
The ETDRS found that IRMAs, multiple retinal hemorrhages, venous beading and loops, widespread capillary nonperfusion, and widespread leakage on fluorescein angiography were all significant risk factors for the development of PDR. Interestingly, cotton–wool spots were not.
Proliferative Diabetic Retinopathy
Although the macular edema, exudates, and capillary occlusions seen in NPDR often cause legal blindness, affected patients usually maintain at least ambulatory vision. PDR, on the other hand, may result in severe vitreous hemorrhage or retinal detachment, with hand-movements vision or worse. Approximately 50% of patients with very severe NPDR progress to PDR within 1 year. Proliferative vessels usually arise from retinal veins and often begin as a collection of multiple fine vessels. When they arise on or within 1 disc diameter of the optic nerve they are referred to as NVD (neovascularization of the disc, Figs. 6.22.5 and 6.22.6 ). When they arise further than 1 disc diameter away, they are called NVE (neovascularization elsewhere) (see Fig. 6.22.6B ). Unlike normal retinal vessels, NVD and NVE leak fluorescein into the vitreous.
Once the stimulus for growth of new vessels is present, the path of subsequent growth taken by neovascularization is along the route of least resistance. For example, the absence of a true internal limiting membrane on the disc could explain the prevalence of new vessels at that location. Also, neovascularization seems to grow more easily on a preformed connective tissue framework. Thus, a shallowly detached posterior vitreous face is a frequent site of growth of new vessels.
The new vessels usually progress through a stage of further proliferation, with associated connective tissue formation. As PDR progresses, the fibrous component becomes more prominent, with the fibrotic tissue being either vascular or avascular. The fibrovascular variety usually is found in association with vessels that extend into the vitreous cavity or with abnormal new vessels on the surface of the retina or disc. The avascular variety usually results from organization or thickening of the posterior hyaloid face. Vitreous traction is transmitted to the retina along these proliferations and may lead to traction retinal detachment.
NVE nearly always grows toward and into zones of retinal ischemia until posterior vitreous detachment occurs (see Fig. 6.22.6 ). Then the vessels are lifted into the vitreous cavity. The end stage is characterized by regression of the vascular tissue. Sometimes there may be contraction of the connective tissue components, development of subhyaloid bands, thickening of the posterior vitreous face, and the appearance of retinoschisis, retinal detachment, or formation of retinal breaks.
Posterior vitreous detachment in diabetics is characterized by a slow, overall shrinkage of the entire formed vitreous rather than by the formation of cavities caused by vitreous destruction. Davis et al. have stressed the role of the contracting vitreous in the production of vitreous hemorrhage, retinal breaks, and retinal detachment. Neovascular vessels do not “grow” forward into the vitreous cavity but are pulled into the vitreous by the contracting vitreous to which they adhere. Confirmation of the importance of the vitreous in the development and progression of PDR comes from the long-term follow-up of eyes that have undergone successful vitrectomy in which neovascularization shrinks, fluorescein leakage decreases, and new areas of neovascularization rarely arise.
It has long been assumed that sudden vitreous contractions tear the fragile new vessels, causing vitreous hemorrhage. However, the majority of diabetic vitreous hemorrhages occur during sleep, possibly because of an increase in blood pressure secondary to early morning hypoglycemia or to rapid eye movement sleep. Because so few hemorrhages occur during exercise, it is not necessary to restrict the activity of patients with PDR. When a hemorrhage occurs, if the erythrocytes are behind the posterior vitreous face, they usually quickly settle to the bottom of the eye and are absorbed. However, when erythrocytes break into the vitreous body, they adhere to the gel, and clearing may take months or years.
A large superficial hemorrhage may separate the internal limiting membrane from the rest of the retina. Such hemorrhages usually are round or oval but also may be boat shaped. The blood may remain confined between the internal limiting membrane and the rest of the retina for weeks or months before breaking into the vitreous. Subinternal limiting membrane hemorrhages were formerly thought to occur between the internal limiting membrane and the cortical vitreous and were called subhyaloid or preretinal hemorrhages. It is now felt that true subhyaloid hemorrhages probably are quite rare. Tight subinternal limiting membrane hemorrhages are dangerous because they may progress rapidly to traction retinal detachment.
As the vitreous contracts, it may pull on the optic disc, causing traction striae involving the macular area or actually drag the macula itself, both of which contribute to decreased visual acuity.
Two types of diabetic retinal detachments occur, those that are caused by traction alone (nonrhegmatogenous) and those caused by retinal break formation (rhegmatogenous). Characteristics of nonrhegmatogenous (traction) detachment in PDR include:
- •
The detached retina usually is confined to the posterior fundus and infrequently extends more than two thirds of the distance to the equator.
- •
The detached retina has a taut and shiny surface.
- •
The detached retina is concave toward the pupil.
- •
No shifting of subretinal fluid occurs.
Occasionally, a spontaneous decrease in the extent of a traction detachment may occur, but this is the exception rather than the rule. Traction on the retina also may cause focal areas of retinoschisis, which may be difficult to distinguish from full-thickness retinal detachment. In retinoschisis the elevated layer is thinner and more translucent.
When a detachment is rhegmatogenous, the borders of the elevated retina usually extend to the ora serrata. The retinal surface is dull and grayish and undulates because of retinal mobility due to shifting of subretinal fluid. Retinal breaks are usually in the posterior pole near areas of fibrovascular change. The breaks are oval in shape and appear to be partly the result of tangential traction from the proliferative tissue, as well as being due to vitreous traction. Determination of the location of retinal holes may be complicated by many factors, particularly poor dilatation of the pupil, lens opacity, increased vitreous turbidity, vitreous hemorrhage, intraretinal hemorrhage, and obscuration of the breaks by overlying proliferative tissue.
Other Ocular Complications of Diabetes Mellitus
Cornea
Corneal sensitivity is decreased in proportion to both the duration of the disease and the severity of the retinopathy. Corneal abrasions are more common in people with diabetes, presumably because adhesion between the basement membrane of the corneal epithelium and the corneal stroma is not as firm as that found in normal corneas. Following vitrectomy, recurrent corneal erosion, striate keratopathy, and corneal edema are more common in diabetics than in nondiabetics.
Glaucoma
The relationship between diabetes and primary open-angle glaucoma is unclear. Some population-based studies have found an association but others have not.
Neovascularization of the iris (NVI) usually is seen only in diabetics who have PDR. PRP not only has protective value against NVI, it also is an effective treatment against established NVI. If the media are clear, PRP should be performed prior to any other treatment for NVI, even in advanced cases. If the media are too cloudy for PRP, transscleral laser or peripheral retinal cryoablation are alternative means of treatment (see later in this chapter). The presence of rubeotic glaucoma is a poor prognostic indicator for visual acuity and for life expectancy. Anti-VEGF therapy is also used as an adjunctive treatment in eyes with NVI.
Lens
The risk of cataract is 2–4 times greater in diabetics than in nondiabetics and may be 15–25 times greater in diabetics under 40 years old.
Patients with diabetes mellitus who have no retinopathy have excellent results from cataract surgery, with 90%–95% having a final visual acuity of 20/40 or better, but chronic CME is about 14 times more common in diabetics than in nondiabetics. The best-known predictor of postoperative success is the preoperative severity of retinopathy. It was hoped that modern surgery, which leaves an intact posterior capsule, would protect the eye from NVI by reducing the diffusion of vasoproliferative factors into the anterior chamber, but several studies have shown that it does not. Furthermore, a neodymium-aluminum-garnet (Nd:YAG) laser capsulotomy does not increase the risk. Other anterior segment complications that are more common in diabetics than in nondiabetics are pupillary block, posterior synechiae, pigmented precipitates on the implant, and severe iritis.
Posterior segment complications of cataract surgery include macular edema, PDR, vitreous hemorrhage, and traction retinal detachment. Unlike prior reports, recent reports suggest that modern uncomplicated cataract surgery may not accelerate progression of diabetic retinopathy in type 2 diabetics with NPDR. Caution should be observed when considering cataract surgery in patients who have diabetic retinopathy, but up to 70% of these patients can attain a final visual acuity of 20/40 or better. In a pilot observational study of eyes with baseline DME at the time of cataract surgery, the DRCR showed that 32% improved four lines and 10% worsened at least two lines by week 16. The study was limited by small enrollment and concluded that it was unlikely that an adequate sample could be recruited within a reasonable time to pursue an interventional trial for eyes with DME in the setting of cataract surgery. For patients without macular edema at the time of cataract surgery, the DRCR found that preoperative noncentral DME or a history of DME treatment may increase the risk of developing central-involved DME 16 weeks after cataract extraction. In addition, visual outcomes were good, with over 85% achieving 20/40 or better visual acuity. However, this was achieved less often in eyes that developed central-involved ME (67%), although these eyes had a lower mean baseline VA (Snellen equivalent 20/63) than eyes that did not develop central-involved ME (Snellen equivalent 20/50).
Cataract surgery in patients with active PDR often results in poorer postoperative visual outcome because of the high risk of both anterior and posterior segment complications. In one series, no patient with active PDR or preproliferative diabetic retinopathy achieved better than 20/80. Most experts recommend aggressive preoperative PRP.
Cornea
Corneal sensitivity is decreased in proportion to both the duration of the disease and the severity of the retinopathy. Corneal abrasions are more common in people with diabetes, presumably because adhesion between the basement membrane of the corneal epithelium and the corneal stroma is not as firm as that found in normal corneas. Following vitrectomy, recurrent corneal erosion, striate keratopathy, and corneal edema are more common in diabetics than in nondiabetics.
Glaucoma
The relationship between diabetes and primary open-angle glaucoma is unclear. Some population-based studies have found an association but others have not.
Neovascularization of the iris (NVI) usually is seen only in diabetics who have PDR. PRP not only has protective value against NVI, it also is an effective treatment against established NVI. If the media are clear, PRP should be performed prior to any other treatment for NVI, even in advanced cases. If the media are too cloudy for PRP, transscleral laser or peripheral retinal cryoablation are alternative means of treatment (see later in this chapter). The presence of rubeotic glaucoma is a poor prognostic indicator for visual acuity and for life expectancy. Anti-VEGF therapy is also used as an adjunctive treatment in eyes with NVI.
Lens
The risk of cataract is 2–4 times greater in diabetics than in nondiabetics and may be 15–25 times greater in diabetics under 40 years old.
Patients with diabetes mellitus who have no retinopathy have excellent results from cataract surgery, with 90%–95% having a final visual acuity of 20/40 or better, but chronic CME is about 14 times more common in diabetics than in nondiabetics. The best-known predictor of postoperative success is the preoperative severity of retinopathy. It was hoped that modern surgery, which leaves an intact posterior capsule, would protect the eye from NVI by reducing the diffusion of vasoproliferative factors into the anterior chamber, but several studies have shown that it does not. Furthermore, a neodymium-aluminum-garnet (Nd:YAG) laser capsulotomy does not increase the risk. Other anterior segment complications that are more common in diabetics than in nondiabetics are pupillary block, posterior synechiae, pigmented precipitates on the implant, and severe iritis.
Posterior segment complications of cataract surgery include macular edema, PDR, vitreous hemorrhage, and traction retinal detachment. Unlike prior reports, recent reports suggest that modern uncomplicated cataract surgery may not accelerate progression of diabetic retinopathy in type 2 diabetics with NPDR. Caution should be observed when considering cataract surgery in patients who have diabetic retinopathy, but up to 70% of these patients can attain a final visual acuity of 20/40 or better. In a pilot observational study of eyes with baseline DME at the time of cataract surgery, the DRCR showed that 32% improved four lines and 10% worsened at least two lines by week 16. The study was limited by small enrollment and concluded that it was unlikely that an adequate sample could be recruited within a reasonable time to pursue an interventional trial for eyes with DME in the setting of cataract surgery. For patients without macular edema at the time of cataract surgery, the DRCR found that preoperative noncentral DME or a history of DME treatment may increase the risk of developing central-involved DME 16 weeks after cataract extraction. In addition, visual outcomes were good, with over 85% achieving 20/40 or better visual acuity. However, this was achieved less often in eyes that developed central-involved ME (67%), although these eyes had a lower mean baseline VA (Snellen equivalent 20/63) than eyes that did not develop central-involved ME (Snellen equivalent 20/50).
Cataract surgery in patients with active PDR often results in poorer postoperative visual outcome because of the high risk of both anterior and posterior segment complications. In one series, no patient with active PDR or preproliferative diabetic retinopathy achieved better than 20/80. Most experts recommend aggressive preoperative PRP.
Optic Neuropathy
As demonstrated by increased latency and decreased amplitude of the visual evoked potential, many diabetic patients without retinopathy have subclinical optic neuropathy. They have an increased risk for anterior ischemic optic neuropathy. In addition, diabetics are susceptible to diabetic papillopathy, which is characterized by acute disc edema without the pale swelling of anterior ischemic optic neuropathy. It is bilateral in one-half of cases and may not show an afferent pupillary defect. Macular edema is a common concurrent finding and is the most common cause of failure of visual recovery in these patients. Visual fields may be normal or show an enlarged blind spot or other nerve fiber defects. The prognosis is excellent, with most patients recovering to 20/50 or better.
Cranial Neuropathy
Extraocular muscle palsies may occur in diabetics secondary to neuropathy involving the third, fourth, or sixth cranial nerves. The mechanism is believed to be a localized demyelinization of the nerve secondary to focal ischemia. Pain may or may not be experienced, and not infrequently extraocular muscle palsy may be the initial clue to a latent diabetic condition. Recovery of extraocular muscle function in diabetic cranial nerve palsies generally takes place within 1–3 months. When the third cranial nerve is involved, pupillary function is usually normal. This pupillary sparing in diabetic third cranial nerve palsy is an important diagnostic feature, helping to distinguish it from an intracranial tumor or aneurysm.
Diagnosis and Ancillary Testing
In nearly all instances, diabetic retinopathy is diagnosed easily via ophthalmoscopic examination. The hallmark lesions are microaneurysms, which usually develop in the posterior pole. Without microaneurysms, the diagnosis of diabetic retinopathy is in doubt. Fasting blood sugar testing, a glucose tolerance test, and HbA1C determinations all can be used to confirm the presence of systemic hyperglycemia.
Intravenous fluorescein angiography is a widely administered ancillary test and is helpful to assess the severity of diabetic retinopathy, to determine sites of leakage in macular edema, to judge the extent of capillary nonperfusion, and to confirm neovascularization. It is a useful preoperative test to evaluate the extent of retinopathy in patients who are to undergo cataract surgery and have media opacity. OCT angiography is being increasingly used as a noninvasive test in diabetic retinopathy to visualize capillary nonperfusion and neovascularization. OCT is widely used to assess and follow macular edema.
Differential Diagnosis
The differential diagnosis is listed in Box 6.22.1 .
- •
Radiation retinopathy
- •
Hypertensive retinopathy
- •
Retinal venous obstruction (central retinal vein occlusion [CRVO], branch retinal vein occlusion [BRVO])
- •
The ocular ischemic syndrome
- •
Anemia
- •
Leukemia
- •
Coats’ disease
- •
Idiopathic juxtafoveal retinal telangiectasia
- •
Sickle cell retinopathy
Pathology
The earliest histopathological abnormalities in diabetic retinopathy are thickening of the capillary basement membrane and pericyte dropout. Microaneurysms begin as a dilatation in the capillary wall in areas where pericytes are absent; microaneurysms initially are thin walled. Later, endothelial cells proliferate and deposit layers of basement membrane material around themselves. Fibrin may accumulate within the aneurysm, and the lumen of the microaneurysm may become occluded ( Fig. 6.22.7 ). In early cases, microaneurysms are present mostly on the venous side of the capillaries, but later they are also seen on the arterial side. Despite the multiple layers of basement membrane, they are permeable to water and large molecules, resulting in water and lipid accumulation in the retina. Because fluorescein passes easily through them, many more microaneurysms are seen on fluorescein angiography than on ophthalmoscopy (see Figs. 6.22.1 and 6.22.3 ).
Treatment
Medical Therapy
Antiplatelet Therapy
The ETDRS reported that aspirin 650 mg daily does not influence the progression of retinopathy, affect visual acuity, or influence the incidence of vitreous hemorrhages. However, there was a significant decrease in cardiovascular morbidity in the aspirin-treated group compared with the placebo cohort. Clopidogrel and ticlopidine, like aspirin, inhibit adenosine diphosphate–induced platelet aggregation. They have been shown to decrease the risk of stroke in patients with transient ischemic attacks, but there is no clear evidence showing an impact on diabetic retinopathy.
Antihypertensive Agents
The Hypertension in Diabetes Study, part of the United Kingdom Prospective Diabetes Study (UKDPS), evaluated the effect of blood pressure control on the progression of diabetic retinopathy. Patients were treated with angiotensin-converting enzyme inhibitors (ACEIs) or beta-blockers to achieve “tight” control of blood pressure (<150/85 mm Hg) or “less tight” control (<180/105 mm Hg). The group with better blood pressure control had a 37% risk reduction in microvascular changes. There was no difference in effect between the two agents used. Lisinopril, an ACEI, has also been shown to decrease the progression of NPDR and PDR in normotensive diabetics. The patients in this study with the better glycemic control benefited more from lisinopril.
Antiangiogenesis Agents
The discovery that VEGF plays a critical role in the initiation of diabetic neovascularization and an important role in DME has revolutionized management of these complications of diabetes. Several pharmacological inhibitors of angiogenesis have been shown to be beneficial in the therapy of center-involving DME. These include:
- •
Pegaptanib sodium, a VEGF aptamer (Macugen, Eyetech Pharmaceuticals, NY).
- •
Ranibizumab (Lucentis, Genentech).
- •
Bevacizumab (Avastin, Genentech).
- •
Aflibercept (Eylea, Regeneron).
All four anti-VEGF agents improve visual acuity and promote normalization of the macular architecture in most eyes with DME. Because of their superior beneficial effect compared to laser and corticosteroids, with a relatively good safety profile, anti-VEGF therapy is now considered primary treatment of fovea-involving DME in most eyes.
The major problems with anti-VEGF therapy are cost and frequency of administration. In addition, the risk of endophthalmitis with anti-VEGF therapy in diabetics appears to be greater than with other ophthalmic conditions (e.g., neovascular age-related macular degeneration, retinal venous occlusion), with several studies suggesting the long-term incidence may approach 1%. In current clinical practice, pegaptanib is not as widely used, as the other agents appear to be more efficacious. The DRCR Protocol T compared the visual and anatomical outcomes of bevacizumab, ranibizumab, and aflibercept for treatment of diabetic macular edema in patients with 20/32–20/320 visual acuity. All three drugs did result in some improvement in visual acuity and reduction of edema by OCT. There were 9–10 treatments given in the first year for all three drugs. Overall, mean change in visual acuity was best for aflibercept (13 letters) which was greater than ranibizumab (11 letters, p = 0.034) or bevacizumab (10 letters, p < 0.001). For visual acuities better than 20/50, there was no significant difference in the visual or anatomic outcomes. However, for visual acuities of 20/50 or worse, aflibercept resulted in a mean gain of 19 letters versus 14 letters for ranibizumab ( p = 0.0031) or 12 letters for bevacizumab ( p < 0.001). More eyes gained 15 or more letters with aflibercept (67%) than ranibizumab (50%, p < 0.001) or bevacizumab (41%, p = 0.0078). There were no significant differences between the drugs in terms of percentage of eyes that lost 10 or more letters or 15 or more letters from baseline. Both ranibizumab and aflibercept resulted in significantly greater reductions in mean CST than bevacizumab. No significant differences in safety were found among the three drugs.
The best dosing regimen for the anti-VEGF remains controversial. It is generally agreed that monthly injections are indicated until the fluid has resolved or until no further improvement occurs, but once that point has been reached, it remains unclear whether further “mandated” monthly injections or long-term monthly injections versus a switch to an as-needed protocol is best. For those clinicians choosing to combine anti-VEGF therapy with focal laser, the optimal timing of the laser and anti-VEGF therapy is not as of yet determined. However, most studies support delayed laser photocoagulation, usually at about 6 months after initiating anti-VEGF therapy. With respect to the role of anti-VEGF therapy combined with immediate or deferred focal laser, the DRCR Protocol I showed that ranibizumab with deferred (≥24 weeks) laser is more efficacious than focal laser alone, intravitreal ranibizumab plus prompt (within 3–10 days of injection) focal laser, triamcinolone acetonide plus prompt laser, or sham plus prompt focal laser treatment. Eyes treated with ranibizumab plus either prompt or deferred focal laser had better visual acuity outcomes than eyes treated with triamcinolone plus prompt laser or sham plus prompt laser. Interestingly, reductions in OCT central subfield thickness were similar between the ranibizumab and triamcinolone groups, suggesting the visual benefits of intravitreal corticosteroids may be tempered by their side effects: cataract and elevated intraocular pressure (IOP). Indeed, for pseudophakic eyes, ranibizumab and triamcinolone groups had similar outcomes. The rate of endophthalmitis was 0.8% in the ranibizumab groups versus none in the corticosteroid group or laser alone eyes.
Bevacizumab
Bevacizumab is in wide usage for treatment of DME. While not studied as rigorously as the other anti-VEGFs, the cost differential favors it so greatly, that for many clinicians it is the first line therapy. The best data on this drug comes from the DRCR Study Protocol T (see earlier). In addition, the Bevacizumab or Laser Therapy (BOLT) Study showed that bevacizumab had superior visual outcomes (20/50 Snellen equivalent) in the 80 enrolled patients compared to laser 54.8 (20/80) at 2 years. The mean change in visual acuity was a gain of 8.6 letters for bevacizumab versus a mean loss of 0.5 letters for laser arm eyes. Forty-nine percent of patients gained 10 or more letters ( p = 0.001), and 32% gained at least 15 letters ( p = 0.004) for bevacizumab versus 7% and 4% for laser eyes. The median number of treatments over 24 months was 13 for bevacizumab and four for laser. A large retrospective study of bevacizumab for DME was performed by the Pan-American Collaborative Retina Study Group (PACORES). This group showed that stability or improvement of visual acuity occurred with 1.25 mg or 2.5 mg bevacizumab. No difference was seen between the two doses. The mean vision gained was 2.4 lines at 24 months for both groups.
Pegaptanib
Pegaptanib sodium (Macugen), a selective VEGF-165 aptamer, was the first anti-VEGF agent used in DME. Although pegaptanib-treated eyes had slightly better visual acuity outcomes and reduction of retinal edema by OCT as compared with sham-treated eyes, the results were not clinically convincing enough to change the standard of care. In the phase 2 macular edema study, a small subset of eyes also had PDR. The PDR regressed during the period of active therapy but recurred at cessation of anti-VEGF treatment. This drug’s comparative efficacy has limited its use.
Aflibercept
In the phase 3 VIVID and VISTA studies, 872 eyes with center-involved macular edema were randomized to receive intravitreal injection of aflibercept 2 mg every 4 weeks, 2 mg every 8 weeks after five monthly doses or macular laser. The mean change in best-corrected visual acuity (BCVA) at week 52 from baseline was the primary endpoint. The primary endpoint was superior for aflibercept compared to laser controls at week 52 and sustained through week 100. The results were similar for both dosing groups. The mean change in vision was a gain of 12.5 for 2 mg q4 and 10.7 for 2 mg q8 versus 0.2 letters for laser ( p < 0.0001) in VISTA, and 10.5 for 2 mg q4 and 10.7 for 2 mg q8 versus 1.2 letters for laser ( p < 0.0001) in VIVID. Eyes receiving aflibercept had more significant reductions in OCT thickness as compared with controls. Significantly more eyes also gained 15 or more letters from baseline in the aflibercept compared with the laser groups. An analysis of the mean visit-to-visit change in BCVA and central retinal thickness during the upload phase from the 2 mg q4 and the 2 mg q8 datasets showed continual functional and anatomical improvements after the fourth and fifth injections. This study suggests that intensive uploading is beneficial.
Additional Medical Therapies
Inhibition of protein kinase C, a compound critical in the cascade that activates VEGF expression, was not shown to be of benefit in the treatment or prevention of DME. An oral inhibitor of protein kinase C has been shown to suppress retinal neovascularization in animal models. Recently, a protein kinase C inhibitor has been shown to reduce diabetes-induced hemodynamic abnormalities in patients with diabetic retinopathy and reduce the risk of vision loss in patients with macular edema. It delayed the progression of edema located more than 100 µm from the foveal center to within 100 µm of the foveal center (68% vs. 50%, p = 0.003). Initial laser treatment for macular edema was 26% less frequent in eyes of ruboxistaurin-treated patients ( p = 0.008). However there was no effect of ruboxistaurin on prevention of progression of diabetic retinopathy.
In addition to anti-VEGF therapies, there are non-VEGF pathways that may be useful targets in the therapy of DME. One of these is the Tie 2 receptor pathway. Inhibition of angiopoietin 2 (Ang2) is being investigated in combination with anti-VEGF therapies. The Boulevard Study is a phase 2 study that is comparing the efficacy of a bispecific antibody (Roche RG7716) to both VEGF and Ang2 with ranibizumab alone. The Ruby Study was a phase 2 study comparing the efficacy of a co-formulation of two drugs (aflibercept and nesvacumab) with aflibercept alone. The Boulevard Study has shown that the bispecific antibody achieved its primary endpoint of efficacy. Mean VA gain at 6 months was significantly improved for the bispecific antibody as compared with ranibizumab alone (3.6 letters difference, p = 0.03); mean gain was 13.9 letters from baseline (written communication from Genentech.) The Ruby Study has shown no difference in the primary endpoint between the coformulated drug and aflibercept (written communication from Regeneron).
Pharmacotherapy for Proliferative Diabetic Retinopathy
The use of anti-VEGF therapy for PDR initially was shown in the pegaptanib sodium for DME studies. In these studies, a few eyes with PDR were inadvertently included, and regression of the PDR occurred. This led to specific studies evaluating the use of anti-VEGF for PDR. In a phase 1, prospective, randomized, controlled, open-label study, 20 active PDR patients were randomly assigned to receive pegaptanib sodium (0.3 mg) every 6 weeks for 30 weeks or PRP laser. In the pegaptanib group, early regression was seen by week 3 (90%) with complete regression by week 12 that was maintained through week 36. In contrast, in the PRP group, 25% showed complete regression, 25% partial, and 50% showed persistent active PDR. Mean change in vision was +5.8 letters in pegaptanib-treated eyes and −6.0 letters in PRP-treated eyes.
The DRCR Protocol S has compared anti-VEGF and PRP for eyes with PDR in a noninferiority study in 394 eyes. Treatment with 0.5 mg ranibizumab (initially every 4 weeks for six injections unless no neovascularization at 4- or 5-month visit) was noninferior to PRP for visual acuity outcomes. Primary outcome was mean change in vision at 2 years compared to baseline. At 2 years, mean visual acuity improved 2.8 letters in the ranibizumab group versus 0.2 letters in the PRP group ( p < 0.001 for noninferiority). Both treatments were effective for controlling the PDR and in preventing NVI and visual loss. However, when improvements in visual acuity were evaluated, anti-VEGF was found to result in superior mean visual acuity over the course of 2 years when an area-under-the-curve analysis was performed. Greater numbers of eyes gained vision. Anti-VEGF also had superior visual field outcomes compared to laser. There was also a decreased need for vitrectomies and lower incidence of center-involved macular edema in the anti-VEGF eyes. PRP was rarely given for failure of anti-VEGF to control PDR. There was a lower amount of reduction in visual field.
An analysis of rate of progression of PDR (defined as first occurrence of vitreous hemorrhage, retinal detachment, anterior segment neovascularization, or neovascular glaucoma) showed that eyes treated with ranibizumab had lower rates (34% vs. 42%) of progression than eyes treated with PRP. Incidentally, the risk of progression was higher for eyes receiving pattern scan laser than for eyes receiving conventional PRP (60% vs. 39%).
Despite these results, one needs to weigh other factors that may affect a patient’s response to anti-VEGF therapy for PDR. If a patient is not likely to follow up and thus not likely to receive the needed anti-VEGF injections, one should probably not use anti-VEGF and instead proceed with PRP. In patients with PDR and DME, anti-VEGF is a reasonable choice.
More recently, the RIDE and RISE studies have shown eyes treated with monthly ranibizumab were more likely to show improvement and less likely than sham eyes to show progression on the ETDRS retinopathy severity as graded on fundus photographs. These eyes were less likely to develop PDR. Interestingly, patients with PDR had regression to NPDR levels. The DRCR showed that there was no difference in the rate of required pars plana vitrectomy 16 weeks later in eyes with vitreous hemorrhage given intravitreal ranibizumab versus sterile saline.
Corticosteroids
Even before clinical trials that investigated the long-term benefit of corticosteroids (triamcinolone acetonide) for DME, this treatment was in wide usage. In the short term, good visual results and improved OCT findings are seen in most eyes. However, with repeated injections over time, complications can occur frequently that limit the initial benefit. The DRCR network studies suggest that over 2 years of treatment, monotherapy with triamcinolone acetonide is not superior to laser photocoagulation for DME. At 2 years, mean visual acuity was better in eyes treated exclusively with laser compared to eyes treated exclusively with either 1 mg triamcinolone ( p = 0.02) or 4 mg triamcinolone ( p = 0.002) groups. As expected, intraocular pressure rise and cataract onset were greater in the triamcinolone group.
Since monotherapy with one agent is not how most patients are treated in the real world, subsequent studies looked at corticosteroids plus laser. As mentioned earlier, except perhaps in pseudophakic eyes, this combination is not recommended for primary treatment of DME if anti-VEGF agents are available.
There have been corticosteroids other than triamcinolone acetonide, evaluated for treatment of DME. A fluocinolone acetonide intravitreal insert was studied in the Famous Study. The intravitreal inserts provide excellent sustained intraocular release of fluocinolone acetonide for one year or more. In a large trial, the FAME study, 29% of eyes receiving the 0.2 µg/day fluocinolone implant gained three or more lines by year 2, versus 16% in the sham group ( p = 0.002). Complications included elevation of IOP related events in 37% treated eyes versus 12% controls. Glaucoma surgery was required in 4.8% treated versus 0.5% controls. Cataract as an adverse event occurred in 82% of treated eyes, and 80% of treated eyes underwent cataract extraction over a period of 36 months versus corresponding data of 50% and 27% in the control group. Eyes with chronic DME had higher proportions of gain of 15 or more letters from baseline compared with controls. The study concluded that the fluocinolone implant is effective in eyes with chronic macular edema. The U.S. Food and Drug Administration (FDA) approval notes that this therapy is not to be used unless the eye has been previously treated with corticosteroids and has not had an elevation in IOP.
The dexamethasone intravitreal implant containing 700 µg dexamethasone (Ozurdex, Allergan, Irvine, CA) in a solid polymer drug delivery system that is also approved by the FDA for treatment of diabetic macular edema, in addition to macular edema from vein occlusions and uveitis. The MEAD study showed that 0.7 mg dexamethasone resulted in 22% treated eyes versus 12% sham eyes gaining 15 or more letters. Also, dexamethasone-treated eyes had more significant reductions in CST on OCT (−112 µg versus −42 µg, p < 0.001).
The BEVORDEX Study compared dexamethasone treated eyes and bevacizumab in a phase 2 study. No significant differences were seen between the drugs for visual acuity or CST change by OCT at 24 months. However, 74% of dexamethasone-treated eyes versus 48% of bevacizumab-treated eyes had a rise in IOP of 5 mm Hg or more.
In vitrectomized eyes, the Champlain study showed that 55 vitrectomized eyes with treatment-resistant DME had statistically and clinically significant improvements in both visual acuity and vascular leakage with treatment. In fact, 30% gained 10 more letters.
Surgical Therapy
Panretinal Photocoagulation
The Diabetic Retinopathy Study proved that both xenon arc and argon laser PRP significantly decrease the likelihood of progression of eyes with high-risk characteristics (HRC) to severe visual loss. Eyes with HRC are defined as those with NVD greater than one fourth to one third of a disc area, those with any NVD and vitreous hemorrhage, or those with NVE greater than one-half the disc area and vitreous or preretinal hemorrhage.
The exact mechanism by which PRP works remains unknown. One hypothesis is that PRP decreases the production of vasoproliferative factors by eliminating areas of hypoxic retina. An alternative hypothesis suggests that by thinning the retina, PRP increases oxygenation of the remaining retina by allowing increased diffusion of oxygen from the choroid. Yet another hypothesis is that PRP leads to an increase in vasoinhibitors by directly stimulating the retinal pigment epithelium to produce inhibitors of vasoproliferation.
The goal of PRP is to arrest or to cause regression of the neovascularization. The recommended therapy is 1200–2000 burns that are 500 µm in diameter delivered through the Goldmann lens, or the equivalent number when using 200 µm burns delivered through the Rodenstock panfundoscope lens or Volk SuperQuad lens. The burns should be intense enough to lightly whiten the overlying retina using a duration of 0.1 second (see Fig. 6.22.5 ). There are newer lasers that deliver shorter bursts of laser energy that result in control of the neovascularization.
Some retinal specialists feel that there is no upper limit to the total number of burns and that treatment should be continued until regression occurs. The only prospective, controlled study found that eyes that received supplementary PRP treatment had no improved outcome over those that received standard PRP only. About two thirds of eyes with HRC that receive PRP have regression of their HRC by 3 months after treatment.
The ETDRS found that PRP significantly retards the development of HRC in eyes with very severe NPDR and macular edema. After 7 years of follow-up, 25% of eyes that received PRP developed HRC as compared with 75% of eyes in which PRP was deferred until HRC developed. Nevertheless, the ETDRS concluded that treatment of severe NPDR and PDR short of HRC was not generally indicated for three reasons.
First, after 7 years of follow-up, 25% of the eyes assigned to deferral of PRP had not developed HRC. Second, when patients are closely monitored and PRP is given as soon as HRC develops, severe visual loss can be prevented. After 7 years of follow-up, 4.0% of eyes that did not receive PRP until HRC developed had a visual acuity of 5/200 or less, compared with 2.5% of eyes assigned to immediate PRP. The difference was neither clinically nor statistically significant. Third, PRP has significant complications. It often causes decreased visual acuity by increasing macular edema or by causing macular pucker. Fortunately, the edema frequently regresses spontaneously over 6 months, but the visual field usually is moderately, but permanently, decreased. Color vision and dark adaptation, which often are already impaired, also are worsened by PRP. However, if both eyes have severe NPDR, the ETDRS reported that PRP was not unreasonable, especially in patients who are unlikely to follow up closely.
Recently the pattern lasers have enabled semi-automated laser photocoagulation. These lasers allow various patterns of laser spots to be given within a fraction of the time usually required when using traditional lasers. The duration of the laser application (usually 10–30 milliseconds) is also shorter than traditional lasers (100–300 milliseconds). Studies have shown that more laser spots are needed with increasing severity of PDR. In a prospective study in which one eye was assigned to pattern laser single session and the other eye to multisession conventional PRP laser, control of PDR was achieved with less associated pain or complications. However, applying the same number of spots used with traditional lasers does not achieve sustained regression of the lesions, probably because more laser spots are required to treat the same area of retina achieved with traditional lasers.
Peripheral Retinal Cryotherapy
Peripheral retinal cryotherapy is used to treat HRC in eyes with media too hazy for PRP. Reported benefits include resorption of vitreous hemorrhages and regression of NVD, NVE, and NVI. The main complication is the development or acceleration of traction retinal detachment in 25%–38% of eyes. Therefore, this treatment should be avoided in patients with known traction retinal detachment, and all patients must be monitored carefully. In the current era of anti-VEGF therapy this option is mostly historical. This option is probably best reserved for those eyes with no visual potential and recalcitrant PDR that has been unresponsive to PRP or in which PRP cannot be applied. The DRCR Protocol AB will investigate the utility of anti-VEGF in eyes with vitreous hemorrhage compared to pars plana vitrectomy.
Focal Laser for Macular Edema
Patz was the first to show that argon laser photocoagulation decreases or stabilizes macular edema. Later, the ETDRS confirmed his results. The ETDRS defined clinically significant macular edema as:
- •
Retinal thickening involving the center of the macula, or
- •
Hard exudates within 500 µm of the center of the macula (if associated with retinal thickening), and
- •
An area of macular edema greater than one disc area but within 1 disc diameter of the center of the macula.
In the era of spectral-domain optical coherence tomography (SD-OCT) imaging, intraretinal cysts are seen on SD-OCT imaging in eyes with relatively good visual acuity and no clinical thickening. There is no clinical trial to date that has addressed management of these “subclinical DME” eyes.
The ETDRS focal laser treatment strategy was to photocoagulate all leaking microaneurysms further than 500 µm from the center of the macula and to place a grid of 100–200 µm burns in areas of diffuse capillary leakage and in areas of capillary nonperfusion ( Fig. 6.22.8 ). After 3 years of follow-up, 15% of eyes with clinically significant macular edema had doubling of the visual angle as opposed to 32% of untreated control eyes. The ETDRS also showed that PRP should not be given to eyes with clinically significant macular edema unless HRC are present. Patients with macular edema who have the best prognosis for improved vision have circinate retinopathy of recent duration or focal, well-defined leaking areas and good capillary perfusion surrounding the avascular zone of the retina. Patients with an especially poor prognosis have dense lipid exudate in the center of the foveola ( Fig. 6.22.9 ). Other poor prognostic signs include diffuse edema with multiple leaking areas, extensive central capillary nonperfusion, increased blood pressure, and CME. Nevertheless, the ETDRS found that even eyes with these adverse findings still benefited from treatment when compared with control eyes. Side effects of the focal laser include some loss of central vision, central scotomas, and decreased color vision. In addition, the retinal pigment epithelium (RPE) and retinal atrophy associated with the laser scars can enlarge over time and may encroach on fixation. Following publication of the ETDRS, clinicians now use laser burns that are lighter and less intense to limit progressive laser scar expansion and the attendant visual side effects. Because of these side effects, investigators turned to pharmacological agents to treat DME. One of the first agents used was intravitreal triamcinolone acetonide. However, as discussed earlier, the effect is transient and there is a high risk of cataract progression and secondary glaucoma. Eyes with macular ischemia and better baseline vision may have less vision improvement with intraocular corticosteroids. Anti-VEGF agents are more commonly employed due to their higher rates of efficacy and lower rates of complications.
