Management of Diabetic Retinopathy: Title and subTitle BreakA Systematic Review FREE

Diabetes mellitus affects 200 million people worldwide,¹ including 20 million in the United States alone.² Diabetic retinopathy (DR), a specific microvascular complication of diabetes, is the leading cause of blindness in working-aged persons in the United States.² The prevalence of DR increases with duration of diabetes,³ and nearly all persons with type 1 diabetes and more than 60% of those with type 2 have some retinopathy after 20 years. The major risk factors for DR have been reported from epidemiologic studies^{3 - 4} and are summarized in the Article .

Consistent Risk Factors

Less Consistent Risk Factors

Diabetic retinopathy can be classified into 2 stages: nonproliferative and proliferative. The earliest visible signs in nonproliferative DR are microaneurysms and retinal hemorrhages (Figure, A). Progressive capillary nonperfusion is accompanied by development of cotton-wool spots, venous beading, and intraretinal microvascular abnormalities. Proliferative DR occurs with further retinal ischemia and is characterized by the growth of new blood vessels on the surface of the retina or the optic disc (Figure, B). These abnormal vessels may bleed, resulting in vitreous hemorrhage, subsequent fibrosis, and tractional retinal detachment. Diabetic macular edema (DME), which can occur at any stage of DR, is characterized by increased vascular permeability and the deposition of hard exudates at the central retina (Figure, A). Diabetic macular edema is now the principal cause of vision loss in persons with diabetes.

Primary interventions, such as intensive glycemic and blood pressure control, can reduce the incidence of DR, while secondary interventions, such as laser photocoagulation, may prevent further progression of DR and vision loss. There are many new interventions, but the evidence to support their use is uncertain. This article provides a systematic review of the literature to determine the best evidence for primary and secondary interventions for DR.

We conducted a literature search to identify English-language randomized controlled trials (RCTs) or meta-analyses evaluating interventions for DR. Articles were retrieved using MEDLINE (1966 through May 2007), EMBASE, Cochrane Collaborations, the Association for Research in Vision and Ophthalmology database, and the National Institutes of Health Clinical Trials Database through May 2007. Search terms included variations of keywords for retinopathy, diabetes, DR, DME, retinal neovascularization, controlled clinical trial, and randomized controlled trial (RCT). This was supplemented by hand searching the reference lists of major review articles. As we were primarily interested in longer-term outcomes, we excluded studies with less than 12 months of follow-up and those failing to separate data of different retinal conditions (eg, macular edema from diabetes vs retinal vein occlusion). We also excluded secondary complications of proliferative DR such as rubeotic glaucoma and tractional detachments, as they were beyond the scope of this review.

We used the Delphi consensus criteria list to select well-conducted studies.²¹ Studies were evaluated on a standardized data extraction form for (1) valid method of randomization, (2) concealed allocation of treatment, (3) similarity of groups at baseline regarding the most important prognostic indicators, (4) clearly specified eligibility criteria, (5) masking of outcome assessor, (6) masking of care provider, (7) masking of patient, (8) reporting of point estimates and measures of variability for outcomes, (9) intention-to-treat analysis, and (10) acceptable loss to follow-up rate unlikely to cause bias. Studies were scored out of a maximum of 10, and studies with a score greater than 5 were considered higher-quality studies. For each intervention, we graded the overall strength of evidence as levels I, II, or III and the ratings for clinical recommendations as levels A, B, and C, using previously reported criteria.²²

For primary interventions, outcome measures included incidence of new DR and rate of adverse effects of intervention. For secondary interventions, measures included progression of DR, changes in visual acuity and macular thickness, and rates of legal blindness and adverse effects. Emphasis was given to studies in which best-corrected visual acuity was measured in a masked fashion using the Early Treatment Diabetic Retinopathy Study (ETDRS) protocol. For some RCTs, both primary (incidence of DR) and secondary (progression of DR) interventions were evaluated.

Studies used different methods to ascertain retinopathy, including clinical ophthalmoscopy, retinal photography, and/or fluorescein angiography.²³ Studies also classified DR differently, with most using the Airlie House classification²⁴ with some modifications.²⁵ This gold-standard assessment involves the grading of seven 30° stereoscopic images of the retina (7 standard fields), with each image compared with standard photographs. A score is then assigned to each eye, ranging from 10 (no retinopathy) to 85 (advanced proliferative DR), and the grades for both eyes are combined into a stepped scale. DME was usually classified as absent or present. Definitions for progression of DR also varied. The Diabetes Control and Complications Trial (DCCT)^{26 - 27} defined progression as at least 3 steps worsening from baseline, while the United Kingdom Prospective Diabetes Study (UKPDS)²⁸ defined progression as a 2-step change from baseline. Other studies used increases in number of microaneurysms or the need for laser photocoagulation as indicators of progression.

A total of 782 citations were accessed, of which 44 studies (including 3 meta-analyses) of interventions for DR met our inclusion criteria

Glycemic Control. Early epidemiologic studies have shown showed a consistent relationship between glycated hemoglobin (HbA_1c) levels and the incidence of DR.^{5 ,7} This important observation has been confirmed in large RCTs demonstrating that tight glycemic control reduces both the incidence and progression of DR (Table 1). The DCCT,^{26 - 27 ,29 ,45} conducted between 1983 and 1993, randomized 1441 patients with type 1 diabetes to receive intensive glycemic or conventional therapy. Over 6.5 years of follow-up, intensive treatment (median HbA_1c, 7.2%) reduced the incidence of DR by 76% (95% confidence interval [CI], 62%-85%) and progression of DR by 54% (95% CI, 39%-66%), as compared with conventional treatment (median HbA_1c, 9.1%).^{26 - 27 ,29 ,45}

The UKPDS³¹ reported similar findings in type 2 diabetes. The UKPDS randomized 3867 persons newly diagnosed as having type 2 diabetes to receive intensive or conventional therapy. Intensive therapy reduced microvascular end points by 25% (95% CI, 7%-40%) and the need for laser photocoagulation by 29%. Data from a subgroup of participants' retinal photographic grading showed a similar association.¹⁷ These findings have been replicated in other studies,^{33 ,46} including a meta-analysis prior to the DCCT³⁴ (Table 1).

Long-term observational DCCT data showed that despite gradual equalization of HbA_1c values after study termination, the rate of DR progression in the former intensively treated group remained significantly lower than in the former conventional group,^{27 ,30} emphasizing the importance of instituting tight glycemic control early in the course of diabetes. This concept is supported by the results of another RCT,⁴⁷ in which participants initially assigned to intensive glucose control vs conventional treatment had lower 10-year incidence of severe retinopathy.⁴⁸

Tight glycemic control has two clinically important adverse effects. First, there is risk of early worsening of DR. In the DCCT, this occurred in 13.1% of the intensive vs 7.6% of the conventional treatment group.⁴⁹ However, this effect was reversed by 18 months, and no case of early worsening resulted in serious visual loss. Similar adverse event rates were reported in a meta-analysis.³⁵ Participants at risk of this early worsening had higher HbA_1c levels at baseline and a more rapid reduction of HbA_1c levels in the first 6 months, suggesting that physicians should avoid rapid reductions of HbA_1c levels where possible. Second, tight glycemic control is a known risk factor for hypoglycemic episodes and diabetic ketoacidosis.³⁴ A meta-analysis of 14 RCTs, including the DCCT,⁵⁰ indicated that intensive treatment is associated with a 3-fold risk of hypoglycemia and 70% higher risk of ketoacidosis as compared with conventional treatment. The risk of ketoacidosis was 7-fold higher among patients exclusively using insulin pumps,⁵⁰ suggesting that multiple daily insulin injection might be a safer strategy.

Blood Pressure Control. Epidemiologic studies have not found blood pressure to be a consistent risk factor for DR incidence and progression.^{8 - 9 ,51 - 52} Evidence from RCTs, however, indicates that tight control of blood pressure is a major modifiable factor for the incidence and progression of DR (Table 2). The UKPDS²⁸ randomized 1048 patients with hypertension to receive tight blood pressure control (target systolic/diastolic pressure, <150/<85 mm Hg) or conventional control (target, <180/<105 mm Hg). After 9 years of follow-up, patients having tight control had a 34% reduction (99% CI, 11%-50%) in DR progression, 47% reduction (99% CI, 7%-70%) in visual acuity deterioration, and 35% reduction in laser photocoagulation compared with those having conventional control.

The UKPDS findings contrast with that of the Appropriate Blood Pressure Control in Diabetes (ABCD) trial,^{54 ,57} which randomized 470 people with type 2 diabetes and hypertension to receive intensive or moderate blood pressure control. Over 5 years, there was no difference in DR progression between the groups. The lack of efficacy in this study may be related to poorer glycemic control, shorter follow-up, and lower blood pressure levels at baseline as compared with the UKPDS. It is unclear if there is a threshold effect beyond which further blood pressure lowering no longer influences DR progression.

The effects of therapy with antihypertensive agents are also apparent among normotensive persons with diabetes. In another group of the ABCD trial,⁵⁷ among 480 nonhypertensive patients with type 2 diabetes, intensive blood pressure control significantly reduced DR progression over 5 years as compared with moderate control. The EURODIAB Controlled Trial of Lisinopril in Insulin-Dependent Diabetes Mellitus (EUCLID)⁵⁶ evaluated the effects of the angiotensin-converting enzyme (ACE) inhibitor lisinopril on DR progression in normotensive, normoalbuminuric patients with type 1 diabetes. Over 2 years, lisinopril reduced the progression of DR by 50% (95% CI, 28%-89%) and progression to proliferative DR by 80%.⁵⁶ EUCLID was limited by differences in baseline glycemic levels between groups (the treatment group had lower HbA_1c levels) and a short follow-up of 2 years. This study, along with another smaller RCT,⁵⁸ suggested that ACE inhibitors may have an additional benefit on DR progression independent of blood pressure lowering. However, data from the UKPDS⁵³ and the ABCD study^{54 ,57} did not find ACE inhibitors to be superior to other blood pressure medications.

Whether newer blood pressure medications have additional beneficial effects is unclear. A recent small RCT (n = 24) with short follow-up (4 months) reported a worsening of DME among patients treated with the angiotensin II receptor blocker losartan, compared with controls.⁵⁹ Two large RCTs are currently ongoing. The Action in Diabetes and Vascular Disease (ADVANCE) study will evaluate the effect of a perindopril-indapamide combination on the incidence of DR,⁶⁰ while the Diabetic Retinopathy Candesartan Trial (DIRECT) will evaluate the angiotensin II receptor blocker candesartan.⁶¹

Lipid-Lowering Therapy. Observational studies suggest that dyslipidemia increases the risk of DR, particularly DME.^{8 ,11} A small RCT conducted among 50 patients with DR found a nonsignificant trend in visual acuity improvement in patients receiving simvastatin treatment,⁶² while another study reported a reduction in hard exudates but no improvement in visual acuity in those with clinically significant DME treated with clofibrate.⁶³

In the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study (Table 3),⁶⁴ among 9795 participants with type 2 diabetes, those treated with fenofibrate were less likely than controls to need laser treatment (5.2% vs 3.6%, P < .001). However, the severity of DR, indications for laser treatment, and type of laser treatment (focal or pan-retinal) were not reported.

The Collaborative Atorvastatin Diabetes Study (CARDS), an RCT of 2830 patients with type 2 diabetes, did not find atorvastatin to be effective in reducing DR progression.^{75 - 76} The study was limited by substantial missing data (only 65% of patients had retinopathy status recorded at baseline) and lack of photographic grading for DR. Several ongoing RCTs, such as the Atorvastatin Study for Prevention of Coronary Endpoints in NIDDM (ASPEN),⁷⁷ will also evaluate the effects of atorvastatin on DR.

Medical Interventions.Antiplatelet Agents. The ETDRS showed that aspirin (650 mg/d) had no beneficial effect on DR progression or loss of visual acuity in patients with DME or severe nonproliferative DR during 9 years of follow-up (Table 3).^{65 - 66} Aspirin treatment was not associated with an increased rate of vitrectomy.^{65 - 66} A smaller RCT evaluating aspirin alone and in combination with dipyridamole reported a reduction in microaneurysms on fluorescein angiograms in both groups as compared with placebo.⁶⁷ A similar trend was observed in a small RCT⁶⁸ evaluating ticlopidine, although results were not statistically significant.

Protein Kinase C Inhibitors. Hyperglycemia induces synthesis of diacylglycerol in vascular cells, leading to activation of protein kinase C (PKC) isozymes. Excessive PKC activation may be involved in the pathophysiology of DR. Ruboxistaurin, an orally active PKC inhibitor, was evaluated in the Protein Kinase C Diabetic Retinopathy Study (PKC-DRS) (Table 3),⁶⁹ which randomized 252 patients with moderate to severe nonproliferative DR to receive ruboxistaurin (8, 16, or 32 mg) or placebo. No significant difference in DR progression was observed after 36 months of follow-up, although patients treated with 32 mg of ruboxistaurin had a significant reduction in the risk of moderate visual loss. Treatment was well tolerated with few adverse events, largely mild gastrointestinal symptoms. A larger study, the PKC-DRS2, which randomized 685 patients, showed similar results.⁷⁰

The PKC-DMES Study (Table 3) reported no significant reduction in progression of DR or incidence of DME in 686 patients with mild to moderate nonproliferative DR and no prior laser therapy.^{71 ,78} There was a trend for a reduction in clinically significant DME among patients treated with 32 mg of ruboxistaurin (P = .04), with a larger effect when patients with HbA_1c levels of 10% or greater were excluded (P = .02).

Aldose Reductase Inhibitors. Aldose reductase is the rate-controlling enzyme in the polyol pathway of glucose metabolism and is involved in pathogenesis of DR. Two aldose reductase inhibitors, sorbinil (Pfizer, New York, New York) and tolrestat (Wyeth-Ayerst, St Davids, Pennsylvania), showed no statistically significant effect in reducing DR incidence or progression in RCTs of 3 to 5 years' duration.⁷²

Growth Hormone/Insulinlike Growth Factor Inhibitors. Observations of improvements in DR following surgical hypophysectomy^{79 - 80} and of increased serum and ocular levels of insulinlike growth factor in patients with severe DR led to studies investigating the use of agents inhibiting the growth hormone/insulinlike growth factor pathway for prevention of DR.⁸¹ A small RCT conducted over 15 months among 23 patients reported reduction in retinopathy severity with octreotide, a synthetic analogue of somatostatin that blocks growth hormone,⁷⁴ but another RCT conducted over 1 year among 20 patients⁸² evaluating continuous subcutaneous infusion of octreotide found no significant benefits. Two larger RCTs currently evaluating long-acting–release octreotide injection^{83 - 84} have reported inconclusive preliminary results,⁸⁵ with significant adverse effects (eg, diarrhea, cholelithiasis, hypoglycemic episodes).

Laser and Surgical interventions for Severe Nonproliferative and Proliferative DR.Pan-Retinal Laser Photocoagulation. Pan-retinal laser photocoagulation (PRP), in which laser burns are placed over the entire retina, sparing the central macula, is an established technique for treating severe nonproliferative and proliferative DR⁸⁶ (Table 4).

The strongest evidence comes from 2 related RCTs in the 1970s and 1980s, the Diabetic Retinopathy study (DRS)^{86 ,88} and the ETDRS.¹⁰² The DRS randomized 1758 patients with proliferative DR in least 1 eye or bilateral severe nonproliferative DR to receive PRP or no treatment. At 2 years, severe visual loss (visual acuity <5/200 on 2 successive visits) was observed in 6.4% of treated vs 15.9% of untreated eyes, with the greatest benefit in eyes with high-risk characteristics (new vessels at the optic disc or vitreous hemorrhage with new vessels elsewhere [Figure, B]), in which the risk of severe visual loss was reduced by 50%.⁸⁶

The ETDRS¹⁰² randomized 3711 patients with less severe DR and visual acuity greater than 20/100 to early PRP or deferral (4-month observation and treatment if high-risk proliferative DR developed). Early PRP treatment decreased the risk of high-risk proliferative DR by 50% as compared with deferral, although the incidence of severe visual loss was low in both the early treatment and the deferral groups (2.6% vs 3.7%). Other RCTs^{91 - 93} and a meta-analysis with combined data of 2243 patients⁸⁷ have confirmed the effectiveness of PRP.

Adverse effects of PRP include visual field constriction (with implications for driving^{103 - 104} ), night blindness, color vision changes, inadvertent laser burn, macular edema exacerbation, acute glaucoma, and traction retinal detachment.¹⁰⁵ There is also the possibility of visual loss immediately following PRP. The DRS reported vision loss of 2 to 4 lines within 6 weeks of PRP in 10% to 23% of patients vs 6% of controls.¹⁰⁶

Surgical Vitrectomy for Vitreous Hemorrhage and Proliferative DR. Vitrectomy has been used for treatment of eyes with advanced DR, including proliferative DR with nonclearing vitreous hemorrhage or fibrosis, areas of traction involving or threatening the macula, and, more recently, persistent DME with vitreous traction (Table 5).¹¹⁵ The Diabetic Retinopathy Vitrectomy Study (DRVS) randomized 616 eyes with recent vitreous hemorrhage and visual acuity of 5/200 or less for at least 1 month to undergo early vitrectomy within 6 months or observation.^{107 - 108 ,116 - 117} After 2 years’ follow-up, 25% of the early vitrectomy group vs 15% of the observation group had 20/40 or greater vision, with the benefits maintained at 4 years and longer in individuals with type 1 diabetes. Vitreoretinal surgery has advanced considerably since the DVRS. These advances include intraoperative fundal imaging and laser treatment and bimanual instrumentation to manipulate the retina. These have widened the indications of vitrectomy and may improve outcomes.¹¹⁸

Laser and Surgical Interventions for Diabetic Macular Edema.Focal Laser Treatment. Like PRP, there is good evidence that focal laser treatment preserves vision in eyes with DME. The ETDRS⁹⁶ randomized 1490 eyes with DME to receive focal laser treatment or observation. At 3 years, treatment significantly reduced moderate visual loss as compared with observation,⁹⁶ with the greatest benefits in eyes with clinically significant DME.¹¹⁹ There is limited evidence that laser type (argon, diode, dye, krypton) or method used influences outcomes.^{97 ,120 - 122} Adverse effects include inadvertent foveal burn, central visual field defect, color vision abnormalities, retinal fibrosis, and spread of laser scars.^{105 - 106}

Surgical Vitrectomy for Diabetic Macular Edema. Widespread or diffuse DME that is nonresponsive to focal laser treatment may benefit from vitrectomy.^{123 - 126} However, the few RCTs to date have had small sample sizes and short follow-up, with inconsistent results (Table 5). An RCT of 28 patients with diffuse DME reported reduced macular thickness and improved visual acuities at 6 months after vitrectomy vs observation.¹²⁷ Vitrectomy was superior to focal laser treatment in 1 RCT¹²⁸ but not in others.^{112 - 113} Complications of vitrectomy include recurrent vitreous hemorrhage, retinal tears and detachment, cataract formation, and glaucoma. The presence of vitreous traction and macular edema—now readily documented with optical coherence tomography—in association with visual impairment is currently a common indication for vitrectomy.

Intravitreal Corticosteroids. Corticosteroids have potent anti-inflammatory and antiangiogenesis effects. Intravitreal triamcinolone (IVTA)—ie, injection of triamcinolone acetonide into the vitreous cavity¹²⁹ —has been used for treatment of DME,^{130 - 132} with a number of RCTs demonstrating significant improvements in DME and visual acuity.^{133 - 138} Many of these, however, had small participant numbers and short follow-up. Additionally, there were substantial adverse effects, include infection, glaucoma, and cataract formation.^{109 ,139 - 142}

In the largest RCT having the longest follow-up yet reported, eyes with persistent DME were randomized to receive 4 mg of IVTA or sham injection (saline injection into the subconjunctival space).¹⁰⁹ After 2 years, 19 of 34 IVTA-treated eyes (56%) had a visual acuity improvement of 5 letters or more compared with 9 of 35 placebo-treated eyes (26%) (P = .007). Overall, IVTA-treated eyes had twice the chance of improved visual acuity and half the risk of further loss. However, many eyes required repeated injections (mean, 2.2), and there was significant intraocular pressure elevation (≥5 mm Hg in 68% of treated eyes vs 10% of controls). Cataract surgery was required in 55% of IVTA-treated eyes. Thus, while this study demonstrated significant efficacy of IVTA in persistent DME, larger RCTs are needed to provide further data on long-term benefits and safety.¹⁴³ Additionally, the ideal dose of triamcinolone remains unclear.¹⁴⁴

More recently, intravitreal or retinal implants have been developed, allowing extended drug delivery. A surgically implanted intravitreal fluocinolone acetonide (Retisert; Bausch & Lomb, Rochester, New York) was evaluated in 97 patients with DME randomized to receive either implantation or standard care (laser treatment or observation).¹¹⁰ At 3 years, 58% of implanted eyes vs 30% of controls had resolution of DME (P < .001) and associated improvement in visual acuity. However, adverse effects included a substantially higher risk of cataract formation and glaucoma than that observed in eyes receiving IVTA, with 5% requiring implant removal to control glaucoma.¹¹⁰ An injectable, biodegradable intravitreal dexamethasone extended-release implant (Posurdex; Allergan, Irvine, California) was evaluated in an RCT, with reported improvements in visual acuity and macular thickness.¹⁴⁵ This study, however, also included eyes with macular edema from other causes (retinal vein occlusion, uveitis, and following cataract surgery) and had relatively short follow-up. A larger RCT of Posurdex for DME is currently under way.

Intravitreal Antiangiogenesis Agents. Several RCTs are currently evaluating 3 agents that suppress vascular endothelial growth factor (VEGF) for treatment of DME. Pegaptanib (Macugen; Pfizer, New York, New York) targets the 165 isoform of VEGF for treatment of neovascular age-related macular degeneration (AMD). An RCT of 172 patients with DME randomized to receive repeated intravitreal pegaptanib or sham injections showed that treated eyes were more likely to have improvement in visual acuity of 10 letters or more (34% vs 10%, P = .03), macular thickness (P = .02), and need for focal laser treatment (P = .04) at 36 weeks.¹⁴⁶ Serious infection occurred in 1 of 652 injections (0.15%) and was not associated with severe visual loss.¹⁴⁶ Retrospective data analysis of 16 eyes with proliferative DR also showed regression of neovascularization.¹⁴⁷

Ranibizumab (Lucentis; Genentech, South San Francisco, California) is another anti-VEGF agent used for treatment of neovascular AMD^{148 - 149} and may also be useful for DR and DME.¹⁵⁰ A phase 2 RCT (the RESOLVE study) is currently evaluating ranibizumab in DME. Finally, bevacizumab (Avastin, Genentech) is an anti-VEGF agent similar to ranibizumab that is approved for the treatment of disseminated colorectal cancer and not licensed for intraocular use. However, bevacizumab appears to show similar efficacy for treatment of neovascular AMD and may also be effective for DME and proliferative DR.^{151 - 154} Bevacizumab has attracted interest because of its low cost, but systemic safety is a concern.¹⁵⁵ An ongoing RCT sponsored by the US National Eye Institute is comparing the effects of laser treatment, intravitreal bevacizumab, and combined intravitreal bevacizumab and laser or sham injection on DME.¹⁵⁶

There is strong evidence that tight glycemic control reduces the incidence and progression of DR (Table 6). For type 1 diabetes, the DCCT showed that each 10% decrease in HbA_1c level (eg, 9% to 8%) reduces the risk of DR by 39%, and this beneficial effect persists long after the period of intensive control. In type 2 diabetes, the UKPDS showed that each 10% decrease in HbA_1c level reduces the risk of microvascular events, including DR, by 25% (95% CI, 7%-40%).

There also is strong evidence that tight blood pressure control in patients with hypertension and diabetes is beneficial in reducing visual loss from DR. The UKPDS showed that each 10-mm Hg decrease in systolic blood pressure reduces the risk of microvascular complications by 13%, independent of glycemic control. The benefit of blood pressure treatment in normotensive patients with diabetes is less clear.

There remains inconclusive evidence about the benefits of lipid-lowering therapy for DR prevention. There also is little evidence that aspirin, other antiplatelet agents, or aldose reductase inhibitors confer any benefit in reducing progression of DR. The role of PKC and growth hormone inhibitors is currently unclear, and results from ongoing trials are pending.

Proliferative DR. There is strong evidence that PRP significantly reduces the risk of severe vision loss from proliferative DR by at least 50%. The benefits are most marked in those with high-risk proliferative DR, in whom PRP should be commenced without delay.⁸⁹ Early vitrectomy should be considered in patients with type 1 diabetes and persistent vitreous hemorrhage or when hemorrhage prevents other treatment. The benefits of vitrectomy are less clear for those with type 2 diabetes. With advances in vitreoretinal surgery, vitrectomy may be indicated earlier in eyes with nonclearing hemorrhage.

Nonproliferative DR. Although there is level I evidence that early PRP reduces the risk of severe visual loss in nonproliferative DR, the absolute risk reduction from early PRP treatment is small, and the risks of deferred treatment are low. In mild to moderate nonproliferative DR, systemic factors such as control of glycemia and blood pressure should be gradually optimized and PRP deferred with careful follow-up. The ETDRS and other RCTs⁹⁵ suggest that PRP should be considered in more severe nonproliferative DR, especially in patients with type 2 diabetes. This benefit for PRP should be balanced against the small risk of vision loss. Early PRP is recommended in these patients if regular follow-up examination is not feasible, if there is significant media opacity or cataract that may affect the ability to apply future laser treatment, or if there are concomitant risk factors (eg, pregnancy) for rapid progression.

Diabetic Macular Edema. There is strong evidence that focal laser photocoagulation reduces the risk of moderate vision loss in DME that poses risk to fixation (or clinically significant DME) by at least 50% and increases the chance of visual improvement. In patients with coexistent proliferative DR and DME, focal laser treatment concurrent with or prior to PRP is recommended.⁸⁹

There is moderate evidence that IVTA may be useful in eyes with persistent DME and loss of vision despite conventional treatment, including focal laser treatment and attention to systemic risk factors. Patients should be warned of adverse effects and the need for reinjection. Further studies are warranted to determine the ideal dose and longer-term efficacy and safety. Intravitreal anti-VEGF agents are being evaluated in several clinical trials; until results are available, there is currently insufficient evidence recommending their routine use.

There is weak evidence that vitrectomy may be beneficial in some patients with DME, particularly in eyes with associated vitreomacular traction, but well-conducted studies with longer follow-up are needed.

Although DR remains the leading cause of preventable blindness in working adults, there are primary and secondary interventions proven effective in limiting visual loss. The indications, efficacy, and safety of newer medical and surgical treatments, however, require further evaluation.