Modern Strategies to Save Sight in Diabetes

Abstract

Owing to the global epidemic of obesity, the incidence of diabetes and its complications are rapidly increasing around the world. Yet, visual impairment caused by diabetic retinopathy is paradoxically on the decline. This improvement is due to better understanding of etiologic mechanisms, increased screening, and advent of newer therapies. Here, we discuss salient developments in the comprehension of the root causes of diabetic retinopathy and the molecular mechanisms underlying current treatment approaches.

Introduction

Diabetic retinopathy is a progressive visual disorder that complicates all forms of diabetes mellitus. The disease is staged based on ophthalmoscopic features, and is commonly divided into non-proliferative and proliferative forms. These features involve varying forms of microvascular pathology, ranging from disintegrity of the blood-retina-barrier, vascular proliferation, fibrosis, and generation of tractional force on the neural retina. Macular edema (See Figure 1), and macular ischemia, two major causes of vision loss in diabetes, can occur during any stage of retinopathy.

Figure 1.

Macula edema from diabetic retinopathy. (A) Color fundus photograph centered on the optic nerve and macula demonstrating center-involving macular edema, demarcated by a ring of hard exudate (white circle). Also pictured are several features of severe non-proliferative diabetic retinopathy, including cotton wool spots (arrowhead) and intraretinal hemorrhage (arrow). (B) Optical coherence tomography (OCT) in a patient with macular edema, showing increased retinal thickness (red bracket) with fluid-filled ‘cystoid’ spaces (asterisks) compared to a relatively unaffected area (yellow bracket).

The incidence of diabetes-related eye disorders continues to grow at alarming rates, despite authentic advances in diabetes screening and care. Such increases are due, in large part, to the growing epidemic of obesity worldwide and the consequent explosion in rates of type 2 diabetes mellitus. Yet, despite these sobering facts, recent studies suggest the current incidence of diabetes-related visual impairment is smaller than that of 25 years ago.¹ A major reason for this improvement is the rapid evolution of diabetic retinopathy management that has occurred over the past ten years. In the following sections, we discuss current therapeutic techniques for diabetic retinopathy and how they curb vision loss in this devastating disease.

Systemic Therapy

Data from several clinical trials suggest a continuous relationship between glycemia and retinopathy progression. In the Diabetes Control and Complications Trial (DCCT), patients with relatively new-onset type 1 diabetes were randomized to intense blood glucose control (target glycated hemoglobin < 7%) or standard control (target <8%). The relative risk of retinopathy progression was 54% and 76% lower in the intensely-controlled patients compared to controls, among those with or without preexisting disease, respectively.² Furthermore, long-term data from the continuation study, Epidemiology of Diabetes Interventions and Complications (EDIC), show that this initial period of intense glucose control had durable effects on retinopathy protection.³ A linear relationship between average blood glucose levels and retinopathy progression was also seen in patients with type 2 diabetes, as shown in the United Kingdom Prospective Diabetes Study (UKPDS)⁴

The robust effects of glycemic control on ocular health, in both type 1 and type 2 diabetes, have led to an increasingly glucose-centric view of pathogenesis in diabetic retinopathy. Preclinical investigation has linked hyperglycemia to multiple potential mechanisms of disease pathogenesis. For example, activation of protein kinase C (PKC) isoforms leads to increased vascular permeability and angiogenesis in animal models of diabetic retinopathy.⁵ Moreover, inhibition of the beta isoform of PKC (PKCβ) with ruboxistaurin ameliorates retinal microvascular complications in animal models and in early phase clinical trials.⁶ Hyperglycemia also leads to increased aldose reductase and polyol enzyme activity, resulting in sorbitol production and nicotinamide dinucleotide phosphate (NADPH) depletion. The deficit of NAPDH, a cofactor in the synthesis of nitric oxide, causes decreased nitric oxide production and, consequently, impaired retinal arteriolar vasodilation.⁷ Hyperglycemia may increase oxidative stress by promoting formation of advanced glycation end products, which subsequently induce reactive oxygen species.⁸ The final common pathway of these intracellular, hyperglycemia-mediated biochemical processes is often retinal capillary endothelium and pericyte apoptosis.

Although these animal and clinical studies strongly argue that glucose homeostasis plays a central role in the development of retinopathy, several lines of data suggest that multiple other causative pathways likely exist. For example, the aldose reductase inhibitor sorbinil failed to prevent visual loss due to diabetic retinopathy in a large clinical trial.⁹ The clinical trials mentioned above establish a correlation between blood glucose levels and retinopathy, but do not provide any evidence that blood glucose itself is directly causing the microvascular disease. In fact, there is no clinical evidence allowing us to separate effects of glucose-lowering and direct effects of insulin supplementation or sensitization on retinopathy. Although the retinal endothelium primarily expresses GLUT1 and GLUT3 glucose transporters, which are insulin-independent, several retinal neurons and glia are responsive to insulin and express insulin receptors.¹⁰ Such data raise the possibility that diabetic retinopathy may result from insufficient insulin-dependent signaling in the neural retina.

While intensive metabolic control using glycated hemoglobin levels as the primary index of therapy has clear microvascular benefits, this approach may also cause severe harm. In the Action to Control Cardiovascular Risk in Diabetes (ACCORD) study, patients in the intensive glucose control group (HbA1c <6%) had a higher mortality rate compared to those in the standard control group, a finding that halted the trial prematurely.¹¹ Additional data from both inpatient and outpatient settings also suggested increased mortality and/or higher incidence of severe hypoglycemia with intensive glucose control.¹² Cumulatively, these studies raise a bona fide concern for increasing mortality with aggressive therapy aimed at lowering blood glucose, and introduce severe limitations to glucose-based therapy in high-risk patients with poor overall health. Therefore, when brittle patients with progressive but sub-threshold retinopathy cannot tolerate aggressive glucose control, what other treatment choices are available for physicians to minimize or prevent the need for ophthalmic therapy?

A promising adjunct to glucose-based therapy to reduce microvascular disease in diabetes is lipid control. Two independent clinical trials have demonstrated the efficacy of fenofibrate in decreasing the progression of diabetic retinopathy. In the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study, a multicenter, multinational trial based in Australia, fenofibrate treatment reduced the requirement for laser therapy for both macular edema and proliferative diabetic retinopathy.¹³ Unexpectedly, this effect was independent of plasma triglyceride concentrations, raising questions about the mechanism by which fenofibrate protects microvessels. These results were corroborated by the ACCORD eye sub-study, which also assessed the role of fenofibrate in protection from diabetic microvascular complications. In ACCORD, fenofibrate reduced the progression of diabetic retinopathy to 6.5% from 10.2% in the placebo group, over a four years period. Surprisingly, fenofibrate therapy decreased the rate of retinopathy progression at a rate similar to intensive glycemic control, as shown in the ACCORD glycemia trial (7.3% and 10.4%, in the intense glycemic control and standard control groups, respectively). As in the FIELD study, the ACCORD eye trial did not show an association between the lipid effects of fenofibrate and the progression of diabetic retinopathy.¹⁴ Despite these data, treatment of diabetic microvascular disease with fibrates has failed to gain widespread traction. Nevertheless, in 2013 Australia became the first nation in the world to approve fenofibrate therapy for the sole indication of preventing retinopathy progression.

There are several potential ways that fenofibrate may shelter the retina from diabetes-induced damage. The drug is a potent ligand for peroxisome proliferator-activated receptor alpha (PPARα). In mouse models, fenofibrate prevents development of diabetic retinopathy in a PPARα-dependent manner, and genetic deficiency of PPARα increases susceptibility for diabetic retinopathy.^3,15 PPARα is expressed in multiple retinal cell types, but its retinal targets remain unclear. Fenofibrate may also reconfigure the local lipid landscape of the retina in a manner that protects it from metabolic stress. For example, it alters lipid and triglyceride levels in a PPARα-dependent fashion to produce an environment inhibitory to angiogenesis and inflammation.¹⁶ Fenofibrate may also upregulate ApoA-1, which acts as a reactive oxygen species scavenger and decreases local oxidative stress.¹⁷ These findings are encouraging, but the precise mechanisms by which fenofibrate protects the retina in diabetes are still elusive. While current investigative efforts are focused on direct actions of the drug in the retina, acting through PPARα, an indirect action occurring via modification of circulating lipids may be responsible for its beneficial ophthalmic effect.

Retinal Laser Photocoagulation

For decades, laser photocoagulation was the mainstay of treatment for advanced diabetic retinopathy. Diffusely scattered peripheral retinal treatment (known as pan-retinal photocoagulation) reduces the risk of severe vision loss in eyes with severe non-proliferative diabetic retinopathy and proliferative diabetic retinopathy by 50% compared to untreated eyes.¹⁸ More directed and limited treatment of the macula (known as focal or grid laser photocoagulation) reduces vision loss in patients with macular edema by nearly 50%.¹⁹ Eyes with more advanced vision loss at baseline (poorer than 20/40 ETDRS), are also more likely to have a visual gain of six or more letters with photocoagulation compared to observation only.^19,20

Although much older than most current therapies for diabetic retinopathy, laser photocoagulation remains the standard against which other therapies are measured. Laser photocoagulation induces a local thermal burn at the site of application, ablating cells in the deep layers of the retina and its supporting epithelium, eventually causing a scar at the treated site. Photocoagulation is thought to improve retinal edema through various actions. For example, it may upregulate the activities of support cells known as the retinal pigment epithelia (RPE), induce proliferation of RPE, or directly promote closure of leaking vascular abnormalities by direct thrombosis. However, evidence to support these proposed mechanisms is lacking. More likely, photocoagulation of all types is effective in diabetic retinopathy because ablation of retinal tissue leads to diminished metabolic demands.

While the efficacy of laser photocoagulation in diabetic retinopathy is well established, this treatment is employed at great expense. Retinal laser treatment is inherently associated with visual loss at the site of application – but the goal of therapy is to spare central vision at the cost of more eccentric vision (akin to amputation of ischemic limbs). Excessive energy used during laser photocoagulation may lead to progressive atrophy around the burn site, even years or decades after treatment. (See Figure 2.) Laser therapy may also lead to premature presbyopia (farsightedness) and nyctalopia (inability to see in low light), since rod photoreceptors are much more numerous in the peripheral retina.

Figure 2.

Extensive chorioretinal atrophy from prior laser photocoagulation. Image from a scanning laser ophthalmoscope showing retinal scarring from pan-retinal photocoagulation encroaching on the macula (asterisks) and focal macular laser scarring (arrows) immediately adjacent to the fovea (arrowhead).

Anti-inflammatory Agents

Diabetes increases numerous cytokines in the retina, including interleukin-1β, tumor necrosis factor α (TNF-α), intercellular adhesion molecule-1, vascular cell adhesion molecule-1, and angiotensin II.²¹ In particular, TNF-α causes retinal injury by promoting angiogenesis, enhancing leukocyte adhesion to endothelial cells, increasing reactive oxygen species, and inducing endothelial cell dysfunction and apoptosis. Furthermore, leukocyte entrapment in retinal vasculature has been implicated as an early finding in the development of retinopathy in animal models.²² Together, these aberrant inflammatory signals impair the integrity of the blood-retina barrier, thereby promoting diabetic macular edema. Corticosteroids block these signaling pathways by inhibiting arachidonic acid formation and downstream inflammatory cascades. Expression of vascular endothelial growth factor (VEGF), a potent effector of microvascular pathology, is also reduced by corticosteroid administration.²³

Corticosteroids are effective in reducing macular edema when delivered into the vitreous in bolus doses.²⁴ Furthermore, periodic intravitreous monotherapy with triamcinolone (at either 1-mg or 4-mg doses) is superior to focal/grid photocoagulation for diabetic macular edema in terms of visual acuity gains among patients who have already had cataract surgery.²⁵ Intraocular corticosteroids are frequently associated with cataract formation and intraocular pressure rise. Consequently, they are most often used in cases of macular edema refractory to other therapeutic modalities or in patients without glaucoma who have previously undergone cataract extraction.

Extended-release steroid implants for vitreous injection or implantation (including dexamethasone 0.7 mg implant and fluocinolone 0.2 g/day insert) are now also approved for use in diabetic macular edema. The dexamethasone implant delivers a low dose of medication over a period of six months²⁶ and the fluocinolone insert is active for up to 36 months,²⁷ which provides an advantage over other therapies that require more frequent administration. At three years follow-up, 22.2% of individuals treated with the 0.7 mg dexamethasone implant experienced greater than 15-letter ETDRS visual acuity improvement compared to 12.0% of patients receiving sham therapy.²⁶ Similarly, at two years of follow up, 28.7% of patients given the 0.2 g/ day fluocinolone insert gained 15 or more ETDRS letters of visual acuity compared to only 16.2% of patients receiving a sham injection.²⁷ While the use of these agents offers the potential of reducing treatment needs, their higher side effect profile compared to vasogenic antagonists usually relegates them to a second tier treatment modality, albeit one that is highly useful when designing individualized therapy based on specific patient features.

Vasogenic Antagonism

A little over ten years ago, the treatment paradigm for diabetic retinopathy was radically altered when a novel class of targeted molecular therapy antagonizing the effects of VEGF was introduced into ophthalmic practice. Multiple retinal disorders are associated with increased VEGF levels, including diabetic retinopathy, suggesting that this molecule plays a central role in their pathogenesis.²⁸ VEGF disrupts the blood retina barrier in diabetic retinopathy by increasing vascular permeability via activation of PKCβ.²¹ This results in phosphorylation of the tight junction protein occludin, which leads to disassembly of tight junctions and increased vascular permeability.²⁹ Macular edema then ensues. In the development of neovascular disease, VEGF acts as a potent mitogen for the retinal endothelium, upregulates expression of cell adhesion molecules, and orchestrates metalloproteinase-mediated basement membrane proteolysis.³⁰

The canonical growth factor signaling network associated with VEGF therefore offers an attractive pathway that can be manipulated to ameliorate retinal disease. Multiple pathway antagonists, including humanized mouse monoclonal antibodies against VEGF (e.g. bevacizumab and ranibizumab) or a chimeric VEGF receptor decoy (aflibercept), are highly effective in treating diabetic macular edema when injected into the vitreous cavity. While each of these three medications is molecularly distinct, utilization of aflibercept, bevacizumab, and ranibizumab result in similar visual acuity improvements in eyes with mild (visual acuity 20/32 – 20/40) center-involved diabetic macular edema. Such superior visual acuity outcomes are amplified among patients with severe macular fluid burdens at baseline.³¹ Further, treatment with any of these agents is associated with better visual outcomes when compared to laser treatment alone.³² It comes as no surprise then that injectable therapy with VEGF pathway blockers has largely supplanted laser monotherapy as the initial treatment choice for central diabetic macular edema among retina specialists. Injections of VEGF antagonists can be used to treat proliferative diabetic retinopathy and, in many patients, these medications are non-inferior to pan-retinal photocoagulation, which is currently the standard-of-care.³³

But the enthusiasm for use of VEGF blockers must be bridled by an understanding of their potential adverse effects. A major limitation of all currently used ophthalmic VEGF antagonists is the need for frequent re-treatment, especially early in the course of therapy. Bevacizumab, a full length IgG molecule, has a half-life of approximately 5.6 days in the vitreous. The half-life of ranibizumab, an immunoglobulin fragment with 1/3 of the molecular mass of bevacizumab, is 3.2 days. Aflibercept, which is both larger and more broadly avid for VEGF isoforms compared to anti-VEGF antibodies, has an estimated half-life of 4.8 days.³⁴ Accordingly, the therapeutic effect of each of these medications lasts no more than six weeks, necessitating frequent re-injection to treat active disease. Multi-year follow-up data involving all three therapies suggest that treatment requirements do in fact decline following the initial year, when study protocols involving combination therapy are strictly followed.³³ Complications of intravitreous injections of anti-VEGF agents are rare. Endophthalmitis, the most visually devastating complication, occurs at a rate of 0.053%.³⁵ While initial clinical trials raised concerns over slightly increased risks of cardiovascular and hemorrhagic events using these agents, subsequent analyses have failed to substantiate them.³⁶ A seldom-discussed undesirable effect of chronic VEGF antagonism is the disruption of trophic homeostasis of the retina. VEGF is secreted by multiple retinal cell types, and receptors for VEGF are littered throughout the healthy neural and vascular retina.³⁷ In long-term studies of VEGF antagonism in age-related macular degeneration, a dose-dependent increase in neuroretinal atrophy is observed.³⁸

Although VEGF levels are elevated in the ocular fluids of patients with diabetes and in animal models of diabetic retinopathy, the precise mechanisms of secretion (including which cell types are responsible) are still being delineated.²⁸ The oxygen-sensitive von Hippel Lindau (VHL) – hypoxia inducible factor alpha (HIF-1α) axis is a primary transcriptional regulator of VEGF.³⁹ However, several other retinal disorders, including retinal vascular occlusion and retinopathy of prematurity, are also dependent on HIF-1α stabilization with subsequent VEGF release, suggesting that this pathway represents a common final effector for a multitude of pathological insults. In diabetic retinopathy, the precise upstream initiators of this cascade have not yet been identified. Potential mechanisms include altered retinal fuel utilization due to deficient insulin action, neuroretinal inflammation from metabolic insults (as discussed above), direct neuroretinal glucotoxicity or lipotoxicity, endothelial dysfunction as a result of diminished insulin signaling or excess glucose, or miscommunication between the neural retina and the retinal microvasculature, which are normally joined in harmony by a sophisticated neurovascular unit – an entity that is still poorly understood.⁴⁰ The future of research into the underlying pathophysiology of diabetic retinopathy offers exciting prospects for discovery related to each of these potential causative pathways.

Conclusion

Over the past three decades, improvements in treating diabetic retinopathy have occurred in parallel with advancements in understanding the pathophysiologic mechanisms responsible for disease onset and progression. Pharmacotherapies, including VEGF pathway inhibitors and glucocorticoids, have provided useful adjuncts to laser treatment. This has resulted in a decreased incidence of diabetes-related vision loss, despite the expanding epidemic of diabetes. Yet, all current ophthalmic therapies are directed at late stages of diabetic retinopathy. Future treatment should be directed towards earlier pathophysiology and may include manipulation of metabolic pathways in the retina, particularly those affected by PPARα signaling.

Biography

Stanford C. Taylor, MD, (left), is a Resident in Ophthalmology, and Rithwick Rajagopal, MD, PhD, (right), MSMA member since 2014, is an Assistant Professor, Department of Ophthalmology and Visual Sciences Washington University School of Medicine, St. Louis.

Contact: rajagopalr@wustl.edu