Diabetic retinopathy and systemic factors.

Diabetic retinopathy, an oculardisease, is governed by systemic as well as local ocular factors. These include primarily chronic levels of blood glucose. Individuals with chronically elevated blood glucose levels have substantially more, and more severe, retinopathy than those with lower blood glucose levels. The relationship of blood glucose to retinopathy is continuous, with no threshold although individuals with hemoglobin A1c levels (a measure of chronic glycemia) <6.5%, generally develop little or no retinopathy. Blood pressure levels have been claimed to influence retinopathy development and progression, but multiple controlled clinical trials of antihypertensive agents in diabetic subjects have produced only weak evidence of benefit from blood pressure lowering on the incidence and progression of diabetic retinopathy. Elevated blood lipids seem to play a role in the progression of retinopathy, and two trials of fenofibrate, a lipid-lowering agent that has not proved effective in preventing cardiovascular disease, have shown benefit in preventing retinopathy progression. The mechanism of this effect may not, however, be directly related to the reduction in blood lipids. Finally, there is strong, but only circumstantial, evidence for a genetic or epigenetic influence on the pathogenesis of diabetic retinopathy. Despite the power of large-scale epidemiologic studies and modern molecular biological and computational techniques, the gene or genes, which predispose or protect against the development and progression of diabetic retinopathy remain elusive.

INTRODUCTION

Diabetes mellitus is a systemic disease in which blood glucose levels become chronically, and often severely, elevated either because insulin is not secreted from the pancreatic islet cells (type 1 diabetes), or because the insulin that is secreted is, for a variety of reasons, less than normally efficacious (type 2 diabetes). Over years of chronic hyperglycemia, many individuals develop severe damage to different organs and tissues: Among others, the kidneys, the peripheral nervous system, the heart and great vessels, and the topic of this discussion, the retina. While treatments to the specific organs that are involved in these complications have been developed, e.g., laser photocoagulation and intraocular injections of antiangiogenic agents for diabetic complications in the retina, it also follows that systemic treatments for this systemic disease should be beneficial as well.

This last statement should be self-evident, but it was heavily debated for many years by investigators who felt that the late complications of diabetes in many organs were genetic, perhaps related in some fashion to the aberrant genes leading to the hyperglycemia of diabetes but independent of the chronic hyperglycemia itself. Recent work, in particular a number of randomized, controlled clinical trials reported over the last 25 years, have shown that this hypothesis is untrue as a single, unifying explanation for the pathogenesis of retinopathy and other complications of diabetes, but, as I will explain, neither is it completely irrelevant. This review will discuss the effects of controlling chronic hyperglycemia, as well as other systemic therapies, on the development and progression of diabetic retinopathy. This topic has also been discussed by Lingam and Wong1 in a recent review in this journal, and I have discussed certain aspects of it in a recently published editorial.2

Systemic medical management for diabetic retinopathy can be divided into four categories:

Control of blood glucose

This issue was hotly debated for many years, but its rigorous investigation could not be undertaken until the development of methods for home blood glucose monitoring using electronic assays based on glucose oxidase and test devices that retained the results in memory; and the recognition that nonenzymatic protein glycation produced long-lived glycated molecules whose blood levels averaged systemic blood glucose values for the lifetime of the protein. These include, in particular, glycated hemoglobin, which has an average lifetime of 120 days, and whose blood level, therefore, represents the average blood glucose concentration for that period of time.34 Glycated hemoglobin levels are usually expressed as the percentage of total hemoglobin, either as the percentage of total glycated hemoglobin, or as the percentage of its largest fraction, hemoglobin A1c (usually abbreviated HbA1c; other, lesser, glycated fractions are HbA1a and HbA1b) period, the level of HbA1c is usually about two percentage points lower than that of total glycated hemoglobin. HbA1c levels in nondiabetic humans range between 4% and 6%, and the risk for development of the microvascular complications of diabetes (retinopathy, nephropathy, and neuropathy) in controlled clinical trials has been shown to decrease steadily with decreasing HbA1c levels.567 A study of over 44,000 individuals from 9 countries,7 which evaluated fasting and 2-h postprandial plasma glucose and HbA1c levels found a curvilinear plot of diabetic retinopathy versus HbA1c, and suggested a threshold for the diagnosis of diabetes, based on several different statistical criteria, defined by the presence of diabetes-specific retinopathy in this large population as a function of HbA1c level. The statistical cutoff point for this diagnosis was determined to be an HbA1c threshold of >6.5%, which is the accepted international standard. This “threshold” is not absolute, however, because individuals with even lower HbA1c levels have, on infrequent occasions, been found to have retinopathy, albeit minimal [Figure 1].

Figure 1

This patient, who had type 1 diabetes of 30 years duration, maintains a vegetarian diet with excellent blood glucose control. His most recent HbA1c value was 6.3%. Nevertheless, he has minimal diabetic retinopathy (note arrows) in both eyes

With the use of these techniques for determining objectively blood glucose levels in individual patients, several randomized, controlled clinical trials of the effect of reducing blood glucose levels on the microvascular complications of the disease over time were undertaken in subjects with type 1 and type 2 diabetes.56891011 All of these studies clearly indicated that chronically lowering blood glucose levels reduced the risk of retinopathy. The largest and most prolonged of these, the Diabetes Control and Complications Trial (DCCT in the United States and Canada) for type 1 diabetes5 and the United Kingdom Prospective Diabetes Study (UKPDS, in the United Kingdom) for type 2 diabetes,6 showed statistically highly significant reductions in the incidence of retinopathy and in its progression by several criteria in patients who were randomized to tight blood glucose control, by comparison with those randomized to the “standard” blood glucose control group over at least 10 years of study. The differential effect of “tight” blood glucose control on retinopathy in the treatment and control groups in these studies was slow, requiring approximately 2 1/2 years from trial entry to become evident.

In addition to the value of blood glucose “control” for prevention of diabetic retinopathy in subjects with type 1 and type 2 diabetes, two other, unexpected results had important implications for the biology of this microvascular disorder. These results were more dramatically shown in studies of type 1 than in those of type 2 diabetes, perhaps because of the greater reduction in HbA1c in the type 1 treatment groups than was evident in subjects with type 2. In two clinical trials,1012 about 10% of individuals with retinopathy at the outset of the study showed transient photographic worsening, primarily the appearance of multiple “cotton wool spots,” lesions that are usually taken to indicate the presence of ischemia in the retina, a tissue that is the most metabolically active in the body in terms of its requirements for glucose and oxygen as energy sources. The explanation for this “early worsening” has been the subject of some speculation by several authors, but I believe the explanation that arose from in vitro experiments in my own laboratory seems most reasonable.13 Using cultured retinal pigment epithelial cells as models for the high metabolic activity of retinal tissue and their production of vascular endothelial growth factor (VEGF), a stimulus for vasoproliferation, as an indicator of ischemia, we found that VEGF production by these cells increased substantially when the oxygen supply in the incubation chamber was reduced. VEGF production in this circumstance could be partially reduced when the glucose concentration of the medium was increased. Alternatively, in a normoxic environment, reducing the glucose concentration in the medium also increased VEGF production, perhaps because this maneuver reduced the cells’ other major energy source, and therefore also led to an ischemic situation. In a human retina that already has some retinopathy, the vascular disease presumably also reduces the blood supply and the retina has become relatively ischemic, but this ischemic situation is partially redeemed by the chronic hyperglycemia in the tissue. If, however, as a condition of the clinical trial, the available glucose is severely reduced, the result, in this already somewhat ischemic tissue, is the imposition of severe ischemia with the resultant appearance of ischemic lesion, namely, “cotton wool spots.”

The second very important result of the long-term follow-up of the DCCT, the Epidemiology of Diabetes Interventions and Complications (EDIC), was the finding that in the 10 years following the conclusion of the treatment phase of the DCCT, when HbA1c values in the original “intensive” and “standard” blood glucose control groups had now come together to an intermediate value of ca. 8.5% for individuals in both groups, retinopathy progression in the original “tight” control group continued to show much slower progression than for those individuals in the “standard” group.14 This long-term process, that in the DCCT/EDIC outlived the period of “tight” blood glucose control imposed by the study, has been called “metabolic memory.”14

The mechanism of “metabolic memory” is unknown but is the subject of speculation: What long-term metabolic processes can be induced by chronic hyperglycemia but then are reversible by normoglycemia? Possibilities include direct changes in the genome, perhaps by acetylation or methylation; epigenetic changes,151617 or modification of proteins, such as the formation of “advanced glycation endproducts”18 that are long-lived and can themselves modify physiologic processes. I would personally speculate that this process, chronic in its onset and its resolution is central to the pathogenesis of diabetic retinopathy itself. Working out this mechanism would, I believe, be an important step to understanding the mechanisms of this disease and its potential reversal.

Control of blood pressure

A large number of studies have evaluated the effect of elevated blood pressure on the development and progression of diabetic retinopathy and conversely, the effect of blood pressure reduction on preventing such progression. Among the more, notable was the UKPDS, which evaluated blood pressure reduction using either an angiotensin-converting enzyme inhibitor or a beta-adrenergic blocker, in conjunction with other drugs as needed to reduce blood pressure, by comparison with controls, who received no antihypertensive medication, in type 2 diabetics.19 Most of these patients were hypertensive (systolic blood pressure more than 150 mm Hg) at the outset of the study. Reduction of blood pressure by either drug prevented retinopathy progression. Other studies, some with diabetic patients who were normotensive at the outset, showed little or no effect of blood pressure reduction. The Action to Control Cardiovascular Risk in Diabetes (ACCORD)-Eye Study, for example, showed that blood pressure reduction in type 2 diabetic patients had no effect on the development or progression of diabetic retinopathy.2021 A recently published systematic review of clinical trials dealing with blood pressure reduction in individuals with type 1 or type 2 diabetes, whether normotensive or hypertensive at the outset, concluded that there is a modest evidence overall for a benefit of blood pressure reduction on the incidence of diabetic retinopathy over a 4–5 years trial period, but, in summarizing results from several clinical trials, there is little overall evidence for a beneficial effect on progression.22

Blood lipid control

A publication several years ago from the Early Treatment Diabetic Retinopathy Study (ETDRS), a clinical trial of argon laser photocoagulation with or without 650 mg/day aspirin treatment for diabetic macular edema and moderate nonproliferative diabetic retinopathy, reported that individuals with macular edema that resolved poorly with the study treatment, and who had extensive macular lipid exudates, had higher blood lipid levels [Figure 2].23 More recently, the ACCORD-Eye study2021 and the FIELD study24 found that treatment of type 2 diabetic subjects with fenofibrate significantly reduced progression of diabetic retinopathy. Although it had been introduced several years ago as a lipid-lowering preventive therapy for atherosclerotic cardiovascular disease, fenofibrate has been little used more recently because of its lack of efficacy for that indication. An explanation for these different results in two different diseases is not apparent, but one possibility is that the lipid-lowering effect of fenofibrate is not the mechanism by which this drug is acting to prevent progression of diabetic retinopathy. For example, diabetes-induced hyperglycemia downregulates the peroxisome proliferator-activated receptor alpha (PPAR α) pathway,25 with adverse effects on retinal vascular cells.26 Fenofibrate is an agonist of PPAR α2728 which may explain its efficacy in preventing progression of diabetic retinopathy without invoking a specific antilipid role. Although the use of systemic drugs, including fenofibrate, is largely not undertaken by ophthalmologists, the apparent special effect of this particular systemic agent on diabetic retinopathy merits a possible re-examination of this custom. I have discussed this point also in a recently published editorial.2

Figure 2

This patient with type 2 diabetes has very extensive lipid deposits in both retinas, with macular edema shown by optical coherence tomography. Although he is using a statin drug, his most recent serum triglycerides were 470 mg/dl, and his serum cholesterol was 325 mg/dl. High-density lipoprotein cholesterol was 42 mg/dl, and low-density lipoprotein cholesterol was not calculated

Genetic studies

The DCCT and UKPDS showed unequivocally the importance of “tight” blood glucose control for reducing the incidence and progression of retinopathy and other microvascular complications of diabetes. But such “tight” control does not entirely prevent these complications. Lachin et al.29 have presented extensive statistical analyses from the DCCT cohort to show that “total glycemic exposure” (HbA1c levels over time) accounts for only about 11% of the reduction in microvascular complications. Another paper from the same study30 showed that diabetic first-degree family members of study subjects who progressed to proliferative or severe nonproliferative retinopathy had a risk ratio for such progression of approximately 3.1 compared to diabetic first-degree family members of study subjects who did not develop such severe progression of retinopathy. Rand et al. reported an association between certain human leukocyte (HLA) antigens and proliferative diabetic retinopathy, with a protective effect of other HLA phenotypes. These associations were negated by the presence of >5 diopters of myopia.31 (Other investigators have also noted the protective effect of myopia on diabetic retinopathy, with increasing degrees of myopia providing steadily increasing protection).32 This strong suggestion of a genetic influence on the development of retinopathy, and similar results suggesting genetic influences on the development of diabetic nephropathy, have led to attempts to find a gene, or genes, responsible for the development of severe retinopathy or nephropathy. These attempts have to date been unsuccessful. One such large-scale, but negative, study in the United States, the Familial Investigation of the Nephropathy in Diabetes (FIND) study evaluated a genetic basis for end-stage diabetic nephropathy and retinopathy in four different ethnic groups in the United States: Caucasians (whites), African-Americans (blacks), Hispanic-Americans, and Native Americans (American Indians).33 A second large investigation, the Candidate Gene Association Resource found some suggestive loci but again, made no definitive genetic associations.34 Similar results were reported from two genome-wide association studies.3536 A paper from the DCCT cohort37 reported modifications in the VEGF gene in study subjects with proliferative retinopathy compared to individuals who did not develop proliferative disease. No genetic variations were reported in subjects who developed macular edema. However, the number of subjects who developed these severe complications in the DCCT cohort was small, and these results have not been repeated in other populations. While there is little doubt that genetic influences are important in the pathogenesis of retinopathy and other complications of diabetes, discovery of these genes, even with the study of large population cohorts and the availability of powerful statistical methods and computational resources, will not be easy.

Other mechanisms, other therapies

Oxidative stress and inflammation are broad-based pathologies that are common to many diseases. A number of investigators have presented evidence that the mechanisms involved in diabetic retinopathy involve inflammatory and/or oxidative insults.38394041 However, attempts to advance these insights by developing preventive therapies or treatments for diabetic retinopathy have to date been unsuccessful. The ETDRS tested aspirin, 650 mg/day, as a treatment for established diabetic retinopathy and found that this effort was unsuccessful,42 although it is of interest that this high dose aspirin therapy did not increase the risk for vitreous hemorrhage in these subjects with diabetic retinopathy,43 an important point because many diabetic patients will be using low-dose aspirin as a prevention for coronary, cerebral and lower extremity vascular occlusions. Another approach that has been used in small clinical studies with uncertain benefit is treatment with tetracycline antibiotics, agents that may have an antioxidant/anti-inflammatory effect in addition to their efficacy as antibiotics.444546 Although the evidence of a beneficial effect from these small studies is unclear, larger clinical trials are probably worthwhile, as would trials using the antioxidant vitamin/zinc combination therapy that was employed in the Age-Related Eye Disease Study, Part 1.47