From Bedside to Bench and Back: Open Problems in Clinical and Basic Research

Abstract

Diabetic retinopathy remains a leading cause of visual loss in the working age population of industrialized countries. Many unsolved problems remain as diabetic patients are still going blind, and many patients in poor countries have no access to screening or effective treatment of diabetic retinopathy. In particular, macular edema is becoming more and more the problem of elderly patients, for which we lack effective, definitive treatments. Clinical trials testing proofs of concept developed from in vitro and animal experiments have so far produced mixed results. Optimizing blood glucose and pressure control, platelet-active agents and blockers of the renin-angiotensin system appear to slow down retinopathy at initial or mild stages, whereas more advanced presentations may not be affected by systemic medication. The only options for sight-threatening retinopathy are either destructive (laser photocoagulation) or invasive (intravitreous administration of steroids and VEGF antagonists). The many questions that remain to be addressed are discussed in this chapter.

Copyright © 2010 S. Karger AG, Basel

Overlooking the last 40 years of research in diabetic retinopathy, it appears that the retina was at the core of studies aiming at the improvement of the outcome of a patient with diabetes. The devastating condition of a blind person dependent on multiple daily injections of insulin always put

a major emphasis on finding a cure for diabetes and its complications.

It was the prevention and secondary intervention of diabetic retinopathy that the Diabetes Control and Complications Trial (DCCT) was primarily designed for, based on findings in diabetic dogs that intensified glucose lowering with insulin was capable of reducing retinopathy [1]. However, the effect of glucose treatment was not 100%, suggesting that even mildly elevated glucose levels would lead to microvascular damage in the eye. Epidemiologic data also suggest that there is a continuous relation between blood glucose and retinal lesions rather than a discontinuous one as indicated by the WHO cut-offs for the diagnosis of diabetes mellitus and impaired glucose tolerance [2].

Recent epidemiologic and daily clinical work suggests that there is a slight but noticeable reduction in incident sight-threatening retinopathy. However, it would be misleading to ease the efforts towards the aims of the St. Vincent Declaration which, at the beginning of the last decade of the 20th century called for a reduction in diabetes-related blindness by one third within the next 5 years. Many unsolved problems

remain as diabetic patients are still going blind, and many patients in poor countries have no access to screening or effective treatment of diabetic retinopathy.

Proliferative retinopathy, although still a severe sight-threatening condition especially for people with type 1 diabetes can be controlled fairly well by scatter, or panretinal, photocoagulation [3], whereas breakdown of the blood-ret- inal barrier and the subsequent development of macular edema affects patients with both type 1 and 2 diabetes. Since type 2 diabetes is at least ten times more prevalent than type 1, macular edema is becoming more and more the sight-threaten- ing problem of elderly patients, for which we lack effective, definitive treatments.

Clinicians and basic scientists embarking on research in retinopathy may wish to address some open problems: Why do retinal capillaries become leaky at some stage(s) of the disease? What triggers growth of new vessels, again at some stages and in some patients only? How does laser work for new vessels and how, as far as it does, in macular edema?

Rather than investigating what happens in advanced retinopathy, however, it is probably more sensible to research what initiates it. Hyperglycemia is necessary, though not sufficient, for diabetic retinopathy to develop, and much work has been directed at establishing how high glucose damages the capillaries and neuroretina. This very rational approach has revealed a number of biochemical alterations caused by high glucose/hyperglycemia and resulted in an elegant unifying hypothesis according to which excess production of reactive oxygen species through the Krebs cycle is at the basis of the four culprits of diabetes-induced tissue damage: accelerated polyol and hexosamine pathways, activation of protein kinase C, and increased advanced glycation end-product formation [4]. Almost as a corollary of this hypothesis, the demonstration that thiamine, by modulating at least three enzymes of glycolysis and the Krebs cycle, may

correct metabolic imbalances induced by high glucose in vitro [5–7] and prevent microalbuminuria [8] and retinal capillary occlusions (though not pericyte loss) [6] in diabetic animals points to a possible simple, inexpensive way of preventing or treating retinopathy.

Food for thought here includes: Which are the first steps in the natural history of retinopathy? Is pericyte loss the earliest event, leading to other changes in a sort of cascade? If so, what microenvironmental changes are causative? If not, what precedes the loss of pericytes? At what stage is the whole process still reversible? Can a clinical trial on the effects of thiamine on retinopathy be organized and carried out?

What has translational research produced so far? Clinical trials testing proofs of concept developed from in vitro and animal experiments have produced mixed results. Inhibition of growth hormone and insulin-like growth factor-1 by a long-acting analogue of somatostatin, octreotide, administered in two dosages versus placebo, was unable to modify the progression of sight-threat- ening retinopathy and reduce the need for laser therapy. A selective inhibitor of PKCβ2, ruboxistaurin, produced marginal improvement in visual acuity in patients with incipient diabetic macular edema [9].

Evidence for involvement of the renin-angio- tensin system (RAS) in retinopathy suggested an intraocular mechanism through which stimulation of AT-1 receptors in the retina might enhance the expression of VEGF, hence edema and angiogenesis. The EUCLID study [10] had suggested that an ACE inhibitor lisinopril reduced progression of retinopathy and the incidence of proliferative retinopathy in patients with type 1 diabetes, probably not driven by a rather small blood pressure reduction of only 3 mm Hg. However, EUCLID was designed to study microalbuminuria and was rather underpowered to address retinopathy; its results were further confounded by differences in HbA1c between the treatment groups.

Another trial, ADVANCE [11]/ADREM [12] showed nonsignificantly protective effects of blood pressure lowering with perindopril, also an ACE inhibitor, associated with a thiazide, indapamide, on progression of retinopathy in 1,241 patients with type 2 diabetes. DIRECT (DIabetic REtinopathy Candesartan Trials), was a program of 3 randomized controlled trials designed specifically to verify if an angiotensin receptor blocker, candesartan, 32 mg/day, administered to 5,231 normoalbuminuric, mostly normotensive patients would: (1) prevent incidence of retinopathy in patients with type 1 diabetes (DIRECT Prevent-1), (2) prevent its progression or cause regression in patients with type 1 diabetes (DIRECT Protect-1), and (3) prevent its progression or cause regression in patients with type 2 diabetes (DIRECT Protect-2) [13, 14]. Mean fol- low-up was 4.7 years. Overall, patients on active treatment ended the trial with less severe retinopathy than those on placebo. Candesartan reduced by 35% the risk of new retinopathy in type 1 diabetes (NNT = 18) and increased by 34% the odds of improvement in type 2 diabetes (NNT = 21), the first ever report of consistent retinopathy regression. The favorable effects of RAS blockade were confirmed by RASS [15], another study of 285 normotensive patients, in which enalapril 20 mg/day and losartan 100 mg/day administered versus placebo reduced the odds of retinopathy progression by 65 and 70%, respectively.

Platelet aggregation has long been suspected to play a role in capillary occlusions [16], but trials with aspirin, dipyridamole and ticlopidine showed a small, clinically nonsignificant reduction in microaneurysm turnover in early retinopathy [17, 18] and no effects in more advanced stages in which, however, aspirin did not increase the risk of bleeding from new vessels [19]. Another molecule with possible vasoactive properties, calcium dobesilate, was recently found to be devoid of therapeutic effect in macular edema [20].

With reference to other possible pathogenic mechanisms, a rather serendipitous result was

reported by the FIELD study [21] in which fenofibrate 200 mg/day reduced progression of existing retinopathy, though not incidence of new retinopathy, and the need for laser treatment for both macular edema and proliferative retinopathy. However, the effect on retinopathy was a tertiary objective of the trial, assessed in only 1,012 out of 9,795 patients enrolled. The effect of fenofibrate was apparently independent of its metabolic action, and did not correlate with glucose, lipid levels or blood pressure.

All in all, the above trials suggest that mech- anism-targeted interventions may work in mild rather than moderate or severe retinopathy, when damage of the capillary wall, and possibly the neuroretina, is far too advanced. Is there a ‘point of no return’ in the natural history of retinopathy? Platelet-active agents appeared to slow down retinopathy at a very initial stage, when only microaneurysms are present [17, 18]. In DIRECT Protect-2 [14], only minimal to mild retinopathy (i.e. microaneurysms with rare hemorrhages and occasional hard exudates or cotton wool spots) was found to regress, whereas more advanced presentations – though still classified as moderate non-proliferative – continued to progress, suggesting that RAS blockade may be effective earlier than previously thought, when capillary occlusions and leakage are not yet predominant. Does RAS blockade work through mechanisms different from VEGF, or does VEGF damage the retinal microcirculation earlier than imagined, and not just by inducing hyperpermeability and angiogenesis?

The results of FIELD [21], if confirmed, will suggest that pathogenesis may be stopped in its steps also in moderate to severe retinopathy. If, as can be hypothesized, different mechanisms preside over subsequent stages of retinopathy, which ones should one elect to study with the best chances of an effective treatment modality in mind?

AsituationinwhichantagonizingVEGFseems definitely effective is when specific antibodies

are administered intravitreously for the treatment of aggressive new vessels and severe macular edema. Several RCTs are currently evaluating three VEGF-suppressing agents: pegaptanib (Macugen; Pfizer) an aptamer which targets the 165 isoform of VEGF, and two antibodies, ranibizumab (Lucentis; Genentech) and bevacizumab (Avastin; Genentech). The latter is approved for the treatment of disseminated colorectal cancer but not licensed, hence used off-label, for intraocular use. VEGF inhibition is effective in reducing both new vessels and edema, which represents a good example of translation from bench to bedside, but so are corticosteroids with their potent anti-inflammatory and antiangiogenic effects. Intravitreal triamcinolone has been used for treatment of diabetic macular edema, with a number of RCTs demonstrating significant improvements in edema and visual acuity [22].

Why are intravitreal steroids effective, albeit temporarily, on both conditions: do edema and new vessels share an inflammatory component or do steroids act through mechanism(s) that are altogether different from their anti-inflammato- ry properties?

Another common opinion is that retinopathy can be prevented by optimizing blood glucose and blood pressure control. Probably as a result of continuously improving clinical attention, drug delivery technology and widespread availability of monitoring systems, some latest epidemiological surveys report decreasing incidence of proliferative retinopathy among patients who developed diabetes in the most recent years [23, 24]. Data from the DCCT/EDIC confirm that, 30 years after enrolment, the cumulative incidence of proliferative retinopathy in the patients who had been on intensive insulin treatment during the trial is 21%, compared to 50% in those who were on conventional treatment [25]. Although encouraging from a public health point of view, the true effects of such measures in individual patients may be less than ideal. Retinopathy may be delayed rather than reduced and, since improved

treatment leads to prolonged life expectancy, the final result might be a shift of the curve, with cumulative incidence adding up later in life. In addition, as reported in a retrospective evaluation of all patients who participated in the DCCT, 10% of those who remained in the best quintile for control (i.e. with an HbA1c ≤6.87%), whether in the active or the control group, developed retinopathy, whereas 43% of those in the worst quintile (i.e. HbA1c ≥9.49%) did not develop any lesions over the study period [26]. This corroborates clinical wisdom and strongly suggests that other factors play a pivotal role in the pathogenesis of this complication. More possible subjects for investigative minds include the search for genetic determinants which might make patients especially prone, or resilient, to microangiopathy.

Another reason why it is not justified to sit smug on the large trials that proved a link between blood glucose levels and retinopathy is the difficulty with which optimized metabolic control is achieved in the diabetic population at large. Surveys from different countries prove that only a minor share of patients do achieve the targets set by scientific societies. In the US [27], France [28], UK (Gill, 2003), Italy [29] and other countries, the percentages of patients with an HbA1c lower than 7.0% are less than half, often less than one third. Patients on insulin do worse than those on oral agents, and both are worse than those on lifestyle intervention (diet) only [30]. Possible reasons include medical inertia, poor patient adherence to treatment, insufficient effect of lifestyle and pharmacological interventions, environmental and socioeconomic obstacles, but none of these appears to entirely account for the difficulty in optimizing metabolic control. Although physicians are not very reactive to abnormally high values of HbA1c or blood pressure [27], a clinic-based intervention study in Liverpool showed virtually no effect in proactively pursuing improved metabolic control, except in patients on diet only [30]. Individual patients may be ‘set’ on different degrees of diabetes severity, hence different levels

of metabolic control, a pragmatic though perhaps slightly heretical suggestion. Among children the situation is even worse, with less than 5% of patients having an HbA1c less than 7.0% and more than 80% above 8.0% [31]. Maybe the therapeutic targets are too ambitious, at least for very young and very old patients. In terms of personal motivation, only exceptional and time-limited circumstances may be powerful enough, as in the case of pregnancy, when 80% of patients achieve levels of 6.5% or less [32]. In any case, data from the 1999–2004 National Health and Nutrition Examination Survey suggest that the percentage of patients with an HbA1c lower than 7.0 in the US is slowly increasing [33].

As optimized control is not always attained and, even when it is, not necessarily effective, the search must go on for a straightforward pathogenic mechanism that would explain the natural history of retinopathy and indicate clear-cut therapeutic targets. There are also solutions which need to be applied to the most urgent questions arising from clinical work with patients. The subsequent paragraphs will address a selection of items from recent roadmaps proposed by the NIH (http://www2.niddk.nih.gov/AboutNIDDK/ ResearchAndPlanning/Type1Diabetes/). Similar activities are ongoing in the EU (http://www. diamap.eu).

(a) Hyperglycemic Memory. The DCCT demonstrated that good glycemic control in type 1 diabetes reduces the incidence and progression of retinopathy to a large extent. After completion of the trial, many patients formerly in the ‘standard’ therapy group intensified their metabolic control and thereby achieved better HbA1c levels. Patients formerly in the ‘intensified’ therapy group experienced a slight worsening of their HbA1c levels. Subsequent follow-up studies called EDIC demonstrated that despite similar post-DCCT HbA1c levels, patients in the former standard treatment group continued to have a higher rate of developing retinopathy, while members of the former well-controlled group remained protected [34].

This phenomenon is called metabolic memory, and the underlying molecular basis which is currently almost unexplored will help to identify possible targets for novel interventions, and answer the clinically important question of the ‘point of no return’ of retinopathy.

(b)Vascular Repair Mechanisms. In the past, much interest has been focused on the damaging mechanisms of chronic hyperglycemia. Adaptive responses had been largely ignored and were underrepresented. Since diabetic retinopathy is not only the result of tissue damage, but also inferior repair responses, much of the basis for improved treatment and prevention may lie in the therapeutic support of repair mechanisms. As in all other target tissues of diabetic complications, retinopathy starts with progressive vascular dropout. Angiogenic responses leading to proliferative diabetic retinopathy are only secondary and occur only in some patients. Repair of damaged vessels is partly promoted by cells from the bone marrow. In diabetic patients, these cells are reduced in number and dysfunctional [35]. The injection of progenitor cells of nondiabetic origin can correct this defect to a certain extent. It is proposed that research in this area will lead to novel drugand cell-based therapies to restore proper vascular function, including the early lesions in the retina.

(c)Epigenetic Factors. The genetic make-up of an individual determines life span and disease susceptibility. Work over the past years has provided much more information on genetic risk for diabetic nephropathy than for diabetic retinopathy, although it is clear that there is a genetic background of retinopathy susceptibility, at least in type 1 diabetes [36, 37]. Recently, permanent molecular changes which can last through life have been identified that can, for example, explain phenotypic differences in identical twins

[38].These epigenetic changes result either from DNA methylation in which a methyltransferase attached a methyl group to DNA, which leads to gene silencing, or from histone acetylation in

which a silenced gene is activated by acetyl modification of regulatory parts. These permanent modifications can be critical for diabetic retinopathy if hyperglycemia can cause them. Novel findings indicate that transient periods of hyperglycemia can permanently turn on genes that modify inflammatory signaling in target cells [39]. The propensity of glucose-derived epigenetic changes may also be relevant in different stages of retinopathy, as many adaptive and maladaptive factors are stage dependent. Therapeutic strategies targeting epigenetic changes are fascinating, as they would allow interventions beyond a specific point of no return.

(d)New Animal Models. Much of the delay in translating innovative therapies into clinically established treatments results from the paucity of animal models that represent human diseases adequately. In the case of diabetic retinopathy, this is particularly the case. There is no animal model that properly reflects hyperglycemia-induced proliferative diabetic retinopathy and diabetic maculopathy. Most of the animals used to investigate the effectiveness of specific treatments only reflect certain aspects of the complex human pathology. Moreover, it has been recently appreciated that even the genetic background of an experimental model – mostly mouse models – have a profound effect on the phenotype in a transgenic or a knockout setting. It is thus concluded that animal models that mimic the human development of advanced diabetic retinopathy are desperately needed for research on mechanisms and for drug development.

(e)Systems Biology. It has become clear – and chapters in this book reflect it – that diabetic retinopathy is more than a microvascular disease. Cell-cell communication under hyperglycemic conditions must be studied in tissue context including the impact of the neuroglia on vascular response to identify relevant signaling pathways addressable by therapeutics. On the cellular level, each glucose-induced abnormality is in a context with other molecules. The same can be viewed

on the tissue level. The emerging field of systems biology is capable of analyzing the many simultaneously occurring events as complex, interconnected circuits. These circuits may have control (check) points amenable to biological interaction. The closer in vitro and in vivo systems will be modeled to diabetic retinopathy the better the answers will be provided with this technology.

Apart from these scientific areas which address current and future questions to be solved for an improved understanding of diabetic retinopathy, there are more urgent areas which have been identified as important. The technique of high throughput screening is a powerful tool to identify disease-relevant target molecules for treatment. However, the design of the model systems by which they are selected is highly critical. The adjustment on retinopathy is a challenging task, as is the identification of early biomarkers of retinopathy. Some biomarkers may help identify early functional or structural lesions in patients prone to progression to sight threatening stages. Others may indicate patients with fast progressing disease, while other biomarkers are sought which reflect relative protection over many years. A third group of markers in patients is essential that is also present in animals and which indicates early (nontransient) structural changes. These markers may also be useful to facilitate noninvasive imaging as the eye is the only organ in diabetes in which blood vessels can be visualized noninvasively.

If these and other activities will have led to the identification of putative therapies, patients with appropriate levels of the disease are essential for the mandatory clinical trials. The lag phase between target discovery and clinical application in routine patients is partly so long because of the paucity of existing collaborative networks that provide infrastructures for rapid outset of clinical trials. A positive exemption is the Diabetic Retinopathy Clinical Research Network (DRCR. net) and other networks which are currently developing.