Alon Harris, PhD, MS Department of Ophthalmology Indiana University School of Medicine 702 Rotary Circle, Room 137 Indianapolis Indiana 46202 USA Tel: + 1 317 278 0177 Fax: + 1 317 278 1007 Email: firstname.lastname@example.org
Diabetic macular oedema (DMO) is an important cause of vision loss in patients with diabetes mellitus. The underlying mechanisms of DMO, on both macrocellular and microcellular levels, are discussed in this review. The pathophysiology of DMO can be described as a process whereby hyperglycaemia leads to overlapping and inter-related pathways that play a role not only in the initial vascular events, but also in the continued tissue insult that leads to chronic DMO. On a macrocellular level, DMO is believed to be in part caused by alterations in hydrostatic pressure, oxygen tension, oncotic pressure and shear stress. Three key components of the microvascular pathways include angiogenic factor expression, inflammation and oxidative stress. These molecular mediators, acting in conjunction with macrocellular factors, which are all stimulated in part by the hyperglycaemia and hypoxia, can have a direct endothelial effect leading to hyperpermeability, disruption of vascular endothelial cell junctions, and leukostasis. The interactions, signalling events and feedback loops between the various molecules are complicated and are not completely understood. However, by attempting to understand the pathways involved in DMO, we can help guide new treatment options targeted towards specific factors or mediators.
Diabetes mellitus (DM) is a chronic and progressive disease affecting all ages of the population (Horton 1995; Chew 2003). Approximately 120 million people are estimated to have diabetes worldwide, including 17 million Americans (Chew 2003). Ocular complications of diabetes include diabetic retinopathy (DR) and diabetic macular oedema (DMO). The latter, which involves retinal oedema in the macular area, is a frequent manifestation of DR and can occur during any stage of the disease (Ciulla et al. 2003), although it occurs more frequently as the duration of diabetes and severity of DR increase (Lopes de Faria et al. 1999). Diabetic retinopathy is the leading cause of blindness among working-age adults (Moss et al. 1998; Williams et al. 2004) and DMO is the most common cause of impaired visual acuity (VA) in diabetes subjects (Moss et al. 1998; Williams et al. 2004). Approximately half the patients with DMO will lose ≥ 2 lines of VA within 2 years (Ferris & Patz 1984). Therefore, understanding the pathogeneses of DMO and DR is important in order to prevent these diabetes complications and to develop new treatments to protect the sight of diabetes patients.
Diabetes mellitus-associated macular oedema is characterized by a thickening of the macular region caused by a breakdown in the blood–retinal barrier (BRB) through dilated capillaries, microaneurysms and loss of pericytes (Ciulla et al. 2003). The Early Treatment Diabetic Retinopathy Study (ETDRS) (1985) defined macular oedema as thickening of the retina and/or hard exudates with retinal thickening within one disc diameter of the centre of the macula. In 2003, the Global Diabetic Retinopathy Project classified disease severity on an international DMO severity scale. The definition included mild, moderate and severe macular oedema according to the amount of involvement of the central macula (Wilkinson et al. 2003).
Population-based studies have helped to elucidate the development of DMO in the context of hyperglycaemia. The Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR) found that the 10-year rate of developing DMO was 20.1% among type 1 diabetes subjects, 25.4% among type 2 diabetes subjects using insulin and 13.9% among type 2 diabetes subjects not using insulin (Klein et al. 1995). A New Jersey-based study found that 15.9% of African-American patients with type 1 DM developed DMO during a 6-year follow-up period (Roy & Affouf 2006). These and other studies, such as the Diabetes Control and Complications Trial (DCCT) and the UK Prospective Diabetes Study (UKPDS), highlight the importance of intensive glycaemic control in preventing DMO. Achieving a haemoglobin A1c (glycosylated haemoglobin) level of 7.5–7.9% (DCCT 1993) or a fasting plasma glucose < 6 mmol/l (haemoglobin A1c, < 7%) (UKPDS 1998a) have been shown to slow the onset and progression of DR in both type 1 and 2 diabetes. With intensive insulin therapy, a reduction in the cumulative incidence of retinopathy among type 1 diabetes subjects of as much as 50% can be demonstrated (DCCT 1993). The current management strategy for DMO and DR requires early detection and optimal glycaemic control to slow the progression of ocular disease.
The contribution of hyperglycaemia to the development of DMO and DR is well established (Vitale et al. 1995), but hypertension and hyperlipidaemia can also contribute to the pathogenesis of ocular disease (Chew et al. 1996; Klein et al. 1998). Intensive blood pressure (BP) control to maintain BP < 150/85 mmHg has been found to reduce the progression of DR in type 2 diabetes (UKPDS 1998b). Tight BP control led to a 34% reduction in the rate of progression of retinopathy by two or more levels of severity and a 47% reduction in the loss of ≥ 3 lines of vision on the ETDRS chart (UKPDS 1998b) and decreased the need for laser photocoagulation (Gillow et al. 1999). In the UKPDS, angiotensin-converting enzyme (ACE) inhibitors and beta-blockers had similar BP-lowering effects and were equally effective in reducing cardiovascular and microvascular end-points (Gillow et al. 1999). The risk of developing DMO increases with advancing age, lower socioeconomic status, longer duration of diabetes, elevated levels of glycosylated haemoglobin, increased systemic BP, elevated total serum cholesterol, and increased severity of DR.
The purpose of this review is to discuss our current understanding of the factors that contribute to the development and progression of DMO and to highlight those that are unknown. We will review changes that occur at the macro level, such as structural changes and changes in the pressure gradient forces involved, as well as the micro aspects by discussing the various molecules secreted, the pathways involved and the cellular disruptions that are implicated.
Several physiological mechanisms have been proposed to contribute to the pathogenesis of DMO. The exact mechanisms by which elevated glucose initiates the vascular disruption and results in the ultimate BRB breakdown in retinopathy remain poorly defined. Several pathways have been implicated, amongst which angiogenic, inflammatory and oxidative stress pathways are pivotal. Our working hypothesis is that during the development of DMO, several events occur as a result of increased levels of blood glucose. The initial insult results in endothelial damage and altered leucocyte function and recruitment (Morigi et al. 1998). The retinal blood vessels manifest morphological changes as elongation and dilation, along with thickening of the capillary basement membrane. Increase in shear stress and decoupling of endothelial cell tight junctions as a result of intracellular dysfunction allow the leakage of proteins into the surrounding tissue. This extravasation leads to an alteration of the extracellular matrix (ECM). During the process there is initiation of the inflammatory response and secretion of inflammatory mediators. This further facilitates cellular hypoxia, with the secretion of growth factors such as vascular endothelial growth factor (VEGF) and activation of oxidative stress reactants.
Vascular disruptions are a hallmark of DMO and DR and include abnormal vascular flow, disruptions in permeability and/or non-perfusion of capillaries (Ciulla et al. 2003). On a macrovascular level consistent with the clinical examination, DMO is characterized by increased retinal vascular permeability leading to the leakage of serum proteins and lipids into the intraretinal space (Knudsen et al. 2002). Hydrostatic pressure, oncotic pressure, shear stress and vasoregulation are contributing factors. From a molecular microvascular approach, hyperglycaemia leads to endothelial cellular dysfunction, hypoxia and the activation of numerous mediators from the angiogenic, inflammatory and oxidative stress pathways. The culmination of macro and micro factors, along with the interaction of feedback loops, leads to disruption of the BRB, loss of pericytes, thickening of the intraretinal space and macular oedema. Although the science is not yet clear on the exact order, interplay and autoregulation of all these pathways, the end result is an eventual impairment of vision.
Starling’s law and macular oedema
Macular oedema involves the accumulation of fluid in the outer retinal layers and within Henle’s layer (Massin et al. 2006). In the late 19th century, Starling suggested the basic concept of two forces that counteract each other: hydrostatic and oncotic pressures. Pressure differences are responsible for the movement of fluid between tissue beds and intravasculature spaces (Cunha-Vaz & Travassos 1984). When vessel integrity is impaired (especially in the BRB), there is leakage of protein and non-protein solutes into the retinal neural layers, which causes an increase in tissue oncotic pressure. This induces fluid movement into the tissue from the intravascular space, leading to macular oedema (Cunha-Vaz & Travassos 1984). Mechanical stress from systemic hypertension and impaired autoregulation can cause capillary hypertension, dilatation of vessels and breakdown of the BRB (Cunha-Vaz & Travassos 1984). Arteriolar dilatation decreases resistance to flow and lessens the degree of intra-arteriolar pressure, which consequently increases the hydrostatic pressure in the capillaries (Kristinsson et al. 1997). High capillary hydrostatic pressure, seen in systemic hypertension and associated with impaired autoregulation, has been reported in patients with DR (Kohner et al. 1995). Animal studies have shown that diabetic-like retinopathy is characterized by increased length and dilatation of arterial vessels (Robison et al. 1995). Kristinsson et al. (1997) reported an increase in the diameter of the temporal vasculature in patients recently diagnosed with DMO. These changes in vessel diameter further illustrate how increases in hydrostatic pressure can contribute to oedema formation. It has been shown that treatment of arterial hypertension can reduce macular oedema by reducing hydrostatic pressure (Stefánsson 2001). Glucocorticoids have been reported to reduce BRB permeability and to change retinal arterial and vein diameters in humans and animals. This can cause a reduction in the hydrostatic pressure gradient across the vessel walls and lead to a decrease in macular oedema (Vinten et al. 2007).
An increase in oxygen tension in the retina results in a compensatory vasoconstriction of the retinal vessels, which can decrease macular oedema by lowering hydrostatic pressure. Grunwald et al. (1984) found that patients with DR tended to have reduced oxygen reactivity. Focal laser in DMO is hypothesized to improve oxygen reactivity. This can effectively reduce growth factor formation, as well as improve retinal haemodynamics (Stefánsson 2006). Vitreous samples from patients undergoing vitrectomy demonstrated increases in oxygen tension within areas treated with laser compared with untreated areas (Stefánsson 2001).
Increases in shear stress over time can result in damage to endothelial cells and cause endothelial decoupling, which can result in outflow of fluids and oedema. Shear stress also contributes to increased production of nitric oxide, which leads to vasodilatation and can result in additional alterations in flow and an increase in hydrostatic pressure, as discussed previously (Kohner et al. 1995). Shear stress leads to an increase in ECM production by the endothelial cells and can contribute to capillary basement membrane thickening and loss of pericytes (Kohner et al. 1995). With the loss of pericytes and their contractile properties which regulate changes in vessel diameter, this damage can impair the autoregulation capabilities of the arterioles (Patel et al. 1992).
Increases in blood glucose levels in animal models were reported to increase retinal blood flow. Patient studies found reduced ocular blood flow in diabetes patients without DR and acute elevations in blood glucose resulted in increased retinal blood flow (Bursell et al. 1996). Grunwald et al. (1996) reported increased ocular blood flow in early DM measured in patients with relatively short duration of diabetes. Patients with mild to moderate proliferative DR demonstrated an increase in blood flow that correlated and has been associated with an increase in disease duration and severity (Feke et al. 1985; Patel et al. 1992; MacKinnon et al. 1997). Increased platelet aggregation, decreased red blood cell deformability and increased viscosity can occur within the capillary beds in diabetes (Patel et al. 1992). The increase in flow, together with the enhanced viscosity, results in an increase in the shear stress on the vessel walls and endothelial cells (Patel et al. 1992; Kohner et al. 1995).
Thickening of the basement membrane, decreases in capillary perfusion, impaired autoregulation and damaged endothelial cells promote capillary closure in this hyperglycaemic environment (Patel et al. 1992; Kohner et al. 1995). The resultant retinal ischaemia stimulates further vasodilatation, with secretion of growth factors, inflammatory mediators and oxidative species, which, in an attempt to correct homeostasis, adds to the pathology on a molecular level.
Hyperglycaemia-induced endothelial dysfunction and vascular damage
Healthy endothelium has an important role in vascular tone and structure as well as in the inhibition of platelet and leucocyte adhesion to vessel walls. Endothelial cells produce vasoconstriction and vasodilatation mediators and have a regulatory role in inflammation and in the production of inflammatory mediators such as the leucocyte adhesion molecule (LAM), the intracellular adhesion molecule (ICAM) and the vascular cell adhesion molecule (VCAM) (Hadi & Suwaidi 2007). Endothelial dysfunction and subsequent endothelial death can accelerate and weaken the intercellular junction, promote attachment of monocytes and leucocytes, produce capillary thrombosis and increase the transendothelial migration and production of adhesion proteins on endothelial cells (Rahman et al. 2007).
Hyperglycaemia and the resultant vascular damage are pivotal to this process in DMO. Glucose can activate the polyol pathway which leads to the accumulation of sorbitol and fructose. Glucose increases intra- and extracellular formation of advanced glycation endproducts (AGEs) which lead to the activation of the protein kinase C (PKC) pathway (Brownlee 2001, 2005). These AGEs can alter the function of intracellular proteins and can interact abnormally with ECM proteins and their integrin receptors on the cell wall. Activation of these oxidative stress reagents by hyperglycaemia, together with activation of the other pathways, leads to endothelial dysfunction and promotes vascular damage (Brownlee 2001, 2005). Hyperglycaemia has been shown to reduce the number of circulating and functioning endothelial progenitor cells, which have a role in the repair of impaired vessels (Chen et al. 2007). Although hyperglycaemia is a major contributor to endothelial dysfunction, adding insulin and lowering the glucose level does not easily reverse these events (Hadi & Suwaidi 2007).
Blood–retinal barrier impairment and cellular components
The BRB is composed of the retinal vasculature and the retinal pigment epithelium (RPE). Endothelial cells are responsible for maintaining the BRB and damage to them results in increased vascular permeability (Ciulla et al. 2003). The inner BRB, composed of retinal capillaries, has endothelial cells with tight junctions that are almost impermeable to protein transport and thus create an effective osmotic gradient (Cunha-Vaz & Travassos 1984). The outer BRB is formed by RPE cells which play an important role in the balance of fluids within the retina (Miyamoto et al. 2007). Anything that disrupts their integrity or function can also lead to oedema. In the early stages of DMO, breakdown of the inner BRB may occur, resulting in the accumulation of extracellular fluids within the macula (Ciulla et al. 2003). Tight junctions between the endothelial cells are important in maintaining the BRB by creating a selective barrier to control water and solute flux between adjacent cells (Felinski & Antonetti 2005). These tight junctions are formed by various transmembrane, scaffolding and signalling proteins, including claudins, occludins and zonula occludins (Erickson et al. 2007). Therefore, alterations in the BRB that occur early in the progression of DR are thought to lead directly to DMO.
Pericytes are an essential cellular component in the regulation of retinal capillary perfusion, endothelial cell proliferation, vessel stabilization and capillary blood flow (Hughes et al. 2006). Pericytes lie outside the BRB and their survival relies on signals from ECM proteins (Yang et al. 2007). Pericytes are sensitive to advanced AGEs, which are increased by high blood sugar levels. Damage to pericytes by hyperglycaemia can lead to abnormal autoregulation of retinal blood flow (Ciulla et al. 2002). Loss of retinal pericytes represents an early feature of DR and correlates with microaneurysm formation (Ciulla et al. 2002).
Basement membranes have a role not only in structure support, but also in the regulation of vascular permeability, cellular adhesion, proliferation and gene expression (Ishii et al. 1998). Basement membrane production, ECM with cellular proliferation (Hempel et al. 1997) and vascular permeability (Lynch et al. 1990) are all altered in diabetes, leading to thickening of the capillary basement membrane and increased deposition of ECM.
Vinten et al. (2007) found a linear correlation between changes in BRB permeability measured with vitreous fluorophotometry and retinal thickness calculated by optical coherence tomography (OCT). In DMO, the inner BRB is impaired and proteins leak from the blood vessels in the inner retina, increasing the oncotic pressure within the tissue. The external limiting membrane anterior to the RPE cells forms a barrier against which fluid accumulates. As fluid accumulates, RPE cells are overwhelmed and unable to clear the fluid, which results in subretinal serous retinal detachments (Soliman et al. 2007).
Vascular endothelial growth factor (VEGF)
Vascular endothelial growth factor mediates angiogenesis by promoting endothelial cell migration, proliferation and survival (Table 1). The VEGF family is made up of VEGF-A, VEGF-B, VEGF-C, VEGF-D, VEGF-E and placental growth factor (PlGF) (Takahashi & Shibuya 2005; Otrock et al. 2007). Amongst the various VEGF family members, VEGF-A, the most well studied, is a critical regulator of ocular angiogenesis and vascular permeability (Shams & Ianchulev 2006). It is involved in the pathogenesis of various ocular diseases, such as neovascular age-related macular degeneration, DMO and DR (Shams & Ianchulev 2006). As a result of alternative splicing, there are at least nine different VEGF-A isoforms, but some isoforms (e.g. VEGF-A165) play a more critical role in the pathogenesis of ocular disease (Usui et al. 2004).
Table 1. Effect of growth factors and inflammatory mediators on vascular permeability and/or leukostasis.
Phosphorylates of junctional proteins (cadherin and vinculin)
Increases expression of endothelin-1 (potent vasoconstrictor)
Increases collagen type IV and fibronectin, which thickens basement membrane
Matrix metalloproteinases (MMPs)
Digests collagen type IV, laminin, fibronectin and gelatin
Signals release of various growth factors
Increases number of apoptotic pericyte cells
Reduces PEDF levels
Induces pericyte migration, hypertrophy, and uncoupling from retinal microvessels
Increases free radical formation
Plays a role in vascular tone through TGF-β1, TNF-α, IGF-1, interleukins, AGE, ROS, and connective tissue growth factors
The actions of VEGF are mediated by the activation of two key cell membrane-bound tyrosine kinase receptors, VEGF receptors (VEGFRs) 1 and 2 (Fig. 1) (Takahashi & Shibuya 2005). VEGFR-1 often serves as a decoy receptor, whereas VEGFR-2 is the primary mediator of the mitogenic, angiogenic and vascular permeability effects of VEGF-A (Ferrara et al. 2003). VEGF and cellular-bound receptor binding activates two possible pathways, a mitogen-activating protein kinase (MAPK) signalling pathway or a calcium influx channel pathway; both have been proposed to increase vascular permeability (Bates & Curry 1997).
VEGFs possess pro-inflammatory properties that mediate chemotaxis and increase adhesion of leucocytes (Melder et al. 1996). VEGF has been found to stimulate expression of intracellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1) (Kim et al. 2001), thus activating the inflammatory cascade, initiating early diabetic retinal leucocyte adhesion and aiding the development of diabetic vasculopathy (Funatsu et al. 2005).
VEGF regulates and induces the formation of vesiculo–vacuolar organelles (VVOs) at sites of augmented vascular leakage (Dvorak & Feng 2001) (Fig. 2). These VVOs are grape-like clusters that occur at intervals in the cytoplasm of endothelial cells that influence permeability. They provide a transcytotic pathway by which soluble macromolecules can extravasate from leaky blood vessels (Dvorak & Feng 2001). VEGF can also induce structural rearrangement of VVOs, transforming them into transendothelial cell pores that allow the passage of even larger molecules out of cells (Feng et al. 1999). Furthermore, VEGF-A has also been shown to increase permeability through this same mechanism involving active transendothelial transport (Hofman et al. 2000).
VEGF and tight junctions
At tight junctions, claudin and occludin comprise the tight junction strands (Felinski & Antonetti 2005) (Fig. 3). They interact with each other at the extracellular surface and with the actin cytoskeleton through peripheral membrane proteins, such as zonula occludin-1 (ZO-1) (Felinski & Antonetti 2005). Occludin is the structural and functional component of tight junctions (Erickson et al. 2007). It controls signal transduction and forms a diffusion barrier by regulating the flux of ions and hydrophilic molecules through the paracellular pathway (Behzadian et al. 2003). The zonula occludins are intracellular proteins that organize the tight junction complex and link to microfilaments by binding to actin (Erickson et al. 2007). Tight junctions maintain the barrier to paracellular passage as well as being apical to the basolateral domain within endothelial cells with a distinct polarity (Behzadian et al. 2003).
Various stimuli affect the amount of occludin at tight junctions. VEGF can cause rapid post-translational modifications to tight junctions by stimulating occludin phosphorylation and by inducing tyrosine phosphorylation of zonula occludin (ZO-1) (Antonetti et al. 1999). In diabetes rat models, VEGF treatment reduced occludin content by half and simultaneously increased vascular permeability (Antonetti et al. 1998).
VEGF and cell–cell junctions
Endothelial cell–cell junctions are complex structures formed by different adhesive molecules that are anchored to the cytoskeleton (Dejana et al. 1999). The functional unit of a cell–cell adhesion is typically composed of cell adhesion molecules (CAMs) and cytoplasmic peripheral membrane proteins (Dejana et al. 1999). Vascular endothelial cadherin (VE-cadherin) interacts with the actin cytoskeleton via a family of catenins and vinculin (Angst et al. 2001; Bazzoni & Dejana 2004). Vinculin is a cytoplasmic peripheral membrane protein that links integral junctional adhesion molecules (e.g. integrin, cadherin) to the actin cytoskeleton (Werth et al. 1983). Stimulation with either VEGF-A165 or endothelial platelet-activating factor (PAF) can induce a rapid and transient phosphorylation of VE-cadherin, disorganization of the endothelial junction proteins and dissociation of the adherens junction, leading to gap formation and increased permeability (Fig. 3) (Esser et al. 1998; Kevil et al. 1998; Hudry-Clergeon et al. 2005).
Placental growth factor
Placental growth factor (PlGF) is a member of the VEGF family which also increases vascular permeability (Otrock et al. 2007). Originally discovered in the placenta, PlGF is associated with increased permeability of the RPE cells in vitro (Miyamoto et al. 2007). Increased PlGF levels in the vitreous fluid of rat eyes can induce RPE cell tight junction alterations and can subsequently lead to subretinal fluid accumulation and retinal oedema (Miyamoto et al. 2007). PlGF mainly activates VEGFR-1 (Otrock et al. 2007). Recent research by Miyamoto et al. (2007) has shown that both hypoxia and insulin induce upregulation of PlGF. Upregulation of PlGF and VEGFR-1 under hypoxic conditions leads to alterations in the RPE barrier (Miyamoto et al. 2007). High levels of PlGF have been reported in the vitreous of patients with proliferative DR (Khaliq et al. 1998; Mitamura et al. 2002).
Hepatocyte growth factor
Hepatocyte growth factor (HGF) is an endothelium-specific growth factor with highly potent mitogenic activity that plays a role in vascular permeability. It mediates retinal angiogenesis, induces VEGF expression and activates PKC (Cai et al. 2000). HGF and its receptors are expressed in retinal endothelial cells and their effect on vascular permeability appears to be mediated by phosphatidylinositol-3 (PI-3) kinase and MAPK (Clermont et al. 2006a). By reducing expression of claudin, occludin, CAMs and cadherin, HGF can lead to endothelial cell separation and increased leakage (Clermont et al. 2006a). In in vivo rat models, eyes receiving HGF therapy demonstrated increases in vitreous fluorescence similar to those induced by VEGF therapy (Clermont et al. 2006a). HGF expression is also upregulated under hypoxic conditions and its levels were increased in vitreous samples from patients with non-proliferative and proliferative DR (Patel et al. 2006).
There is evidence that inflammatory stimulation plays an important role in the pathogenesis of DMO and DR. Leucocytes possess a natural tendency to adhere to the vascular endothelium (leukostasis) and a capacity to generate toxic superoxide radicals and proteolytic enzymes (Miyamoto et al. 1999). Leukostasis can directly increase vascular permeability and damage endothelial cells by the release of free radicals, enzymes and cytokines (Schröder et al. 1991). Furthermore, inflammatory stimulation can increase occludin phosphorylation, thereby also increasing vascular permeability (Hirase et al. 2001). Preclinical and clinical models have demonstrated that leucocytes present in a hyperglycaemic environment are less deformable, become activated more readily and acquire the ability to adhere more tightly to endothelial cells (Miyamoto et al. 1998), thus leading to distal capillary non-perfusion (Schröder et al. 1991; Miyamoto et al. 1999). One study showed that diabetic vascular leakage and non-perfusion are temporally and spatially associated with retinal leukostasis in streptozotocin (glucose derivative)-induced diabetic rats (Miyamoto et al. 1999). Even the mild breakdown of the BRB that is demonstrated early in the onset of diabetes may in part reflect leukostasis (Vinores et al. 2007).
Not only do leucocytes play a role in the pathogenesis of DMO and DR, but platelets and erythrocytes can also be involved in this process (Ciulla et al. 2003). Capillary occlusions by cellular thrombi produce capillary dropout or degeneration associated with perivascular leucocytosis and focal areas of retinal ischaemia or hypoxia (Miyamoto et al. 1999). This retinal ischaemia is now the stimulus for further migration of inflammatory cells, formation of reactive oxygen species (ROS) and the release of angiogenic growth factors, such as VEGF (Aiello et al. 1994; Miller 1997). These angiogenic factors act not only as neovascular chemokines, but also as vascular trophic factors affecting tight junctions, fenestrations and increasing vascular permeability (Hofman et al. 2000).
Tumour necrosis factor-α
Tumour necrosis factor-α (TNF-α) is a pro-inflammatory cytokine that causes breakdown of the BRB (Table 1). It probably acts through the stimulation of leukostasis, which results in damage to endothelial cells (Penfold et al. 2002). The ability of VEGF to induce leukostasis is reduced in the absence of TNF-α, but the breakdown of the BRB is not reduced in the absence of TNF-α, indicating that BRB breakdown is not solely the result of leukostasis (Penfold et al. 2002). TNF-α activity is associated with increased expression of VEGF and ICAM-1 in choroidal endothelial cells (Penfold et al. 2002). VEGF can induce TNF-α and, in turn, TNF-α can also increase VEGF levels (Penfold et al. 2002). TNF-α has been found to be elevated in the vitreous of patients with proliferative DR and in the plasma of diabetes patients (Ben-Mahmud et al. 2006).
Transforming growth factor-β
Transforming growth factor-β (TGF-β) is a cytokine that is involved in endothelial cell proliferation and adhesion, and deposition of ECM (Harada et al. 2006). In vitro TGF-β increases the permeability of retinal capillary endothelial cells in a process thought to be mediated by the expression of matrix metalloproteinase-9 (MMP-9) (Behzadian et al. 2001). TGF-β has been found to induce expression of VEGF by fibroblasts (Harada et al. 2006). TGF-β is upregulated with high glucose levels in retinal endothelial cells (Harada et al. 2006). TGF-β plasma levels were increased in type 2 DM patients compared with controls and were associated with disease states such as retinopathy and nephropathy (Pfeiffer et al. 1996). By contrast, reduced vitreous levels of active TGF-β in patients with proliferative DR are postulated to demonstrate some anti-angiogenic effects of TGF-β in the normal vitreous (Pfeiffer et al. 1997). Its role, if any, in DMO is not fully understood.
Intercellular adhesion molecule-1
The ICAM-1 is an intracellular adhesion molecule necessary for the adhesion of leucocytes to capillary endothelium (Meleth et al. 2005). The expression of ICAM-1 is increased by VEGF (Lu et al. 1999) and AGEs. Activation of the AGE receptors induces oxidative stress, which activates nuclear factor (NF)-κ and various other proteins, including ICAM-1 (Kaji et al. 2007). Blocking ICAM-1 reduced the breakdown of the BRB and the diabetes-induced leukostasis in a diabetes rat model (Miyamoto et al. 1999). Immunohistochemistry in a donor eye with severe non-proliferative DR and DMO demonstrated increased ICAM-1 levels compared with a normal donor eye (Meleth et al. 2005).
Interleukin-6 (IL-6) has been shown to cause enhanced expression of VEGF and to increase vascular permeability (Funatsu et al. 2002b; Funatsu & Yamashita 2003). IL-6 acts as a mediator of inflammation, a hepatocyte stimulating factor, a regulator of physiological molecules, a neuroprotective agent and a B-cell stimulating factor (Harada et al. 2006). IL-6 has been found to increase endothelial permeability in human umbilical vein endothelial cells through the redistribution of the tight junction proteins, ZO-1 and cytoskeletal actin, increased cell contractions, and disorganization of the intercellular borders (Desai et al. 2002). The inhibition of PKC attenuated this response (Desai et al. 2002). Moreover, hypoxic endothelial cells are known to produce IL-6 (Desai et al. 2002). Vitreous levels of VEGF and IL-6 were found to be increased in patients with DMO and exhibited a positive correlation to the degree of DMO disease severity (Funatsu et al. 2002b, 2003).
Diabetes has been shown to increase superoxide, nitric oxide (NO) and peroxynitrite formation and VEGF expression, leading to breakdown of the BRB and increased vascular permeability (Marumo et al. 1999; El-Remessy et al. 2003). Hyperglycaemia stimulates superoxide formation and increases intracellular calcium to activate endothelial NO synthase (eNOS) (Caldwell et al. 2005). As activated eNOS generates NO, superoxide and NO combine to form peroxynitrite, a highly reactive oxidant. Peroxynitrite not only oxidizes cellular components, but increases the formation of additional superoxide and peroxynitrite molecules (Caldwell et al. 2005). Generation of superoxide and peroxynitrite can lead to a self-perpetuating cycle of free radical formation. eNOS activation and NO formation are probably required for the early phase of the VEGF effect (Behzadian et al. 2003). VEGF stimulates VEGFR-2, which has been shown to transduce the mitogenic signal and the formation of NO and superoxide anions (Caldwell et al. 2005). Furthermore, ROS can increase diacylglycerol (DAG) formation, activate PKC and aldose reductase and increase AGE formation, as well as activate the transcription factor NF-κβ to increase vascular permeability (Nishikawa et al. 2000).
Diabetes-related products, including AGE and ROS, as well as direct effects of hyperglycaemia, are involved in an increase in matrix metalloproteinases (MMPs) in animals with diabetes (Giebel et al. 2005) (Table 1). The MMPs are a family of proteinases that degrade structural components of the ECM as well as other non-extracellular proteins (Navaratna et al. 2007). These proteinases are critical in extracellular remodelling and repair, and in angiogenesis. MMP-9 and MMP-2 are gelatinases located in endothelial cells that are capable of digesting the basement membrane proteins (Behzadian et al. 2001). MMPs play a role in vascular endothelial barrier dysfunction, not only through degradation of the ECM, but also through the release of growth factors (Behzadian et al. 2003), thereby contributing to vascular permeability (Navaratna et al. 2007). In vitro models have shown that activation of MMP-2 leads to increased numbers of apoptotic pericyte cells (Yang et al. 2007) and MMP-mediated proteolytic activities reduce the levels of pigment epithelium-derived factor (PEDF) (Notari et al. 2005). MMP-9 is often produced in conjunction with endothelial cell activation (Behzadian et al. 2001) or AGE stimulation (Navaratna et al. 2007).
MMPs are upregulated in early DR during times of increased BRB permeability. In rat models, streptozotocin decreased endothelial cadherin levels and increased permeability that was markedly blocked by MMP inhibitors (Navaratna et al. 2007). This suggests that during the development of DR, hyperglycaemia creates an environment that facilitates increased microvascular permeability associated with the upregulation of MMP expression and the loss of VE-cadherin (Navaratna et al. 2007). MMPs degrade the tight junction protein occludin and MMP-9 and MMP-2 expression was found to be increased in isolated endothelial cells under high glucose exposure (Giebel et al. 2005).
Protein kinase C
Protein kinase C is a family of enzymes composed of at least 12 members, although not all forms are expressed in all cell types (Mellor & Parker 1998). PKC increases vascular permeability through various mechanisms. PKC activation is involved in the induction of VEGF (Williams et al. 1997) and TGF-β (Koya et al. 1997). PKCβ1 overexpression augments phorbol ester-induced increase in endothelial permeability (Nagpala et al. 1996). PKC plays a role in the phosphorylation of junctional proteins (e.g. cadherin, vinculin) associated with cell–cell and cell–substrate junctions, thereby increasing vascular permeability (Werth et al. 1983; Vuong et al. 1998; Dejana et al. 1999). PKC can reduce blood flow, either by a direct vasoconstriction effect or by an increase in the expression of a potent vasoconstrictor, endothelin-1 (Ishii et al. 1998). Levels of PKC are increased in vascular endothelial cells after exposure to oxidative stress (Ciulla et al. 2003).
PKC has been implicated in the development of microvascular damage and diabetic changes in the eye (Ciulla et al. 2003). PKC and TGF-β increase messenger ribonucleic acid (mRNA) expression of collagen type IV and fibronectin, which is associated with a thickened capillary basement membrane, a classic structural abnormality in diabetes (Ishii et al. 1998). Nevertheless, results of treatment with a PKC selective inhibitor in DMO were disappointing, with no statistically significant effect compared with placebo. These results question the exact role of PKC in DMO (Protein Kinase C β Inhibitor Diabetic Macular Eedema Study Group 2007).
Wistrand et al. (1986) demonstrated the presence of carbonic anhydrase (CA) types 1 and 2 in the human eye. CA type 1 is found in the endothelium of the choroids and CA type 2 is found in Müller cells and within the RPE. In order to study the role of CA on vascular permeability, Gao et al. (2007) injected purified human CA type 1 into rat vitreous and demonstrated increases in retinal fluorescein leakage, which was inhibited after co-injection of acetazolamide. The magnitude of the effect of CA type 1 on vascular permeability was found to be similar to that of VEGF and additive with the effects of VEGF (Gao et al. 2007).
Various CA inhibitors are being used for the treatment of other causes of macular oedema. In cystoid macular oedema, CA inhibitors are thought to change the polarity of ionic transport in the RPE by blocking CA or by inhibiting γ-glutamyltransferase (γ-GT) (Marmor 1990). This is thought to promote the influx of fluid from the intraretinal space through the retinal epithelium, thereby reducing the extent of oedema (Marmor 1990). Gao et al. (2007) found increased vitreous levels of CA types 1 and 2 in patients with moderate to severe non-proliferative DR and proliferative DR. However, DMO patients do not respond well to CA inhibitors. This failure to respond is hypothesized to reflect a dysfunction of the RPE cells or high extravasation from damaged vessels which cannot be cleared by healthy RPE cells (Marmor 1990).
Systemic hypertension is a known risk factor for both the development and progression of DR (Norgarard et al. 1991; UKPDS 1998b). Angiotensin-II has been shown to stimulate VEGF overexpression and VEGF-associated hyperpermeability retinopathy characteristic of diabetes (Gilbert et al. 2000). Angiotensin-II induces pericyte migration and hypertrophy, pericyte uncoupling from retinal microvessels, and increased free radical formation (Wilkinson-Berka 2006). Angiotensin-II plays a role in smooth muscle cell regulation of structural support and regulation of vascular tone (Wilkinson-Berka 2006). Diabetes rat models have been used to demonstrate the presence of angiotensin converting enzyme (ACE) on the retinal vessel wall (Okada et al. 2001).
ACE inhibitors not only lower systemic BP, but have been postulated to directly affect retinal blood flow via vascular endothelial ACE receptors (Funatsu & Yamashita 2003). The activation of angiotensin-II type 1 (AT1) receptors expressed on retinal endothelial cells and retinal and choroidal pericytes has been implicated in the pathophysiology of microvascular abnormalities observed in DR (Funatsu & Yamashita 2003). Vitreous analysis from patients with DMO revealed increased levels of angiotensin-II and VEGF that correlated to hyperfluorescent DMO (Funatsu et al. 2002a). Studies of systemic ACE inhibitors in normotensive diabetes subjects have also produced evidence of beneficial effects on retinopathy (Gillow et al. 1999; Clermont et al. 2006b). The Diabetic Retinopathy Candesartan Trails (DIRECT) programme has enrolled 5231 patients with type 1 and 2 diabetes to examine the incidence, progression and prevention of DR when blocking AT1 receptors with the oral ARB candesartan. In type 1 diabetes, Candesartan decreased incidence of diabetic retinopathy but failed to decrease progression rates. In type 2 diabetes, candesartan did not affect the incidence of proliferative diabetic retinopathy or clinically significant macular oedema. The results of the study suggest that Candesartan has the potential to improve outcome only in patients with early stages of diabetic retinopathy (Chaturvedi et al. 2008, Sjoolie et al. 2008). These results question the role of ACE inhibitor in DMO treatment beyond their blood pressure lowering effect.
Diabetic macular oedema is a major cause for decreased visual acuity (VA) in diabetes patients. It has significant economic and personal impact on the younger working population. Large, randomized studies, such as the DCCT, UKPDS and WESDR, have shown that systemic improvement in glucose and BP control can reduce the incidence of DMO and DR and the occurrence of decreased VA from macular oedema. The standard treatments for DR, including laser photocoagulation, are aimed at stabilizing vision once DMO and vision loss have occurred. Because of the laser’s destructive qualities, and with emerging understanding of the pathophysiology underlying the oedema, therapy is now being targeted directly at chemical mediators, such as growth factors and inflammatory components. Despite our manipulation and inhibition of these components with anti-VEGF agents and steroids, oedema is only lessened or temporarily improved at best. In some patients VA continues to deteriorate despite these treatment strategies. Investigation into the longterm efficiency and adverse event profiles of these drugs will require the completion of large, randomized, controlled trials. Currently, it seems that treatment modalities are only targeted at one or limited components responsible for DMO formation. These treatments target VEGF and inflammation once macular oedema has occurred. It is important to develop treatment modalities that can target the underlying causes of DMO, perhaps even prior to visual loss. Diabetic macular oedema involves multiple pathways that can contribute to the disease. It seems that by targeting only one pathway, we fail to offer sufficient beneficial effect to patients. The relative contribution of the different pathways may also differ between patients and the optimal treatment strategy may involve multi-drug therapies (Aiello 2008).
As elucidated in this paper, the underlying pathophysiology of DMO is complicated and involves numerous interconnected pathways regulated by multi-factorial feedback loops. As we have outlined herein, we hypothesize that increased levels of blood glucose cause morphological changes in blood vessels. Increased shear stress associated with the mechanical disruption of blood vessels leads to the decoupling of endothelial cells. This, in turn, causes leakage of fluids into the surrounding tissue. The process initiates an inflammatory response and the secretion of inflammatory mediators. This further facilitates the secretion of growth factors such as VEGF. Various regulatory mediators play numerous roles depending on timing with respect to the underlying diabetic disease condition. Early in the disease process, DMO may be driven more by hyperglycaemia, AGE and ischaemic injury with ROS, whereas other growth factors such as VEGF or PlGF play a greater role once proliferative DR develops. Emerging research and greater scientific knowledge will improve our understanding of the mechanisms responsible for the development of DMO and DR and will allow us to target new and better treatment options more efficiently with the aim of improving patient outcomes.
This study was partly supported by an unrestricted research grant from Research to Prevent Blindness, New York, NY. BW is employed as a medical director at Pfizer Inc., New York, NY.