Diabetic macular oedema (DMO) is a major cause of blindness worldwide and its incidence is increasing due to the greater numbers of people with diabetes.1–4 The disease is a result of changes in retinal capillaries that leak and changes in the retinal pigment epithelium causing macular thickening, oedema and poor eyesight, leading to vision loss if left untreated.5,6 DMO is particularly serious due to its onset affecting a young/middleaged age group and disabling them during their productive working lives.2 Laser photocoagulation treatments have been the mainstay of treatment over the past few decades, but this can be destructive and destroys photoreceptors around the retinal areas affected. Steroids given intravitreally are also used to treat DMO, but these can have undesirable side effects, such as cataracts and increased intraocular pressure (IOP).7 Early DMO is associated with substantially increased vascular endothelial growth factor (VEGF) secretion, and treatments that block VEGF, particularly ranibizumab, have shown significant improvements vision in patients with DMO compared with laser treatment and shows good long-term safety.8,9 This medication is proving to be a valuable option in DMO treatment. This review will outline the pathology and burden of DMO, its associations with both types I and II diabetes and different approaches to treatment.
The Prevalence and Burden of Diabetic Macular Oedema
Prevalence estimates of DMO from numerous studies vary widely, largely as a result of differing populations included and different definitions of DMO. Epidemiological data on DMO are sparse and reported incidence varies between countries.2 A systematic review of 359 studies found the prevalence among people with diabetes to be in the range 6–16 % for type I diabetes, 0.6–26.1 % for type II and 4.3–10.0 % for mixed cohorts.3 A meta-analysis of 35 studies conducted in the US and Europe found the age-standardised prevalence of DMO in populations aged 20–79 years to be 6.81 %.4 The worldwide prevalence of DMO has been estimated to be 21 million and, as a result of the projected increase of diabetes due to ageing populations, sedentary lifestyles and dietary changes, this population is expected to rise to 100 million by 2030.10 DMO generally occurs at a higher frequency in type I than type II diabetes (14.25 % versus 5.57 %)4 but 70– 90 % of diabetes cases are type II.11 DMO prevalence also increases with duration of diabetes as shown in Figure 1: within 5 years of onset 0 % and 3 % of type I and type 2 diabetes, respectively, have DMO whereas after 20 years this rises to 29 % and 28 %.3
In developed countries, DMO is the leading cause of blindness in the working population in individuals aged 20–70 years although it occurs mostly in individuals aged 40–69 years, with similar frequency in males and females.2 It is associated with dyslipidaemia, hyperglycaemia and hypertension.12 Between 30–35 % of cases spontaneously resolve within 6 months;13,14 however, the ETDRS (sponsored by the US National Eye Institute and started in 1979) (see Box for definitions of study name acronyms) has shown that without treatment, 30 % of eyes with DMO develop visual loss of three lines or more at 3 years.15 It has consequently been described as a costly disease that has a negative effect on patients’ health-related quality of life.1
Pathophysiology of Diabetic Macular Oedema
DMO is a chronic disease and a frequent complication of diabetic retinopathy. The pathophysiology of DMO is not fully understood but increases in various factors that occur in diabetes are believed to contribute to several interrelated inflammatory processes. Chronic hyperglycaemia is a feature of diabetes that results in increased levels of free radicals, advanced glycation end product (AGE) proteins, protein kinase C (PKC) and other inflammatory factors such as interleukin 6.6,16 Hyperglycaemia also stimulates the secretion of various cytokines including insulin like growth factor-1: the combination of these factors increases the expression of VEGF in retinal pigment epithelial cells.17 Other factors such as metalloproteases, pigment epithelium-derived factor (PEDF), angiotensin II, basic fibroblast growth factor (b-FGF) and platelet-derived growth factor (PDGF) are also believed to be involved in the pathogenesis of DMO by increasing vascular permeability and disrupting the blood– retina barrier.18,19 This leads to accumulation of excess extracellular fluid in the macula (MO) and consequent reduction in visual acuity (VA). In addition, hyperglycaemia can also directly increase PKC and angiotensin II, which stimulates endothelin synthesis leading to vasoconstriction and worsening of hypoxia. These processes are often reversible during early disease, but prolonged oedema causes irreversible vision loss.
Studies using optical coherence tomography (OCT) have revealed the structural changes and damage that occurs at the retina in patients with DMO.20 There is generally an increase in central retinal thickness, which is moderately correlated with the level of VA. In addition, cystoid spaces, sponge-like retinal swelling, hyper-reflective foci, photoreceptor damage and serious retinal detachment have been observed and these effects worsen with time.
DMO occurs as focal or diffuse types according to the pattern of macular leakage observable on a fluorescein angiogram and it is important to classify which type is present prior to treatment.21,22 In focal DMO, leakage of intravascular liquid into interstitial spaces due to vasopermeability and microaneurysms is visible as discrete points of retinal hyper-fluorescence. These points of leakage are frequently seen within circular deposits of hard exudates. Where there are multiple microaneurysms, multifocal MO may occur and this can be confused with diffuse MO. Diffuse DMO is characterised by areas of diffuse leakage that are visible as a result of any or a combination of the following factors: intra-retinal leakage from a dilated capillary bed, leakage from intra-retinal microvascular abnormalities and leakage from arterioles and venules without foci of leaking microaneurysms.23 A further subtype, cystoid diabetic MO (CMO), occurs as a result of widespread degradation of the inner blood retinal barrier with fluid accumulation in the outer plexiform layer.24
Screening for Diabetic Retinopathy and Diabetic Macular Oedema
In patients with diabetes, screening for DMO and determination of the risk of DMO is essential since early detection and initiation of treatment is vital to prevent irreversible vision loss.12 Initiatives for screening and ensuring rapid treatment of DMO are not given priority in many parts of the world, often because other types of preventable blindness receive more attention or are considered to be more important.25 Some screening programmes fail to identify a large proportion of high-risk patients.5 Risk factors for DMO that should be monitored include duration of disease, severity of hyperglycaemia, hypertension and hyperlipidaemia, insulin resistance, obesity, urbanisation, access to healthcare systems, genetic susceptibility and epigenetic factors.25 The importance of controlling hyperglycaemia was demonstrated by the Diabetes Control and Complications Trial (DCCT) in which maintaining glycated haemoglobin (HbA1C) below 7 % resulted in marked decreases in development and progression of diabetic retinopathy in type I diabetes.26 In addition, some racial groups such as South Asian, African, Latin American and some indigenous tribal groups show a higher prevalence of DMO compared with white populations, making screening efforts more urgent.25
Success of screening has been demonstrated in the US where rates of non-proliferative diabetic retinopathy are declining as a result of these programmes.25 In the UK, the NHS Diabetic Eye Screening Programme provides annual digital fundus photography for all patients with diabetes over 12 years of age.27 This programme aims to detect all forms of diabetic retinopathy and maculopathy and has achieved a reported take-up of 79 %. In cases where disease is detected, the patient is referred to a specialist eye unit for further assessment and treatment within a pre-specified timeframe. Screening for diabetic retinopathy and DMO is becoming more widespread and accessible with the use of teleretinal scanning. Several studies have found that using cameras remote from treatment centres is an efficient and reliable means of screening diabetic populations, particularly those in remote locations.28,29
Photocoagulation as a Treatment for Diabetic Macular Oedema
Laser photocoagulation been used for many years in the treatment of DMO and has proved effective in stabilising visual function reducing the rate of vision loss. The ETDRS group showed that laser treatment halved the 3-year risk of moderate visual loss (a decrease in VA score of 15 or more letters or moving from 20/20 to 20/40 in the optotypes scale) a reduction from 24 % in untreated eyes to 12 % in treated eyes.15
Laser treatment of DMO is divided into focal and grid types.13,30 Focal laser treatment is suitable for specific points of leakage such as microaneurysms or intra-retinal microvascular abnormalities that are between 500 and 3,000 μm from the centre of the macula. The intensity of laser treatment depends on the size of the lesion treated. The laser exposure of 0.05–0.1 second duration results in burns of 50–100 μm of moderate intensity. Effective treatment is indicated by the appearance of whitening or darkening of focal lesions. Grid laser treatment covers a wider area for a duration of 0.05–0.5 second with a spot size of 50–200 μm, to achieve mild retinal pigment epithelium whitening. The power is adjusted to prevent the burns from spreading to more than 200 μm in diameter. The size of the grid can be up to two disc diameters (3,000 μm) from the centre of the macula.
It is not fully understood how laser photocoagulation improves the symptoms of DMO and it is not clear what effects this treatment has on macular VEGF levels. It has been proposed that oxygen-consuming peripheral-retina photoreceptors are destroyed by the laser resulting in greater oxygen in the macular area photoreceptors. An alternative explanation is that oxygen diffusion can occur through the laser scars to the inner retina resulting in autoregulatory vasoconstriction that may decrease fluid leakage and swelling thus improving DMO. It has also been suggested that biological activities are stimulated in the areas around the burned region that did not receive a lethal dose of laser radiation.31
Despite some efficacy in DMO, laser photocoagulation can cause scarring and progressive atrophy in surrounding tissue leading to further oedema and vision loss at a later stage. In such areas the possibility of retreatment is limited. Since 1990, micropulsed laser treatment has emerged. This uses a longer wavelength (810 nm) and longer off-times during treatment resulting in less heat generation at the retina and less damage to surrounding tissues. Subthreshold photocoagulation has been developed using this technique that causes little or no damage on the surrounding retina.32,33 Laser surgery has been the ‘gold standard’ treatment for DMO and is widely used. In recent years, however, there has been a move away from laser treatments towards the use of medications although combination treatments of both laser photocoagulation and medications have been shown to be effective in various clinical trials.34
Surgical Approaches for Diabetic Macular Oedema
Vitrectomy involves the surgical removal of all or part of the vitreous humour, which may have become fogged with blood or exudate (including AGE proteins), and is often carried out with peeling of the internal limiting membrane (ILM).35–37 Studies show that vitrectomy is useful only in cases of vitreomacular traction,38 but results have been variable and inconclusive. A study on 40 eyes showed that vitrectomy with or without ILM did not improve the eyesight in patients with type II diabetes and DMO and no vitreoretinal traction. Another study conducted by the DRCRNet on 87 eyes with DMO showed that 6 months after vitrectomy 68 % had at least a 50 % reduction in central field thickness.39 VA improved by ≥10 letters in 38 % and deteriorated by ≥10 letters in 22 %. It was concluded that 28–49 % of eyes with characteristics similar to those included in this study are likely to have improvement of VA, whereas 13–31 % are likely to have worsening. Some animal model work has indicated that there may be greater clearance of exogenously applied VEGF after vitrectomy40 but this needs elucidation as a mechanism in laser treatment of DMO.
The value of vitrectomy with or without ILM peeling was further investigated in a Cochrane analysis of four clinical studies conducted in Europe and Hong Kong (n=47, 80, 49 and 141).41 This analysis found there was no difference in acuity of distance vision at 6 or 12 months after surgery between patients receiving or not receiving ILM peeling. Various complications may occur following vitrectomy including cataract, retinal detachment, epiretinal membrane, glaucoma and vitreous haemorrhage.42 The use of vitrectomy therefore tends to be limited to tractional or taut DMO for which laser treatment is inappropriate.
Diabetic Macular Oedema Steroid Treatments
Steroid treatments have been successfully used intravitreally, in the treatment of DMO for many years due to their anti-angiogenic, antioedematous, anti-inflammatory, anti-apoptotic and anti-proliferative effects.7 Steroids may reduce the concentration of inflammatory cytokines and growth factors, and increase vascular permeability. Systemic or topical application often fails to achieve intraocular therapeutic concentrations so doses need to be high but this can result in severe systemic side effects. Intravitreal injections can be effective but can cause cataracts and elevated IOP. Intravitreal steroid implants, however, can be used as an alternative to frequent injections.7
The evidence supporting steroid use in DMO, however, is variable. Example data come from a US study of 840 eyes that were treated with intravitreal triamcinolone 1 mg or 4 mg or focal/grid photocoagulation.43 Three years after treatment, the change in VA letter score from baseline was +5 in the laser group and 0 in each triamcinolone group. The cumulative probabilities of cataract surgery in these groups were 46 %, and 83 % and 31 %, respectively. These results indicated no long-term benefit of triamcinolone treatment over photocoagulation.
Steroid treatments for DMO have been more successful when administered as intravitreal implants. Several small European studies have reported favourable findings of dexamethasone implants in terms of improved eyesight over 6 months after treatment. Improvements were seen in ETDRS vales (18.80 at baseline changed to 26.15 after 1 month; p=0.04) and the central macular thickness (518.80 μm at baseline to 412.75 μm, at day 3 and 292.0 after 1 month; p<0.0001).44 These implants have also been shown to provide significant improvements in best-corrected VA (BCVA) and central macular thickness in patients with persistent or refractory DMO 1 month after injection.45 Another steroid used in DMO treatment is fluocinolone acetonide given as intravitreal injections or implants.46 Fluocinolone acetonide implants have shown efficacy significant improvements in VA as shown in the FAME study (n=953),47 but have been associated with cataract formation and elevated intraocular pressure. The fluocinolone acetonide insert has been approved for use in Europe48 but it is generally reserved for cases in which first-line therapy has failed.
The exact role of steroids for diabetic retinopathy and MO remains to be fully elucidated: their reported efficacy is variable but they continue to have a valuable role in treatment. Steroids are often used in conjunction with laser photocoagulation treatment and have been reported to prevent further vision loss.49 In addition, steroid treatments have been used as adjunct or comparator treatments to anti-VEGF treatments in multiple clinical trials50,51 and in some reports the combination has an improved efficacy over the anti-VEGF drug used alone.
Anti-vascular Endothelial Growth Factor Treatments in Diabetic Macular Oedema
There are several anti-VEGF agents that have been used in the treatment of DMO but ranibizumab (Lucentis®, Novartis) is the only one that has been approved by the US Food and Drug administration (FDA) and European Medicines Association (EMA) for use in this indication. Ranibizumab is approved as a 0.3 mg intravitreal monthly dose, but in the EU it is approved as a 0.5 mg intravitreal monthly dose. The use of this drug in DMO is supported by an extensive series of clinical trials that have included over 2,000 patients and these are reviewed in Table 1 – extensions of these studies are ongoing.
Several of the major studies (RESOLVE, RESTORE, REVEAL, DRCRNet52–56) compared ranibizumab with laser treatment or included arms in which patients were treated with both ranibizumab and laser. In each case, ranibizumab improved VA (BCVA letter score) and in some studies decreased central retinal thickness to a greater extent than laser treatment alone. In the DRCRNet study, both prompt or deferred laser treatment in combination with ranibizumab treatment, provided greater mean changes in VA at 2 years compared with prompt laser treatment alone or prompt laser treatment and triamcinolone (see Figures 2 and 3).52 The mean change in VA letter score from baseline was 3.7 letters greater in the ranibizumab with prompt laser group, 5.8 letters greater in the ranibizumab with deferred laser treatment and 1.5 letters worse in the triamcinolone plus prompt laser group. There was also a greater reduction in central subfield retinal thickness (measured by OCT) after 2 years, the percentages of eyes with central subfield thickness ≥250 μm were: 43 % in the ranibizumab + prompt laser group, 42 % in the ranibizumab + deferred laser group, 52 % in the triamcinolone + prompt laser group and 59 % in the prompt laser alone group.
In the RIDE and RISE studies, after 24 months ,18.1 % of sham-treated patients gained ≥15 letters versus 44.8 % of 0.3 mg (p<0.0001) and 39.2 % of 0.5 mg ranibizumab-treated patients (p<0.001).57 Ranibizumab-treated patients also showed significant improvements in MO (measured by OCT) and needed significantly fewer macular laser procedures. More recently, 3-year long-term results from the DRCRNet study and 2-year results from the RESTORE study showed long-term maintenance of visual improvements.56,58 The 3-year data show that long-term ranibizumab is well tolerated with no retinal toxicity up to 3 years. Longer term follow-up in this trial is in progress.
The RESTORE study has also shown significant improvements in quality of life with ranibizumab compared with laser alone and improvements in National Eye Institute Visual Functioning Questionnaire-25 (NEI VFQ-25) composite score. Analysis of the RESTORE trial 12-month data suggests that ranibizumab is cost-effective relative to laser treatment but the economics of combined therapy is less certain.59
The ongoing RETAIN study is comparing two ‘treat and extend’ (TE) regimens of ranibizumab 0.5 mg (with or without laser photocoagulation) and a further group receiving the drug as needed (pro re nata [PRN]) in the treatment of 372 patients with DMO.60 The advantage of TE regimens is that patients can be treated at any of their follow-up visits but the interval can be increased until the next visit. With PRN dosing, patients can be treated according to individual disease activity, but with fixed monthly monitoring schedules. Interim results show no differences in outcomes between PRN and TE groups. There was a small increase in the number of injections and significantly fewer visits in the TE groups.
In these trials, ranibizumab showed a favourable safety profile and patient tolerability in the treatment of DMO and that adverse events were mainly confined to ocular symptoms. In the DRCRNet study, retinal detachment was reported in 1 % of patients compared with <1 % in those treated with laser alone and ocular vascular events were reported in 1 % in both these groups compared with 2 % for laser with triamcinolone.52 Endophthalmitis is also less common adverse event associated with ranibizumab although this was absent in the RESTORE trial54,58 but reported in 1 % of ranibizumab-treated patients in the DRCRNet trial52 and in 1.6 % in the RIDE and RISE trials.57 The DRCRNet study also showed that ranibizumab does not increase the incidence of cardiovascular events compared with sham treatment or triamcinolone (non-fatal myocardial infarction, non-fatal cerebrovascular accident or vascular death).52 Other systemic effects were either not reported ranibizumab or no more frequent than those occurring in sham- or laser-treated patients.52,54
Bevacizumab (Avastin®, Genentech) has also been used in DMO treatment but is not approved for this indication and treatment is therefore off label. Various medium and smaller studies have shown improvement in VA and retinal thickness compared with laser treatment alone (see Table 2).61–63 The data supporting bevacizumab in DMO, however, are generally of less high quality than that supporting ranibizumab. Bevacizumab is associated with various safety concerns including necrotising fasciitis, haemorrhage, gut perforation and wound-healing problems and serious systemic side effects.64 Longterm safety and tolerability data are needed to support its use in DMO.
Aflibercept (Eylea®, Bayer) has also been used for DMO treatment, but it is also not yet approved for this indication (see Table 2). Different monthly dosing schedules of 0.5 mg and 2.0 mg aflibercept showed notable efficacy compared with laser treatment alone in the DA VINCI study.65 AEs included: conjunctival haemorrhage, eye pain, ocular hyperaemia and increased IOP systemic AEs: hypertension, nausea and congestive heart failure. The ongoing VIVID and VISTA studies comparing aflibercept with laser treatment in DMO have shown marked improvement in BCVA for aflibercept after 1 year of treatment.66,67 Long-term efficacy and safety data from these trials will provide better evidence supporting the use of this treatment in DMO.
Pegaptanib sodium (Macugen®, Eyetech Pharmaceuticals), a treatment approved for neovascular (wet) age-related macular degeneration, is not approved for DMO but has recently been studied in a phase II trial for this indication.68 Patients who were treated for 36 weeks with 0.3, 1 and 3 mg pegaptanib (n=172) showed significant improvements in VA and significant decreases in mean central retinal thickness compared with sham injections. Laser photocoagulation was required in 25 % of the 0.3 mg group compared with 48 % of the sham injection group (p=0.04) and the injections were well tolerated.
With the substantial rise in both type I and type II diabetes in most populations, the incidence and prevalence of DMO is increasing but more effective therapies to tackle it are evolving. The established laser photocoagulation approach is being replaced or augmented with newer treatments. Steroid treatments can be given as implants and pulsed or sub-threshold laser treatments are less damaging to retinal tissues. The principal change in DMO therapy, however, is the use of anti-VEGF treatments, in particular, ranibizumab. An extensive body of clinical trial evidence now supports ranibizumab; follow-up results now extend to 3-years and show sustained efficacy and tolerability. Experience with this drug is also increasing with clinical use. Other anti-VEGF agents, bevacizumab, aflibercept and pegaptanib have also shown efficacy in DMO, but the data supporting these are, as yet, limited.
DMO is an urgent problem and a substantial cause of vision loss worldwide. The anti-VEGF treatments, in particular, ranibizumab are increasingly important as a means of controlling it and are more effective than other medications currently available. Ranibizumab has been shown to be cost-effective and improves quality of life relative to laser treatment: it is therefore likely to have a central role in DMO therapy for the foreseeable future.