Graefe's Archive for Clinical and Experimental Ophthalmology

, Volume 248, Issue 7, pp 915–930

A review of clinical trials of anti-VEGF agents for diabetic retinopathy

Authors

    • Cole Eye InstituteCleveland Clinic
  • Andrew P. Schachat
    • Cole Eye InstituteCleveland Clinic
Review Article

DOI: 10.1007/s00417-010-1315-z

Cite this article as:
Nicholson, B.P. & Schachat, A.P. Graefes Arch Clin Exp Ophthalmol (2010) 248: 915. doi:10.1007/s00417-010-1315-z
  • 1.6k Views

Abstract

Background

Diabetic retinopathy (DR) is a leading cause of vision loss in the working-age population worldwide. Many observational and preclinical studies have implicated vascular endothelial growth factor (VEGF) in the pathogenesis of DR, and recent successes with anti-VEGF therapy for age-related macular degeneration (AMD) have prompted research into the application of anti-VEGF drugs to DR. Here we review the numerous early studies that suggest an important potential role for anti-VEGF agents in the management of diabetic retinopathy.

Conclusions

For diabetic macular edema, phase II trials of intravitreal pegaptanib and intravitreal ranibizumab have shown short-term benefit in visual acuity. Intravitreal bevacizumab also has been shown to have beneficial short-term effects on both visual acuity and retinal thickness. For proliferative diabetic retinopathy (PDR), early studies suggest that intravitreal bevacizumab temporarily decreases leakage from diabetic neovascular lesions, but this treatment may be associated with tractional retinal detachment (TRD). Furthermore, several studies indicate that bevacizumab is likely to prove a helpful adjunct to diabetic pars plana vitrectomy (PPV) for TRD. Finally, three small series suggest a potential beneficial effect of a single dose of bevacizumab to prevent worsening of DME after cataract surgery. Use of anti-VEGF medications for any of these indications is off-label. Despite promising early reports on the safety of these medications, we eagerly await the results of large, controlled trials to substantiate the safety and efficacy of anti-VEGF drugs for diabetic retinopathy.

Keywords

Diabetic retinopathyPegaptanibRanibizumabBevacizumabVascular endothelial growth factor (VEGF)

Introduction

The leading cause of vision loss in older patients in Europe and the US is age-related macular degeneration (AMD), and diabetic retinopathy (DR) is a leading cause in working-age patients worldwide [1, 2]. Recent advances in anti-VEGF therapy in AMD have prompted research into its application to DR.

The treatment of neovascular AMD has been revolutionized by the results of the MARINA and ANCHOR trials. MARINA showed that treating minimally classic or occult neovascular AMD with the anti-VEGF medication ranibizumab results in a 95% rate of visual stabilization or improvement at 12 months. Furthermore, nearly 40% of these patients had a 15-letter (approximately 3-line) or greater gain in acuity at 12 months [3]. ANCHOR demonstrated similar results in eyes with predominantly classic choroidal neovascularization in AMD [4]. These outcomes were a substantial improvement over those achieved with photodynamic therapy with verteporfin (49–61% stabilization at 12 months) [5, 6]. Given this remarkable clinical success, a search for additional applications of anti-VEGF medications has ensued.

Diabetic retinopathy is an attractive potential application with high world-wide prevalence. The American Diabetes Association reports that 8% of Americans are diabetic [7]. The Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR), a landmark study of diabetic retinopathy, showed a 98% rate of retinopathy in Americans with type 1 diabetes who had the disease for greater than 15 years, and a rate of 78% for those with type 2 disease [8, 9]. WESDR is now nearly 30 years old, but more recent studies have continued to show the great burden of diabetic retinopathy in a variety of communities. For example, Proyecto VER (2001) showed a nearly identical rate of DR after 15 years with diabetes and an overall prevalence of DR of 48% in a Hispanic diabetic population [10]. The Beijing Eye Study (2009) reports a 37% rate of DR among all diabetic patients and a 54% rate after 10–19 years with the disease [11].

Our greatest tool for preventing vision loss from diabetic retinopathy is perhaps timely screening and identification of patients at risk. Regular office visits can serve to educate patients on the importance of tight blood sugar and blood pressure control in both type 1 and type 2 diabetes. We know from the Diabetes Control and Complications Trial (DCCT) that tight blood sugar control results in a 76% reduction in the development of DR in type 1 disease, and a 54% reduction in the progression of DR over the 9-year study [12]. Furthermore, for type 2 diabetics the United Kingdom Prospective Diabetes Study (UKPDS) showed a 29% reduction in the need for laser surgery for DR with tight blood glucose control, and also showed a 47% reduction at 9 years in 3-line loss in acuity with tight blood pressure control [13, 14].

Regular examinations are also the means for timely application of known effective treatments for DR. For proliferative diabetic retinopathy (PDR) with high-risk characteristics, the Diabetic Retinopathy Study (DRS) showed a greater than 50% decrease in the rate of severe vision loss with pan-retinal photocoagulation (PRP) [15]. When new vessels respond to PRP by regressing within the first 3 months after treatment, the visual outcome tends to be excellent [16]. However, the limitations of PRP have been well-documented. They include poor response to treatment, pain during treatment, nyctalopia, loss of peripheral vision, uveal effusions, worsening of macular edema, and difficulty in treating eyes with vitreous hemorrhage [1619]. New treatment modalities would therefore be valuable in treating PDR.

Diabetic macular edema (DME) has been an area of particular interest for clinical investigators, in large part because the Early Treatment Diabetic Retinopathy Study (ETDRS) indicated that few patients improved after focal laser treatment. Although PDR is the cause of most severe vision loss, DME is more prevalent and is the leading cause of moderate vision loss in patients with DR.

It came as a surprise to some when the Diabetic Retinopathy Clinical Research (DRCR) Network’s randomized trial comparing intravitreal triamcinolone with focal/grid photocoagulation showed laser treatment to be not only safer, but also more effective at the primary 2-year endpoint [20]. Focal laser treatment remains the standard treatment for DME, and the DRCR Network study showed that 51% of patients in the focal laser arm had at least 5 letters of improvement at 2 years. A review of the data from the ETDRS has demonstrated that while only 10% of subjects improved with focal laser, about 40% of those with a baseline acuity worse than 20/40 had gained 6 or more letters at 3 years [2123].

Although focal laser photocoagulation is probably more effective in treating DME than originally thought, the search for new treatments rightfully continues. Many prospective studies of anti-VEGF medications for DME have been recently published, and some are currently underway. The remainder of this review examines these studies of DME, prospective trials of anti-VEGF medications for PDR, and trials of anti-VEGF agents for other specific problems in the treatment of diabetic retinopathy.

Preclinical rationale for anti-VEGF therapy for diabetic retinopathy

There are seven members of the VEGF family of glycoproteins: VEGF-A, VEGF-B, VEGF-C, VEGF-D, VEGF-E, VEGF-F, and PlGF (placental growth factor). For the purposes of this discussion, we refer to VEGF-A as VEGF. Numerous VEGF isoforms have been described, and they result from alternative mRNA splicing. VEGF121, VEGF165, VEGF189, and VEGF206 are the most abundant isoforms [24]. VEGF acts via two distinct classes of receptors. The VEGFR family of receptor tyrosine kinases plays roles in angiogenesis, lymphangiogenesis, and hematopoiesis [25, 26]. The second class is the neuropilins, which are found on vascular endothelial cells, tumor cells, and within the nervous system [24, 27]. Numerous retinal cell types produce VEGF, including RPE cells, astrocytes, Müller cells, vascular endothelial cells, pericytes, and ganglion cells [2831].

VEGF levels in ocular tissues from patients with diabetes are greater than in non-diabetic subjects [32]. This finding begs the question of how exactly VEGF fits into the biology of DR. Numerous preclinical studies point to a central role for VEGF in the pathogenesis of both DME and PDR. Animal models of nonproliferative DR have been known for a decade to have increased retinal VEGF levels and increased VEGFR2 expression [33]. Roberts and Palade showed in 1995 that topical or intradermal administration of VEGF in rats was followed by a rapid increase in capillary and post-capillary venule permeability [34]. This phenomenon causes breakdown of the blood–retinal barrier (BRB). BRB breakdown can be experimentally prevented in diabetic rats by the administration of the VEGF-neutralizing compound VEGF TrapA40 [35].

Humans with PDR have high levels of VEGF in vitreous samples, and these levels decline significantly after laser photocoagulation treatment [36]. In mouse models of ischemic retinopathy, blocking VEGF activity prevents development of proliferative disease [37]. Intravitreal injection of VEGF has been shown to cause iris neovascularization in nonhuman primates [38]. These studies provide the pre-clinical basis for investigations of anti-VEGF agents for PDR.

DR is increasingly recognized as an inflammatory process, and VEGF has been identified as a pro-inflammatory mediator. VEGF causes increased expression of the cellular adhesion molecule ICAM-1, and is chemotactic for monocyte/macrophage lineage cells [39, 40]. VEGF therefore promotes leukostasis and increased leukocyte counts in the retinas of diabetic animals [41]. Leukostasis has been found to correlate temporally with leakage [42]. Leukocytes are particularly abundant at the leading edge of areas of pathologic neovascularization [43]. Blockage of VEGF decreases retinal leukocyte counts in experimental diabetes [44]. ICAM-1 expression and leukocyte counts are also increased in human diabetic retinas [45]. These findings provide additional rationale for clinical trials of VEGF blockade for DR.

Anti-VEGF agents and safety

Three anti-VEGF pharmacologic agents are currently available commercially. Pegaptanib (Macugen, OSI/Eyetech, Melville, NY, USA) is a pegylated aptamer that targets the VEGF165 isoform. It has been shown to inhibit VEGF’s endothelial mitogen activity and its vascular permeability effects [42, 46]. The US Food and Drug Administration (FDA) has approved Macugen for the treatment of neovascular AMD. The VEGF Inhibition Study in Ocular Neovascularization (VISION) trial established its safety and efficacy in neovascular AMD [47].

Ranibizumab (Lucentis, Genentech, Inc., South San Francisco, CA, USA) is a recombinant, humanized antibody fragment that binds all isoforms of VEGF, whereas bevacizumab (Avastin, Genentech, Inc.) is a recombinant, full-length, humanized antibody that also binds all VEGF isoforms. Lucentis is currently FDA-approved for neovascular AMD, while Avastin is used on an off-label basis for a variety of ophthalmic conditions. Large clinical trials of Avastin are currently underway for AMD, DME, and vein occlusions, but the safety and efficacy of Avastin for intraocular use remains to be demonstrated.

The safety data to date for these three medications is reassuring, although a large prospective trial with methodical surveillance of adverse events is still lacking for bevacizumab. The largest data set for bevacizumab is a retrospective study of 1,173 patients who received intravitreal bevacizumab and were followed for 12 months [48]. This study yielded the following adverse effects: seven cases of acute elevation of blood pressure, six strokes, five myocardial infarctions, five deaths, seven cases of bacterial endophthalmitis, seven cases of tractional retinal retachment, and four cases of uveitis. These numbers are similar to those found in the prospective, controlled studies of the other anti-VEGF agents. Mason and colleagues retrospectively studied their series of 5,233 intravitreal bevacizumab treatments for the incidence of acute post-injection endophthalmitis [49]. They found just one case.

The most dramatic safety concerns surrounding the use of bevacizumab come from studies of its intravenous use in cancer therapy. Side-effects in these studies included arterial thromboembolism, gastrointestinal perforation, hemorrhage, hypertensive crisis, and nephrotic syndrome [5052]. Small, early trials of intravenous bevacizumab in AMD patients rather than cancer patients revealed only increased rates of hypertension, which was readily controlled with medication [53, 54]. The relatively miniscule (1/400th of the intravenous) intravitreal dose, as one might expect, appears to have a far better side-effect profile [55].

The VISION trial has provided 3 years of safety data for intravitreal pegaptanib in patients with neovascular AMD [47, 56, 57]. No systemic side effects were attributed to treatment over the course of the study. Patients receiving sham injections in the first 2 years of the study were more likely to experience the most common ocular side effects, including punctate keratitis, than patients receiving pegaptanib. Serious ocular complications were thought to be related to the injection procedure rather than the medication. These included endophthalmitis (0.06% to 0.16% per injection), retinal detachment (0.03% to 0.17% per injection), and lens injury (0% to 0.7% per injection).

Data from the MARINA and ANCHOR trials for neovascular AMD support the safety of intravitreal ranibizumab. The 2-year data from MARINA showed no increase in systemic adverse effects with ranibizumab [3]. Pooled 1-year safety data from PIER, MARINA and ANCHOR showed an increased rate of vascular events (2.1% rate of myocardial infarction and stroke) in the ranibizumab arms versus the control (1.1%) [58]. These differences may be clinically irrelevant given the 2-year safety data. Rates of adverse ocular effects were low, and similar to rates reported for intravitreal bevacizumab and pegaptanib.

The majority of the safety data for these medications is derived from studies of patients with neovascular AMD, and new safety concerns may arise with the use of these medications in different patient populations. The population with diabetes tends to be younger, with more heart and kidney disease, and the ocular status is different. Diabetic eyes may have neovascularization and fibrous tissue that can contract and cause an altogether different set of adverse effects. For these reasons, we eagerly await the results of appropriate safety studies in diabetic populations.

Trials of anti-VEGF agents for DME

Pegaptanib

Numerous prospective studies of anti-VEGF medications for DME have been recently published (Table 1). Much of the clinical impetus for these ongoing trials has come from the work of the Macugen Diabetic Retinopathy Study Group, who conducted a phase II trial of pegaptanib for fovea-involving DME [59]. One hundred and seventy-two patients at multiple centers were randomized to four study arms: 0.3 mg intravitreal pegaptanib, 1 mg, 3 mg, or sham. Injections were given at weeks 0, 6 and 12. Additional injections could be administered to subjects after week 12 at the discretion of the masked investigators. Similarly, investigators could choose to treat with focal laser beginning in week 13. Patients were also masked to their treatment arm.
Table 1

Trials of anti-VEGF drugs for diabetic macular edema

Paper

Drug

Design

N

Follow-up

Population studied

Authors’ conclusions

Macugen Diabetic Retinopathy Study Group, 2005 [59]

Pegaptanib

Randomized, double-masked, dose-ranging, controlled

172

36 weeks

Center-involving DME, no previous treatment, mean 56 letters

Pegaptanib groups had better, VA, CRT, and less need for laser

Nguyen et al. 2009 [63]

Ranibizumab

Randomized, double-masked, three treatment arms

126

6 months

Center-involving DME in type 1 and 2 diabetics

Significantly better visual outcomes over 6 months in ranibizumab groups versus laser

DRCRnet, 2007 [60]

Bevacizumab

Randomized, partially masked, five treatment arms

121

24 weeks

Center-involving DME, mean 64 letters

Promising data in bevacizumab arms, phase III trial warranted

Ahmadieh et al. 2008 [65]

Bevacizumab (triamcinolone)

Randomized, partially masked, placebo-controlled

115

24 weeks

CSME with failed focal laser treatment, no previous intravitreal steroid

Three consecutive bevacizumab injections were beneficial in refractory DME

Faghihi et al. 2008 [66]

Bevacizumab (triamcinolone)

Randomized, three treatment arms

130

16 weeks

Center-involving DME in type 2 diabetics, no previous treatment, mean BCVA 0.75

Bevacizumab and bevacizumab/IVTA significantly reduced CRT versus laser; small acuity benefit in bevacizumab/IVTA group only

Soheilian et al. 2009 [67, 68]

Bevacizumab (triamcinolone)

Randomized, double-masked, three treatment arms

150

24 weeks

CSME, no previous treatment, mean BCVA 0.66

Significantly better BCVA improvement in bevacizumab arms versus laser at 24 weeks

Kook et al. 2008 [72, 73]

Bevacizumab

Open-label, single dose, uncontrolled

126

6 months (59 completed 12 months)

Diffuse, chronic CSME that failed prior treatment, mean 40 letters

Multiple bevacizumab injections are associated with a decrease in CRT and with improved BCVA at some time points

Lam et al. 2009 [61]

Bevacizumab

Randomized, dose-ranging

52

6 months

Diffuse center-involving DME, mean BCVA 0.61

Three monthly injections were associated with decreased CRT and increased BCVA in both dosing groups

Paccola et al. 2008 [70]

Bevacizumab (triamcinolone)

Randomized, two treatment arms

28

24 weeks

Center-involving DME, failed previous laser treatment, mean BCVA 0.94

IVTA may offer more short-term CRT and BCVA benefits than bevacizumab

Fang et al. 2008 [62]

Bevacizumab

Open-label, single dose, uncontrolled

38

12 weeks

Center-involving DME, mean BCVA 0.65

Temporary beneficial effect on CRT and BCVA irrespective of previous treatment

Shimura et al. 2008 [71]

Bevacizumab (triamcinolone)

Patients with bilateral disease received one drug in each eye

28

24 weeks

Fovea-involving DME, no previous treatment, mean BCVA 0.62

Both groups had short-term improvements in CRT and VA. IVTA produced better results at 1 and 24 weeks.

Kumar and Sinha, 2007 [74]

Bevacizumab

Open-label, single dose, uncontrolled

20

7.5 months

Diffuse DME, no ischemic disease, mean BCVA 1.34

No adverse events; significant effects on BCVA and CRT at 3 and 6 months after the second injection

Velez-Montoya et al. 2009 [75]

Bevacizumab

Open-label, single dose, uncontrolled

23

4 weeks

Bilateral diffuse DME, no previous treatment, Mean BCVA in treated eye 22 letters

No effect of intravitreal bevacizumab on DME in contralateral eye

Bonini-Filho et al. 2009 [77]

Bevacizumab

Open-label, single dose, uncontrolled

10

54 weeks

DME and severe capillary loss within 1500 µm of the foveola

Bevacizumab may improve BCVA and CMT in DME with severe capillary loss

[Abbreviations: N (number of eyes); DME (diabetic macular edema); VA (visual acuity); CRT (central retinal thickness); CSME (clinically significant macular edema); BCVA (best-corrected visual acuity); IVTA (intravitreal triamcinolone)]

Eyes in the pegaptanib groups did better than the sham arm, and this was particularly true for the 0.3 mg arm. After 36 weeks of follow-up, the pegaptanib-treated eyes had better visual acuity (P = 0.04 for the 0.3 mg group vs sham), more reductions in central retinal thickness (P < 0.01 for the 0.3 mg group vs sham), and less need for macular laser photocoagulation (P = 0.042 for the 0.3 mg group vs sham). These benefits were detectable despite the fact that 23% more sham-treated eyes received focal or grid laser treatment between weeks 12 and 36. Some visual improvement occurred in 73% of patients in the 0.3 mg pegaptanib group versus 51% in the sham group. The mean improvement in the 0.3 mg group was 4.7 letters, and 18% gained 3 Snellen lines or more.

This study notably included only patients with no previous history of treatment for DME, unlike many of the more recent trials. Some subsequent trials have concluded that treatment-naïve eyes respond better to anti-VEGF therapy [6062]. This may be because more difficult eyes would be more likely to meet study inclusion criteria. The study provides some data about the appropriate dosing interval for pegaptanib, although this remains a significant area for future study. Patients were seen at 6-week intervals, and some effect of the medication apparently persisted on average over the 6-week period between visits. Longer or shorter intervals may show more benefit in future studies. The treatment regimen in the study was less standardized after the week 12 evaluation, so the usefulness of pegaptanib for longer-term management of DME remains unclear. Furthermore, the optimal combination of focal laser treatment and pegaptanib will require further study.

The size of this study prevented detection of a statistical difference between the three pegaptanib groups, and the study was not powered for the detection of adverse effects that might occur at a less than 3% rate. There was a trend that suggested that the 0.3 mg group faired better than the 1 mg and 3 mg groups. Do the higher doses block known beneficial physiologic effects of VEGF on retinal neurons, RPE cells, or choroidal endothelial cells [24]? While the VISION trial of pegaptanib for AMD did not reveal any unexpected serious adverse effects of intravitreal pegaptanib, might there be a rare adverse effect in the diabetic population? Is it possible, since pegaptanib does not appear to be as efficacious as ranibizumab in patients with AMD, that it might be a “less strong,” and therefore safer, agent? No trials of pegaptanib for PDR have been reported to date, but we wonder whether this drug might promote less retinal traction. We doubt though that there will be a trial of pegaptanib vs ranibizumab or bevacizumab any time soon.

Ranibizumab

Nguyen and colleagues have recently reported the six month results of the Ranibizumab for Edema of the Macula in Diabetes (READ-2) Study of ranibizumab for DME [63]. Acting upon the favorable results of their own phase I trial, the group randomized 126 eyes with DME to three groups: the first received 0.5 mg intravitreal bevacizumab at baseline and months 1, 3, and 5; the second underwent focal/grid laser photocoagulation at baseline and again at 3 months if needed; the third group underwent laser at baseline and ranibizumab injections at baseline and 3 months [64]. Their primary end point was best-corrected visual acuity (BCVA) at 6 months, and at this time point the ranibizumab-only group had done best. The ranibizumab-only group gained a mean of 7.24 letters, the laser-only group lost 0.43 letters, and the combination group had gained 3.80 letters. The difference between the ranibizumab and laser groups was statistically significant, but that between the two ranibizumab groups did not attain significance (P = 0.08).

Among the secondary outcomes of this study, 22% of ranibizumab-only eyes gained 3 or more lines of acuity at 6 months, versus none of the laser-only eyes and 8% of combined treatment eyes. One ranibizumab-treated patient died of a cerebrovascular accident 6 weeks after the first injection. The authors doubt a connection to ranibizumab treatment given the patient’s substantial pre-existing cardiovascular risk factors and the 6-week period between the injection and the death. No other substantial adverse events occurred.

This study included eyes with a previous history of treatment for DME. It also included both type 1 and type 2 diabetics. While the number of injections differed between groups, the authors tested just one dose. Since data was only reported for the 3- and 6-month visits, this study does not answer the question of the duration of effect of a single dose of ranibizumab. Future studies will need to answer remaining questions about the apparent difference between the combined treatment group and the ranibizumab group; the comparison of these two groups is probably biased in favor of the ranibizumab-only group, since these eyes had a final injection just 1 month before the final data point. Future studies should also address the long-term efficacy of treatment and the potential need for long-term repeat dosing. Like the pegaptanib study, this study was not powered for the detection of rare adverse effects, and its limited duration precludes detection of long-term adverse outcomes. Numerous other clinical trials of ranibizumab for DME are underway currently, including a large Genentech-sponsored trial.

Bevacizumab

Bevacizumab, for reasons of cost and availability, is currently the best-studied anti-VEGF medication for DME. Several studies, beginning with the work of the DRCR Network, indicate that this drug warrants phase III investigation for DME. The DRCR Network conducted a randomized study of 121 eyes over a 12-week period (safety data is reported for 24 weeks) [60]. The study involved five treatment arms: focal photocoagulation, two intravitreal injections of 1.25 mg bevacizumab at 0 and 6 weeks, two intravitreal injections of 2.5 mg bevacizumab at 0 and 6 weeks, 1.25 mg bevacizumab at week 0 followed by a sham injection at week 6, and lastly 1.25 mg bevacizumab at 0 and 6 weeks combined with focal photocoagulation at 3 weeks. 69% of eyes in this study had undergone previous treatment for DME.

The two groups that received two bevacizumab injections without laser had a statistically significant improvement in vision over the laser-only group, and this difference persisted through the 12-week monitoring period. The median gain in vision following the second injection (week 9) was seven letters for the 1.25 mg group and eight for the 2.5 mg group. These two groups also had a greater improvement in central subfield thickness (CST) at the 3-week visit, with a trend suggesting a similar finding at subsequent visits. There were no detectable differences between the 1.25 mg and 2.5 mg doses. The single injection group had no advantage over the photocoagulation group in this study. Interestingly, the combination of laser and bevacizumab yielded comparable results to the laser-only group, with a trend toward worse short-term acuity outcomes than the eyes that received two bevacizumab injections.

The DRCR Network study suggests that bevacizumab is an effective treatment for DME in eyes that have been previously treated and those that have not. The paper does, however, report that previously untreated eyes had a greater (P = 0.04) improvement in vision and a trend toward a better response in terms of CST. Another trend from this study was that eyes with subretinal fluid may have a greater improvement in acuity (P = 0.06) after bevacizumab treatment. About half of the bevacizumab-treated eyes in the study had a significant (>11%) decrease in retinal thickness. No new safety concerns were identified.

The study was not designed to be an adequate comparison of bevacizumab to focal laser. A true head-to-head comparison will require years of follow-up since the primary endpoint of the ETDRS was 3 years post-treatment [22]. Such a lengthy study will require repeat injections, and the dosing schedule used by the DRCR Network is already informing the design of current and future trials. Six weeks post-injection, mean CST had plateaued or worsened from the 3-week value, suggesting that a dosing interval of less than 6 weeks may be appropriate for future trials.

Of the other published trials of bevacizumab for DME, three of the largest have compared intravitreal bevacizumab and intravitreal triamcinolone (IVTA) [6568]. Of these three studies, the work of Ahmadieh and colleagues has the longest follow-up (24 weeks) but suffers from a baseline statistical difference in central macular thickness (CMT) between the control and treatment groups. The paper also does not report baseline visual acuities. Nonetheless, the results support the further study of bevacizumab for DME in eyes that have failed prior laser treatment, a group not specifically studied in the pegaptanib study or the DRCR Network study.

The design involved randomizing 115 eyes to one of three study arms: a bevacizumab-only arm, an IVTA/bevacizumab combination arm, and a placebo arm. The two treatment arms received three 1.25 mg bevacizumab injections separated by 6 weeks, with the IVTA/bevacizumab group receiving an additional injection of 2 mg triamcinolone at the baseline visit only. The two treatment groups had better BCVA versus placebo at all time points, with the exception of the bevacizumab-only group at the first 6-week follow-up. They detected no difference between the bevacizumab and IVTA/bevacizumab groups in BCVA or CMT. The effect of the injections lasted for 12 weeks after the final injection in this study, with no clear trend toward worsening acuity or edema over that period. Mean CMT reductions at 24 weeks were 96 microns (P = 0.012) in the bevacizumab-only group, and 92 microns (P = 0.022) in the bevacizumab/IVTA group.

Faghihi and colleagues also compared intravitreal bevacizumab with IVTA, but instead studied a group of eyes with no history of treatment for DR [66]. They randomized 130 eyes of type 2 diabetic patients to one of three arms: a bevacizumab arm, an IVTA/bevacizumab arm, and a macular photocoagulation arm. Injections were given at the baseline visit only, and doses of 1.25 mg bevacizumab and 2 mg triamcinolone were used. Each of the three groups had significant improvements in CMT at both the 6- and 16-week visits versus baseline. Likewise, they each had significant improvements in BCVA at both visits, with the exception of the bevacizumab group at 16 weeks. The bevacizumab group outperformed the laser group in CMT and BCVA at 6 weeks, but not at 16. The bevacizumab/IVTA group outperformed the laser group in CMT and BCVA at both 6 and 16 weeks.

Like the DRCR Network study, this study is an incomplete head-to-head comparison of focal laser and these two pharmacotherapies. It does offer helpful data, however, by suggesting that a single bevacizumab injection will generally not last 16 weeks. Furthermore, the study indicates that a combination with IVTA may reduce the initial frequency of injections. The bevacizumab may not add any benefit to the IVTA at the 16-week visit, since the acuity improvement reported by Faghihi and colleagues is similar to that reported by the DRCR Network at 4 months for IVTA alone [69]. While IVTA may add some benefit to bevacizumab, combining the two drugs eliminates the potential to avoid the known adverse effects of IVTA.

The work of Soheilian and colleagues features a similar design [67, 68]. In fact, the three treatment arms are the same as in the trial conducted by Faghihi and colleagues. This study benefits from randomization and careful masking of both subjects and investigators. Unfortunately, their photocoagulation group had a significantly better mean BCVA at baseline. After looking at 150 previously untreated eyes, they concluded that both bevacizumab groups had similar, significant improvements in acuity only versus photocoagulation. However, the baseline differences between their study groups hinder final interpretation.

Two smaller studies have also compared IVTA and intravitreal bevacizumab. Paccola and colleagues administered one baseline injection of either 1.5 mg bevacizumab or 4 mg triamcinolone to 28 eyes with histories of failed laser treatment and relatively poor vision [70]. While both groups had significant improvements in CMT and BCVA at 4 weeks, the improvements were greater and lasted longer in the IVTA group (12 weeks). Shimura and colleagues treated 14 patients with bilateral, previously untreated DME with 1.25 mg bevacizumab in one eye and 4 mg triamcinolone in the opposite eye [71]. They also found that both groups had improvements in CMT and acuity at 1 and 4 weeks, but that improvement in the IVTA group lasted longer (24 weeks).

Among the larger studies of bevacizumab for DME, Haritoglou and colleagues investigated perhaps the most diverse group of eyes, with no exclusions based on previous treatment, ischemia, or poor acuity. They conducted a noncomparative trial of 1.25 mg bevacizumab at baseline, with subsequent repeat dosing based on the presence of a beneficial optical coherence tomography (OCT) or acuity response to the initial injection [72, 73]. All 126 eyes had failed previous treatment, and had diffuse and chronic DME. At 6 months, mean central retinal thickness had decreased from 463 to 374 µm (P < 0.001). The improvement in mean acuity was not significant at 6 months. The group reports data over a 12-month follow-up period, but more than half of the study eyes were lost to follow-up after 6 months. Baseline retinal thickness, previous treatment, and diameter of the foveal avascular zone (FAZ) did not correlate with responses to treatment. While the diverse baseline characteristics of their study eyes and the variable number of treatments given may confound application of their results, their data supports the notion that bevacizumab is a good target for further study in eyes that have failed previous treatment.

The DRCR Network study of bevacizumab detected no difference between 1.25 mg bevacizumab and 2.5 mg, and similar results are reported by Lam and colleagues in their study of 52 eyes [60, 61]. This study involved three monthly injections. It offers a longer follow-up period of 6 months. Both groups had significant reductions in central foveal thickness (CFT) at all visits, peaking at the 3- and 4-month visits. Both groups also had significant improvements in BCVA at all visits excluding the 1-week visit. The maximal mean improvement was 0.15 logMAR units at 4 months in the 1.25 mg group and 0.13 units at months 3 and 6 in the 2.5 mg groups. The two study groups had statistically similar results throughout the 6 months. Subgroup analysis suggested that the 17 eyes with histories of previous DME treatment had less improvement at 6 months.

Several smaller studies to date provide further data in support of future investigation of bevacizumab for DME. Kumar and Sinha studied two 1.25 mg injections in 20 eyes with relatively poor baseline acuity that had failed previous laser treatment [74]. They found significantly improved BCVA and CMT at 3 and 6 months after the second injection. Fang and colleagues investigated a single 1 mg injection in 38 eyes, 29 of which had histories of previous treatment for DME [62]. They report significant improvements in BCVA and central retinal thickness (CRT) for the group as a whole, with a measurable improvement persisting longer (8 weeks) in the nine eyes with no history of previous treatment. Velez-Montoya and colleagues studied 23 patients with bilateral DME to determine whether a 2.5 mg bevacizumab injection in one eye would affect the untreated eye [75]. In the treated eye, they found significant improvements in CRT but not acuity. They detected no effect in the untreated eyes.

ETDRS Report Number 19 suggests the possibility of a trend toward less treatment effect for focal laser in eyes with diabetic macular edema and severe capillary loss [76]. Hoping to bolster treatment options for such eyes, Bonini-Filho and colleagues conducted a pilot study of intravitreal bevacizumab for macular edema in ten eyes with severe capillary loss [77]. All ten eyes underwent a 1.5 mg injection at baseline, and were re-treated at follow-up visits based on the presence of intraretinal or subretinal fluid on OCT. CMT and BCVA improved significantly throughout the 54-week study (approximately 20/125 to 20/80). Follow-up fluorescein angiogram revealed no progression of capillary loss at the end of the study.

We draw several conclusions about the short-term use of bevacizumab from this group of studies. First, intravitreal bevacizumab has some beneficial short-term effects on treatment-naïve eyes with DME. Likewise, it has beneficial short-term effects on eyes with laser-refractory DME, although most groups did not require multiple treatments over 1 to 2 years before diagnosing “failure.” Multiple small studies suggest that treatment-naïve eyes are more responsive to the drug, either because they truly are more responsive or because of an ascertainment bias and an overabundance of severe cases in studies that enroll previously treated subjects who did not “respond.” Bevacizumab treatment is associated with improvements in both acuity and CMT. Repeat dosing increased the average beneficial effect of the drug. The optimal timing of repeat dosing is unclear, but is almost certainly between 3 and 12 weeks. A repeat dosing interval of 3 to 6 weeks seems most likely to produce maximal benefit. A 2.5 mg dose appears to have no benefit over 1.25 mg. Perhaps most importantly, the absence of serious side-effects in these reports should not lull the patient or physician into the sense that the safety of this use of bevacizumab has been established. However, it seems that any adverse effect of this medication for DME will be rare, late, or beyond the scope of previous studies.

Trials of anti-VEGF agents for PDR

While no sizeable prospective trials of pegaptanib or ranibizumab for PDR have yet been published, several groups have conducted trials of bevacizumab (Table 2). The largest trial to date is by Mirshahi and colleagues [78]. Forty type 2 diabetes patients with bilateral PDR with high-risk characteristics underwent scatter laser treatment according to the ETDRS protocol and had a single 1.25 mg bevacizumab injection in one eye; in the contralateral control eye, they received scatter laser and sham injection. Eyes were randomly assigned to treatment or sham.
Table 2

Trials of anti-VEGF drugs for proliferative diabetic retinopathy

Paper

Intervention

Design

N

Follow-up

Population studied

Authors’ conclusions

Mirshahi et al. 2008 [78]

Bevacizumab/PRP combination

Partially masked, sham controlled in fellow eye

80

16 weeks

Bilateral high-risk PDR, no previous laser or TRD

Bevacizumab temporarily augmented response to PRP

Cho et al. 2009 [81]

Bevacizumab/PRP combination

Randomized, two treatment arms

41

3 months

High-risk PDR, no previous laser

Bevacizumab improves outcomes when combined with PRP, especially in absence of CSME

Moradian et al. 2008 [80]

Bevacizumab

Open label, single dose, uncontrolled

38

20 weeks

Eyes with fibrovascular tissue refractory to PRP or fresh vitreous hemorrhage preventing PRP

Treatment resulted in significant improvement of vitreous hemorrhage, and active neovascularization, may increase risk of tractional retinal detachment

Tonello et al. 2008 [82]

Bevacizumab/PRP combination

Randomized, two treatment arms

30

16 weeks

High-risk PDR, no previous laser treatment or previous thromboembolic event

Bevacizumab plus PRP temporarily reduces area of NV leakage versus PRP alone

Jorge et al. 2006 [83]

Bevacizumab

Open label, single dose, uncontrolled

15

12 weeks

Eyes with active NV unresponsive to previous PRP, worse than 20/40 vision

Bevacizumab temporarily reduced leakage from active NV, further study warranted

[Abbreviations: N (number of eyes); CSME (clinically significant macular edema); PRP (panretinal photocoagulation); PDR (proliferative diabetic retinopathy); TRD (tractional retinal detachment); NV (neovascularization)]

The group found that at week 6 87.5% of bevacizumab-treated eyes had complete regression of neovascularization (NV) versus 25% in the sham group (p < 0.005). They defined complete regression as a lack of angiographic leakage from the neovascular lesion on the 2-minute image. By week 16, the difference between the two groups had disappeared. In fact, ten of the 40 patients had complete regression angiographically at 16 weeks in both eyes, and the remaining 30 patients had some residual active neovascularization in both eyes. Rates of partial regression were similar at 16 weeks. Only hemoglobin A1C levels were correlated with recurrence of NV in bevacizumab-treated eyes. The group detected no side-effects over the course of the study.

This study provides further evidence that bevacizumab has a substantial effect on new vessels, but the significance of these findings remains to be elucidated. The anti-permeability effects of anti-VEGF agents have been well-documented. One must then question whether decreased leakage on fluorescein angiography indicates regression of new vessels or whether this is simply a manifestation of a known effect of the drug. A longer-term study might assess the value of repeat injections and attempt to determine whether this decreased leakage from new vessels correlates with a decreased rate of vitreous hemorrhage, tractional retinal detachment (TRD), or vision loss. Furthermore, the baseline severity of proliferative disease is not reported for this cohort, except to say that all had at least PDR with high-risk characteristics. Intravitreal bevacizumab may be a better treatment for early high-risk PDR, given other reports that suggest an increased risk of TRD with this medication [79, 80].

The work of Cho and colleagues looked primarily at visual acuity and CMT in the immediate post-PRP period [81]. They randomized 41 eyes with high-risk PDR to PRP alone or PRP plus 1.25 mg of intravitreal bevacizumab. Twenty-two eyes also had CSME at the time of randomization. They distributed these eyes equally between the two treatment arms, and administered standard focal/grid laser treatment for CSME at the time of initiation of PRP.

The PRP-only group had significantly worse visual acuity when compared to baseline at 3 months (P = 0.041). The PRP/bevacizumab group had a significantly decreased CMT at 1 and 3 months. Eyes with CSME when the study began had no significant change in acuity in either treatment group, but their CMT was significantly improved versus baseline at 1 and 3months in the combined treatment group. In eyes without CSME, acuity worsened at 1 and 3 months compared to baseline in the PRP-only group, but did not change in the combined treatment group. The PRP-only eyes with no CSME also had increased CMT 1 and 3 months post-PRP. Four eyes in the PRP-only group had worsening neovascularization and vitreous hemorrhage during the study period, while none of the combined treatment eyes had similar progression. The authors detected no adverse events in patients receiving intravitreal bevacizumab.

This study aimed to evaluate the efficacy of intravitreal bevacizumab for prevention of short-term vision loss and macular edema associated with PRP. McDonald and Schatz showed that 43% of patients in a large series had increased macular edema 6 to 10 weeks after PRP, and that 8% developed chronic macular edema after the treatment that was associated with significant vision loss [18]. Preventing these adverse outcomes, particularly persistent vision loss, is an attractive potential application for bevacizumab. This study suggests that bevacizumab can prevent short-term vision loss after PRP in patients not treated with focal laser at the time of PRP. The study was brief, however, and eyes with prior history of laser treatment were excluded.

Tonello and colleagues studied two similar treatment groups, but looked at visual acuity and the total area of fluorescein leakage from areas of neovascularization [82]. They randomized 30 eyes with high-risk PDR and no previous history of laser treatment to PRP alone or PRP plus 1.5 mg bevacizumab. Acuity did not change from baseline in either group, nor did the groups’ mean acuity differ at any time point. The area of active fluorescein leakage was significantly better at 4, 9, and 16 weeks in the combined treatment group than in the PRP-only group. The mean area of active leakage in the combined group had improved most at the 4- and 9-week visits, and it increased substantially by the 16-week visit. The paper does not report any substantial adverse effects in the bevacizumab group.

As in the Mirshahi paper, the area of active leakage from neovascular lesions decreased when bevacizumab was added to PRP. As mentioned previously, the practical implications of this finding will require future study. The increase in the area of fluorescein leakage between weeks 9 and 16 in this study indicates that re-injection might be most appropriate between these two time points. In contrast to the paper by Cho and colleagues, no changes in visual acuity occurred in either treatment group. While Tonello and colleagues studied a relatively small number of eyes over a short period, this finding might suggest that there can be important differences between patient populations. Tonello and colleagues studied a Latin population with a relatively high mean HbA1C (9.6), while Cho and colleagues studied an Asian population with a mean HbA1C of approximately 7.

Moradian and colleagues evaluated intravitreal bevacizumab for cases that were not amenable to traditional PRP [80]. They included eyes that had failed previous PRP, and eyes with vitreous hemorrhage that prevented adequate laser treatment. All 38 eyes received a 1.25 mg bevacizumab injection at baseline, and were re-injected at 6 or 12 weeks on an as-needed basis. They sought primarily to assess clearance of vitreous hemorrhage and regression of active fibrovascular tissue. They assessed vitreous hemorrhage using a grading system, and the extent of fibrovascular tissue using standard photos with an associated grading system.

They found significant resolution of vitreous hemorrhage versus baseline at 1, 12, and 20 weeks, with a trend toward significance at 6 weeks (P = 0.06). No significant change in the extent of fibrovascular tissue occurred, but most eyes were not able to be assessed for this variable because of media opacity. At week 6, only 14 of 38 eyes were suitable for assessment, and fewer were suitable at every other time point. All five eyes with iris neovascularization (NVI) at baseline had complete resolution of these lesions at the final 20-week evaluation, as did one new case of NVI that occurred during the study. Two eyes with moderate fibrovascular tissue at baseline that each underwent two injections developed TRD and required surgery.

The improved vitreous hemorrhage that this group noted in their patients as early as 1 week post-injection certainly seems noteworthy, but in the absence of controls no clear conclusions should be drawn. Since the natural history of vitreous hemorrhage is generally to resolve spontaneously, controls would be an essential component of a future study. Perhaps the most remarkable finding in this study was the occurrence of two TRDs (5.3% of study eyes). This rate corresponds with the report by Arevalo and colleagues that 11/211 (5.2%) eyes with PDR developed TRD after adjunctive intravitreal bevacizumab before vitrectomy [79].

Jorge and colleagues assessed intravitreal bevacizumab in eyes with persistent, active PDR in a noncomparative trial [83]. They administered one baseline 1.5 mg bevacizumab injection to 15 eyes and followed them for 12 weeks. They found that BCVA improved significantly from baseline at all time points (1, 6, and 12 weeks), from 20/160 at baseline to approximately 20/125 at 12 weeks. The mean area of active fluorescein leakage also improved significantly at all time points, peaking at 6 weeks, when none of the 15 eyes were found to have active leakage. By week 12, 14/15 eyes had recurrence of at least some angiographic leakage. The extent of DME at baseline is unclear. No significant adverse events occurred.

The causes for the improvement in acuity in this study are elusive, and this prevents us from drawing firm conclusions. Bevacizumab may be found to be an important intervention for eyes with refractory PDR, and this study demonstrates a substantial short-term effect on angiographic leakage. Recurrence of leakage between weeks 6 and 12 suggests that re-dosing in future trials should be performed within that time range.

Taken as a group, these studies suggest that intravitreal bevacizumab decreases leakage from diabetic neovascular lesions in newly diagnosed and refractory disease, has a temporary effect, may be associated with TRD, and may have beneficial effects on vitreous hemorrhage, post-PRP macular edema, and NVI. These studies collectively illustrate the difficulty in objectively measuring the effect that this drug has on active neovascular lesions; two techniques used for this purpose (measuring the area of angiographic leakage or employing a grading system) fail to provide functionally meaningful conclusions in brief studies. Nonetheless, intravitreal bevacizumab presents an attractive potential benefit for eyes with no other treatment options if the risk for TRD is judged to be low. Future study of this drug for PDR will benefit from these early contributions.

Bevacizumab as an adjunct to vitrectomy

Several groups have prospectively evaluated intravitreal bevacizumab as a pre-operative adjunctive therapy for patients undergoing vitrectomy (Table 3). Yang and colleagues conducted one of the first such studies, acting upon the observation made by Avery and colleagues that six patients with PDR who received intravitreal bevacizumab before vitrectomy had noticeable reductions in intraoperative bleeding [84, 85]. They prospectively enrolled 16 eyes with active diabetic proliferative disease and vitreo-retinal adhesions at three or more sites for pre-operative intravitreal bevacizumab, and compared them with 24 eyes that had previously undergone the same procedure. Controls had a similar extent of fibrovascular proliferation. One week before surgery, the 16 eyes were injected with 1.25 mg bevacizumab. They then performed a standard three-port pars plana vitrectomy (PPV), concluded in each case with infusion of C3F8 10%. Primary outcome measures were the severity of intraoperative bleeding (assessed by a grading scale), vitreous clear-up time, percentage of prolonged clear-up, and rates of recurrent hemorrhage.
Table 3

Trials of anti-VEGF drugs as adjuncts to diabetic vitrectomy

Study

Intervention

Design

N

Follow-up

Population studied

Authors’ conclusions

Yang et al. 2008 [84]

Bevacizumab prior to vitrectomy

Non-randomized, controlled, single dose

16

6 months

Active diabetic fibrovascular proliferation, eyes suitable for C3F8 10%

Bevacizumab may be effective adjunct to accelerate vitreous clear-up

Ahmadieh et al. 2009 [86]

Bevacizumab prior to vitrectomy

Randomized, controlled, single dose

68

1 month

Any indication for diabetic PPV, excluded eyes with internal tamponade

Bevacizumab may reduce the frequency of post-operative vitreous hemorrhage, and may obviate surgery in some cases

Yeh et al. 2009 [88]

Bevacizumab prior to vitrectomy

Randomized, controlled, single dose

41

6 months

Severe active diabetic fibrovascular proliferation, eyes suitable for silicone oil

Bevacizumab may decrease intra- and post-operative bleeding; possible increase in subretinal bleeding

Romano et al. 2009 [90]

Bevacizumab prior to vitrectomy

Open label, single dose, uncontrolled

32

6 months

Persistent vitreous hemorrhage with active PDR

Bevacizumab may be a safe adjuvant treatment for preventing recurrent VH

Rizzo et al. 2008 [87]

Bevacizumab prior to vitrectomy

Randomized, controlled, single dose

22

6 months

Severe PDR with TRD, tractional/rhegmatogenous RD or TRD with VH

Bevacizumab was well-tolerated, reduced active neovascularization, and facilitated surgery

Lucena et al. 2009 [89]

Bevacizumab prior to vitrectomy

Randomized, controlled, single dose

20

No post-operative follow-up reported

PDR with a macula-involving TRD of recent onset

Bevacizumab was associated with less intraoperative bleeding

[Abbreviations: N (number of eyes); PDR (proliferative diabetic retinopathy); TRD (tractional retinal detachment); PPV (pars plana vitrectomy); RD (retinal detachment); VH (vitreous hemorrhage)]

They found that intraoperative bleeding was less in the treatment group, vitreous clear-up time was significantly shorter, and fewer patients had prolonged vitreous clear-up. They detected no significant difference in either early or late post-operative vitreous hemorrhage. The mean time to vitreous clear-up was 7.2 days in the treatment group versus 15.2 days in controls (P = 0.04). No increased retinal traction occurred during the period between injection and surgery, and no retinal detachments occurred post-operatively.

This study provides support for the observations of Avery and colleagues. Besides its small sample size, it is limited by its heavy reliance on subjective grading scales to assess bleeding, clear-up, and the extent of pre-operative disease. The sample size in this paper is too small to detect a clinically relevant rate of worsening retinal traction after bevacizumab injection. However, the group showed a substantial benefit associated with this use of bevacizumab, and several other groups have subsequently reported similar findings.

The largest trial of bevacizumab as an adjunct to vitrectomy has been recently reported by Ahmadieh and colleagues [86]. They randomized 68 eyes scheduled to undergo PPV for PDR to either 1.25 mg intravitreal bevacizumab 1 week before PPV or to sham injection. Of these 68 eyes, just 34 completed the study, in large part because a substantial number of the bevacizumab eyes had a resolution of vitreous hemorrhage during the week after the injection (nine treatment eyes versus two control eyes, P = 0.028). They also found that fewer bevacizumab eyes had postoperative vitreous hemorrhage at both 1 week and 1 month. The rate of intraoperative bleeding was significantly less in the bevacizumab group (ten versus 17 eyes, P = 0.035), as was the mean frequency of intraoperative endodiathermy use. No significant treatment-related complications occurred. The group excluded eyes that underwent internal gas or oil tamponade, thereby eliminating a potentially confounding variable.

Rizzo and colleagues randomized 22 eyes with severe PDR and TRD to 1.25 mg intravitreal bevacizumab 5 to 7 days before PPV or to a control group [87]. They aimed primarily to assess the difficulty of the surgeries in the two groups by recording operative times, number of instrument exchanges, number and severity of intraoperative bleeds, dissection techniques, and intraoperative retinal tears. While they did not mask the investigators or perform inferential statistical analysis, the group found that each of these parameters was more favorable in the bevacizumab group. The mean number of tool exchanges was 27 (versus 53), and the number of intraoperative retinal tears was zero (versus four). Among their secondary outcome measures, visual acuity at the 6-month follow-up was 0.88 logMAR in the bevacizumab group and 2.01 logMAR in controls (p = 0.01). No worsening of retinal traction occurred after bevacizumab injection.

Yeh and colleagues also evaluated 1.25 mg bevacizumab as an adjuvant therapy 1 week before vitrectomy [88]. They prospectively enrolled 41 eyes with severe PDR and active fibrovascular proliferation extending to the periphery in at least three quadrants, and all eyes were infused with silicone oil at the conclusion of the PPV. The authors randomized these eyes to a bevacizumab group and a control group.

They found that their grading of intraoperative bleeding from proliferative tissue was significantly worse in the control group, but intraoperative subretinal hemorrhage was significantly more common in the bevacizumab group. In fact, nine of 20 eyes in the treatment group (versus one control) developed subretinal hemorrhage intraoperatively (P = 0.004). Immediate post-operative preretinal hemorrhage was significantly less in the bevacizumab group, blood reabsorption time was less (mean 11.1 versus 34.8 days), and the bevacizumab group had a greater improvement in vision at 6 months. One eye in the bevacizumab group converted from a TRD to a combined rhegmatogenous/TRD in the interval between the injection and vitrectomy, while no such progression occurred in the control group. At the final follow-up visit, one eye in the treatment group and two in the control group had recurrent TRD.

This study provides early data for bevacizumab before PPV with silicone oil infusion in patients with particularly severe PDR. Although an increased rate of intraoperative subretinal hemorrhage occurred in the bevacizumab group, the numerous potential benefits of the drug outweighed the observed adverse effects. The possibility of worsening retinal traction after bevacizumab injection is again raised by this study, but these events did not impact the 6-month outcomes. One may reasonably hypothesize that the risks of worsening traction are mitigated by the short time between injection and surgical relief of traction.

Lucena and colleagues conducted a study to assess quantitatively the amount of intraoperative bleeding in eyes injected with 1.5 mg bevacizumab 2 weeks before PPV for macula-involving TRD [89]. They concluded that less intraoperative bleeding occurred in treatment eyes after performing an automated count of red blood cells in the vitrectomy cassette. While this methodology is not foolproof, the greater than ten-fold difference in the RBC counts suggests some treatment effect.

Romano and colleagues have reported on the use of intravitreal bevacizumab before PPV for non-clearing VH in patients with DR [90]. They conducted a noncomparative trial, injecting 32 eyes with 2.5 mg bevacizumab 1 week before PPV. They then assessed the rate of recurrence of VH, visual acuity, and the incidence of complications over a 6-month follow-up period. Just one or two eyes had recurrent severe VH at any post-operative time point. Acuity improved significantly at 6 months. Twelve of 22 phakic eyes developed cataract and seven eyes developed posterior synechiae at 1 month, but no other adverse events occurred. The group subjectively reports less intraoperative bleeding. The study did not evaluate the extent of retinal traction before injection or before surgery.

These recent investigations indicate that bevacizumab is likely to prove a helpful adjunct to diabetic PPV for TRD, and potentially for non-clearing VH. Administration of the drug 1 or 2 weeks before PPV did not ultimately lead to any reported adverse outcomes in these small studies. The suggestion of earlier clearance of blood, fewer intraoperative complications, decreased operative time, less post-operative bleeding, and better visual outcomes warrants further study.

Bevacizumab as an adjunct to cataract surgery

Three groups have investigated bevacizumab as an adjunct to cataract surgery in patients with diabetes (Table 4). Cheema and colleagues randomized 68 eyes with any type of DR to 1.25 mg bevacizumab at the conclusion of cataract surgery or surgery with no injection [91]. At baseline, 39 eyes had CSME, and all were treated pre-operatively with focal laser. They found that 15 control eyes had progression in the severity of DR over 6 months versus only four treatment eyes (P = 0.002). They also found that significantly more control eyes had progression of macular edema. Despite this data, they detected no difference between the groups in visual acuity, CMT, or rates of post-operative laser treatment. Little baseline data is presented to establish the similarity of the two treatment groups; indeed, their treatment group has many more patients with severe and very severe NPDR. Nonetheless, the group tested a novel hypothesis and found an indication for further study of this drug.
Table 4

Trials of anti-VEGF drugs for prevention of diabetic macular edema after cataract surgery

Study

Intervention

Design

N

Follow-up

Population studied

Authors’ conclusions

Cheema et al. 2009 [91]

Bevacizumab at end of cataract surgery

Randomized, partially masked, controlled, single dose

68

6 months

Any DR; patients with previous PRP excluded

Bevacizumab prevented progression of DR and diabetic maculopathy

Takamura et al. 2009 [92]

Bevacizumab at end of cataract surgery

Randomized, double-masked, controlled, single dose

42

3 months

Type II DM, CRT > 300 μm, fovea-involving DME

Bevacizumab has beneficial short-term effects on retinal thickness and acuity

Lanzagorta-Aresti et al. 2009 [93]

Bevacizumab at end of cataract surgery

Randomized, controlled, single dose

26

6 months

Type II DM, moderate NPDR with fovea-involving diffuse DME

Bevacizumab prevents exacerbation of pre-existing DME

[Abbreviations: N (number of eyes); DME (diabetic macular edema); CRT (central retinal thickness); PRP (panretinal photocoagulation); DR (diabetic retinopathy); NPDR (non-proliferative diabetic retinopathy)]

Takamura and colleagues also injected 1.25 mg bevacizumab at the conclusion of cataract surgery in a cohort of diabetic subjects [92]. They studied 42 eyes of type 2 patients with pre-existing center-involving DME, randomizing these eyes to a treatment group and a control group. None of these eyes had laser treatment within 12 months of the surgery. While the control eyes had a significant worsening of retinal thickness post-operatively at both 1 and 3 months (mean 351.6 μm at baseline and 379.9 μm at 3 months), treatment eyes had a significant improvement versus pre-operative measurements (mean 355.0 μm at baseline and 330.0 μm at 3 months). Both groups had improved acuity at 1 and 3 months, but the improvement was significantly better in the treatment group.

Lanzagorta-Aresti and colleagues performed a similar study in patients with moderate NPDR and center-involving DME at the time of surgery [93]. Twenty-six eyes that underwent focal/grid laser followed by uncomplicated cataract surgery received either an intraoperative bevacizumab injection or a sham injection. The treatment group had a significant improvement in BCVA at 3 and 6 months, and no change in CMT. The sham group had a significant worsening of visual acuity versus baseline at 6 months, and a significant increase in CMT. The BCVA and CMT differences between the two groups were also significant at 3 and 6 months. The authors do not disclose their dose of drug. These small series suggest a powerful potential effect of just a single dose of bevacizumab. Future studies are needed to substantiate these findings, to establish safety in this population, and to determine long-term outcomes.

Conclusions

Early studies have suggested an important potential role for anti-VEGF agents in the treatment of diabetic retinopathy. For diabetic macular edema, intravitreal pegaptanib has some short-term benefit on visual acuity in treatment-naïve eyes. At 30 weeks, eyes in the 0.3 mg pegaptanib group had gained a mean of 5.4 letters. This is the nearest reported time point to the 6-month data of the READ-2 trial for ranibizumab, when ranibizumab-only eyes had gained a mean 7.2 letters. In both trials, the injection-only groups had better short-term outcomes than the combined injection/laser groups. The trials of both these medications suggest a need for phase III studies.

Intravitreal bevacizumab has some beneficial short-term effects on treatment-naïve eyes with DME and on eyes with laser-refractory DME. Bevacizumab treatment is associated with improvements in both acuity and CMT. Repeat dosing has been shown to be necessary, and the optimal timing is somewhere between 3 and 12 weeks. Reported side-effects for each of the anti-VEGF medications for DME have been rare and procedure-related rather than medication effects.

For PDR, early studies suggest that intravitreal bevacizumab decreases leakage from diabetic neovascular lesions in newly diagnosed and refractory disease. The effect is temporary, and it may be associated with TRD. Bevacizumab also holds promise for vitreous hemorrhage, post-PRP macular edema, and NVI. Several studies to date also indicate that bevacizumab is likely to prove a helpful adjunct to diabetic PPV for TRD, and potentially for non-clearing vitreous hemorrhage. Small studies have associated intravitreal bevacizumab with earlier clearance of blood, fewer intraoperative complications, decreased operative time, less post-operative bleeding, and better visual outcomes. Finally, three small series suggest a potential beneficial effect of a single dose of bevacizumab to prevent worsening of DME after cataract surgery.

Use of anti-VEGF medications for any of the above indications remains off-label. The safety and efficacy of these drugs for DR has not yet been established, and as such we eagerly await the results of large, controlled trials to address these questions.

Copyright information

© Springer-Verlag 2010