A meta-analysis of the effect of a dexamethasone intravitreal implant versus intravitreal anti-vascular endothelial growth factor treatment for diabetic macular edema

Macular edema (ME) is not an independent disease, but a common phenomenon in various retinal diseases in which fluid and protein accumulate in the extracellular space within the retina [1, 2]. Diabetic macular edema (DME) is macular thickening secondary to diabetic retinopathy (DR) that may be present in any of the stages of this disease, although it manifests more commonly in the non-proliferative diabetic retinopathy stage. In patients with DR, aged 20 to 79 years, the global prevalence for DME is 6.8% [3]. The prevalence of DME is reported to be related to the duration of the diabetes [4, 5]. DME is the foremost cause of central vision loss, and even blindness, and has a great influence on life quality of patients. Thus, reduction of ME may be associated with improved vision.

However, the treatment of DME remains controversial among vitreoretinal specialists. In addition to glycemic control, a variety of treatment alternatives exist for patients presenting with DME, including focal or grid photocoagulation, which has been the standard therapy since the 1970s, and more recently, intravitreal injection (anti-VEGF or corticosteroids) has been applied to DME, and vitrectomy in patients with DME with vitreomacular traction. As understanding of the pathophysiology of DME has improved, focal or grid photocoagulation is no longer the first choice for the treatment of DME. Treatment now targets the causal factor specifically, VEGF. To date, anti-VEGF drugs, including aflibercept, ranibizumab, and bevacizumab, have been proven in many clinical trials with efficacy for DME. The RESOLVE [6], RISE/RIDE [7, 8], and READ-2 [9] studies have all shown that ranibizumab is a good choice for the treatment of DME. The BOLT [10] randomized trial showed that bevacizumab is superior to laser monotherapy for persistent center-involving clinically significant macular edema.

Anti-VEGF drugs is still the first-line for treating DME, but also may impose a significant burden for patients who either do not have good respond to anti-VEGF or have recurrent ME, require frequent anti-VEGF injections [11]. Indeed, a study by the Diabetic Retinopathy Clinical Research Network (DRCR. Net), Protocol I revealed that more than 40% of ranibizumab-treated eyes still had CST ≥ 250 μm at 2 years post-treatment. The limited visual gains or resolution of DME in those naive patients are believed to be related to the pathophysiology of DME. Although the contribution of VEGF to the development of DME is indisputable, the role of other non-VEGF pathways has also been considered. Many studies have demonstrated that the inflammation is involved in the DR progression. With the increased recognition of the role of inflammation in the development of DME, sustained-release implants of steroids have shown good anatomical benefit and have also, to some extent, reduced the number of intravitreal injections that is needed in most cases.

Dexamethasone, it has the highest relative clinical efficacy of any corticosteroid applied to ophthalmological practice, and exerts its multiple-effects via its influence on multiple signal transduction pathways. The 0.7 mg intravitreal DEX implant is a biodegradable solid polymer drug-delivery system (Ozurdex®, Allergan, Inc.), which uses the following characteristic mode of drug release by diffusion, in a biphasic fashion: an initial high-concentration phase and a second low-concentration phase, which facilitates continued efficacy of the treatment for up to 6 months [12]. The U.S. Food & Drug Administration (FDA) first approved the Ozurdex for retinal vein occlusion-induced ME treatment in 2009. It was then approved for non-infectious uveitis treatment. In 2014, the FDA and most European countries approved Ozurdex for the treatment of DME, based on results of the MEAD study [13]. Studies of treatment of DME have demonstrated that Ozurdex may be an alternative treatment for patients who do not have good respond to serial anti-VEGF injections or in recalcitrant cases [14‐19]. In addition, Ozurdex may be considered as primary treatment for DME [20].

To date, no systematic review has discussed the therapeutic effect and safety of intravitreal anti-VEGF versus DEX implant in DME. We performed a systematic review and meta-analysis to quantify the effect of these two treatments on BCVA and CST in DME. Additionally, we report the adverse events described with these therapies.

In our study, we evaluated four RCTs to evaluate the efficacies of DEX implants and anti-VEGF in the treatment of DME. We found that both DEX implant and anti-VEGF could achieve significant functional and anatomical improvement during early treatment. We did not find statistically significant differences in terms of BCVA and with no statistical heterogeneity between these two treatment arms at 6 months. However, the anti-VEGF group revealed significant improvement in BCVA at 12 months, compared to the DEX implant group, and without statistical heterogeneity. These findings are in line with the Protocol I in DRCR.​net, whereby the ranibizumab group had better visual acuity than the triamcinolone group.

Even with good treatment effects, repeated injections carry increased risks, such as infectious endophthalmitis, intraocular inflammation, and even stroke or myocardial infarction [28]. Therefore, anti-VEGF treatment may not be a good therapy for all patients. DME has been shown to be a complex multifactorial disease; in addition to VEGF, inflammation may be another pathophysiological feature of these treatment-naive patients. Diabetics are found to have high concentrations of pro-inflammatory cytokines, such as interleukin 6 (IL-6), IL-1β, tumor necrosis factor α (TNFα), and intracellular adhesion molecule (ICAM)-1. All of these cytokines induce retina with persistent chronic inflammation, which leads to leukostasis, increased vascular permeability and dysfunction of the blood-retinal barrier (BRB) [29, 30]. Additionally, according to an investigation by Jonas et al. [31], DME was shown to be related to elevated cytokines in the aqueous humor or vitreous, such as ICAM-1, IL-6, transforming growth factor beta (TGF-β) and monocyte chemotactic protein 1 (MCP-1). Among the cytokines, ICAM-1 is closely related with diabetes characteristics. Thus, treatment principally aims to block the effect of these two pathogenic pathways. Corticosteroids have anti-inflammatory, anti-permeability, and angiostatic effects when treating DME [32]. Thus, DEX implant could be a better alternative for DME.

Our analysis demonstrated that DEX implant treatment could significantly reduce CST at 6 months, compared to anti-VEGF treatment. Unfortunately, this effect can not last until 12 months. This can be explained by the characteristics of the DEX implant. There are two phases in the DEX implant treatment. Higher concentrations of DEX are found in the initial phase, followed by lower concentrations in the final phase. The drug reaches its peak efficacy less than two months after administration. Afterwards, the drug continues to provide treatment, but, at lower levels in months 2–6 [12]. Regarding heterogeneity, we could not perform a subgroup analysis to interpret the potential source of heterogeneity, due to the limited studies. We suspect that, at 6 months, the heterogeneity mainly derives from the small sample size of the study by Gallemore et al. Moreover, the results of CST may be affected by other factors, including initial retinal thickness, the dosage of bevacizumab, and previous treatment (laser or anti-VEGF). The mean change in CST at 12 months showed heterogeneity mainly caused by the type of anti-VEGF drugs used. In the Allergan 2015 study [27], patients received 0.5 mg ranibizumab. However, in the BEVORDEX study [23], 0.5 mg bevacizumab was given to the patients. It has been reported that ranibizumab and bevacizumab do not differ in terms of functional outcomes, but that ranibizumab was more effective in terms of CST reduction [33]. Factors such as race, dosage, age, and baseline states may have attributed to heterogeneity. To some extent, these factors were unavoidable. In terms of CST, at 6 months, Ozurdex showed superior reduction in CST, which indicates that patients with DME can achieve superior anatomical outcomes with fewer injections. Similar to our results, the results from the MEAD study evaluating the DEX implant showed that a mean number of 4.1 DEX treatments were administered after 3 years of follow-up. Compared to the sham group, the DEX group acquired a CST reduction of 112 μm, nearly three times than the sham group (42 μm).

Notably, the improved anatomical outcomes in our data did not translate to improved visual acuity outcomes. Results from other clinical trials were consistent with the four studies included here. In the CHROME study [34], significant reduction in retinal thickness were occurred in the Ozurdex group: 190.9 ± 23.5 μm for DME (P < 0.0001). The greatest mean improvements in BCVA in terms of number of lines of vision observed in the eyes with DME (0.7 ± 0.5; P > 0.05). Similarly, compared to monthly IVB monotherapy (− 30 μm), the combination therapy with DEX and bevacizumab led to a significantly greater change of CST (− 45 μm) in the study by Maturi et al.; nervertheless, visual acuity improvement was not statistically significant [35]. The limited visual gains in these patients may have been related to the duration of macular edema, neural damage, retinal pigment epithelium changes, and subretinal fibrosis caused by chronic macular edema prior to treatment, as well as the result of structural damage from repeated macular laser therapy, and the natural course of DR [7].

Additionally, we suspect that it may be due to DEX-induced progression of cataract, based on previous studies. In the DRCR Protocol I, when controlling for cataract formation, in pseudophakic eyes, the group that received triamcinolone observed similar BCVA results to the ranibizumab group. Similar to this study, the CHROME study also showed that the pseudophakic eyes with DME acquired a average gain of 1.4 ± 0.5 lines, but a average loss of 0.6 ± 0.6 lines was shown in phakic eyes [34]. These results are also in agreement with those of the BEVORDEX study, which found exhibited no significant effect for pseudophakic eyes. Compared to 10.4 letters in the dexamethasone group, the mean change in BCVA for the bevacizumab group was 7.7 letters (P = 0.47). Moreover, visual acuity may be affected by post-operative complications, such as Irvine−Gass syndrome or cataract secondary to pars plana vitrectomy prior to DEX or anti-VEGF treatment.

In addition to having different efficacies for treating DME, DEX implants and anti-VEGF injection are associated with varying degrees of increased risk of systemic and/or local complications over the period of treatment. The systemic adverse event rate has been reported to be higher with anti-VEGF treatment in some clinical trials [36]. In the study by Avery et al., the increased potentially cerebrovascular accidents may be link to anti-VEGF treatment, especially after two years therapy [37]. Similar to previous studies, our data demonstrated that a lower incidence of SAEs in the DEX implant group, but this was not statistically significantly different between the DEX and anti-VEGF group. These results suggest that we should be cautious in using anti-VEGF in patients with myocardial infarction and stroke [38]. The deterioration of hypertension was the most frequent systemic adverse event encountered in the BEVORDEX study. Other adverse events, such as cardiac disorders, also occurred in the studies included in this meta-analysis, except in the study by Gallemore et al. High IOP and secondary cataract are the most common ocular adverse events of DEX implants. Our meta-analysis agrees with these findings, which demonstrated a statistically significant difference between the two groups in terms of increased IOP and cataract. The groups receiving DEX had a higher risk of a rise in IOP and cataract progression than the anti-VEGF groups for DME. This suggests that the ophthalmologist should take care when using DEX implants in patients with high IOP or in young patients with a clear lens.

The relative benefits and costs should also be considered when applying therapies. In terms of cost-effectiveness, intravitreal corticosteroid injections are relatively cheaper than anti-VEGF therapies, although this is not true for ranibizumab. One study calculated the amount of money (USD) saved per line of visual acuity improvement for each of the various therapies as follows: DEX implant—$5666, bevacizumab—$1329–$2246, and ranibizumab—$11,372–$11,609. In addition, the dollars per quality-adjusted life year for these therapies were as follows: DEX implant—$9446, bevacizumab—$2013–$4260, and ranibizumab—$19,251–$23,119 [39].

Currently, dexamethasone delivery systems and anti-VEGF therapies have a positive effect on the course of DME. However, these two different types of drugs have different pharmacological properties and side-effect characteristics. Given the results of clinical trials and the pathophysiology of DME, Ozurdex is considered to be the preferred treatment for patients who have chronic DME and are anti-VEGF-resistant, as an alternative to switching between anti-VEGF drugs [40, 41]. Ozurdex may be recommended as a first choice for the following cases: 1. pseudophakic eyes, or patients who are under consideration for cataract surgery in the near future; 2. patients who are anti-VEGF-resistant [42]; 3. patients who have a history of cardiovascular and cerebrovascular diseases [42]; 4. post-vitrectomy patients [18]; 5. patients without a high IOP risk at baseline; 6. patients who are reluctant to receive frequent injections. In all cases, the IOP should to be monitored frequently. Reinjection of Ozurdex can be considered after approximately 3−6 months if remains evidence of impaired vision and residual ME.

Our study was limited by the following factors: (1) We only included four studies assessing a total of 521 eyes. (2) The clinical trails duration was quite short in some of the that were included, and thus we may have underestimated the drug-induced adverse events. (3) Heterogeneity was inevitable due to the different regimens of anti-VEGF therapies used. Previous studies indicated that the efficacy of aflibercept, ranibizumab or bevacizumab for DME was different but the relative efficacy depended on baseline BCVA. At the 1-year follow-up, aflibercept exhibited some advantage over bevacizumab and ranibizumab, especially among patients with an initial baseline BCVA letter score of less than 69. However, ranibizumab and bevacizumab did not show significant differences [28, 33]. A previous meta-analysis study also confirmed that bevacizumab and ranibizumab did not differ in terms of BCVA, but ranibizumab was more effective in terms of CST reduction with a low-certainty of evidence [43]. To reinforce the validity of our meta-analysis, clinical trials comparing the 3 anti-VEGF agents with the DEX implant as well as extended follow-up trials should be conducted in the future.

In summary, this meta-analysis of data from four randomized clinical trials revealed that despite some ocular adverse events, DEX-treated eyes had relatively superior anatomic outcomes compared with anti-VEGF, and showed similar rates of vision improvement, while requiring fewer injections, especially in pseudophakic patients. However, considering for the restrictions of indications, the DEX implant may not be recommended as a first-line therapy for DME. In the future, randomization of these treatments would allow a definite conclusion about whether switching to a DEX implant is more beneficial rather than anti-VEGF treatment. Additionally, new treatments (monotherapy or combined therapy) should be investigated to optimize clinical efficacy and reduce side-effects.