Short-Term Outcomes of Trabeculectomy With or Without Anti-VEGF in Patients With Neovascular Glaucoma: A Systematic Review and Meta-Analysis

Abstract

Objectives: The aim of this study was to compare the safety and efficacy of trabeculectomy alone or combined with intravitreal injections of anti–vascular endothelial growth factor (anti-VEGF) agents for the treatment of neovascular glaucoma.

Methods: We conducted a systematic review and meta-analysis to compare the effects of trabeculectomy alone or combined with intravitreal injections of anti-VEGF agents for the treatment of neovascular glaucoma. We searched four databases (PubMed, Cochrane Library, Embase, and Web of Science) up to January 2023 and extracted data on three surgical outcomes: postoperative intraocular pressure, success rate and complications. We used a random-effects model to calculate pooled relative risk (RR) or standardized mean difference (SMD) estimates and 95% confidence intervals (CIs). We assessed publication bias using Begg and Egger tests.

Results: We included seven studies with 353 eyes. Compared to trabeculectomy alone, trabeculectomy with anti-VEGF had a lower risk of postoperative complications (RR, 0.60; 95% CI, 0.41–0.89) and higher success rate (RR, 1.19; 95% CI, 1.02–1.40). The intraocular pressure reduction was significantly greater in the trabeculectomy with anti-VEGF augmentation group than the trabeculectomy group from 1 week (SMD, −1.36; 95% CI, −2.76 to 0.04) to 6 months (SMD, −0.79; 95% CI, −1.50 to −0.07) after surgery.

Conclusions: According to current evidence, adding intravitreal injection of anti-VEGF agents to trabeculectomy may improve the short time outcomes of patients with neovascular glaucoma.

Introduction

Neovascular glaucoma (NVG) is a vision-threatening type of secondary glaucoma. It can lead to blindness.¹ NVG features neovascularization in the iris, high intraocular pressure (IOP), poor visual prognosis, and hyperplasia of fibrovascular tissue in the anterior chamber.² Among secondary glaucoma cases, the prevalence of NVG ranges from 9% to 17.4%. NVG results from various eye lesions that cause retinal ischemia, which is the main pathogenesis of most NVG cases.² 

NVG has a high risk of poor prognosis. The management of patients with NVG should aim to (1) reduce ocular ischemia and treat its underlying cause, (2) lower elevated IOP effectively, and (3) control the inflammation process. Prevention and early treatment are the best strategies for NVG. The treatment of patients with NVG should be done with caution. 

NVG results from posterior ocular ischemia, which causes neovascularization that obstructs the aqueous humor outflow.³ The main cause of NVG is retinal ischemia due to vascular occlusion or diabetic retinopathy.⁴ It involves the formation and growth of a fibrovascular membrane on the iris and anterior chamber surface.⁵ According to the European Union figures, NVG affects 75,000 to 113,000 people in Europe. Patients with NVG make up about 3.9% of all patients with glaucoma.⁶ In the United States, about 17,500 diabetic patients have iris neovascularization. Most of them have proliferative diabetic retinopathy.⁷ The incidence of iris neovascularization in patients with proliferative diabetic retinopathy is as high as 65%.⁸ The incidence of diabetic iris neovascularization without proliferative diabetic retinopathy is only 1% to 17%.^9,10 

Ocular neovascularization in patients with NVG is a slow and complex process that involves multiple angiogenic factors. It occurs mainly due to the imbalance between proangiogenic factors (such as vascular endothelial growth factor [VEGF]) and antiangiogenic factors (such as pigment epithelium–derived factor).¹¹ VEGF has a key role in promoting intraocular neovascularization in patients with ischemic retinal diseases.¹² Other potential angiogenic initiators have been studied previously, such as interleukin 6,¹³ basic fibroblast growth factor,¹⁴ transforming growth factor β1 and β2,¹⁵ nitric oxide,¹⁶ and endothelin 1.¹⁷ 

VEGF has a major role in the formation of ocular neovascularization. Some studies have suggested that anti-VEGF therapy can improve the prognosis of patients with NVG.^18–20 In 2020, intravitreal injection of aflibercept was approved as a new treatment for patients with NVG.²¹ Moreover, the clinical guidelines for glaucoma by the Glaucoma Society of Japan recommend intravitreal injection of anti-VEGF before or after NVG surgery to prevent intraoperative or postoperative complications of glaucoma surgery.²² 

Currently, several randomized controlled trials and retrospective comparative studies have evaluated the postoperative outcomes of trabeculectomy alone or combined with intravitreal injection of anti-VEGF. However, there is a lack of systematic reviews and meta-analyses to analyze the postoperative effects of these two treatments in patients with NVG comprehensively. 

Therefore, we performed a meta-analysis to compare the two different arms in terms of postoperative IOP, surgical success rate, and postoperative complications. The aim of this study was to assess the efficacy and safety of trabeculectomy alone or combined with anti-VEGF and to assist ophthalmologists in developing better treatment strategies for patients with NVG. 

Methods

We followed the process and methods suggested by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines to conduct this systematic review and meta-analysis.²³ This study design has been registered in PROSPERO (CRD42023388416). 

Data Sources

We searched PubMed, Cochrane Library, Embase, and Web of Science for relevant literature published until January 1, 2023. We did not apply any restrictions on publication year, language, or author. We used Medical Subject Headings and keywords for searching, such as “neovascular glaucoma,” “neovascular,” “vascular endothelial growth factors,” “VEGFs,” “brolucizumab,” “bevacizumab,” “ranibizumab,” “conbercept,” “aflibercept,” and “trabeculectomy.” We also checked the references of the retrieved publications to identify any potential additional relevant publications that were missed by the PubMed search (Supplementary Materials 1–4). Finally, we included seven English publications. One of them was a randomized controlled trial and the other six were retrospective comparative studies. 

Inclusion Criteria

Study inclusion criteria were based on the following criteria: (1) study design: randomized controlled trials or retrospective comparative studies; (2) study population: patients with NVG diagnosed based on symptoms, signs, and ophthalmic testing; (3) intervention factor: studies that compared trabeculectomy with or without intravitreal anti-VEGF agents in patients with NVG; (4) outcome variable: studies that reported relevant outcomes such as IOP, success rate, or complications or could provide at least one of these outcomes from the authors; or (5) studies that provided sufficient data for meta-analysis or could obtain them from the authors. 

Exclusion Criteria

Studies with the following criteria were excluded: (1) small sample size (n < 10), (2) drug dose-related interventions, (3) interventions not meeting predefined standards (e.g., anti-VEGF alone versus anti-VEGF augmented trabeculectomy), (4) conference abstracts or full texts without raw data, (5) reviews or letters without original data, or (6) duplicate publications. 

Outcome Measure

The primary outcome measure was the postoperative IOP between the two different treatments. We used the mean difference or standardized mean difference (SMD) to analyze postoperative IOP as a continuous variable using the random-effects model. We used the outcome data of the last time or closest to the follow-up if the required data were not reported at the end of the follow-up duration. The secondary outcome measures were the surgical success rate and postoperative complications. We used the relative risk (RR) to calculate an overall RR for each outcome using the random-effects model. We adopted the definitions of surgical success rate and postoperative complications as defined by the authors of the studies. We used the available data at final follow-up for statistical analysis if no common time point of surgical success rate and postoperative complication was reported. 

Data Extraction

Two researchers independently extracted the following data from the eligible studies: (1) demographic characteristics: such as the author’s name, the publication year, the study design, the sample size, the mean age, the gender ratio, and the follow-up duration and (2) clinical outcomes: such as pre- and postoperative IOP means and mean difference or SMD. We collected postoperative IOP data at different time points postoperatively, such as 1 week, 1 month, 3 months, 6 months, and 12 months. We also extracted other surgical outcomes like success rate and postoperative complications. We resolved any discrepancies by consensus or by consulting the other reviewer. We assessed the data quality using the Cochrane risk of bias tool. 

Quality Assessment

We used the Cochrane bias risk tool to assess the quality of randomized controlled trial. We used the Newcastle-Ottawa Scale to assess the quality of retrospective comparative studies. In this scale, we scored poor, medium, and good quality as 0 to 3, 4 to 6, and 7 to 9, respectively. We included studies with the Newcastle-Ottawa Scale scores above 4 points in the final analysis. Two authors independently performed the quality assessment and resolved any disagreement by discussion or consultation with the other author. We used the leave-one-out method to perform sensitivity analysis to evaluate the impact of individual studies on the pooled results. 

Statistical Analysis

The Stata software/MP, version 17.0 (Stata Corporation, College Station, TX, USA), was used for all statistical analyses. Continuous variables were analyzed using SMD and 95% confidence intervals (CIs), while dichotomous variables were analyzed using RR and 95% CI. We assessed heterogeneity between studies using the Cochran I² statistic, with I² values of 25%, 50%, and 75% indicating low, moderate, and high heterogeneity, respectively.^24,25 We used a random-effects model to pool the results across studies. We assessed publication bias by constructing a funnel plot of each study's effect size against the standard error and testing for funnel plot asymmetry using the Begg and Egger tests. A P value <0.05 was considered indicative of significant publication bias.²⁶ 

Results

Screening of the Relevant Studies

We identified 376 references in the initial search. Figure 1 shows the filtration process. We excluded 90 references for duplication, 211 references based on titles and abstracts, 44 references for noncomparative design, 3 references for incorrect trial design, 12 references for inappropriate intervention, and 9 references for review or case report. We included seven English studies in the final data extraction and analysis. 

Figure 1.

 

Flowchart of literature selection for this meta-analysis.

Characteristics of the Included Studies

We included seven studies that compared trabeculectomy assisted by mitomycin with or without anti-VEGF augmentation for 525 patients. The sample size ranged from 14 to 88. One study was a randomized controlled trial and six were retrospective comparative studies. All studies used the same trabeculectomy surgical technique for both groups, except that the experimental group received an intravitreal injection of anti-VEGF. The control group had 271 eyes that underwent trabeculectomy alone, and the experimental group had 254 eyes that received trabeculectomy with anti-VEGF intravitreal injection. 

The Table shows the demographic characteristics of the patients. Follow-up duration ranged from 6 to 12 months across the studies. Three studies followed up for about 6 months and four studies for about 12 months. The authors defined surgical success criteria, and they varied among the studies (see Supplementary Material 6). 

Quality Assessment

The only randomized controlled trial used a computer-generated random block permutation method stratified by age and gender for randomization.²⁷ Nilforushan et al.²⁷ reported the trial as open-label rather than double-blind. The Cochrane bias risk tool was used to assessed its quality. The detailed results are in Supplementary Material 7. As for the retrospective comparative studies, we used the Newcastle-Ottawa quality assessment scale to evaluate the quality these studies. The scoring details are shown in Supplementary Material 8. 

Results of the Meta-Analysis

Postoperative IOP

We analyzed postoperative IOP data from four studies involving 373 eyes to evaluate the effect of trabeculectomy with anti-VEGF augmentation in patients with NVG. The other studies were excluded because they did not report the means or standard deviations of IOP. We compared and analyzed the experimental group and the control group outcomes at six time points: baseline, 1 week, 1 month, 3 months, 6 months, and 12 months postoperatively. Forest plots were generated to illustrate the results. No significant difference in IOP between the two groups was found at baseline. However, the difference in IOP between the two groups increased from 1 week (SMD, −1.36; 95% CI, −2.76 to 0.04; P = 0.06) to 6 months (SMD, −0.79; 95% CI, −1.50 to −0.07; P = 0.03) postoperatively and then decreased at 12 months (SMD, −0.30; 95% CI, −1.61 to 1.02; P = 0.66) postoperatively. This suggests that trabeculectomy with anti-VEGF augmentation achieved better outcomes than trabeculectomy alone within 6 months postoperatively, but the effect may diminish after 6 months (Figs. 2–7). 

Figure 2.

 

Forest plot of preoperative IOP in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma. The forest plot shows the meta-analysis results of the data on IOP of preoperative baseline in the included studies. The SMD and its 95% CI are shown for each study and the pooled estimate. RCT, randomized controlled trial.

Figure 3.

 

Forest plot of IOP reduction in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma at 1 week postoperatively. The forest plot shows the meta-analysis results of the data on IOP of 1 week postoperatively in the included studies. The SMD and its 95% CI are shown for each study and the pooled estimate.

Figure 4.

 

Forest plot of IOP reduction in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma at 1 month postoperatively. The forest plot shows the meta-analysis results of the data on IOP at 1 month postoperatively in the included studies. The SMD and its 95% CI are shown for each study and the pooled estimate. The pooled estimate shows a significant difference in IOP reduction between the two groups.

Figure 5.

 

Forest plot of IOP reduction in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma at 3 months postoperatively. The forest plot shows the meta-analysis results of the data on IOP at 3 months postoperatively in the included studies. The SMD and its 95% CI are shown for each study and the pooled estimate. The pooled estimate shows a significant difference in IOP reduction between the two groups.

Figure 6.

 

Forest plot of IOP reduction in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma at 6 months postoperatively. The forest plot shows the meta-analysis results of the data on IOP at 6 months postoperatively in the included studies. The SMD and its 95% CI are shown for each study and the pooled estimate. The pooled estimate shows a significant difference in IOP reduction between the two groups.

Figure 7.

 

Forest plot of IOP reduction in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma at 12 months postoperatively. The forest plot shows the meta-analysis results of the data on IOP at 12 months postoperatively in the included studies. The SMD and its 95% CI are shown for each study and the pooled estimate.

Complications

We also analyzed postoperative complications. All seven studies involving 525 eyes were included in this analysis. We compared and analyzed the total number of postoperative complications between the trabeculectomy with anti-VEGF augmentation group and the trabeculectomy group, as well as the individual complications reported in at least two studies. 

We performed a meta-analysis to compare the rates of different complications (Figs. 8–15). We only included complications reported in more than two studies in the meta-analysis. We separately examined the following eight complications: bleb leakage (RR, 1.20; 95% CI, 0.13–11.29; P = 0.87), phthisis bulbi (RR, 0.29; 95% CI, 0.03–2.56; P = 0.27), shallow anterior chamber (RR, 0.50; 95% CI, 0.21–1.21; P = 0.12), anterior chamber exudation (RR, 1.33; 95% CI, 0.14–12.43; P = 0.80), anterior chamber hemorrhage (RR, 0.48; 95% CI, 0.24–0.98; P = 0.04), choroidal detachment (RR, 1.26; 95% CI, 0.66–2.39; P = 0.49), maculopathy with low intraocular pressure (RR, 1.18; 95% CI, 0.26–5.24; P = 0.83), and recurrence of NVG (RR, 0.45; 95% CI, 0.10–1.92; P = 0.28). According to the analysis results, only the incidence of anterior chamber hemorrhage was significantly lower in the trabeculectomy with anti-VEGF augmentation group than in the trabeculectomy group. The incidence of the other seven complications was not significantly different between the two groups. 

Figure 8.

 

Forest plot of postoperative complications of bleb leak in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma. The forest plot shows the meta-analysis results of the data on postoperative complications of bleb leak from the included studies. The horizontal lines represent the 95% CI. The diamonds represent the pooled RR and its 95% CI.

Figure 9.

 

Forest plot of postoperative complications of phthisis bulbi in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma. The forest plot shows the meta-analysis results of the data on postoperative complications of phthisis bulbi from the included studies. The horizontal lines represent the 95% CI. The diamonds represent the pooled RR and its 95% CI. The pooled estimate shows a significant difference in postoperative hyphema between the two groups.

Figure 10.

 

Forest plot of postoperative complications of shallow anterior chamber in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma. The forest plot shows the meta-analysis results of the data on postoperative complications of shallow anterior chamber from the included studies. The horizontal lines represent the 95% CI. The diamonds represent the pooled RR and its 95% CI. The pooled estimate shows a significant difference in postoperative hyphema between the two groups.

Figure 11.

 

Forest plot of postoperative complications of anterior chamber exudation in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma. The forest plot shows the meta-analysis results of the data on postoperative complications of anterior chamber exudation from the included studies. The horizontal lines represent the 95% CI. The diamonds represent the pooled RR and its 95% CI.

Figure 12.

 

Forest plot of postoperative complications of hyphema in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma. The forest plot shows the meta-analysis results of the data on postoperative complications of hyphema from the included studies. The horizontal lines represent the 95% CI. The diamonds represent the pooled relative risk and its 95% CI. The pooled estimate shows a significant difference in postoperative hyphema between the two groups.

Figure 13.

 

Forest plot of postoperative complications of choroidal detachment in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma. The forest plot shows the meta-analysis results of the data on postoperative complications of choroidal detachment from the included studies. The horizontal lines represent the 95% CI. The diamonds represent the pooled RR and its 95% CI.

Figure 14.

 

Forest plot of postoperative complications of hypo-ocular pressure macular disease in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma. The forest plot shows the meta-analysis results of the data on postoperative complications of hypo-ocular pressure macular disease from the included studies. The horizontal lines represent the 95% CI. The diamonds represent the pooled RR and its 95% CI.

Figure 15.

 

Forest plot of postoperative complications of recurrence of neovascularization in trabeculectomy combined with anti-VEGF and trabeculectomy alone for neovascular glaucoma. The forest plot shows the meta-analysis results of the data on postoperative complications of recurrence of neovascularization from the included studies. The horizontal lines represent the 95% CI. The diamonds represent the pooled RR and its 95% CI.

Figure 16 shows the meta-analysis results of the total number of complications. We compared and reported the number of complications in the trabeculectomy with anti-VEGF augmentation group and the trabeculectomy group (RR, 0.60; 95% CI, 0.41 to 0.89; P = 0.01). We found a statistically significant difference in total postoperative complications between trabeculectomy with anti-VEGF augmentation and trabeculectomy alone, with fewer complications in the former group than in the latter group. 

Figure 16.

 

Comparison of total postoperative complications between trabeculectomy combined with anti-VEGF and trabeculectomy alone in neovascular glaucoma. A meta-analysis of the total number of postoperative complications reported in the included studies was performed. The figure shows the RR and its 95% CI for each study (squares and horizontal lines) and the pooled RR and its 95% CI (diamond). The pooled estimate indicates a significant difference in the total number of complications between the two groups, with trabeculectomy augmented by anti-VEGF agents having a lower incidence than trabeculectomy alone.

Surgical Success Rate

Another outcome measure was the surgical success rate. We compared and reported the postoperative success rate between the trabeculectomy with anti-VEGF augmentation group and the trabeculectomy group (RR, 1.19; 95% CI, 1.02–1.40; P = 0.03), and Figure 17 shows the results. We observed a statistically significant difference in postoperative success rate between trabeculectomy with anti-VEGF augmentation and simple trabeculectomy, with a higher success rate in the former group than in the latter group. 

Figure 17.

 

Meta-analysis of the effect of trabeculectomy combined with anti-VEGF and trabeculectomy alone on the surgical success rate for neovascular glaucoma. The RR and 95% CI were calculated for each study and pooled using a random-effects model. The figure shows the RR and its 95% CI for each study (squares and horizontal lines) and the pooled RR and its 95% CI (diamond). The pooled estimate indicates a significant difference in the surgical success rate between the two groups, with trabeculectomy augmented by anti-VEGF having a higher rate than trabeculectomy alone.

IOP data from 6 to 12 months after surgery showed significant heterogeneity in the study by Takihara et al.,²⁸ but the number of included studies was too small to allow subgroup meta-analysis of the studies on either side of the null line in the forest plot. Therefore, we explored the possible reasons for this high heterogeneity study. We found no significant bias in this study except for relatively higher preoperation IOP of participants and longer follow-up duration in the Takihara et al.²⁸ study. The mean preoperative IOP was 40.3 ± 10.8 mm Hg in the trabeculectomy with anti-VEGF augmentation group and 35.8 ± 10.6 mm Hg in the trabeculectomy group in the Takihara et al.²⁸ study. The IOP values of the two groups of patients in this study differed significantly from those reported in other studies. 

Moreover, female patients were underrepresented in this study. The male-to-female ratio was 44:6, which deviated considerably from other studies. Except for the ones described above, other factors in this study were comparable to those reported in other studies. One possible explanation for these findings is that anti-VEGF drugs may be less effective in male patients than in female patients. Another possibility is that the high preoperative IOP in the trabeculectomy with anti-VEGF augmentation group may lead to a higher postoperative IOP than in the trabeculectomy group, which may increase the heterogeneity of the Takihara et al.²⁸ study in this meta-analysis. 

The Takihara et al.²⁸ study was the first large comparative study on the benefits and limitations of anti-VEGF drugs with filtering surgery for NVG. Their longest follow-up time was 35 months. Anti-VEGF drugs lowered IOP only in the early postoperative period after trabeculectomy and did not affect the long-term success rate. Takihara et al.²⁸ suggested that this might be due to the transient nature of anti-VEGF drugs. 

This meta-analysis has a small sample size, which limits the strength of the conclusions and requires cautious interpretations of the results. More long-term studies are needed to determine the clinical significance of the postoperative difference between trabeculectomy with anti-VEGF augmentation and trabeculectomy alone. 

Publication Bias

We assessed publication bias with funnel plots and Begg and Egger tests (Fig. 18). We used data on postoperative IOP at 6 months for analysis because two studies did not report data at 12 months postoperatively. The results of the Begg test (P = 0.31) and Egger test (P = 0.32) for postoperative IOP at 6 months showed no significant publication bias. The Begg and Egger tests for operative success rate and complications rate also showed no significant bias, with P values of 0.76 and 0.55, and 1.00 and 0.93, respectively. 

Figure 18.

 

Publication bias analysis of the included studies. The funnel plots of the SMD or the RR were used to assess the publication bias for postoperative IOP (A), success rate (B), and total complications (C), respectively. The Begg and Egger methods were applied to test the asymmetry of the funnel plots. No significant publication bias was detected. OR, odds ratio.

Adverse Events

This review analyzed the occurrence of 14 complications, such as bleb leak, hypotony, and choroidal detachment. The statistics showed that anterior chamber hemorrhage (42 cases) and anterior chamber exudation (2 cases) were the most and least common complications, respectively. The probabilities of each complication were relatively low but varied among studies. The types of complications reported also differed among studies, which should be interpreted with caution. 

The Table in Supplementary Material 9 shows the specific incidence of each complication in the 7 studies. One study reported seven cases of postoperative ocular pain, with five in the trabeculectomy group and two in the trabeculectomy with anti-VEGF augmentation group. Another study reported low intraocular pressure after surgery, with five in the trabeculectomy group and two in the trabeculectomy with anti-VEGF augmentation group. Two studies reported leaking filtering bleb, with one in each group. Three studies reported intraocular hypertension, with two in the trabeculectomy group and none in the trabeculectomy with anti-VEGF augmentation group. Two studies reported ocular atrophy, with three in the trabeculectomy group and none in the trabeculectomy with anti-VEGF augmentation group. 

Table.

 

Characteristics of the Included Studies

 

Eight cases of visual acuity decrease were reported in total, with six in the trabeculectomy group and two in the trabeculectomy with anti-VEGF augmentation group. One study reported three cases of corneal edema, all in the trabeculectomy with anti-VEGF augmentation group. Four studies reported shallow anterior chamber, with 15 in the trabeculectomy group and 7 in the trabeculectomy with anti-VEGF augmentation group. Two cases of anterior chamber exudation were reported, with one in each group. Forty-two cases of anterior chamber hemorrhage were reported, with 29 in the trabeculectomy group and 13 in the trabeculectomy with anti-VEGF augmentation group. 

Moreover, one study reported four cases of vitreous hemorrhage, with two in each group. Two studies reported 27 cases of choroidal detachment, with 11 in the trabeculectomy group and 16 in the trabeculectomy with anti-VEGF augmentation group. Two studies reported seven cases of hypotony, with three in the trabeculectomy group and four in the trabeculectomy with anti-VEGF augmentation group. Three studies reported 23 cases of neovascular glaucoma recurrence, with 17 in the trabeculectomy group and 6 in the trabeculectomy with anti-VEGF augmentation group (Supplementary Material 9). 

Discussion

This review evaluated the safety and effectiveness of trabeculectomy with or without anti-VEGF augmentation for NVG and concluded that trabeculectomy with anti-VEGF augmentation is superior in short-term outcomes. 

There is no consistent standard for defining the surgical success or failure of NVG, and it is also difficult to collect an appropriate cohort.²⁹ The criteria for surgical success are largely based on previous studies and personal experience. After reviewing literatures, we found that IOP is one of the most important criteria for assessing the success of surgery. Other criteria included the percentage reduction in IOP, the type of antiglaucoma medication compared between pre- and postsurgery, and so on. Different criteria for assessing surgical success may affect the results of this study. In the future, developing a single set of criteria to evaluate the success of glaucoma surgery may help to advance glaucoma treatment research. Supplementary Material 6 lists the metrics for surgical success evaluation of included studies in this meta-analysis. 

Patients with NVG have a poor response to antiglaucoma drugs, severe complications, and low success rates of conventional trabeculectomy surgery. Therefore, drainage valve implantation has become an increasingly popular option for treating NVG.³⁰ Drainage valve implantation is also considered the gold standard for treating NVG.³¹ However, relevant research shows that the success rate of drainage valve implantation is still not satisfactory enough.^32,33 Bowden et al.³⁴ conducted a pooled analysis of three prospective, multicenter, randomized clinical trials to investigate the risk factors associated with shunt surgery failure. Their results showed that the cumulative probability of failure after tube shunt surgery was 38.3% after 5 years. They also identified some predictors of tube shunt failure, and NVG was one of the significant reasons. 

Previous clinical studies have reported conflicting results on the comparative efficacy of drainage valve implantation and trabeculectomy for patients with NVG. While Tokumo et al.³⁵ found that trabeculectomy with mitomycin C (MMC) was superior to drainage valve implantation, Iwasaki et al.³⁶ observed a higher success rate for drainage valve implantation than for trabeculectomy with mitomycin. These findings suggest that drainage valve implantation may not be universally effective for refractory NVG. 

Intraocular anti-VEGF agents have been increasingly used as an adjunctive therapy for refractory NVG to improve the outcome of glaucoma surgery in high-risk patients.²⁷ The VEGA and VENERA studies supported the approval of intravitreal aflibercept injection in 2020 as a treatment for patients with NVG in Japan.^37,38 A prospective randomized pilot study by Kahook³⁹ compared trabeculectomy with anti-VEGF augmentation to standard trabeculectomy alone and found a better outcome for the former group. However, their study only included patients with primary open-angle glaucoma, not patients with NVG. Kahook³⁹ also suggested that anti-VEGF agents can modulate the neovascular drive in diseases such as NVG. 

The timing of anti-VEGF administration may influence the surgical outcome. In this meta-analysis, only one study administered anti-VEGF drugs at the end of surgery for patients with mild neovascularization due to diabetic retinopathy.²⁷ The other six studies administered anti-VEGF drugs at least 1 day before trabeculectomy. The interval between anti-VEGF and trabeculectomy may vary depending on the severity and stage of neovascularization. 

The dosage and type of anti-VEGF drugs may also influence the surgical outcome. All seven trials used intravitreal injections, but the doses and types differed slightly. Bevacizumab was used in five studies and conbercept in two studies. The dose of bevacizumab was 1.25 mg in three studies and 2.5 mg in two studies, while the dose of conbercept was 0.5 mg in two studies. Guo et al.⁴⁰ found similar effects of conbercept and ranibizumab combined with trabeculectomy for NVG. Gupta et al.⁴¹ found no significant difference between 1.25 mg and 2.5 mg intracameral anti-VEGF on surgical outcomes of trabeculectomy for NVG. 

The location of the anti-VEGF injection is essential for the effectiveness of the procedure. All the studies in this meta-analysis used intravitreal injections, but other approaches such as subconjunctival and anterior chamber injections are also used clinically. Guven Yilmaz et al.⁴² reported that drainage valve implantation with preoperative anti-VEGF anterior chamber injection may be an effective alternative for refractory NVG. Bhagat et al.⁴³ found that intracameral injection was the most effective for IOP control. 

Anti-VEGF treatment alone is not enough for refractory glaucoma management. Treating the cause is also necessary to slow disease progression and reduce the complications. The underlying cause of NVG is retinal ischemia, especially before anti-VEGF therapy.⁴⁴ Panretinal photocoagulation is a main therapy for retinal ischemic disease that improves retinal oxygenation by creating thermal burns in the peripheral retina.⁴⁵ Five studies used panretinal photocoagulation, one study did not use it, and one study did not report relevant data. 

Anti-inflammatory therapy (steroid and nonsteroidal anti-inflammatory drugs) may inhibit neovascularization by reducing inflammatory factors such as cytokines and chemokines. Other treatments, such as alpha-lipoic acid, lutein, and ARA290,⁴⁴ may protect retinal nerve cells from apoptosis. Treating retinal ischemia is indispensable for refractory NVG management and can improve glaucoma surgery outcomes. 

The use of antimitotic drugs in this study was essential, as trabeculectomy without them would fail within a few days.^46–48 The dose of antimitotic drugs varied. All studies in this meta-analysis used MMC as the antimitotic drug in trabeculectomy, with different concentrations and durations of action. Four studies used 0.4 mg/mL MMC for 3 to 5 minutes, two studies used 0.2 mg/mL MMC for 2 to 3 minutes, and one study used 0.5 mg/mL MMC for 3 to 5 minutes. 

Our study has several limitations. First, we discussed only three outcome measures (postoperative IOP, success rate, and complications) because they were reported in all studies. Second, patients included in this meta-analysis had different primary diseases, such as proliferative diabetic retinopathy, retinal vascular occlusion, and ocular ischemic syndrome. Third, we only included anti-VEGF drugs used in clinical practice and published studies. Anti-VEGF drugs in preclinical development (such as squalamine, AKB-9778, nesvacumab, and RO6867461)⁴⁴ and ongoing studies or unpublished studies were not included in this meta-analysis. 

Conclusions

This meta-analysis suggested that anti-VEGF agents can enhance trabeculectomy outcomes for NVG in the short term (up to 6 months) but not in the long term (more than 6 months). This finding has implications for the optimal timing and frequency of anti-VEGF injections in refractory NVG management. 

Acknowledgments

Supported by the “Medical Excellence Award” funded by the Creative Research Development grant from the First Affiliated Hospital of Guangxi Medical University (YD). 

Disclosure: X. Zhou, None; J. Chen, None; W. Luo, None; Y. Du, None