Introduction

Neovascular glaucoma (NVG) is known to cause severe vision loss due to an imbalance of angiogenic factors, resulting in neovascularization of the anterior chamber structures and progressive angle closure.¹ This condition often arises from certain ocular diseases, particularly ischemic retinal conditions such as proliferative diabetic retinopathy (PDR), central retinal vein occlusion (CRVO), and ocular ischemic syndrome (OIS).² The primary treatment objective is to limit and minimize ocular and retinal ischemia using laser pan-retinal photocoagulation (PRP) and intravitreal anti-vascular endothelial growth factor (anti-VEGF) injection.^1,2 Nevertheless, it is crucial to note that this treatment strategy alone may not be sufficient for intraocular pressure (IOP) control. When antiglaucoma medications are maximally tolerated, trabeculectomy and glaucoma drainage device (GDD) implantation are the treatments of choice, with comparable surgical success and IOP reduction in NVG.^3,4 However, trabeculectomy requires fewer surgical instruments and shorter operation time, and may represent a more cost-effective measure compared with GDD implantation.

Several surgical techniques of trabeculectomy for NVG had been modified to improve surgical outcome. For example, intraoperative antimetabolites have been used to improve trabeculectomy outcomes. In addition, mitomycin-C (MMC), a more potent fibroblast inhibitor than 5-fluorouracil (5-FU), downregulates subconjunctival fibrosis and enhances surgical success.⁵ The use of anti-VEGF has been recommended as an adjunct to trabeculectomy because it has demonstrated to enhance surgical success and reduce postoperative complications.^6,7 VEGF is well recognized as a core factor that mediates inflammation and both physiological and pathological angiogenesis.⁸ It plays a significant role in the pathogenesis of intraocular neovascularization in NVG.⁹ The coexistence of systemic vascular diseases is commonly observed in most patients with NVG. VEGF gene expression is also present in arteries from the early to advanced stages of atherosclerosis, and the degree of expression is associated with the severity of the disease.¹⁰

A number of preoperative baseline characteristics in NVG patients were reported as prognostic factors for trabeculectomy outcome; PDR, persistent neovascularization of the iris (NVI), a higher number of anti-VEGF injections, delayed trabeculectomy,¹¹ a greater extent of peripheral anterior synechia (PAS),¹² a lower preoperative IOP (<30 mmHg), vitrectomy after trabeculectomy,¹³ and history of vitrectomy^12,14 were associated with poorer surgical outcomes. However, the impact of age to surgical success is inconclusive.^11,14 Furthermore, postoperative hyphema was significantly correlated with the outcome of trabeculectomy for NVG.¹⁵

The primary objective of this study was to evaluate the surgical outcomes and identify prognostic factors of trabeculectomy with MMC as a treatment for NVG. Notably, our investigation incorporated an assessment of systemic risk factors, a unique aspect that, to the best of our knowledge, apart from age and sex, has not been addressed in previous studies.

Material and Methods

Patients

We retrospectively reviewed the medical records of Thai patients with neovascular glaucoma aged >18 years who were referred to two tertiary centers (1, Burapha University Hospital, Chonburi, eastern Thailand; 2, Rajavithi Hospital, Bangkok, central Thailand) between 2013 and 2022. Only patients who underwent primary trabeculectomy with MMC were included. The exclusion criteria were: 1) no light perception, 2) eyes with any other prior glaucoma surgery, and 3) less than 6 months postoperative follow-up period. This study was approved by the Institutional Review Board of both hospitals (Ethics approval numbers 026/2566 and 110/2567, respectively). The informed consents were obtained from all participants. However, the lost-to-follow participants by the data collection period, informed consents were waived by the Institutional Review Board ethics committee of both hospitals regarding to retrospective nature of this study. The study adhered to patient data confidentiality and was conducted in compliance with the principles of the Declaration of Helsinki.

NVG was categorized based on the presence of neovascularization of the iris (NVI) and/or neovascularization of the angle (NVA), coupled with an IOP exceeding 21 mmHg. If both eyes satisfied the qualifying criteria, only the left eye was included in the study.

The following information was collected at baseline. Systemic factors included age, sex, laterality, glomerular filtration rate (GFR-mL/min/1.73 m²), and underlying diseases (diabetes mellitus, hypertension, coronary artery disease, stroke, and dyslipidemia). Ocular factors include the cause of NVG (PDR, CRVO, and OIS), presence of concurrent vitreous hemorrhage, prior vitrectomy, IOP (mmHg - examined by Goldmann applanation tonometer [GAT]), antiglaucoma medications, peripheral anterior synechiae (PAS-quadrant), prior pan-retinal photocoagulation (PRP), number and time of anti-vascular endothelial growth factor (anti-VEGF) injection, lens status, and visual acuity (LogMAR) taken at the nearest visit before the surgery. Eyes without form vision were assigned the following scores: counting finger = 2.3, hand motion = 2.6, light perception = 2.9, and no light perception = 4.0.

Postoperative assessments, including visual acuity, IOP, and the number of antiglaucoma medications, were documented for 12 months. Additionally, other postoperative complications and subsequent management procedures, such as the administration of anti-VEGF, PRP, needling, and subconjunctival anti-metabolites (MMC and 5-FU), were recorded.

Surgical Technique

PRP was administered whenever feasible prior to the surgery. Additionally, intravitreal anti-VEGF injections (1.25 mg of bevacizumab) were delivered using a sterile technique under topical anesthesia before surgery. Trabeculectomy was performed by glaucoma specialists or in-training glaucoma fellows under supervision in a fornix-based manner. There are two MMC application techniques: subconjunctival injection and sub-tenon-soaked sponge, depending on the surgeon’s preferences. The subconjunctival injection technique involved 10–20 µg / 0.1–0.2 mL after conjunctival closure. The sub-tenon-soaked sponge technique used multiple pieces of 0.4 mg/mL MMC-soaked sponge directly applied to the sub-tenon’s pocket without touching the conjunctival edges for 2–4 minutes. After removing the sponges, irrigation with balanced salt was performed. A rectangular or square scleral flap, approximately 40–60% of the scleral thickness was created. A 2×2 mm sclerostomy was then created using a punch, and peripheral iridectomy was performed. The scleral flap was sutured using at least one stitch of 10–0 nylon. Postoperatively, each patient was properly educated about postoperative care by assigned expert medical personnel and was prescribed topical antibiotics four times a day for one month, coupled with a topical steroid administered in a gradually decreasing dosage for a minimum of three months. Laser suture lysis could be performed according to IOP status and bleb morphology.

Main Outcome

Trabeculectomy was considered a complete success if the IOP remained at or below 21 mmHg without the need for ocular hypotensive medication, and the visual acuity was light perception or better. Conversely, qualified success was achieved if the same criteria were met with the use of medications. Failure was defined as IOP exceeding 21 mmHg or demonstrating less than a 20% reduction below baseline on two consecutive visits 3 months after the procedure. Additionally, failure criteria included IOP remaining at or below 5 mmHg on two consecutive visits after 3 months, the need for additional intraoperative procedures (excluding needling) to control the IOP, or loss of light perception.

Statistical Analysis

Data analysis was conducted using SPSS software (IBM SPSS Statistics 28, IBM Corp., New York). Continuous data are presented as means and standard deviations, while categorical data are presented as percentages (%). The cumulative probability of success was determined using Kaplan–Meier survival curves. Risk factors for failure, postoperative complications, and postoperative manipulations were assessed through Cox proportional hazards regression analysis. Statistical significance was established with a two-sided p-value of less than 0.05.

Results

A total of 106 patients met the inclusion criteria. The mean age of the patients was 57.07 ± 11.79 years, ranging from 27 to 87 years. Sixty-three patients (59.43%) were male and 43 patients (40.57%) were female. Underlying diseases included diabetes mellitus (80.19%), hypertension (73.58%), and major vascular events (coronary artery disease and stroke, 16.98%). Then mean GFR was 60.34 ± 30.83 mL/min/1.73 m² with a range from 4.2 to 124.0, stratified into chronic kidney disease in different stages including stage 1 (22.64%), stage 2 (27.36%), stage 3a (19.81%), stage 3b (9.43%), stage 4 (11.32%), and stage 5 (9.43%), according to Kidney Disease Improving Global Outcomes (KDIGO) criteria. Among these patients, seven underwent renal replacement therapy, with two receiving hemodialysis (HD) and five undergoing continuous ambulatory peritoneal dialysis (CAPD). The baseline systemic characteristics are summarized in Table 1.

Table 1 Systemic Baseline Characteristics

Of the ocular baseline characteristics, 22 patients had bilateral NVG and 6 underwent bilateral trabeculectomy. Only the left eyes in patients who satisfied the inclusion criteria were included in the study. Overall, there was an equal distribution between the right and left eyes (50.94% and 49.06%, respectively). The primary cause of NVG was PDR in 63 (59.43%) eyes. This was followed by CRVO in 39 eyes (36.97%) and OIS in 4 eyes (3.77%). The average visual acuity was 1.89 ± 0.86 Log MAR (range, 0.1–2.9 Log MAR). Most eyes (76.42%) had severe visual loss, with a visual acuity of less than 20/200. Only 10.38% of patients had a visual acuity of 20/70 or better. The mean IOP before surgery was 38.78 ± 13.01 (range, 10–74) mmHg. Eighty-one eyes (76.42%) had an IOP of at least 30 mmHg, whereas 25 eyes (23.58%) had an IOP <30 mmHg. Fifty-one eyes (48.11%) required only topical IOP-lowering medications; all of them took at least two medications, but the majority (41 eyes, 80.39%) had four topical medications. However, 55 eyes (51.89%) received adjunct systemic IOP-lowering medications including oral acetazolamide and/or glycerin. Almost all eyes (94 eyes, 88.68%) had gonioscopic angle closure, defined as the presence of PAS in 2–4 quadrants in 5.66%, 9.43%, and 73.58% of eyes, respectively. Twenty-five eyes (23.58%) had a history of cataract surgery and 13 (12.26%) had undergone vitreoretinal surgery. Thirty-seven patients (34.91%) had concurrent vitreous hemorrhage. Eighty-two eyes (77.36%) had undergone laser PRP preoperatively, and 100 eyes (94.34%) had intravitreous anti-VEGF within 7 days in 24%, 7–30 days in 43%, and after 30 days in 33% of cases. Ocular demographic data of the patients are summarized in Table 2.

Table 2 Ocular Baseline Characteristics

Kaplan–Meier analysis exhibited that the overall success rates at one year with and without medication were 73.6% and 54.7%, respectively (Figure 1).

Figure 1 Kaplan–Meier analysis of trabeculectomy with MMC for NVG (red line-complete success, blue line-qualified success). Abbreviation: NVG, neovascular glaucoma.

From the clinical data, the leading cause of trabeculectomy failure owing to failure criterion was uncontrolled IOP (defined as IOP >21 mmHg or less than a 20% reduction below baseline on two consecutive study visits after 3 months) in 25 eyes (52.1%), followed by reoperation, loss of light perception, and hypotony (defined as IOP ≤5 mmHg on 2 consecutive study visits after 3 months) in 13, 8, and 2 eyes (27%, 16.7%, and 4.2%) respectively (Table 3). The most common reoperative procedure was GDD implantation in 10 eyes (76.9%), followed by cyclophotocoagulation and revised trabeculectomy in 2 eyes (15.4%) and 1 eye (7.7%), respectively (Table 4).

Table 3 Failure Reason

Table 4 Reoperation Procedure

We evaluated the systemic risk factors for trabeculectomy failure (Table 5). Cox proportional hazards regression analysis demonstrated that hypertension was correlated with the risk of trabeculectomy failure (crude hazard ratio [HR], 2.164, p=0.046). Major vascular events (crude HR 2.695, p=0.002) were also significant prognostic factors of surgical failure. The age of the patients (<50 or ≥50 years) and sex did not seem to affect the survival (p=0.485, 0.333 respectively). Other investigated underlying systemic diseases, including diabetes mellitus, dyslipidemia, CKD stage, and renal replacement therapy, also had no significant effect on survival rate in the univariate analysis.

Table 5 Cox Proportional Hazard Regression Analysis Assesses Systemic Factors

Kaplan–Meier survival analysis was performed to compare the surgical success rates between eyes with and without major vascular events (Figure 2; p=0.002).

Figure 2 Kaplan–Meier analysis of trabeculectomy with MMC for NVG (red line-major vascular event, blue line-no major vascular event). Abbreviation: NVG, neovascular glaucoma.

In multivariate Cox regression analysis, only major vascular events were ascertained as predictive factors for surgical failure independent of each other (HR, 2.778, p=0.001).

The potential ocular prognostic factors influencing survival time are listed in Table 6. The prognostic factors found to be significant were the number of antiglaucoma medications (p=0.026) and additional systemic IOP-lowering medications (HR, 1.987, p=0.022) in the univariate analysis. However, significant differences were found only for additional systemic IOP-lowering medications (HR, 1.837, p=0.045) in the multivariate analysis. Other preoperative ocular factors studied were laterality, cause of NVG, visual acuity, IOP, PAS, stage of NVG, lens status, previous vitreoretinal surgery, concurrent vitreous hemorrhage, preoperative PRP, and preoperative anti-VEGF. They had no significant effect on trabeculectomy outcomes.

Table 6 Cox Proportional Hazard Regression Analysis Assesses Ocular Factors

Kaplan–Meier survival analysis was performed to compare the surgical success rates between eyes with and without combined systemic medications (Figure 3, p=0.045).

Figure 3 Kaplan-Meier analysis of trabeculectomy with MMC for NVG (red line-combined systemic medication, blue line-only topical medication).

Cox proportional hazards regression analysis demonstrated that postoperative manipulation with anti-VEGF, PRP, and 5-FU had no significant effect on survival rate. Needling and MMC subconjunctival augmentation were correlated with trabeculectomy failure in the univariate analysis, p=0.007 and p<0.001 respectively. In multivariate Cox regression analysis, postoperative subconjunctival MMC was identified as significant association with trabeculectomy failure (Table 7, HR, 3.100, p<0.001). Survival analysis revealed that patients who underwent postoperative MMC subconjunctival injection had significantly poorer outcomes than those who did not receive MMC (Figure 4, p<0.001).

Table 7 Cox Proportional Hazard Regression Analysis Assesses Postoperative Manipulation

Figure 4 Kaplan-Meier analysis of trabeculectomy with MMC for NVG (red line-MMC, blue line-no MMC).

Complications encountered during the postoperative period are listed in Table 8. The most common postoperative complications included hyphema in 14.15% of the cases followed by hypotony, choroidal detachment (CD), and bleb leak accounting for 11.32%, 9.43%, and 5.66%, respectively. However, no significant differences were observed in the univariate analysis.

Table 8 Postoperative Complication

Discussion

This study investigated the surgical success rate and systemic and ocular prognostic factors of trabeculectomy with MMC for NVG in Thailand. Survival analysis showed that the overall complete and qualified success rates at one year were 54.7% and 73.6%, respectively. Our complete success rate was favorable compared to that of a previous study,¹¹ where a complete success rate of 50% at one year was observed in the NVG group with proliferative diabetic retinopathy. However, the results were much higher in a study by Higashide et al in 2015,¹³ which reported a complete success rate of 86.9% in 61 eyes treated with NVG. This discrepancy in outcomes may be attributed to the differences in baseline characteristics among the study populations. Notably, over 90% of the participants in our study were in the advanced stage of neovascular glaucoma (closed angle), in contrast to the 59% reported in a previous study. This finding is also supported by the Kiuchi study,¹² which demonstrated a significant correlation between extensive PAS and increased surgical failure rates. In terms of qualified success, these research results aligned consistently with prior studies (62.6–93.4%).^{11–14,16–18} Researchers must exercise caution when interpreting and comparing success results, because previous studies may have possibly employed varying definitions of failure. While some studies included hypotony (IOP ≤5 mmHg) as a failure criterion,^11,13 others did not.^{12,14,16–18} Furthermore, some studies¹³ have focused on achieving a specific percentage reduction in the IOP.

Several studies have investigated the ocular risk factors of trabeculectomy in patients with NVG. The prognostic factors varied among the studies. Takihara et al concluded that a younger age and a history of vitrectomy were predictive factors for surgical failure.¹⁴ According to Senthil et al, factors associated with trabeculectomy failure include older age, PDR, persistent NVI, a higher number of anti-VEGF injections, and delayed trabeculectomy.¹¹ Kiuchi et al found that patients with NVG with a greater extent of peripheral anterior synechia and a history of vitrectomy had poorer outcomes,¹² whereas Higashide et al reported that a lower preoperative IOP (<30 mmHg) and vitrectomy after trabeculectomy were associated with higher trabeculectomy failure in NVG.¹³ To the best of our knowledge, no study has examined the systemic risk factors other than age and sex in patients with NVG and surgical failure of trabeculectomy with MMC.

In this study, underlying major vascular events (HR, 2.778; p=0.001) were identified as significant risk factors for surgical failure according to the complete success criteria in multivariate analysis. There have been no previous reports of cardiovascular disease and stroke associated with trabeculectomy failure in NVG. However, three cases have been reported on the incidence of neovascular glaucoma (one case from OIS¹⁹ and two cases from PDR^20,21 after the early postoperative coronary artery bypass graft [CABG] procedure in cardiovascular disease patients). Oda et al reported a case of stroke in one month after NVG due to PDR.²² According to Nielsen, neovascular glaucoma is also significantly associated with macrovascular diseases, including cardiovascular disease, hypertension, cerebrovascular stroke, and gangrene amputation.²³

From both the molecular and clinical perspectives, it is evident that NVG and ischemic stroke share similar pathogenesis, supported by elevated VEGF levels observed during the progression of both conditions.^9,24–28 VEGF levels are increased in the aqueous humor of patients with glaucoma and rabbit models, indicating its role in the disease process. Additionally, studies have demonstrated that VEGF stimulates fibroblast proliferation in vitro, potentially leading to scar formation after filtration surgery.²⁹ Park et al highlighted a noteworthy association between higher levels of VEGF in Tenon’s capsule at the time of surgery and a greater risk of failure and poorer IOP outcomes at one year in patients with primary open-angle glaucoma (POAG).³⁰ The study by Kim et al in 2009 also confirmed the same relationship in NVG patients who underwent Ahmed implantation.³¹ These findings suggest the potential benefits of adjunctive anti-VEGF therapy in trabeculectomy to enhance surgical outcomes, correlating with a recent systematic review of 2023.⁷

Major vascular events, especially myocardial infarction, may secrete the circulatory cytokines growth differentiation factor 3 (GDF-3),³² galectin-3,^33,34 and arginase-1³⁵ in response to ischemia. These cytokines promote cell proliferation and fibroblast activation in the ischemic heart. Angiogenesis has also been established to compensate for the reductions in blood flow and oxygen supply.³⁶ Angiogenesis is essential for recovery and protection following cerebral ischemia. The expression of activin receptor-like kinase 5 (ALK5) was significantly elevated after ischemic stroke and could promote extracellular matrix production, stimulate mesenchymal cell differentiation, and induce vascular maturation.³⁷ These secretory cytokines from compensatory mechanisms promote local tissue remodeling. However, the effects contributing to ocular tissue fibrosis and angiogenesis, resulting in adverse consequences at the trabeculectomy site, need to be further elucidated.

In our multivariate analysis, we discovered that preoperative systemic medication use was associated with an increased risk of trabeculectomy failure. The use of systemic medication often indicates more severe NVG, which could independently affect surgical outcomes. The administration of systemic medications, such as oral acetazolamide and/or glycerin, indicated that NVG eyes already had maximal tolerable topical medications, potentially harming conjunctival tissue and affecting trabeculectomy outcomes.^38–40 However, in our study, the number of preoperative topical medications showed a significant association with the risk of failure in the univariate analysis but failed to show significance in the multivariate analysis.

Our retrospective cohort study also found that postoperative MMC injections were significantly associated with surgical failure (HR, 3.100; p<0.001). This likely results from the higher possibility of postoperative subconjunctival MMC injections observed in cases prone to failure.

This study had some limitations. First, the retrospective design and lack of a control group are the major limitations. Second, the involvement of multiple surgeons across two tertiary care centers may have hindered the standardization of surgical procedures. Additionally, there was a potential for selection bias, because the study was conducted in a tertiary center where more complex or advanced cases are often referred. Furthermore, the relatively short follow-up period might have restricted our ability to identify additional long-term prognostic factors.

The strength of our study lies in its pioneering analysis of preoperative systemic factors as prognostic contributors to trabeculectomy with MMC outcomes in NVG patients, providing valuable additional insights. This study identified associations between baseline clinical characteristics, including major vascular events, and the risk of trabeculectomy with MMC failure in NVG. These results may initiate the design of an ideally prospective randomized study with a larger sample size and longer follow-up period to validate the association of major vascular events, including cardiovascular disease and stroke, after trabeculectomy with MMC survival in patients with NVG.

Conclusion

We found that trabeculectomy with MMC effectively reduced elevated IOP associated with NVG in Thailand. Among the risk factors identified for trabeculectomy with MMC failure using the complete success criteria, underlying major vascular events and adjunct preoperative systemic IOP-lowering medications were significant negative prognostic factors.