Efficacy and Safety of Different Types of Intraocular Pressure-Lowering Surgeries in Patients with Primary Angle Closure (PAC) or PAC Glaucoma: Systematic Review and Network Meta-Analysis of Randomized Clinical Trials

ABSTRACT Objective To compare the intraocular pressure (IOP)-lowering effect of different types of surgery available in the literature using a network meta-analysis (NMA) based on a systematic review. Methods PubMed and the Cochrane database were searched. Randomized clinical trials involving surgical interventions for high IOP for PAC (primary angle closure) or PACG (primary angle closure glaucoma) were included. Descriptive statistics and outcomes were extracted. Bayesian NMA was performed to compare the IOP-lowering effect and the change in the number of antiglaucoma drugs required between baseline and endpoint, as well as success rates. Results This NMA included 21 articles with 1237 eyes with PAC or PACG. Interventions were characterised as phacoemulsification (phaco), trabeculectomy, goniosynechialysis (GSL) with viscoelastic or blunt device, goniosurgery (GS) (trabeculotomy or goniotomy), micro-bypass stent (Istent®), endocyclophotocoagulation (ECPL) or various combinations of these procedures. Phaco+GSL [−1.73 (95%CrI: −3.53 to −0.13)] and phaco+GSL+GS [−3.92 (95%CrI: −6.91 to −1.31)] provided better IOP lowering effects than phaco alone. Phaco+trabeculectomy [−3.11 (95%CrI: −5.82 to −0.44)] was inferior to phaco+GSL+GS. Phaco+trabeculectomy [−0.45 (95%CrI: −0.81 to −0.13)] provided a better outcome in terms of reducing the need for antiglaucoma drug compared to phaco alone. There were no differences between the other surgeries in terms of reduction of antiglaucoma drug number and IOP lowering effect. Success rates were similar for all surgical procedures. Conclusions Phaco+GSL+GS showed the most promising results for lowering IOP. Phaco+trabeculectomy resulted in a significant reduction in the number of antiglaucoma drugs compared to phaco alone.


INTRODUCTION
Primary angle closure glaucoma (PACG) accounts for approximately half of all glaucoma-related blindness. 1The term "angle closure" is used to describe the positional approximation of the iris and ciliary processes.A shallow anterior chamber, an anteriorly positioned lens or a crowded angle are the predisposing risk factors for this condition.The risk of angle closure increases with age, hyperopia and female sex.The distribution of glaucoma subtypes and their prevalence varies by geographical region and ethnic group.In general, PACG has a lower prevalence than primary open-angle glaucoma; however, it is more common in Asia and may be underdiagnosed in other populations. 1,2rimary angle closure disease, the leading cause of irreversible blindness in East Asia, is projected to affect 32 million adults aged 40-80 years by 2040. 3 Although it is a less common form of glaucoma, surgery is more prominent and urgent in its treatment.Therefore, it is important to determine the type of surgery to be performed.
Trabeculectomy, lens extraction or angle-related surgery such as goniosynechialysis are the current options available to surgeons.Many randomised trials conducted in recent years have compared both combined and stand-alone surgery and shown that they can effectively lower intraocular pressure (IOP).The aim of this trial was to determine the most effective IOP-lowering surgery for primary angle closure (PAC) or PACG.

METHODS
We used the Preferred Reporting Items notification checklist for Systematic Review and Meta-Analyses (PRISMA) as a guide for reporting the systematic review. 4The conduct of systematic reviews and meta-analyses was based on the 'Cochrane Handbook for Systematic Reviews of Interventions'. 5

Search Methods for Identification of Studies and Criteria for Considering Studies
Trials were retrieved from the PubMed and Cochrane databases on 10 August 2022.The search was restricted to title/ abstract fields to reduce irrelevant literature.The terms "angle closure", "closed angle" and "narrow angle" were searched.No other filters were used.Details of the search strategies are provided in Supplementary file 1.The second search was conducted on 14 March 2023.
All prospective randomised clinical trials (RCTs) comparing the IOP-lowering effect of at least two surgical interventions in patients with PAC and/or PACG and reporting results at least six months after surgery were included.All included trials had a two-arm intervention design; there were no multiarm intervention trials.
Trials comparing surgery with iridotomy/iridoplasty as a stand-alone procedure and interventions performed during the acute angle closure crisis were excluded.Studies with fewer than 10 treated patients, follow-up of less than six months, and non-English literature were excluded.Meeting abstracts, case reports, letters to the editor, reviews, recommendations and expert opinions were also excluded.

Data Collection and Outcome Measures
Data were reviewed and extracted separately by two independent authors (HS and DGS) using EndNote X8 software (Clarivate, Philadelphia, USA).The third author (CE) resolved any disagreements.The following data were extracted: first author's name, year of study, country in which the study was conducted, subject characteristics (gender and age), intervention arms, outcome time, and type of outcome measure.
The primary outcome measure was the change in IOP (mmHg) between baseline and endpoint, and the secondary outcome measure was the change in the number of antiglaucoma medications between baseline and endpoint.Tertiary outcomes were surgical failure rate and surgical complications.

Risk of Bias Assessment and Confidence Assessment
The authors used the Cochrane risk of bias 2 (ROB2) tool to assess methodological quality.ROB2 assesses bias due to the randomisation process, bias due to deviations from the intended interventions, bias due to missing outcome data, bias in the measurement of the outcome and bias in the selection of the reported outcome. 6The Confidence In Network Meta-Analysis (CINeMA) tool was used to assess the confidence in the results.The CINeMA approach assesses within-study bias, reporting bias, indirectness, imprecision, heterogeneity and inconsistency. 7The Risk Of Bias due to Missing Evidence in Network meta-analysis (ROB-MEN) tool was used for the reporting bias domain in CINeMA. 8

Data Synthesis and Analysis
Network meta-analysis (NMA) was performed for all comparisons using the Bayesian framework random effects model of MetaInsight v4.0. 9Pooled estimates for IOP change and number of antiglaucoma medications were presented as mean difference and 95% credible intervals (CrI).Pooled estimates for success rates were presented as risk ratio and 95% CrI.The reference intervention was "phaco" for NMA.When medians and ranges were reported in the study, we estimated the mean and standard deviation from the sample size, median, range, mid-range and/or interquartile range and mid-quartile range. 10,11If the mean and standard deviation of the change from baseline to endpoint were reported in the original articles, these were used in the analysis.If the mean and standard deviation of the change from baseline to endpoint were not reported in the original articles, they were calculated using the following equations. 12 a conservative approach, we used the correlation coefficient (r) = 0.4. 12he area under the cumulative ranking curve (SUCRA) graph indicates the probability of a given rank.Phaco was used as the reference group in all rankings and interventions were ranked from best to worst performance. 13,14nconsistency was assessed for each comparison using the node-splitting approach.A p value < 0.05 was considered significant for the inconsistency test.Methodological and clinical comparability was assessed for transitivity.
Comprehensive Meta-Analysis Software version 3 (Biostat, New Jersey, USA) was used for pooled estimation for all treatments.An I 2 value greater than 50% was considered significant heterogeneity.Pooled estimates are presented as mean and 95% confidence intervals (CI).
As the included studies reported non-standardised success criteria, success rates were reviewed using qualitative and quantitative synthesis.The surgical complications were reviewed using qualitative synthesis.

Results of the Search
The literature search identified 1046 studies (PubMed: 367, Cochrane: 679).After removing duplicate studies, the titles and abstracts of 822 studies were screened, and the full texts of 103 studies were examined.Twenty-one studies  were included in the final quantitative review. The sudy selection process is shown in the flow diagram (Figure 1).

Study Characteristics and Quality Assessment of Studies
A total of 1237 eyes were included in the analysis.One trial did not define whether patients had cataract or not, 33 one trial had cataractous and non-cataractous eyes, 27 and another trial had all eyes non-cataractous. 19In the all other trials, patients all had cataracts.Interventions were characterised as phacoemulsification (phaco), trabeculectomy, goniosynechialysis (GSL) with viscoelastic or blunt device, goniosurgery (GS) (trabeculotomy or goniotomy), micro-bypass stent (Istent®), endocyclophotocoagulation (ECPL) or various combinations of these procedures.Seven different combinations of procedures were classified.Mean follow-up ranged from 6 to 24 months.All intervention and study characteristics are shown in Supplementary file 2. The network of comparisons is shown in Figure 2.  Overall, the calculated ROB in the 21 included studies was high in five studies, moderate in eight studies and low in eight studies.The ROB summary is shown in Figure 3.The quality of the evidence estimates from the NMA ranged from low to very low according to the CINeMA approach.The evidence ratings using the CINeMA tool are presented in Supplementary file 3.

IOP-Lowering Effect
All 21 trials were included in this analysis.  The oled IOPlowering effect estimate with the random effects model for all surgical interventions was −6.6 mm/Hg (95%CI: −7.7 to −5.5) from baseline to endpoint (Supplementary file 4).The intervention-specific IOP-lowering effect is shown in Table 1.Phaco was inferior to both phaco+GSL+GS [−3.92 (95%CrI: −6.91 to −1.31)] and phaco+GSL [−1.73 (95%CrI: −3.53 to −0.13)].Furthermore, phaco+trabeculectomy was inferior to phaco +GSL+GS [−3.11 (95%CrI: −5.82 to −0.44)].There was no statistical difference in IOP reduction between the other interventions.The forest plot of NMA is shown in Figure 4 According to the SUCRA value (92.08), phaco+GSL+GS had the highest IOP lowering effect.The ranking probabilities for each surgical procedure were ranked from highest to lowest performance and are presented in Supplementary file 5.The SUCRA plot is shown in Figure 5.The higher the SUCRA value and the faster the curve rises, the better the surgical intervention performs [12,13].
There was a wide range in the initial baseline IOP values of the subjects in the included studies, which may have     influenced the results.However, when the baseline IOPs were compared using the NMA, there was no statistical difference between the baseline IOPs (Supplementary file 6).

Inconsistency and Heterogeneity
The node-split approach was used to assess the inconsistency between direct and indirect evidence, and the inconsistency node was not found to be significant (Supplementary file 7).For the IOP-lowering effect, the pairwise meta-analysis result of phaco+trabeculectomy versus phaco was highly heterogeneous (I 2 = 80.1, tau 2 = 0.254, Q = 25.2, p < .001),but the results of phaco+GSL versus phaco was homogeneous (I 2 = 29.6,tau 2 = 1.522,Q = 8.5, p < .202)].

DISCUSSION
Pairwise meta-analyses 36,37 and many RCTs have compared surgical interventions for PACG and/or PAC, but a head-tohead comparison of all surgical interventions in one study is not possible due to the wide range of interventions.Therefore, we used NMA techniques with RCTs comparing surgical interventions to provide further guidance on the surgical treatment of angle closure.The Bayesian framework was chosen because it can derive the rank probability of each intervention. 38n this analysis comparing the IOP lowering effect of surgical procedures in PAC and/or PACG patients, phaco+GSL and phaco+GSL+GS were better than phaco alone.The mean postoperative IOP reduction was 6.6 mmHg for all procedures.Phaco+trabeculectomy was superior to phaco alone in reducing the number of antiglaucoma medications required.The mean postoperative number of antiglaucoma medications decreased by 1.7 for all procedures.Other surgical procedures showed comparable effects to phaco in terms of IOP reduction and number of antiglaucoma medications.According to the criteria of complete and qualified success, all surgical procedures were similar.
Angle closure is often associated with a small corneal diameter, a shallow anterior chamber, a thick and anteriorly located lens and a short axial length.0][41][42][43] These factors lead to crowding of the angle region.As a result, the iris covers the trabecular meshwork and peripheral cornea, leading to peripheral anterior synechiae (PAS), which damages the trabecular meshwork and impairs outflow. 3Surgery aims to restore the anatomy of the anterior chamber.After phacoemulsification, the anterior chamber deepens and the angle widens.However, if angle closure persists and becomes chronic, PAS will occur and although phaco would deepen the anterior chamber centrally, PAS would persist and the angle would not be open unless angle-based surgery is added to phaco.We believe that the superiority in IOP reduction with angle surgery combined with phaco compared to phaco alone shows that although the anterior pushing effect of the lens is reduced with the removal of the thick lens, the adhesions formed in the angle during the chronic period can only be released by angle surgery.Trabeculectomy, GS and micro-bypass stenting restore the impaired outflow function.Therefore, these procedures can be combined to achieve the most effective IOP reduction according to the patient's needs.
Treatment recommendations for primary angle closure are based on several guidelines, including the European Glaucoma Society (EGS) guidelines and the Primary Angle Closure Preferred Practice Pattern (PPP) guidelines.According to the EGS guideline, 44 recommended interventions for primary angle closure include lens extraction or laser peripheral iridotomy, depending on the disease spectrum and the presence of cataract.The Effectiveness in Angle-Closure Glaucoma of Lens Extraction (EAGLE) study showed that clear lens extraction provided better IOP control and quality of life benefits compared to laser peripheral iridotomy in PACG and PAC patients. 45Tham et al. suggested that phacoemulsification may be an alternative to trabeculectomy as the first surgical option in eyes with CACG without cataract. 19The PPP guideline 46 also suggests that peripheral anterior synechiae (PAS), adhesions between the iris and trabecular meshwork, may cause persistent or progressive IOP elevation.GSL to increase aqueous outflow has been proposed as a treatment option to improve IOP control.We found that phaco+GSL +GS surgery was the best in terms of IOP reduction.The SUCRA value of this combination was 92.0 and showed significantly more IOP reduction than trabeculectomy (SUCRA value = 59.7).We attribute this successful performance to the fact that the phaco+GSL+GS combination addresses all the impaired anatomical components in angle closure.
Trabeculectomy is the recommended glaucoma surgery for pseudophakic patients according to the EGS guideline. 44rabeculectomy does not correct angle narrowing and may lead to further shallowing of the anterior chamber.8][49] Trabeculectomy or surgery combined with trabeculectomy is associated with serious complications such as malignant glaucoma or choroidal detachment in some cases. 18,19,23,25,27,31,34,35However, these complications were not seen in GSL and/or GS.Minimal, self-limiting and nonsight-threatening complications were mostly seen in nontrabeculectomy surgery.Non-trabeculectomy procedures were found to be relatively safer than trabeculectomy alone or phaco+trabeculectomy. Furthermore, non-trabeculectomy procedures had similar success rates to trabeculectomy.Surgical procedures with lower rates of complications and less severe complications may be preferred when success rates are statistically similar.
High IOP is a major risk factor for glaucoma progression and lowering IOP is the primary goal of glaucoma treatment. 50chieving a lower IOP with surgery can reduce the need for antiglaucoma drugs, thereby reducing the risk of side effects. 51,52By reducing the burden of antiglaucoma drug, it can improve medication compliance and quality of life. 53All surgical procedures provide a significant reduction in the number of antiglaucoma drugs.Phaco+trabeculectomy seems to be more effective than phaco in reducing the number of antiglaucomatous drugs.However, the phaco+ECPL combination ranked first in terms of ranking probability and was most successful in reducing the number of antiglaucoma drugs.
The study has some limitations.Iridotomy may affect the angle configuration and this may have influenced our results.There were two different GSL procedures, blunt and viscoelastic.
In addition, there were two trabeculectomy procedures, with and without mitomycin C, which may have influenced the observed complication rates.Different PAS levels of the eyes, timing of surgery after diagnosis, delay in diagnosis, degree of angle closure and ethnic differences may also have influenced our results. 54n conclusion, phaco combined with angle-based surgery showed a significantly better IOP-lowering effect than phaco alone, with minimal complications.These procedures may be the most appropriate option for patients considering IOP-lowering surgery.Other procedures were similar to phaco and to each other.The number of antiglaucoma druglowering effects of the procedures were similar, although phaco+trabeculectomy was better than phaco.Complete and qualified success rate was similar in all surgical procedures.Due to the increasing prevalence of angle closure worldwide, we believe that our findings may be clinically important for the surgical management of IOP in these patients.We would encourage further studies to determine the most appropriate approach to the treatment of a patient with PAC glaucoma.

Figure 2 .
Figure 2. Network plot of IOP-lowering effect (A) and antiglaucoma drug number lowering effect (B).The size of the nodes depends on the number of cases in each treatment arm.The thickness of edges depends on the number of trials conducted.

Figure 3 .
Figure 3. Risk-of-bias 2 assessment tool."Traffic light" plots of domain-level judgements for each individual study; and weighted bar plots of the distribution of risk-ofbias judgements within each bias domain.

Figure 4 .
Figure 4. Forest plot of network meta-analysis.Reference surgery is phacoemulsification.IOP-lowering effect (A) and antiglaucoma drug number lowering effect (B).

Figure 5 .
Figure 5. Litmus Rank-O-Gram: Higher SUCRA (Surface Under the Cumulative Ranking Curve) values and cumulative ranking curves nearer the top left indicate better performance.IOP-lowering effect (A) and antiglaucoma drug number lowering effect (B).

Figure 6 .
Figure 6.Forest plot of network meta-analysis.Reference surgery is phacoemulsification.Complete success rate (A) and qualified success rate (B).

Table 1 .
Comparison of all treatment pairs.