Meta-Analysis of Phacoemulsification and Laser Peripheral Iridotomy in the Treatment of Primary Angle-Closure Glaucoma

Background In this meta-analysis, we aimed to systematically compare the efficacy and safety of phacoemulsification and laser peripheral iridotomy (LPI) in the treatment of primary angle-closure glaucoma (PACG). Method We searched PubMed, MEDLINE, EMBASE, the Cochrane Library, the Chinese Journal Full-text Database (CNKI), the Wanfang database, and the China Science and Technology Journal Database for randomized controlled trials (RCTs) of phacoemulsification and LPI in the treatment of PACG published up to September 30, 2022. Postoperative intraocular pressure (IOP), anterior chamber depth (ACD), complications, corneal endothelial cell count, and best-corrected visual acuity (BCVA) were compared. The effective quantity of measurement data was measured by the mean difference (MD) and 95% confidence interval (CI). The effect of counting data was measured by the odds ratio (OR). Result A total of 1731 potential studies were identified, and after screening, 8 RCT studies were included. The results of the meta-analysis showed that, compared to the LPI group, the patients in the phacoemulsification group showed lower IOP six and twelve months after operation (MD-3.39, 95% CI −4.15∼−2.63, P < 0.00001; −2.29, −3.52∼−1.06, 0.0003). The ACD in the phacoemulsification group was significantly deeper than that in the LPI group (1.59, 1.10∼2.09, 0.00001). Meanwhile, the incidence of complications in the phacoemulsification group was lower than that in the LPI group (OR = 0.46, 0.29∼0.72, 0.0006). There was no statistically significant difference between the phacoemulsification group and the LPI group in corneal endothelial cell count and BCVA at 6 and 12 months after operation (P=0.38; 0.11; 0.81). Conclusion Compared with LPI, phacoemulsification is safer and more effective in the treatment of PACG, especially in controlling IOP and minimizing postoperative complications.


Introduction
Glaucoma is one of the leading causes of irreversible blindness worldwide [1]. It was estimated that there were 20.2 million cases of PACG in 2013, and it was projected that this number would rise to 32.04 million in 2040 due to the anticipated increase in the aging population [2]. Although primary open-angle glaucoma (POAG) is more prevalent, PACG is more invasive and on average carries a three-fold higher risk of severe, bilateral visual impairment compared to POAG [3]. Te prevalence of glaucoma varied across ethnicity, as PACG is more common in Asians [4]. Moreover, growing amounts of data from recent studies have shown that PACG is 2-4 times more prevalent than previously reported in Europe and the USA [2,5]. Terefore, it is particularly important to choose the appropriate PACG therapy and treatment.
LPI is commonly performed on PACG patients to relieve pupil blockage. LPI, however, is unable to open the closed anterior chamber angle in up to 58% of PACG patients [6,7].
Cataract phacoemulsifcation has improved in recent years due to the development of phacoemulsifcation and a more reliable fuid fow system [8]. Te indications for cataract surgery have been expanded to include the treatment of eye diseases other than cataracts due to improvements in surgical safety. Currently, a growing number of studies indicate that cataract extraction can play a signifcant role in decreasing IOP and minimizing the application of glaucoma medicines [9][10][11][12][13][14][15][16][17][18][19][20][21].

Inclusion and Exclusion
Criteria. Studies were included if they met the following criteria: (1) the RCT study of patients with PACG; (2) interventions only included phacoemulsifcation and LPI; and (3) studies published in Chinese or English.
Studies were excluded if they met the following criteria: (1) they were not RCT studies; (2) studies contained other interventions in addition to phacoemulsifcation and LPI; (3) studies published in languages other than Chinese or English; and (4) studies with incomplete resources, insufcient data, or lack of author information.

Data Extraction.
Te relevant data from the articles were extracted by two researchers independently according to the designed data extraction table. Te specifc information includes the following: (1) general characteristics of the study: including author, publication time, and publication journal; (2) the general characteristics of the subjects included in the study; (3) intervention measures, outcome indicators, etc; and (4) statistical methods of research. Inconsistencies and diferences of opinion in the process of extracting data shall be solved through joint discussion with the third researcher.

Assessment of Study Quality.
Te two researchers independently evaluated the quality of the selected studies according to the evaluation criteria. In case of disagreement, it shall be solved through discussion or by inviting a third researcher. Te quality of the RCTstudy was evaluated by the Cochrane bias risk tool according to the following seven domains: (1) random sequence generation (selection bias); (2) allocation concealment (selection bias); (3) blinding of participants and personnel (implementation bias); (4) blinding of outcome assessment (measurement bias); (5) incomplete outcome data (follow-up bias); (6) selective reporting of research results (reporting bias); and (7) other bias. Each domain was graded into "low risk of bias," "high risk of bias," or "unclear risk of bias." 2.5. Statistical Analysis. All analyses were performed with the RevMan5.4 software. Te heterogeneity among studies was evaluated by the Q test and I 2 test. For the studies with signifcant heterogeneity (P < 0.1 or I 2 > 50%), the random efect model was used for analysis. When signifcant heterogeneity existed, sensitivity analysis was used to eliminate studies with large diferences or studies with a high risk of bias to test the stability of the combined results. For studies with low heterogeneity (P > 0.1 or I 2 < 50%), the fxed efect model was used for analysis. IOP, ACD, corneal endothelial cell counting, and BCVA were measured. Te mean difference (MD) and 95% confdence interval (CI) were used as statistical analysis variables. Te complications during treatment were binary variables, using the odds ratio (OR) and 95% CI as statistical analysis variables. A P value less than 0.05 was considered to be statistically signifcant.

Search Results.
A total number of 1731 potential studies were identifed from an electronic-based search, of which 1570 studies remained after eliminating duplicates. Fiftyfour studies remained after analyzing the titles and abstracts of the studies and removing any with inconsistent content or incomplete main indicators. Fifty-four full-text articles were identifed as potentially relevant for analysis in terms of the inclusion and exclusion criteria. Finally, eight studies [22][23][24][25][26][27][28][29] involving 894 eyes were included in the analysis. Tere was no signifcant diference in age, gender composition ratio, and IOP before treatment between the two groups (P > 0.05). Te article screening process is shown in Figure 1, and the basic information of the included studies is shown in Table 1.

Te Characteristics of the Included Studies. Te 8 in-
cluded studies were all RCTs. Te Cochrane Risk of Bias Assessment Tool was used to assess the quality and bias risk of the included RCTs. Te results are displayed in Figure 2. Six studies [23,[25][26][27][28][29] mentioned the random sequence generation scheme, indicating a low risk of bias. Two studies [22,24] were rated as uncertain because the random sequence generation scheme was not specifed in the text. Among the selection bias related to allocation concealment, four studies [22][23][24][25] did not report it in the paper, so the risk of bias was uncertain. In the implementation bias, two studies [22,24] could not judge the bias risk, and one study [22] was evaluated as high risk in reporting the source of bias. Among measurement bias, follow-up bias, and other biases, all studies [22][23][24][25][26][27][28][29] were determined as low risk.

IOP at Diferent Times after the Operation.
A total of three studies [24,27,29] recorded the IOP values of the two groups six months after treatment. Because no statistical heterogeneity was found among the studies (P � 0.61 and I 2 � 0%), we used the fxed efect model for analysis. Te results showed that the IOP six months after operation in the phacoemulsifcation group was signifcantly lower than that in the LPI group, and the diference was statistically signifcant (MD-3.39, 95% CI −4.15∼−2.63, and P < 0.00001; Figure 3(a)).
A total of three studies [26,27,29] recorded the IOP values of the two groups of patients twelve months after treatment. Tere was signifcant heterogeneity between the studies (P � 0.12 and I 2 � 52%), so the random efect model was used for analysis. Te results showed that the IOP 12 months after operation in the phacoemulsifcation group was signifcantly lower than that in the LPI group, and there was statistically signifcant diference (MD-2.29, 95% CI −3.52∼−1.06, and P � 0.0003; Figure 3(b)). A sensitivity analysis was conducted for the heterogeneity among studies, and the results showed that the heterogeneity decreased signifcantly and the combined efect value did not change signifcantly (I 2 � 0%, MD-2.63, 95% CI −3.31 ∼ −1.95, and P < 0.00001; Figure 3(c)) after excluding the study of Husain et al. [26]. It showed that the results of the metaanalysis were stable.

ACD after the Operation.
A total of four studies [22,23,25,28] recorded the ACD of the two groups three months after the operation. Tere was signifcant heterogeneity between the studies (P < 0.00001 and I 2 � 99%), so the random efect model was used for analysis. Te results showed that the central ACD in the phacoemulsifcation group was signifcantly deeper than that in the LPI group three months after the operation, and there was statistically signifcant diference (MD 1.59, 95% CI 1.10∼2.09, and P < 0.00001). A sensitivity analysis was conducted for the heterogeneity among studies, and the results showed that the heterogeneity decreased signifcantly and the combined efect value did not change signifcantly (I 2 � 0%, MD 1.79, 95% CI 1.72∼1.86, and P < 0.00001) after excluding the study of Min et al. [22]. It showed that the meta-analysis results were stable (Figure 4).

Probability of Complications during Treatment.
A total of three studies [24,26,29] recorded the complications during postoperative treatment. Tere was no signifcant heterogeneity between the studies (P � 0.73 and I 2 � 0%), so the fxed efect model was used for analysis. Te results showed that the incidence of complications in the phacoemulsifcation group was signifcantly lower than that in the LPI group, and the diference was statistically signifcant (OR � 0.46, 95% CI 0.29∼0.72, and P � 0.0006; Figure 5).

Corneal Endothelial Cell Counting after the Operation.
A total of two studies [24,26] recorded corneal endothelial cell counting of the two groups six months after the operation. Tere was no signifcant heterogeneity between the studies (P � 0.69 and I 2 � 0%), so the fxed efect model was   (61·0-73·0) 122 121 6, 12, 24, and 36 months used for analysis. Te results showed that there was no signifcant diference in corneal endothelial cell counting between the two groups six months after the operation (P � 0.38, Figure 6).

BCVA after the Operation.
A total of two studies [26,27] recorded the BCVA of the two groups six months after the operation. Tere was no signifcant heterogeneity between the studies (P � 0.37 and I 2 � 0%), so the fxed efect model was used for analysis. Te results showed that there was no signifcant diference in the BCVA six months after the operation between the two groups (P � 0.11, Figure 7(a)).
A total of two studies [27,29] recorded the BCVA of the two groups of patients twelve months after the operation. Tere was no signifcant heterogeneity between the studies (P � 0.72 and I 2 � 0%), so the fxed efect model was used for analysis. Te results showed that there was no signifcant      Figure 4: Forest plots comparing the ACD between the Phaco groups and LPI groups after operation. (a) Tree months after the operation. (b) Tree months after the operation (excludes the study of Min et al. [22]). diference in the BCVA twelve months after operation between the two groups (P � 0.81, Figure 7(b)).

Discussion
In PACG, the relative pupillary block is a common mechanism that contributes to the closure of the anterior chamber angle, but it is not the only mechanism. According to the World Glaucoma Expert Consensus, [30] the mechanism for angle closure in PACG primarily involves four elements: the iris, ciliary body, lens, and posterior lens. Some patients experience multiple factors contributing to their condition, and lens-related factors play a role in all the mechanisms except the one at the ciliary body level, highlighting the signifcance of the lens in the pathogenesis of glaucoma.
A multicenter research indicated that clear lens extraction showed greater efcacy and should be considered to be the frst-line treatment of PACG [29]. Although phacoemulsifcation has been used to treat PACG in China 20 years ago, [31] some domestic scholars disagree that clear lens extraction is the frst choice for PACG. Tey believe that LPI can relieve pupil block. As a result, there is no reason to remove the normal lens. It has been suggested that, as an internal eye surgery, the risk of cataract phacoemulsifcation is higher than that of LPI, there may be more complications, and the visual function of patients may be dramatically worse than before. How should we select a PACG treatment?
Firstly, in most PACG cases, there are some pathological changes even when there is no obvious lens opacity. In these patients, the lens is generally thicker or positioned too far  forward, which is a signifcant pathological mechanism leading to angle closure. Lens extraction can theoretically eliminate the anatomical factors that contribute to this closure. In recent years, several clinical trials have demonstrated that cataract extraction can signifcantly broaden the anterior chamber angle [32][33][34][35][36][37]. By comparing ACD between the two groups after treatment, we observed that the ACD in the phacoemulsifcation group was signifcantly deeper than that in the LPI group, and the diference was statistically signifcant, which was consistent with the above research results. Tese fndings confrmed that phacoemulsifcation could efectively deepen the anterior chamber and inhibit the mechanism of angle closure. Secondly, our study found that there was no signifcant diference in corneal endothelial cell count between the two groups six months after treatment (P � 0.38 > 0.05), indicating that, despite being an internal eye operation, cataract phacoemulsifcation did not cause additional endothelial cell damage compared to the LPI group. With the advancement of modern cataract phacoemulsifcation technology and the protection provided by viscoelastic agents, the impact of cataract surgery on the corneal endothelium has been gradually reduced. Te results revealed that the occurrence of complications was lower in the phacoemulsifcation group than that in the LPI group, and the diference was statistically signifcant (P � 0.0006 < 0.05). It further proves that phacoemulsifcation is a safe and efective treatment option for PACG.
At the same time, phacoemulsifcation has more advantages than LPI in terms of IOP control. At six and twelve months after the operation, the IOP in the phacoemulsifcation group was lower than that in the LPI group, and the diference was statistically signifcant.
Finally, this study analyzed the BCVA of the two groups six and twelve months after treatment. Te results showed that there was no signifcant diference in BCVA between the two groups (P � 0.11 > 0.05), indicating that there were no complications afecting visual acuity in the phacoemulsifcation. It demonstrated that phacoemulsifcation did not increase the risk of visual loss while maintaining good IOP control. Tis further suggests that phacoemulsifcation is expected to be the frst choice for the treatment of angle closure glaucoma.

Conclusion
Based on the above evidence, cataract phacoemulsifcation may be a crucial surgical option for the treatment of PACG. Phacoemulsifcation is more efective than LPI for PACG treatment, particularly in controlling IOP and minimizing complications. However, there are few studies that have examined the efectiveness of the two treatments over the long term. Terefore, additional research is required for long-term follow-up observation to evaluate its safety and validate the advantages already observed so far.

Data Availability
Te data used to support the fndings of this study are included within the article.