Efficacy and Adverse Event Profile of the iStent and iStent Inject Trabecular Micro-bypass for Open-angle Glaucoma: A Meta-analysis

ABSTRACT Aim This meta-analysis explores the efficacy and adverse event profile of the iStent, an ab interno implant for the treatment of open-angle glaucoma. Methods A systematic literature search of Ovid MEDLINE and EMBASE was used to identify peer-reviewed original studies that provided efficacy data on the first or second generation iStent for at least five eyes. Intraocular pressure (IOP) was the primary efficacy endpoint, while the number of medication classes was the secondary outcome. Weighted mean differences were reported for continuous endpoints, while a relative risk was computed for dichotomous variables. Review Results The search revealed 545 results, of which 1767 eyes from 28 studies were included. The cohort age was 71.4 ± 5.4 years, and 44.9% of patients were male. There was a significantly greater IOP reduction after the use of two first-generation stents compared to one, irrespective of phacoemulsification status (p < 0.001). Additionally, there was a significantly greater IOP reduction following iStent alone relative to phaco-iStent for the first-generation iStent (p < 0.001) and the iStent inject (p < 0.001). For the first generation stent, combined phaco-iStent provided a greater level of IOP reduction (p < 0.001) and reduction in the number of medication classes relative to phacoemulsification alone (p < 0.001). In total, 22.5% of eyes that received iStent implantation sustained some type of adverse event. The most common adverse events were intraocular pressure elevation, stent blockage or obstruction, stent malposition and hyphema. Conclusion and Clinical Significance Statistically significant differences in efficacy outcomes exist between different numbers of stents and the presence or absence of concurrent phacoemulsification. How to cite this article: Popovic M, Campos-Moller X, Saheb H, Ahmed IIK. Efficacy and Adverse Event Profile of the iStent and iStent Inject Trabecular Micro-bypass for Open-angle Glaucoma: A Meta-analysis. J Curr Glaucoma Pract 2018;12(2):67-84.


BACKGROUND
Given the irreversible retinal ganglion cell damage resulting from open-angle glaucoma (OAG), current treatment modalities are focused on preserving the structural integrity of the optic nerve and visual function. [1][2][3] Prospective evaluations in glaucoma have demonstrated that the reduction of IOP leads to significant sparing of vision: namely, every 1 mm Hg reduction of IOP is correlated with an approximate 10% decrease in the risk of glaucomatous progression. 4 In OAG, IOP elevation is often a result of reduced aqueous humor flow through the trabecular meshwork 5 In early stages, ocular hypotensive medications and laser trabeculoplasty have been shown to attenuate glaucoma progression; however there are well known issues with compliance, tolerability, persistence, and difficulty of proper instillation. 3,5 In the situations in which these treatments are insufficient in reducing IOP to target pressures according to disease severity, ab externo filtering procedures are utilized to provide a more significant IOP reduction. Unfortunately, these techniques are higher risk options that may result in a bleb-related complication, hemorrhage, hyphema, hypotony, infection, inflammation, loss of vision or reoperation. 6,7 Recently, there has been increasing interest in the ability of microinvasive glaucoma surgery (MIGS) devices to provide a significant level of IOP reduction with less severe postoperative adverse events. 8 One such device, the iStent ® (Glaukos Corporation, San Clemente, California), is the first ab interno glaucoma implant that has been approved for the management of mild-to-moderate OAG. 9 The iStent works by allowing aqueous humor to drain directly from the anterior chamber into Schlemm's canal, thus bypassing a portion of the trabecular meshwork and reducing IOP. 10 Currently, the iStent has only received ORIGINAL ARTICLE food and drug administration approval for use combined with cataract surgery.

Literature Search and Data Collection
A systematic literature search was performed on Ovid MEDLINE (2006-Week 1 2018) and Ovid EMBASE (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018) Week 3). The search strategy that was used can be found in Table 1A and B. Further, Google, Google Scholar and the reference lists of past reviews were manually searched to elicit further relevant literature. Any original prospective or retrospective clinical study that provided relevant efficacy data (i.e., IOP and number of medication classes) on the implantation of the iStent for at least five eyes was included. Only peer-reviewed journal articles were included. Non-english studies, letters to the editor, correspondences, editorials, reviews, opinions, case reports, articles reporting on other surgical procedures and studies that contained repeat data or less than 4 week follow-up were excluded. Studies were screened first by consulting titles and abstracts and afterwards by examining full-text versions. To assist with the screening process, a quality assessment of articles was performed. The Cochrane criteria were used in the assessment of randomized controlled trials, while the National Institute for Health and Care Excellence tool was used to evaluate case series. 38,39 In both cases, studies were excluded if there was a high risk of bias in at least half of the assessment categories.
Variables that were included for the baseline demographic evaluation were country of origin, study design, distribution of right and left eyes, age, gender, ethnicity, cup-to-disc ratio, visual field, mean deviation and time of follow-up. The primary efficacy endpoint, IOP, was collected as a continuous variable (i.e., IOP postoperatively and reduction pre-to post-operatively). The postoperative number of hypotensive medication classes and pre-to post-operative reduction in the number of medication classes was the secondary endpoint. For the efficacy analysis, data on the number of iStents and phacoemulsification status (i.e., whether concomitant phacoemulsification was performed) were extracted. For adverse event analysis, the number of events and the four most prevalent events for each study arm were recorded.

JOCGP
Postoperative outcomes were collected at last followup.

Statistical Analysis
Weighted mean differences (WMD) and corresponding 95% confidence intervals (95% CI) were reported in the analysis of primary and secondary endpoints. Throughout the analysis, the number of eyes (i.e., sample size) was used as a weighted variable. Alongside a random effects model, the inverse variance method was used in the meta-analysis. The weighted mean was defined as while the weighted standard deviation was computed using the formula Due to the differential reporting of included studies, each unique endpoint contains data from a different collection of studies. A consequence of this is that the WMDs of IOP and medication class reduction will likely not equal the difference between the preoperative and postoperative values for IOP and medication class count. In the test for overall effect, a p-value of less than 0.05 was considered statistically significant. The main analysis was performed based on whether patients had 1, 2 or 3 iStents implanted and whether they did or did not receive combined phacoemulsification and iStent. All statistical analyses were performed using Review Manager (RevMan 5.3; The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark) and Microsoft ® Excel (Microsoft Corporation, Redmond, Washington).

Study Inclusions and Baseline Demographics
The systematic search revealed 545 results. Upon title and abstract screening, the number of potential articles was reduced to 135. Afterwards, full-text screening resulted in 28 studies that met al.l inclusion criteria (Fig. 1). 3, Baseline characteristics and the results of quality assessment for included studies are reported on  (Tables 2B and C). No study met the a priori condition for exclusion based on the quality assessment.
Of the 1767 eyes included in the efficacy and adverse event analysis, a total of 1217 (68.9%) underwent combined iStent implantation and phacoemulsification, while 497 eyes (28.1%) underwent iStent implantation alone (Table 3). More than half of included eyes had one iStent implanted (999, 56.5%), while 685 eyes had two (38.8%) and 63 eyes received three (3.6%). Overall, the vast majority of eyes (1398, 79.1%) received a first generation iStent, while only 369 eyes (20.9%) received an iStent inject. The distribution of relevant clinical features between groups is presented in Table 4.

Number of iStents-First Generation
Not accounting for phacoemulsification status, metaanalysis was only possible to evaluate the effect of the number of stents on IOP and medication class reduction for first generation iStents ( n/a -11.5 ± 8.0 n/a n/a n/a *MD = Mean deviation; dB = Decibels; n/a = Not available. Contd...

Phacoemulsification Status-First Generation
Next, studies were categorized by whether phacoemulsification was performed, irrespective of the number of first-generation iStents (Table 6A, Figs 3A and B). Data revealed that the iStent alone group produced a significantly more pronounced reduction in IOP than the phaco-iStent cohort [WMD = -7.44 mm Hg, 95% CI = (-7.82 mm Hg, -7.06 mm Hg), p < 0.001]. The iStent alone group also had a significantly greater preoperative IOP than the phaco-iStent cohort [WMD = -5.72 mm Hg, 95% CI =   The combination of phacoemulsification and a first generation iStent was also compared to phacoemulsification alone (Table 6B,

Adverse Event Analysis
Overall, a total of 261 out of 1159 eyes (22.5%) that received iStent implantation sustained some type of adverse event (Table 8). In order from most to least common, the following adverse events were reported: IOP elevation or spike (reported in 12 of 27 papers; 44.4%), stent blockage or obstruction (8/27; 29.6%), stent malposition (

DISCUSSION
The efficacy and adverse event profile of the iStent device have been explored in a variety of different settings. To evaluate the efficacy and adverse events following iStent Another meta-analysis by the same team aimed to investigate the reduction of IOP after phaco-iStent compared to phacoemulsification alone. 41 40,41 The greater IOP reduction with multiple iStents compared to one has been documented in previous laboratory studies and was also confirmed by the findings of the present meta-analysis. 42 For instance, both postoperative IOP and IOP reduction were significantly improved in the two-stent comparator relative to one. We hypothesize that a selection bias may have influenced these findings, as the higher initial IOP or more severe disease seen in the two-stent comparator may have contributed to the greater IOP reduction following stent implantation. For patients with high preoperative IOP (average of 22.5 mm Hg), three stents provided a more pronounced level of   n/a n/a n/a n/a n/a n/ 96 0 n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a Significant improvement Gallardo et al., 2016 134 0 n/a n/a n/a n/a 83% of eyes achieved a BCVA of 20/40 or better after surgery relative to 20% preoperatively Ferguson et al., 2016 350 n/a IOP spike n/a n/a n/a n/a Lindstrom et al.

57
1 Progression of cataract n/a n/a n/a Stable El wardani et al.

31
n/a n/a n/a n/a n/a n/a 2nd study arm 22 n/a n/a n/a n/a n/a n/a Contd... IOP reduction (9.3 mm Hg) relative to one or two stents. However, interpretations of the three-stent data should be made with caution, as data from only 63 eyes existed for this comparison.
Regardless of the number of implanted iStents, the cohort that underwent first-generation iStent implantation alone saw a more pronounced IOP reduction and lower postoperative IOP than the phaco-iStent group. However, this comparison considers two different patient populations, namely (1) patients receiving iStent alone, who normally do not have cataracts and are receiving the device specifically for IOP reduction, and (2) patients undergoing combined phacoemulsification and iStent, who are receiving the treatment for both their cataracts and an elevated IOP. As such, the finding of a higher preoperative IOP in the iStent alone group may have influenced the difference in IOP reduction between comparators. Even though some included studies contained both patients who received phaco-iStent and iStent alone, subgroup analysis analyzing the differences in outcomes between these two groups was never performed in individual studies. 15,18 As such, the conclusions derived from comparing phaco-iStent versus iStent alone have not been previously established.
Analysis of phaco-iStent compared to phacoemulsification alone revealed that there was a greater IOP reduction following phaco-iStent relative to phacoemulsification alone. This aligns with the findings of Malvankar-Mehta et al., who also showed that there was a significantly greater IOP reduction following phaco-iStent relative to phacoemulsification alone [SMD = -0.46, 95% CI = (0.87, -0.06)]. 41 Despite the similarity, it is important to note that uncontrolled, one-armed studies examining the efficacy of phacoemulsification alone were included in the previous analysis but were excluded in the present article. 41 Instead, we limited our analysis of phaco-iStent versus phacoemulsification only to the studies that had a phaco-iStent arm and a phacoemulsification only comparator, thus resulting in a more controlled analysis. Beyond analysis of IOP, both meta-analyses concluded that phaco-iStent was statistically superior relative to phacoemulsification alone in the reduction of medication class number pre-to post-operatively. 37 0 n/a n/a n/a n/a 76% of eyes achieved a BCVA of 20/40 or better after surgery relative to 68% Preoperatively 2nd study arm 41 0 n/a n/a n/a n/a 66% of eyes achieved a BCVA of 20/40 or better after surgery relative to 61% Preoperatively 3rd study arm 38 0 n/a n/a n/a n/ n/a n/a n/a n/a n/a n/a Stable Iop spike Need for additional surgery n/a n/a n/a n/a Trabulectomy n/a n/a n/a n/a
The adverse event analysis revealed that fewer than 25% of eyes carried some type of adverse event postoperatively, most of which were not serious nor visually threatening. This compares favorably with the postoperative adverse event rates of both trabeculectomy and the Baerveldt glaucoma implant. 43 However, due to differential reporting of adverse events between individual studies, caution should be used when interpreting these findings. In our cohort, IOP elevation, stent blockage or obstruction, stent malposition and hyphema were the most common adverse events following iStent implantation.
Beyond the efficacy and adverse event profile, the cost-effectiveness of the iStent relative to topical glaucoma medications has been studied by Iordanous and colleagues. 44 Following implantation of two iStents, the authors analyzed cost differences at 6 years postoperatively. At 6 years, the iStent was $20.77 more expensive relative to monodrug therapy but was cheaper by $1272.55 compared to bidrug treatment and $2124.71 versus tridrug therapy. The authors concluded that the iStent may offer a modest cost saving when compared to glaucoma medications.
Given that past meta-analyses included lower numbers of eyes receiving iStent implantation (first article: 5 studies, n = 248; second article: 10 studies, n = 396), the present work (28 studies, n = 1767) represents the largest quantitative synthesis of efficacy and adverse event data for the iStent device. 40,41 The large statistical power provided by such a high sample size allowed us to conduct certain analyses that were novel to the published literature; for example, an analysis comparing phaco-iStent to iStent alone. We only included published articles, thus ensuring that the rigors of peer-review were met for each included study.
Limitations of the analysis include the fact that there was no restriction of studies based on design. As such, baseline values for included endpoints were significantly different between comparator arms. As shown in Table 4, the relevant clinical features were often not balanced between groups. As noted by Kaplowitz et al., variation in study design and implementation such as length of follow-up, etiology of disease and baseline clinical indicators may account for the high degree of heterogeneity upon meta-analysis. 45 Further, since some articles did not include sociodemographic and clinical characteristics of their study cohorts (e.g. surgeon experience), it is uncertain whether there was a balance of these factors between comparator arms. For instance, there is variable reporting of surgeon experience in the literature: two articles 19,20 noted that the study surgeon was in an early stage in the learning curve, one noted that the data incorporate the surgeon learning curve, 3 and another hypothesized how the learning curve influenced the greater number of adverse events in an initial set of patients. 22 Two studies reported that their surgeons were experienced, 24,30 while another found no significant difference in outcomes between initial and late procedures. 28 Another limitation was that the lack of available studies prevented us from performing a robust meta-analysis for some endpoints, such as IOP reduction following three stents, where there was only 63 included patients. Limited reporting of adverse event severity across studies prevented us from analyzing severity in the adverse event analysis. Studies were variable in how they handled medication washout before stent implantation, which made it impossible to analyze the effect of preoperative medications on baseline IOP. Given that data was extracted from study cohorts, conclusions should be limited to the level of the cohort.

CONCLUSION AND CLINICAL SIGNIFICANCE
The following meta-analysis has shown that there may be differences in treatment response for the iStent due to varying parameters, including the number of iStents and phaco-iStent compared to either iStent alone or phacoemulsification alone. In our analysis, two stents delivered a greater response in terms of IOP reduction relative to one and iStent alone had a significantly greater IOP reduction compared to phaco-iStent. Combined phaco-iStent was statistically superior relative to phacoemulsification alone in the reduction of IOP and medication classes preto post-operatively. Future research should determine whether similar conclusions are reached following metaanalysis in a more controlled environment.

ETHICAL APPROVAL
This article does not contain any studies with human participants or animals performed by any of the authors. As such, there was no informed consent process needed for this study.