Article Text

Review and meta-analysis of ab-interno trabeculectomy outcomes
1. Kevin Kaplowitz1,
2. Igor I Bussel2,
3. Robert Honkanen3,
4. Joel S Schuman2,4,
5. Nils A Loewen2
1. 1Department of Ophthalmology, VA Loma Linda Healthcare System, Loma Linda, CA, USA
2. 2Department of Ophthalmology, UPMC Eye Center, Eye and Ear Institute, Ophthalmology and Visual Science Research Center, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
3. 3Stony Brook University School of Medicine, Ophthalmology, Stony Brook, NY, USA
4. 4Department of Bioengineering, Swanson School of Engineering, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
1. Correspondence to Dr Kevin Kaplowitz, VA Loma Linda Healthcare System, Department of Ophthalmology, 11201 Benton St 112F, Loma Linda, CA 92357, USA; kevin.kaplowitz{at}va.gov

## Abstract

We analysed all of the PubMed publications on ab-interno trabeculectomy (AIT) with the Trabectome (Neomedix, Irvine, California, USA) to determine the reduction in intraocular pressure (IOP) and medications following the procedure. For IOP outcomes, PubMed was searched for ‘trabectome’, ‘ab interno trabeculotomy’ and ‘ab interno trabeculectomy’ and all available papers retrieved. The meta-analysis used a random-effects model to achieve conservative estimates and assess statistical heterogeneity. To investigate complications, we included all abstracts from the American Glaucoma Society, American Academy of Ophthalmology, American Society of Cataract and Refractive Surgery and the Association for Research in Vision and Ophthalmology. The overall arithmetic mean baseline IOP for standalone Trabectome was 26.71±1.34 mm Hg and decreased by 10.5±1.9 mm Hg (39% decrease) on 0.99±0.54 fewer medications. Defining success as IOP ≤21 with a 20% decrease while avoiding reoperation, the overall average success rate after 2 years was 46±34%. For combined phacoemulsification-Trabectome, the baseline IOP of 21±1.31 mm Hg decreased by 6.24±1.98 mm Hg (27% decrease) on 0.76±0.35 fewer medications. The success rate using the same definition at 2 years was 85±7%. The weighted mean IOP difference from baseline to study endpoint was 9.77 mm Hg (95% CI 8.90 to 10.64) standalone and 6.04 mm Hg (95% CI 4.95 to 7.13) for combined cases. Despite heterogeneity, meta-analysis showed significant and consistent decrease in IOP and medications from baseline to end point in AIT and phaco-AIT. The rate of visually threatening complications was <1%. On average, trabectome lowers the IOP by approximately 31% to a final IOP near 15 mm Hg while decreasing the number of medications by less than one, with a low rate of serious complications. After 2 years, the overall average success rate is 66%.

• Glaucoma
• Intraocular pressure
• Physiology
• Treatment Surgery
View Full Text

## Introduction

Ab-interno trabeculectomy (AIT) with the Trabectome (Neomedix, Tustin, California, USA) is a surgical technique with an extensive body of experience following Food and Drug Administration approval in 2004.1 It is a plasma surgery method with a bipolar 550 kHz electrode ablating 30°–180° of trabecular meshwork (TM). Our aim was to systematically review all published literature on Trabectome, focusing on intraocular pressure (IOP) outcomes and safety profile. AIT has been performed for over 10 years but many publications report overlapping data and no randomised control trials (RCTs) have yet been done. We also aim to recommend the minimal size needed for an RCT comparing IOP outcomes between AIT and phaco-AIT. AIT refers to Trabectome surgery done as the only procedure in either phakic or pseudophakic cases and phaco-AIT to combined phacoemulsification with Trabectome.

### Indications

In the initial clinical description, AIT was described for use in adult and primary open-angle glaucoma (POAG) (open >20°) and uncontrolled IOP on maximally tolerated medical therapy. Additionally, the patient must be able to rotate in order to assure an adequate gonioscopic view.

Of all published cases that detail glaucoma subtype, POAG accounts for two-thirds. Pseudoexfoliation glaucoma is the second largest subgroup. Numerous secondary glaucomas, from uveitic to intraocular lymphoma,2 have been treated in numbers too small for subgroup analysis. In order of prevalence in the reported literature these were pigment dispersion, uveitic, steroid-induced, anti-vascular endothelial growth factor (VEGF) agent-induced, status postsurgery and traumatic glaucoma. Although narrow angles were originally a contraindication, one study showed that cases with Shaffer grade ≤2 had a similar IOP decrease and success rate to cases with open angles.3

The only relative contraindications still unstudied are active neovascular glaucoma, elevated episcleral venous pressure, and angle dysgenesis, presumably due to concern for arterial bleeding or flowback and aberrant location of Schlemm's canal. There is no evidence that using Trabectome in any type of glaucoma is harmful.

## Methods

The meta-analysis and review were performed based on a predetermined protocol. We followed the recommended guidelines as described by the Cochrane Handbook for Systematic Reviews of Interventions.4

### Literature search

PubMed and Google Scholar were searched for studies published before February 2015 for ‘trabectome’, ‘ab interno trabeculotomy’ and ‘ab interno trabeculectomy’. Abstracts of the following meetings were also searched: American Glaucoma Society, American Academy of Ophthalmology, American Society of Cataract and Refractive Surgery and the Association for Research in Vision and Ophthalmology. All references were followed.

### Inclusion and exclusion criteria

The following articles were considered eligible: (a) type of study: prospective or retrospective case series or cohort; (b) type of participants: glaucoma subjects without restriction for age, gender, ethnicity, comorbidities, use of adjunctive medications; (c) type of intervention: AIT with or without phacoemulsification; (d) type of outcome variables: IOP, anti-glaucoma medications, complications. For publications with overlapping populations, only the longest follow-up was included. All available papers in any language on PubMed were included. Case reports and reviews were examined for content but excluded from the meta-analysis.

### Outcome measures

Primary outcomes were change in IOP and anti-glaucoma medications. Secondary outcomes were adverse events.

### Statistical analysis

Meta-analysis was conducted using the Stata software package (V.11.0; Stata, College Station, Texas, USA). Given variation in clinical and study characteristics, we assumed that heterogeneity was present and used a random-effects model to achieve conservative estimates. Statistical heterogeneity was examined by the I-squared value, a consistency statistic describing total variation across studies due to heterogeneity rather than chance.5 I-squared values of 25%, 50% and 75% reflect heterogeneity that is low, moderate and high, respectively. p <0.05 was considered statistically significant. We also calculated the sample size required for an RCT between AIT and phaco-AIT to detect a 3 mm Hg difference between groups with a power of 90%.

### Sensitivity analysis

While there were too few studies to conduct sensitivity analysis, a limited sensitivity analysis evaluated the effect of trial type and length. Publication bias was assessed by funnel plot asymmetry. It was not meaningful to apply Begg and Egger statistical tests of funnel plot asymmetry with combined data sets.6

### Surgical technique

No published literature reported comparisons of surgical technique. Most studies follow the same key steps. Briefly, the surgeon sits temporally as for clear cornea phacoemulsification. The ablation is carried out starting at 0.8 mW. Up to 90° can be ablated in both directions. Once ablation is complete, a viscoadaptive substance can be injected over the ablation arc to minimise postoperative hyphema.

The only likely source of variation in technique is the length of the ablation arc. This was studied in two abstracts. One attempted to correlate the length of the ablation arc (mean 87°) with the IOP decrease and found no significant correlation.7 The second analysed the Trabectome database and split results based on whether or not ablation >90° was reported.8 Again, no statistically significant difference was found. In narrow angles with synechiae, the smooth base plate of the handpiece has been reported to be used for synechiolysis.9

Most studies report discontinuing glaucoma medications on the day of surgery. The postoperative regimen consists of a topical fluoroquinolone, prednisolone acetate 1% and pilocarpine 2%. Flattening the peripheral iris using pilocarpine may reduce peripheral anterior synechiae formation.10

## Results

### Literature search

The initial literature search returned 109 articles, but only 64 were related to Trabectome. Figure 1 provides a flow diagram of the search results.

Figure 1

Flow diagram of included studies for the meta-analysis.

### Characteristics of included studies

The studies included in the meta-analysis were published from 201011 to 2014.12 The total number of patients was 5091. The mean sample size for AIT studies was 355±400 and for phaco-AIT it was 204±196. The mean duration of follow-up for AIT was 26.6±20.2 months and for phaco-AIT it was 19.5±13.2.

### Overall results

The published results of AIT surgery are shown in table 1. There were no RCTs involving Trabectome. The largest data set comes from the global Trabectome study database sponsored by the device manufacturer, which analyses the first 20 cases of any Trabectome surgeon who voluntarily sends in de-identified clinical data as postmarket surveillance. This data is stored by the company. The most recent update had 4659 cases (AIT±phacoemulsification) and detailed a mean IOP decrease of 26% from a baseline of 23.1±8 on 1.5 fewer medications at 7.5 years.23

Table 1

Intraocular pressure (IOP) and medication outcomes following Trabectome

The only common definition of success was adapted from the Tube versus Trabeculectomy (TVT) Study24 of having a final IOP ≤21 mm Hg with a 20% decrease from baseline, without reoperation, and this definition was used in seven studies.11 ,14 ,18 ,20 ,22 ,23 ,25 For AIT, the average success rate was 61±17% after 1 year (based on n=5 studies) and 46±34% after 2 years (n=2). For phaco-AIT, the average success rate was 85±17% after 1 year (n=6) and 85±7% after 2 years (n=2). In the longest available analysis (Trabectome database) the success rate neared 85% for phaco-AIT after 5 years, and 56% for AIT after 7.5 years, for an overall rate of 66%.23 Only 7% of those cases required reoperations.

There were two studies that reported success rates significantly lower than the average. The first used the TVT definition and found that only 36% of AIT cases were successful after 1 year and 22% after 2 years.14 The second study used a less stringent definition of success of IOP ≤21 mm Hg or ≥20% reduction and found that only 38% of cases were successful after 1 year and 36% after 2 years.16 This low success rate seemed disparate from the mean 36% IOP decrease seen in this same group after 2 years. As with the rest of the papers, they did not state that failures were censored from IOP averages, except in cases where a second surgery was done. The authors stated that mean IOP could have a significant decrease despite low success rates because the successful cases disproportionately lower the mean IOP.26 The average IOP decrease in the cases designated as successful was 11 mm Hg on two fewer medications, whereas those designated as failures had a mean decrease <1 mm Hg on one less medication. This highlights the difficulty of estimating a success rate based on average IOP, particularly when different indications exist (since some consider a reduction in medication in someone who needed cataract surgery to be a success). Also, these studies use success criteria that were originally published to study penetrating surgeries that have significantly higher rates of success, complications and vision loss.14 Ideally future trials could include a score or index that includes the decrease in IOP as well as the decrease in medication, and possibly rates of vision loss.

### Meta-analysis

Combined analysis with a random-effects model indicated that IOP decreased significantly from baseline to end point; AIT weighted mean difference (WMD)=−9.77 mm Hg (95% CI −8.90 to −10.64) and phaco-AIT WMD=−6.04 mm Hg (95% CI −4.95 to −7.13) (figure 2). Medications decreased as well; AIT WMD=−1.11 (95% CI −0.59 to −1.64) and phaco-AIT WMD=−0.73 (95% CI −0.51 to −0.94) (figure 3). These point estimates are based on data with high heterogeneity (for IOP and medication in AIT and phaco-AIT, I-squared statistic >75%, p<0.001). Despite heterogeneity, a limited meta-analysis showed significant decrease in IOP and medications for both AIT and phaco-AIT. Since this review brought together diverse studies, heterogeneity is expected. As such, the consistent direction of effect across all studies is not negated by the degree of heterogeneity. We reason that the high heterogeneity is likely due to clinical trial variation in sample size, glaucoma subtype, follow-up duration and baseline IOP. It was not feasible to explore sources of heterogeneity due to limited amount of trials, variable study design and lack of participant-specific data. With these limitations in data sources, along with overlapping data sets, we were unable to analyse sources of bias including sampling bias, attrition bias and other reporting biases.

Figure 2

Weighted mean difference (WMD) of the reduction in intraocular pressure (IOP) between baseline and final measurements by ab-interno trabeculectomy (AIT) alone, phacoemulsification combined with AIT (phaco-AIT), and trials that only reported one set of results for both standalone and combined cases (combined).

Figure 3

Weighted mean difference (WMD) of the reduction in medications between baseline and final measurements by ab-interno trabeculectomy (AIT) alone, phacoemulsification combined with AIT (phaco-AIT), and trials that only reported one set of results for both standalone and combined cases (combined).

A limited sensitivity analysis was performed to evaluate robustness by trial type (prospective vs retrospective) and trial length (<12M, 12M-24M, 24M-48M, >48M). Random-effect estimates were generally similar in demonstrating decrease in IOP and medications, suggesting high stability in the meta-analysis (data not shown).

Publication bias was assessed by funnel plot (figure 4). The asymmetry of WMD against SE indicates probable multiple sources of bias. There is likely a reporting and language bias. A small study effect is noted since smaller studies are more likely to indicate a greater intervention effect than larger studies.

Figure 4

Funnel plot of weighted mean difference (WMD) versus SE of WMD indicated probable multiple sources of bias including a reporting and language bias.

### Risk factors for failure

Possible reasons for failure to reach target IOP following an apparently successful procedure include incomplete engagement with the footplate or re-approximation of the edges,27 likely by limbal corneal endothelial cells.28 ,29 Fellman and Grover proposed testing for functional patency downstream of Schlemm's canal by decompressing the anterior chamber and then irrigating directly adjacent to the cleft to observe for blanching of the episcleral, conjunctival or aqueous veins.30 Most failures occur by postoperative month 6.17 ,31 ,32

The biggest risk factor for failure is a lower baseline IOP, with an HR of 0.96 per 1 mm Hg (mean baseline IOP 28.1±8.6 mm Hg).14 Two studies found a somewhat linear trend for greater IOP reduction with higher baseline IOPs.19 ,33 The superior IOP reduction may be due to a floor effect in that most successful cases will finish with an IOP in the low to mid-teens, regardless of the baseline IOP.

The second main risk factor is younger age (HR 0.98 per year of age, mean age in that group was 63.6±16.6 years).14 One study reported that body mass index >25 kg/m2 may be a borderline risk factor for failure, possibly because of increased episcleral vein resistance.34 Questionable risk factors include previous laser trabeculoplasty.12 Conflicting results have also been reported about lens status. Ahuja et al16 found no difference in IOP reduction between phakic and pseudophakic cases, while another study found higher success rates with AIT compared with phaco-AIT.35

### Reoperations

Following a failed AIT surgery, every type of conventional surgery and laser has been reported subsequently.14 ,16 ,35 In one study, the reoperation rate after 4 years was 27%.16 The mean IOP at the visit prior to reoperation was 26.2 mm Hg, 21% higher than the IOP prior to AIT. Limited selective laser trabeculoplasty following failed AIT (mean 55 shots) was ineffective in a small series.36 Repeat AIT has been reported in small numbers.14 ,35 Previous failed AIT was not a risk factor for failure of subsequent trabeculectomy.32 Neodymium-doped yttrium aluminium garnet goniopuncture (0.2–0.6 mJ for 3–15 shots) was used for eight patients with post-AIT IOP elevation.10 Following goniopuncture, the IOP decreased 21% 11 months later.

### Complications

Table 2 lists the published complications of AIT. The most common complication is hyphema. Twelve patients presented with transient vision loss 2–31 months following AIT from a spontaneous rebleed associated with a mean IOP 12 mm Hg higher than the preceding visit.37 This may be under-reported: a case report using continuous IOP monitoring showed nocturnal IOP spikes ascribed to transient microhyphemas.38 Knape and Smith noted intraoperative hyphema during a trabeculectomy done after failed AIT,39 and recommended avoiding hypotony during second surgeries.

Table 2

Complications reported after AIT

The second most common complication was peripheral anterior synechiae, in 24% of patients.1 Transient IOP spikes ≥10 mm Hg were seen in 4%40– 10%.13 AIT in phakic eyes does not seem to cause cataract progression. After 30 months, only 1.2% of 86 cataracts had significant progression.1

To investigate the rate of serious complications following AIT, an extended literature search was conducted as described above. The total number of AIT cases reported in PubMed and abstract databases was difficult to determine because of overlapping data sets. The total number of reported cases is approximately 10 902. The most common serious complication was 10 cases of hypotony (IOP<5 mm Hg) 4 weeks after surgery, or 0.09% of all reported cases.23 No details were given as to the possible aetiology, what the target IOP was, and whether this was clinically significant. Other complications were four cases of aqueous misdirection (0.04%). Details were sparse but in at least one, miochol was used.35 Six cyclodialysis clefts were reported (0.06%), with only one reporting further intervention (closure by laser).11 ,41 There are three cases of a postoperative IOP spike associated with hyphema35 ,37 requiring surgical intervention (0.03%). Finally there is a single case of choroidal haemorrhage (0.01%) with no details given15 and a single case (0.01%) of endophthalmitis after phaco-AIT where the culture grew Enterococcus faecalis.9 Persistent hypotony or sympathetic ophthalmia has not yet been reported after AIT. These rates likely overestimate complication rates because not all AIT outcomes are published, but a rare complication is more likely to be published.

Phaco-AIT was reported to have similar refractive outcomes to phacoemulsification.42 Most of the studies directly specify that no patient lost two lines of vision. One paper noted that 13 of 246 cases (5%) lost >2 lines of Snellen visual acuity, but there are no details specifying if the vision loss was related to the surgery or glaucoma, a comorbidity, a complication or whether those patients had received a combined surgery.16

### Comparison with other surgeries

Unlike with microsurgical bypass stents (eg, iStent,43 Glaukos, Laguna Hills, California, USA), AIT has not been compared with phaco-AIT in an RCT. Studies suggest that standalone phacoemulsification usually decreases the IOP by 1.5–2 mm Hg in both healthy patients and those with glaucoma.44 One non-randomised study compared phacoemulsification in healthy patients with 114 phaco-AIT cases,22 but the baseline IOPs were too different to yield convincing conclusions.

One of the main benefits of performing AIT is that there is no bleb formation with the perpetual increased risk of infection nor is there any hardware left in the eye to erode or become infected. The main disadvantage is that AIT should not be relied upon for an IOP in the low teens. Two studies compared AIT surgery with trabeculectomy with mitomycin C. The first found a 52% IOP decrease following trabeculectomy versus a 30% decrease following AIT.40 The second reported a success rate at 2 years of 43% for AIT versus 76% with trabeculectomy.14 However, excluding hyphema, the complication rate was only 4% with AIT versus 35% with trabeculectomy.

AIT has been compared with another surgery in three other studies. Two compared phaco-AIT with trabecular aspiration and found lower IOPs with phaco-AIT.21 ,45 An abstract compared 52 phaco-AITs with 49 phacoemulsification procedures where 2 iStents were inserted. After 12 months, the IOP was <18 mm Hg in 14% of AITs versus 39% of double iStents.46

Since studies for different minimally invasive glaucoma surgeries (MIGS) have different patient populations and different indications for surgery, the results cannot be directly compared. AIT can be expected to lower the IOP by approximately 36% to a final average IOP around 16 mm Hg. To try to give some type of comparison from the available data, a single iStent has been shown to reduce the IOP by 8%43–27%47 with an overall mean final IOP near 17 mm Hg with significantly greater reductions reported when inserting more than one stent.46 Endoscopic cyclophotocoagulation used for POAG (as opposed to neovascular glaucoma) decreases the IOP by 8%48–47%,49 to a final average near 15 mm Hg. In the TVT Study, the Baerveldt glaucoma drainage device lowered the IOP by 41% (mean IOP at 5 years was 14.4 mm Hg) and trabeculectomy by 50% (mean IOP at 5 years was 12.6 mm Hg).24

## Discussion

AIT has outcomes published out to 10 years and lowers the IOP by approximately 36% to a final average near 16 mm Hg on less than one fewer medication. After 2 years, the overall average success rate is 66% using a strict criteria designed to measure success in penetrating surgeries. The improved safety profile with MIGS means that they can be employed earlier than trabeculectomy. Combined cases may be done in patients with lower baseline IOPs to reduce reliance on medication, making it especially challenging to study the success rates. Future trials ideally could include some type of score or index that includes the decrease in IOP in mmHg as well as the decrease in medication. This would allow for a better comparison of surgeries with different indications. The best way to determine how much further IOP lowering AIT offers in addition to phacoemulsification is an RCT, which is also needed to compare the IOP-lowering efficacy of AIT with other penetrating and MIGS. We used observed SD data from prior 1-year outcomes between AIT and phaco-AIT to calculate the number of subjects required for an RCT. To detect a ≥3 mm Hg difference between AIT and phaco-AIT with 90% power, each group requires ≥33 subjects.

Unlike with trabeculectomy or aqueous shunts, removing the TM with AIT only allows for increased flow into Schlemm's canal since there is no route created for aqueous to diffuse out through subconjunctival space or lymphatics.50 ,51 While AIT does not usually lower the IOP as much as trabeculectomy or aqueous shunts, the rate of visually threatening complications is very low at <1%. Most of the complications besides hyphema overlap with phacoemulsification.

Cost-effectiveness has also been studied.52 A cost analysis indicated that the cost of AIT over 6 years offered a cumulative savings of $C279.23,$C1572.55 and \$C2424.71 per patient versus monodrug, bidrug and tridrug therapy, respectively.53 The study did not adjust for success rates decreasing over time of any of the proposed treatments. The only data on Trabectome success rates after 6 years comes from the company database, which reported it to be 66%.23 These savings were greater than with iStent but lower than with endoscopic cyclophotocoagulation.

Future directions with AIT may include intra- and postoperative imaging of Schlemm's canal and perhaps even distal structures with optical coherence tomography or longer-wavelength studies.54 Ideally, preoperative evaluation could identify regions near collector channels or other segments that displayed dilations in Schlemm's canal, and intraoperative imaging could assure that the TM was ablated over those regions. Postoperatively, optical coherence tomography or canalography could identify the source of obstruction to aqueous flow. Another research area is adjunctive use of AIT, such as for cases after failed trabeculectomy.55

## Conclusion

Trabeculectomy ab interno with the AIT is a mature surgical technique with an extensive body of experience since 2004. AIT can be expected to lower the IOP by approximately 36% to a final average IOP around 16 mm Hg while decreasing the number of medications by less than one. After 2 years, the overall average success rate is 66%. The rate of visually threatening complications is <1%.

View Abstract

## Footnotes

• Contributors All authors contributed to the study concept, analysis and drafting. All authors approve this submission.

• Competing interests NAL: Neomedix (L). JSS: Zeiss (P).

• Provenance and peer review Not commissioned; externally peer reviewed.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.