The Ahmed Versus Baerveldt Study

The Ahmed Versus Baerveldt Study Five-Year Treatment Outcomes Panos G. Christakis, MD,1 Jeffrey W. Kalenak, MD,2 James C. Tsai, MD,3 David Zurakowski, PhD,4 Jeffrey A. Kammer, MD,5 Paul J. Harasymowycz, MD,6 Juan J. Mura, MD,7 Louis B. Cantor, MD,8 Iqbal I.K. Ahmed, MD1 Purpose: To compare 2 frequently used aqueous shunts for the treatment of glaucoma. Design: International, multicenter, randomized trial. Participants: Patients aged 18 years or older with uncontrolled glaucoma despite maximum tolerated medical therapy, many of whom had failed or were at high risk of failing trabeculectomy. Methods: Eligible patients were randomized to receive an Ahmed-FP7 valve implant (New World Medical, Inc, Rancho Cucamonga, CA) or a Baerveldt-350 implant (Abbott Medical Optics, Inc, Santa Ana, CA) using a standardized surgical technique. Main Outcome Measures: The primary outcome was failure, defined as intraocular pressure (IOP) outside the target range (5e18 mmHg) or reduced <20% from baseline for 2 consecutive visits after 3 months, severe vision loss, or de novo glaucoma surgery. Secondary outcomes measures included IOP, medication use, visual acuity, complications, and interventions. Results: A total of 238 patients were randomized; 124 received the Ahmed-FP7 implant, and 114 received the Baerveldt-350 implant. Baseline characteristics were similar between groups. Mean preoperative IOP was 31.4
10.8 mmHg on 3.1
1.0 glaucoma medications. At 5 years, the cumulative failure rate was 53% in the Ahmed group and 40% in the Baerveldt group (P ¼ 0.04). The main reason for failure in both groups was high IOP, and the cumulative de novo glaucoma reoperation rate was 18% in the Ahmed group and 11% in the Baerveldt group (P ¼ 0.22). Hypotony resulted in failure in 5 patients (4%) in the Baerveldt group compared with none in the Ahmed group (P ¼ 0.02). Mean IOP was 16.6
5.9 mmHg in the Ahmed group (47% reduction) and 13.6
5.0 mmHg in the Baerveldt group (57% reduction, P ¼ 0.001). Mean medication use was 1.8
1.5 mmHg in the Ahmed group (44% reduction) and 1.2
1.3 mmHg in the Baerveldt group (61% reduction, P ¼ 0.03). The 2 groups had similar complication rates (Ahmed 63%, Baerveldt 69%) and intervention rates (Ahmed 41%, Baerveldt 41%). Most complications were transient, and most interventions were slit-lamp procedures. Conclusions: Both implants were effective in reducing IOP and the need for glaucoma medications. The Baerveldt group had a lower failure rate and a lower IOP on fewer medications than the Ahmed group, but had a small risk of hypotony that was not seen in the Ahmed group. Ophthalmology 2016;123:2093-2102 ª 2016 by the American Academy of Ophthalmology. Supplemental material is available at www.aaojournal.org.

The mainstay of glaucoma treatment is reducing intraocular pressure (IOP) to prevent disease progression and vision loss.1,2 Treatment often begins with the use of topical medications and laser trabeculoplasty, with surgery reserved for advanced or refractory cases, or for patients at high risk of failing medical therapy.3 Traditional filtration surgery is trabeculectomy, which removes a block of limbal tissue to allow aqueous humor to flow through a scleral flap to an adjacent subconjunctival bleb. However, rates of failure of approximately 10% per year have been reported, as well as complications including hypotony.4e6  2016 by the American Academy of Ophthalmology Published by Elsevier Inc.

Patients who have failed trabeculectomy with antimetabolite or who have high risk disease (i.e., neovascular or uveitic glaucoma) may benefit from aqueous shunts.4,7 These implants shunt aqueous humor from the anterior chamber using a long tube connected to a subconjunctival equatorial end-plate that drains into the venous plexus. The 2 most commonly used aqueous shunts are the Ahmed valve (New World Medical Inc, Rancho Cucamonga, CA) and the Baerveldt implant (Abbott Medical Optics, Santa Ana, CA). These shunts differ in that the Ahmed valve implant has a built-in Venturi-based flow restrictor designed to prevent postoperative hypotony and its complications. However, the http://dx.doi.org/10.1016/j.ophtha.2016.06.035 ISSN 0161-6420/16

2093

Ophthalmology Volume 123, Number 10, October 2016 Ahmed implant has been associated with bleb encapsulation, resulting in inadequate IOP reduction and the need for postoperative glaucoma medications or additional glaucoma surgery.8,9 The Baerveldt implant lacks built-in flow restriction and requires the surgeon to create a temporary tube ligature to allow for healing to occur around the end plate to regulate flow. This limits postoperative flow until the ligature dissolves or is removed, at which point the Baerveldt implant achieves excellent IOP control because of its large end-plate surface area (350 mm2).4,7 However, there is a risk of hypotony should there be inadequate healing around the plate to regulate flow when the tube opens.10,11 A 2008 American Glaucoma Society survey showed a significant change in practice patterns between 1996 and 2008, with trabeculectomy use decreasing from 81% to 46% and aqueous shunt use increasing from 18% to 51%.12 The impetus for this change may be based on the results of the Tube Versus Trabeculectomy (TVT) study, a multicenter randomized trial that showed that Baerveldt-350 implantation had a higher success rate with fewer complications than trabeculectomy with mitomycin C after 5 years.4,5 Furthermore, a 2015 meta-analysis comparing Ahmed valve implantation with trabeculectomy showed similar success rates, IOP reduction, and medication use, with fewer adverse effects seen in the Ahmed group.13 Until recently, selecting which aqueous shunt to use was dependent on surgeon and site preference given the lack of prospective data comparing the devices. Retrospective studies have yielded inconclusive results and have had modest sample sizes, and pooling them is difficult given their heterogeneous patient populations, use of different device models, and varying success criteria.9,14e16 The Ahmed Versus Baerveldt (AVB) study is an international, multicenter, randomized trial that has provided evidence to support a surgeon’s choice.17e19 Three-year results demonstrated that the Baerveldt implant achieves a lower IOP on fewer medications, but may have more safety issues, including hypotony.19 This article will present the 5-year treatment results of the AVB study.

Methods The AVB study methodology is described in detail in our baseline article17 and will be summarized in the current article. Institutional review board approval was obtained at each clinical center, and the study protocol is registered on www.ClinicalTrials.gov (NCT00940823).

Inclusion Criteria 1. Age 18 years or older. 2. Inadequately controlled glaucoma defined as IOP greater than clinical target despite maximum tolerated medical and laser therapy. 3. Failed trabeculectomy with antimetabolite or disease at high risk of failing trabeculectomy (e.g., neovascular or uveitic glaucoma) and scheduled for aqueous shunt implantation. 4. Eyes with prior surgeries or significant conjunctival scarring were included. 5. One eye enrolled per patient.

2094

Exclusion Criteria 1. Requires an additional procedure at the time of implantation (e.g., cataract surgery, corneal transplant). 2. Unwilling or unable to provide informed consent or adhere to the study requirements, including implant randomization and follow-up visits.

Recruitment and Treatment Patients were recruited from 6 international clinical centers by 9 surgeons from 2005 to 2009 and randomized to an Ahmed-FP7 valve implant or a Baerveldt-350 implant (Appendix, available at: www.aaojournal.com). The surgical technique was standardized, and Baerveldt implants were ligated intraoperatively with a dissolvable or releasable suture. Tube fenestrations were placed anterior to the ligation at the discretion of the performing surgeon in cases deemed requiring immediate postoperative IOP reduction. Follow-up occurred at scheduled intervals, including 9 appointments in the first postoperative year, 2 appointments in the second year, and annually through 5 years. At each visit, IOP, glaucoma medication use, visual acuity, and any complications or interventions related to the implant were recorded.

Outcome Measures The primary outcome was failure, defined as any one of the following: 1. IOP out of target range (5e18 mmHg inclusive) or <20% reduction from baseline for 2 consecutive visits after 3 months. 2. De novo glaucoma surgery required (e.g., cyclodestructive procedure, additional tube shunt). 3. Removal of the implant. 4. Severe vision loss related to the surgery (endophthalmitis, suprachoroidal hemorrhage with vision loss, enucleation, evisceration, or phthisis bulbi) or progression to no light perception for any reason. Success was the absence of failure and was further classified as complete or qualified. Complete success required IOP to be within target range (5e18 mmHg) at all visits after 3 months without the use of glaucoma medications, without any additional surgical interventions, and without significant vision loss (within 2 Snellen lines). Qualified success allowed nonconsecutive visits to be outside of the target IOP range, allowed the use of glaucoma medications, and allowed surgical interventions (including revisions), provided they were not de novo glaucoma procedures. Alternate IOP success criteria of 21 mmHg and 14 mmHg also were analyzed as recommended by the World Glaucoma Association.20 Secondary analyses compared groups on the basis of IOP, medication use, visual acuity, postoperative complications, and interventions required.

Data Censoring Patients meeting the criteria for failure were included in secondary analyses unless they underwent de novo glaucoma procedures or device explantation, or experienced severe vision loss affecting treatment goals. In these cases, IOP and medication use were censored to prevent confounding, but visual acuity, complications, and interventions related to the original surgery were included. Patients who underwent evisceration or enucleation, or progressed to no light perception vision had their visual outcome carried forward for analysis.

Christakis et al



The AVB Study: 5-Year Results

Statistical Analysis Statistical analysis was performed using IBM SPSS (version 23.0, IBM Corp, Armonk, NY) using intention-to-treat analysis. All tests were 2-sided, and statistical significance was defined as P  0.05. Continuous and quantitative variables were compared between groups using the Student t test or ManneWhitney U test. Discrete and qualitative variables were compared using Pearson’s chisquare test, except in cases when >25% of contingency cells had an expected value <5 and Fisher exact test was used. Snellen visual acuity was converted to logarithm of the minimum angle of resolution (logMAR) acuity for analysis, with 0.1 logMAR corresponding to 1 Snellen line. For the purpose of statistical analysis, count fingers was considered 1/100, hand motions was considered 1/800, light perception was considered 1/1600, and no light perception was considered 1/3200.20 Repeated-measures, mixed model analysis of variance was used to compare change in quantitative variables from baseline over time. KaplaneMeier analysis was used to compare failure rates between groups using the log-rank test (Mantel-Cox), as well as to assess risk factors for treatment failure. Multivariate analysis was performed using Cox proportional hazard regression analysis with forward stepwise elimination.

Results Baseline Characteristics A total of 238 patients were recruited and randomized: 124 (52%) to the Ahmed group and 114 (48%) to the Baerveldt group. The baseline characteristics of the study group were similar and are presented in Table 1. Mean patient age was 66 years old with uncontrolled disease as evidenced by a mean preoperative IOP of 31 mmHg on 3 classes of glaucoma medications. Many patients had failed trabeculectomy with antimetabolite (37%) or had highrisk disease, including neovascular (21%) or uveitic (10%) glaucoma. All patients received the implant to which they were randomized. There were few intraoperative complications (4% in each

group), including 1 suprachoroidal hemorrhage requiring drainage in the Ahmed group and 2 serous choroidal detachments in the Baerveldt group that did not require drainage.18 Retention rates were high in both groups, with 72% of patients in each group reaching 5 years of follow-up (Fig 1). There was no difference in the rates of loss to follow-up between the 2 groups (P ¼ 0.51, chi-square test excluding deaths).

Treatment Outcomes At 5 years, the cumulative probability of failure (standard error) was 53.2% (4.7%) in the Ahmed group and 40.0% (5.0%) in the Baerveldt group (P ¼ 0.037, log-rank test) (Fig 2). The most common reason for failure in both groups was high IOP, which resulted in 56 (45%) Ahmed failures and 26 (23%) Baerveldt failures (Table 2). De novo glaucoma surgery was required in 19 patients (15%) in the Ahmed group who failed because of high IOP and 11 patients (10%) in the Baerveldt group. Hypotony resulted in 5 (4%) failures in the Baerveldt group and none in the Ahmed group. Failure due to complications (severe vision loss related to surgery, explantation, evisceration, enucleation, or phthisis bulbi) constituted 7 (6%) Ahmed failures and 9 (8%) Baerveldt failures. The rates of complete success were low in both groups, with only 3 patients (2%) in the Ahmed group and 5 patients (4%) in the Baerveldt group meeting these stringent criteria (P ¼ 0.49). At 5 years, 15 patients (12%) in the Ahmed group and 22 patients (19%) in the Baerveldt group were qualified successes on no glaucoma medications (P ¼ 0.12). Univariate analysis found neovascular glaucoma to be a risk factor for failure (56% failure rate vs. 40% for non-neovascular cases, P ¼ 0.04), although multivariate analysis showed no difference in success rates between devices for neovascular cases (P ¼ 0.20). Not having had a prior trabeculectomy also was a risk factor for failure (52% failure rate vs. 29% for patients with a prior trabeculectomy, P ¼ 0.001). This remained significant even when accounting for known risk factors for failure, including glaucoma subtype (e.g., neovascular, uveitic), young age, and black race.21 However, there was no interaction between device and prior trabeculectomy status (P ¼ 0.54).

Table 1. Baseline Demographic and Ocular Characteristics

Age, mean
SD Sex: female Ethnicity: white Glaucoma diagnosis Open-angle Neovascular Uveitic IOP (mmHg), mean
SD Glaucoma medications, mean
SD Previous surgery, mean
SD Cataract surgery Trabeculectomy Previous lasers, mean
SD Trabeculoplasty Visual acuity, median Snellen

Overall (n [ 238)

Ahmed (n [ 124)

Baerveldt (n [ 114)

P Value

66
16 132 (55%) 170 (71%)

65
17 59 (48%) 91 (73%)

67
15 73 (64%) 79 (69%)

0.29* 0.011y 0.90y

119 (50%) 50 (21%) 23 (10%) 31.4
10.8 3.1
1.0 1.7
1.2 172 (72%) 89 (37%) 0.9
1.1 60 (25%) 20/100

64 (52%) 28 (23%) 10 (8%) 31.1
10.5 3.1
1.0 1.8
1.3 90 (73%) 41 (33%) 0.8
1.1 27 (22%) 20/100

55 (48%) 22 (19%) 13 (11%) 31.7
11.1 3.1
1.1 1.6
1.1 82 (72%) 48 (42%) 1.0
1.1 33 (29%) 20/100

0.82y

0.71* 0.60* 0.35* 0.91y 0.15y 0.17* 0.20y 0.67z

IOP ¼ intraocular pressure; SD ¼ standard deviation. *Student t test. y Pearson chi-square. z ManneWhitney U test.

2095

Ophthalmology Volume 123, Number 10, October 2016 Table 2. Primary Reasons for Failure*

High IOP (>18 mmHg or <20% reduction) De novo glaucoma surgery required Hypotony Device explantation No light perception Enucleation/evisceration Suprachoroidal hemorrhage with vision loss Phthisis bulbi Total Figure 1. Patient retention at 5 years.

When using an alternate IOP criteria of 21 mmHg, the cumulative probability of failure (standard error) was 49.1% (4.7%) in the Ahmed group and 38.3% (5.1%) in the Baerveldt group (P ¼ 0.057) (Fig 3, available at: www.aaojournal.com). When using an IOP criteria of 14 mmHg, the cumulative probability of failure was 83.5% (3.6%) in the Ahmed group and 72.2% (4.5%) in the Baerveldt group (P ¼ 0.050) (Fig 4, available at: www.aaojournal.com).

Intraocular Pressure Both implants were effective in lowering IOP (Table 3, Fig 5). Mean IOP in the Ahmed group decreased from 31.1
10.5 mmHg preoperatively to 16.6
5.9 mmHg at 5 years (47% reduction, P < 0.001). Mean IOP in the Baerveldt group decreased from 31.9
11.0 mmHg preoperatively to 13.6
5.0 mmHg at 5 years (57% reduction, P < 0.001). When comparing the 2 implants, the Ahmed group had a lower mean IOP in the

Ahmed (n [ 124)

Baerveldt (n [ 114)

56 19 0 1 2 3 1 0 63

26 11 5 2 3 0 2 2 40

IOP ¼ intraocular pressure. *The first reason for failure is reported (e.g., if a patient fails because of high IOP and later progresses to no light perception, high IOP is considered the primary reason for failure).

early postoperative period (1 day, 1 week, 2 weeks) and the Baerveldt group had a lower mean IOP at all visits from 1 year onward (P < 0.05, Table 3). Repeated-measures, mixed-model analysis of variance demonstrated that both groups had significant mean IOP reduction over time (both P < 0.001), but that the Baerveldt group had a greater IOP reduction than the Ahmed group at all follow-up visits beginning at 1 year and continuing to 5 years (P ¼ 0.02).

Glaucoma Medication Use Both implants were effective in lowering mean glaucoma medication use (Table 3, Fig 6). Mean glaucoma medication use in the Ahmed group decreased from 3.1
1.0 mmHg preoperatively to 1.8
1.5 mmHg at 5 years (44% reduction, P < 0.001). Mean glaucoma medication use in the Baerveldt group decreased from 3.1
1.0 mmHg preoperatively to 1.2
1.3 mmHg at 5 years (61% reduction, P < 0.001). When comparing the 2 implants, Table 3. Mean Intraocular Pressure and Medication Use

Baseline IOP (mmHg) Glaucoma medications 1 year IOP (mmHg) Glaucoma medications 2 years IOP (mmHg) Glaucoma medications 3 years IOP (mmHg) Glaucoma medications 4 years IOP (mmHg) Glaucoma medications 5 years IOP (mmHg) Glaucoma medications Figure 2. Cumulative failure rate (5 mmHg  intraocular pressure [IOP] criteria 18 mmHg). KaplaneMeier analysis using the log-rank test. Censored patients are denoted by vertical tick marks. Errors bars showing the standard error of the mean are included at 1 year, 3 years, and 5 years.

2096

Ahmed

Baerveldt

P Value*

31.1
10.5 3.1
1.0

31.9
11.0 3.1
1.0

0.60 0.75

16.5
5.1 1.7
1.3

13.8
4.6 1.2
1.3

<0.001 0.017

16.5
5.1 1.8
1.4

13.8
4.6 1.0
1.2

0.026 <0.001

16.2
4.8 1.8
1.4

14.6þ5.0 1.1
1.3

0.047 <0.001

16.0
4.5 1.9
1.5

14.4
4.8 1.2
1.3

0.049 0.004

16.6
5.9 1.8
1.5

13.6
5.0 1.2
1.3

0.001 0.033

IOP ¼ intraocular pressure. Data presented as mean
standard deviation. *Student t test.

Christakis et al



The AVB Study: 5-Year Results

Figure 5. Mean intraocular pressure (IOP) over 5 years of follow-up. Error bars represent standard deviation.

the Ahmed group required fewer glaucoma medications in the early postoperative period (1 day, 1 week, 2 weeks) and the Baerveldt group required fewer medications at all visits after 1 month (P < 0.05). At 5 years, the Baerveldt group required a median of 1 glaucoma medication (confidence interval, 0e2) compared with 2 glaucoma medications (confidence interval, 0e2) in the Ahmed group (P ¼ 0.038, ManneWhitney U test).

Visual Acuity Both groups had a moderate but similar reduction in visual acuity over 5 years of follow-up (P ¼ 0.88). Mean logMAR visual acuity in the Ahmed group decreased from 1.3
1.0 at baseline to 1.6
1.0 at 5 years (P < 0.001). Mean logMAR visual acuity in the Baerveldt group decreased from 1.2
1.0 at baseline to 1.6
1.2 at 5 years (P < 0.001). The distribution of change in Snellen visual acuity from baseline to 5 years is shown in Figure 7.

De Novo Glaucoma Procedures The cumulative proportion of patients requiring de novo glaucoma surgery (standard error) was 17.7% (3.7%) in the Ahmed group and 11.0% (3.2%) in the Baerveldt group (P ¼ 0.22) (Fig 8). To evaluate for a potential bias in the decision to reoperate, the mean IOP and medication use at the time de novo glaucoma surgery was scheduled were compared between devices. At this time, the 2 groups had similar mean IOPs (Ahmed 30.0
10.8 mmHg, Baerveldt 28.9
10.9 mmHg, P ¼ 0.79) and were taking a similar mean number of glaucoma medications (Ahmed 2.8
1.2, Baerveldt 1.8
1.5, P ¼ 0.06). The most common procedure was cyclodestruction, which was required in 10 patients (8%) in the Ahmed group and 3 patients (3%) in the Baerveldt group (P ¼ 0.09). Transscleral cyclophotocoagulation was used in all cases except for 2 cases of malignant glaucoma in which endocyclophotocoagulation was used. Placement of an additional aqueous shunt (all Baerveldt implants) was required in

Figure 6. Mean glaucoma medication use over 5 years of follow-up. Error bars represent standard deviation.

2097

Ophthalmology Volume 123, Number 10, October 2016

Figure 7. Distribution of change in visual acuity from baseline to 5 years.

7 patients (6%) in the Ahmed group and 4 patients (4%) in the Baerveldt group. A gold shunt was required in 1 patient (1%) in the Ahmed group and 3 patients (3%) in the Baerveldt group. Selective laser trabeculoplasty was performed in 1 patient (1%) in each group with an IOP modestly higher than target.

Censoring A total of 43 patients were censored from analysis: 25 (20%) in the Ahmed group and 18 (16%) in the Baerveldt group (P ¼ 0.25). The reasons for censoring included additional glaucoma surgery

(Ahmed 15%, Baerveldt 10%), device explantation (Ahmed 2%, Baerveldt 3%), enucleation or evisceration (Ahmed 2%, Baerveldt 0%), and vision loss altering treatment goals (Ahmed 0%, Baerveldt 4%). Censoring due to vision loss occurred in 4 patients in the Baerveldt group; 2 patients developed phthisis bulbi, 1 patient had a large suprachoroidal hemorrhage, and 1 patient progressed to no light perception and glaucoma medications were withdrawn.

Complications After 5 years of follow-up, 78 patients (63%) in the Ahmed group and 79 patients (69%) in the Baerveldt group experienced a complication (P ¼ 0.30) (Table 4). The most common complications were shallow anterior chamber (15% Ahmed, 17% Baerveldt), choroidal effusions (13% Ahmed, 16% Baerveldt), tube complications (13% Ahmed, 16% Baerveldt), persistent corneal edema (11% Ahmed, 12% Baerveldt), and persistent iritis (7% Ahmed, 12% Baerveldt). Clinical bleb encapsulation was defined as visible fibrosis of the endplate in the setting of an elevated IOP and was more common in the Ahmed group (11% Ahmed, 4% Baerveldt, P ¼ 0.023). Refractory hypotony needing surgery was required in 1 patient (1%) in the Ahmed group and 6 patients (5%) in the Baerveldt group (P ¼ 0.057). Of these patients, 1 in each group had early postoperative overfiltration requiring temporary tube ligation with an absorbable suture; in both cases, hypotony resolved and patients were qualified successes at last follow-up. Persistent diplopia or a motility disorder occurred in 6 patients (5%) in the Ahmed group and 2 patients (2%) in the Baerveldt group (P ¼ 0.28).

Interventions

Figure 8. Cumulative de novo glaucoma reoperation rates. KaplaneMeier analysis using the log-rank test. Censored patients are denoted by vertical tick marks. Errors bars showing the standard error of the mean are included at 1 year, 3 years, and 5 years.

2098

After 5 years of follow-up, 64 patients (51%) in the Ahmed group and 58 patients (51%) in the Baerveldt group required an intervention (P ¼ 0.99) (Table 5). The most common interventions were anterior chamber reformation (Ahmed 11%, Baerveldt 13%), paracentesis (4% Ahmed, 14% Baerveldt, P ¼ 0.007), tube-related interventions (10% Ahmed, 17% Baerveldt), cataract surgery (26% Ahmed, 38% Baerveldt), vitrectomy (7% Ahmed, 7% Baerveldt), and corneal transplantation (7% Ahmed, 4% Baerveldt).

Christakis et al



The AVB Study: 5-Year Results

Table 4. Complications in 5 Years of Follow-up Complication Shallow anterior chamber Choroidal effusions Tube complications Tube obstruction Tube malposition Tube erosion Corneal edema Iritis Cataract progression Encapsulated bleb Hyphema Motility disorder Aqueous misdirection Suprachoroidal hemorrhage Phthisis bulbi Retinal detachment Endophthalmitis Progression to no light perception Hypotony requiring surgery High IOP requiring de novo surgery Other Total

Ahmed (n [ 124)

Baerveldt (n [ 114)

P Value

18 16 17 7 8 5 14 9 11 14 4 6 2 2 1 1 1 7 1 19 6 78

19 18 19 10 7 2 14 14 13 4 6 2 4 3 2

0.65y 0.53y 0.41y 0.35y 0.92y 0.45z 0.81y 0.19y 0.49y 0.023y 0.43y 0.28z 0.43z 0.67z 0.61z 1.0z 1.0z 1.0y 0.057z 0.19y 0.66y 0.30y

(15%) (13%) (14%) (6%) (6%) (4%) (11%) (7%) (32%)* (11%) (3%) (5%) (2%) (2%) (1%) (1%) (1%) (6%) (1%) (15%) (5%) (63%)

7 6 11 7 79

(17%) (16%) (17%) (9%) (6%) (2%) (12%) (12%) (41%)* (4%) (5%) (2%) (4%) (3%) (2%) 0 0 (6%) (5%) (10%) (6%) (69%)

IOP ¼ intraocular pressure. *Corrected for number of phakic patients. y Pearson chi-square test. z Fisher exact test.

Table 5. Interventions in 5 Years of Follow-up Interventions

Ahmed (n [ 124)

Baerveldt (n [ 114)

P Value

Anterior chamber reformation Paracentesis Tube interventions Laser for obstruction Iris sweep Tube irrigation Tube ligation Tube reposition Tube patch graft Tube explanted Phacoemulsification Vitrectomyx Corneal transplant Bleb needling Laser iridotomy Nd:YAG capsulotomy Nd:YAG hyaloidotomy Drainage suprachoroidal hemorrhage Device explanted Enucleation/evisceration De novo glaucoma surgery Other Total

13 (11%) 5 (4%) 12 (10%) 0 1 (1%) 1 (1%) 2 (2%) 6 (5%) 2 (2%) 1 (1%) 9 (26%)* 8 (7%) 8 (7%) 4 (3%) 2 (2%) 5 (4%) 0 1 (1%) 3 (3%) 3 (3%) 19 (15%) 4 (4%) 63 (51%)

15 16 19 3 4 2 4 7

0.52y 0.007y 0.11y 0.11z 0.20z 0.61z 0.43z 0.66y 0.50z 1.0z 0.92y 0.86y 0.48y 1.0z 1.0z 1.0z 0.48z 0.61z 0.92z 0.25z 0.19y 1.0z 0.99y

1 12 8 5 3 2 4 1 2 3 11 4 58

(13%) (14%) (17%) (3%) (4%) (2%) (4%) (6%) 0 (1%) (38%)* (7%) (4%) (3%) (2%) (4%) (1%) (2%) (3%) 0 (10%) (4%) (51%)

Nd:YAG ¼ neodymium doped yttrium aluminum garnet. *Corrected for phakic patients. y Pearson chi-square test. z Fisher exact test. x Includes unplanned vitrectomies during initial surgery.

Device explantation was required in 3 patients (3%) in each group. In the Ahmed group, explantation occurred for malposition and corneal decompensation (1 patient) and to allow for subsequent replacement with a Baerveldt implant (2 patients). In the Baerveldt group, explantation occurred for refractory hypotony (1 patient), persistent motility disorder (1 patient), and corneal decompensation (1 patient). Enucleation was required in 1 patient in the Ahmed group for endophthalmitis, and evisceration for a blind painful eye was required in 2 patients in the Ahmed group.

Discussion The AVB study is an international, multicenter, randomized trial comparing the 2 most frequently used aqueous shunts. Enrolled patients had uncontrolled glaucoma despite maximum tolerated medical therapy, with 37% of patients having failed trabeculectomy with mitomycin C and 31% of patients having neovascular or uveitic glaucoma. Aqueous shunts are being increasingly used in this patient population.12 An IOP target of 18 mmHg was selected on the basis of Advanced Glaucoma Intervention Study data suggesting that higher pressures may be insufficient to prevent disease progression, and our patient population required on average a 40% reduction in IOP to meet this target.2 De novo glaucoma surgery, severe vision loss secondary to a complication of surgery, and progression to no light perception were considered failures. After 5 years of follow-up, the Ahmed group had a higher failure rate of 53% compared with 40% in the Baerveldt group (P ¼ 0.04). This failure rate is similar to the 10% failure rate per year quoted by the American Academy of Ophthalmology report on aqueous shunts.7 The main reason for failure in both groups was high IOP, and the cumulative glaucoma reoperation rate was 18% in the Ahmed group and 11% in the Baerveldt group (P ¼ 0.22). The most common de novo glaucoma surgery was cyclodestruction (8% Ahmed, 3% Baerveldt) and Baerveldt implantation (6% Ahmed, 4% Baerveldt). Visual potential guided de novo glaucoma procedure selection; the majority of patients who underwent cyclodestruction had count fingers or worse vision compared with 20/60 or better vision in patients who underwent additional Baerveldt implantation. Failure due to hypotony was rare, occurring in 5 patients (4%) in the Baerveldt group and zero patients in the Ahmed group (P ¼ 0.02). Hypotony is a known complication of Baerveldt implantation given the lack of built-in flow restriction and may occur if there is a ligature malfunction or insufficient healing around the plate to regulate flow once the tube is patent.10 Neovascular glaucoma was a risk factor for failure, as has been reported in the literature,22 although no device conferred an advantage in these patients. Not having had a prior trabeculectomy also was a risk factor for failure, which we attribute to a selection bias in that patients scheduled for a primary tube likely have disease at higher risk for failure. Although our multivariate model attempted to correct for the presence of known risk factors for failure, including neovascular glaucoma, younger age, and black race, factors such as poor conjunctival health

2099

Ophthalmology Volume 123, Number 10, October 2016 and long-standing antiglaucoma medication use are difficult to account for.21 However, an important finding is that patients in our study cohort who had previously failed trabeculectomy had a 71% success rate at 5 years, suggesting that tube surgery may be an effective treatment option in these patients. Both implants were effective in lowering IOP, but the Baerveldt provided a 3 mmHg lower IOP at 5 years (Ahmed 16.6 mmHg, Baerveldt 13.6 mmHg, P ¼ 0.001). Both implants also reduced the need for glaucoma medications, and the Baerveldt group required a median of 1 glaucoma medication at 5 years compared with 2 glaucoma medications in the Ahmed group (P ¼ 0.04). The lower IOP and need for medications may be attributed to the Baerveldt implant having a larger end-plate size (184 mm2 Ahmed, 350 mm2 Baerveldt), and clinical studies have shown a relationship between end-plate size and filtration.23,24 The Ahmed group also had a higher rate of clinical bleb encapsulation than the Baerveldt group (11% Ahmed, 4% Baerveldt, P ¼ 0.02), as has been shown in previous retrospective comparisons.9 The early exposure of the Ahmed bleb to mechanical compression from aqueous flow, as well as exposure to proinflammatory cytokines incited by surgery, has been postulated to stimulate bleb encapsulation and influence the fibrosis pathway.25e29 In addition, although both implants are made of silicone, electron microscopy has shown that the Ahmed implant topography has more a 10-fold greater roughness than the Baerveldt implant, which in vitro was associated with increased tenon fibroblast adhesion that may predispose to fibrous encapsulation.30 Both groups experienced a moderate but similar reduction in visual acuity over the duration of the study, with 43% of the Ahmed group and 46% of the Baerveldt group losing 2 or more lines of vision. Visual outcomes were modestly better in patients with 20/50 or better vision at baseline (n ¼ 80), with 36% of patients losing 2 or more lines of vision after 5 years of follow-up. Ascertaining the cause of vision loss was difficult because of the high rates of concomitant eye disease (e.g., diabetic retinopathy, corneal pathology). Progression to no light perception occurred in 6% of both groups; half (7/14) of these patients had neovascular glaucoma. Both implants had high rates of postoperative complications, although most were transient or required minimal intervention. The most common early postoperative complication was hypotony, resulting in shallow anterior chambers (Ahmed 15%, Baerveldt 17%) and choroidal effusions (Ahmed 13%, Baerveldt 16%), which resolved over time. Early postoperative IOP spikes requiring paracentesis were more common in the Baerveldt group while the tubes were ligated (Ahmed 4%, Baerveldt 14%, P ¼ 0.007). Tube complications occurred in 14% of the Ahmed group and 17% of the Baerveldt group, with tube obstruction and tube malposition being the most common. Tube interventions were required in 10% of the Ahmed group and 17% of the Baerveldt group (P ¼ 0.11), with tube reposition being the most common. Serious early postoperative complications included suprachoroidal hemorrhage, which occurred in 3 patients (3%) in the Baerveldt group, in whom 2 required

2100

drainage. One patient in the Ahmed group had an intraoperative suprachoroidal hemorrhage that required drainage, and 1 patient had a suprachoroidal hemorrhage after a tube reposition at 18 months. The most common long-term complication was corneal edema, which affected 11% of the Ahmed group and 12% of the Baerveldt group, of whom 7% of the Ahmed group and 4% of the Baerveldt group required a corneal transplant. This is similar to the TVT study, which reported a 16% persistent corneal edema rate at 5 years, with a 9% corneal graft rate.5 The high rates of corneal decompensation are likely a result of endothelial cell failure from multiple prior surgeries, postoperative IOP fluctuations (especially hypotony and flat anterior chamber, tube-cornea touch) and chronic inflammation, which can accelerate endothelial cell loss.29,31 Persistent iritis occurred in 7% of patients in the Ahmed group and 12% of patients in the Baerveldt group (P ¼ 0.19). The rate of motility disorder or persistent diplopia was 5% in the Ahmed group and 2% in the Baerveldt group (P ¼ 0.28), which is lower than the 6% reported in the TVT study and may reflect the worse vision of our patient population, revealing less symptomatic diplopia.5 Cataract progression occurred frequently in both groups, with 26% of phakic Ahmed cases and 38% of phakic Baerveldt cases requiring cataract surgery within 5 years of device implantation. Our results are comparable to the Ahmed Baerveldt Comparison (ABC) study, a concurrent independent randomized trial comparing the Ahmed implant with the Baerveldt implant.32,33 Baseline patient characteristics in the ABC study were similar to our study (AVB), with the study group having a mean preoperative IOP of 31.5 mmHg (AVB: 31.4 mmHg) on an average of 3.4 glaucoma medications (AVB: 3.1). Many patients had secondary glaucoma, including 29% with neovascular glaucoma (AVB: 21%) and 7% with uveitic glaucoma (AVB: 10%). Eighty percent of patients had previously undergone ocular surgery (AVB: 89%), including 42% who had failed prior trabeculectomy (AVB: 37%). Study design and outcome measures were similar between studies, although the primary IOP criterion for the ABC study was 6 to 21 mmHg inclusive compared with 5 to 18 mmHg inclusive for the AVB study. However, both study groups followed the World Glaucoma Association consensus guidelines and analyzed alternate IOP criteria to allow for valid comparison with one another. At 5 years, the ABC study reported failure rates of 60% in the Ahmed group and 46% in the Baerveldt group (using an IOP criterion of <18 mmHg), with the majority of failures in both groups secondary to inadequate IOP reduction.33 The Baerveldt achieved a 2 mmHg lower IOP, and a greater proportion of patients in the Ahmed group required de novo glaucoma surgery.33 However, more patients in the Baerveldt group failed because of persistent hypotony, device explantation, and loss of light perception.33 Study Limitations There are several limitations that should be considered when interpreting the results of the AVB study. Although IOP

Christakis et al



The AVB Study: 5-Year Results

criteria are used in glaucoma surgical trials because they are the only known modifiable risk factor for disease progression, they remain a surrogate marker of success.1,2 An IOP target of 18 mmHg was selected on the basis of Advanced Glaucoma Intervention Study data2; however, each patient had his/her own individual target IOP that may have been higher or lower than this and guided treatment decisions. A stable patient with an IOP of 19 mmHg on no medications would be considered a failure, whereas a progressing patient with an IOP of 15 mmHg receiving maximum tolerated medical therapy would be considered a success. Likewise, hypotony was defined as <5 mmHg as per the World Glaucoma Association guidelines,20 although it should be a clinical diagnosis that incorporates visual function and whether there are any hypotonyrelated complications, including shallow anterior chamber, corneal edema, or kissing choroidal effusions. Simply using an IOP cutoff would classify a patient with an IOP of 4 mmHg and excellent vision as a failure, whereas a patient with an IOP of 6 mmHg and poor vision secondary to hypotony maculopathy would be classified as a success. In addition, averaging IOP over a population may not appropriately reflect clinical outcomes and is susceptible to skew from outliers or bias introduced from censoring patients who underwent de novo glaucoma surgery or experienced visionthreatening complications. Thus, applying our results to clinical decision making requires careful balancing of success criteria, IOP, medication use, de novo glaucoma surgery and device-related complications, and interventions required. Furthermore, although IOP is useful clinically to monitor glaucoma given that it is quantifiable and reproducible, the main goal of glaucoma therapy is preserving visual function. Unfortunately, the poor baseline visual acuity of our patient population made analyzing visual fields difficult, although they were used to guide clinical management. Assessing structural outcomes such as glaucomatous cupping also would provide useful data if there was a standardized means of comparison; however, the study commenced before the routine use of ocular tomography. Our results also may have been influenced by our surgeon’s relative experience and skill with each device. To minimize this bias, only surgeons with extensive experience in implanting both shunts participated in the study as determined by the study chair (IIKA). In conclusion, both the Ahmed FP7-valve implant and the Baerveldt-350 implant were effective in lowering IOP and the need for glaucoma medications in patients with refractory or high-risk glaucoma. When comparing the implants, the Baerveldt implant had a lower failure rate and yielded a moderately lower IOP on fewer medications than the Ahmed implant. Thus, the Baerveldt implant may be a good choice in patients with a low long-term IOP target or in patients intolerant to topical medications. However, patients need to be followed closely in the early postoperative period while the tube is ligated because IOP volatility is common, and there is a risk of hypotony. The Ahmed valve implant may be a good choice in patients who need an immediate postoperative reduction in IOP and have moderate long-term IOP targets. However, glaucoma medications and additional glaucoma surgery may be necessary to achieve this goal. Ultimately,

selecting a device should balance patient factors, such as goals of therapy, target IOP, and medication compliance, with surgeon factors, including familiarity with each device and personal outcomes.

References 1. Heijl A, Leske MC, Bengtsson B, et al. Reduction of intraocular pressure and glaucoma progression: results from the Early Manifest Glaucoma Trial. Arch Ophthalmol 2002;120: 1268–79. 2. The Advanced Glaucoma Intervention Study (AGIS): 7. The relationship between control of intraocular pressure and visual field deterioration. The AGIS Investigators. Am J Ophthalmol 2000;130:429–40. 3. Prum BE Jr, Rosenberg LF, Gedde SJ, et al. Primary OpenAngle Glaucoma Preferred Practice Pattern((R)) Guidelines. Ophthalmology 2016;123:P41–111. 4. Gedde SJ, Schiffman JC, Feuer WJ, et al. Treatment outcomes in the Tube Versus Trabeculectomy (TVT) study after five years of follow-up. Am J Ophthalmol 2012;153:789–803.e2. 5. Gedde SJ, Herndon LW, Brandt JD, et al. Postoperative complications in the Tube Versus Trabeculectomy (TVT) study during five years of follow-up. Am J Ophthalmol 2012;153:804–814.e1. 6. Five-year follow-up of the Fluorouracil Filtering Surgery Study. The Fluorouracil Filtering Surgery Study Group. Am J Ophthalmol 1996;121:349–66. 7. Minckler DS, Francis BA, Hodapp EA, et al. Aqueous shunts in glaucoma: a report by the American Academy of Ophthalmology. Ophthalmology 2008;115:1089–98. 8. Ayyala RS, Zurakowski D, Smith JA, et al. A clinical study of the Ahmed glaucoma valve implant in advanced glaucoma. Ophthalmology 1998;105:1968–76. 9. Tsai JC, Johnson CC, Kammer JA, Dietrich MS. The Ahmed shunt versus the Baerveldt shunt for refractory glaucoma II: longer-term outcomes from a single surgeon. Ophthalmology 2006;113:913–7. 10. Stein JD, McCoy AN, Asrani S, et al. Surgical management of hypotony owing to overfiltration in eyes receiving glaucoma drainage devices. J Glaucoma 2009;18:638–41. 11. Nguyen QH, Budenz DL, Parrish RK 2nd. Complications of Baerveldt glaucoma drainage implants. Arch Ophthalmol 1998;116:571–5. 12. Desai MA, Gedde SJ, Feuer WJ, et al. Practice preferences for glaucoma surgery: a survey of the American Glaucoma Society in 2008. Ophthalmic Surg Lasers Imaging 2011;42:202–8. 13. HaiBo T, Xin K, ShiHeng L, Lin L. Comparison of Ahmed glaucoma valve implantation and trabeculectomy for glaucoma: a systematic review and meta-analysis. PLoS One 2015;10:e0118142. 14. Goulet RJ 3rd, Phan AD, Cantor LB, WuDunn D. Efficacy of the Ahmed S2 glaucoma valve compared with the Baerveldt 250-mm2 glaucoma implant. Ophthalmology 2008;115: 1141–7. 15. Syed HM, Law SK, Nam SH, et al. Baerveldt-350 implant versus Ahmed valve for refractory glaucoma: a case-controlled comparison. J Glaucoma 2004;13:38–45. 16. Wang JC, See JL, Chew PT. Experience with the use of Baerveldt and Ahmed glaucoma drainage implants in an Asian population. Ophthalmology 2004;111:1383–8. 17. Christakis PG, Tsai JC, Zurakowski D, et al. The Ahmed Versus Baerveldt study: design, baseline patient characteristics, and intraoperative complications. Ophthalmology 2011;118:2172–9.

2101

Ophthalmology Volume 123, Number 10, October 2016 18. Christakis PG, Kalenak JW, Zurakowski D, et al. The Ahmed Versus Baerveldt study: one-year treatment outcomes. Ophthalmology 2011;118:2180–9. 19. Christakis PG, Tsai JC, Kalenak JW, et al. The Ahmed versus Baerveldt study: three-year treatment outcomes. Ophthalmology 2013;120:2232–40. 20. Heuer DK, Barton K, Grehn F, et al. Consensus on definitions of success. In: Shaarawy TM, Sherwood MB, Grehn F, eds. Guidelines on Design and Reporting of Surgical Trials. Amsterdam, The Netherlands: Kugler Publications; 2008. 21. Broadway DC, Chang LP. Trabeculectomy, risk factors for failure and the preoperative state of the conjunctiva. J Glaucoma 2001;10:237–49. 22. Lavin MJ, Franks WA, Wormald RP, Hitchings RA. Clinical risk factors for failure in glaucoma tube surgery. A comparison of three tube designs. Arch Ophthalmol 1992;110:480–5. 23. Heuer DK, Lloyd MA, Abrams DA, et al. Which is better? One or two? A randomized clinical trial of single-plate versus double-plate Molteno implantation for glaucomas in aphakia and pseudophakia. Ophthalmology 1992;99:1512–9. 24. Lloyd MA, Baerveldt G, Fellenbaum PS, et al. Intermediate-term results of a randomized clinical trial of the 350- versus the 500mm2 Baerveldt implant. Ophthalmology 1994;101:1456–64. 25. Molteno AC, Fucik M, Dempster AG, Bevin TH. Otago Glaucoma Surgery Outcome Study: factors controlling capsule fibrosis around Molteno implants with histopathological correlation. Ophthalmology 2003;110:2198–206.

26. Tsai JC, Johnson CC, Dietrich MS. The Ahmed shunt versus the Baerveldt shunt for refractory glaucoma: a singlesurgeon comparison of outcome. Ophthalmology 2003;110: 1814–21. 27. Mahale A, Othman MW, Al Shahwan S, et al. Altered expression of fibrosis genes in capsules of failed Ahmed glaucoma valve implants. PLoS One 2015;10:e0122409. 28. Freedman J, Iserovich P. Pro-inflammatory cytokines in glaucomatous aqueous and encysted Molteno implant blebs and their relationship to pressure. Invest Ophthalmol Vis Sci 2013;54:4851–5. 29. Anshu A, Price MO, Richardson MR, et al. Alterations in the aqueous humor proteome in patients with a glaucoma shunt device. Mol Vis 2011;17:1891–900. 30. Choritz L, Koynov K, Renieri G, et al. Surface topographies of glaucoma drainage devices and their influence on human tenon fibroblast adhesion. Invest Ophthalmol Vis Sci 2010;51: 4047–53. 31. McDermott ML, Swendris RP, Shin DH, et al. Corneal endothelial cell counts after Molteno implantation. Am J Ophthalmol 1993;115:93–6. 32. Barton K, Gedde SJ, Budenz DL, et al. The Ahmed Baerveldt Comparison Study methodology, baseline patient characteristics, and intraoperative complications. Ophthalmology 2011;118:435–42. 33. Budenz DL, Barton K, Gedde SJ, et al. Five-year treatment outcomes in the Ahmed Baerveldt comparison study. Ophthalmology 2015;122:308–16.

Footnotes and Financial Disclosures Originally received: January 18, 2016. Final revision: June 8, 2016. Accepted: June 9, 2016. Available online: August 18, 2016.

Financial Disclosure(s): The author(s) have made the following disclosure(s): P.J.H.: Personal fees e Abbott Medical Optics. Manuscript no. 2016-114.

I.I.K.A.: Personal fees e Abbott Medical Optics and New World Medical.

Department of Ophthalmology and Vision Sciences, University of Toronto Faculty of Medicine, Toronto, Canada.

The AVB study is supported by a research grant from the Glaucoma Research Society of Canada, Toronto, Canada (I.I.K.A., P.G.C.) and a departmental challenge grant from the Research to Prevent Blindness, Inc, New York, New York (J.C.T.). Our research funding sponsors had no role in the design or conduct of the study.

1

2

Drs. Massaro & Kalenak SC, Milwaukee, Wisconsin.

3

Department of Ophthalmology, Icahn School of Medicine at Mount Sinai, New York, New York.

4

Department of Anesthesiology, Children’s Hospital Boston, Harvard Medical School, Boston, Massachusetts.

Author Contributions: Conception and design: Christakis, Kalenak, Tsai, Zurakowski, Kammer, Harasymowycz, Mura, Cantor, Ahmed

5

Department of Ophthalmology and Visual Sciences, Vanderbilt University School of Medicine, Nashville, Tennessee.

Data collection: Christakis, Kalenak, Tsai, Zurakowski, Kammer, Harasymowycz, Mura, Cantor, Ahmed

6

Department of Ophthalmology, University of Montreal, Montreal, Canada.

Analysis and interpretation: Christakis, Kalenak, Tsai, Zurakowski, Kammer, Harasymowycz, Mura, Cantor, Ahmed

7

Obtained funding: Not applicable

Clinic of Las Condes, Santiago, Chile.

8

Eugene and Marilyn Glick Eye Institute, Indiana University School of Medicine, Indianapolis, Indiana. Presented at: American Academy of Ophthalmology, October 22e25, 2011, Orlando, Florida; November 10e13, 2012, Chicago, Illinois; November 13e17, 2015, Las Vegas, Nevada; American Glaucoma Society, March 1e4, 2012, New York, New York; Canadian Ophthalmological Society, June 26e29, 2010, Quebec City, Canada; June 9e12, 2011, Vancouver, Canada; June 26e30, 2012, Toronto, Canada.

2102

Overall responsibility: Christakis, Kalenak, Tsai, Zurakowski, Kammer, Harasymowycz, Mura, Cantor, Ahmed Abbreviations and Acronyms: ABC ¼ Ahmed Baerveldt Comparison; AVB ¼ Ahmed Versus Baerveldt; IOP ¼ intraocular pressure; logMAR ¼ logarithm of the minimum angle of resolution; TVT ¼ Tube Versus Trabeculectomy. Correspondence: Iqbal I.K. Ahmed, MD, 3200 Erin Mills Parkway, Mississauga, Ontario L5L 1W8. E-mail: [email protected].