CASE-MATCHED COMPARISON OF VITRECTOMY, PERIPHERAL RETINAL ENDOLASER, AND ENDOCYCLOPHOTOCOAGULATION VERSUS STANDARD CARE IN NEOVASCULAR GLAUCOMA KYLE V. MARRA, BS,*† SUSHANT WAGLEY, AB,‡ AHMED OMAR, MD,*§¶ TAIGA KINOSHITA, MD,* KYLE D. KOVACS, AB,** PAOLO SILVA, MD,§ MARK C. KUPERWASER, MD,* JORGE G. ARROYO, MD, MPH* Purpose: To evaluate the efficacy of combination pars plana vitrectomy, endoscopic peripheral panretinal photocoagulation, and endocyclophotocoagulation (ECP) as compared with standard care in patients with neovascular glaucoma. Methods: This age-matched case-controlled retrospective series of 54 eyes compared the clinical outcomes between a consecutive series of combination pars plana vitrectomy/ panretinal photocoagulation/ECP (n = 27) versus the current standard of care (n = 27) for patients with neovascular glaucoma. “Standard” treatments for patients with neovascular glaucoma include panretinal photocoagulation, intravitreal bevacizumab, filtration surgery, pars plana vitrectomy, and Ahmed valve placement. Results: After 1 year, mean intraocular pressure reduced from 40.7 ± 12.40 mmHg preoperatively to 12.3 ± 4.84 mmHg (P , 0.001) in the ECP group and from 34.7 ± 12.38 mmHg to 23.2 ± 12.34 mmHg in the control group (P = 0.002). Compared with controls, the mean drop in intraocular pressure in the ECP group was significantly greater at all postoperative visits. Logarithm of the minimal angle of resolution visual acuity outcomes were similar in both groups. There were 2 cases (7.4%) of postoperative phthisis bulbi in each group. Conclusion: Endoscopic pars plana vitrectomy, panretinal photocoagulation, and ECP seem to control intraocular pressure to a greater extent than standard glaucoma treatments in patients with neovascular glaucoma. In this aged-matched comparative case series, there was no significant difference between the two treatments’ effects on visual acuity. RETINA 35:1072–1083, 2015

N

In 2001, Sivak-Callcottet al reviewed previously published reports of NVG due to ischemic retinopathies. The authors concluded that complete panretinal photocoagulation (PRP), antiangiogenic drugs (such as bevacizumab), medical reduction of IOP (glaucoma drops) and inflammation (such as steroid drops or injections), and glaucoma surgery (such as filtration and Ahmed valve placement) were the most recommended treatments of NVG.1 These treatments have since become the standard managements for NVG. Unfortunately, filtration surgery or tube shunt placement shows limited IOP reduction in profoundly ischemic eyes associated with NVG

eovascular glaucoma (NVG) is a visionthreatening disease caused by retinal or ocular ischemia, most commonly due to proliferative diabetic retinopathy (PDR) or central retinal vein occlusion (CRVO). Neovascular glaucoma often results in loss of vision and severe pain caused by rapid elevation of intraocular pressure (IOP). Surgical procedures to reduce IOP are generally classified into the following 2 approaches: 1) increasing outflow, which includes filtration surgery or tube shunt placement and 2) reducing inflow, which includes transscleral cyclophotocoagulation, cyclocryotherapy, or endoscopic cyclophotocoagulation (ECP). 1072

ENDOCYCLOPHOTOCOAGULATION FOR NVG  MARRA ET AL

possibly because of excess angiogenic factors that lead to neovascularization and inflammation.2–8 However, as reduction of inflow by targeted destruction of the ciliary body through ECP shows marked IOP reduction, this treatment may be best suited for managing severe NVG.9 Endoscopic cyclophotocoagulation allows a surgeon to endoscopically treat the ciliary body with the least collateral tissue damage currently possible. Thus, this technique may be superior to transscleral cyclocoagulation for the treatment of NVG.7,10–13 Furthermore, the endoscopic laser through the pars plana approach facilitates the delivery of 360° of near-confluent peripheral retinal laser treatment up to the ora serrata, whereas such extensive treatment is more difficult to accomplish using the standard externally delivered endolaser PRP in the operating room or slit-lamp laser in the clinic. Excess photocoagulation of the ciliary body, however, may cause early complications such as severe inflammation and intense pain or late complications such as hypotony or phthisis and marked visual field constriction.9,14,15 Our hypothesis was that endoscopic peripheral retinal photocoagulation and ECP would better control ocular ischemia and IOP than standard treatments. This comparative age-matched study evaluates the efficacy and complications of combination of pars plana vitrectomy (PPV), endoscopic PRP, and ECP surgery versus standard treatments (PRP, injection of bevacizumab [IVB], filtration surgery, or Ahmed valve placement) for severe NVG. Participants and Methods This study was submitted to the Beth Israel Deaconess Medical Center (BIDMC) Institutional Review Board for approval to ensure compliance with the tenets of the Declaration of Helsinki. Informed consent was waived because of its retrospective nature. A case-matched comparative case series of 54 eyes from 53 patients with iris neovascularization (INV) and glaucoma secondary to retinal ischemia From the *Retina Service, Division of Ophthalmology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts; †University of California, San Diego School of Medicine, San Diego, California; ‡College of Human Medicine, Michigan State University, East Lansing, Michigan; §Beetham Eye Institute, Joslin Diabetes Center, Harvard Medical School, Boston, Massachusetts; ¶Department of Ophthalmology, Faculty of Medicine, Assiut University, Assiut, Egypt; and **Albert Einstein College of Medicine, New York, New York. Supported by the Grimshaw-Gudewicz Charitable Foundation. None of the authors have any conflicting interests to disclose. Reprint requests: Jorge G. Arroyo, MD, MPH, Retina Service, Beth Israel Deaconess Medical Center, Harvard Medical School, 330 Brookline Avenue, CC-5, Boston, MA 02215; e-mail: [email protected]

1073

from PDR or CRVO was retrospectively reviewed and selected for inclusion. The control group received standard treatments for NVG, including PRP (n = 12), PRP/PPV (n = 7), IVB (n = 3), PRP/IVB (n = 4), and PRP/IVB/Ahmed valve (n = 1). The ECP group was composed of 27 eyes from 27 patients treated with combination PPV/endoscopic PRP/ECP surgery in addition to previous treatments similar to the control group (Table 1). All patients were treated with ECP in the Division of Ophthalmology, BIDMC, Boston, MA, between August 2005 and March 2013 by a single surgeon (J.G.A.). A review of electronic medical records was conducted to match ECP cases with suitable controls. An electronic medical records review from January 2006 to August 2012 at the Joslin Diabetes Center, Boston, MA, and at BIDMC (before arrival of endoscopic laser) found 65 cases of NVG that received standard glaucoma treatments. Eyes with preoperative INV and at least 6 months of follow-up were included as controls. Based on case matching and inclusion criteria, 27 eyes from 26 patients were selected as controls. Patients in the control group received PRP either externally or using an endolaser probe connected to a Coherent green laser (Coherent Inc, Palo Alto, CA). In addition to a history of external or endolaser PRP, patients in the ECP group received combined PPV, as well as endoscopic PRP and ECP using the E2 combined xenon illuminator, endoscope, and laser (Endo Optiks Inc, Little Silver, NJ). During combination surgery, a retrobulbar injection or posterior sub-Tenon’s infusion of 0.75% Marcaine and 2% lidocaine was administered before surgery to minimize intraoperative and postoperative pain. Sub-Tenon’s injection of 0.9 mL of triesence (40 mg/mL) was also given to help control postoperative inflammation. A 20- or 23-gauge 3-port PPV was then performed. In 3 phakic eyes, phacoemulsification with intraocular lens implantation or pars plana lensectomy was performed immediately before the vitrectomy. The ECP probe was inserted through each superior scleral incision for the treatment of both the peripheral retina and ciliary processes. In all eyes, extensive near-confluent 360° PRP treatment from the equator to the ora serrata was performed using the E2 unit. The entire surface of each ciliary body was coagulated from anterior to posterior until completely white, taking care not to cause vaporization and bubble formation on the surface of the ciliary processes, which would occur with overexposure to the laser. All ECP eyes received a full 360-degrees of ECP. Figure 1 shows typical endoscopic PRP and ECP treatment. Patients were left with 0.1 mL (40 mg) of triesence in the vitreous cavity to help suppress postoperative inflammation.

1074 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES



2015  VOLUME 35  NUMBER 6

Table 1. Patient Preoperative Demographics

Number of cases Age, years Mean ± SD Range Gender (M:F) Ocular ischemic causes Previous treatments None PRP PRP, PPV IVB PRP, IVB PRP, PPV, IVB PPV, PRP, filtration PRP, IVB, Ahmed valve Preoperative BCVA (logMAR value) ± SD Preoperative IOP ± SD, mmHg

ECP Group

Control Group

P

27 eyes, 27 cases

27 eyes, 26 cases

N/A

65.9 ± 13.4 41–91 21:6 PDR, 17; CRVO, 10

64.1 ± 14.4 36–86 17:10 PDR, 24; CRVO, 3

0.716

1 5 3 3 8 5 2 0 1.83 ± 0.45 40.7 ± 12.4

0 12 7 3 4 0 0 1 1.37 ± 0.78 34.7 ± 12.4

0.233 0.024 0.054

0.054 0.134

CRVO, central retinal vein occlusion; F, female; IVB, intravitreal bevacizumab injection; M, male; PDR, proliferative diabetic retinopathy; PPV, pars plana vitrectomy; PRP, panretinal photocoagulation.

Postoperatively, patients were placed on a regimen of topical antibiotics, steroids (prednisolone acetate 1%), nonsteroidal antiinflammatory drugs (NSAIDs, such as ketorolac tromethamine), cycloplegics (atropine), and their preoperative glaucoma medications except for miotics. Antibiotics were discontinued after 1 week, and the steroids, NSAIDs, and cycloplegics were tapered as inflammation subsided over a few weeks. Postoperative visits were scheduled for 1 day, 1 week, 1 month, 3 months, 6 months, and 1 year after surgery. At each postoperative visit, routine ophthalmic examinations were performed including tonometry with a Goldmann applanation tonometer (Reichert, Buffalo,

Fig. 1. A. Extensive PRP performed thoroughly up to the ora serrata, with E2 sparing the long posterior ciliary vessels. B. The entire surface of each ciliary body whitened by laser ablation as observed endoscopically.

NY), Snellen acuity measurements, biomicroscopy, and direct and indirect ophthalmoscopy. Snellen bestcorrected visual acuity (BCVA) was converted into its logarithm of the minimal angle of resolution (logMAR) format for analysis (counting fingers vision was coded as 1.6, hand motions as 2.0, light perception as 2.5, and no light perception as 3.0). Antiglaucoma medications were discontinued according to the IOP changes. The following information was also collected: age, gender, preexisting ischemic retinal disorders (PDR and CRVO), previous treatments (including PRP, PPV, IVB, trabeculectomy, and Ahmed valve placement), follow-up period, postoperative regression of INV, postoperative

ENDOCYCLOPHOTOCOAGULATION FOR NVG  MARRA ET AL

adverse events, and any subsequent surgeries. To standardize the time of follow-up measurements for each group, postoperative BCVA and IOP measurements used for comparison were from the 1 month, 3 months, 6 months, 1 year, and most recent follow-up. Regression of INV and the number of topical antiglaucoma medications were recorded from the patients’ most recent visit to handle the large number of missing data for these variables at 1 year postoperatively. Statistical analysis was performed with Version 20.0 of SPSS software for Windows (IBM Corp, Chicago, IL). Nonparametric Wilcoxon signed-rank tests were used to compare the change from preoperative to postoperative measurements of BCVA, IOP, and the number of antiglaucoma medications within the same group. Mann–Whitney U tests were used to test for statistical significance between the groups’ preoperative IOP and preoperative BCVA. Mann–Whitney U tests were also run to compare age and the preoperative to postoperative changes in IOP and BCVA between the two groups. Chi-square test was used to compare gender, ocular ischemic causes, previous procedures, presence or recurrence of INV, complication rates, and rate of visual acuity deterioration. Probability values of ,0.05 were considered significant. Subgroup analysis was conducted to assess the effects of each treatment on cases where PDR, as compared with CRVO, was the ischemic cause of their NVG. Nonparametric Wilcoxon signed-rank tests were used to compare outcomes in IOP and BCVA within each subgroup. Mann–Whitney U tests were run to determine any significant difference between the PDR and CRVO in regards to the preoperative measurements, postoperative measurements, and recorded change in IOP, BCVA, and the number of glaucoma medications.

Results Of the 54 patients, 38 were male and 16 were female. The mean age was 65.0 ± 13.8 years (range, 36–91 years). Twenty-seven eyes from 27 patients were included in the ECP group, and 27 eyes from 26 patients were included in the control group. The comparisons between preoperative data from each group are shown in Table 1. There was no significant difference in age, gender, and preoperative BCVA and IOP between the case-matched ECP and control group. Within the ECP group, there were no significant differences when preoperative and postoperative BCVA or IOP measurements in patients who underwent ECP surgery combined with cataract extraction were compared with those of patients who underwent ECP surgery alone. Analysis

1075

of ocular ischemic causes showed a significantly greater number of CRVO in the ECP group than in controls, where PDR was much more prevalent (P = 0.024). Within-group analysis showed significant reduction of IOP in both groups. Compared with the mean preoperative IOP (40.7 mmHg), mean IOP in the ECP group was significantly reduced at postoperative Months 1, 3, and 9, at 1 year, and at the most recent visit (1 month: 13.7 mmHg, P , 0.001; 3 months: 14.1 mmHg, P , 0.001; 6 months: 12.5 mmHg, P , 0.001; 1 year: 12.3 mmHg, P , 0.001; and final follow-up: 12.6 mmHg, P , 0.001). Similarly, compared with mean preoperative IOP (34.7 mmHg), the mean IOP in the control group was also significantly reduced during each of these postoperative visits (1 month: 21.9 mmHg, P , 0.001; 3 months: 19.4 mmHg, P , 0.001; 6 months: 23.3 mmHg, P = 0.001; 1 year: 23.2 mmHg, P = 0.002; and final follow-up: 20.4 mmHg, P , 0.001). Between-group analysis showed that the drop in mean IOP between preoperative and each postoperative visit in the ECP group, as compared with the control group, was significantly greater at all postoperative visits (Figure 2, for data and P values). At 1 year postoperatively, the mean IOP of the ECP group was significantly lower than that of the controls (12.3 vs. 23.2 mmHg, P , 0.001). The drop between the preoperative IOP and the IOP at 1 year postoperatively was significantly greater in the ECP group than in controls (−28.5 vs. −11.4 mmHg, P , 0.001). Also, at 1 year postoperatively, ocular hypertension (defined as an IOP .21 mmHg) was observed in no eyes of the ECP group (0%) but in 9 eyes (33.3%) of controls, a significant difference (P = 0.001). The mean preoperative BCVA in the ECP group was worse compared with the control group, but this difference was not statistically significant (1.83 vs. 1.37, P = 0.055). Within-group analysis of postoperative BCVA showed no significant change from the mean best-corrected preoperative acuity for both groups. Compared with the mean preoperative BCVA in the ECP group (1.83), the mean BCVA was not significantly different at any follow-up visit (1 month: 1.80, P = 0.983; 3 months: 1.86, P = 0.709; 6 months: 1.72, P = 0.682; 1 year: 1.84, P = 0.899; and final follow-up: 1.99, P = 0.262). Similarly, compared with the mean preoperative BCVA of the control group (1.37), there was no significant change in the mean BCVA at any follow-up visit (1 month: 1.18, P = 0.235; 3 months: 1.16, P = 0.190; 6 months: 1.34, P = 0.825; 1 year: 1.38, P = 0.887; and final followup: 1.46, P = 0.331). At 1 year postoperatively, there was no statistically significant difference between the mean BCVA of the

1076 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES



2015  VOLUME 35  NUMBER 6

Fig. 2. Scattergram of preoperative to postoperative IOP for the ECP group (marked as “x”) and control group (marked as “o”) before and after combined PPV/PRP or PPV/ECP surgery. At all postoperative intervals, both the ECP group and the control group exhibited significant reduction of mean IOP relative to each group’s respective mean preoperative IOP. Between-group comparison of the drop in IOP is shown below the x-axis.

ECP group, which had an observation recorded for 19 eyes (70.4%), and the control group, which had an observation recorded for 22 eyes (81.5%) (1.84 vs. 1.38, P = 0.079). On the Snellen chart, these BCVA measurements correlate to slightly better than hand motions in the ECP group and slightly better than counting fingers in the control group, a difference of approximately 1 increment on this scale. When marked visual change was defined as a change of three or more lines on the Snellen chart, between-group comparison showed no difference in the number of cases that experienced positive, negative, or no marked visual change at 1 year postoperatively. These data and P values are shown in Table 2. Figure 3 shows a plot of mean BCVA before procedures, and at 1 month, 3 months, 6 months, and 1 year after procedures for both groups. The graph also shows acuity at the last recorded follow-up and also compares the BCVA between groups for each

follow-up visit. This graph was repeated for IOP in Figure 2, which shows a significantly greater IOP drop in the ECP group as compared with that of the control group. The mean, median, and range in the duration of follow-up were 21.8, 17, and 6 to 62 months, respectively, in the ECP group compared with 36.4, 34.5, and 6 to 116 months, respectively, in the control group. The mean number of antiglaucoma medications in the ECP group was significantly decreased from 3.12 to 0.54 (P , 0.001) but increased in the control group (1.30–1.89; P = 0.136). In addition, the drop in number of glaucoma medication required was significantly greater in the ECP group than in controls (−2.58 vs. +0.59 drops, P , 0.001). In the ECP group, 20 of 27 eyes (74.0%) no longer required glaucoma medications to maintain IOP within normal limits, whereas only 5 of 27 eyes (18.5%) in the control group no longer required such glaucoma medications

Table 2. Visual Changes Visual Change

ECP Group (n = 27), Number of Eyes (%)

Control Group (n = 27), Number of Eyes (%)

P

Reduced postoperative BCVA (.2 lines of vision loss) Negligible visual changes Visual improvement (.2 lines of vision gain)

3 eyes (11.1) 22 eyes (81.5) 2 eyes (7.4)

3 eyes (11.1) 20 eyes (74.1) 4 eyes (14.8)

1.000 0.331 0.386

1077

ENDOCYCLOPHOTOCOAGULATION FOR NVG  MARRA ET AL

Fig. 3. Scattergram of preoperative to postoperative BCVA for the ECP group (marked as “x”) and control group (marked as “o”). The mean, median, and range of postoperative duration are 23.7, 16, 3 to 62 months (ECP group) and 36.4, 25, 3 to 116 months (control group). Counting fingers vision was coded as 1.6, hand motions as 2.0, light perception as 2.5, and no light perception as 3.0.

(P , 0.001). A summary of each treatment’s effects on the number of prescribed glaucoma medications is provided in Table 3. As an inclusion requirement for this study, INV was detected in all cases preoperatively. Postoperative regression of INV measured at the most recent follow-up was seen in significantly more eyes (P = 0.004) in the ECP group (22 eyes, 91.7%) than in the control group (12 eyes, 48.0%). There were 6 eyes in the ECP group (22.2%) and 5 eyes in the control group (18.5%) that experienced postoperative complications of varying severity. The complication rates were not statistically significant between groups (P = 0.685). Table 4 provides a summary of postoperative complications.

In the ECP group, hyphema occurred in 3 eyes (11.1%) and retinal detachment in 1 eye (3.7%). One of the 3 eyes with hyphema in the ECP group experienced a stained corneal endothelium that led to diffuse corneal opacity. This eye did not receive subsequent surgical intervention because of poor preoperative vision and successful pain relief from reduction in IOP after surgery. The second case of hyphema in the ECP group occurred 7 months after PPV/PRP/ECP and was successfully treated with anterior chamber washout and PPV. The third ECP case with hyphema underwent PPV soon after ECP surgery, but the patient died before substantial postoperative data could be collected. One eye in the ECP group also suffered retinal detachment with inferior peripheral break at postoperative Month 2, and an additional PPV with a silicone oil tamponade was performed to successfully

Table 3. Drop in the Number of Glaucoma Medications ECP Group (n = 27)

Controls (n = 27)

Change in number of −2.58 (0.54 from 3.12) glaucoma medications (postoperative mean from preoperative mean) Within-group P , 0.001 significance Between-group P , 0.001 significance

0.59 (1.89 from 1.30)

P = 0.136

Table 4. Postoperative Complications

Complications Hyphema Retinal detachment Vitreous hemorrhage Phthisis bulbi

Control Group ECP Group (n = 27), Number (n = 27), Number of Eyes (%) of Eyes (%) 3 eyes (11.1) 1 eye (3.7) 0 eyes (0)

0 eyes (0) 0 eyes (0) 3 eyes (11.1)

2 eyes (7.4)

2 eyes (7.4)*

*One case of phthisis bulbi progressed from a case of vitreous hemorrhage.

1078 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES

reattach the retina. Two cases of prolonged hypotony, or phthisis bulbi, were reported in the ECP group (7.4%). Patients were diagnosed with phthisis bulbi if their IOP remained under 5 mmHg for 3 consecutive visits anytime after 6 months after initial treatment. No instances of endophthalmitis or sympathetic ophthalmia were observed in either group. Vitreous hemorrhage occurred in 3 eyes (11.1%) of the control group. These 3 eyes were initially observed. Two of these eyes with persistent vitreous hemorrhage were subsequently treated with PPV, and phthisis bulbi developed in 1 of them. There were a total of 2 eyes that developed phthisis bulbi in the control group (7.4%). The ECP group and control group were each divided into 2 subgroups: 1 where PDR was the ischemic cause of NVG, and the other where CRVO was the underlying cause. Subgroup analysis was conducted both within and between the ECP group and the control group. Within the ECP group, there was no significant difference in mean preoperative IOP (38.5 vs. 44.5 mmHg, P = 0.170) and BCVA (1.70 vs. 2.03, P = 0.103) between subgroups with PDR (n = 17) versus those with CRVO (n = 10) as the underlying ischemic cause of NVG. At 1 year after ECP surgery, the PDR subgroup showed significantly lower mean IOP (12.7 from 38.5 mmHg preoperatively, P = 0.003) and slightly better mean BCVA (1.57 from 1.70 preoperatively, P = 0.384). The CRVO subgroup similarly showed a significantly reduced mean IOP (11.6 from 44.5 mmHg preoperatively, P = 0.012) at 1 year postoperatively. There was no significant change in BCVA (2.20 vs. 2.03 preoperatively, P = 0.066) in the CRVO subgroup at 1 year postoperatively. At all follow-up intervals, there was no statistically significant difference between the PDR and the CRVO subgroups of the ECP group regarding mean postoperative IOP and BCVA. There were also no significant differences shown at any follow-up visit when the mean IOP drop and BCVA change were compared between PDR and CRVO subgroups. In addition, there was no significant difference between the preoperative and postoperative decrease in glaucoma medications in the PDR subgroup as compared with the CRVO subgroup (−2.88 vs. −2.10, P = 0.351). Within the control group, there was no significant difference in mean preoperative IOP (33.5 vs. 48.7 mmHg, P = 0.119) and BCVA (1.73 vs. 1.32, P = 0.483) between the PDR (n = 24) and CRVO (n = 3) subgroups. The PDR subgroup of the controls showed significant reductions in mean IOP (20.6 from 33.5 mmHg preoperatively, P = 0.002) and no significant visual changes (1.22 from 1.32 preoperatively, P = 0.393) at 1 year postoperatively. The CRVO subgroup exhibited no



2015  VOLUME 35  NUMBER 6

significant changes in IOP (50.0 from 48.7 mmHg preoperatively, P = 0.655) and BCVA (3.00 vs. 1.73 preoperatively, P = 0.180) at 1 year postoperatively. Comparison of mean postoperative IOP and BCVA between subgroups of controls trends toward better outcomes in the PDR subgroup, and this trend becomes significant at 1 year for both IOP (20.6 vs. 50.0 mmHg, P = 0.017) and BCVA (1.22 vs. 3.00, P = 0.002). When the mean change in IOP and BCVA from preoperative to postoperative visits was compared between the PDR and CRVO subgroups of controls, no significance was shown at any follow-up visit. Finally, there was a nonsignificant increase in the number of glaucoma medications in the PDR and CRVO subgroups of the controls (+0.45 vs. +1.67, P = 0.204). Between-group analysis then compared the PDR subgroup in the ECP group (n = 17) with that of the controls (n = 24), and this analysis was repeated for the CRVO subgroups. In the comparison of PDR subgroups, the ECP group exhibited a significantly lower IOP (12.7 vs. 20.2 mmHg, P = 0.005) but a nonsignificantly different mean BCVA (1.57 vs. 1.22, P = 0.227) at 1 year as compared with the control group. For the PDR subgroups, ECP caused a significantly larger drop in mean IOP (−26.8 vs. −12.7, P = 0.007) relative to controls but showed a similar improvement of vision (−0.16 vs. −0.11, P = 0.611) at 1 year postoperatively. Furthermore, when the change in the number of prescribed glaucoma medications was compared between the PDR subgroups of ECP versus controls, the analysis showed that ECP significantly reduced the number of medications (−2.88 vs. +0.45, P , 0.001). Subgroup analysis of the treatments’ effects on glaucoma medications has been summarized in Figure 4. However, for the CRVO subgroup, ECP treatment significantly reduced mean IOP at 1 year (11.6 from 44.5 mmHg preoperatively, P = 0.012) while the standard care in the controls did not control IOP (50.00 from 48.7 mmHg, P = 0.655). In addition, the CRVO subgroup of the ECP group exhibited a trend toward a more improved visual outcome than the CRVO subgroup of controls (2.20 vs. 3.00, P = 0.178), but this was not statistically significant. At 1 year postoperatively, ECP caused a significant drop in mean IOP while standard care treatments were ineffective in controlling IOP (−22.9 vs. 1.5 mmHg, P = 0.030). In their CRVO subgroups, both the ECP group and the control group showed no significant changes in BCVA at 1 year postoperatively, but the ECP group showed a nonsignificant trend toward less visual decline at 1 year postoperatively (+0.29 vs. +1.30, P = 0.089). Finally, the change in the number of glaucoma medications was

ENDOCYCLOPHOTOCOAGULATION FOR NVG  MARRA ET AL

1079

Fig. 4. Subgroup analysis of the treatments’ effects on the number of prescribed glaucoma medications. Because of a large amount of missing data regarding glaucoma medications at 1 year postoperatively, data were recorded at the final postoperative visit.

nonsignificantly more favorable in the PDR subgroup of the ECP group as compared with the PDR subgroup of controls (−2.10 vs. +1.67, P = 0.077)All data regarding glaucoma medication are provided in Figure 4. Discussion Severe retinal ischemia due to PDR or CRVO often leads to rubeosis, angle neovascularization, and formation of fibrovascular tissue on the trabecular meshwork, obstructing aqueous outflow from the eye and rapidly increasing IOP.16,17 Some cases continue to experience increases in IOP and severe pain despite PRP and medical glaucoma treatment. Glaucoma surgery is typically used to reduce IOP to alleviate the patient’s ocular pain; however, reported success rates in traditional glaucoma surgery have been mixed.2–8 Ciliary body destructive procedures such as ECP have shown marked effects in reducing IOP.9,10,14,15 Our study found that ECP showed better IOP control when compared with standard treatments for severe cases of NVG. Endoscopic cyclophotocoagulation treatment maintained an IOP within normal limits (under 21 mmHg) in significantly more cases than control treatments (27 vs. 15 eyes, P = 0.001), whereas the number of unresolved complications of phthisis bulbi was the same in both groups (2 eyes). In general, external laser cyclodestructive procedures are used predominantly for pain relief in eyes with little or no visual potential because these blind treatments have some risk for overtreatment and undertreatment.9,10,14,15 Endoscopic cyclophotocoagulation,

however, is a precisely controlled destructive procedure that avoids potential overtreatment and collateral tissue damage by making it possible to visualize the treatment targeted tissue directly and to deliver laser power specifically to the epithelium of the ciliary body. Another approach to treating ciliary processes is through transvitreal ECP, which is performed using scleral depression and direct observation. Most of the ciliary body, however, stretches out anatomically from anterior to posterior. Because of this anatomical arrangement, each ciliary process can be coagulated more thoroughly under direct visualization without scleral depression using an intraocular endoscope rather than transvitreal cyclophotocoagulation using a standard endolaser probe and scleral depression. Previous studies have reported the effectiveness of transvitreal cyclophotocoagulation. For instance, in a retrospective case review of 73 eyes, Haller et al reported that 48 of 55 eyes (87.3%) with NVG treated with transvitreal cyclophotocoagulation were successfully controlled the IOP at postoperative 12 months. Comparatively, in our case series, ECP successfully controlled the IOP in 25 of 27 eyes (92.6%) with NVG. Of note, both of these studies had similar preoperative mean IOP (the study by Haller et al with 38 mmHg and our study with 40.7 mmHg), but ECP achieved a greater drop in IOP (12.7 mmHg) as compared with that achieved by the treatment technique of transvitreal cyclophotocoagulation (16 mmHg) at 1 year postoperatively. Furthermore, our study of ECP reported 2 eyes (7.4%) to have sustained IOP ,5 mmHg after 6 months, whereas Haller et al reported such sustained low pressure in 6 eyes (8.4%) after

1080 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES

transvitreal cyclophotocoagulation.18 Visual acuity improved or remained stable in 53 of 73 eyes (73%) in the study by Haller et al and in 25 of 27 eyes (92.6%) in our case series of ECP. At the last follow-up visit, 36 of 55 eyes (65.5%) no longer required glaucoma medications after transvitreal cyclophotocoagulation in the study by Haller et al. In our study, 20 of 27 eyes (74.0%) in the ECP group no longer required glaucoma medications to maintain IOP within normal limits. This comparison suggests that ECP may trend toward improved IOP and fewer required glaucoma medications than transvitreal cyclophotocoagulation, but a more appropriately designed direct comparison between these modalities would be necessary to make any such conclusions. A few studies have reported favorable results of ECP treatment for NVG.11–13 In agreement with our study, these studies (Uram, Lima et al, and Murthy et al) all report a significant decrease in IOP, the number of glaucoma medications, and the rates of complications. In all 3 studies combined, there was only 1 case of phthisis bulbi, which was reported by Uram11 in the case series of 10 patients who underwent ECP. Overall, the rates of complications were significantly lower in these studies’ ECP group and significantly so when compared with Ahmed valve placement, according to Lima et al.12 In addition to exhibiting higher rates of complications, Ahmed valve placement was reported to significantly decrease BCVA, whereas no significant decrease was reported in cases that received ECP for NVG. Murthy et al13 corroborate this finding by reporting a significant increase in BCVA in the postoperative period. In our study, the effect of ECP as measured by IOP and number of glaucoma medications required were similarly favorable. Compared with our control group, the ECP group began with slightly worse preoperative IOP, experienced significantly higher rate of maintained normal pressures at 1 year, and saw a significantly greater drop in IOP at all postoperative visits (1 month, P = 0.004; 3 months, P = 0.009; 6 months, P = 0.001; 1 year, P , 0.001; final, P = 0.001). One year postoperatively, the ECP group exhibited a mean IOP that was significantly lower than that of the controls (12.3 vs. 23.2 mmHg, P = 0.003). When the mean drop in IOP in the ECP group was compared with that of the control group at time-matched intervals, ECP showed a significantly greater and more rapid impact on IOP than routine glaucoma care at all postoperative visits (Figure 2). In addition, the mean drop between preoperative and postoperative IOP at 1 year was significantly greater in the ECP



2015  VOLUME 35  NUMBER 6

group than in controls (−28.5 vs. −11.4 mmHg, P , 0.001). The significantly higher occurrence of ocular hypertension at 1 year in the control group (9 eyes, 33.3%) further suggests that ECP (0 eyes) may be more effective in controlling IOP (P = 0.001). The ECP group also demonstrated a significant reduction in the number of antiglaucoma medications from 3.12 preoperatively to 0.54 postoperatively (P , 0.001), whereas the control group exhibited a nonsignificant increase in number of medications (1.30–1.89, P = 0.136). Thus, ciliary body ablation with ECP technique resulted in marked reduction and stabilization of IOP, as well as a decreased number of required glaucoma medications, even in more severe cases of NVG. The overall visual outcomes in our cases are worse than in previous reports,11–13,19 but the preoperative BCVA in our ECP group were also poorer than these studies, ranging from light perception to hand motions. In our study, there were no significant differences in BCVA between treatment groups at any postoperative follow-up (Figure 3). In addition, the incidence of significant visual change, defined as the number of cases with a change of 3 or more lines on the Snellen chart, was compared between groups. When measured in this way, once again, no significant difference in visual outcome was found between groups (Table 2). Visual outcomes in both groups were similar and poor; it is important to note that the primary goal of treating NVG is to control IOP, which was more effectively accomplished by ECP in this study. Preoperative information such as IOP, BCVA, age, gender, and previous treatments were similar between the 2 groups (Table 1). This finding was to be expected given the case-matched nature of our study. The retrospective nature of this study did present some limitations in data collection. Because of the different time frames in follow-up periods between the 2 groups, changes in IOP and BCVA (Figures 2 and 3, respectively) were plotted and compared at standardized postoperative times—1 month, 3 months, 6 months, and 1 year. There were a limiting number of data points at 1 year for the number of glaucoma medications and INV regression postoperatively. To increase the number of data points to allow for statistical analysis, these measurements were taken at the last recorded measurement on record. Comparison of the number of glaucoma medications and regression of INV at a standardized time after treatment would have provided a more logical comparison, if that data had been available. As controls were collected from a well-known diabetes center (Joslin Diabetes Center [JDC]), the

ENDOCYCLOPHOTOCOAGULATION FOR NVG  MARRA ET AL

control group had significantly more cases of PDR than the ECP group (P = 0.024). By comparing IOP and BCVA outcomes of patients with PDR versus patients with CRVO, subgroup analysis explored the effect of each treatment on cases with differing causes of the retinal ischemia underlying NVG. This analysis showed that ECP and standard treatment of NVG, where PDR is the ischemic cause, both result in significantly reduced mean IOP 1 year after treatment (ECP group: 12.7 from 38.5 mmHg, P = 0.003; control group: 20.6 from 33.5 mmHg, P = 0.002). In the PDR subgroup of both the ECP group and controls, there was a slightly improved BCVA at 1 year postoperatively (ECP group: 1.57 from 1.70, P = 0.384; controls: 1.22 from 1.32, P = 0.393). For cases where PDR was the ischemic cause of NVG, however, ECP caused a significantly greater drop in IOP (−26.8 vs. −12.7, P = 0.007) when compared with controls at 1 year postoperatively. In addition, ECP showed a significantly more favorable decrease in the number of glaucoma medications required (−2.88 vs. +0.45, P , 0.001) for the cases where PDR was the underlying ischemic cause of NVG. Loss of vision is always an unfavorable result; however, the primary goal of treating such severe NVG is the reduction of IOP, which was exhibited in both groups but significantly greater in the ECP group. Vision in both groups was nonetheless poor, and the PDR subgroup of the ECP group had nonsignificantly worse preoperative BCVA than that of controls, which may be contributory to worsened postoperative outcomes. Thus, the data suggest that both ECP surgery and standard glaucoma care are fairly effective in achieving the primary goal of treatment (lowering IOP) for cases of NVG where PDR is the ischemic cause. However, the results from this study suggest that in cases of NVG where PDR is the ischemic cause, ECP more effectively controls IOP at 1 year postoperatively. In addition, ECP significantly reduced the number of glaucoma medications required (from 3.38 to 0.50 preoperatively, P = 0.001) while resulting in a favorable IOP outcome. The PDR subgroup of the controls, however, exhibited an increase in the number of required glaucoma medications (1.75 from 1.29, P = 0.281). Subgroup analysis of cases where CRVO was the ischemic cause of NVG showed similarly improved outcomes after ECP as compared with standard glaucoma treatments. At 1 year postoperatively, mean IOP was significantly reduced in the CRVO subgroup of the ECP group (11.6 from 44.5 mmHg, P = 0.012), whereas the mean IOP in the CRVO subgroup of the controls increased (50.0 from 48.7 mmHg, P = 0.655). Compared with the controls at 1 year, the ECP group showed a significantly greater drop in mean IOP

1081

(−22.09 vs. 1.5 mmHg, P = 0.030). In addition, visual acuity slightly worsened in both the ECP group (2.20 from 2.03, P = 0.066) and the control group (3.00 from 1.73, P = 0.180) at 1 year. Although not statistically significant, the worsening of visual acuity was not as severe in the CRVO subgroup of the ECP group than in the controls (0.29 vs. 1.30, P = 0.089). Finally, although ECP significantly reduced the number of glaucoma medications required for the CRVO subgroup (0.60 from 2.7, P = 0.014), the control treatment resulted in a nonsignificant increase in the prescribed number of medications (3.00 from 1.33, P = 0.180). As the primary goal of NVG treatment is the reduction of IOP, these data may suggest that ECP shows favorable IOP outcomes compared with standard glaucoma care for patients where CRVO is the underlying ischemic cause of NVG. This study’s small sample size, however, limits the power of this conclusion. Larger studies would be necessary to further explore whether ECP may be especially effective when CRVO is the cause of NVG. The destructive nature of ECP has raised concern over the incidence of complications such as hypotony or phthisis bulbi.9,14,15 Although some studies report only favorable outcomes after ECP treatment of NVG,11–13 other smaller reports warn that ECP may produce unpredictable results.20 Our study suggests that the prevalence of phthisis bulbi is similar in eyes treated with standard care plus ECP versus eyes treated with standard care alone. Overall, complication rates were similar between the groups (6 vs. 5 eyes, P = 0.685). In our study, two cases of hyphema that required surgical evacuation and one retinal detachment that required surgical repair occurred in the ECP group postoperatively. In the control group, three nonclearing vitreous hemorrhages occurred with two requiring revision vitrectomy. These complications were not unexpected given the severity of the underlying disease process. With the exception of one patient who died before substantial postoperative data could be acquired, all of the complications in the ECP group were resolved on further intervention. Endoscopic PRP allowed us to easily and completely treat 360° of the retinal periphery up to the ora serrata. This endoscopic treatment resulted in improved rates of INV regression. Where 12 of 27 eyes in the control group (48.0%) experienced postoperative disappearance of INV, the disappearance of INV occurred in 22 of the 27 eyes (91.7%) in the ECP group (P = 0.004). This trend suggests that endoscopic near-confluent PRP from the equator to the ora serrata was more effective in decreasing the amount of retinal ischemia present as compared with standard retina and glaucoma care.

1082 RETINA, THE JOURNAL OF RETINAL AND VITREOUS DISEASES

Vascular endothelial growth factor plays a major role in mediating active intraocular neovascularization in patients with retinal ischemia.21 Some recent studies reported that intravitreal injections of bevacizumab, a specific inhibitor of vascular endothelial growth factor, showed short-term regression of neovascularization and IOP reduction in NVG cases.22–25 In our study, three eyes in the ECP group had undergone a previous intravitreal IVB. An intravitreal IVB was administered in 2 of 3 eyes 4 to 5 days before PPV/ PRP/ECP in an attempt to decrease IOP and intraoperative bleeding. In the third eye, IVB was performed 4 months before PPV/PRP/ECP; however, NVG recurred 2 months later. The effect of bevacizumab disappears even earlier in cases that underwent filtration surgery presumably because of increased aqueous outflow.26 As ECP decreases aqueous inflow, it might sustain the effective duration of IVB. Further assessment is needed of combined ECP with a postoperative bevacizumab injection in preventing recurrence of NVG. In this study, we investigate the safety and efficacy of ECP in NVG cases with severe elevation of IOP and poor vision, and our findings indicate that ECP may be useful in NVG.5,6 To obtain a representative control group, we matched patients receiving ECP at BIDMC to patients from the well-known diabetes center (JDC), which did not use ECP. The limited number of cases in each group and the retrospective nature of this study limit the interpretation of these data. Given the different institutions and approaches to treatment, selection bias may have played a role. Indeed, there were many more PDR patients in the control group because of the JDC’s intentional focus on patients with diabetes. Furthermore, BIDMC institution obtained the ECP equipment in 2006, and most of the standard PPV/PRP treatments were performed before this date whereas all of the PPV/ endoscopic PRP/ECP was performed after this date, resulting in different amounts of follow-up for each group. To account for this discrepancy, visual outcomes were compared at corresponding postprocedural times (1, 3, 6, and 12 months and most recent visit). In summary, our findings indicate that ECP combined with PPV and endoscopic PRP seems to be more effective than standard care alone for treating acute NVG. Especially in cases where PDR is the ischemic cause of NVG, this study shows that ECP more effectively decreased and stabilized IOP and reduced the number of required glaucoma medications when compared with controls. There were no significant differences in long-term visual outcomes, but the control group showed a trend toward improved, but nonetheless poor, BCVA. In the control group, there were two late complication of phthisis bulbi, one



2015  VOLUME 35  NUMBER 6

of which developed from the three vitreous hemorrhages that occurred. Similarly, two cases of phthisis bulbi progressed in the ECP group. Three cases of hyphema and one retinal detachment occurred and were successfully treated in the ECP group, with no presentations of such complications in the control group. Further studies are necessary to determine the long-term safety and efficacy of ECP, late-onset adverse events, and potential combination with other treatments such as IVB or intravitreal or sub-Tenon’s injection of triamcinolone acetonide. Key words: case-matched, ciliary body, endocyclophotocoagulation, endoscope, neovascular glaucoma, pan retinal photocoagulation, surgical intervention. Acknowledgments The authors thank Lloyd Paul Aiello for his editorial assistance. References 1. Sivak-Callcott JA, O’Day DM, Gass JD, Tsai JC. Evidencebased recommendations for the diagnosis and treatment of neovascular glaucoma. Ophthalmology 2001;108:1767–1776. 2. Tsai JC, Feuer WJ, Parrish RK II, Grajewski AL. 5-Fluorouracil filtering surgery and neovascular glaucoma. Long-term follow-up of the original pilot study. Ophthalmology 1995;102:887–892; discussion 892–893. 3. Allen RC, Bellows AR, Hutchinson BT, Murphy SD. Filtration surgery in the treatment of neovascular glaucoma. Ophthalmology 1982;89:1181–1187. 4. Mandal AK, Majji AB, Mandal SP, et al. Mitomycin-Caugmented trabeculectomy for neovascular glaucoma. A preliminary report. Indian J Ophthalmol 2002;50: 287–293. 5. Kiuchi Y, Nakae K, Saito Y, et al. Pars plana vitrectomy and panretinal photocoagulation combined with trabeculectomy for successful treatment of neovascular glaucoma. Graefes Arch Clin Exp Ophthalmol 2006;244:1627–1632. 6. Takihara Y, Inatani M, Fukushima M, et al. Trabeculectomy with mitomycin C for neovascular glaucoma: prognostic factors for surgical failure. Am J Ophthalmol 2009;147: 912–918. 7. Goulet RJ III, Phan AD, Cantor LB, WuDunn D. Efficacy of the Ahmed S2 glaucoma valve compared with the Baerveldt 250-mm2 glaucoma implant. Ophthalmology 2008;115: 1141–1147. 8. Every SG, Molteno AC, Bevin TH, Herbison P. Long-term results of Molteno implant insertion in cases of neovascular glaucoma. Arch Ophthalmol 2006;124:355–360. 9. Pastor SA. Cyclophotocoagulation: a report by the American Academy of Ophthalmology. Ophthalmology 2001;108:2130–2138. 10. Goldenberg-Cohen N, Bahar I, Ostashinski M, et al. Cyclocryotherapy versus transscleral diode laser cyclophotocoagulation for uncontrolled intraocular pressure. Ophthalmic Surg Lasers Imaging 2005;36:272–279. 11. Uram M. Ophthalmic laser microendoscopic ciliary process ablationin the management of neovascular glaucoma. Ophthalmology 1992;99:1823–1828.

ENDOCYCLOPHOTOCOAGULATION FOR NVG  MARRA ET AL 12. Lima FE, Magacho L, Carvalho DM, et al. A prospective, comparative study between endoscopic cyclophotocoagulation and the Ahmed drainage implant in refractory glaucoma. J Glaucoma 2004;13:233–237. 13. Murthy GJ, Murthy PR, Murthy KR, et al. A study of the efficacy of endoscopic cyclophotocoagulation for the treatment of refractory glaucomas. Indian J Ophthalmol 2009;57: 127–132. 14. Bloom PA, Tsai JC, Sharma K, et al. “Cyclodiode.” Transscleral diode laser cyclophotocoagulation in the treatment of advanced refractory glaucoma. Ophthalmology 1997;104: 1508–1519. 15. Kosoko O, Gaasterland DE, Pollack IP, Enger CL. Long-term outcome of initial ciliary ablation with contact diode laser transscleral cyclophotocoagulation for severe glaucoma. The Diode Laser Ciliary Ablation Study Group. Ophthalmology 1996;103:1294–1302. 16. John T, Sassani JW, Eagle RC Jr. The myofibroblastic component of rubeosis iridis. Ophthalmology 1983;90:721–728. 17. Gartner S, Taffet S, Friedman AH. The association of rubeosis iridis with endothelialisation of the anterior chamber: report of a clinical case with histopathological review of 16 additional cases. Br J Ophthalmol 1977;61:267–271. 18. Haller JA. Transvitreal endocyclophotocoagulation. Trans Am Ophthalmol Soc 1996;94:589–676.

1083

19. Chen J, Cohn RA, Lin SC, et al. Endoscopic photocoagulation of the ciliary body for treatment of refractory glaucomas. Am J Ophthalmol 1997;124:787–796. 20. Gorovoy IR, Eller AW. Endocyclophotocoagulation as an adjuvant to vitreoretinal surgery in cases with concomitant glaucoma. Ophthalmic Surg Lasers Imaging Retina 2013; 144:243–247. 21. Aiello LP, Avery RL, Arrigg PG, et al. Vascular endothelial growth factor in ocular fluid of patients with diabetic retinopathy and other retinal disorders. N Engl J Med 1994;331:1480–1487. 22. Davidorf FH, Mouser JG, Derick RJ. Rapid improvement of rubeosis iridis from a single bevacizumab (Avastin) injection. Retina 2006;26:354–356. 23. Mason JO III, Albert MA Jr, Mays A, Vail R. Regression of neovascular iris vessels by intravitreal injection of bevacizumab. Retina 2006;26:839–841. 24. Illiev ME, Domig D, Wolf-Schnurrbursch U, et al. Intravitreal bevacizumab (Avastin) in the treatment of neovascular glaucoma. Am J Ophthalmol 2006;142:1054–1056. 25. Yazdani S, Hendi K, Pakravan M. Intravitreal bevacizumab (Avastin) injection for neovascular glaucoma. J Glaucoma 2007;16:437–439. 26. Moraczewski AL, Lee RK, Palmberg PF, et al. Outcomes of treatment of neovascular glaucoma with intravitreal bevacizumab. Br J Ophthalmol 2009;93:589–593.