Incisional glaucoma surgical procedures produce greater intraocular pressure (IOP) reduction and less IOP variability than medical treatment. We sought to determine the efficacy of glaucoma surgery in decreasing localized and global rates of visual field (VF) progression.
Retrospective, interventional case series.
Subjects in the New York Glaucoma Progression Study with glaucomatous optic neuropathy, repeatable VF loss, and 10 or more Swedish interactive threshold algorithm standard VF examinations were assessed for eligibility. Patients who underwent successful glaucoma surgery (not requiring further surgical intervention and IOP < 18 mm Hg) in either eye and who were followed up for at least 2 years before and after surgery were enrolled. Automated pointwise linear regression analysis was used to calculate global and localized rates of progression before and after surgery. Eyes with other ocular conditions likely to affect the VF and an insufficient number of VF to create a slope before and after surgery were excluded. Comparisons were performed within the same eyes before and after surgery (Student paired t test).
We enrolled 28 eyes of 28 patients (mean age, 61.2 ± 14.5 years). The mean number ± standard deviation of VF was 13.4 ± 2.3, spanning 7.1 ± 1.2 years (range, 4 to 9 years). Mean IOP ± standard deviation decreased from 19.0 ± 3.9 mm Hg before surgery to 11.3 ± 3.7 mm Hg after surgery (40% reduction; P < .01). Mean global progression rates decreased from −1.48 ± 1.4 dB/year before surgery to −0.43 ± 0.8 dB/year after surgery (70% reduction; P = .01). Twelve eyes (42.8%) had at least 1 significantly progressing point before surgery, whereas only 2 (7.1%) had at least 1 progressing point after surgery. Each 1 mm Hg of IOP reduction after surgery resulted in a 0.1 dB/year decrease in the global rate of progression.
Successful IOP reduction after glaucoma surgery greatly reduces both the number of progressing points and the localized and general rates of VF progression.
Intraocular pressure (IOP) is the most important known risk factor for the onset and progression of glaucoma, and its reduction remains the only evidence-based intervention shown to slow or halt disease progression. The Early Manifest Glaucoma Trial was among the first large, randomized controlled trials to show that effective IOP reduction with nonsurgical intervention led to a significant decrease in the number of eyes to reach a progression end point. The Ocular Hypertension Treatment Study demonstrated a significant reduction in the incidence of developing open-angle glaucoma in a large cohort of eyes with ocular hypertension randomized to observation or topical treatment. The European Glaucoma Prevention Study failed to detect a significant difference between medical therapy and placebo in reducing the progression to glaucoma in ocular hypertensives.
Although medical therapy delays visual field (VF) loss, greater IOP reduction often can be achieved with incisional surgery. Surgical intervention may produce greater IOP reduction and less IOP fluctuation than medical treatment, thus better preventing future VF loss than medical therapy. In the Advanced Glaucoma Intervention Study, mean IOP and range of fluctuation both were reduced after surgery. Further analysis suggested that VF progression was prevented with substantial IOP reduction. In the Collaborative Initial Glaucoma Treatment Study, surgery as initial treatment was more effective than medical therapy in reducing IOP and preventing progression.
All of the major National Eye Institute clinical trials in glaucoma use an event-based analysis of VF loss to determine if progression has occurred, rather than assessing the rate of disease progression. Trend analysis, using pointwise linear regression (PLR), is an alternative VF assessment strategy that uses the rate of loss in decibels per year (dB/year) to describe change. Furthermore, the major clinical trials determine VF loss with parameters that reflect global or hemifield progression. Analysis of localized points of active glaucoma progression could be a more accurate method of quantifying true disease progression. The purpose of this study was to evaluate the effect of glaucoma surgery on the rates (dB/year) of both global and localized VF progression.
The New York Glaucoma Progression Study consists of 43 660 consecutive subjects (132 512 VF tests) evaluated in the glaucoma referral practice of the authors (J.M.L., R.R., C.T.) from January 1999 through December 2008. After an initial visit that consisted of a complete ophthalmologic examination, perimetry (24-2 Swedish interactive threshold algorithm standard automated perimetry, Humphrey Field Analyzer II; Carl Zeiss Meditec, Inc, Dublin, California, USA) and optic disc stereophotographs, patients were reexamined, usually at 3- to 6-month intervals, and the same tests were repeated within 6 to 12 months. The study was approved by the New York Eye and Ear Infirmary Institutional Review Board and followed the tenets of the Declaration of Helsinki.
We evaluated the charts of 441 glaucoma patients with 10 or more Swedish interactive threshold algorithm standard 24–2 fields (Swedish interactive threshold algorithm standard automated perimetry, Humphrey Field Analyzer II; Carl Zeiss Meditec, Inc) in either eye. From this cohort, we enrolled those eyes that underwent successful reduction of IOP after glaucoma surgery in 1 eye, had a minimum of 5 preoperative and 5 postoperative VFs, and were followed up for a minimum of 2 years both before and after surgery. Glaucoma surgery was indicated whenever there was suspected VF progression or the IOP was considered unsatisfactory for the extent of glaucomatous damage. Successful IOP reduction was defined as all follow-up IOP measurements less than 18 mm Hg beginning 8 weeks after the procedure without any further operating room surgical interventions. Suture lysis, bleb needling revision, 5-fluorouracil subconjunctival injection, and topical ocular hypotensive medications were used at the discretion of the treating surgeon as clinically indicated. Patients with other ocular conditions known to affect the VF (including anterior segment, vitreoretinal, and optic nerve disease), and an insufficient number of VF to create a slope before or after surgery were excluded. All patients had at least 2 reliable baseline glaucomatous VF tests. The minimum criteria for VF abnormality were glaucoma hemifield test results outside normal limits on at least 2 consecutive, reliable examinations or the presence of at least 3 contiguous test points on the pattern standard deviation (SD) plot with P < .01, with at least one at P < .005, not including points on the edge of the field or those directly above and below the blind spot. Baseline VF tests had reliability indices less than 25% fixation losses, false-positive responses, or false-negative responses. The same criteria were used to determine reliability for baseline VF tests before and after surgery.
Each medical record was reviewed to collect patient demographics, type of glaucoma, and method of surgical intervention. Preoperative and postoperative IOP measurements using Goldmann applanation tonometry at every visit were recorded for calculation of mean IOP. Measurements from all visits before surgery then were averaged to provide an IOP mean and SD before surgery. IOP measurements from all visits starting 8 weeks after the date of surgery were averaged to provide the mean and SD after surgery. IOP fluctuation was defined as the IOP SD during that period. Visits within 2 months after surgery were excluded to eliminate outliers caused by the large variability in IOP that characteristically occur in the immediate postoperative period.
Automated PLR analysis was performed using Progressor software (version 3.3; Medisoft, Inc, London, United Kingdom) providing slopes (dB/year) of progression both globally and locally for each point, as well as its level of significance ( P values). A Gaussian filter was applied to reduced measurement variability without recourse to additional testing or exclusion of noisy tests. Therefore, all available VF tests (other than baseline examinations) were included in the analysis irrespective of reliability criteria. A test point was identified as progressing if the slope of sensitivity over time exceeded 1 dB loss/year (with P < .01). For edge points, a stricter slope criterion of > 2 dB loss/year (also with P < .01) was used.
To determine whether the change in slopes was related to regression to the mean compared with surgical intervention, we evaluated a separate control group of concurrent, consecutive, unoperated New York Glaucoma Progression Study patients with at least 10 VF tests. One eye was enrolled randomly. We divided the set of tests in 2 halves, also using the same criteria as previously described for PLR analysis. We further compared the global and localized progression slopes of the first half (≥ 5 tests) and second half (≥ 5 tests) of the VF series.
Statistical Analysis and Main Outcome Measures
Categorical variables were compared using the chi-square and McNemar tests. Global rates of progression were compared within the same subject (Student paired t test) before and after surgery. The same approach was used to evaluate the IOP mean and fluctuation. The Wilcoxon rank-sum test was used to compare the number of glaucoma medications before and after intervention. Assuming a 5% chance of type I error for a test power of 80% and a global slope SD of 0.8 dB/year, sample size calculation determined that to detect a minimum difference of 0.5 dB/year in rates of VF progression before and after surgery using 2-tailed probability, a minimum number of 22 eyes was required before and after surgery. To investigate the influence of IOP reduction on changes on the rate of progression, we performed linear regression analysis comparing the absolute amount of IOP reduction and the decay in the global slope. Statistical significance was defined at P < .05. Computerized statistical analysis was performed using MedCalc software (Medcalc, Inc, Mariakerke, Belgium).
Twenty-eight eyes of 28 patients (mean age 61.2 ± 14.5 years) were enrolled ( Table 1 ). Sixteen (57%) were women and 24 (86%) were of European ancestry. The most common diagnosis was primary open-angle glaucoma (46%). The mean number of analyzed VFs was 13.4 ± 2.3, spanning a mean of 7.1 ± 1.2 years (range, 4 to 9 years). The average mean deviation (MD) and pattern standard deviation (PSD) by the time of surgery were −9.6 ± 6.0 dB and 6.4 ± 3.9 dB, respectively. The mean number of VF tests was 6.6 ± 1.4 and 6.8 ± 2.1 before and after surgery, respectively. Subjects were followed up for a mean of 3.6 ± 1.2 years (range, 2 to 6 years) before surgery and 3.5 ± 1.0 years (range, 2 to 6 years) after surgery. Procedures included trabeculectomy with mitomycin C (13 eyes), combined phacoemulsification and trabeculectomy with mitomycin C (13 eyes), and glaucoma drainage device implantation (2 eyes). Twenty-two eyes were phakic and 6 eyes had posterior chamber pseudophakia. Of the eyes having trabeculectomy alone, 38% (5/13) already had undergone cataract extraction before the date of the first VF entered in the analysis.
|Gender (F), n (%)||16 (57%)|
|Age (yrs) a||61.2 ± 14.5|
|Diagnosis, n (%)|
|Primary open-angle glaucoma||13 (46%)|
|Angle-closure glaucoma||7 (25%)|
|Pseudoexfoliation glaucoma||4 (14%)|
|Visual Field a|
|Mean deviation (dB)||−9.6 ± 6.0|
|PSD (dB)||6.4 ± 3.9|
|Total number||13.4 ± 2.3|
|Preoperative number||6.6 ± 1.4|
|Postoperative number||6.8 ± 2.1|
|Total follow-up (mos)||85.2 ± 14.4|
|Preoperative follow-up (mos)||43.2 ± 14.4|
|Postoperative follow-up (mos)||42.0 ± 12.0|
Significant IOP reduction occurred after glaucoma filtering surgery ( Figure ). Mean IOP ± SD decreased from 19.0 ± 3.9 mm Hg before surgery to 11.3 ± 3.7 mm Hg after surgery, resulting in a mean ± SD IOP reduction of 7.7 ± 5.6 mm Hg (40% reduction; P < .01). Of the eyes with a mean preoperative IOP less than 18 mm Hg (13/28), the range of percent IOP reduction was 22% to 61%. The mean IOP fluctuation ± SD also decreased from 4.8 ± 3.2 mm Hg to 2.8 ± 1.9 mm Hg ( P < .01; Table 2 ). The mean number of glaucoma medications ± SD was 3.2 ± 0.8 (range, 1 to 5) before surgery and 0.7 ± 1.1 (range, 0 to 3) after surgery ( P < .01). Eighteen eyes (64%) were receiving no glaucoma medication at the time of their last follow-up visit.
|Study End Points||Before Surgery||After Surgery||P Value|
|Mean IOP (mm Hg)||19.0 ± 3.9||11.3 ± 3.7||<.01|
|Mean IOP fluctuation (mm Hg)||4.8 ± 3.2||2.8 ± 1.9||<.01|
|Global rate of progression (dB/year)||−1.48 ± 1.4||−0.43 ± 0.8||.01|
|No. of significantly progressing points||1.4 ± 3.6||0.07 ± 0.3||<.01|
The mean ± SD global rate of progression decreased from −1.48 ± 1.4 dB/year before surgery to −0.43 ± 0.8 dB/year after surgery (70% reduction; P = .01; Table 2 ). A global progression rate of more than 1.5 dB/year was present in 9 eyes (32%) before surgery and in 2 eyes (7.1%) after surgery (78% reduction; P = .04).
Twelve eyes (42.8%) had at least 1 significantly progressing point before surgery, whereas only 2 (7.1%) had at least 1 progressing point after surgery ( P < .01, McNemar test). Of the 2 eyes that showed significant progression before and after surgery, one had its local slope reduced from −4.1 to −1.7 dB/year, and the other from −2.5 to −1.4 dB/year. The mean number of significantly progressing points per eye ± SD decreased from 1.4 ± 3.6 to 0.07 ± 0.3 after surgery ( P < .01; Table 2 ). Considering only eyes receiving 0 medications after surgery (18/28), the mean IOP decreased from 19.7 ± 4.2 mm Hg to 11.0 ± 3.2 mm Hg ( P < .01), and the mean IOP fluctuation decreased from 5.4 ± 3.3 mm Hg to 3.1 ± 2.1 mm Hg ( P < .01). The mean global progression rate also was reduced significantly from −1.6 ± 1.6 dB/year to −0.3 ± 0.4 dB/year ( P < .01). Eight eyes (44%) had at least 1 significantly progressing point before surgery, whereas 1 eye (5.5%) maintained sustained progression after the procedure ( P = .02). Because the observed reduction in the rates of progression of the operated group might have been skewed by a regression to the mean biased by eyes with a faster rate of progression before surgery, we performed a median split of the global rates of progression and compared the slopes before and after surgery within each group. The group with faster global rate (> −0.9 dB/year) went from a mean rate of progression of −2.5 ± 1.3 dB/year to −0.6 ± 1.1 dB/year before and after surgery, respectively (76% reduction; P < .01). In the group with a slower progression, the global rate decreased from −0.4 ± 0.3 dB/year to −0.2 ± 0.3 dB/year (50% reduction; P = .02).
When analyzing the control group (≥ 10 VF and no surgery), the average rate ± SD for the first period was −0.2 ± 0.4 dB/year and 8 eyes reached a progression end point, whereas in the second period, the mean rate was −0.3 ± 0.4 dB/year ( P = .55) and 15 eyes reached a progression end point ( P = .16).
Linear regression analysis comparing the absolute IOP lowering (in millimeters of mercury) and the decay in global rate of progression (dB/year) was positive and significant ( r 2 = 0.20; y = 0.1 x + 0.3; P = .01). Each 1 mm Hg of IOP reduction after surgery resulted in 0.1 dB/year decrement in the global rate of progression.