To compare rim area rates in patients with and without the visual field (VF) progression endpoint in the Canadian Glaucoma Study and determine whether intraocular pressure (IOP) reduction following the endpoint altered rim area rate.
Prospective multicenter cohort study.
setting : University hospitals. patient population : Two hundred and six patients with open-angle glaucoma were examined at 4-month intervals with standard automated perimetry and confocal scanning laser tomography. intervention : After the endpoint, IOP was reduced by ≥20%. outcome measures : Univariate analysis for change in rim area rate and multivariable analysis to adjust for independent covariates (eg, age, sex, and IOP).
Patients with an endpoint (n = 59) had a worse rim area rate prior to the endpoint compared to those without (n = 147; median [interquartile range]: −14 [−32, 11] × 10 −3 mm 2 /y and −5 [−14, 5] × 10 −3 mm 2 /y, respectively, P = .02). In univariate analysis, there was no difference in rim area rate before and after the endpoint (median difference [95% CI], 8 (−10, 24) × 10 −3 mm 2 /y), but the muItivariate analysis showed that IOP reduction >2 mm Hg after the endpoint was strongly linked to a reduction in rim area rate decline (8 × 10 −3 mm 2 /y for each additional 1 mm Hg reduction).
Patients with a VF endpoint had a median rim area rate that was nearly 3 times worse than those without an endpoint. Lower mean follow-up IOP was independently associated with a slower decline in rim area.
Estimating the rate of visual field (VF) change in patients with glaucoma is a key prognostic indicator for predicting the risk of future visual disability. In recent years, there has been considerable research directed toward estimating the rate of VF change in clinical studies and trials, as well as in routine clinical practice. There are also several recent reports on how fast neuroretinal rim area and retinal nerve fiber layer thickness change in glaucoma. Confocal scanning laser tomography (CSLT) provides a useful and objective method for obtaining rim area measurements of the optic nerve head. With this technique, average rates of rim area loss in glaucoma patients of between −5 × 10 −3 mm 2 /year (y) and −15 × 10 −3 mm 2 /y have been reported. In contrast, the average longitudinal normal age-related rim area loss measured with CSLT is about −1 × 10 −3 mm 2 /y.
The Confocal Scanning Laser Ophthalmoscopy Ancillary Study to the Ocular Hypertension Treatment Study (OHTS) found that patients who reached a study endpoint and developed manifest glaucoma had a rim area rate that declined, on average, 5 times faster compared to those who did not. Similarly, in another study, glaucoma suspects who developed VF loss had a more rapid rate of rim area decline than patients who did not develop VF loss. On the other hand, Alencar and associates demonstrated that rim area rates with CLST were not different among glaucoma patients and suspects with or without progression determined by VF change or optic disc change determined subjectively by stereophotography.
The goal of glaucoma treatment is to slow the rate of disease progression to minimize the chance of visual disability. There is now ample evidence from randomized clinical trials that lowering intraocular pressure (IOP) slows the rate of VF change. However, similar evidence on the effect of treatment on the rate of structural loss in patients with glaucoma is lacking. Ascertaining how treatment intervention affects structural change over time will establish the utility of CSLT for monitoring patients with glaucoma and guide its appropriate use.
The Canadian Glaucoma Study was undertaken to identify risk factors for VF progression in glaucoma. We previously showed that the rate of mean deviation (MD) change was worse in patients who reached an event-based VF progression endpoint compared to those without an endpoint, and that IOP lowering mandated after the endpoint ameliorated the rate of MD decline. As an analogous investigation, the current study had 2 specific aims with regard to the rim area rate in Canadian Glaucoma Study patients: first, to determine whether the rim area rate was different in patients with a VF endpoint compared to those without an endpoint; and second, to determine whether the level of IOP was associated with a change in rim area rate.
The Canadian Glaucoma Study is a prospective multicenter interventional cohort study involving 5 Canadian hospital-based university departments. Detailed methodology for the Canadian Glaucoma Study has previously been described and is summarized here. Open-angle glaucoma patients were followed up with a standardized protocol for IOP control and examined at 4-month intervals with standard automated perimetry (full-threshold 30-2 program, Humphrey Field Analyzer; Carl Zeiss Meditec, Dublin, California, USA), short-wavelength automated perimetry (Humphrey Field Analyzer), and CSLT (Heidelberg Retina Tomograph, HRT Classic; Heidelberg Engineering GmbH, Heidelberg, Germany). Patients with a study endpoint defined as confirmed VF change assessed with an event-based criterion received an additional ≥20% reduction in IOP with a standardized treatment protocol.
The Canadian Glaucoma Study is a registered clinical trial (NCTT00262626) that was approved by the research ethics board of each participating center. Written informed consent was obtained from each patient. The study and data accumulation were carried out with approval from the appropriate Institutional Review Board (IRB). The details of the trial design and primary analysis have been published elsewhere.
Inclusion and Exclusion Criteria
Patients with either newly or previously diagnosed open-angle glaucoma (including pseudoexfoliation glaucoma) were enrolled. Inclusion criteria were best-corrected visual acuity of 6/10 or better (Early Treatment Diabetic Retinopathy Study [ETDRS] visual acuity chart); photographically documented glaucomatous optic disc change; glaucomatous VF damage with MD better than −10 dB and a positive glaucoma hemifield test; and nonoccludable anterior chamber angles determined gonioscopically. Exclusion criteria were significant comorbid ocular disease (other than glaucoma); chronic use of ocular medication (other than for glaucoma); other ocular or systemic disease affecting the VF; spherical equivalent >6 diopters (D) or astigmatism >2.5 D; and previous incisional glaucoma surgery.
Visual Field Endpoint
The VF endpoint was based on event-based progression criteria with automated perimetry. Progression was suspected when ≥8 VF locations, with ≥4 clustered within a single hemifield, were identified in the total deviation change map. A first confirmation examination was then performed to verify the change within 7–10 days. Progression was confirmed when there was an overlap of ≥4 locations, with ≥2 of these clustered within 1 hemifield. If progression was not confirmed at this examination, a third examination was conducted within 7–10 days. Hence, progression had to be demonstrated in 2 examinations. These VF change criteria were subsequently shown to have high specificity, while identifying progression relatively early compared to other criteria available at the time. The endpoint was taken as the time when progression was first suspected, and IOP intervention introduced at the visit the VF change was confirmed. A new baseline VF status was established and additional endpoints required confirmation in the same manner.
Intraocular Pressure Protocol
In newly diagnosed patients, a target IOP of ≥30% reduction from the pretreatment IOP level was mandated, whereas previously diagnosed patients entered the study at an ophthalmologist-defined target IOP. The untreated IOP for the latter group was obtained from the referring ophthalmologist or optometrist. For patients reaching an endpoint, the target IOP was reset, with an additional ≥20% reduction after each endpoint. IOP lowering was achieved with a predefined treatment protocol starting with monotherapy and advancing to adjuvant topical therapy, oral carbonic anhydrase inhibitors or argon laser trabeculoplasty, and trabeculectomy, as required to achieve target IOP.
The mean IOP for each patient was calculated for all follow-up visits if there were no endpoints, and for the follow-up before and after endpoints in patients who had an endpoint. IOP changes after intervention for each patient were measured by computing the difference between the mean IOP of all visits before and after intervention.
CSLT was performed at baseline and then at 4-month intervals thereafter. Three consecutive images were acquired, each capturing a 10 × 10 degree image centered on the optic nerve head. The software allows for image alignment in a scan series to correct for the effect of small eye movements. The mean topography of the 3 images was used for analysis. After acquisition, the contour line was drawn in the baseline image to depict the clinically visible optic disc margin and subsequently exported to all follow-up images. Rim area was computed with the reference plane located 50 μm posterior to the temporal disc margin at baseline. The axial distance between the reference plane and the reference ring (located in the image periphery) was maintained for the follow-up images. Global rim area was used in the analysis. Unless there were obvious image artifacts, all images were included in the analysis.
Statistical analyses included both univariate and multivariable analysis. Univariate estimates provided unadjusted estimates of rim area rates and rate differences. A multivariable analysis was used to deduce factors that are independently associated with rim area rate and provide rim area rate estimates adjusted for these covariates. In patients without an endpoint, only those with ≥4 CSLT examinations were included in the analysis. In patients with an endpoint, only those with ≥4 CSLT examinations before and after the endpoint were included in the analysis.
A univariate analysis was performed to provide a population estimate of rim area rate in patients who reached the first endpoint and those who did not. For patients with an endpoint, the rim area rates before and after the endpoint were compared, while in those patients with no endpoints, all images available were included in the analyses to mirror the previously reported VF analysis. Rim area rate was derived with robust linear regression, which overcomes some of the limitations of ordinary least squares regression, including minimizing the influence of outlier observations and heteroscedasticity. Rim area rates were compared between patients with and without an endpoint with the Mann-Whitney test. In patients with an endpoint, rim area rates were compared before and after the endpoint with the Wilcoxon test.
In the multivariable linear mixed-effects model, we included age, sex, and positive anticardiolipin antibody test, the previously identified independent risk factors for VF progression. Random effects included a random intercept and slope for patients before and after each endpoint and IOP intervention (if one occurred), with an additional random intercept for study center, effectively allowing each patient’s rim area change to vary about that of the mean rim area change. Confidence intervals and statistical significance of fixed effects estimates were derived using a parametric bootstrap procedure based on the fitted model.
The analysis was based on the study eye, which was chosen by a random selection technique at the beginning of the study if both eyes were eligible. All statistical analyses were performed within R (R Foundation for Statistical Computing, Vienna, Austria; www.R-project.org ) with the robust package for robust regression and the lme4 package for mixed-effect modeling.
Of the 258 patients enrolled in the Canadian Glaucoma Study, 206 (79.8%) met the inclusion criteria for the number of CSLT examinations. The median (interquartile range) follow-up was 6.0 (4.7, 7.3) years. Fifty-nine patients (28.6%) had ≥1 VF endpoint, while 147 patients (71.4%) did not have an endpoint. Baseline characteristics of patients with and without an endpoint are shown in Table 1 . The patients with an endpoint were older ( P = .01) and had lower global rim area at baseline ( P = .01) compared to patients without an endpoint. Sex, untreated IOP, and optic disc area were similar for the 2 groups ( P > .20). The median (interquartile range) mean pixel height standard deviation (MPHSD, an indicator of CSLT image quality, with lower values indicating better image quality) for all patients and examinations was 25 μm (18, 37), with 90% of images having MPHSD of ≤50 μm. The median MPHSD was similar in the 2 groups ( P > .99).
|VF Endpoint (N = 59)||No VF Endpoint (N = 147)|
|Age (y)||68.1 (63.0, 74.5)||62.5 (53.0, 70.7)|
|Global rim area (mm 2 )||1.00 (0.88, 1.17)||1.12 (0.94, 1.37)|
|Optic disc area (mm 2 )||2.05 (1.70, 2.33)||2.05 (1.81, 2.41)|
|Untreated IOP (mm Hg)||25.0 (22.5, 27.0)||26.0 (23.0, 28.0)|
|MPHSD (μm)||24 (18, 36)||25 (18, 36)|
The distribution of univariate rim area rates in patients with a VF endpoint (derived with data only up to the first endpoint) and without an endpoint is shown in Figure 1 . The median (95% confidence interval [CI]) rim area rate was significantly ( P = .02) more negative in the patients with an endpoint (−14 [−32, 11] mm 2 × 10 −3 /y) compared to patients without an endpoint (−5 [−14, 5] mm 2 × 10 −3 /y).
There were 43 patients with ≥4 CSLT examinations before and ≥4 examinations after IOP-lowering intervention. The median (95% CI) change in rim area rate was 8 (−10, 24) mm 2 × 10 −3 /y ( Figure 2 ) and not significantly different from zero ( P = .72). The relationship between the rim area rate difference and IOP reduction was not statistically significant ( r = 0.24, P = .10).
Table 2 shows the independent associations between rim area rates and the fixed-effects covariates in the multivariable model. Although patients with a VF endpoint had a more negative rim area rate prior to the endpoint (−12 × 10 −3 mm 2 /y, P = .05, compared to 0) than patients without an endpoint (−6 × 10 −3 mm 2 /y, P = .09), the difference was not statistically significant ( P = .32). For patients with an endpoint, the mean (standard deviation [SD]) IOP reduction over the follow-up was 2.3 (2.7) mm Hg after intervention. On average, the rim area rate changed nonsignificantly from −11 × 10 −3 mm 2 /y to −12 × 10 −3 mm 2 /y after the endpoint ( P = .88). However, for every additional 1 mm Hg reduction in IOP above the mean after the endpoint, rim area rate changed significantly by 8 × 10 −3 mm 2 /y ( P < .01). Therefore a reference patient ( Table 2 ) who had a reduction of 3 mm Hg would have an estimated rim area rate of −13 + 8, or −5 × 10 −3 mm 2 /y; a patient with a 4 mm Hg reduction would have a rim area rate of −5 + 8, or 3 × 10 −3 mm 2 /y after the endpoint; and so on. Hence, a reduction of IOP greater than 2 mm Hg was strongly linked to a reduction in the rate of rim area decline.
|RA Rate Estimate |
(95% CI, mm 2 × 10 −3 /y)
|Patients without an endpoint||−6 (−13, 1)|
|Difference in RA rate between patients (before endpoint or without endpoint) with IOP increase, per 1 mm Hg increase||−2 (−4, 0)|
|Patients with an endpoint|
|Before endpoint||−11 (−22, 0)|
|After endpoint. Change in RA rate for mean IOP reduction of 2.3 mm Hg||−1 (−30, 27)|
|After endpoint. Change in RA rate with each additional 1 mm Hg IOP reduction||8 (3, 13)|
|Change in RA rate with age for each additional decade||−3 (−7, 1)|
|Difference in RA rate when male||3 (−6, 12)|
|Difference in RA rate with presence of anticardiolipin antibody||−1 (−22, 21)|