Factors for Glaucoma Progression and the Effect of Treatment: Title and subTitle BreakThe Early Manifest Glaucoma Trial FREE

THE FACTORS related to the progression of open-angle glaucoma (OAG) have been evaluated in many studies, with variable findings reported.^{1 - 11} This variability may be explained by differences in study design, methods of data collection, specific factors evaluated, or approaches used for statistical analyses. Some studies have been based on retrospective analyses of patient data, which are subject to various limitations, while others have provided stronger evidence by evaluating risk factors prospectively. In addition, the number and type of patient characteristics studied have ranged widely—from a broad spectrum of variables to only a few data items. At times, the level of intraocular pressure (IOP) has been used to select patients for study, with some reports focusing only on "high pressure" or "normal pressure" in patients with glaucoma, and in others, including a continuum of IOP. It is not surprising, therefore, to find that studies have reached different conclusions regarding the relative importance of factors influencing outcome.

The role of IOP-lowering treatment on progression was recently assessed in the Early Manifest Glaucoma Trial (EMGT).^{12 - 13} The EMGT is a randomized clinical trial designed to evaluate the effect of immediate treatment on glaucoma progression, as compared with no initial treatment or later treatment. All EMGT patients had early and previously undetected glaucoma with visual field defects, and most were identified through a large (N = 44 243) population-based screening in Malmö and Helsingborg, Sweden. Study participants were randomized to argon laser trabeculoplasty plus betaxolol, or to no immediate treatment. Patients were examined every 3 months for at least 4 years, and glaucoma progression was determined through specific visual field and optic disc criteria, as described in detail elsewhere.^{12 - 14} The EMGT thus allows a prospective evaluation of factors for progression, following a standardized protocol, within the context of a randomized trial comparing treated and control arms.

The rationale and specific details of the EMGT design have been reported, as have the major results of the trial.^{12 - 13} The EMGT showed that treated patients had significantly lower rates of progression than controls, and that immediate IOP-lowering treatment significantly delayed the progression of OAG.¹³ This article adds to these main results by presenting multivariate analyses that jointly evaluate the effect of treatment and patient-related factors on progression. This matter has been difficult to assess conclusively, since most glaucoma clinical trials have evaluated the effect of different modes of treatment, rather than of treatment itself.^{15 - 20} In the Collaborative Normal Tension Glaucoma Study (CNTGS), which was limited to patients with a median IOP of 20 mmHg or less, the intent-to-treat analyses yielded no differences between treated and untreated patients, but a significant effect of treatment was reported after censoring for cataract outcomes.¹⁹ The role of other characteristics affecting glaucoma progression has been reported for patients in the untreated arm of that study, ¹⁰ but not for treated patients.

The first aim of this article is to estimate and evaluate the magnitude of the treatment effect in EMGT, while controlling for other factors. This aim was achieved by comparing progression rates in both study groups during at least 4 years of follow-up, while adjusting for other variables possibly related to progression. In addition to providing a quantitative measure of the IOP-lowering effect of EMGT treatment, further analyses explored possible interactions with treatment.

The second aim is to identify clinically relevant factors that are independently related to glaucoma progression in EMGT. This aim was achieved by evaluating the role of demographic, systemic, familial, and ocular factors on EMGT progression. The intent was to provide predictive information on the risk of progression according to specific patient characteristics at baseline. The possible role of longitudinal changes in clinical findings at follow-up was also examined, as the results could contribute to our understanding of the factors influencing glaucoma progression.

Details on the study design have been described elsewhere.¹² The trial setting included a clinical center (Malmö University Hospital, Malmö), a satellite clinical center (Helsingborg Hospital, Helsingborg), a data center (School of Medicine at Stony Brook, Stony Brook, NY), and a disc photography reading center (Lund University Hospital, Lund, Sweden). The study is regularly monitored by a Data and Safety Monitoring Committee, which oversees all aspects of the trial. Funding was provided by the National Eye Institute (Bethesda, Md) and the Swedish Research Council (Stockholm).

Briefly, men and women aged 50 to 80 years who had newly diagnosed, previously untreated, early manifest OAG were eligible for inclusion. The OAG diagnosis (including chronic simple glaucoma, normal-tension glaucoma, and exfoliative glaucoma) required repeatable glaucoma visual field defects in at least one eye that were not explained by other causes, and that were assessed by computerized perimetry.^{21 - 23} Patients with the following were excluded: (1) advanced visual field defects; (2) visual acuity worse than 0.5; (3) mean IOP greater than 30 mmHg, or any IOP greater than 35 mmHg in at least one eye; (4) lens opacities²⁴ or any condition precluding reliable visual field or disc photography, or use of study treatments or 4-year follow-up. The EMGT protocol includes 4 prerandomization visits (ie, 2 postscreening visits and 2 baseline visits), as well as laser treatment visits for the treated patients. All patients have follow-up visits every 3 months, with data being collected by trained and certified EMGT examiners, following a standardized protocol.

Progression, as defined by EMGT, is evaluated by standardized, independently determined criteria that are based on perimetry or optic disc assessment. Perimetric criteria were objectively determined and defined as significant changes from baseline in at least 3 of the same progressing points in 3 consecutive visual fields, as assessed by pattern deviation–based Glaucoma Change Probability Maps.²⁵ As was necessary for the design of the trial, these criteria were designed to be highly sensitive to detect visual field changes—an aim that was achieved.¹⁴ Optic disc progression was evaluated at a masked reading center and was based on photographic criteria designed to have high specificity. These criteria required clear change on an optic disc follow-up photograph, as detected by flicker chronoscopy and confirmed by side-by-side gradings in 2 consecutive visits. Progression as EMGT defines it, is patient-specific, occuring when at least one eye meets progression criteria. For patients with one eligible eye at baseline, only that eye was considered in the analyses. For patients with 2 eligible eyes at baseline, time of progression for the first progressing eye was used in life-table and multivariate analyses.

For univariate analyses, the unit of analysis was based on the patient for person-based covariates (eg, age and sex), and based on the eye for eye-based covariates (eg, exfoliation status and IOP). The multivariate analyses strictly used the person as a unit of analysis; for patients with 2 eligible eyes, the following selection criteria were used. If one eye progressed first, then that eye was considered for the analyses. If neither eye progressed (or if both progressed at the same time), then the worse of the 2 eye-based covariate measurements at baseline was considered (eg, IOP, mean deviation [MD], disc hemorrhages, and exfoliation). To evaluate the potential effects of the "worse-eye" selection criteria on the results, an additional approach was to randomly select one eye of patients with 2 eligible nonprogressing eyes (or 2 eyes progressing at the same time), rather than the eye with the worse measurement. This selection scheme yielded results indistinguishable from those using our initial selection criteria. The results of patient-based analyses were also compared with those obtained by eye-based analyses, accounting for intereye correlation.²⁶ Again, the results were essentially the same as those from the patient-based analyses, with similar hazard ratio(HR) magnitudes.

Univariate analyses were based on summarizing percent progression according to a specific variable (eg, IOP at baseline) and did not include simultaneous adjustment for other covariates. The Pearson χ² test was used to test for a significant difference in percent progression for categorical variables, and the t test was used for differences in progression on continuous variables.

Based on the constancy of the HR throughout follow-up time, multivariate analyses used Cox proportional hazard models, ²⁷ with Breslow adjustment for ties in time to progression, ²⁸ to(1) model the hazard (follow-up specific conditional probability of progression) of the treated group as a constant multiple of the hazard in the control group, while (2) simultaneously adjusting for other study covariates. Criteria for model selection were guided by: (1) the findings of the univariate analysis(eg, P≤.20); (2) simultaneous adjustment for important covariates; and (3) identification of the most parsimonious, clinically interpretable, and statistically fitting model, which involved using backward, forward, and stepwise variable selection algorithms in SAS.²⁹ The model selection also included testing for interaction between the study groups and the different levels of each covariate.

As a first step, the main effects (independent) models were pursued to define the baseline factors that were significantly associated with progression of glaucoma. All 3 model-selection criteria identified the same variables for inclusion in the final model. As a second step, longitudinal changes in IOP, and clinically assessed disc hemorrhages at follow-up visits were also evaluated while adjusting for significant baseline factors. As a third step, 2-factor interaction models were fit using those factors identified as significantly associated with glaucoma progression. Results are expressed as HRs with 95% confidence intervals (CIs). Analyses conducted separately in each study group yielded the same conclusions.

Study covariates were selected based on their potential clinical importance and statistical association with glaucoma progression. In addition to study group (treated or control), other baseline factors were evaluated. Demographic variables were age, sex, and clinical center.

Ocular variables were number of eligible eyes, IOP (average of 2 baseline Goldmann measurements), perimetric MD (average of 2 baseline fields), exfoliation(dilated examination), and refractive error (automated refractor); clinically observed disc hemorrhages (dilated examination); and central corneal thickness(ultrasonic pachymeter; measured after baseline¹³ ).

Medical and family history variables were casual systolic and diastolic blood pressure measurements; hypertension (systolic >160 mmHg or diastolic>95 mmHg, or use of antihypertensive medications) self-reported history of cardiovascular disease, use of hypertension medications, low blood pressure, migraine, Raynaud disease, smoking, and family history of glaucoma (in either parent or any sibling).

The follow-up covariates evaluated were: (1) initial IOP change from baseline to the first follow-up visit (IOP at baseline minus IOP at 3 months' follow-up); (2) IOP at the first follow-up visit (IOP at 3 months' follow-up);(3) later IOP change beyond the first follow-up visit (IOP at 3 months' follow-up minus IOP at progression [for those who progressed] or at the end of follow-up[for those who did not progress]); (4) mean follow-up IOP (average IOP at all follow-up visits until progression or until the end of follow-up); and(5) percent of all patient visits with clinically assessed disc hemorrhages until progression or until the end of follow-up.

The 255 patients enrolled in EMGT had a median age of 68 years, and 66% were female. As reported previously, both groups were balanced at baseline with regard to major variables.¹³ Retention was excellent, with 227 of 255 patients (89%) completing follow-up through September 2001. At that time, the median length of follow-up was 6 years and was similar in both groups. Death was the major reason for losses to follow-up(n = 22); only 6 patients (2%) were lost to follow-up for other reasons. Of the EMGT-designated visits, 99% were fully completed, and the missed-visit rate was very low at only 5% (289/5744 patients).

Overall, 53% (136/255) of the patients progressed, and 47% (119/255) did not progress during the follow-up period. Table 1 presents univariate comparisons of percent progression according to several patient characteristics; the study groups were balanced at baseline in all the factors presented. Progression was significantly lower in the treated group than the control group (45% vs 62%; HR = 0.60). Patients above the median age had a higher percentage of progression, as did patients with higher median IOP, worse median MD, exfoliation, and both eyes eligible for the trial. Analyses based on continuous variables and evaluated as such, gave similar results as analyses based on median values. No significant associations (P<.05) were found with the other factors evaluated.

Table 2 presents results of multivariate analyses evaluating associations of EMGT progression with the baseline factors presented in Table 1. Study group (treated vs control) was significantly associated with progression (P<.001). In analyses based on the hazard function, which adjusts for censoring and other covariates, the treated group had half the risk of progression (HR = 0.50) as compared with controls.

An IOP at or above the median at baseline increased the risk of progression(HR = 1.70), as well as the presence of exfoliation (HR = 2.31) and having 2 eligible eyes at baseline (HR = 1.93). Increased progression risks were also found in patients with worse median baseline MD (HR = 1.55) and patients above the median age (HR = 1.43). Similar estimated treatment effects were found when we repeated these analyses using continuous values of IOP, MD, and age, rather than median values. In these analyses, the risk of progression increased by 5% with each millimeter of mercury of higher baseline IOP (HR= 1.05; 95% CI, 1.01-1.10). Results were also consistent with the higher progression of patients with worse MD and older age (MD: HR = 1.03 per 1 d B [decibel] of worse MD; [95% CI, 0.98-1.09]); age: HR = 1.01 per 1 year of age; [95% CI, 0.98-1.05]). No significant relationships were found with other factors in Table 1. In addition, no statistically significant interactions were found with treatment.

Patients randomized to treatment had a substantial lowering of IOP. At the first follow-up visit after randomization (3 months' follow-up), there was an average reduction of 5.1 mmHg, or 25% from baseline in the treated group, with no changes in the control group.¹³ After this visit, the difference in IOP between groups was generally maintained over time. Both in treated and control patients, average differences of less than 1 mmHg were observed between the IOP at the first follow-up visit and the IOP at the time of progression (for those who progressed) or the end of follow-up (for those who did not progress). For that reason, we separately evaluated associations with each of the follow-up IOP variables defined in the "Methods" section (ie, the initial IOP change [baseline minus 3 months' follow-up]; the 3 months' follow-up IOP itself, representing the posttreatment baseline IOP; the later IOP change [beyond the IOP at 3 months]; and the mean IOP at follow-up). Table 3 presents the results of each of these multivariate analyses, evaluating the association between EMGT progression and those measures of longitudinal changes in IOP after baseline, while controlling for IOP, exfoliation, number of eligible eyes, MD, and age.

The first row of Table 3 substantiates that progression is related to the magnitude of initial IOP change from baseline to the first follow-up visit. This initial change in IOP was strongly and inversely associated with progression. Thus, an IOP reduction of 1 mmHg from baseline decreased the risk of progression by about 10% in these analyses.

The second row of Table 3 similarly indicates that progression was strongly associated with the initial IOP reached after treatment or with no treatment (IOP at 3 months' follow-up). This IOP, which reflected the effects of study group assignment and of baseline IOP, was a significant predictor of progression, with an estimated 11% higher risk for every millimeter of mercury higher IOP—a result consistent with the estimate presented in the first row of Table 3. The later IOP change was not a significant factor when accounting for the 3 months' IOP. The latter result would be expected, given the small changes in IOP after the initial follow-up visit.

The third row of Table 3 further confirms that the mean follow-up IOP achieved after baseline is related to progression, with an estimated 13% higher risk per each millimeter of mercury higher IOP.

In all these analyses, baseline IOP was not significantly associated with progression. When study group was added to the model, results were essentially the same, although study group was no longer retained as a significant factor.

The final row of Table 3 presents results that examine associations with disc hemorrhages observed during follow-up, while controlling for the mean IOP during the follow-up period and the significant variables. The percent of patient visits with disc hemorrhages was strongly related to progression, with a 2% increase in risk for every percentage point.

Several baseline factors were independently related to EMGT-defined progression (Table 2). Treated patients had half the risk of progressing of control patients, indicating the efficacy of EMGT treatment. Other factors included a higher IOP at baseline, the presence of exfoliation, having 2 eyes eligible for the trial, worse median MD, and older median age. No other baseline factors evaluated were related to progression in these analyses.

Progression was closely linked to the magnitude of the initial IOP reduction with treatment. The initial change in IOP (from baseline to the initial follow-up visit) was strongly associated with progression, with about a 10% lowering of the risk with each mmHg of IOP reduction (Table 3). Consistent with and related to this finding, the IOP level achieved after this initial change (ie, the 3-months IOP) was also a strong predictor of progression (Table 3), as was the mean IOP at follow-up (Table 3), with about a 10% higher risk with each mmHg of higher IOP. In analyses that included the posttreatment IOP, neither baseline IOP nor later change in IOP were significantly related to progression. While interpretation of these nonsignificant findings must consider the high degree of dependence among all these factors, the results support the major prognostic importance of the IOP achieved after the initial reduction, representing the posttreatment baseline (Table 3, row 2). Frequent disc hemorrhages at follow-up were also an independent factor for progression(Table 3).

The multivariate analyses have quantified the magnitude of the IOP-lowering effect of treatment, which halved the risk of EMGT progression after adjusting for baseline IOP and other variables (Table 2). These analyses thus contribute to the main results of the trial¹³ and provide new quantitative information on the effect of IOP-lowering treatment to reduce progression in early OAG, while accounting for other relevant factors. As such, it addresses the main study question regarding the effect of IOP reduction on progression. Such knowledge can be obtained only from rigorous randomized trials with an untreated control arm that were designed to evaluate this issue.^{30 - 31} To our knowledge, the EMGT provides the first estimate that meets these criteria. The previous estimates of treatment efficacy in glaucoma have been based on nonrandomized studies with various limitations³² or on randomized trials comparing various treatments, ^{15 - 16 ,19} which addressed a different research question.

Of the 2 previous randomized glaucoma trials that measured treatment effects directly by including an untreated control arm, the first had a small sample size and yielded negative results.³³ The second trial was the CNTGS, which randomized 145 patients (average IOP <20mmHg at baseline) to "treatment" or "no treatment." The intent-to-treat analyses, which are most comparable to the main EMGT analyses, ¹³ did not directly show an effect of treatment.¹⁸ Visual field progression developed at similar rates in the pressure-lowered and untreated arms (22/66 [33%] vs 31/79 [39%], respectively). Significant differences favoring treatment were detected only after additional analyses censoring for cataract(8/66 [12%] vs 21/79 [26%]). The CNTGS results, therefore, did not yield a measure of treatment effects that is comparable to the EMGT results.

In the Collaborative Initial Glaucoma Treatment Study (CIGTS), the role of IOP lowering on progression was not clearly established, as medically treated patients had visual field outcomes similar to surgically treated patients, yet IOP was 3 mmHg lower in the latter group.³⁴ Other comparisons within clinical trials, although not based on intent-to-treat analyses, have provided suggestive evidence that IOP lowering induced by treatment decreases progression. In the other CNTGS report¹⁷ (not based on intent-to-treat), the course of treated patients was followed from the time they achieved a 30% reduction in IOP (mean ± SD = 210 ± 158 days after randomization), to the course of control patients followed up since randomization. These results indicated that treated patients reached progression less frequently than controls (7/61 [12%] vs 28/79 [35%]), supporting a favorable treatment effect.

The Advanced Glaucoma Intervention Study (AGIS) investigated the issue using nonrandomized comparisons of patients achieving different levels of IOP reduction, ³¹ finding a consistent association between lower IOP and decreased visual field progression. Few changes in AGIS field scores were seen in patients who achieved reductions to IOP of less than 14 mmHg early in the study (predictive analyses), as well as in patients who maintained an IOP lower than 18 mmHg throughout follow-up (associative analyses). In AGIS, the IOP achieved in the early follow-up period was positively correlated with subsequent IOP levels. The results of these analyses are very similar to our findings of a strong association between progression and initial changes in IOP after baseline (Table 3), as well as the IOP level reached at the first follow-up visit (Table 3), and the mean IOP after baseline (Table 3). We evaluated these factors separately, as they are very closely linked. Since the baseline IOP, the reduction in IOP, and IOP at 3 months' follow-up are all interdependent, they will compete for statistical significance if entered jointly into a model; hence, careful interpretation is needed for lack of associations in the various models. Still, the IOP achieved after the initial reduction emerged as a major predictor of future progression, indicating the importance of the initial changes following EMGT treatment.

The AGIS results are also consistent with our observation that IOP was relatively stable after the first follow-up visit. For this reason, later changes in IOP were not related to progression after accounting for the IOP level reached at the first follow-up visit (Table 3). Further follow-up of EMGT patients is needed to evaluate this issue, particularly since later visits in AGIS were not as highly correlated with the initial posttreatment IOP as were earlier visits (eg, r = 0.63 at 24 months vs r = 0.35 at 96 months). Additional data on IOP will be provided in a future EMGT report.

Further support for the magnitude of the estimated EMGT treatment effect comes from the Ocular Hypertension Treatment Study (OHTS).³⁵ This trial randomized ocular hypertensives to IOP-lowering treatment or to no treatment, and compared the development of OAG in both study groups. After 5 years, the cumulative probability of developing glaucoma was 4.4% in the treatment group vs 9.5% in the untreated group, for an HR of 0.40, which is similar to the HR of 0.50 that was found in our EMGT analyses. While the OHTS is addressing a different research question among ocular hypertensives and not glaucoma patients, and while it has a somewhat shorter follow-up period, its results are highly consistent with EMGT findings.

Most (83%) of the 23 patients with exfoliation at baseline progressed(Table 1), and this condition was a major factor predicting glaucoma progression (Table 2). While known to have a wide geographic distribution, ³⁶ the prevalence of exfoliation is reported to be particularly high in Scandinavia, accounting for more than half of the OAG cases in a population study in Sweden.³⁷ In contrast, exfoliation was present in fewer than 10% of EMGT patients—a fact most likely related to the study's eligibility criteria for early glaucoma. Another important reason for the low frequency in EMGT is age, given the late age of onset of exfoliation, with prevalences usually reported in individuals older than 60 years, ³⁸ as it is seen rarely in persons younger than 50 years.³⁸ Despite this relatively low frequency, exfoliation emerged as an important and independent predictor of progression, which more than doubled the risk. This increased risk was an expected finding, being consistent with the more severe clinical course of exfoliation glaucoma.³⁹ Since persons with this type of glaucoma have higher IOP than others, ⁴⁰ their increased risk could be attributed, at least in part, to the elevated IOP. In fact, 20 of the 23 EMGT patients with exfoliation had an IOP of 21 mmHgor more. It is important to emphasize, however, that the strong association with exfoliation was present while controlling for IOP. As such, exfoliation itself seems to confer a greatly increased risk.⁴¹ Vascular factors may be the possible mechanisms contributing to this increased progression, given the reports of altered hemodynamics in patients with exfoliation glaucoma.^{42 - 44}

Having 2 eligible eyes in the study was also a strong predictor of glaucoma progression, increasing the risk just under 2-fold. Progression was observed in 72% of the patients with bilateral disease, as opposed to 47% of those with one eye eligible (Table 1).Such patients had both eyes at risk for progression since the beginning of the study, and thus, they had a higher probability of progression than other patients. Patients with manifest bilateral visual field defects at enrollment might also have a more aggressive disease than patients with only one eye eligible. While EMGT patients with 2 eligible eyes had similar age and IOP to those with one eligible eye, their average MD was worse by 2.17 d B, suggesting more visual field damage for these patients.

Patients in EMGT generally had mild visual field loss, with a median MD of −4 d B. Those with MD worse than this level were at increased risk of progression as compared with patients with better MD. However, MD has not been always associated with subsequent progression.⁹ The EMGT results are consistent with those of previous studies, which also found that the extent of initial visual field damage is related to subsequent damage.^{2 ,34} Possibly, a worse MD in newly identified patients in our population screening may indicate a less favorable disease course.

Older patients showed an increase in their risk of progression as compared with younger patients. Several studies have reported similar results, such as CIGTS, which found older age to be associated with increased visual field scores at follow-up, ^{9 ,34} while in other studies, such as CNTGS, age was not associated with progression.^{2 ,10 - 11}

Disc hemorrhages are a well-known sign of glaucoma damage, ^{45 - 47} and their presence during follow-up is related to progression.^{48 - 50} While EMGT results confirmed these observations, they are offered with the caveat that our report is based on clinical assessment of disc hemorrhages. This method is subject to interobserver variation and may lead to considerable underascertainment, as compared with standardized photographic assessment of disc hemorrhages, which will be the subject of a future EMGT report.

Because of the limitations of the method of assessment, it is difficult to interpret the lack of an association with the presence of disc hemorrhages at baseline, which was recorded in 12% to 13% of the patients.¹³ To assess their role during follow-up, we attempted to quantify this variable by evaluating the percentage of patient visits with disc hemorrhages. As the percentage of patient visits with disc hemorrhages increased, the risk of progression correspondingly increased so that each percentage point implied a 2% higher progression risk. Frequent disc hemorrhages at follow-up were confirmed as an important sign and conferred a worse prognosis.

Although many variables were evaluated, no additional factors, other than those reported here, were found in our main patient-based analyses or our eye-based analyses. Of interest, with the exception of age, all the factors associated with progression were eye related (Table 1). In an investigation of progression factors, which was restricted to patients in the untreated arm of the CNTGS, progression was related to female gender, migraine, and Raynaud disease.¹⁰ None of those factors were significant in EMGT, a divergence that could be due to the difference in study populations. Migraine and Raynaud disease are both considered to be manifestations of vasospasm, which has been related to glaucoma in previous studies. However, EMGT patients reporting any of these conditions did not overlap (more than expected by chance). While they showed a nonsignificant trend to higher progression in univariate analyses, neither migraine nor Raynaud disease were significant factors in the multivariate analyses.

In the OHTS, factors related to the onset of OAG were older age, larger cup-disc ratios, higher IOP, greater pattern standard deviation, and thinner central corneal measurements.⁵¹ Of these variables, age, IOP, and visual field damage were also significantly related to glaucoma progression in EMGT, but not to central corneal thickness (Table 1). The latter result could be explained by the many differences in study populations between OHTS and EMGT, including IOP levels and age.

Randomized clinical trials provide the strongest evidence to assess the effects of treatment. The EMGT was carefully designed to meet this goal and assess the extent to which IOP-lowering treatment affected the progression of newly diagnosed, early OAG. The visual field and optic disc criteria to assess progression were predefined and did not change during the study, as well as being independently determined and confirmed. The detection of the early stages of field progression, a necessary safety feature, was achieved. The EMGT perimetric criterion has high sensitivity¹⁴ and detected visual field changes in EMGT patients earlier than other measures of progression.¹³ To facilitate interpretation, separate analyses have estimated the amount of visual field loss associated with EMGT-defined progression.¹⁴

Thorough attention was given to the overall conduct of the trial, as well as to ensuring high retention. Few patients were lost to follow-up, rigorous quality control protocols were implemented, extensive monitoring of study activities was conducted, and data quality was high throughout the study.¹³ An expert data safety and monitoring committee met regularly to provide independent oversight of the trial. We also made efforts to address potential methodologic issues in our statistical analyses, with similar results obtained from patient-based and eye-based approaches. For these reasons, we believe that EMGT provides firm evidence to answer its main study questions.

As in all trials, it is necessary to assess the generalizability of EMGT results. The study was based on previously undetected patients with early glaucoma field defects, who were mainly identified by a population-based screening. Since EMGT patients were newly diagnosed and had early disease, results are not directly applicable to patients with a more advanced stage of the disease, but it is likely that our results also apply to patients with more advanced disease.¹³ The study was conducted in Sweden and predominantly involved white people, so that appropriate caution is needed when extending EMGT results to other populations. Interpretation of the results must also consider the specific protocols for treatment and follow-up in EMGT, which are not necessarily applicable to all clinical situations.

The EMGT provides conclusive evidence to confirm that reduction in IOP lowers the risk of progression in early OAG. Furthermore, the analyses reported here estimate that the pressure reduction achieved in EMGT decreases the risk in half. The magnitude of the initial IOP reduction achieved after treatment emerged as a strong predictor of progression. Since on average, no marked differences existed between the IOP achieved after the initial reduction and subsequent IOP levels, the mean follow-up IOP was similarly related to progression.

Our analyses also identified clinical characteristics other than IOP, which were important and independent factors for progression. The trial also provides solid scientific evidence on the effectiveness of treatment to reduce glaucoma progression in a randomized clinical trial, which is needed to support the value of early detection and subsequent treatment of persons with glaucoma.^{52 - 56} These data have not been available in the past and have led to considerable uncertainties, not only in the clinical domain, but also pertaining to the rationale and merits of glaucoma screening, ^{55 - 57} a topic that we plan to address separately. The results of the EMGT, therefore, have clinical and public health implications.

Submitted for publication July 3, 2002; final revision received August 22, 2002; accepted September 26, 2002.

This study was supported by grants U10EY10260, U10EY10261, and K2002-74X-10426-10A from the National Eye Institute (Bethesda, Md) and the Swedish Research Council(Stockholm).

Early Manifest Glaucoma Trial Group

Clinical Center

Department of Ophthalmology, Malmö University Hospital, Malmö, Sweden: Anders Heijl, MD, PhD (study director); Bo Bengtsson, MD, PhD (screening director); Karin Wettrell, MD, PhD (ophthalmologist; 1992-2000); Peter Åsman, MD, PhD, (ophthalmologist); Boel Bengtsson, PhD (investigator; since 2001); Margareta Wennberg, BA (clinic coordinator); Gertie Ranelycke (technician); Monica Wollmer, RN (technician); Gunilla Lundskog, RN (technician); Katarina Magnusson (secretary).

Data Center

Department of Preventive Medicine, State University of New York at Stony Brook: M. Cristina Leske, MD, MPH (director); Leslie Hyman, PhD (deputy director); Mohamed Hussein, PhD (senior biostatistician); Qimei He, PhD (biostatistician; since 2001); Eugene Komaroff, PhD (biostatistician; since 2001); Ling-Yu Pai, MA (data manager); Lisa Armstrong (assistant data manager; since 1999).

Satellite Clinical Center

Department of Ophthalmology, Helsingborg Hospital, Helsingborg, Sweden: Kerstin Sjöström, MD (director); Lena Brenner, MD (ophthalmologist); Göran Svensson, MD (ophthalmologist); Ingrid Abrahamson, RN (head nurse); Nils-Erik Ahlgren, RN (technician); Ulla Andersson, RN (technician); Annette Engkvist, RN (technician); Lilian Hagert(secretary/clinic coordinator).

Disc Photography Reading Center

Department of Ophthalmology, Lund University Hospital, Lund, Sweden: Anders Bergström, MD (director; since 1997); Catharina Holmin, MD (director; 1993-1997); Anna Glöck, RN (photograder); Catharina Dahling Westerberg, RN (photograder); Inger Karlsson, RN (center coordinator).

National Eye Institute, Bethesda, Md

Carl Kupfer, MD (director; until 2000); Donald Everett, MA (program director).

Steering Committee

Bo Bengtsson, MD, PhD; Donald Everett, MA; Anders Heijl, MD, PhD; Leslie Hyman, PhD; M. Cristina Leske, MD, MPH.

Data Safety and Monitoring Committee

Curt Furberg, MD, PhD (chairman); Richard Brubaker, MD; Berit Calissendorff, MD, PhD; Paul Kaufman, MD; Maureen Maguire, PhD; Helge Malmgren, MD, PhD.

Corresponding author and reprints: M. Cristina Leske, MD, MPH, Department of Preventive Medicine, Stony Brook University School of Medicine, Health Sciences Center, L3 086, Stony Brook, NY 11794-8036 (e-mail: cleske@notes.cc.sunysb.edu).