The cumulative probability of conversion to POAG was significantly lower in the treatment group versus the observation group during the course of the study (hazard ratio, 0.40; p < 0.0001; Fig. 6.1). Conversion to POAG
at 5 years was 4.4% in the treatment group versus 9.5% in the observation group.

The OHTS demonstrated that IOP reduction reduces the risk of conversion from ocular hypertension to POAG. However, the total number of patients developing glaucoma was quite small (9.5% in the untreated group at 5 years). This resulted in a large number needed to treat (NNT) of 20 to prevent one patient from developing early POAG. This study also highlighted some important risk factors that are predictive of conversion to POAG, such as thin CCT (especially when combined with higher
baseline IOP). Hence, using these predictive factors and risk calculations now allows clinicians to more accurately predict risk in individual patients in order to refine treatment decisions. Notably, while the OHTS found that increased pattern standard deviation and cup-to-disc ratio were predictive factors of conversion to POAG, patients with these findings at baseline may have already had early glaucoma. This would not be expected to alter the main findings of this trial, but could affect its generalizability.

In the treatment group, the mean IOP was reduced from 20.6 to 15.5 mmHg (25% reduction) and the reduction was maintained throughout the study period. IOP was not significantly changed in the observation group. The median follow-up times were 66 and 69 months for the treatment and control groups, respectively. Progression of glaucoma was more common in the control patients, occurring in 62% of patients (78/126) compared to 45% (58/129) of treated patients (p = 0.007). Figure 6.3 shows progression over time in all study patients and demonstrates an early separation between study arms that was maintained throughout. Median time to progression (Kaplan-Meier cumulative survival) was 48 months in controls compared to 66 months in treated patients, indicating a delay in progression with treatment. All except one patient that met the criteria for progression did so based on VF criteria (with or without optic disc changes).

The EMGT was the first large RCT (and likely the last) in which patients with manifest glaucoma and elevated IOP could be randomized to observation alone. This study showed that a pressure reduction of 25% in newly diagnosed glaucoma patients decreases the risk of progression, with an NNT of approximately 6 (over approximately 6 years). The EMGT enrollment protocol, which consisted of a population screening program, yielded a study patient sample that was highly reflective of the underlying population with glaucoma. This, combined with the high patient retention and rigorous study protocol, supports the validity and generalizability of the results. While the high rate of glaucoma progression observed may have resulted from overly sensitive VF progression criteria, these criteria were necessary from an ethical standpoint in order to detect the earliest progression in the observation group. Because this effect was nondifferential, it would not be expected to influence the overall study conclusions.

In contrast to the OHTS, which identified conversion to POAG in the majority of patients based on optic disc changes,³ the EMGT found progression by VF criteria in an overwhelming majority of the patients that progressed. This highlights differences between the studies in the relative sensitivity of the VF and optic nerve criteria used to define progression.

During the study period, IOP was unchanged at 16.0 mmHg in the observation group and was reduced to an average of 10.6 mmHg in the treatment group (37% reduction). Using the primary VF endpoints, 35% of the control group and 12% of the treatment group progressed over 5 years (p = 0.0001). Using the “four-of-five” analysis of progression, the difference was smaller, however still significant, with 30% of the control group and 18% of the treatment group showing progression (p = 0.01). It is important to note that when an intent-to-treat analysis was performed, there was no significant difference in progression between the two groups, with 39% and 33% of the observation and treatment groups showing progression, respectively (p = 0.21; Fig. 6.4). However, after adjusting the VF data to account for cataract development, the difference in progression between the two groups in the intent-to-treat analysis became significant, with 27% and 13% of patients in the observation and treatment groups, respectively, showing progression at 5 years (p = 0.0018; Fig. 6.5). Indeed, cataract developed in 14% of the control group and 38% of the treatment group (p = 0.0011), which was attributable to the treatment patients who underwent glaucoma surgery.

The CNTGS, like the EMGT, was an important study as it compared IOP reduction to no treatment in a group of patients with manifest glaucoma. The CNTGS conclusively demonstrated that there is an IOP-dependent component of disease progression in many patients with IOPs in the statistically “normal” range.

The findings of the CNTGS indicate that treating patients with normal-tension glaucoma to achieve an IOP reduction of 30% delays progression of disease, with an NNT of approximately 7 over a 5-year period. However, it is important to highlight that many patients in the observation arm of the CNTGS did not progress. Further, 85 of 230 patients (37%) initially assessed as having glaucomatous discs
and repeatable VF defects failed to show any change in VF and thus were never randomized. Hence, the NNT in standard clinical settings is, in reality, much higher. At the other end of the spectrum, several patients progressed in the treatment arm despite a 30% reduction in IOP, supporting the concept of IOP-independent mechanisms of disease progression in patients with normal-tension glaucoma. These findings highlight the importance of identifying patients at risk for progression to visual disability in order to tailor clinical management accordingly.

The AGIS reported primary results stratified by race. Overall, at the 10-year report, there was a trend toward more VF progression in the TAT sequence in the black subpopulation; however, this did not reach statistical significance. In contrast, the white subpopulation demonstrated greater VF loss with the ATT sequence, which became statistically significant at follow-up years 8 to 10. The proportion of eyes with decreased visual acuity was greater in the TAT sequence in both black
and white subpopulations in the early follow-up period but was not significantly different between the TAT and ATT groups from years 2 through 10.

While the objective of AGIS was to compare different sequences of interventional surgical management in advanced glaucoma, the most often quoted and arguably the most sustainably useful data generated from this study relate to IOP control and its relationship to VF progression. AGIS addressed this relationship through two analyses. First, a predictive analysis investigated whether IOP level during early follow-up was predictive of subsequent VF deterioration over the next 7 years. A total of 738 patients were divided into three groups based on their average IOP during the first 18 months of follow-up, and rates of subsequent VF deterioration among each group were compared. Figure 6.6 demonstrates the results of this predictive analysis and shows a considerable difference in the rate of VF progression dependent on early IOP levels. Second, an associative analysis (Fig. 6.7) was designed to determine the effect of elevated IOP measurement frequency on VF progression. Specifically, 586 eyes were divided into four groups defined by the percent of visits with an IOP less than 18 mmHg, and the rates of VF progression were compared. A striking relationship between IOP control and VF stability
was demonstrated, with patients who had an IOP less than 18 mmHg at all visits (group A, mean IOP of 12.3 mmHg) showing no change in mean VF score.

The AGIS also showed that the rate of cataract development was higher in the TAT sequence than in the ATT sequence (56% and 47%, respectively). Additionally, younger age and higher preintervention IOP were risk factors for both ALT and trabeculectomy failure, while diabetes and postoperative complications were associated with trabeculectomy failure. Male gender was found to be a risk factor for bleb encapsulation.

Strengths of AGIS include its large size, long follow-up period, racially heterogeneous patient population, and numerous subgroup analyses. Criticisms of AGIS include the lack of well-defined endpoints for treatment failure as well as the changing use of antimetabolites in filtration surgery during the study period. The types of medications available at the time of AGIS were also much more limited than today, reducing the generalizability of some of the findings.

While AGIS provided some clinically applicable data regarding ALT and trabeculectomy outcomes, the associative and predictive analyses
of IOP control and VF progression have become instrumental in generating IOP targets in advanced glaucoma. Specifically, the associative analysis demonstrated that IOPs consistently below 18 mmHg stabilize the VF in many patients. This subgroup had an average IOP of 12.3 mmHg, which has resulted in many clinicians choosing 12 mmHg as a target in very advanced POAG. However, it is important to note that while there was no change in mean VF score in this subgroup as a whole, individually some patients showed VF deterioration (14.4% had worse VF defect score at 7 years) despite never having an IOP above 18 mmHg. The overall stability of this subgroup’s mean IOP score was due to a counter-balancing improvement in many patients’ VFs.

Table 6.1 shows the mean change in IOP for each drug at peak and trough.

These results support the shift among many clinicians to utilizing prostaglandin analogs as first-line agents in medical monotherapy for POAG and ocular hypertension. However, the results underscore the efficacy of all of the listed drugs. Hence, an assessment of variables such as cost, likelihood of compliance, and side effects is also necessary in making rational therapeutic choices for individual patients.