PHASE 2 STUDY OF THE SAFETY AND EFFICACY OF BRIMONIDINE DRUG DELIVERY SYSTEM (BRIMO DDS) GENERATION 1 IN PATIENTS WITH GEOGRAPHIC ATROPHY SECONDARY TO AGE-RELATED MACULAR DEGENERATION

Abstract

Purpose: 

To evaluate the safety and efficacy of Brimonidine Drug Delivery System (Brimo DDS), a biodegradable intravitreal implant, in the treatment of geographic atrophy (GA) secondary to age-related macular degeneration.

Methods: 

Phase 2, randomized, multicenter, double-masked, 24-month study. Study eyes were treated (Day 1; Month 6 retreatment) with Brimo DDS 132 µg (n = 49), Brimo DDS 264 µg (n = 41), or sham procedure (n = 23). The primary timepoint for efficacy analysis was Month 12.

Results: 

Mean GA area growth at Month 12 was 1.78 mm², 1.59 mm², and 2.19 mm² in the Brimo DDS 132 µg, 264 µg, and sham groups, respectively. Geographic atrophy area growth was consistently smaller with Brimo DDS 132 and 264 µg than sham; between-group differences were significant (P ≤ 0.032) at Month 3. In patients with baseline lesion area ≥6 mm² (two-thirds of patients), GA lesion area and effective radius growth was reduced with Brimo DDS 132 and 264 µg at Month 12 (P ≤ 0.050 vs. sham). Treatment-related adverse events were usually injection procedure–related.

Conclusion: 

Brimo DDS demonstrated a favorable safety profile and reduced GA lesion area growth at Month 3. Lesion growth at Month 12 was reduced in patients with baseline GA lesion area ≥6 mm². The results support Phase 3 development.

In Brief

In this Phase 2 study in patients with geographic atrophy secondary to age-related macular degeneration, an intravitreal, biodegradable implant that provides sustained release of brimonidine demonstrated a favorable safety profile and potential for reducing geographic atrophy lesion growth, particularly in patients with larger geographic atrophy lesions at baseline.

Age-related macular degeneration (AMD) is the leading cause of irreversible moderate or severe visual impairment in people 50 years of age or older worldwide.¹ Geographic atrophy (GA), an advanced form of AMD, presents as a discrete area of retinal pigment epithelium (RPE) loss with associated loss of the overlying photoreceptors. Geographic atrophy is present in at least one eye in more than 50% of individuals with advanced AMD² and accounts for approximately 25% of the severe visual impairment attributed to AMD.³ An estimated 973,000 individuals in the United States had GA in 2000, including 6.9% of the population over the age of 80 years.² Race/ethnicity is a risk factor for AMD, with the incidence of early and advanced AMD highest in white individuals.⁴ Among the white population of the United States, an estimated 160,000 individuals develop GA each year,⁵ and many of these individuals experience progressive and severe loss of vision. In a study of the natural history of GA, the cumulative risk of a 6-line decrease in best-corrected visual acuity (BCVA) at 2 years was 21% and the cumulative risk of legal blindness at 4 years was 27% in patients who presented with BCVA better than 20/50.⁶

Geographic atrophy has a complex pathogenesis. Oxidative and metabolic stress, chronic inflammation, and complement activation all are believed to play a role in the development and progression of GA.⁷ There are currently no approved treatments for GA. A strategy of cyto/neuroprotection to protect the structure and function of the retina has been proposed as a potential approach to therapy for GA⁸ (see Figure, Supplemental Digital Content 1, https://links.lww.com/IAE/B195, which shows a schematic of the proposed pathophysiology of GA and potential approaches to therapy, and Figure, Supplemental Digital Content 2, https://links.lww.com/IAE/B196, which shows the rationale for cyto/neuroprotection). This approach may be particularly useful for degenerative diseases that have a multifactorial etiology, such as AMD. The goal of cyto/neuroprotection in GA is to reduce the rate of GA progression and stabilize loss of visual function by protecting RPE cells and photoreceptors (increasing their resistance to injury) and preventing, or delaying, cell death.

Brimonidine, an alpha2-adrenergic agonist, has demonstrated cytoprotective and neuroprotective activity in cultured cells and a variety of animal models of retinal and optic nerve disease.^9–11 In vitro, pretreatment with brimonidine reduced reactive oxygen species production and improved the viability of human RPE and Müller cells (ARPE-19 and MIO-M1 cell lines) exposed to hydroquinone, a cigarette smoke toxin.⁹ In vivo, topical administration of brimonidine reduced photoreceptor loss in a rat model of blue light phototoxicity,¹⁰ and systemic or topical administration of brimonidine has also been shown to promote the survival of retinal ganglion cells in multiple rodent models of optic nerve injury, including injury caused by optic nerve crush, retinal ischemia, and elevated intraocular pressure (IOP).¹¹ Cyto/neuroprotection by brimonidine was demonstrated in a nonhuman primate model of GA induced by blue light. In this model, intraocular injection of a brimonidine implant reduced hypoautofluorescence (i.e., protected RPE cells) and preserved outer nuclear layer thickness as measured by optical coherence tomography, suggesting an effect on photoreceptors as well (Rajagopalan L, et al. IOVS 2019; 60: ARVO E-Abstract 2,993). Furthermore, in the Low-Pressure Glaucoma Treatment Study, brimonidine and timolol eye drops were similarly effective in lowering IOP, but the proportion of patients who had visual field progression during treatment was smaller in the brimonidine group than in the timolol group (P = 0.001), suggesting a possible neuroprotective effect of brimonidine in humans independent of its IOP-lowering effects.¹²

An intravitreal implant containing brimonidine in a poly-(d,l-lactide) biodegradable polymer matrix, Brimonidine Drug Delivery System (Brimo DDS; Allergan plc, Dublin, Ireland), has been developed for potential treatment of GA. Brimo DDS Generation 1 (Gen 1) contains a dose of 132 µg or 264 µg brimonidine (formulated as 200 µg or 400 µg of brimonidine tartrate). The implant is administered via intravitreal injection using a 22-gauge needle and a proprietary applicator system. Brimonidine diffuses out of the implant into the vitreous humor over a period of several months as the polymer matrix degrades.

A Phase 2 study was conducted to evaluate the safety and effects of two administrations of Brimo DDS Gen 1 containing 132 or 264 µg brimonidine on retinal structure and visual function in patients with GA secondary to AMD. A secondary study objective was to use the study data to develop a better understanding of the disease and help design future clinical trials in GA.

Methods

This Phase 2 study was a randomized, multicenter (25 clinical sites in 7 countries), double-masked, dose-ranging, parallel-group, sham-controlled evaluation of Brimo DDS Gen 1 in patients with GA secondary to AMD. The study was conducted in compliance with the International Conference on Harmonisation Guideline for Good Clinical Practice. An institutional review board or ethics committee approved the study at each investigational site, and all participating patients provided written informed consent. The study is registered with the identifier NCT00658619 at clinicaltrials.gov.

Patient Selection

Eligible patients were 50 years of age or older with GA (defined as partial or complete depigmentation of the RPE on assessment with digital stereoscopic color fundus photography) in both eyes that was attributed to AMD. In both eyes, at least one site of GA had to be between 0.75 and 12 disc areas (2.02–32.28 mm²) and reside at least in part within 1,500 µm of the foveal center at screening. All sites of macular GA had to be visible within 30° images centered on the fovea (Field 2 macular photographs) without extension to the margin of the color fundus images. Other key inclusion criteria included BCVA letter score at screening, measured with the Early Treatment Diabetic Retinopathy Study method, between 70 and 35 letters (20/40 and 20/200 Snellen equivalent) in the study eye and at least 25 letters (20/320 Snellen equivalent) in the fellow eye. The ocular media was required to be clear enough for good quality fundus photographs in both eyes, allowing visualization of areas of GA.

Key exclusion criteria included history or evidence of choroidal neovascularization in either eye based on fluorescein angiography at the screening visit, known allergy to brimonidine, uncontrolled systemic disease, any intravitreal injection within the 6 months before screening in either eye, periocular injection of corticosteroid within the 4 months before screening in either eye, infective condition in either eye, intraocular surgery or laser in either eye within 6 months before Day 1, and any condition that in the investigator's opinion could put the patient at significant risk, confound the study results, or interfere significantly with the patient's participation in the study.

If both eyes met the study eye eligibility criteria, the worse eye, based on visual acuity at Day 1 and the area of depigmentation of the RPE at the screening visit, was selected as the study eye.

Visits, Randomization, and Intervention

Study visits were scheduled at screening (Day −21 to −1), Day 1 (treatment visit), and Months 1, 3, 4, 5, 6 (retreatment visit), 7, 8, 9, 10, 11, 12, 18, and 24/exit (see Figure, Supplemental Digital Content 3, https://links.lww.com/IAE/B197, which shows a schematic of the study design). In addition, safety visits were scheduled on Day 7 after each treatment, and follow-up telephone calls to collect adverse event reports were made on Days 2 and 14 after each treatment and at Months 15 and 21.

On Day 1, patients were randomized in a 2:2:1 ratio to receive treatment in the study eye with Brimo DDS 132 µg, Brimo DDS 264 µg, or sham procedure on Day 1 and Month 6. The fellow eyes in all treatment groups received sham procedure on Day 1 and Month 6. Brimo DDS was administered intravitreally through the pars plana using standard sterile technique and a single-use 22-gauge applicator system (see Figure, Supplemental Digital Content 4, https://links.lww.com/IAE/B198, which shows photographs of a Brimo DDS implant and applicator). For the sham procedure, a needleless applicator was pressed against the temporal bulbar conjunctiva of the eye. Topical and subconjunctival anesthetics were used before the Brimo DDS and sham procedures, and a broad-spectrum topical antibiotic was administered before and for 3 days after the procedures. All patients were masked to the treatment received. Patients were followed through Month 24.

Assessments and Outcome Measures

The primary efficacy measure was the change in the GA lesion area, based on stereoscopic fundus photography, as measured from digital images by the central reading center (Doheny Image Reading Center, Los Angeles, CA). Stereoscopic fundus photography and fluorescein angiography were performed at screening (baseline assessment) and Months 3, 6, 9, 12, 15, 18, and 24. The primary endpoint was the change in GA lesion area from baseline at Month 12.

Secondary efficacy measures for each eye included BCVA measured using the Age-Related Eye Disease Study modification of the Early Treatment Diabetic Retinopathy Study method at 4 m,¹³ contrast sensitivity measured using a Pelli-Robson contrast sensitivity chart at 1 m with +0.75 diopters added to the refraction determined for distance BCVA, and reading speed measured using Bailey-Lovie word charts at 25 cm with a +4.00-diopters reading add worn over the distance refraction. Analysis of the secondary efficacy measures evaluated the change from baseline, with baseline BCVA measured on Day 1 before study treatment, and baseline contrast sensitivity and reading speed measured at screening. All site personnel responsible for performing the secondary efficacy measures, as well as the reading center personnel involved in the assessment of GA, were masked to the study treatment.

Key safety measures included adverse events, biomicroscopy, indirect ophthalmoscopy, and IOP. An adverse event was determined to be treatment-related if in the judgment of the investigator, there was a reasonable possibility that the adverse event may have been caused by the study treatment (i.e., the implant, applicator, or insertion procedure).

Data Analysis

Statistical analysis was performed using SAS version 9.2 or 9.3 (SAS Inc, Cary, NC) and an alpha level of 0.05. The preplanned analyses of primary and key secondary efficacy parameters (change from baseline in GA lesion area and BCVA) used the intent-to-treat (ITT) population of all randomized patients, with patients with no baseline data excluded from analysis, and missing data imputed using the last-observation-carried-forward method. Comparisons between study and fellow eyes used paired t-tests, and comparisons of study eyes between treatment groups used two-sample t-tests that incorporated the Welch approximation when the variances were unequal. There was no adjustment for multiple comparisons.

Post-hoc, supplementary analysis of change from baseline in GA lesion area used observed values in the modified ITT population of all randomized patients with data at baseline and at least one follow-up visit, and mixed-effects models for repeated measures (MMRMs). In some of these models, the GA lesion area data were transformed to reduce skewness. Transformation of data used the square root of the area of GA¹⁴ or the effective radius of the area of GA, defined as $\sqrt{\raisebox{1ex}{$\left(\text{GA lesion area}\right)$}\!\left/ \!\raisebox{-1ex}{$\Pi $}\right.}$ and representing the radius of a circular GA lesion with the same total area. Sensitivity analysis of the change in GA lesion size from baseline in study eyes used square root–transformed observed GA area data and an MMRM model that included fixed effects of treatment, visit, square root-transformed baseline area as a covariate, the treatment-by-visit interaction, and the square root-transformed baseline area-by-visit interaction. An additional exploratory MMRM analysis used effective-radius–transformed observed data and evaluated the effects of Brimo DDS on GA lesion progression in patients with baseline GA area of <6 mm² (lowest tertile of baseline lesion area) or ≥6 mm² (combined middle and highest tertiles of baseline lesion area). The MMRM model included fixed effects of treatment, visit, baseline lesion area stratum (<6 or ≥6 mm²), baseline effective radius of GA lesion area as a covariate, and the treatment-by-visit, effective-radius–transformed baseline GA area-by-visit, treatment-by-stratum, and treatment-by-stratum-by-visit interactions. Each MMRM analysis assumed an unstructured covariance matrix.

Safety analyses used the safety population of all randomized patients who received study treatment. Adverse events were coded using the Medical Dictionary for Regulatory Activities version 10.0 nomenclature. An adverse event was determined to be related to treatment if in the judgment of the investigator there was a reasonable possibility that it may have been caused by the study treatment.

The planned enrollment was approximately 110 patients with an anticipated 12-month dropout rate of 10%. A sample size of 40 patients in the Brimo DDS 264-µg group was estimated to provide 80% power to detect a reduction in lesion growth of 0.98 mm² (approximately 38%) in the study eye relative to the fellow eye at Month 12, assuming a mean increase in the GA area from baseline of 2.60 mm² at Month 12 in the fellow eye with a SD of the paired-eye difference of 2.17 mm² (based on data from untreated eyes¹⁵), and a paired t-test with a two-sided alpha level of 0.05. The study was not powered for statistical comparisons between treatment groups. Multiple P values are presented as less than or equal to the largest exact P value.

Results

A total of 113 patients were randomized to treatment in the study eye with Brimo DDS 132 or 264 µg or sham procedure. Baseline demographic and study eye characteristics were similar across treatment groups (Table 1). The mean age of the study population was 76.8 years, 88.5% of patients were white, and 59.3% were female. Mean (SD) GA lesion size was 11.28 (7.81) mm² in study eyes (n = 104) and 11.52 (8.72) mm² in fellow eyes (n = 104). The GA lesion in the study eye typically had a subfoveal location.

Table 1. - Baseline Patient Characteristics (ITT Population)

Characteristic	Brimo DDS 132 µg, (n = 49)	Brimo DDS 264 µg, (n = 41)	Sham, (n = 23)
Age, mean (SD), years	77.0 (9.1)	75.6 (8.8)	78.4 (5.8)
Range	53–94	55–90	65–86
Gender, n (%)
Female	28 (57.1)	27 (65.9)	12 (52.2)
Male	21 (42.9)	14 (34.1)	11 (47.8)
Race/ethnicity, n (%)
White	41 (83.7)	38 (92.7)	21 (91.3)
Asian	6 (12.2)	3 (7.5)	1 (4.3)
Hispanic	1 (2.0)	0 (0.0)	1 (4.3)
Other (Armenian)	1 (2.0)	0 (0.0)	0 (0.0)
BCVA in study eye, mean (SD), ETDRS letters	52.1 (13.9)	54.8 (12.9)	53.7 (10.7)
Snellen equivalent	∼20/100	∼20/80	∼20/80
GA Lesion area in study eye, mean (SD), mm2*	12.19 (8.20), n = 46	10.99 (6.98), n = 38	9.76 (8.46), n = 20
Range	0.65–34.70	0.37–27.78	1.15–33.63
GA Lesion area in fellow eye, mean (SD), mm2*	11.70 (8.53), n = 44	11.16 (8.63), n = 40	11.82 (9.70), n = 20
Range	0.40–35.25	0.46–39.21	1.52–35.24
Subfoveal location of GA in study eye, n (%)
Yes (definite or probable)	33 (67.3)	33 (80.5)	18 (78.3)
No	14 (28.6)	6 (14.6)	4 (17.4)
Not gradable or not reported	2 (4.1)	2 (4.9)	1 (4.3)
Lesion complexity, n (%)
Unifocal	25 (51.0)	22 (53.7)	10 (43.5)
Multifocal	21 (42.9)	16 (39.0)	10 (43.5)
Not reported	3 (6.1)	3 (7.3)	3 (13.0)

^*GA lesion area in all eyes with screening data.

Brimo DDS, brimonidine drug delivery system; ETDRS, Early Treatment Diabetic Retinopathy Study.

The study completion rate was higher in the sham group (91.3%) than in the Brimo DDS 132-µg and 264-µg groups (75.5% and 75.6%, respectively) because of a larger number of discontinuations due to adverse events in the Brimo DDS groups (see Table, Supplemental Digital Content 5, https://links.lww.com/IAE/B199, which shows patient disposition). However, none of the adverse events leading to early study exit were considered to be related to the study treatment. Adverse events leading to patient discontinuations included acute myocardial infarction, cerebrovascular accident, choroidal neovascularization, retinal hemorrhage, macular edema, myocardial infarction, and road traffic accident in the Brimo DDS 132-µg group; renal failure, urosepsis, acute myocardial infarction, cerebrovascular accident, congestive cardiac failure, and pulmonary vascular disorder in the Brimo DDS 264-µg group; and hip fracture and laryngeal cancer in the sham group.

Efficacy Outcomes

During the 24-month study period, progression in the area of GA was observed in both study eyes and sham-treated fellow eyes in all treatment groups (Table 2). The mean change in GA lesion area from baseline was statistically significant (P ≤ 0.002) at all follow-up visits, except in study eyes in the Brimo DDS 132-µg group at Month 3, when there was no statistically significant change in GA lesion area from baseline (P = 0.235). There were no statistically significant paired-eye differences between study and fellow eyes in baseline GA lesion area or in change from baseline in GA lesion area at any follow-up visit, except in the sham group at Month 3, when the mean change in GA lesion area from baseline was greater in the sham-treated study eyes than in the sham-treated fellow eyes (P = 0.010). At Month 12, the mean change in GA lesion area from baseline was 1.78 mm², 1.59 mm², and 2.19 mm² in study eyes and 1.92, 1.47, and 1.88 mm² in fellow eyes in the Brimo DDS 132 µg, Brimo DDS 264 µg, and sham groups, respectively.

Table 2. - Change From Baseline in Area of Geographic Atrophy in Study and Fellow Eyes (ITT Population)^*

Visit	Mean (SD) Change From Baseline in GA Lesion Area, mm2
	Brimo DDS 132 µg, Study Eye (n = 46)	Brimo DDS 132 µg, Fellow Eye (n = 44)	Brimo DDS 264 µg, Study Eye (n = 38)	Brimo DDS 264 µg, Fellow Eye (n = 40)	Sham, Study Eye (n = 20)	Sham, Fellow Eye (n = 20)
Month 3	0.21 (1.17)†	0.42 (0.58)	0.32 (0.59)‡	0.47 (0.63)	0.71 (0.66)§	0.34 (0.40)
Month 6	1.01 (1.02)	1.10 (1.20)	0.85 (1.12)	0.81 (0.90)	1.08 (0.97)	0.73 (0.87)
Month 9	1.41 (1.14)	1.50 (1.32)	1.13 (1.21)	1.13 (1.03)	1.51 (1.28)	1.29 (1.09)
Month 12	1.78 (1.39)	1.92 (1.64)	1.59 (1.33)	1.47 (1.38)	2.19 (2.06)	1.88 (1.33)
Month 18	2.50 (1.82)	2.99 (2.46)	2.31 (1.80)	2.08 (1.91)	3.00 (2.51)	2.74 (1.74)
Month 24	3.52 (2.56)	3.75 (2.70)	3.08 (2.48)	2.87 (2.80)	3.85 (3.01)	3.36 (2.14)

^*Eyes with baseline GA lesion area data (from the screening visit) as determined by the central reading center were included in the analysis. Missing values during follow-up were imputed with the last-observation-carried-forward method.

^†P = 0.032 versus sham study eye.

^‡P = 0.028 versus sham study eye.

^§P = 0.010 versus sham fellow eye.

Brimo DDS, brimonidine drug delivery system.

Figure 1 shows the mean change from baseline in GA lesion area in the study eye at each follow-up visit for the ITT population using the last observation carried forward as the imputation method for missing values. At Month 3, the growth in GA lesion area was significantly smaller in the Brimo DDS 132-µg group (P = 0.032) and the Brimo DDS 264-µg group (P = 0.028) compared with the sham group. At subsequent time points through Month 24, the mean values for growth in GA lesion area were consistently lower in both Brimo DDS groups than in the sham group, but the differences were not statistically significant. The mean (SEM) change from baseline in GA lesion area in study eyes at Month 12 (primary endpoint) was 1.78 (0.20) mm² in the Brimo DDS 132-µg group, 1.59 (0.22) mm² in the Brimo DDS 264-µg group, and 2.19 (0.46) mm² in the sham group. These results indicate a reduction in GA progression rate from baseline to Month 12 of 19% with Brimo DDS 132 µg and 28% with Brimo DDS 264 µg compared with sham.

Fig. 1.: Change from baseline in GA lesion area in the study eye. Data shown are the mean ± SEM in the ITT population with last observation carried forward for missing values. Arrows indicate timing of study treatments. n is 46 for Brimo DDS 132 µg, 38 for Brimo DDS 264 µg, and 20 for sham. At Month 12, the differences between treatment groups were not statistically significant (P = 0.420 for Brimo DDS 132 µg vs. sham; P = 0.246 for Brimo DDS 264 µg vs. sham). *P = 0.028 and †P = 0.032 versus sham (independent samples t-test).

The Figure in Supplemental Digital Content 6, https://links.lww.com/IAE/B200, shows the results of an MMRM analysis of GA lesion area conducted using observed data in the modified ITT population. In general, the results were similar to those presented in Figure 1, with Brimo DDS reducing the progression rate over time, but there were no statistically significant differences between groups (P ≥ 0.077). At the primary endpoint, Brimo DDS 132 and 264 µg reduced the progression rate by 19% and 23%, respectively. At Month 24, the difference between the Brimo DDS 264-µg group and the sham group was 22%.

A sensitivity analysis was conducted using the square root transformation strategy¹⁴ to evaluate growth of GA lesion area in study eyes (see Figure, Supplemental Digital Content 7, https://links.lww.com/IAE/B201, which shows the results of the sensitivity analysis). The analysis used observed data in an MMRM model that adjusted for baseline lesion size. The sensitivity analysis confirmed that GA lesion growth was consistently reduced in the Brimo DDS groups compared with the sham group, but none of the differences between treatment groups were statistically significant (P ≥ 0.089). At Month 12, square root–transformed GA lesion growth from baseline was reduced by 8% with Brimo DDS 132 µg and by 11% with Brimo DDS 264 µg compared with sham. At Month 24, square root–transformed GA lesion growth from baseline was reduced by 20% with Brimo DDS 264 µg compared with sham (least squares mean [SE] change from baseline of 0.52 [0.05] mm compared with 0.65 [0.07] mm). In both the primary analysis and the sensitivity analysis, dose-dependent reductions in lesion growth were observed after Month 3, with larger effects on lesion growth observed in the Brimo DDS 264-µg group than in the Brimo DDS 132-µg group.

In sham-treated study eyes, there was a strong association between baseline GA lesion area and the change from baseline in GA lesion area at Month 12 (r_s = 0.699, P < 0.001) (Figure 2). This association between baseline GA lesion area and the Year 1 rate of GA progression was dose-dependently reduced by Brimo DDS (Figure 2). The Spearman correlation coefficient for the relationship between baseline GA lesion area and change from baseline in GA lesion area at Month 12 was 0.279 (P = 0.089) and 0.161 (P = 0.395) for Brimo DDS 132 µg- and Brimo DDS 264 µg-treated eyes, respectively. In addition, the slope of the linear regression for the sham group was +0.16 and significantly different from zero (P = 0.001), whereas the slope of the linear regression was reduced in a dose-dependent manner to +0.033 and −0.017 with Brimo DDS 132 and 264 µg, respectively, and was not significantly different from zero (Figure 2). The linear regression of the data suggested the possibility that Brimo DDS effects on GA lesion progression are more apparent in patients with larger GA lesions at baseline, and therefore the effects of Brimo DDS can be observed more readily in patients with a faster progression rate. Overall, these data suggest that a steeper rate of GA progression in the study patient population results in a faster ability to detect a study drug–induced change.

Fig. 2.: Relationship between baseline GA area and change from baseline GA area at Month 12 in the study eye of patients in the sham, Brimo DDS 132-µg, and Brimo DDS 264-µg treatment groups. Correlation coefficients (r_s) shown are from Spearman rank-order correlation analysis. Lines indicate the linear regressions with shading showing the 95% confidence intervals. The slope of the linear regression was +0.16 for sham (P = 0.001 vs. zero), +0.033 for Brimo DDS 132 µg (P = 0.227 vs. zero), and −0.017 for Brimo DDS 264 µg (P = 0.640 vs. zero).

A subsequent exploratory MMRM analysis evaluated study eye GA area change from baseline in patients with baseline GA area in the lowest tertile or in the combined middle and highest tertiles. Baseline GA area ranged from 0.37 mm² to 5.97 mm² in the lowest tertile (<6 mm²) and from 6.04 mm² to 34.7 mm² in the combined middle and highest tertiles (≥6 mm²). In patients with baseline GA lesion area ≥6 mm² (two-thirds of all patients), mean GA area growth was consistently reduced in both Brimo DDS-treated groups compared with the sham group (Figure 3). The change from baseline in GA area was significantly reduced in the Brimo DDS 132-µg group compared with the sham group at Months 3 (P = 0.029) and 12 (P = 0.040) and was significantly reduced in the Brimo DDS 264-µg group compared with the sham group at Month 12 (P = 0.029). At the Month 12 primary time point, growth of GA area was reduced by 32% with Brimo DDS 132 µg and by 36% with Brimo DDS 264 µg compared with sham. In contrast, Brimo DDS treatment was not demonstrated to be effective in patients with baseline GA lesion area <6 mm² (Figure 3B). Among these patients, there were no significant differences in GA lesion area growth between the Brimo DDS groups and the sham group. These results are consistent with the idea that patients with a slower progression rate would require longer follow-up to detect a benefit of Brimo DDS treatment.

Fig. 3.: Change from baseline in GA area in the study eye in (A) the lowest tertile (<6 mm²) and (B) the combined middle and highest tertiles (≥6 mm²) of baseline GA lesion area. Data shown are least-squares means ± SE from a mixed-effects repeated-measures model using observed values (with no imputation for missing values) and factors of treatment, visit, baseline GA lesion area stratum (<6 or ≥6 mm²), treatment-by-visit interaction, treatment-by-strata interaction, the baseline GA area-by-visit interaction, the treatment-by-strata-by-visit interaction, and the baseline GA lesion area as a covariate. Arrows indicate timing of study treatments. *P ≤ 0.040 versus sham.

Results were similar in an MMRM analysis that evaluated GA effective radius change from baseline in the study eye in patients with baseline GA area of <6 or ≥6 mm². Mean GA effective radius growth was consistently reduced in both Brimo DDS-treated groups compared with the sham group in patients with baseline GA lesion area ≥6 mm² (Figure 4). The change from baseline in GA effective radius was significantly reduced in the Brimo DDS 132-µg group compared with the sham group at Months 3 (P = 0.041), 12 (P = 0.050), and 24 (P = 0.047) and was significantly reduced in the Brimo DDS 264-µg group compared with the sham group at Month 12 (P = 0.049). At the Month 12 primary time point, growth of GA effective radius was reduced by 29% with Brimo DDS 132 µg and by 31% with Brimo DDS 264 µg compared with sham. Brimo DDS treatment was not demonstrated to be effective in patients with baseline GA lesion area <6 mm² (Figure 4). Among patients with baseline GA lesion area <6 mm², there were no significant differences in GA effective radius change from baseline between Brimo DDS 264 µg and sham, whereas the change from baseline in GA effective radius was significantly larger in the Brimo DDS 132-µg group compared with the sham group at Month 12 (P = 0.026) (Figure 4B).

Fig. 4.: Change from baseline in effective radius of GA in the study eye in the combined middle and highest tertiles (A) and lowest tertile (B) of baseline GA lesion area. Data shown are least-squares means ± SE from a mixed-effects repeated-measures model using observed values (with no imputation for missing values) and factors of treatment, visit, baseline GA lesion area stratum (<6 or ≥6 mm²), treatment-by-visit interaction, treatment-by-strata interaction, the effective radius of baseline GA area-by-visit interaction, treatment-by-strata-by-visit interaction, and the effective radius of baseline GA area as a covariate. Arrows indicate timing of study treatments. *P ≤ 0.050 versus sham.

There was no consistent improvement or worsening over time in BCVA, contrast sensitivity, or reading speed in study or fellow eyes in any treatment group. Changes from baseline in these parameters were small with no meaningful differences between treatment groups (see Table, Supplemental Digital Content 8, https://links.lww.com/IAE/B202, which shows baseline and change from baseline values for the visual function parameters at Month 12 in study and fellow eyes in each treatment group).

Safety Outcomes

The percentage of patients with one or more adverse events was comparable across treatment groups (Table 3). However, ocular adverse events were more frequent in both Brimo DDS groups compared with the sham group (Table 3). Most treatment-related ocular adverse events in the Brimo DDS groups were attributed to the injection procedure (Table 3). The most common ocular adverse events were conjunctival hemorrhage and conjunctival hyperemia (Table 4), which were generally related to the injection procedure. Choroidal neovascularization was reported in 3 (6.3%) patients in the Brimo DDS 132-µg group, 1 (2.5%) patient in the Brimo DDS 264-µg group, and 0 (0%) patients treated with sham. There were no adverse event reports of endophthalmitis in any treatment group, and no nonocular treatment-related adverse events were reported in any treatment group.

Table 3. - Incidence of Adverse Events (Safety Population)

Parameter, n (%)	Brimo DDS 132 µg (n = 48)	Brimo DDS 264 µg (n = 40)	Sham (n = 23)
Any adverse event	44 (91.7)	35 (87.5)	20 (87.0)
Ocular*	37 (77.1)	31 (77.5)	12 (52.2)
Study eye	32 (66.7)	29 (72.5)	10 (43.5)
Fellow eye	21 (43.8)	17 (42.5)	9 (39.1)
Nonocular	29 (60.4)	28 (70.0)	18 (78.3)
Treatment-related adverse event	19 (39.6)	11 (27.5)	2 (8.7)
Ocular*	19 (39.6)	11 (27.5)	2 (8.7)
Treated eye	17 (35.4)	11 (27.5)	2 (8.7)
Attributed to injection procedure	16 (33.3)	10 (25.0)	2 (8.7)
Attributed to applicator	1 (2.1)	0	0
Attributed to drug/implant	1 (2.1)	3 (7.5)	0
Fellow eye	6 (12.5)	4 (10.0)	0
Nonocular	0	0	0

^*An adverse event was determined to be ocular if the investigator indicated it was associated with the eye or if it was coded to a MedDRA preferred term in the System Organ Class of Eye Disorders.

Brimo DDS, brimonidine drug delivery system.

Table 4. - Ocular Adverse Events Reported in 2 or More Patients in Any Treatment Group (Safety Population)

Adverse Event (MedDRA Preferred Term)	No. (%) of Patients
	Brimo 132 µg (n = 48)	Brimo 264 µg (n = 40)	Sham (n = 23)
Conjunctival hemorrhage	22 (45.8)	11 (27.5)	2 (8.7)
Conjunctival hyperemia	8 (16.7)	3 (7.5)	3 (13.0)
Retinal hemorrhage	6 (12.5)	4 (10.0)	2 (8.7)
Cataract	3 (6.3)	3 (7.5)	2 (8.7)
Punctate keratitis	1 (2.1)	6 (15.0)	1 (4.3)
Visual acuity reduced	3 (6.3)	3 (7.5)	1 (4.3)
Vitreous detachment	3 (6.3)	2 (5.0)	2 (8.7)
Eye pain	0	5 (12.5)	1 (4.3)
Vitreous floaters	2 (4.2)	4 (10.0)	0
Choroidal neovascularization	3 (6.3)	1 (2.5)	0
Macular degeneration	1 (2.1)	3 (7.5)	0
Vitreous hemorrhage	2 (4.2)	2 (5.0)	0
Corneal edema	0	2 (5.0)	1 (4.3)
Posterior capsule opacification	0	3 (7.5)	0
Blepharitis	2 (4.2)	0	0
Conjunctival edema	0	2 (5.0)	0
Eyelids pruritus	0	2 (5.0)	0

Brimo DDS, brimonidine drug delivery system.

The incidence of serious adverse events was similar across treatment groups. There were 9 deaths during the study, four in each of the Brimo DDS groups and one in the sham group. None were considered to be related to the study treatment. The only serious adverse event considered by the investigator to be potentially related to treatment was a severe decrease in visual acuity (from 20/50 to 20/250) in the study eye of a patient in the Brimo DDS 264-µg group. The event occurred 14 months after the last administration of Brimo DDS and resolved spontaneously within 10 days, and therefore, was unlikely to be related to study treatment.

Findings of other safety assessments including IOP, biomicroscropy, and ophthalmoscopy were unremarkable. No safety concerns were indicated.

Discussion

Brimo DDS (Gen 1) was well tolerated in this study and demonstrated an ability to reduce GA lesion growth in patients with GA secondary to AMD. Preplanned and post-hoc analyses of GA lesion area consistently showed a trend for a reduced rate of GA lesion growth after Brimo DDS treatment. Although the study was not powered to show statistically significant differences between treatment groups, in the preplanned primary analysis (mean change in GA lesion area from baseline in the ITT population with last observation carried forward for missing values), GA lesion growth was significantly reduced in both Brimo DDS groups compared with sham at Month 3. At Month 12 (primary endpoint), GA lesion growth was reduced by 19% and 28% compared with sham in the Brimo DDS 132-µg and 264-µg groups, respectively; the effects at Month 12 were not statistically significant.

Differences between treatment arms in patient dropout rates were evident by Month 6 of the study. Because these differences were due to the occurrence of adverse events that were not related to the study treatment, they would not be expected to affect conclusions drawn from the preplanned primary efficacy analysis, which used last observation carried forward for missing values. Nonetheless, all post hoc analysis of the data used MMRM modeling and observed values with no imputation for missing values.

The MMRM sensitivity analysis that was performed used the square root of the GA area. This recently has become an accepted methodology for analysis of GA lesion growth, because it reduces skewness of the data and compensates for potentially different rates of progression at different stages of the disease.^14,16,17 Results of the sensitivity analysis were confirmatory in showing consistently lower rates of GA lesion progression in the Brimo DDS groups compared with the sham group.

There was a wide range of GA lesion sizes within each treatment group at baseline. Consistent with previous studies showing a relationship between baseline GA lesion size and the rate of lesion growth,^18–20 a strong positive correlation was observed between baseline GA area and the change from baseline in GA lesion area at Month 12 in sham-treated study eyes. This relationship was dose-dependently reduced by Brimo DDS treatment. The reduction in the association between baseline lesion area and progression rate at Month 12, and the reduction in the slopes of the regression lines, seemed to be due to reduced lesion growth in Brimo DDS-treated patients with large lesions at baseline, that is, those who generally have a faster rate of disease progression. These results were not unexpected; given the precision and reliability of GA size measurements, it would be difficult to detect any treatment-related reduction in lesion growth rate when the growth rate is slow. Post-hoc analysis using GA lesion effective radius indicated that Brimo DDS 132 and 264 µg reduced the growth of GA lesions from baseline at Month 12 by 29% and 31%, respectively, in patients with baseline GA lesion area of ≥6 mm². These effects were statistically significant. In contrast, in the cohort of patients with baseline GA lesion area smaller than 6 mm², lesion growth at Month 12 was greater with Brimo DDS 132 µg than with sham. The absence of a similar dose-dependent effect with Brimo DDS 264 µg suggests that the between-group difference in lesion growth in the cohort with baseline GA lesion area <6 mm² most likely occurred by chance. Interestingly, the effect size for Brimo DDS increased progressively as baseline GA lesion area increased from Tertile 1 to Tertile 3. In the third tertile (i.e., lesion area at baseline 13–32 mm²) GA lesion area growth at Month 12 was reduced by 38% and 59% in the Brimo DDS 132-µg and 264-µg groups, respectively, and the percent reduction in lesion effective radius growth was 26% and 47%, respectively (data not shown).

In our study, the progression rate in sham-treated patients was ∼2.2 mm²/year (and ∼0.32 mm/year when looking at the square root of the area), and Brimo DDS reduced the GA progression rate at Month 12. The effect at Month 12 was not statistically significant, but the study was not powered for parallel-group comparisons. The FILLY study that evaluated APL-2, a complement component C3 inhibitor, also showed a reduction in the rate of progression of GA at Month 12. The data reported were the change from baseline in the square root transformation of lesion area and did not include the growth of GA lesions expressed in area units. The growth in lesions in square root units in sham-treated patients in the FILLY study was 0.35 mm at Month 12,²¹ similar to the 0.32 mm seen in our study. Therefore, we can assume that progression rate in the patient populations in these studies was similar. The data showing reduction of GA growth at Month 12 highlight that in these two studies the progression rate observed was sufficiently fast to allow identification of potential treatment effects. Two additional studies seem to have evaluated populations with progression rates in the range of our Phase 2a study and the FILLY study. In the COMPLETE study, which evaluated eculizumab, a systemic inhibitor of complement component C5, the rate of change in square root of GA lesion area in placebo-treated patients was 0.37 mm over 12 months.²² In the Phase 3 studies Chroma and Spectri, which evaluated complement factor D inhibitor lampalizumab for treatment of GA secondary to AMD, the mean baseline GA lesion size in sham-treated patients was smaller than in the overall population of our study, and the progression rate in sham-treated patients was 2.0 mm² over 48 weeks.²³ These two studies did not show an alteration in the GA progression rate with treatment. Lampalizumab blocks only the alternative complement cascade, and this has been postulated as a potential explanation for its lack of efficacy.²⁴ In contrast, eculizumab is expected to block membrane attack complex formation and, therefore, would be expected to impair complement activation pathways similarly to APL-2. Nevertheless, the COMPLETE study of eculizumab did not show an effect of treatment on the GA progression rate, suggesting that other potential components of the complement cascade upstream of C5 may be involved in the pathogenesis of GA. These upstream components are inhibited by APL-2.

In the Phase 3 study (SEATTLE) of emixustat, an inhibitor of the RPE65 visual cycle isomerase, mean baseline GA lesion area was smaller than that in our Phase 2a study, and the progression rate was 1.69 mm²/year.²⁵ Similarly, in a Phase 2 study of intravenous GSK933776, a humanized monoclonal antibody directed against the N-terminal amino acids of beta-amyloid, the baseline mean lesion area was 8.1 mm² and the progression rate was 1.74 mm²/year.²⁶ The progression rate in sham-treated patients of our Phase 2a study was ∼2.2 mm² over 12 months overall, and when we consider participants with baseline GA lesion area of ≥6 mm², the progression rate increased to ∼3.2 mm² over 12 months. The observation that Brimo DDS effects were seen more readily in patients with larger baseline lesions, which tend to progress more rapidly, suggests that to identify a reduction in progression after 12 months, the sham group may have to progress at a rate of at least 2 mm²/year. In addition, an effect may be more readily observed (i.e., sooner) when the progression rate reaches 3 mm² to 4 mm²/year.

Brimo DDS demonstrated a favorable safety profile in this study. There were no unexpected safety findings, and most treatment-related adverse events were attributed to the injection procedure. Although repeat administration of Brimo DDS was shown to be well tolerated over 24 months, additional studies with multiple repeat treatments will be needed to confirm the long-term safety of Brimo DDS therapy.

The main limitation of the study was that the study was not powered to show statistically significant differences between the study treatments because of the small sample size. Although the study was powered for comparisons between study and fellow eyes, the US Food and Drug Administration has clarified that it does not consider fellow eyes to be adequate comparator controls in clinical trials of GA progression.²⁷ In addition, paired-eye comparisons of study eyes and fellow eyes were likely confounded because there were different selection criteria for study and fellow eyes, and the selection of the study eye was not random. Finally, the study design included a single Brimo DDS repeat treatment with an interval between treatments of 6 months. Brimo DDS 132 and 264 µg significantly reduced the growth in lesion area from baseline at Month 3, but more frequent dosing may be needed for sustained effects on GA lesion progression.

A secondary objective of this study was to use the study data to improve the design of future clinical trials in GA, and some insights were gained regarding the patient selection and types of analysis that may be most appropriate for evaluation of effects of therapy in GA in future studies. Inclusion of patients with larger lesions (and correspondingly faster rates of GA progression) may enable evaluation of efficacy at earlier time points. Square root or effective radius transformation of the data was applied to reduce the skewness seen in the raw data before applying the MMRM model. A potential advantage of use of effective-radius–transformed data, as compared with values obtained from square root transformation of raw GA lesion area, is more intuitive interpretability of changes from baseline in lesion size. Larger datasets may provide more ability to discriminate differences between these measures and potential advantages of data transformation.

In summary, in this Phase 2 study, Brimo DDS (Gen 1) treatment was well tolerated and seemed to have beneficial effects on GA lesion growth in patients with GA secondary to AMD, particularly in patients with GA lesions 6 mm² or larger at baseline. Additional studies, powered to detect a clinically meaningful effect with sound protection from Type I and II errors, are needed to confirm inhibition of GA lesion progression by Brimo DDS. A larger Phase 2b study (BEACON, clinicaltrials.gov NCT02087085) evaluated a modified formulation of the implant, Brimo DDS Gen 2 400 µg, administered at 3-month intervals in patients with GA secondary to AMD (WR Freeman et al. Presented at the 2018 American Academy of Ophthalmology Retina Subspecialty Day, October 26, 2018, Chicago, IL). The BEACON study was terminated early because the rate of GA lesion progression was slow (∼1.6 mm²/year) in the enrolled population, which had a mean baseline GA lesion area of ∼5 mm². Nevertheless, the implant significantly reduced lesion growth at the final (Month 30) time point. On the basis of the promising results from the Phase 2 and Phase 2b studies, Brimo DDS is continuing in Phase 2/3 development.