Pharmacokinetic/pharmacodynamic analysis of geographic atrophy lesion area in patients receiving pegcetacoplan treatment or sham

Abstract

Pegcetacoplan is a complement C3/C3b inhibitor indicated for the treatment of geographic atrophy (GA). A population pharmacokinetic (PK)/pharmacodynamic (PD) analysis of pegcetacoplan used GA lesion area measurements from three clinical studies to determine the effect of pegcetacoplan exposure on GA progression. A base disease progression model was developed using data from sham‐treated eyes and untreated fellow eyes, followed by treatment effect assessment in dose–response and PK/PD models. In total, 1501 patients from FILLY (NCT02503332), OAKS (NCT03525613), and DERBY (NCT03525600) received intravitreal pegcetacoplan 15 mg monthly or every other month (EOM) or sham treatment monthly or EOM and were included in the population analysis of lesion area. Disease progression over time was adequately described as linear‐with‐time over the 24‐month maximal study duration. Disease‐specific covariates associated with slower lesion growth were unilateral, unifocal, and subfoveal GA lesions and >20 intermediate or large drusen groups (≥63 μm) at baseline. The dose–response model estimated 0.80‐fold (95% CI: 0.75, 0.84) and 0.83‐fold (95% CI: 0.78, 0.87) reductions in GA lesion growth rate with pegcetacoplan monthly and EOM, respectively, versus sham. A relationship between vitreous humor concentration and GA lesion growth rate was quantified as 2.6% per unit of log‐transformed vitreous pegcetacoplan concentration in the PK/PD model. PK/PD predictions of treatment effect based on exposure (pegcetacoplan monthly: 0.80 [90% CI: 0.77, 0.84]; pegcetacoplan EOM: 0.83 [90% CI: 0.80, 0.86]) were consistent with predictions based on dose response. These results support the benefit of pegcetacoplan administered monthly or EOM in slowing GA lesion growth.

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

Pegcetacoplan, a synthetic pegylated peptide inhibitor of complement components C3 and C3b, significantly reduced the rate of geographic atrophy (GA) lesion area progression in phase II and III trials in patients with age‐related macular degeneration.

• WHAT QUESTION DID THIS STUDY ADDRESS?

Data were pooled across the 12‐month phase II FILLY trial and two 24‐month phase III trials (OAKS and DERBY) to quantitatively characterize disease progression in sham‐treated study eyes and untreated fellow eyes, and to develop both dose–response and pharmacokinetic (PK)/pharmacodynamic (PD) models of the relationship between pegcetacoplan treatment and GA lesion area, to inform posology.

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

Progression of GA lesions was approximately linear with respect to time over the 24‐month maximal study duration. Baseline disease characteristics associated with the rate of lesion growth were identified. The reduction in the rate of lesion area growth demonstrates rank‐order correlation with dosing intensity, and a PK/PD relationship between predicted pegcetacoplan concentration in vitreous humor was quantified providing supportive evidence for dosing.

• HOW MIGHT THIS CHANGE DRUG DISCOVERY, DEVELOPMENT, AND/OR THERAPEUTICS?

Identification of baseline disease characteristics associated with faster GA progression may assist in identifying patients with GA secondary to age‐related macular degeneration (AMD) who would benefit from pegcetacoplan treatment. Intravitreal pegcetacoplan 15 mg monthly or every‐other‐month dosing regimens are appropriate for treatment of patients with GA secondary to AMD. Covariate analyses indicate intrinsic or extrinsic factors are not expected to affect pegcetacoplan posology, including concomitant anti‐vascular endothelial growth factor administration.

INTRODUCTION

Geographic atrophy (GA) is an advanced form of nonexudative age‐related macular degeneration (AMD) characterized by progressive and irreversible degeneration of the macula affecting photoreceptors, retinal pigment epithelium, and choriocapillaris. ¹ GA is estimated to affect >5 million people worldwide, including >1 million in the United States, and accounts for ~ 20% of all cases of legal blindness in North America. ² , ³ Clinically, GA can lead to significant visual function deficits in reading and low‐light adaptation with dense, irreversible scotomas in the visual field. ⁴ , ⁵ , ⁶ The underlying pathology of GA is complex and thought to be triggered by intrinsic and extrinsic oxidative stressors of the retinal pigment epithelium, leading to inflammation mediated by multiple pathways including the complement cascade. ⁷ Chronic inflammation resulting from overactivation or misregulation of the complement cascade has been shown to be strongly associated with GA lesion development and disease progression in human biochemical, genetic, and preclinical studies. ⁷ , ⁸ As the point of convergence for all three pathways of complement cascade activation, C3 has been identified as a central component driving GA progression and has become a therapeutic target for development of GA treatments. ⁸

Pegcetacoplan, a US Food and Drug Administration (FDA)–approved treatment for GA, is a synthetic pegylated peptide medication for the treatment of patients with GA secondary to AMD. ⁹ Pegcetacoplan binds to and inhibits complement component C3 and its activation fragment C3b, preventing cleavage of C3 into C3a and C3b. ¹⁰ By preventing cleavage of C3 and binding to C3b, pegcetacoplan regulates all downstream effectors of complement activation including phagocytosis, inflammation, and opsonization, regulating the overactive complement pathways. ¹¹

Pegcetacoplan is administered as a 15‐mg solution (0.1 mL of 150 mg/mL solution) by intravitreal injection to the affected eye once every 25–60 days. ⁹ Following repeated administration in patients with GA, the model‐predicted steady‐state geometric mean (% coefficient of variation) serum maximal concentration was 2.2 μg/mL (28.7%) when administered monthly and 1.5 μg/mL (58.1%) when administered every other month (EOM). ⁹ Model‐predicted vitreous exposure is predicted to be >1300‐fold higher than serum exposure within minimal accumulation predicted in either serum or vitreous humor from first dose to steady‐state (ratio of geometric mean exposure: 1.10–1.50). ¹²

Three clinical trials were conducted to investigate the efficacy, safety, and tolerability of pegcetacoplan in patients with GA secondary to AMD with patients randomized 2:2:1:1 to pegcetacoplan monthly, pegcetacoplan EOM, sham monthly, and sham EOM. In the 12‐month phase II FILLY trial (ClinicalTrials.gov identifier: NCT02503332) and the 24‐month phase III OAKS (NCT03525613) and DERBY (NCT03525600) trials, intravitreal pegcetacoplan was associated with significant reductions in GA lesion growth, with increasing effects over time compared with sham treatment. ⁶ , ¹³ Pegcetacoplan also demonstrated an acceptable safety profile with 2 years of follow‐up and ~ 12,000 injections administered to >1200 patients in the phase III trials combined.

The objectives of this analysis were to quantify the progression of lesion area and assess the impact of baseline disease characteristics and pegcetacoplan treatment using patient‐level data pooled across the FILLY, OAKS, and DERBY trials. Disease progression models were developed to characterize lesion area progression in the eyes not receiving pegcetacoplan and to investigate the association of baseline disease characteristics with disease progression followed by the assessment of pegcetacoplan treatment effect in dose–response and pharmacokinetic (PK)/pharmacodynamic (PD) models.

MATERIALS AND METHODS

Study design and patient population

The PK/PD analysis included lesion area measurements from three clinical trials (phase II FILLY, phase III OAKS, and phase III DERBY) in patients with GA following intravitreal administration of pegcetacoplan or sham treatment. Study methodologies were published previously. ⁶ , ¹³ Briefly, eligible patients aged ≥50 years (FILLY) or ≥60 years (OAKS and DERBY) were randomized 2:2:1:1 to receive a 100 μL intravitreal injection of pegcetacoplan 15 mg monthly, pegcetacoplan 15 mg EOM, sham monthly, or sham EOM for ≤24 months administered to the eye under treatment, defined as the study eye of the patient. The patient's fellow eye was the eye that was not included in the study and was not randomized to receive either sham treatment or pegcetacoplan. The fellow eye served as the patient's own control for the patient's study eye. Pegcetacoplan treatment was administered using a 27‐ or 29‐gauge thin‐walled needle. Sham treatment was performed in the same manner as pegcetacoplan treatment except no injection was administered, and the study eye was touched only with the blunt end of the syringe. Key study inclusion criteria (study eye) included best‐corrected visual acuity of 24 letters or better using Early Treatment Diabetic Retinopathy Study charts (20/320 Snellen equivalent), diagnosis of GA secondary to AMD confirmed by fundus autofluorescence imaging, and total GA lesion area between 2.5 and 17.5 mm² with ≥1 lesion of ≥1.25 mm² if GA was multifocal. Patients with both nonsubfoveal and subfoveal lesions in the study eye were included in the analysis. Key exclusion criteria included GA due to causes besides AMD, history or presence of active choroidal neovascularization, and non‐AMD retinopathies. The primary end point of the three clinical trials was the change in total GA lesion area from baseline to month 12; the change in total GA lesion area from baseline to month 24 was a key secondary end point in OAKS and DERBY.

Software

Analysis data set assembly and pre‐ and post‐processing of model outputs were performed using SAS (version 9.4 or later; SAS Institute Inc., Cary, NC, USA) or R (version 3.6.1 or later; R Core Team, Vienna, Austria) software. Model fitting was performed using NONMEM (version 7.3; ICON Plc, Dublin, Ireland).

Clinical end point

The primary efficacy end point for this analysis was GA lesion area as measured by fundus autofluorescence. All patients with ≥1 nonmissing lesion area measurement were included in the efficacy analysis set for GA lesion area.

Pegcetacoplan exposure

A population PK model devised using data from 11 clinical studies, including two phase III studies, displayed linear PKs over the range of doses evaluated in the included studies. ¹⁴ The predicted geometric mean (coefficient of variation %) of clearance for pegcetacoplan is 0.36 L/day (30%). The median (interquartile range) effective elimination half‐life of pegcetacoplan is 8.6 (6.9–11.0) days in patients with paroxysmal nocturnal hemoglobinuria. No meaningful effect on the PK profile of pegcetacoplan was seen with concomitant administration of eculizumab, age, sex, Asian race, Japanese ethnicity, renal function, and hepatic function (as evaluated by total bilirubin, albumin, aspartate aminotransferase, or alanine aminotransferase). Pegcetacoplan kinetics was influenced by body weight with 1.20‐fold (90% confidence interval [CI]: 1.16–1.23) higher and 0.827‐fold (90% CI: 0.800–0.853) lower average exposure at the 5th (53 kg) and 95th (94 kg) percentiles of body weight (reference: 70 kg). ¹⁴ The proposed dosing regimen for treatment of adult patients with GA secondary to AMD was deemed to be appropriate based on data from two clinical pharmacology studies in patients with neovascular AMD and three clinical studies in patients with GA secondary to AMD. ¹⁵ Individual predicted pegcetacoplan concentrations in both serum and vitreous humor were derived from this population PK model for all patients using their actual dosing history, empirical Bayesian estimates of PK parameters, and observed covariate vectors. For patients with missing or nonevaluable PK data, typical PK parameter estimates were used to generate population‐predicted pegcetacoplan concentrations, accounting for actual dosing history and the effects of patient‐level covariates included in the PK model. For calculations of pegcetacoplan vitreous concentration, the volume of the vitreous compartment was set to 4 mL. ¹⁶

Disease progression base model

The first step in model development was development of a disease progression model that used 12 months of treatment data and 6 months of follow‐up data from study eyes receiving sham treatment and fellow eyes in the phase II FILLY trial and 24 months of treatment data from study eyes receiving sham treatment and fellow eyes from the phase III OAKS and DERBY trials (Figure 1). Fixed‐effect parameters describing disease progression were estimated separately for the study and fellow eyes, with the presence of GA in the fellow eye at baseline (bilateral GA) included as a structural covariate of disease progression (Table S1). Fellow eye observations were excluded from patients with unilateral GA lesions at baseline who developed bilateral disease during the study. Standard diagnostic plots, model stability, and changes in the objective function value were used to determine the most appropriate disease progression model.

FIGURE 1.

Flowchart of model development. VPC, visual predictive check.

Disease progression covariate model

Following identification of the base disease progression model, a full disease progression covariate model was developed to explore the effect of disease‐specific covariates that were not included as structural covariates in the disease progression base model.

Covariate effects were assessed on parameters for initial lesion area and the rate of disease progression in study eyes only using 12 months of treatment data and 6 months of follow‐up data from the phase II FILLY trial and 24 months of treatment data from the phase III OAKS and DERBY trials. Covariates for consideration in the lesion area disease progression full model were lesion location, focality, number of intermediate or large drusen groups (diameter ≥ 63 μm), ¹⁷ and low‐luminance deficit. The relationship between continuous covariates and the typical value of disease progression parameters was described initially using power models:

$${\theta }_{\mathrm{TV},\mathit{ij}}={\theta }_{\mathrm{REF}}{\left(\frac{x_{\mathit{ij}}}{x_{\mathrm{REF}}}\right)}^{{\theta }_x}$$

where θ _REF and θ _x are the fixed‐effect parameters, x _REF is a reference value of the covariate x, and x _ij is the value of covariate x for patient i at time j. For time‐invariant (stationary) covariates, the values of x _ij and θ _TV,ij were constant within individual i at all time points. The approximate median value in the population was used for x _REF.

The relationship between categorical covariates (x _ij ) and the typical value of disease progression parameters was described initially as:

$${\theta }_{\mathrm{TV},\mathit{ij}}={\theta }_{\mathrm{REF}}·\left(1+{\theta }_x·x_{\mathit{ij}}\right)$$

where θ _REF and θ _x are fixed‐effect parameters and x _ij is the indicator variable that is equal to 1 or 0 dependent on the category of the covariates for patient i at time j. For time‐invariant categorical covariates, the value of θ _TV,ij was constant within individual i at all time points.

All covariate–parameter relationships of interest were included simultaneously in a single full model. Following the identification of a stable full covariate model, a stepwise backward elimination procedure was used to identify a parsimonious disease progression model containing similar information content as the full model, but with fewer covariates (Table S2). The covariate–parameter relationship that had the lowest change in objective function value (ΔOFV) and did not meet the inclusion criteria ΔOFV < 10.8 (p > 0.001) was eliminated and the stepwise backward elimination procedure was repeated until all covariate parameters met the inclusion criteria. A visual predictive check was performed on the preliminary final disease progression model prior to development of an exposure–response model to evaluate the impact of pegcetacoplan dose and exposure on the rate of disease progression.

PK/PD model

Following identification of the final covariate disease progression model, the effect of pegcetacoplan treatment on disease progression was assessed using 12 months of treatment data and 6 months of follow‐up data from study eyes receiving sham treatment and fellow eyes not receiving treatment in the phase II FILLY trial and 24 months of treatment data from study eyes receiving sham treatment and fellow eyes not receiving treatment from the phase III OAKS and DERBY trials (Table S3). Initially, pegcetacoplan dosing regimen was used as a surrogate for pegcetacoplan exposure in a dose–response model. Subsequent models evaluated instantaneous individual predicted vitreous pegcetacoplan concentration as the input for pegcetacoplan exposure following a PK/PD modeling approach. The PK/PD model used to generate individual predicted vitreous pegcetacoplan concentration in this compartment was based on an assumed volume of 4 mL and the individual empirical Bayesian estimates of the vitreous‐to‐serum absorption rate constant (the elimination rate constant for the vitreous compartment). Individual predicted vitreous pegcetacoplan exposure was preferred to serum, as it is more mechanistically related to lesion area. A visual predictive check was performed on the preliminary final exposure–response model prior to finalization.

Post hoc covariate PK/PD model development

Additional covariates not included in the primary covariate analysis were assessed in prespecified post hoc analyses using 12 months of treatment data and 6 months of follow‐up data from study eyes receiving sham treatment and fellow eyes not receiving treatment in the phase II FILLY trial and 24 months of treatment data from study eyes receiving sham treatment and fellow eyes not receiving treatment from the phase III OAKS and DERBY trials and the final PK/PD model as the reference model. The effect of concomitant administration of anti‐vascular endothelial growth factor (anti‐VEGF) medications on disease progression was simultaneously tested on the parameters underlying both the rate of disease progression and treatment effect and the influence of pegcetacoplan immunogenicity was evaluated as a covariate of treatment effect only.

RESULTS

Observed data

In total, 1501 patients with GA who participated in the FILLY, OAKS, and DERBY trials were included in the PK/PD analysis. Of those, 505 patients received pegcetacoplan monthly, 498 patients received pegcetacoplan EOM, and 498 patients received sham treatment. Baseline disease characteristics were well‐balanced across the three trial populations, supporting the pooled analysis strategy (Table 1). The patient population was primarily female (62%) and White (94%), with median age of 79 years (range: 60–100 years). At baseline, most patients had bilateral disease (81%), with subfoveal lesions (63%) and multifocal lesions (70%) in study eyes. The presence of pseudodrusen was observed in 82% of study eyes and >20 intermediate or large drusen were present in 45% of study eyes. Median baseline lesion area was consistent across studies, ranging from 7.56–7.69 mm² for study eyes and 7.49–8.04 mm² for fellow eyes. Use of concomitant anti‐VEGF medication was uncommon across treatment groups in both the study (8%) and fellow eyes (16%). The prevalence of serum antidrug antibodies against the polyethylene glycol (PEG) or peptide moieties of pegcetacoplan—detectable at baseline or at any time during the study—was 67% and 5%, respectively.

TABLE 1.

Demographics and baseline disease characteristics of the exposure–response PK/PD analysis population.

Characteristic	FILLY	OAKS	DERBY	Total
	n = 246 a	n = 636	n = 620	n = 1502 a
Continuous covariates, median (range)
Age, years	80 (60–97)	79 (60–100)	80 (60–96)	79 (60–100)
GA lesion area b , mm2	8 (3–17)	8 (2–18)	8 (2–18)	8 (2–18)
Low‐luminance deficit b , c	24 (−3, 69)	24 (−12, 76)	24 (−7, 77)	24 (−12, 77)
Categorical covariates, n (%)
Age
Elderly (≥65 years)	240 (98)	612 (96)	601 (97)	1453 (97)
Not elderly (<65 years)	6 (2)	24 (4)	19 (3)	49 (3)
Sex
Female	154 (63)	391 (61)	379 (61)	924 (62)
Male	92 (37)	245 (39)	241 (39)	578 (38)
Race
White	241 (98)	581 (91)	583 (94)	1405 (94)
Black	2 (0.8)	2 (0.3)	2 (0.3)	6 (0.4)
Asian	1 (0.4)	6 (0.9)	1 (0.2)	8 (0.5)
Other	2 (0.8)	0	1 (0.2)	3 (0.2)
Missing	0	47 (7)	33 (5)	80 (5)
GA laterality
Unilateral	40 (16)	112 (18)	134 (22)	286 (19)
Bilateral	206 (84)	524 (82)	486 (78)	1216 (81)
GA lesion location b
Subfoveal	146 (59)	410 (64)	386 (62)	942 (63)
Nonsubfoveal	100 (41)	226 (36)	234 (38)	560 (37)
GA focality b , d
Unifocal	79 (32)	195 (31)	181 (29)	455 (30)
Multifocal	166 (67)	441 (69)	439 (71)	1046 (70)
No. of intermediate or large drusen groups b , e
≤20	140 (57)	326 (51)	354 (57)	820 (55)
>20	106 (43)	308 (48)	266 (43)	680 (45)
Concomitant anti‐VEGF	26 (11)	46 (7)	47 (8)	119 (8)

Abbreviations: GA, geographic atrophy; PD, pharmacodynamics; PK, pharmacokinetics; VEGF, vascular endothelial growth factor.

^a Includes 1 patient without quantifiable PK samples who was excluded from the PK/PD modeling analyses.

^b Study eye.

^c Two patients in OAKS and seven patients in DERBY had missing data.

^d One patient in FILLY had missing data.

^e Two patients in OAKS had missing data.

Disease progression model

The disease progression model was developed using data from study eyes receiving sham treatment and untreated fellow eyes. Growth in lesion area was characterized using a zero‐order linear disease progression model with prespecified structural covariate effects for bilateral GA on study eye initial lesion area and time slope, and Box‐Cox transformation of the interindividual (patient‐level) random‐effect parameters. Three nonstructural covariates representing baseline disease status (subfoveal involvement, focality, and number of intermediate or large drusen) were included in the parsimonious covariate disease progression model. Fixed‐effect model parameters in the disease progression model were well estimated, with relative standard error (RSE) values <40% (Table S4). Model performance assessments indicated the disease progression base model adequately predicted the change in lesion area over time in both eyes in the absence of pegcetacoplan treatment (Figure S1).

The disease progression model described an approximately linear increase in GA lesion area over the 24‐month maximal duration of the phase III trials. Lesion growth was estimated at 2.02 mm²/year from an estimated baseline lesion area of 7.73 mm² in patients with bilateral GA, subfoveal involvement, multifocal distribution, and fewer (≤20) intermediate or large drusen groups. The effect of baseline disease factors on disease progression was evaluated at month 24. Lack of subfoveal involvement was associated with a 1.12‐fold (90% CI: 1.07, 1.16) higher predicted median growth in lesion area compared with subfoveal involvement (Figure 2). Unilateral GA lesions (0.860‐fold lower; 90% CI: 0.808, 0.913), unifocal GA lesions (0.837‐fold lower; 90% CI: 0.800, 0.872), and the presence of more (>20) intermediate or large drusen groups (0.869‐fold lower; 90% CI: 0.837, 0.904) were associated with lower predicted lesion area growth than bilateral GA lesions, multifocal GA lesions, and fewer (≤20) intermediate or large drusen groups, respectively.

FIGURE 2.

Influence of covariates on change in GA lesion area ratio at month 24. CI, confidence interval; GA, geographic atrophy.

Dose–response model

A dose–response model was developed using the parsimonious covariate disease progression model as the starting point. This model included separate step function estimates of the drug effect for each active treatment regimen (pegcetacoplan 15 mg intravitreal monthly and EOM) on disease progression. Model parameters were well estimated (RSE: <25%) and consistent with values from the disease progression model (Table S5). Parameter estimates from the dose–response model for pegcetacoplan‐mediated reduction in the rate of GA lesion growth versus sham showed rank‐order correlation with dosing frequency (Figure S2). The model‐predicted rate of disease progression relative to sham was 0.80‐fold lower (95% CI: 0.75, 0.84) with pegcetacoplan 15 mg intravitreal monthly and 0.83‐fold lower (95% CI: 0.78, 0.87) with pegcetacoplan 15 mg intravitreal EOM (Figure 3). The dose–response model quantified point estimates of 20.4% and 17.2% reductions in lesion area change from baseline with intravitreal pegcetacoplan 15 mg monthly and EOM, respectively, compared with sham treatment.

FIGURE 3.

Influence of treatment effect on change in GA lesion area ratio at month 24. CI, confidence interval; GA, geographic atrophy; PEOM, pegcetacoplan every other month; PM, pegcetacoplan monthly.

PK/PD model

Following dose–response assessment, PK/PD models were fitted to quantify the relationship between predicted vitreous pegcetacoplan concentration and the reduction in lesion growth rate. A linear with log‐transformed concentration model was selected as the functional form for the PK/PD relationship between vitreous pegcetacoplan concentration and lesion area progression, as all parameters in a nonlinear, maximal effect (E _max) model could not be precisely estimated. Parameters in the PK/PD model were well estimated (RSE values <25%) (Table S6), with initial lesion area and rate of progression similar to the parameter estimates generated by the disease progression and dose–response models (Figure S3).

The typical model‐predicted PK/PD relationship for pegcetacoplan‐mediated reduction in disease progression is illustrated in Figure 4. The rate of GA lesion area growth was estimated to decrease by 2.6% per unit of log‐transformed vitreous pegcetacoplan concentration. This PK/PD curve was consistent with the point estimates for dose response within the predicted range of exposures for pegcetacoplan 15 mg monthly and EOM dosing regimens.

FIGURE 4.

Typical predicted treatment effect on GA lesion area disease progression. The continuous PK/PD relationship is illustrated using fixed‐effect parameter final estimates without uncertainty in estimation or interindividual variability and overlaid on the 90% prediction interval of individual patient vitreous average concentration for each dosing regimen. Horizontal reference lines are included for estimated effects on lesion area from the dose–response models. C _avg,ss, average concentration at steady state; DR, dose–response; ER, exposure–response; GA, geographic atrophy; PD, pharmacodynamics; PEOM, pegcetacoplan every other month; PI, prediction interval; PK, pharmacokinetics; PM, pegcetacoplan monthly.

The influence of additional disease‐specific covariates was assessed in prespecified post hoc covariate analyses using the final PK/PD model as the reference model. Concomitant anti‐VEGF treatment was not predicted to have a clinically meaningful impact on disease progression (Figure 5). Similarly, anti‐PEG immunogenicity was not predicted to meaningfully influence the treatment effect (Figure 6). The prevalence of anti‐pegcetacoplan antibody positivity was low across studies and treatment arms and was therefore insufficient to support a model‐based assessment of impact.

FIGURE 5.

Influence of concomitant anti‐VEGF medication on change in GA lesion area ratio at month 24. CI, confidence interval; GA, geographic atrophy; PEOM, pegcetacoplan every other month; PM, pegcetacoplan monthly; VEGF, vascular endothelial growth factor.

FIGURE 6.

Influence of anti–polyethylene glycol antibodies on change in GA lesion area ratio at month 24. CI, confidence interval; GA, geographic atrophy; PEG, polyethylene glycol; PEOM, pegcetacoplan every other month; PM, pegcetacoplan monthly.

DISCUSSION

In the present analysis, growth in GA lesion area in eyes not receiving pegcetacoplan (study eyes receiving sham and fellow eyes) was adequately characterized using a linear (zero‐order) disease progression model through up to 24 months of follow‐up. In the absence of treatment, the estimated GA lesion growth rate was 2.02 mm²/year for a baseline GA lesion, with an estimated area of 7.73 mm² for patients with bilateral GA, subfoveal involvement, multifocal distribution, and ≤20 intermediate or large drusen groups. Patients with unilateral GA (vs. bilateral), unifocal GA (vs. multifocal), subfoveal GA (vs. nonsubfoveal) lesions, and the presence of >20 intermediate or large drusen groups (vs. ≤20 intermediate or large drusen groups) had lower rates of disease progression at 24 months. A 0.80‐fold and 0.83‐fold lower rate of disease progression was observed with pegcetacoplan 15 mg intravitreal monthly and pegcetacoplan 15 mg intravitreal EOM, respectively, compared with sham treatment. The dose–response model quantified point estimates of 20.4% and 17.2% reductions in lesion area change from baseline with intravitreal pegcetacoplan 15 mg monthly and EOM, respectively, compared with sham treatment. Intrinsic or extrinsic factors are not predicted to affect pegcetacoplan posology, as disease‐specific covariates were shown to affect only the underlying rate of disease progression, not the magnitude of the treatment effect. A clinically meaningful impact was not predicted with concomitant anti‐VEGF treatment. Anti‐PEG immunogenicity was not predicted to meaningfully influence pegcetacoplan treatment effect.

Among >1500 patients with GA in the FILLY, OAKS, and DERBY trials, disease progression over time in sham‐treated study eyes and fellow eyes was adequately described by a linear‐with‐time disease progression model. The estimated GA lesion growth rate of 2.02 mm²/year in this analysis is slightly higher than the estimated 1.66 mm²/year (95% CI: 1.44, 1.89) GA lesion growth rate reported in a recent meta‐analysis. ¹⁸ However, the meta‐analysis was not adjusted for disease‐specific covariates and is consistent with the estimate from the unadjusted base disease progression model (1.88 mm²/year). A moderately strong correlation (ρ = 0.705) was observed in the interindividual variability of the rate of disease progression between study and fellow eyes with GA, suggesting incorporation of fellow eye data may help describe intrapatient variability in lesion growth. The baseline disease characteristics of unilateral GA, unifocal GA, and subfoveal GA lesions identified as having lower rates of disease progression at 24 months in this study are consistent with previously published findings. ¹ , ¹⁹ , ²⁰ , ²¹ Unilateral GA was included as a structural covariate of disease progression based on prior knowledge that bilateral GA is an indicator of accelerated atrophy. ²²

A linear with log‐transformed vitreous concentration PK/PD model parameterization demonstrated adequate fit of the observed data and sufficient precision in parameter estimation. The typical model‐predicted PK/PD relationship was globally consistent with dose–response model estimates at corresponding vitreous average exposures. The PK/PD model was used to generate individual predicted vitreous concentrations of pegcetacoplan as opposed to exposure–response analysis, as the vitreous humor is inaccessible in humans. Quantification of an exposure‐related reduction in the rate of GA lesion growth provides additional proof of pharmacology and evidence in support of the 15 mg monthly to EOM posology. The rapid decrease in the dynamic portion of the curve outside of the 90% prediction interval of vitreous concentrations with pegcetacoplan dosing at 15 mg EOM and the overlapping CIs for the pegcetacoplan 15 mg monthly and EOM dosing regimens generated in the PK/PD model indicate that the exposure–response is likely saturated. Given the one‐compartment kinetics, assumed volume of the vitreous humor, and assessment of lesion area at dosing visits, limited additional value was seen in using derived exposure metrics as opposed to instantaneous predicted vitreous concentration. In addition, the injection volume with these dosing regimens is near the maximal acceptable for intravitreal administration (0.1–0.2 mL), making more frequent dosing infeasible. Indeed, Figure 3 shows a predicted reduction greater than 15% in the rate of lesion area growth at the 5th percentile of vitreous average concentration was observed with pegcetacoplan 15 mg EOM, suggesting benefit in reducing the rate of disease progression even at the lowest extreme of predicted exposure. The dose–response model quantified point estimates of 20.4% and 17.2% reductions in lesion area change from baseline with intravitreal pegcetacoplan 15 mg monthly and EOM, respectively, align with the reduction in GA growth rates seen with pegcetacoplan 15 mg monthly and EOM reported in the phase II and phase III clinical trials of pegcetacoplan. A 29% and 20% reduction in GA lesion area growth with pegcetacoplan monthly and EOM, respectively, compared with sham treatment was reported in the phase II FILLY trial. ⁶ The combined results of the phase III OAKS and DERBY reported a 30% and 24% reduction in GA lesion area growth with pegcetacoplan monthly and EOM, respectively, compared with sham treatment. ¹³ The results of this analysis can be used to provide external validation of long‐term extension data from the GALE trial (NCT04770545) and assessment of nonlinearity in the PK/PD relationship in follow‐up times beyond 24 months.

Limitations of this study include the brief overview of the population PK model used to predict the vitreous concentration of pegcetacoplan and use of this model as opposed to performing an exposure–response analysis.

AUTHOR CONTRIBUTIONS

R.L.C., K.P., F.G., R.R., S.C., and E.L. wrote the manuscript. R.L.C., K.P., F.G., S.C., and E.L. designed the research. R.L.C. and E.L. performed the research. R.L.C., K.P., F.G., R.R., S.C., C.R.B., and E.L. analyzed the data.

FUNDING INFORMATION

This study was funded by Apellis Pharmaceuticals, Waltham, Massachusetts, USA.

CONFLICT OF INTEREST STATEMENT

CRB is an employee of Apellis Pharmaceuticals and may hold stock and/or stock options. RLC, KP, FG, and RR are former employees of Apellis and may hold stock and/or stock options. RLC, EL, and SC are paid consultants for A2‐Ai LLC, a full‐service clinical pharmacology consultancy service provider.

Supporting information

Data S1

Crass RL, Prem K, Gaudreault F, et al. Pharmacokinetic/pharmacodynamic analysis of geographic atrophy lesion area in patients receiving pegcetacoplan treatment or sham. CPT Pharmacometrics Syst Pharmacol. 2025;14:257‐267. doi: 10.1002/psp4.13264

DATA AVAILABILITY STATEMENT

The data reported in this article are available from the sponsor upon reasonable request. Individual participant data from clinical trials will not be made available. Requests for data access should be addressed to medicalresearch@apellis.com. The study protocol will be available with no end date. All proposals requesting data access will need to specify how the data will be used.