Ocular Pharmacodynamics of Intravitreal Faricimab in Patients With Neovascular Age-Related Macular Degeneration or Diabetic Macular Edema

Abstract

Purpose: Evaluate the ocular pharmacodynamics (PD) of intravitreal faricimab, a bispecific inhibitor of angiopoietin-2 (Ang-2) and vascular endothelial growth factor-A (VEGF-A), in patients with neovascular age-related macular degeneration (nAMD) or diabetic macular edema (DME).

Methods: Aqueous humor (AH) samples (1025 free Ang-2 concentrations and 1345 free VEGF-A concentrations) were collected from approximately 300 faricimab-treated patients with nAMD or DME in phase 2/3 trials. A population pharmacokinetic pharmacodynamic (popPKPD) model was developed to describe the dynamic effect of faricimab on free AH Ang-2 and VEGF-A.

Results: Mean baseline Ang-2 concentrations were 8.1 and 13.4 pg/mL in patients with nAMD and DME, respectively. The corresponding mean baseline VEGF-A concentrations were 58 and 135 pg/mL, respectively. Overall, approximately 79% of Ang-2 (84% within 8 weeks postdose and 55% beyond 12 weeks postdose) and 7% of VEGF-A postdose observations were below the lower limit of quantification. Model-derived Ang-2 and VEGF-A concentration-time profiles for patients on every 4-week/every 8-week dosing were predicted to maintain greater than 50% suppression of Ang-2 concentrations for the entire dosing period. Patients on every 12-week/16-week dosing were predicted to have greater than 50% Ang-2 suppression for 12 or more weeks, whereas 50% VEGF-A suppression was maintained for 9 to 10 weeks. At 8 weeks postdose, the median Ang-2 concentrations remained suppressed by approximately 80%. At 16 weeks postdose, the median VEGF-A concentrations returned to baseline, but median Ang-2 levels remained below baseline.

Conclusions: A popPKPD analysis demonstrated faricimab's rapid and sustained suppression of AH Ang-2 and VEGF-A.

Translational Relevance: A popPKPD analysis suggested that sustained suppression of ocular Ang-2 contributes to faricimab's extended durability, observed in clinical trials.

Introduction

The angiopoietin-2 (Ang-2) and tyrosine kinase with immunoglobulin-like and epidermal growth factor homology domains (Tie) signaling pathway is a key regulator of vascular stability in the retinal vasculature. Under physiological conditions, Ang-1 mediates endothelial cell survival and cell junction integrity via Tie2 receptors.^1,2 However, in retinal vascular disease, upregulation of Ang-2 leads to the displacement of Ang-1 binding to Tie2, thus neutralizing the vasoprotective effects of Ang-1/Tie2 signaling.^1,2 Ang-1 is involved in regulating endothelial cell survival and the integrity of cell junctions via Tie2 receptors. This led to the hypothesis that dual Ang-2/vascular endothelial growth factor-A (VEGF-A) inhibition might improve vascular stability and/or provide a prolonged treatment effect compared to that seen with current standard of care anti-VEGF monotherapies,³ and thereby improve patient outcomes. Faricimab is the first humanized, bispecific, immunoglobulin G monoclonal antibody that acts by independently binding and neutralizing both Ang-2 and VEGF-A with high specificity and potency. The fragment crystallizable (Fc) region of faricimab was engineered to reduce systemic exposure and proinflammatory responses by abolishing the binding to the neonatal Fc receptor and Fc-gamma receptor.⁴ 

A population pharmacokinetic pharmacodynamic (popPKPD) model has recently been described for the anti-VEGF, bevacizumab, in patients with neovascular age-related macular degeneration (nAMD).⁵ However, to date, there are no published popPKPD models for patients with diabetic macular edema (DME). Herein, we report for the first time a popPKPD model for faricimab, a dual Ang-2/VEGF-A inhibitor, in patients with nAMD or DME. We describe the ocular PD of intravitreal faricimab from randomized controlled trials in patients with nAMD or DME. A popPKPD model was developed to relate faricimab concentrations in the vitreous humor (VH) to the dynamics of associated targets in the aqueous humor (AH). Specifically, sampling data from phase 1, 2, and 3 clinical trials were used to characterize the extent and duration of suppression of Ang-2 and VEGF-A concentrations in the AH, and the correlation between AH Ang-2 and VEGF-A suppression time profiles and the observed durability in faricimab phase 3 trials. The popPKPD model was then used to understand and inform the dosing regimen for faricimab-treated patients with nAMD or DME. 

Materials and Methods

Data Acquisition

Data were obtained from three phase 2 (AVENUE [NCT02484690], BOULEVARD [NCT02699450], and STAIRWAY [NCT03038880]) and four phase 3 (TENAYA [NCT03823287], LUCERNE [NCT03823300], YOSEMITE [NCT03622580], and RHINE [NCT03622593]) trials. Trial designs, patient characteristics, and efficacy and safety results, have been reported previously.^6–14 Patients with intraocular surgery, such as cataract surgery or vitrectomies, within 3 months before the day 1 study visit, were excluded. Any patient with vitreous or preretinal hemorrhage, or concurrent intraocular conditions that, in the opinion of the investigator, could reduce the potential for visual improvement or require surgical intervention during the study were also excluded. All trials included in the analyses described herein complied with the Declaration of Helsinki and all patients provided informed consent in writing. 

AH samples were collected at baseline, before the first intravitreal injection, and at multiple timepoints from faricimab-treated patients who provided additional consent (see Supplementary Figs. S1 and S2 in the Supplementary Material). Each sample from each patient was considered to be an independent observation and was referenced to the most recent dose. Samples were collected and stored at −70°C within 1 hour. Samples were aliquoted and shipped on dry ice to the bioanalytical laboratory. On arrival, samples were stored at −70°C until analysis. Overall, there were 769 AH samples (VEGF-A and Ang-2 concentrations) collected during phase 2 trials and 1601 AH samples (VEGF-A and Ang-2 concentrations) collected during phase 3 trials. Details regarding AH sample timing for each trial are included in Supplementary Table S1. 

Free (faricimab-unbound) VEGF-A concentrations were assessed in AH samples using a validated bead-based sandwich immunoassay on a single molecule array (SIMOA) platform. The samples were analyzed without any pretreatment as the antibodies used recognize free VEGF-A. Based on the data obtained during validation, the method used was considered to be sufficiently precise, selective, and sensitive for the determination of free VEGF-A in human AH. For free VEGF-A in the AH, the lower limit of quantification (LLOQ) was 1.46 pg/mL and the upper limit of quantification (ULOQ) was 4800 pg/mL. 

In the phase 2 trials, free (unbound) Ang-2 concentrations were assessed in AH samples using immunological multi-parametric chip technique technology. Before measurement, Ang-2 molecules complexed to faricimab were removed from AH samples by an immunodepletion step, leaving only the free Ang-2 fraction in the supernatant. The assay was not validated and therefore not used for phase 3 sample analysis, but the method was considered to be sufficiently precise, selective, and sensitive for the determination of free Ang-2 concentrations in human AH. For free Ang-2 in the AH, LLOQ was 2.24 pg/mL and the ULOQ was 25,914 pg/mL. 

In the phase 3 trials, free Ang-2 concentrations were measured using a validated bead-based immunoassay on the SIMOA platform.¹⁵ The method involved two steps. In the first step, Ang-2–faricimab complexes were removed using paramagnetic beads coated with an anti-idiotypic antibody targeting the anti-VEGF-A part of faricimab. The high binding affinity of this antibody largely excluded any impact of VEGF on the efficiency of immunodepletion and the determination of free Ang-2.¹⁶ In the second step, the remaining free Ang-2 was then quantified using the SIMOA immunoassay. The LLOQ was 4.04 pg/mL and the ULOQ was 1750 pg/mL, defining the quantification range of the assay in 100% human AH. 

PopPKPD Model Development Approach

A sequential approach was followed. First, the previously established population PK model of faricimab in plasma, VH, and AH (based on data from the trials mentioned above and including the phase 1 studies NCT01941082 and JP39844) was used to predict drug concentrations in the VH for each patient (Diack et al., manuscript submitted). Second, the indirect response models were subsequently developed to relate the predicted VH concentrations to the concentrations of free Ang-2 and VEGF-A in the AH, as described elsewhere in this article. 

Patients

All faricimab-treated patients who had faricimab PK parameters estimated in the popPK model and had at least one quantifiable biomarker observation (Ang-2, VEGF-A), were included in the PK-Ang-2 and PK-VEGF-A models. Observations from patients that contributed more than one sample were handled as independent observations and were modeled as the time since the first dose. 

Handling of Below the Limit of Quantification Observations

The M3 method (likelihood-based method) for handling observations that were below the limit of quantification (BLQ) was applied for postdose BLQ observations.¹⁷ For graphical displays, values BLQ were set to LLOQ. 

Indirect Response Model

Because faricimab neutralizes both Ang-2 and VEGF-A, the drug concentrations in the VH were assumed to inhibit the production rates of both targets in AH. The model (indirect response model) was described by Equations (1) and (2):  

\begin{eqnarray}\begin{array}{@{}l@{}} \frac{{d{B_{AH}}}}{{dt}} = {K_{in}} \cdot \left( {1 - EFF} \right) - {K_{out}}{B_{AH}}\\ {K_{in}} = BAS{E_{AH}} \cdot {K_{out}}\\ {B_{AH}}\left( 0 \right) = BAS{E_{AH}} \end{array},\end{eqnarray}

(1)

where K_in is the production constant of Biomarker B, B_AH is AH concentrations of a biomarker B (i.e., VEGF-A or Ang-2), B_AH(0) is B_AH at time 0, BASE_AH is the baseline concentration of the biomarker in AH, k_out is the elimination rate constant of B_AH from AH, and EFF is the inhibitory effect of faricimab. 

The effect of faricimab on biomarker production rate was described by a sigmoid model:  

\begin{eqnarray}EFF = \frac{{{E_{Max}} \cdot C_{VH}^\gamma }}{{EC_{50}^\gamma + C_{VH}^\gamma }},\end{eqnarray}

(2)

where E_MAX is the maximum inhibitory effect, C_VH is the faricimab concentration in VH, EC₅₀ is the VH faricimab concentration that provides half of the maximum inhibitory effect, and γ is the sigmoidicity (Hill) coefficient. 

Separate models were developed for PK-Ang-2 and PK-VEGF-A (Fig. 1). Structural model refinement was driven by the data, and the adequacy of the model was assessed using various goodness-of-fit indicators (e.g., observed vs. predicted concentrations, visual predictive check simulations), the minimum objective function value (a measure of the likelihood of the model), and plausibility and precision of the parameter estimates. Inter-individual variability in structural model parameters was estimated using a log-normal distribution for the random effects. The initial residual variability was a combined additive and proportional error model. Covariates, such as age, sex, presence of antidrug antibodies, disease indication (nAMD vs. DME), and baseline concentrations of targets (VEGF-A and Ang-2) in AH were investigated in all model parameters. Significant covariates were included in the PK-VEGF and PK-Ang-2 models. Additional exploratory diagnostics were conducted by graphical exploration of all measured covariate effects (plots of estimates of individual random effects [η_i] from the base and final models vs. covariates and diagnostic plots stratified by the covariates of interest [not shown in this article]), and the models were revised, as necessary, guided by model diagnostics and change in the minimum objective function value. 

Figure 1.

 

Schematic of the final model for ocular PKPD of faricimab. BASEVH and BASEAH are the baseline values of the biomarker BAH (VEGF-A or Ang-2) concentrations in the VH and AH compartments. EFF, effect of faricimab on biomarker production rate; k, elimination rate; k_AH, elimination rate constant for AH; k_in, zero-order formation rate constant; k_plasma, elimination rate constant for plasma; K_out, first-order elimination rate constant; k_VH, elimination rate constant for VH; PKPDs, pharmacokinetics pharmacodynamics; VEGF-A, vascular endothelial growth factor-A; VH, vitreous humor.

Software

The popPKPD analyses were conducted via nonlinear mixed-effects modeling using NONMEM software, Version 7.5.0 (ICON Development Solutions, Dublin, Ireland).¹⁸ Graphical and all other statistical analyses, including evaluation of NONMEM outputs, were performed using R for Windows (R project, http://www.r-project.org/). 

Results

Datasets

A total of 1345 VEGF-A AH observations (413 observations, phase 2; 932 observations, phase 3) from 302 patients and 1025 Ang-2 AH observations (356 observations, phase 2; 669 observations, phase 3) from 225 patients were available for the PKPD analyses (Supplementary Table S1). Among the VEGF-A observations, 82 (1 at baseline and 81 postdose; 6.1%) were BLQ. Among the Ang-2 observations, 639 (46 at baseline and 593 postdose; 62.3%) were BLQ. The continuous covariates for patients included in the PKPD analyses are summarized in Supplementary Table S2 and Supplementary Table S3. The categorical covariates are summarized in Supplementary Table S4 (VEGF-A datasets) and Supplementary Table S5 (Ang-2 dataset). 

Observed Target Suppression in the AH

After intravitreal injection, faricimab is eliminated slowly from the VH and is distributed within the AH, where it binds to both VEGF-A and Ang-2. Predicted faricimab VH concentrations postdose were derived from both the initial faricimab injection (concentration) and the observed AH samples at different timepoints postdose (Fig. 2). The mean baseline VEGF-A concentrations were more than 2-fold higher in patients with DME (135 pg/mL) compared with patients with nAMD (58 pg/mL). The mean baseline Ang-2 concentrations were 65% higher in patients with DME (13.4 pg/mL) than in patients with nAMD (8.11 pg/mL). 

Figure 2.

 

Observed target suppression of AH VEGF-A and Ang-2 from samples obtained at baseline and various time points postdose, from phase 3 trials. In the boxplots, the middle line represents the median, the lower part of the box represents the 25th percentile, and the top part of the box represents the 75th percentile. The minimum and maximum values of the boxplot are shown as 25th percentile + 1.5*(75th percentile − 25th percentile) and 75th percentile + 1.5 × (75th percentile − 25th percentile), respectively. The red circles represent observed samples and the black diamonds represent outliers. AH, aqueous humor; Ang-2, angiopoietin-2; DME, diabetic macular edema; nAMD, neovascular age-related macular degeneration.

There was nearly complete suppression of VEGF-A after initiation of faricimab dosing for all patients, which rebounded when faricimab VH concentrations (predicted to be 113.00, 11.60, 1.65, and 0.28 mg/mL at weeks 4, 8, 12, and 16 postdose, respectively, in patients with DME) decayed. The predicted concentrations are higher for patients with nAMD because these patients are older. Furthermore, in both diseases, the AH concentration of faricimab is approximately 10-fold lower than that of the VH. The mean VEGF-A concentrations in patients with nAMD remained suppressed by approximately 50% relative to baseline 16 weeks postdose (Fig. 2). Free AH VEGF-A concentrations returned to baseline in patients with DME by week 12 postdose. Observed concentrations of free AH Ang-2 decreased from baseline to undetectable levels shortly after faricimab administration and then increased when faricimab VH concentrations decreased. Approximately 80% of free AH Ang-2 measurements were BLQ by 8 weeks postdose, and approximately 40% were BLQ by 12 weeks postdose (Fig. 2). 

The relative change from baseline in VEGF-A over time by treatment interval is shown in Figure 3A. In both diseases, the median time profile for the relative change from baseline in free AH VEGF-A up to 8 weeks postdose was similar between patients on every-8-week (Q8W) and every-12-week (Q12W)/every-16-week (Q16W) dosing in nAMD and also between patients on every-4-week (Q4W)/Q8W and Q12W/Q16W dosing in DME. Over the 8-week postdose period, more than 50% of the free AH Ang-2 concentrations were BLQ in patients with nAMD or DME, for all treatment intervals. Therefore, the median of free AH Ang-2 in all patient groups was also below or at the LLOQ over the 8-week postdose period. Hence, it is more relevant to use the percentage of free AH Ang-2 BLQ as a comparator between different treatment intervals. Figure 3B shows the relative change from baseline in the percentage of free AH Ang-2 BLQ. A higher percentage BLQ indicates greater suppression of free Ang-2. Suppression of free Ang-2 was greater with Q12W/Q16W treatment intervals than Q4W/Q8W treatment intervals in patients with nAMD or DME. 

Figure 3.

 

(A) Median observed concentration of AH VEGF-A relative to baseline, and (B) proportion of AH Ang-2 samples BLQ at time postdose relative to baseline (in faricimab-treated patients with nAMD or DME). BLQ, below the limit of quantification; Q4W, every-4-weeks; Q8W, every-8-weeks; Q12W, every-12-493 weeks; Q16W, every-16-weeks.

The phase 3 clinical trials included an aflibercept Q8W comparator arm. The free Ang-2 concentrations in AH from patients who had received aflibercept Q8W were also analyzed. Baseline Ang-2 levels were similar between faricimab- and aflibercept-treated patients (Fig. 4). However, aflibercept did not suppress free AH Ang-2 concentrations in patients with nAMD and DME. 

Figure 4.

 

Observed AH Ang-2 suppression in patients with nAMD or DME treated with faricimab or aflibercept. Approximately 80% or more of the Ang-2 concentrations in patients treated with faricimab were BLQ within the 8-week postdose time frame. In the boxplots, the middle line represents the median, the lower part of the box represents the 25th percentile, and the top part of the box represents the 75th percentile. The minimum and maximum values of the boxplot are shown as 25th percentile + 1.5*(75th percentile − 25th percentile) and 75th percentile + 1.5 × (75th percentile − 25th percentile), respectively.

PKPD Models

Two models with identical structure as defined by the previously described Equations (1) and (2) were developed to characterize the dynamics of VEGF-A and Ang-2, respectively. All model parameters from Equations (1) and (2) were estimated. Predicted faricimab concentrations in the VH were used as the driver of the dynamics of both Ang-2 and VEGF-A. The interindividual random effects were included on all parameters (k_out, BASE, EC₅₀, and Hill coefficient), except the E_MAX. During model development, observed baseline values of VEGF-A and Ang-2 in AH were used to predict the respective baseline parameters. The predicted baseline values of VEGF-A and Ang-2 were estimated for patients who had missing baseline values. 

VEGF-A Model

The popPKPD analysis showed that concentrations of free VEGF-A in the AH are well described by an indirect response model with inhibition of VEGF-A production described by a sigmoid (Hill) function of faricimab VH concentrations. The final PKPD model for VEGF-A did not include random effect on k_out. The residual variability was described by a proportional error model. Different Hill coefficients between phase 2 and phase 3 studies were included in the model. The only statistically significant covariate on the Hill parameter was baseline VEGF-A. All model parameters were estimated with good precision (see Relative Standard Error in Table 1). The goodness-of-fit diagnostics and the prediction-corrected visual predictive check simulations (Supplementary Fig. S3 and Fig. S4 in the Supplementary Material) indicated a good agreement between the observed and simulated data, confirming the ability of the model to capture both the central tendency and the interindividual variability of VEGF-A kinetics in AH. 

Table 1.

 

Parameter Estimates of the Faricimab Population PK-VEGF-A Model (Final Model)

 

The EC₅₀, the faricimab VH concentration that provided 50% suppression of VEGF-A, was predicted to be 2.59 µg/mL (Table 1). This concentration was reached between 9 and 10 weeks postdose. The model was used to simulate target suppression profiles for the different treatment intervals. After faricimab 6-mg dosing with Q12W or Q16W treatment intervals, greater than 50% suppression of free AH VEGF-A from baseline was predicted to be maintained for a median of approximately 73 days for patients with nAMD and 62 days for patients with DME (Table 2). Additional suppression cutoffs and corresponding durations of suppression for free VEGF-A are shown in Table 2. The predicted dynamics of AH VEGF-A suppression in faricimab-treated patients with nAMD or DME are shown in Figure 5. 

Table 2.

 

Predicted Duration of VEGF-A Suppression at Steady State Following 6-mg Dosing, by Dosing Regimen (Patients from Phase 3 Studies): Median (95% Prediction Interval)

 

Figure 5.

 

Predicted dynamics of intraocular VEGF-A and Ang-2 suppression through week 16 postdose (ASRS 2023 PKPD, Muni). ^aThe graphs are median AH, Ang-2, and VEGF-A concentrations over time derived from the popPKPD model by disease population. popPKPD, population pharmacokinetic pharmacodynamic.

Ang-2 Model

The popPKPD analysis showed that concentrations of free Ang-2 in the AH are well-described by an indirect response model with inhibition of Ang-2 production described by a sigmoid (Hill) function of faricimab VH concentrations. The model parameters were precisely estimated (see Relative Standard Error, Table 3). The standard goodness-of-fit plots were of limited use for a dataset with more than 60% of BLQ observations and suggested systematic bias (which is not due to model misspecification but a reflection of the high proportion of BLQ observations). However, the visual predictive check simulations indicated good agreement between the observed and simulated data, confirming the ability of the model to capture both the central tendency and the interindividual variability of Ang-2 kinetics in AH (Supplementary Fig. S5 in the Supplementary Material). 

Table 3.

 

Parameter Estimates of the Faricimab Population PK-Ang-2 Model (Final Model)

 

The faricimab VH concentration that provided 50% suppression of Ang-2 was predicted to be 0.0835 µg/mL (Table 3). This concentration is reached between 14 and 16 weeks postdose. The model was used to simulate target suppression for the different treatment intervals. After faricimab 6-mg Q16W dosing, free AH Ang-2 suppression of greater than 50% from baseline was predicted to be maintained for a median of 104 days for patients with nAMD, and for a median of 93 days for patients with DME. Additional suppression cutoffs and corresponding durations of suppression for free Ang-2 are shown in Table 4. The predicted dynamics of AH Ang-2 suppression in faricimab-treated patients with nAMD or DME are shown in Figure 5. 

Table 4.

 

Predicted Duration of Ang-2 Suppression at Steady-State Following 6-mg Dosing, by Dosing Regimen (Patients from Phase 3 Studies): Median (95% Prediction Interval)

 

Discussion

The faricimab popPKPD models described in this manuscript were developed and used to characterize and quantify the effect of faricimab on the intraocular suppression of Ang-2 and VEGF-A in patients with nAMD or DME from three phase 2 and four phase 3 trials. The data used for the current analyses were derived from a large number of patients and samples per patient (analyzed using validated assays), and hence provided a high-quality dataset representative of the target nAMD and DME populations. To the best of our knowledge, this popPKPD model is the first reported for both nAMD and DME that can be used to understand and inform the dosing regimen for faricimab-treated patients with nAMD or DME. 

In both nAMD and DME, the observed free AH VEGF-A and Ang-2 concentrations decreased from baseline to undetectable levels shortly after intravitreal injection of faricimab, thereafter rebounding when faricimab VH concentrations decreased. Free AH VEGF-A concentrations returned to baseline by 12 to 16 weeks postdose. In contrast, free AH Ang-2 concentrations did not return to baseline before 16 weeks postdose, with the majority (approximately 80%) of concentrations remaining BLQ 8 weeks postdose. These data demonstrate that treatment with faricimab resulted in the rapid and sustained suppression of both Ang-2 and VEGF-A in AH. The mean duration of AH VEGF-A suppression below the LLOQ has previously been reported as 34 days (approximately 5 weeks) for ranibizumab and 67 days (approximately 10 weeks) for aflibercept.¹⁹ 

The observed AH data from faricimab-treated patients suggest that there is no obvious correlation between dosing interval and free AH VEGF-A concentrations. In contrast, we did find a correlation between dosing interval and free AH Ang-2 concentrations. We also found that the duration of AH Ang-2 suppression was longer and the relative change in the percentage of AH Ang-2 levels that were BLQ was greater for patients on longer (Q12W/Q16W) vs. shorter (Q4W/Q8W) dosing intervals. These findings suggest that the Ang-2 inhibiting effects of faricimab may play an important role in mediating vascular stabilization. However, owing to the heterogeneity of the Ang-2 data (i.e., different assays and LLOQ concentrations for the phase 2 and phase 3 trials) and because 62.3% of observations were BLQ, interpretation of the observed data was challenging. To overcome these challenges and to account for variability in PK and PD data, the sparse sampling, and the information from BLQ values, popPKPD models were developed. The predicted dynamics of free AH Ang-2 and VEGF-A following multiple intravitreal faricimab administration in patients with nAMD and DME were described by two indirect response models (Fig. 5). The model parameters were precisely estimated, and the models predicted accurately the faricimab-mediated inhibition of VEGF-A and Ang-2 production in VH. The models predicted AH target suppression of 50% or more from baseline for a duration of approximately 14 weeks for Ang-2 and approximately 9 weeks for VEGF-A, with Ang-2 suppressed by faricimab for a longer duration than VEGF-A. Indeed, in both diseases, the predicted median VEGF-A concentrations returned close to baseline by 16 weeks postdose. In contrast, the predicted median Ang-2 concentrations remained below baseline through week 16 postdose. This is partly explained by the EC₅₀ of the PK-Ang-2 model, which is estimated to be 31-fold lower than that of the PK-VEGF-A model. 

The level of target inhibition required to prevent disease activation remains unknown. However, in a mouse model of ischemic retinopathy, induction of Ang-2 or concomitantly high levels of Ang-2 and VEGF, promote neovascularization.²⁰ Furthermore, preclinical studies have shown that Ang-2 acts as a permissive factor which helps to regulate the responsiveness of retinal vessels to VEGF-A.²¹ Thus, Ang-2 inhibition through 16 weeks, as shown by the popPKPD model, may play an important role in the durability signal seen in faricimab phase 3 studies. 

Our analysis does have several limitations. Although an appropriate M3 method was used to describe BLQ data, the conclusions of the PK-Ang-2 analysis should be interpreted with caution. The LLOQ levels were relatively high at 2.24 ng/L for the phase 2 assay (22% of the mean baseline Ang-2 level) and 4.04 ng/L for the phase 3 assay (40% of the mean baseline Ang-2 level). The robustness of both models (PK-VEGF and PK-Ang-2) could be enhanced in the future by incorporating additional AH data, including from the faricimab retinal vein occlusion phase 3 trials, BALATON (NCT04740905) and COMINO (NCT04740931). 

Conclusions

The faricimab popPKPD model showed rapid and sustained suppression of VEGF-A and Ang-2 concentrations in the AH following intravitreal injection of faricimab in patients with DME and nAMD. Notably, VEGF-A concentrations returned to baseline approximately 16 weeks after dosing, whereas Ang-2 concentrations remained suppressed through 16 weeks after dosing, which is consistent with the extended durability of faricimab seen in clinical trials. These robust data constitute evidence that faricimab provides intraocular target suppression of VEGF-A and Ang-2 in patients with nAMD and DME. 

Acknowledgments

Supported by F. Hoffmann-La Roche Ltd., Basel, Switzerland. The sponsor participated in the study design; the collection, analysis, and interpretation of data; the writing of the report; and the decision to submit the paper for publication. Funding was provided by F. Hoffmann-La Roche Ltd. for the study and third-party writing assistance, which was provided by Neil Norcross, PhD, and Luke Carey, PhD, CMPP, of Envision Pharma Group. 

Disclosure: C. Diack, F. Hoffmann-La Roche Ltd. (E, I); R.L. Avery, 4DMT (C), AbbVie (C), Alcon (C), Alimera (C), Allergan (C), Amgen (C), AsclepiX (C), Astellas (C), Avicida (C), Bausch + Lomb (C), Cardinal Health (C), Clearside Biomedical (C), Coherus (C), EyePoint (C), Forwardvue (C), Genentech Inc. (C), Glaukos (C), InFocus Capital Partners (C), Imprimis (C), Ingenia (C), Kriya (C), KYS Vision (C), Merit (C), Novartis (C), NVasc (C), Ocular Therapeutix (C), Outlook (C), Pixium (C), Pr3vent AI (C), PulseMedica (C), Regenxbio (C), Replenish (C), Re-Vana (C), Santen (C), Tenpoint Therapeutics (C), Vial (C), Visionary Ventures (C), AbbVie (I), Adverum (I), Alcon (I), Aldeyra (I), Apellis (I), EyePoint (I), InFocus Capital Partners (I), Iveric Bio (I), Kodiak Sciences (I), Novartis (I), NVasc (I), Ocular Therapeutix (I), Outlook (I), Regeneron (I), Replenish (I), Re-Vana (I), Verana Health (I), Visionary Ventures (I), Aviceda (I), Forwardvue (I), Ingenia (I), KYS Vision (I), NVasc (I), Genentech Inc. (S); C.M.G. Cheung, Allergan (C), Bayer (C), Boehringer Ingelheim (C), Novartis (C), Roche (C), Samsung (C), Topcon (C); K.G. Csaky, AbbVie (C), Adverum (C), Annexon (C), Cognition Therapeutics (C), Endogena (C), EyeBio (C), Genentech Inc./Roche (C), Heidelberg Engineering (C), Johnson & Johnson (C), Merck (C), NGM Bio (C), Novartis Pharma AG (C), Ocular Therapeutix (C), Ribomic (C), Alexion (F), Annexon (F), Boehringer Ingelheim (F), Gyroscope (F), Iveric Bio (F), NGM Bio (F); L. Gibiansky, F. Hoffmann-La Roche Ltd. (C); F. Jaminion, F. Hoffmann-La Roche Ltd. (E, I); E. Gibiansky, F. Hoffmann-La Roche Ltd. (C); D. Sickert, F. Hoffmann-La Roche Ltd. (E, I); I. Stoilov, Genentech Inc. (E); V. Cosson, F. Hoffmann-La Roche Ltd. (E, I); K. Bogman, F. Hoffmann-La Roche Ltd. (E, I)