Changes in Aqueous Angiopoietin-1/2 Concentrations During the Induction Phase of Intravitreal Faricimab Injections for Diabetic Macular Edema

Abstract

Purpose: The purpose of this study was to investigate the dynamic changes in aqueous concentrations of angiopoietin (Ang)-1/2 and vascular endothelial growth factor (VEGF) during injection in treatment-naïve patients with diabetic macular edema (DME) receiving faricimab during the induction phase (3 consecutive monthly doses) and retrospectively analyze the data.

Methods: Thirty-five eyes of 26 patients (age = 63.1 ± 12.9 years) with treatment-naïve DME received faricimab injections monthly, 3 consecutive times. Additionally, 59 eyes of 59 patients (age = 63.9 ± 8.8 years) who underwent cataract surgery were recruited as controls. Aqueous humor samples were collected from each injection or surgery and stored at −80°C, and the concentration of each cytokine was quantified using a multiple enzyme-linked immunosorbent assay (Luminex). The clinical parameters of best-corrected visual acuity (BCVA) and central foveal thickness (CFT) at each visit were also recorded.

Results: Three induction phases of faricimab significantly suppressed each aqueous cytokine, rapidly for VEGF, gradually for Ang-2, and slightly for Ang-1. The Ang-1/2 ratio was lower (<1.0) at baseline and gradually increased, but did not reach a control ratio of 1.58. The baseline CFT correlated with VEGF, but not with the Ang-2, Ang-1, or Ang-1/2 ratios. After three injections, CFT did not correlate with VEGF, but it positively correlated with Ang-2 and negatively correlated with Ang-1, and it strongly negatively correlated with the Ang-1/2 ratio.

Conclusions: The Ang-1/2 ratio in the aqueous humor significantly negatively correlated with the degree of residual edema after faricimab treatment for DME.

Translational Relevance: The Ang-1/2 ratio in aqueous humor is thus a useful biomarker of the treatment response for DME.

Introduction

Diabetic macular edema (DME) is a leading cause of central vision loss among people with diabetes mellitus of working age, and it can occur at any stage of diabetic retinopathy.¹ Intravitreal injection of anti-vascular endothelial growth factor (VEGF) has proven to be effective in treating DME and improving visual acuity,² thus, in recent years, anti-VEGF drugs have become the first choice for the treatment of DME.³ However, the response to anti-VEGF drugs varies widely between individuals, with up to 30% of cases not responding adequately to treatment,^4,5 and other factors, such as inflammation, angiopoietin, and PlGF, are involved in the pathogenesis of DME.⁶ 

Dysregulation of angiopoietin (Ang)/Tie-2 signaling has been linked to diabetic retinopathy as it regulates the permeability of the blood-retinal barrier and the function of retinal pericytes.⁷ Ang-1 binds to activate the Tie-2 receptor and plays a central role in vascular stabilization,⁸ whereas Ang-2 can be induced under conditions of hypoxia, hyperglycemia, and oxidative stress.^9,10 It also competitively binds to the Tie-2 receptor to inactivate its signaling and disrupt vascular stability. Thus, the inhibition of Ang-2 and activation of Ang-1 play pivotal roles in vascular stabilization. 

Faricimab was designed as the first intraocular bispecific antibody that independently inhibits Ang-2 and VEGF, and it has demonstrated sustained visual and anatomic improvements in patients with DME.¹¹ Although faricimab suppressed the intraocular concentration of Ang-2,¹² and (may) disrupt Ang-2 upregulation to activate the Ang-1/Tie-2 signaling pathway, it is not entirely clear how VEGF and Ang-2 are involved in the clinical pathogenesis. 

This study included treatment-naïve patients with DME who received three consecutive monthly doses of faricimab for the first time. Dynamic changes in VEGF, Ang-1, and Ang-2 concentrations in the anterior chamber fluid obtained at the time of administration were measured and correlated with the clinical parameters, such as visual acuity and edema. 

Patients and Methods

Data Acquisition

This retrospective study was conducted at the Department of Ophthalmology of the Tokyo Medical University Hachioji Medical Center. From April 2023 through March 2024, the 35 eyes of 26 patients with treatment-naïve DME received 3 consecutive monthly doses of intravitreal faricimab (IFI; 6.0 mg in 0.05 mL, Vabysmo; Chugai Phramaceutical, Tokyo, Japan) as an induction phase. Of the 35 DME eyes, 4 eyes were classified as proliferative diabetic retinopathy (PDR), and 31 eyes as non-PDR. Patients were excluded if their medical history included any of the following conditions: a history of glaucoma, uveitis, retinal diseases other than DME, vitreous hemorrhage, rubeosis iridis, ocular infections, laser photocoagulation, intraocular surgery (including cataract surgery and anti-VEGF therapy), or severe kidney injury (e.g. the need for renal dialysis). 

At the time of each injection, after topical anesthesia with 0.4% oxybuprocaine eye drops (Benoxil; Santen Pharmaceutical, Osaka, Japan), a 30-gauge needle attached to an insulin syringe was used to obtain 0.1 mL of aqueous humor (mean value) from the anterior chamber by limbal paracentesis; the samples were subsequently stored in sterile plastic tubes at −80°C until the time of analysis. IFI was performed at a distance of 3.5 mm from the limbus through the pars plana. After IFI, the eyes were treated for 3 days with antibiotic eye drops. Full eye examinations and optical coherence tomography (OCT) were performed at each visit. The best-corrected visual acuity (BCVA) was measured using a Snellen chart and converted to the logMAR scale, and central foveal thickness (CFT) was measured using spectral domain OCT (Spectralis; Heidelberg Engineering, Heidelberg, Germany) and obtained from automated measurement of the retinal thickness based on analyses of the central 1000 µm subfield using the OCT mapping program. 

Aqueous humor collected at the time of surgery from the eyes of 59 patients who underwent cataract surgery was used as a control. 

The ethics committee of Tokyo Medical University Hachioji Medical Center approved the study (approval no. T2023-0039), and was performed in accordance with the principles of the Declaration of Helsinki. 

Cytokine and Growth Factor Measurements

The aqueous humor cytokine levels (VEGF, Ang-1, and Ang-2) were measured using enzyme-linked immunosorbent assays (xMAP; Luminex Corp. Austin, TX, USA), as reported previously.¹³ VEGF, Ang-1, and Ang-2 were detected using Beadlyte capture bead kits (Upstate Biotechnology, Lake Placid, NY, USA) according to the manufacturer's instructions in 25 µL of undiluted aqueous humor samples after incubation in a dark room at room temperature for 2 hours. The reference cytokine concentrations in the measurement kits were used to generate duplicate standard cytokine curves. All samples from a patient, before and after IFI, were analyzed in the same run to avoid potential run-related heterogeneity and were compared with the reference curves. All factors were at high enough levels to be detected (minimum detectable concentrations = VEGF = 0.99 pg/mL; Ang-1 = 9.43 pg/mL; and Ang-2 = 17.1 pg/mL). 

Ang-1 and Ang-2 act competitively on Tie-2 receptors, thus contributing to vascular stability. Thereafter, the Ang-1/2 ratio was calculated. 

Statistical Analysis

The SAS System software program (version 9.4; SAS Institute Inc., Cary, NC, USA) was used to analyze the data. Results are shown as the mean ± standard deviation (SD). A repeated analysis of variance (ANOVA) and post hoc analysis of Bonferroni test were used to compare continuous variables at baseline and after the anti-VEGF therapy. The correlation coefficients between the clinical parameters and each cytokine were calculated using the least-squares method. To examine the relationships among the variables, Spearman's rank-order correlation analysis or Pearson's correlation analysis was performed, as appropriate. Two-tailed P values < 0.05 were considered to indicate statistical significance. 

Results

A total of 35 eyes with treatment-naïve DME received monthly faricimab injections for 3 consecutive months. For the study of cytokine concentration measurements in the anterior chamber, anterior chamber fluid obtained from 59 eyes of non-diabetic cataract surgery patients with no other ocular diseases were used as controls. The basic characteristics of each group are presented in the Table. 

Table.

 

Basic Characteristics

 

Dynamic Changes in the Clinical Parameters

The BCVA at baseline was 0.32 ± 0.31 logMAR, but improved over time with successive injections of faricimab, to 0.23 ± 0.29 logMAR at the third injection (Fig. 1A: P < 0.01). Thus, the improvement in BCVA after 3 injections of faricimab was −0.09 ± 0.20 logMAR. CFT at baseline was 459.4 ± 143.4 µm, and regressed over time to reach 344.8 ± 100.9 µm at the final injection (Fig. 1B: P < 0.01); thus, the regression of CFT was –150.5 ± 146.8 µm. 

Figure 1.

 

Changes in the best-corrected visual acuity (BCVA) (A) and central foveal thickness (CFT) (B) after 3 consecutive doses of faricimab for diabetic macular edema (DME). Each point and vertical bar shows the mean and standard deviation of the data. Asterisks (*) indicate statistically significant differences (P < 0.05). The relationship between BCVA and CFT at the first injection (C) and the third injection (D) of faricimab. The theoretical line shows the correlations based on linear regression equations.

Relationship Between CFT and BCVA

The CFT and BCVA at baseline significantly correlated with each other (Fig. 1C: r = 0.572, P < 0.01), but they showed no correlation at the second (r = 0.189, P = 0.28) or third injection (Fig. 1D: r = 0.225, P = 0.19). 

Dynamic Changes of the Aqueous Concentrations of Cytokines

The aqueous concentration of VEGF in eyes with DME at baseline was 93.8 ± 67.0 (1.99–253.4) pg/mL, which was significantly higher than that in the control eyes (15.0 ± 7.4 [0.99–1.99] pg/mL, P < 0.01). One month after the first injection of faricimab, it significantly decreased to 5.9 ± 23.9 (0.99–142.8) pg/mL (P < 0.01), and at the final third injection, it was 1.4 ± 0.3 (3.1–38.6) pg/mL (P < 0.01), which was significantly lower than in the control eyes (P < 0.01; Fig. 2A). 

Figure 2.

 

Changes in the aqueous concentrations of vascular endothelial growth factor (VEGF) (A), angiopoietin-2 (Ang-2) (B), Ang-1 (C), and Ang-1/2 ratio (D) during 3 consecutive doses of faricimab for DME. Black circles and vertical bars indicate the means and standard deviations in DME, whereas the white circle indicates the value in normal controls. Asterisk (*) indicates statistically significant difference (P < 0.05).

The aqueous concentration of Ang-2 in eyes with DME at baseline was 142.5 ± 237.6 (27.4–1383.0) pg/mL, which was significantly higher than that in the control eyes (29.4 ± 12.4 [17.1–79.9] pg/mL, P < 0.01). One month after the first injection of faricimab, it decreased to 31.6 ± 15.1 (17.1–87.6) pg/mL, and at the final third injection, it was 28.0 ± 8.8 (17.1–49.3) pg/mL, which showed no significant difference from the concentration in control eyes (29.4 ± 12.4 [17.1–79.9] pg/mL; Fig. 2B). 

The aqueous concentration of Ang-1 in DME eyes at baseline was 45.5 ± 14.3 (20.2–94.5) pg/mL, which was not significantly different from that in control eyes (48.0 ± 20.3 [31.4–114.9] pg/mL, P = 0.50). One month after the first injection of faricimab, it decreased to 37.0 ± 14.0 (20.2–77.6) pg/mL (P < 0.01), and, at the final third injection, it was 36.0 ± 12.4 (20.0–70.7) pg/mL (P < 0.01), but this was still not significantly different from the control eyes (P = 0.37; Fig. 2C). 

The calculated Ang-1/2 ratio in DME eyes was 0.57 ± 0.29 (0.05–1.37), which was significantly lower than that in control eyes (1.78 ± 0.69 [0.39–4.44], P < 0.01). After the injection, it increased over the course of treatment, eventually rising to 1.36 ± 0.60 (0.60–2.20, P < 0.01), but was still significantly lower than in the control eyes (P < 0.01; Fig. 2D). 

Relationships Between CFT/BCVA and Cytokines at Baseline

The relationships between CFT/BCVA and aqueous concentrations of VEGF, Ang-2, and Ang-1 were calculated. According to ANOVA, the baseline CFT significantly correlated with aqueous VEGF (P < 0.01), but not with Ang-2 (P = 0.99) or Ang-1 (P = 0.929). The calculated correlation coefficients between the baseline CFT and each cytokine were 0.495 (VEGF: Fig. 3A), 0.142 (Ang-2: Fig. 3B), and 0.106 (Ang-1: Fig. 3C). Interestingly, the baseline CFT negatively correlated with the Ang-1/2 ratio (Fig. 3D: r = −0.345, P = 0.04). 

Figure 3.

 

The relationships between CFT at baseline (at the time of the first injection) and the aqueous concentration of VEGF (A), Ang-2 (B), Ang-1 (C), and the Ang-1/2 ratio. Each theoretical line indicates significant correlations based on linear regression equations.

The baseline BCVA did not significantly correlate with the aqueous VEGF (r = 0.273, P = 0.10), Ang-2 (r = −0.010, P = 0.69), Ang-1 (r = −0.044, P = 0.65), or Ang-1/2 ratio (r = −0.230, P = 0.18). 

Relationships Between CFT and Cytokines at the Third Injection

A multivariate analysis by ANOVA revealed that the CFT at the third injection did not correlate with the aqueous VEGF (P = 0.48). In contrast, the CFT at the third injection significantly correlated with the aqueous Ang-2 (P < 0.01) and Ang-1 (P < 0.01). The calculated correlation coefficients were 0.047 (Fig. 4A: VEGF), 0.264 (Fig. 4B: Ang-2), and −0.344 (Fig. 4C: Ang-1). The CFT at the third injection strongly and negatively correlated with the Ang-1/2 ratio (Fig. 4D: r = −0.657, P < 0.01). 

Figure 4.

 

The relationships between CFT at the time of the third injection and the aqueous concentration of VEGF (A), Ang-2 (B), Ang-1 (C), and Ang-1/2 ratio. Each theoretical line indicates significant correlations based on linear regression equations.

However, BCVA at the third injection did not significantly correlate with the aqueous VEGF (r = −0.112, P = 0.52), Ang-2 (r = 0.143, P = 0.15), Ang-1 (r = 0.179, P = 0.50), or the Ang-1/2 ratio (r = −0.206, P = 0.23). 

Discussion

In this study, three consecutive doses of faricimab in treatment-naïve DME were shown to improve both visual acuity and macular edema over time, as previously reported in the clinical setting,¹⁴ and to reduce not only the VEGF and Ang-2 levels in the anterior chamber, but also the Ang-1 levels, although only slightly. In addition, pretreatment CFT positively correlated with the aqueous VEGF concentration, whereas post-treatment CFT positively correlated with aqueous Ang-2, negatively correlated with aqueous Ang-1 concentration, and strongly negatively correlated with the aqueous Ang-1/2 ratio. 

As for the results of 3 consecutive doses of faricimab, which improved BCVA by −0.09 ± 0.20 logMAR (4.5 letters gain) and regression of edema by −150.5 ± 146.8 µm in treatment-naïve DME, the results do not appear to compare favorably with the results of a previous large-scale prospective study.¹⁵ In our study, the mean baseline BCVA and CFT were 0.32 logMAR (69 letters) and 459.4 µm, which is mild to moderate DME. The eyes included in this study were more mildly affected than those in previous studies,^11,14 which may have led to this clinical outcome owing to the floor and ceiling effects. 

However, it is interesting to note that the relationship between BCVA and CFT significantly correlated before the intervention, but not after the injection of faricimab. It has been reported that in DME, the correlation between BCVA and OCT is modest and that an improvement in edema is not always a surrogate for an improvement in BCVA.⁷ Clinically, we often experience DME cases in which edema improves after anti-VEGF treatment, but the patient's vision does not; however, irreversible visual dysfunction may have already occurred before the initiation of treatment. Anti-VEGF drugs are expected to act on retinal vessels to improve edema by inhibiting leakage, with an improvement in visual acuity as a secondary effect. Therefore, CFT can be used to evaluate the efficacy of anti-VEGF agents directly. 

It has already been reported that the intraocular VEGF and Ang-2 levels are higher in diabetic eyes than in the control eyes prior to therapeutic intervention,¹⁶ and similar results were obtained in the present study. Faricimab is a bispecific antibody against VEGF and Ang-2 and is believed to inhibit both DME and age-related macular degeneration (AMD).¹⁷ A previous study revealed the suppression of aqueous VEGF and Ang-2 after switching from aflibercept to faricimab in patients with AMD.¹⁸ 

In the present study, we confirmed that in clinical practice, the continuous administration of faricimab reduced the anterior levels of both cytokines over time; however, VEGF was significantly lower than that in the control eyes, whereas Ang-2 remained at the same level as that in the control eyes. Although the effects of an excessive suppression of VEGF in the eye by continuous faricimab administration are not clear, the long-term impact on the clinical presentation should be noted. 

A unique aspect of this study is that the kinetics of Ang-1 concentration in the anterior chamber were investigated. In DME before intervention, Ang-1 was not significantly different from that in the control eyes; however, interestingly, faricimab treatment was also associated with a slight but significant reduction in the Ang-1 levels. Relatively little data are available regarding the regulation of native Ang-1, probably because Ang-1 is produced by specific cell types, such as pericytes, smooth muscle cells, and hematopoietic stem cells. A previous in vitro study of cultured pericytes suggested that hypoxia upregulates Ang-1 through hypoxia-inducible factor-2 alpha in this cell type.¹⁹ Another report also demonstrated that the Ang-1 levels were elevated in vitreous samples from patients with PDR.²⁰ In addition to VEGF and Ang-2, Ang-1 is induced under hypoxic and inflammatory conditions. Faricimab treatment for VEGF and Ang-2 might therefore improve retinal hypoxic and inflammatory responses while downregulating the Ang-1 expression in pericytes. However, the precise mechanism underlying these actions is not clear, and further investigation is necessary. Ang-1 is secreted by platelets and pericytes, it then binds to and activates Tie-2 receptors on vascular endothelial cells, and is thought to play a role in vascular homeostasis by maintaining vascular endothelial cell adhesion and pericyte function and inhibiting VEGF receptor 2 activation.⁸ In contrast, Ang-2 is secreted by vascular endothelial cells in response to ischemia or vascular injury, it then competitively binds to Tie-2 receptors, and interferes with their activity.^9,10 

Therefore, the retinal vascular status in DME is likely to be established by the balance between Ang-1 and Ang-2. In the present study, we calculated the Ang-1/2 ratio and investigated whether it could be a biomarker for the DME pathogenesis. 

Interestingly, the Ang-1/2 ratio in DME was significantly lower than that in the control before intervention and increased over time with the continuous administration of faricimab, but it did not recover to control levels. 

Previously, we have already reported that CFT before intervention treatment was significantly correlated with the aqueous VEGF concentration,²¹ and we were able to corroborate this finding. It is interesting to note that the baseline CFT did not correlate with either aqueous Ang-2 or Ang-1, however, it modestly but significantly correlated with the Ang-1/2 ratio, thus indicating that edema was exacerbated when the Ang-1/2 balance tilted toward Ang-2 dominance. This may indicate that the Ang-1/2 balance can be a biomarker for DME severity. 

Furthermore, at the third faricimab injection, although CFT at the third dose of faricimab positively but modestly correlated with the aqueous Ang-2 concentration and negatively but modestly correlated with aqueous Ang-1, the Ang-1/2 ratio negatively and strongly correlated with the CFT, thus indicating that Ang-1/2 may also be an appropriate biomarker for the efficacy of continuous administration of faricimab in improving DME. 

The ratio of Ang-1/2 in serum has been investigated for its efficacy as a biomarker in several pathological conditions, including respiratory disease, cardiovascular disease, kidney disease, and metabolic disorders, such as diabetes. In diabetic retinopathy, the Ang-1/2 ratio in serum decreases with the progression of retinopathy, and it has also been reported to be a more sensitive biomarker, showing a decrease in the early stages of retinopathy compared to the individual measurements of an Ang-1 decrease or an Ang-2 increase.²² To the best of our knowledge, this is the first study to show the Ang-1/2 ratio in the intraocular fluid of patients with diabetic retinopathy and demonstrate its efficacy as a biomarker for diabetic macular edema and a prognostic factor for the treatment course. 

A large prospective comparative study of faricimab and aflibercept showed no difference in the clinical course of BCVA or CFT and no clinical advantage over VEGF suppression plus Ang-2 suppression.¹¹ However, given that the Ang-1/2 ratio strongly correlated with the edema status and improvement, and that faricimab gradually improves the Ang-1/2 ratio during the initial treatment period, but still not reaching control values, it may be worth considering, for example, giving Ang-1 in addition to faricimab to improve the Ang-1/2 ratio over time. 

Although our retrospective results are unique in that we examined not only VEGF and Ang-2, but also the Ang-1 and Ang-1/2 ratios in DME, they are still only preliminary findings due to the limited number of patients in our study. Our results should be confirmed in future large-scale, prospective studies. 

Acknowledgments

Funding information was received by a Grant-in-aid for clinical research in Tokyo Medical University 2022-3. 

Disclosure: M. Shimura, Chugai Pharmaceutical (F, C), Roche (C), Bayer (F, C), Novartis Pharma (F, C), Senju Pharmaceutical (F, C), Santen Pharmaceutical (F), Kowa (F), Otsuka Pharmaceutical (F), Boehringer-Ingelheim (C), Nikki Holdings (C); S. Sasaki, None; R. Nonaka, None; I. Kashiwagi, None; K. Yasuda, None; H. Noma, Chugai Pharmaceutical (F, C), Roche (C), Bayer (F, C); H. Takagi, None