Efficacy and safety of intravitreal anti-VEGF therapy in diabetic retinopathy: what we have learned and what should we learn further?

Abstract

Introduction:

Diabetic retinopathy (DR) is one of the most frequent microvascular complications of diabetes that can lead to blindness. Laser treatment has been the gold standard treatment for diabetic macular edema (DME) and proliferative diabetic retinopathy (PDR) for many years. Recently, the role of vascular endothelial growth factor (VEGF) has been established in the pathogenesis of DR, and the use of intravitreal anti-VEGF therapy has gained popularity for the management of DR.

Areas covered:

This review includes a brief overview of the efficacy and safety of currently available (bevacizumab, ranibizumab, and aflibercept) and potential future (brolucizumab, faricimab, and KSI-301) anti-VEGF agents in patients with DR based mainly on publicly available data from phase 1, 2 and 3 clinical trials.

Expert opinion:

Clinical trials investigating the efficacy of intravitreal bevacizumab, ranibizumab, and aflibercept injections demonstrated favorable functional and anatomical outcomes in patients with DME. Moreover, the use of these anti-VEGF agents showed a significant improvement in the severity of DR. Recent clinical research for future anti-VEGF molecules aims to provide higher target-protein binding affinity and prolonged therapeutic effect. Brolucizumab, faricimab, and KSI-301 are three novel anti-VEGF agents that demonstrate promising data for the management of DME and potentially DR.

1. Introduction

Diabetic retinopathy (DR) is among the most common complications of diabetes and is the leading cause of visual impairment and blindness among the working-age population (aged 15–64) worldwide [1–4]. Diabetic macular edema (DME) by far is the most frequent vision-threatening complication of DR that can develop at any stage of DR and therefore creates a major challenge for the health-care system. Complications of proliferative DR (PDR) include vitreous hemorrhage, tractional retinal detachment, and neovascular glaucoma, all of which are other significant reasons for vision loss in DR [5,6].

Although the complex pathophysiology of DR is not entirely understood, vascular endothelial growth factor (VEGF), particularly VEGF-A, is known to be a key mediator in disease progression and subsequent development of vision threatening complications. In DR, chronic hyperglycemia is thought to damage endothelial cells and basement membrane proteins in retinal vasculature. Such damage, accompanied by vascular hyperpermeability and production of proinflammatory cytokines, can lead to retinal ischemia that can result in a maladaptive increase in VEGF expression. The breakdown of the retinal-blood barrier and resulting angiogenesis from increased VEGF is thought to play a major role in visual loss from DR [7,8]. This understanding has led physicians to use intravitreal anti-VEGF therapy to reduce DR-associated complications and improve vision in affected patients. To date, several VEGF inhibitors, such as bevacizumab (BVZ), ranibizumab (RBZ) and aflibercept (AFB), have been administered by intravitreal injection and have been studied in prospective, randomized phase 2 and phase 3 clinical trials with favorable results in patients with DME and DR.

In this review, we aim to conduct a comprehensive overview summarizing the role, efficacy and safety of several intravitreal anti-VEGF therapies and promising therapeutic targets in patients with DME and DR.

2. Agents in clinical usage

2.1. Bevacizumab

2.1.1. Structure and mechanism of action

BVZ is a 148 kilodalton (kDa) recombinant whole humanized monoclonal immunoglobulin G1κ antibody produced in Chinese hamster ovary (CHO) mammalian cells [9]. It can generally be described as consisting of two main regions, the fragment antigen-binding (Fab) and fragment crystallizable (Fc) regions. The antibodies consist of two light chains (VL-CL) and two complete heavy chains (VH-CH1-CH2-CH3) (kappa subclass) which provide two binding sites for VEGF [10]. Each heavy chain contains an N-linked glycan at a consensus site in the Fc region, which is the site for glycosylation (Figure 1).

Figure 1.

The structure of a full-length IgG1 bevacizumab antibody. Yellow stars represent glycosylation sites.

Fc, fragment crystallizable; Fab, fragment antigen-binding; Variable domains, VL and VH; Constant domains, CL, CH1, CH2, CH3.

The mechanism of action of BVZ is its binding to all active VEGF-A isoforms. Such binding prevents the interaction of VEGF-A with its receptors (VEGFR1 and VEGFR2) on the surface of endothelial cells, reducing endothelial cell proliferation, vascular leakage, and angiogenesis [9]. Table 1 summarizes the structural features and pharmacodynamics of BVZ.

Table 1.

Comparison of structural properties, pharmacodynamics, advantages and disadvantages, and status of different anti-VEGFs used in the clinical trials.

Drug Name	Structure	Size	IV dose (mg)	Biological Target	Advantages	Disadvantages	Status
Bevacizumab	Humanized monoclonal G1ΰ antibody	148kDa	1.25	All isoforms of VEGF-A	Cost-effective	No preservative free vials for intravitreal administration	Off label use
Ranibizumab	Humanized IgG1 kappa isotype monoclonal antibody fragment	48kDa	0.5	All isoforms of VEGF-A	Small molecular weight allows for excellent retinal penetration,Less systemic absorption	One binging site for VEGF, Shorter systemic half-life	FDA approved
Aflibercept	Human recombinant fusion protein: VEGFR-1 and VEGFR-1 extracellular domains fused with IgG1 Fc	115kDa	2	All isoforms of VEGF-A, VEGF-B, PIGF-1 and PIGF-2	Great affinity for VEGF-A, VEGF-B, PIGF-1 and PIGF-2, Less systemic absorption, Allows for less frequent intravitreal injections	Incidence of intraocular inflammation	FDA approved
Brolicizumab	Humanized single-chain antibody fragment	26kDa	6 mg	All isoforms of VEGF-A	Low molecular weight allows for a higher molecular dosing and more durability	Concerns over inflammatory profile, particularly occlusive retinal vasculitis	FDA approved
Faricimab	Humanized bispecific antibody	150kDa	6 mg	All isoforms of VEGF-A and Ang-2	Dual pathway inhibition allows extended dosing intervals	Incidence of intraocular inflammation greater numerically compared to Aflibecept	FDA approved
KSI-301	Anti-VEGF antibody-biopolymer conjugate	950kDa	5 mg	All isoforms of VEGF-A	Large molar dose of drug is designed for more bioavailability and durability	Efficacy and safety are still under investigation	Not-FDA approved

Ang-2, Angiopoietin-2, Fc, fragment crystallizable, FDA, Food and Drug Administration, IV, intravitreal, kDa, kilodalton, mg, milligram, PIGF, placental growth factor, VEFG, vascular endothelial growth factor, VEGFR, vascular endothelial growth factor receptor.

2.1.2. Bevacizumab for diabetic macular edema

A survey to evaluate the safety and efficacy of BVZ for DME was first conducted as phase 2, multicenter, randomized controlled trial of the diabetic retinopathy clinical research (DRCR) network investigators. In this study, subjects with DME were randomized into five arms: 1. focal laser group, 2. 1.25 mg BVZ at baseline and week 6, 3. 2.5 mg BVZ at baseline and week 6, 4. 1.25 mg BVZ at baseline and sham at week 6 and 5. 1.25 mg BVZ with photocoagulation at week 3. Patients treated with both 1.25 mg and 2.5 mg BVZ had a greater reduction in central macular thickness (CMT) at 3 weeks than those with focal laser. However, no statistical difference in CMT was found between patients treated with BVZ and focal laser in long-term follow-up. Patients treated regularly with BVZ had greater improvement in best-corrected visual acuity (BCVA) over one line compared to baseline over the 12 weeks [11]. Another interventional retrospective multicenter study by the Pan-American Collaborative Retina Study Group (PACORES) showed significant improvement in BCVA and CMT at 12 months in both two BVZ regimens (1.25 and 2.5 mg) in patients with DME (101 eyes in 82 patients) [12]. No differences were seen between two different BVZ doses in both studies [11,12]. This positive data have been supported by a prospective single-center, randomized 2-year trial, intravitreal BVZ (1.25 mg, q6) study (BOLT) showing significant improvement in VA (mean BCVA improvement of 8.6 letters) in BVZ arm compared to macular laser treatment arm (mean BCVA loss of 0.5 letters) over 24 months [13]. Table 2 summarizes the clinical trials using intravitreal BVZ therapy in the management of DME.°Results for pooled RISE and RIDE study *Result for the subset of patients with a VA score of 20/50 or worse

Table 2.

List of clinical trials summarizing intravitreal anti-VEGF therapy in the management of diabetic macular edema

Study	No. of Patients (Study Eyes)	Duration, mo	Inclusion Criteria	Intervention	BCVA Change, letters, (Mean)	CMT Change, (Mean)	No of injections, (Mean)
DRCR.net [Citation11]	121	12	BCVA 20/32–20/320 CMT ≥275 μm	1. Focal laser 2. BVZ (1.25 mg, q6) 3. BVZ (2.5 mg, q6) 4. BVZ at baseline (1.25 mg) and sham at 6 weeks 5. BVZ (1.25 mg, q6) with photocoagulation at 3 weeks	−1†, (−6,+5) +5†, (+1. +12) +7†, (+4, +11) +4†, (−3, +7) 0†, (−5, +8)	−40 μm†, (−146, +85) −56 μm†, (−120, −6) −47 μm†, (−125, −16) −5 μm†, (−4,+53) −40 μm†, (−103, +33)	N/A
BOLT [Citation13]	80	24	BCVA 20/40–20/320 CMT ≥270 μm	1. BVZ (1.25 mg, q6) 2. Laser therapy	8.6 ± 9.1 –0.5 ± 10.6	−146 ± 171 –118 ± 112	12†, (11,15)
READ 2 [Citation27]	74	36	BCVA 20/40–20/320 CMT ≥250 μm	1. RBZ (0.5 mg, q8, after initial 2q4 doses) 2. Laser alone 3. RBZ (0.5 mg, q12) + laser	10.3 1.4 8.9	−132 −193 −243	5.4 2.3 3.3
READ 3 [Citation28]	152	24	BCVA 20/40–20/320 CMT ≥250 μm	1. RBZ (0.5 mg, PRN, after 6 initial q4 doses) 2. RBZ (2 mg, PRN, after 6 initial q4 doses)	11.06 6.78	−192.53 –170.64	18.4 17.3
RESOLVE [Citation29]	151	12	BCVA 20/40–20/160 CMT ≥300 μm	1. RBZ (pooled 0.3 and 0.5 mg, q4) 2. Sham	7.8 ± 7.7 –0.1 ± 9.8	−194.2 ± 135.1 –48.4 ± 153.4	10.2 ± 2.5 8.9 ± 3.5
RISE [Citation37]	377	36	BCVA 20/40–20/320 CMT ≥275 μm	1. RZB (0.3 mg, q4) 2. RZB (0.5 mg, q4) 3. Sham	11.0 ± 12.9 14.2 ± 12.8 4.3 ± 14.9	−261.2 ± 196.2 –269.1 ± 178.9 –200.1 ± 215.6	10.6° ±2.6 10.9° ± 2.2 10.0° ± 2.0
RIDE [Citation37]	382	36	BCVA 20/40–20/320 CMT ≥275 μm	1. RZB (0.3 mg, q4) 2. RZB (0.5 mg, q4) 3. Sham	11.4 ± 16.3 10.6 ± 12.9 4.7 ± 13.3	−261.8 ± 180.8 –266.7 ± 207.8 –213.2 ± 193.5	10.6° ±2.6 10.9° ± 2.2 10.0° ± 2.0
Protocol I [Citation30]	235	60	BCVA 20/32–20/320	1. RZB (0.5 mg, q4) +prompt laser 2. RZB (0.5 mg, q4) +deferred laser	8 ± 13 10 ± 13	−167 ± 168 –165 ± 165	13† (9,24) 17† (11, 27)
RESTORE [Citation34]	345	12	BCVA 20/32–20/160	1. RZB (0.5 mg, q4) +sham laser 2. RZB (0.5 mg, q4) + laser 3. Sham injection + laser	6.1 ± 6.43 5.9 ± 7.92 0.8 ± 8.56	−118.7 ± 115.07 –128.3 ± 114.43 –61.3 ± 132.29	7.0 ± 2.81 6.8 ± 2.95 7.3 ± 3.22
REVEAL [Citation35]	396	12	BCVA 20/32–20/160	1. RZB (0.5 mg, q4) +sham laser 2. RZB (0.5 mg, q4) + laser 3. Sham injection + laser	6.6 ± 7.68 6.4 ± 10.67 1.8 ± 8.27	−134.6 –171.8 −57.2	7.8 ± 2.94 7.0 ± 3.07 7.4 ± 3.14
DA VINCI [Citation55]	176	12	BCVA 20/40–20/320 CMT ≥250 μm	1. AFB (0.5 mg, q4) 2. AFB (2 mg, q4) 3. AFB (2 mg, q8) 4. AFB (2 mg, PRN) 5. Laser	11.0 13.1 9.7 12.0 –1.3	−165.4 –227.4 −187.8 –180.3 −58.4	11.7 ± 2.49 10.8 ± 2.87 7.2 ± 1.74 7.4 ± 3.19 N/A
VISTA [Citation56]	461	36	BCVA 20/40–20/320	1. AFB (2 mg, q4) 2. AFB (2 mg, q8) 3. Laser	10.4 10.5 1.4	−200.4 –190.1 −109.8	29.6 ± 9.8 18.1 ± 4.8
VIVID [Citation57]	404	36	BCVA 20/40–20/320	1. AFB (2 mg, q4) 2. AFB (2 mg, q8) 3. Laser	10.3 11.7 1.6	−215.2 –202.8 −122.6	32.0 ± 9.7 18.1 ± 11.9
Protocol V [Citation58]	702	24	BCVA >20/25 CMT ≥250 μm	1. AFB (2 mg, q4) 2. Laser 3.Observation	0.9 ± 6.4 0.1 ± 6.3 –0.4 ± 6.4	−48 ± 65 –41 ± 75 –42 ± 75	8† (6–11) 7† (5–9) 9† (6–11)
Protocol T [Citation59]	660	24	BCVA 20/40–20/320	1. AFB (2 mg, 4q) 2. BVZ (1.25 mg, q4) 3. RBZ (0.3 mg. q4)	18.1 ± 13.8* 13.3 ± 13.4* 16.1 ± 12.1*	−211 ± 155* −185 ± 158* −174 ± 159*	15† (11, 17) 16† (12–20) 15† (11–19)

^†Median.

AFB, aflibercept; BCVA, best corrected visual acuity; BVZ, bevacizumab; CMT, central macular thickness; mo: month; PRN, pro re nata; q, every; RBZ, ranibizumab; VEGF, vascular endothelial growth factor.

2.1.3. Bevacizumab for diabetic retinopathy

Panretinal photocoagulation (PRP) is an established treatment for PDR and severe non-proliferative DR (NPDR) described by the Early Treatment Diabetic Retinopathy Study (ETDRS) Research Group [14]. However, since the introduction of intravitreal VEGF inhibitors, numerous studies have sought to further elucidate the most effective treatments for DR.

One prospective study (IBeHi) of patients with high-risk PDR compared patients who were randomized to PRP (PRP group, n = 15) to those randomized to PRP plus intravitreal injection (PRP plus, n = 15). This study showed that the adjunctive use of intravitreal BVZ with PRP in patients with high-risk PDR was associated with a greater reduction in active neovascularization (NV) than PRP alone, but showed no significant difference in BCVA compared to PRP alone [15]. In a separate retrospective review of 60 consecutive eyes (30 patients) with severe PDR, PRP was performed in two sessions in all eyes in both the control group and the interventional group, and intravitreal BVZ 1.25 mg was performed at 1 week before initial PRP in only the interventional group. Though seven eyes (23.3%) in the control group had worse vision by 2 lines and increased foveal thickness by over 50 μm at 24 weeks, none in the interventional group with BVZ had either worse vision or a significant increase in foveal thickness (P = 0.011) [15]. As a result, it was concluded that intravitreal BVZ before PRP might be able to prevent PRP-induced ocular complications.

A prospective, paired-eye, randomized pilot study was conducted for 12 months to investigate intravitreal BVZ monotherapy without prior PRP for symmetric untreated patients with severe NPDR (n = 5) and patients with PDR (n = 5). The right eye was randomly assigned to treatment with PRP or intravitreal BVZ 2.5 mg every 2 months, and the left eye received the other treatment. Final CMT was significantly decreased in eyes treated with BVZ (197 ± 17 μm) compared to those treated with PRP (243 ± 49 μm), respectively (p = 0.012). There was no difference in BCVA. Moreover, none of 10 eyes treated with BVZ injections every 2 months developed a complication, though two eyes treated with PRP developed vitreous hemorrhage (VH) and three of 10 eyes treated with PRP developed macular edema [16].

A subsequent representative retrospective multicenter interventional case series study was conducted by the PACORES group to evaluate the effects of intravitreal BVZ on retinal NV in patients with PDR [17]. Eighty-one (81) consecutive patients (97 eyes) with retinal NV due to PDR who received at least 1 intravitreal BVZ injection were reviewed in this study. Reinjections of BVZ were performed if there had been a recurrence of NV on ophthalmic examination on fluorescein angiography (FA). In the evaluation of leakage on FA, complete NV regression occurred in 58 (59.7%) eyes, partial resolution in 17 (17.7%) eyes, and no regression in 22 (22.6%) eyes at the end of follow-up. There was a statistically significant change in mean BCVA from 20/125 (0.8 ± 0.4 logMAR) at baseline to 20/60 (0.5 ± 0.4 logMAR) at the last follow-up visit (p < 0.0005). Final BCVA remained stable in 49 (50.5%) eyes, improved by 2 or more Snellen lines in 42 (43.3%) eyes, and decreased by 2 or more Snellen lines in 6 (6.2%) eyes. In particular, it was shown that 8 (42.1%) eyes out of 19 eyes with only intravitreal BVZ treatment without laser therapy or surgery achieved a complete regression of PDR during the 24 months of follow-up. Due to limitations from study design, this study could not answer the question of whether BVZ or PRP was a superior management approach for DR. However, the authors recommended that intravitreal BVZ alone provides a reasonable chance of control of retinal NV in a patient with prior PRP.

Most recently, at the 2021 American Academy of Ophthalmology Annual Congress (New Orleans, November 12–15), it was reported that investigative ophthalmic bevacizumab-vikg formulation (Outlook Therapeutics, ONS-5010) significantly improved visual acuity versus RBZ in patients with neovascular age-related macular degeneration (nv-AMD) over 12 months, according to the phase 3 NORSE findings and outcomes (NCT04516278) [18]. The phase 3 superiority pivotal trial assessed 12-month efficacy and safety of ONS-5010 1.25 mg monthly intravitreal injections (IVIs) versus RBZ IVIs in patients with nv-AMD [18]. Given the positive results, it is very likely that ONS-5010 will be evaluated in DR and DME in the near future to elucidate further the protean applications of intraocular BVZ. If approved, bevacizumab-vikg (Lytenaya) would be the first ophthalmic formulation of BVZ approved to treat retinal conditions. It is a full-length, humanized anti-VEGF recombinant monoclonal antibody that inhibits VEGF and associated angiogenic activity.

2.1.4. Safety of bevacizumab

Ocular and systemic side effects of BVZ have been addressed by a number of studies [19–21]. The most common ocular adverse events (AE) are subconjunctival hemorrhage, regurgitation of drug from the site of injection, transient rise of intraocular pressure, uveitis, lens injury, conjunctival chemosis, and iatrogenic VH [19]

Goldberg et al. reported an outbreak of infectious endophthalmitis in 12-patients after an intravitreal BVZ. The outbreak was attributed to contamination of the BVZ during syringe preparation by the compounding pharmacy [20]. However, a study on 1376 BVZ IVIs by Afarid et al. showed no ocular adverse effects after injections [22]. Sangroongruangsri et al. in the prospective observational study on 5975 patients who received intravitreal BVZ and 397 patients who received intravitreal RBZ, categorized serious systemic adverse events (SAE) into three subgroups due to low prevalence of the adverse systemic events. All-cause mortality, arterial thromboembolic events (ATE), and non-fatal heart failure (HF) were the three main subgroups, which included 66 (1.10%), 46 (0.77%), and 20 (0.33%) of patients, respectively. They reported that there were no significant differences between serious SAE of intravitreal BVZ and RBZ. Their data also showed that 6 treated eyes (0.08%) got endophthalmitis [23].

2.2. Ranibizumab

2.2.1. Structure and mechanism of action

RBZ, which is produced in prokaryotic Escherichia coli, is a 48 kDa recombinant humanized IgG1 kappa isotype monoclonal antibody fragment. It contains only Fab fragment compared to BVZ, which has both Fab and Fc fragments. The fragment consists of a light chain (VL-CL) covalently connected to one heavy chain (VH-CH1), which provides only one binding site for VEGF [10]. Because it lacks the Fc fragment, RBZ does not carry any glycosylation sites (Figure 2).

Figure 2.

The structure of a ranibizumab IgG1 antibody fragment composed of the affinity maturated Fab of bevacizumab.

Fab, fragment antigen-binding; Variable domains, VL; Constant domains, CL, CH1.

The mechanism of action for RBZ is binding with high affinity to the receptor-binding site of all active VEGF-A isoforms, including the biologically active, cleaved form of this molecule, VEGF110. This prevents the interaction of VEGF-A with its receptors (VEGFR1 and VEGFR2) on the surface of endothelial cells, reducing endothelial cell proliferation, vascular leakage, and angiogenesis [24]. Table 1 summarizes the structural features and pharmacodynamics of RBZ.

2.2.2. Ranibizumab for diabetic macular edema

RBZ has undergone significant evaluation in various clinical studies assessing its efficacy and safety. The Ranibizumab for Edema of the Macula in Diabetes – Study 2 (READ-2) was one of the earliest phase 2 clinical trials evaluating 0.5 mg RBZ against laser therapy alone and in combination with RBZ in 1:1:1. The outcomes of the study demonstrated superiority of RBZ therapy (BCVA improvement of 7.4 letters) over the laser therapy alone (BCVA improvement of 0.5 letters) at month 6 [25]. At the end of two years, however, the changes between the three groups were not statistically significant despite RBZ only group showing better improvement in BCVA [26]. Additionally, the combination of laser therapy with RBZ resulted in lower frequency of intravitreal injection at the end of two years. The three-year extension of the READ-2 study showed that aggressive treatment with RBZ therapy maintained the improvement in BCVA and CRT [27]. Following the READ-2 study, the READ-3 Study was conducted, which was a double masked, multicenter clinical trial that evaluated the role of higher dose of RBZ (2.0 mg vs 0.5 mg). The outcomes of the READ-3 clinical trial did not show any additional benefit of 2.0 mg dose of RBZ compared to 0.5 mg dose at the primary end point (ΔBCVA/ΔCRT of 7.01 letters/ 159.70 μm and 9.42 letters/168.58 μm for 2.0 mg and 0.5 mg, respectively). However, the improvement in BCVA at the end of month 24 in the READ-3 study was significantly more in the lower dose group (11.06 letters) compared to the high-dose group (6.78 letters) establishing 0.5 mg as more ideal dose for the management of DME [28]. The change in CRT between the two groups, however, was not significantly different highlighting the fact that the change in CRT does not always directly correlate with the visual improvement.

Safety and efficacy of RBZ in diabetic macular edema (RESOLVE) study was a randomized, sham-controlled, double-masked, multicenter phase 2 clinical trial that compared the two doses of RBZ (0.5 mg and 0.3 mg) against sham treatment [29]. The pooled analysis for the RBZ groups demonstrated significant improvement in BCVA (10.39 letters) compared to sham group (−1.4 letters). Protocol I by the DRCR.net was a phase 3 clinical trial that compared 0.5 mg dose of RBZ along with prompt or deferred laser (≥ 24 weeks) therapy, or triamcinolone plus prompt laser therapy to sham plus prompt laser therapy [30]. The study demonstrated superiority of RBZ plus prompt/deferred laser treatment at the end of 1 and 2-year compared to sham plus prompt laser treatment [30,31]. The study was extended to compare the RBZ plus prompt against RBZ plus deferred laser treatments at 3 and 5-year [32,33]. At both time-points the group with RBZ plus prompt laser treatment demonstrated lower improvement in BCVA compared to RBZ plus deferred treatment (difference in BCVA between RBZ plus deferred vs prompt was 2.9 and 2.6 at 3 and 5 years, respectively). However, the difference was only statistically significant at 3 years (p = 0.02). RESTORE and REVEAL studies were two phase 3 clinical trials that demonstrated the advantage of RBZ-only therapy over laser therapy (Δ BCVA in RESTORE (RBZ group = 6.1 vs laser group = 0.8) and in REVEAL (RBZ group = 5.9 vs laser group = 1.4) [34,35]. Both studies also showed that addition of laser therapy to the RBZ treatment did not provide any additional benefit.

Two landmark phase 3 multicenter clinical trials that led to approval of RBZ by the United States Food and Drug Administration (FDA) were RISE and RIDE. These studies compared two doses of monthly treatment with RBZ (0.3 mg and 0.5 mg) of RBZ to sham injections in patients with DME at the end of month 24. In both the studies, the percentage of patients achieving ≥15 letters improvement from the baseline was significantly higher for patients treated with RBZ (RISE (0.3 mg: 44.8% and 0.5 mg: 39.2%) and RIDE (0.3 mg: 33.6% and 0.5 mg: 45.7%)) compared to sham group (RISE: 18.1% and RIDE: 12.3%) [36]. The studies were extended to month 36 and the patients in the sham groups were allowed to cross over to receive 0.5 mg RBZ. The month 36 outcomes demonstrated sustenance in the efficacy of monthly treatment with RBZ. However, the group that received delayed treatment with RBZ (Crossover from sham group) gained fewer letters compared to the patients receiving RBZ treatment from the start of the study (Sham:2.8 vs. 0.3 mg: 10.6 and 0.5 mg: 11.1 letters) [37]. These results suggested that chronic macular edema may result in permanent vision loss which may not improve with delayed treatment with RBZ.

In various clinical trials, monthly dosing of RBZ has been utilized. However, despite being the optimal regimen, monthly dosing may not be very practical in real-world scenarios. Therefore, some trials have been conducted looking at the pro re nata (PRN) or treat-and-extend (T&E) dosing regimen of RBZ and this has also been studied in extension phases of trials with initial monthly treatment. In the Protocol I, the number of injections received by the patients progressively declined over the 5-year period despite maintenance of BCVA benefits in the RBZ + prompt laser and RBZ + deferred laser arms (8–9 injections in year 1, 2–3 injections in year 2, 1–2 injections in year 3, 0–1 injection in year 4, and no injections in year 5 in RBZ + prompt laser and RBZ + deferred laser arms, respectively). Thus, around 52% to 62% of the patients in the two arms received no injection during year 5. Similarly, year-3 extension of the RESTORE study also demonstrated efficacy of RBZ in maintaining BCVA while lowering the number of injections needed using PRN regimen [38]. The RELIGHT study was a phase 3b, open-label, single-arm study evaluating individualized re-treatment with 0.5 mg RBZ after initial three monthly doses, along with bimonthly follow-up from month 6 to month 18. The results of the study showed that BCVA gain achieved during initial 6 months with monthly follow-ups was maintained at the end of month 18 after a year of bimonthly follow-up [39]. The randomized trial of treat-and-extend (T&E) RBZ with and without navigated laser for DME: TREX-DME study was a multicenter, prospective, RCT with three groups comparing monthly dosing of 0.3 mg RBZ with T&E algorithm using 0.3 mg RBZ with and without angiography guided macular laser photocoagulation [40]. The primary endpoint results of the study showed that there was no significant difference in BCVA gains between the monthly (8.6 letters), T&E (9.6 letters), and T&E with angiography guided macular laser photocoagulation (9.5 letters) groups. In addition, both T&E groups needed significantly lower injections (10.7 and 10.1) compared to monthly treatment arm (13.1, p < 0.001). The addition of laser photocoagulation, however, did not infer any additional benefit to the T&E regimen. Table 2 summarizes the clinical trials using intravitreal RBZ therapy in the management of DME.

2.2.3. Ranibizumab for diabetic retinopathy

RBZ has been shown to not only slow the progression of DR but also improve the severity the DR in various studies. Monthly dosing of RBZ has recently been approved for the management of all forms of DR. In the post-hoc analysis of the RISE and RIDE clinical trial, 37.2% and 35.9% of the patients receiving 0.3 mg and 0.5 mg RBZ demonstrated ≥ 2-step improvement in the DR severity score compared to 5.4% in the sham group at the end of month 24 which persisted at the end of extension phase at 36 months [41,42]. The post-hoc analysis of the READ-3 study demonstrated that monthly dosing of RBZ has shown improvement DR severity similar to other clinical trials as early as month 6 and these changes persisted at the end of the month 24 [43,44]. The improvement in DR status by RBZ was also shown in the DRCR.net Protocol I [45]. Among patients with NPDR the improvement in DR severity was shown by 29–32% of the people along the time of study.

For the management of PDR, DRCR.net protocol I demonstrated efficacy of RBZ in improving the severity in eyes of these patients [45]. The improvement in DR severity among patients with PDR was 23–38% during the course of the study. Protocol S was a noninferiority trial by the DRCR.net which was structured to compare the standard PRP treatment for PDR against 0.5 mg [46,47]. The results of the study demonstrated non-inferiority of RBZ compared to PRP in terms of BCVA improvement in eyes with PDR both at 2- and 5-year end points. The mean BCVA improvement was +2.8 and +0.2 letters at 2 year in the RBZ and PRP group, respectively [46,47]. Surprisingly, the 5-year results showed that the difference in improvement in BCVA between the RBZ and PRP groups had resolved (+3.1 letters for the RBZ group and +3.0 letters for the PRP group, respectively) [47]. It is important to note that approximately 53% of the patients in the PRP group of the study received RBZ for DME during the course of the study that may have some contribution toward the VA changes in this group. Table 3 summarizes the Protocol S study.

Table 3.

List of clinical trials summarizing intravitreal anti-VEGF treatment in diabetic retinopathy.

Study	No. of Patients (Study Eyes)	Duration, mo	Inclusion Criteria	Intervention	Regression of DRP (%)	BCVA Change, letters, (Mean)	CMT Change, (Mean)	No of Injections
Protocol S [Citation46]	394	60	PDR BCVA≥20/320	1. RBZ (0.5 mg, q4 PRN) 2. PRP	46 N/A	3.1 ± 14.3 3.0 ± 10.5	−139 ± 157 –59 ± 102	19.2 ± 10.9 5.4 ± 7.9
CLARITY [Citation63]	232	12	PDR BCVA≥20/320	1. AFB (2 mg, q4 PRN, after 3 initial q4 doses) 2. PRP	64 34	1.3 ± 0.6 –2.9 ± 0.7	−8.9 ± 2.3 24.0 ± 5.5	4.4 ± 1.7-
PANORAMA [Citation64]	402	24	DRSS score 47 and 53 BCVA ≥20/40	1. AFB (2 mg, q16, after 3 initial q4 doses and one q8 interval) 2. AFB (2 mg, q8, after 5 initial q4 doses) 3. Sham	65.2 79.9 15.0	1.7 ± 3.5 1.3 ± 3.49 0.5 ± 3.01	N/A N/A N/A	N/A N/A N/A

AFB, aflibercept; BCVA, best corrected visual acuity; CMT, central macular thickness; DRP, diabetic retinopathy; DRSS, diabetic retinopathy severity scale; mo, month; PDR, Proliferative diabetic retinopathy; PRP, panretinal photocoagulation; q, every; RBZ, ranibizumab.

2.2.4. Safety of ranibizumab

2.2.4.1. Ocular adverse events.

Most recently, the REFINE study evaluated the safety of the RBZ versus laser photocoagulation in 384 DME patients [48]. The most common observed ocular AEs was increased IOP (RBZ 5.2%), followed by vitreous hemorrhage (RBZ 1.6%; laser 5.3%), conjunctival hemorrhage (RBZ 3.6%; laser 1.3%), and dry eye (RBZ 3.6%; laser 1.3%). Ocular serious AEs occurred in 0.3% of patients in RBZ arm due to cataract. The authors reported no cases with endophthalmitis [48]. These findings were consistent with what have been reported from the READ-3 study, which compared the safety outcomes between the 0.5 mg and 2.0 mg RBZ in 152 eyes. They found no ocular SAEs in either of the study groups [28].

In a 24-month follow-up of the RISE and RIDE studies, the most common SAE was vitreous hemorrhage, which occurred in 4 sham-treated and 2 RBZ-treated eyes in RISE and in 3 sham-treated eyes in RIDE. Serious intraocular inflammation (IOI) was uncommon among RBZ-treated patients, occurring only once. SAEs arising from the injection procedure were also uncommon; 1 case of endophthalmitis occurred in RISE and 3 in RIDE, along with three cases of traumatic cataract and one rhegmatogenous retinal detachment out of 10,584 IVIs [36]. The ocular AEs after a 36-month follow-up of the RISE and RIDE study were generally consistent with the results seen at month 24 [37].

2.2.4.2. Systemic adverse events.

Since the SAEs of RBZ are rare, studies with an extremely large study population are required to address the small difference between the systemic complications. The present evidence are not powered sufficiently to project these AEs due to relatively small sample sizes. However, in 2014 Yanagida et al. launched a meta-analysis of the six trials, including 2456 patients with DME, to evaluate the systemic safety of the RBZ. Risk ratio for cerebrovascular accident, myocardial infarction, vascular death, and overall mortality were 0.80 (95% confidence interval, 0.37 1.73; P = 0.57), 0.91 (95% confidence interval, 0.46–1.80; P = 0.78), 1.29 (95% confidence interval, 0.58–2.86; P = 0.53), and 1.92 (95% confidence interval, 0.78–4.73; P = 0.16), respectively. The meta-analysis showed that RBZ is systemically safe in patients with eligible systemic vascular conditions (history of cerebrovascular accident, myocardial infarction, uncontrolled blood pressure, uncontrolled diabetes mellitus, or renal failure were considered as exclusion criteria in the six included trials) [49]. In the REFINE study, hypertension (6.2%), nasopharyngitis (6.5%), upper respiratory tract infection (8.5%), and cough (5.9%) considered as the SAEs. The overall incidence of SAEs was 16.3% [48].

2.3. Aflibercept

2.3.1. Structure and mechanism of action

AFB is a 115 kDa human recombinant decoy receptor, consisting of portions of human VEGF receptor 1 and 2 extracellular domains fused to the Fc portion of human IgG1, produced in Chinese hamster ovary (CHO) K1 cells. It has five putative N-glycosylation sites on each polypeptide chain. The affinity of AFB for VEGF is higher than their natural receptors and is greater than that of RBZ and BVZ [50,51] (Figure 3).

Figure 3.

The structure of aflibercept composed of domains from VEGFR1 and 2 fused to an IgG1 Fc-region.

VEGFR1, vascular endothelial growth factor receptor 1; VEGFR2, vascular endothelial growth factor receptor 2; IgG1, immunoglobulin class G; Fc, fragment crystallization; Constant domains, CH2, CH3.

A major functional distinction between AFB and other anti-VEGF agents is that it blocks VEGF-B, PlGF1, and PlGF-2 in addition to VEGF-A isoforms. AFB antagonizes a broader spectrum of growth factors, and the potency of AFB for blocking VEGF-mediated signaling by VEGF121 and VEGF165 is greater than that for RBZ and BVZ by several orders of magnitude; in particular, the affinity of AFB for VEGF-A165 is 94 times greater than RBZ and approximately 120 times greater than BVZ [50]. VEGF normally binds its receptors as homo- or heterodimers [9]. AFB has a unique binding action and binds to both sides of the VEGF dimer, ‘like two hands on a football,’ forming an inert 1:1 complex, also termed a VEGF trap which minimizes the potential for VEGF to interact with more than one AFB molecule. Compared to anti-VEGF antibodies, such as BVZ and RBZ, AFB binds in a manner that allows the VEGF dimer to interact further with other molecules [51,52]. Table 1 summarizes the structural features and pharmacodynamics of AFB.

2.3.2. Aflibercept for diabetic macular edema

Do and colleagues published the first study worldwide of intravitreal AFB for DME [53]. Subsequently, the DA VINCI study is a randomized, double-masked, multicenter, phase 2 clinical trial that compared the efficacy and safety of intravitreal AFB at 4 different dosing regimens (0.5 mg every 4 weeks (0.5q4); 2 mg every 4 weeks (2q4); 2 mg for every 8 weeks (2q8) after 3 initial monthly doses; 2 mg on an as-needed basis after 3 initial monthly doses) with macular laser treatment [n = 44] in patients with DME [54,55]. The results showed significant improvement in BCVA and CRT (ΔBCVA/ΔCRT of 11.0 letters/ 165.4 μm, 13.1 letters/ 227.4 μm, 9.7 letters/ 187.8 μm, and 12.0 letters/ 180.3 μm for 0.5q4, 2q4, 2q8 and 2 PRN, respectively) from baseline to week 24 (primary endpoint) and week 52 for each AFB regimen compared to macular laser photocoagulation (ΔBCVA/ΔCRT of −1.3 letters/ 58.4 μm). Although all AFB dosing regimens showed significant treatment effects compared with laser treatment, 2q4 had the highest percentages of eyes with VA improvements. However, the differences between the 2q4 and 2q8 groups were attributed to the baseline differences between the two AFB groups instead of the dosing regimens.

Subsequently, two similarly designed, randomized, double-masked, multicenter phase 3 trials, VISTA and VIVID, compared the efficacy of intravitreal AFB injections of 2q4 and 2q8 after 5 initial monthly doses with macular laser treatment in patients with DME [56,57]. Both VISTA and VIVID groups demonstrated significant improvements in mean BCVA gain, proportion of eyes that gained ≥ 15 letters, and diabetic retinopathy severity scale (DRSS) score (visual and anatomical outcomes) with similar overall efficacy in two intravitreal regimens (2q4 and 2q8 groups) compared to laser group at week 52 and 100. Mean CRT changes also were significantly greater in eyes treated in both AFB arms compared with laser group at week 52 and 100. These favorable results maintained at week 148. Although VISTA (ΔBCVA/ΔCRT of 10.4 letters/ 200.4 μm, and 10.5 letters/ 190.1 μm for 2q4 and 2q8, respectively) and VIVID (ΔBCVA/ΔCRT of 10.3 letters/ 215.2 μm, and 11.7 letters/ 202.8 μm for 2q4 and 2q8, respectively) studies showed similar efficacy between 2q4 and 2q8 regimens in terms of visual and anatomic outcomes through week 148, mean injections from baseline to week 148 for 2q8 arms were being noted fewer (18.1 for both VISTA and VIVID) compared to 2q4 arm (29.6 and 32.0 in VISTA and VIVID, respectively). Favorable results obtained from clinical trials have led to the approval of intravitreal AFB for the treatment of DME in July 2014 by the FDA [56,57].

Intravitreal anti-VEGF therapies has been shown to be effective in the treatment of center involved-DME (ci-DME) in patients with decreased visual acuity (VA) (20/32 or worse) in many clinical trials [11,13,27–29,35,37,55,57]. Recently Protocol V study, a randomized and multicenter clinical trial, compared visual loss in patients with good VA (20/25 or better) and ci-DME among eyes managed with monthly intravitreal AFB, laser photocoagulation and observation. No statistically significant difference was observed in mean VA changes and CRT among three groups at 2 years [58]. However, the study was not designed as a comparison of monotherapy of these approaches, and patients in the laser and observation groups received intravitreal AFB during follow-up if they meet the worsening criteria (VA worsened by 1 line at 2 consecutive visits or by 2 lines at 1 visit). Nevertheless, the majority of eyes in the observation group (two-thirds) and laser group (three-fourths) did not meet the worsening criteria for VA and did not receive AFB during 2 years of follow up. Therefore, observation without treatment only if VA worsened was considered as a reasonable approach in the management of ci-DME in eyes with good VA [58]. Table 2 summarizes the clinical trials using intravitreal AFB therapy in the management of DME.

The DRCR.net has also conducted a multicenter randomized, double-masked, multicenter, phase 3 study clinical trial, Protocol T, to compare the efficacy and safety of AFB (2.0 mg), BVZ (1.25 mg) and RBZ (0.3 mg) for ci-DME with vision impairment [59]. All three intravitreal groups showed improved BCVA and decreased the number of injections at 2 years. The subset of patients with better VA (20/32–20/40) had similar visual outcomes among three intravitreal groups (BCVA improvement of 7.8, 6.8, and 8.6 letters for AFB, BVZ, and RBZ, respectively). On the other hand, AFB demonstrated better visual outcomes compared to BVZ and RBZ in the subset of patients with worse VA (20/50–20/320) at year one. The significant difference between AFB (BCVA improvement of 18.1 letters) and BVZ (BCVA improvement of 13.3 letters) continued to be the same at year 2. However, AFB showed no superiority over RBZ in terms of VA (BCVA improvement of 16.1 letters for RBZ) at year 2 [59]. Recently, 5-year follow-up outcomes of Protocol T was published [60]. About two-thirds of the Protocol T study participants were eligible for the continuation of the study. However, outcomes among three anti-VEGF regimens were not comparable as half of the eyes were treated with a different anti-VEGF agent after the initial study ended at 2 years. Overall, 5-year outcomes showed a better mean BCVA than baseline but worse than 2 years (BCVA improvement of 12.1 and 7.4 letters for year 2 and 5, respectively), although there was no significant difference in the mean CRT between 2 and 5 years [60]. Table 2 summarizes the protocol T.

2.3.3. Aflibercept for diabetic retinopathy

AFB not only improves the DR severity but also slows down the progression of DR. In the DA VINCI study, eyes treated with each AFB regimen at week 52 demonstrated more significant improvement in the DR severity score (40%, 31%, 64%, and 32% improvement for 0.5q4, 2q4, 2q8, and 2PRN regimens, respectively) compared to the laser group (12%). Moreover, worsening of DRSS score was rare compared to the laser group (0%, 13%, 0%, and 14% in the 0.5q4, 2q4, 2q8, and 2PRN groups and 24% in the laser group) [55].

The improvement of DR by AFB was also shown in VISTA and VIVID trials. Intravitreal AFB injection has demonstrated a two-step or greater regression in DRSS score at week 100 both in VISTA (37%, 2q4 and 37.1%, 2q8) and VIVID (29.3%, 2q4 and 32.6%, 2q8) compared to laser treatment [56]. The number of patients treated with PRP through week 100 was smaller in two intravitreal AFB groups compared to laser group in both VIVID and VISTA [61]. Based on positive data in VIVID and VISTA, the use of intravitreal AFB in the treatment of DR with DME was approved by the FDA.

Similarly, AFB has been found to be associated with DR regression (31.2% at year 1 and 24.8% at year 2) in patients with NPDR in Protocol T study [59]. Additionally, a greater proportion of eyes with PDR had experienced improvement in DR after treated with AFB (75.9%) compared to RBZ (55.2%) or BVZ (31.4%) at each of the two annual visits [62].

CLARITY is a randomized, single-blinded, multicenter, controlled, phase 2b, non-inferiority trial comparing intravitreal AFB (2 mg on an as-needed basis after 3 initial monthly doses) to PRP in patients with PDR. The study has shown that AFB was non-inferior to PRP, showing a difference of 30% regression in DR ([95% CI: 16–42], p < 0.0001) and 3.9 letters in mean BCVA ([95% CI: 2.3–5.6], p < 0.0001) favoring AFB at week 52. Moreover, vitreous hemorrhage was found to be twice as high in the PRP group compared to AFB treatment group (p = 0.034) [63]. These successful outcomes of AFB may be attributed to its increased affinity for VEGF binding and blockade of other angiogenic pathways such as placental growth factor and galectin-1 [63]. Table 3 summarizes the CLARITY study.

Most recently, PANORAMA (NCT02718326), a randomized multicenter phase 3 clinical trial, evaluated the improvement of diabetic retinopathy in patients with moderately severe and severe NPDR in the absence of ci-DME among two intravitreal AFB 2q8 and 2q16 regimens and sham group. The result of the study showed a 65% and 80% ≥ 2-step improvement in DRSS scores for 2q16 and 2q8 groups, respectively, versus 15% in the sham group at two years (P < 0.0001 for both). The risk of developing vision-threatening complications such as PDR or anterior segment neovascularization significantly reduced by 85% and 88% in both 2q16 and 2q8 groups, respectively, compared to sham. In addition to the findings mentioned above, the proportion of ci-DME was lower in the 2q16 (7%) and 2q8 (8%) groups compared to the sham group (26%), (P < 0.0001 for both) [64]. Table 3 summarizes the PANORAMA study.

2.3.4. Safety of aflibercept

2.3.4.1. Ocular adverse events.

VISTA and VIVID are the two pivotal clinical trials that evaluated the safety and efficacy of the intravitreal AFB injection (IAI) in DME during a 2-year follow-up [56]. In 2016, Ziemssen et al. carried out a post hoc pooled analysis of the VISTA and VIVID trials [65]. There was no significant difference between the ocular and non-ocular AEs across treatment groups. The most common ocular AEs were conjunctival hemorrhage (31.1%), cataract (11.6%), and eye pain (10.7%). No endophthalmitis was reported in the treated eyes. Also, the ocular AEs after 148-week follow-up were consisted with the previous findings [57]. Well et al. performed a multi-center clinical trial on 660 adults with DME to compare the safety and efficacy of AFB, BVZ, and RBZ during a 1-year follow-up. Contrary to VIVID and VISTA studies, they reported 1 (1%) eye with endophthalmitis and 3 (2%) eyes with inflammation after intravitreal AFB injection. Other ocular AEs were increased IOP (12%), vitreous hemorrhage (4%), and cataract (1%) [66].

In 2016, Andrade et al. reported a short-term, prospective clinical trial assessing the efficacy and safety of intravitreal injection of ziv-AFB in DME therapy [67]. Seven patients with DME received intravitreal ziv-AFB every 4 weeks for 6 months. They concluded that the intravitreal injection of the ziv-AFB is safe with no systemic or ocular complication. These findings were supported by the very first pilot study conducted by Do et al. which reported no ocular toxicity after the intravitreal AFB injection [53]. Despite the satisfactory safety outcomes of the two previous studies, it should be considered that these studies were primarily investigation with small sample size, short-term follow-up, and had low internal validity due to the lack of the control group.

2.3.4.2. Systemic adverse events.

According to the VIVID and VISTA study, the most common SAEs were hypertension, nasopharyngitis, and urinary tract infection [56]. Overall rates of arterial thromboembolic events, including vascular deaths, which were defined by Anti-Platelet Trialists’ Collaboration criteria, were low and comparable across the treatment groups. The incidence of death in the intravitreal AFB injection after 4 and 8 weeks, and in the control groups was 5.2, 2.6, and 1.9%, respectively, in VISTA, and 2.9, 4.4, and 0.8% in VIVID [56].

In 2011, the DA VINCI study compared the efficacy and safety of AFB at different dosing with traditional laser photocoagulation in eyes with DME [54]. The most common SAEs in the intravitreal AFB groups were hypertension (9.7%), myocardial infarction (1.1%), cerebrovascular accident (1.1%) and death (1.7%). However, these results should be interpreted with caution, since the SAEs maybe attributed to the underlying disease in these patients rather than the drug itself.

3. Selected VEGF antagonists with strong promise for diabetic retinopathy

3.1. Brolucizumab

Brolucizumab (Novartis Pharmaceuticals, Basel, Switzerland) is a humanized single-chain variable fragment that binds to and inhibits activity of VEGF-A on the VEGF receptor 1 and 2 [68]. Since it is only a single-chain variable fragment, the molecular mass of brolucizumab is only 26kDa, which makes it the smallest of the anti-VEGF agents relative to BVZ (148kDa), RBZ (48kDa) and AFB (97–115 kDa). Table 1 summarizes the structural features and pharmacodynamics of brolucizumab. Because of its small size and therefore high solubility, brolucizumab can be concentrated easily, resulting in delivery of 6 mg of drug delivered via a single 0.05 ml intravitreal injection. The effective dose delivered is about 22 and 11 times more than RBZ and AFB, respectively [68,69]. Such property can potentially translate to longer duration of therapeutic effects, thereby increasing that time period needed between injections. HAWK and HARRIER were two double masked, randomized, clinical trials that indeed demonstrated non-inferiority of brolucizumab administered q8 or even q12 weekly after a loading doze compared to AFB and resulted in its FDA approval for the treatment of neovascular-AMD in 2019 [70,71].

Based on the success of brolucizumab in patients with nv-AMD, KITE (NCT03481660) and KESTERL (NCT03481634) were two phase-3, double masked, multicenter, randomized controlled trials that were launched to compared the efficacy and safety of 6 mg brolucizumab (KITE and KESTERL) and 3 mg brolucizumab (KESTREL) compared to AFB in treatment naïve patients with DME [72]. The 52-week outcomes of the KITE and KESTERL trials were recently reported at the 2021 virtual Stanford Retina Innovation Summit (SRIS, 17 July 2021), among other congresses. Brolucizumab 6 mg was non-inferior to AFB in mean change in BCVA at week 52, with fewer injections. Significant improvements in central subfield thickness (CST) from baseline were achieved with brolucizumab 6 mg. There was a higher proportion of patients with fluid resolution (IRF/SRF) on brolucizumab 6 mg at week 52. More than half of brolucizumab 6 mg patients were maintained on q12w treatment interval up to week 52 immediately after the loading phase [72].

Despite the promised efficacy and increased duration between injections associated with brolucizumab treatment due to higher concentration of drug delivered, recent safety data from the HAWK and HARRIER studies as well as other case reports and case series have shown an increased risk of IOI and occlusive retinal vasculitis (RV) in these patients [73–76]. As reported at the SRIS on 17 July 2021, in KESTREL, one subject in the brolucizumab 6-mg arm had both RV and retinal vascular occlusion (RO); both events resolved without treatment and BCVA at week 52 had increased by 14 letters compared with baseline [72]. In KITE, both RO events were reported as retinal artery occlusion and were not associated with IOI or RV. Thus, despite its strong efficacy, given the potential risks and benefits, and the availability of other therapeutic options, more information is being collected and analyzed to enable clinician-scientists to determine the role of brolucizumab in the management of nv-AMD and DR [72].

3.2. Faricimab

Faricimab (Roche/Genentech, Basel, Switzerland) is a humanized antibody with a bispecific antibody design that targets both VEGF-A and angiopoietin 2 (Ang-2). Angiopoietins regulate vasculogenesis by binding to tyrosine kinase with immunoglobulin and epidermal growth factor homology domain (Tie-2) receptor. Upregulation of Ang-2 has been suggested to play a role in the pathogenesis of macular edema and other retinal vascular diseases [77,78]. By targeting both VEGF-A and Ang-2 pathways, faricimab is designed to improve efficacy and durability in patients with DME. Table 1 summarizes the structural features and pharmacodynamics of faricimab.

A phase 2 randomized, active comparator-controlled, double-masked clinical trial, BOULEVARD, compared the efficacy and safety of faricimab 1.5 mg and 6 mg with RBZ 0.3 mg. Statistically significant improvement in BCVA was achieved in the faricimab 6 mg group (mean gain of +13.9 letters) compared to RBZ 0.3 group (mean gain of +10.3 letters) in DME-naïve patients at week 24. Faricimab also demonstrated dose-dependent improvement in CST and DRSS score with a longer time to retreatment compared with RBZ [79].

Subsequently, YOSEMITE and RHINE, two phase 3 clinical trials, compared faricimab 6 mg every 8 weeks and faricimab 6 mg at a personalized treatment interval (every 4 weeks up to every 16 weeks based on disease activity) with AFB 2 mg every 8 weeks in patients with DME. Both faricimab 6 mg every 8 weeks (ΔBCVA of 10.7 letters and 11.8 letters in YOSEMITE and RHINE, respectively) and faricimab 6 mg at the personalized treatment interval (ΔBCVA of 11.6 letters and 10.8 in YOSEMITE and RHINE, respectively) were not inferior to AFB 2 mg every 8 weeks (ΔBCVA of 10.9 and 10.3 letters in YOSEMITE and RHINE, respectively) at 1 year. Data from the personalized treatment interval group showed improved VA and anatomical outcomes with extended dosing intervals, with more than 50% of patients receiving faricimab every 16 weeks at 52 weeks and more than 70% receiving faricimab every 12 weeks or longer. With regard to safety outcomes, intraocular inflammation was found to be higher in the faricimab 6 mg every 8 weeks group (YOSEMITE n = 5 [1 · 6%], RHINE n = 3 [0 · 9%]) and faricimab 6 mg at the personalized treatment interval group (YOSEMITE n = 7 [2 · 2%], RHINE n = 2 [0 · 6%]) as compared to the AFB 2 mg every 8 weeks group (YOSEMITE n = 3 [1 · 0%], RHINE n = 1 [0 · 3%]). Data from these clinical trials have led to FDA approval of faricimab for the management of DME [78].

3.3. KSI-301

KSI-301 (Kodiak Sciences, Palo Alto, CA) is a novel anti-VEGF antibody biopolymer conjugate (ABC platform), with a 950 kDa molecular weight, that is designed to maintain increased bioavailability and durability in ocular tissues for the treatment of nv-AMD and retinal vascular diseases. Table 1 summarizes the structural features and pharmacodynamics of KSI-301. A phase 1a ascending dose escalation study (1.25 mg, 2.5 mg, or 5 mg) in 9 patients with DME demonstrated a better tolerability without any adverse event in all doses at 12 weeks [80]. The ongoing NCT03790852 randomized, open label, phase 1b study is comparing 2.5 mg and 5 mg KSI 301 in patients with nv-AMD, DME, and retinal vein occlusion. The results at week 20 showed +7.4 letters mean BCVA improvement and −129 μm mean CRT change in the DME group, with 72% of the patients have been extended longer than 3 months after the loading dose without retreatment [81]. A phase 2/3 clinical trial (NCT04049266), DAZZLE, has been conducted to investigate the efficacy and safety of 5 mg KSI 301 at 12–20 weeks intervals and compare with 2 mg AFB every 8 weeks after the loading dose in patients with nv-AMD [82]. Most recently, in February 2022, Kodiak announced that the DAZZLE Study did not meet its primary endpoint of non-inferiority in BCVA, even though the majority of KSI-301-treated patients achieved durable visual gains. The failure to meet the primary endpoint is thought to be in large part due to the impact of undertreatment in some patients. Additional analyses of the DAZZLE Study are being conducted.

4. Conclusion

Although the complex pathophysiology of DR is not entirely understood, VEGF has an important role in disease progression and subsequent development of vision-threatening complications. Phase 2 and 3 clinical trials investigating the efficacy of intravitreal administration of anti-VEGFs demonstrated significant improvement in mean BCVA and reduction in mean CRT compared to traditional laser treatment, especially in patients with worse BCVA with DME. In DME, the combination of laser and anti-VEGF treatments did not show superiority compared to anti-VEGF therapy alone. Moreover, anti-VEGF therapy resulted in a remarkable reduction in DRSS score and NV in patients with NPDR and PDR. These favorable outcomes and the regression of vascular-related pathologies in DR from intravitreal anti-VEGF treatment support the crucial role of VEGF mediators in the pathogenesis of DR.

The safety of BVZ, RBZ and AFB has been evaluated in large-scale clinical trials. The most common ocular AEs include subconjunctival hemorrhage, eye pain, elevated intraocular pressure, and floaters. Intravitreal injection-related ocular SAE such as cataract, endophthalmitis, and rhegmatogenous retinal detachment were uncommon.

Systemic AE such as arterial thromboembolism, myocardial infarction, stroke, and hypertension due to intravitreal administration of anti-VEGF have been found to be lower in clinical trials, though these studies were not powered to assess the safety risks of SAE and have not enrolled patients at increased risk for serious SAEs.

5. Expert opinion

Intravitreal anti-VEGF treatments are the first-line treatment options for diabetic macular edema (DME). Key clinical trials BOLT investigating bevacizumab (BVZ), RIDE and RISE employing ranibizumab (RBZ), and VIVID and VISTA evaluating aflibercept (AFB) have shown significant superiority in visual acuity (VA) and anatomical improvement compared to laser treatment in patients with DME [13,36,56,57]. Protocol T is the first clinical trial comparing the efficacy and safety of currently available anti-VEGFs (BVZ, RBZ, and AFB) in the management of DME. In Protocol T, all anti-VEGF agents were effective equally for eyes with mild VA loss (20/32–20/40). However, AFB was found to be more efficacious for eyes with moderate or worse VA loss (20/50–20/320) [59]. Unlike studies investigating the efficacy of anti-VEGFs in DME with mild to severe VA loss, Protocol V study indicated that observation without treatment only if VA worsens maybe a reasonable and cost-effective strategy for eyes with good VA (>20/25) and center involved-DME [58].

Despite successful outcomes with anti-VEGF treatment in the management of DME, some eyes show incomplete anatomic and/or visual responses to treatment. In regard to inadequate response to anti-VEGF treatment, post hoc analyses of clinical trials have shown that continued treatment with RBZ injections based on the clinical trial algorithm provides a reduction in the CST or CFT and improved VA in eyes with persistent DME through 24 months [83,84]. Similarly, post hoc analysis of the Protocol T study indicated that continued treatment in accordance with the DRCR.net retreatment algorithm, particularly with AFB and RBZ, provided resolution of DME despite initial limited response following 3 or more anti-VEGF injections. The study also demonstrated considerable improvement in VA for all anti-VEGF groups with a low risk of vision loss in eyes with persistent DME through 24 weeks [85]. To address the issue of persistent DME, a phase 2 multi-center randomized clinical trial was conducted to compare combination therapy of RBZ and dexamethasone implant with RBZ treatment alone in patients who had persistent DME following at least 3 anti-VEGF injections. Although the reduction in CST was more likely in eyes treated with dexamethasone administration in addition to continued RBZ IV injections (mean, −110 ± 86 μm) compared to RBZ injections alone (mean, −62 ± 97 μm), the mean change in VA remained similar in both groups at the 24-week follow-up. Additionally, 29% of eyes developed increased intraocular pressures in the combination group, whereas no intraocular pressure elevation occurred in the RBZ group [86].

Current clinical trials have not addressed the issue of switching to other anti-VEGF treatments in eyes who have suboptimal or worsening responses to treatment with one particular anti-VEGF alone. Tachyphylaxis or tolerance has been described as the phenomenon that could be attributed to the lack of response to over repeated one anti-VEGF agent, particularly in NV-AMD. To overcome this resistance, switching agents to another anti-VEGF agent has been suggested to provide improvement in the treatment response [87,88]. Similarly, in DME, some eyes can be unresponsive or show limited response to one particular anti-VEGF but show considerable response when treated with a different anti-VEGF agent. There are studies investigating switching from BVZ and/or RBZ to AFB, which demonstrated improved functional and anatomical outcomes in DME [88–92]. On the other hand, a retrospective study that compared the efficacy of sustained single anti-VEGF and double anti-VEGF switching intravitreal injections in persistent DME following 3 initial consecutive monthly anti-VEGF injections showed no significant visual outcomes between the two treatment strategies. The retrospective nature of the study and the small number of patients poses a limitation for the study. Further prospective randomized studies are required to investigate the effect of changing anti-VEGF agents in the management of persistent DME [93].

Clinical trials for anti-VEGF treatments in DME have shown further benefits in improving the severity of diabetic retinopathy (DR) [41–43,54,55,57,59,61,62]. Most recently, the PANORAMA study (NCT02718326) demonstrated a significant improvement in the DRSS score in patients with moderately severe to severe non-proliferative DR without DME [64]. For proliferative DR (PDR), anti-VEGFs alone or as an adjunctive with pan-retinal photocoagulation (PRP) demonstrated non-inferiority to PRP alone and provides a reasonable reduction in retinal NV, PDR-related complications as well as DME [15,16,46,47,63,94]. However, cost-effectiveness and patients’ compliance for repeated treatments with anti-VEGF are the most important factors to consider compared to PRP in patients with PDR. Therefore, the use of intravitreal anti-VEGF administration may be a better treatment option for PDR patients with concurrent DME.

Among currently available three anti-VEGFs (BVZ, RBZ, and AFB), BVZ is originally developed for the treatment of meta-static colorectal cancers [95,96]. Although BVZ has not been approved for the treatment of retinal vascular diseases by the FDA, it is used off-label by physicians throughout the world due to its economic efficiency in comparison to RBZ and AFB.

Pharmacokinetics of intravitreally injected anti-VEGF agents are important considerations in determining the optimal intraocular concentration and dosing frequency to achieve the maximal therapeutic effect of anti-VEGF. Vitrectomy has been shown to increase clearance and reduce the half-life of anti-VEGF agents in both animal models and human eye studies [97–104]. The enhanced clearance of anti-VEGF after vitrectomy may have an important effect on the dosing regimen as vitrectomized eyes may need more frequent injections and closer follow-ups. However, there is some controversy as to whether the level of anti-VEGF agent may reach an adequate level and duration in vitrectomized eyes as the vitreous concentration of VEGF also reduces following vitrectomy [97,105]. In MODEVA, a prospective, non-comparative, and multicenter clinical trial, a significant VA gain and CST reduction were achieved with 5 monthly injections followed by a PRN regimen in vitrectomized eyes with DME at one year. Mean interval injection of AFB was 5.8 weeks [106]. The outcomes of clinical trials including VITCLEAR (NCT02174211) assessing a pharmacokinetic study of RBZ, AFB, and the effect of vitrectomy in age-related macular degeneration [107] will likely provide a better understanding in the management of DME in vitrectomized eyes.

Favorable results from large-scale clinical trials for anti-VEGF treatments in retinal vascular diseases have led to the development of new anti-VEGF molecules. The goals for future anti VEGF molecules are to bind and inhibit VEGF receptors with a higher affinity and provide longer duration of therapeutic effect and to attack other targets that play a role in the management DR and other retinal vascular diseases. These features are expected to lessen the overall number of IVIs to reduce the treatment burden for DR patients. Brolucizumab, Faricimab, and KSI 301 are three promising anti-VEGF agents currently under investigation for the treatment of retinal vascular diseases.

Brolucizumab is a humanized single-chain variable fragment inhibitor of VEGF-A with a small molecular size of 26 kDa. Its small size enables better target-tissue penetration and allows use of a higher concentration of the agent, leading to a rapid onset of action and a longer duration of therapeutic effects [68,69]. In 2019, the United Sates Food and Drug Administration (FDA) approved brolucizumab for the management of nv-AMD based on the successful results from HAWK and HARRIER phase 3 clinical trials [70,71]. However, the same studies have reported higher IOI rates for brolucizumab compared to AFB (1.4–2.2% vs 0–0.3%, respectively). In addition to mild–moderate IOI that has been reported in these HAWK and HARRIER trials, on 23 February 2020, the American Society of Retina Specialists alerted members to reported 14 cases of RV, of which 11 were identified as occlusive RV after brolucizumab treatment [76]. More recently, the 52-week outcomes of the KITE and KESTREL for DME were reported, demonstrated clear efficacy of brolucizumab for DME [72]. Safety profile at 52-week showed less evidence of IOI than in HAWK and HARRIER [71]. Therefore, a careful evaluation of any active IOI sign is recommended before the initiation of brolucizumab due to increased risk of IOI associated with occlusive RV. If IOI occurs, based on what have been reported, treatment with anti-inflammatory therapy should be initiated as promptly and aggressively as possible.

Faricimab is a bispecific antibody that inhibits both VEGF-A and angiopoietin-2 (Ang-2) pathways. The angiopoietins, Ang-1 and Ang-2, bind to tyrosine kinase with immunoglobulin and epidermal growth factor homology domain (Tie-2) receptor. While Ang-1 provides vascular stability and feedback inhibition of Ang-2 production, Ang-2 has been shown to inhibit the stabilization or maturation features of Ang-1 [108,109]. Therefore, dual blockade of VEGF-A and Ang-2 pathways may act synergistically and result in increased vascular stability and improved outcomes in DME and other retinal vascular disorders. BOULEVARD, a phase 2 clinical trial, showed a significant improvement in BCVA with 6 mg faricimab compared to RBZ 0.3 mg in patients with DME [79]. Subsequently, YOSEMITE and RHINE, two phase 3 clinical trials, established comparable results with intravitreal injection of 6 mg faricimab every 8 weeks, 6 mg of faricimab with personalized treatment interval, and 2 mg AFB every 8 weeks in the management of DME. The incidence of intraocular inflammation was found to be numerically greater in both faricimab groups as compared to AFB group in the same two phase 3 clinical trials [78].

KSI 301 is a novel anti-VEGF antibody biopolymer conjugate, a combination of monoclonal anti-VEGF antibody and a high molecular weight phosphorylcholine biopolymer, that is designed to increase durability and bioavailability in ocular tissue. Phase 1a and ongoing phase 1b studies showed promising results on efficacy, safety and durability of intravitreal KSI 301 administration in patients with DME, nv-AMD and retinal vein occlusion [80,81]. Based on the successful outcomes from phase 1 studies, a double-masked, randomized, multicenter phase 2 trial (NCT04049266), DAZZLE, has been conducted to compare the efficacy and safety of 5 mg KSI 301 at 12–20 weeks intervals and 2 mg AFB every 8 weeks after the loading dose in patients with nv-AMD [82]. However, in February 2022, it was announced that the DAZZLE Study did not meet its primary endpoint; additional analyses are being conducted.

Article highlights.

• Vascular endothelial growth factor (VEGF), particularly VEGF-A, is known to be a key mediator in the pathophysiology of diabetic retinopathy.

• Intravitreal anti-VEGF therapy is the first-line treatment for center-involved diabetic macular edema.

• Currently available intravitreal anti-VEGF agents, bevacizumab, ranibizumab, and aflibercept, have shown significant superiority in visual acuity (VA) and anatomical improvement compared to laser treatment for center-involved diabetic macular edema in the key clinical trials.

• Intravitreal anti-VEGF agents have demonstrated additional benefits in improving severity of diabetic retinopathy in patients with both non-proliferative and proliferative diabetic retinopathy.

• Brolucizumab, faricimab, and KSI 301 are three novel anti-VEGF pharmacologic agents that have demonstrated promising results on efficacy and durability in patients with diabetic macular edema as well as other retinal vascular diseases.

• Recent safety data from the HAWK and HARRIER studies as well as the KITE and KESTREL studies, and other case reports and case series, have shown an increased risk of intraocular inflammation and occlusive retinal vasculitis following intravitreal brolucizumab administration, more among eyes with neovascular age-related macular degeneration than eyes with diabetic macular edema.

Abbreviations

AE

adverse events

AFB

aflibercept

Ang-2

Angiopoietin-2

ATE

arterial thromboembolic events

BCVA

best-corrected visual acuity

BVZ

bevacizumab

CHO

Chinese hamster ovary

CMT

central macular thickness

CRT

Central retinal thickness

DME

diabetic macular edema

DR

diabetic retinopathy

DRCR

diabetic retinopathy clinical research

DRSS

diabetic retinopathy severity scale

ETDRS

early treatment diabetic retinopathy study

FA

fluorescein angiography

Fab

fragment antigen-binding

Fc

fragment crystallizable

FDA

Food and Drug Administration

HF

heart failure

IOI

intraocular inflammation

IVIs

intravitreal injections

kDa

kilodalton

NPDR

non-proliferative diabetic retinopathy

NV

neovascularization

nv-AMD

neovascular age-related macular degeneration

PACORES

Pan-American Collaborative Retina Study Group

PDR

proliferative diabetic retinopathy

PRP

panretinal photocoagulation

PRN

pro re nata

RBZ

ranibizumab

RO

retinal vascular occlusion

RV

retinal vasculitis

SAE

systemic adverse events

T&E

treat-and-extend

VA

visual acuity

VEGF

vascular endothelial growth factor