Efficacy, Safety, and Injection Frequency with Novel Aflibercept 8 mg in Neovascular Age-Related Macular Degeneration: A Comparison with Existing Anti-VEGF Regimens Using a Bayesian Network Meta-Analysis

Abstract

Introduction

Aflibercept 8 mg administered in extended dosing intervals has shown non-inferior visual gains and comparable safety profile to aflibercept 2 mg in the PULSAR pivotal randomized clinical trial and has the potential to reduce the treatment burden of treating neovascular age-related macular degeneration (nAMD). This study aimed at gathering robust evidence to assess the comparative efficacy, safety, and treatment burden of aflibercept 8 mg against other anti-vascular endothelial growth factor (VEGF) agents as ranibizumab, brolucizumab, faricimab, and bevacizumab in patients with nAMD.

Methods

A systematic literature review (SLR) was conducted, targeting clinical trials of anti-VEGF agents in patients with nAMD. The results of the SLR were included in a network meta-analysis (NMA) comparing aflibercept 8 mg to other anti-VEGF treatments in nAMD, considering a 1-year time horizon. Treatment efficacy was assessed based on the change in best-corrected visual acuity (BCVA) from baseline, the proportion of patients gaining or losing 15 Early Treatment Diabetic Retinopathy Study (ETDRS) letters and changes in anatomical outcomes measured as change in central retinal thickness (CRT) or central subfield thickness (CST). Safety was assessed considering the incidence of ocular and non-ocular adverse events. Treatment burden was defined as the mean number of intravitreal injections over the study period.

Results

The base-case NMA involving 21 studies did not show significant differences between aflibercept 8 mg and comparators regarding BCVA change from baseline and proportion of patients with a gain or loss of ≥ 15 letters. On the anatomical endpoints, aflibercept 8 mg was associated with statistically significant improvement in CRT/CST change from baseline compared with ranibizumab in fixed and pro re nata regimens. No significant differences were identified versus the other anti-VEGF. The analysis of the safety outcomes did not identify any significant differences between aflibercept 8 mg and any of the comparators. During the first year of treatment, patients treated with aflibercept 8 mg (following 12- or 16-week injection intervals) received on average 5.9 and 5.1 injections, respectively. For the same period, patients treated with faricimab received from 6.2 to 6.7 injections, patients treated with ranibizumab from 7.62 to 12.14 injections, and patients treated with aflibercept 2 mg up to 7.67 injections.

Conclusion

Aflibercept 8 mg demonstrates a comparable efficacy and safety to currently available anti-VEGF treatments for nAMD, with the potential added benefit of requiring fewer injections. These results suggest that aflibercept 8 mg could be a favourable treatment option for nAMD, achieving sustained disease control while alleviating the burden of injections on patients, caregivers, and healthcare providers.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40123-025-01098-y.

Key Summary Points

Aflibercept 8 mg administered in extended dosing intervals has shown non-inferior visual gains and a comparable safety profile to aflibercept 2 mg in the neovascular age-related macular degeneration (nAMD) population, while also having the potential to reduce the treatment burden.

This study aimed to assess the comparative efficacy, safety, and treatment burden of aflibercept 8 mg against other anti-vascular endothelial growth factors in patients with nAMD in a 1-year horizon.

The base-case NMA did not show significant differences in best-corrected visual acuity (BCVA) change from baseline or in safety outcomes between aflibercept 8 mg and its comparators, while the patients receiving aflibercept 8 mg required fewer injections compared to those on alternative treatments.

These findings suggest that aflibercept 8 mg could be a favourable treatment option for nAMD, providing sustained disease control while reducing the burden of injections for patients, caregivers, and healthcare providers.

Introduction

Neovascular age-related macular degeneration (nAMD) stands as one of the most prevalent causes of blindness worldwide, estimated to impact around 200 million people across the globe, and its prevalence is anticipated to further increase as populations age [1].

Vision loss in nAMD stems from abnormal blood vessel growth and leakage due to vascular endothelial growth factor (VEGF). Therapeutic strategies involving inhibition of VEGF using recombinant anti-VEGF-A proteins have proven efficacious in managing patients with nAMD [2]. Treatment with these agents focuses on neutralizing VEGF binding, reducing fluid leakage, and restraining vessel growth in the retina [3, 4]. Anti-VEGF drugs, such as aflibercept, ranibizumab, bevacizumab, brolucizumab, and faricimab [5, 6], have significantly enhanced patients’ visual outcomes and overall disease prognosis [2, 3], and are now the standard approach to managing patients with nAMD. Anti-VEGF drugs are administered via intravitreal injection or implant [6–15], and treatment involves regular injections into the eyes and frequent patient monitoring. The high treatment burden often results in low patient compliance, leading to suboptimal clinical responses [16, 17]. In contrast, ranibizumab was initially proposed with a monthly dosing in nAMD treatment. In order to reduce the burden of fixed dosing regimens, alternative flexible dosing regimens such as pro re nata (PRN) and treat-and-extend (T&E) were subsequently introduced. The best results in clinical practice have been obtained with proactive regimens such as T&E, where treatment is administered before disease reactivation, allowing the longest interval between injections without jeopardizing the patient’s visual acuity. On the other hand, in a reactive approach such as PRN, the patient is monitored relatively frequently and treatment is given only after disease recurrence [18, 19]. There is still a demand for further burden reduction while maintaining outcomes through newer anti-VEGF agents which may have increased durability. In the PULSAR pivotal clinical trial, treatment-naïve patients with nAMD were randomized to one of three treatment groups—aflibercept 2 mg every 8 weeks, aflibercept 8 mg every 12 weeks, or aflibercept 8 mg every 16 weeks—each regimen after three initial monthly injections. Aflibercept 8 mg demonstrated non-inferior vision gains and safety profile compared to aflibercept 2 mg, achieved with fewer injections spaced at intervals of 12 or 16 weeks (Q12 or Q16) during the first year, and with extensions up to 20 and 24 weeks during the second year of therapy [20, 21].

While robust evidence is available on the novel aflibercept 8 mg compared to the current standard of care, aflibercept 2 mg, there is no direct evidence comparing aflibercept 8 mg to other anti-VEGF agents. Therefore, indirect evidence comparing aflibercept 8 mg with further approved treatment options for nAMD needs to be generated.

In the absence of evidence on head-to-head comparisons, the objective of this analysis was to compare efficacy and safety profiles, as well as treatment burden, for aflibercept 8 mg administered in extended dosing intervals versus anti-VEGF agents approved by the European Medicines Agency (EMA) and commonly used for treatment of nAMD.

Methods

A systematic literature review (SLR) was conducted to identify publications of randomized clinical trials (RCTs) describing the efficacy, safety, and treatment burden of anti-VEGF therapy in treatment-naïve patients diagnosed with nAMD.

Eligibility Criteria and Search Strategy

The selection criteria are reported in Supplementary Table 1, and the search strategy is outlined in Supplementary Table 2 and Supplementary Table 3. The investigated interventions for nAMD involved aflibercept (2 mg, 8 mg), bevacizumab, ranibizumab, brolucizumab, and faricimab. Although bevacizumab for the treatment of ophthalmic conditions was not approved by EMA at the time of the analysis, it has been commonly used in many countries. Therefore, data on bevacizumab were collected and included in a scenario analysis. Specific requirements for the RCTs included a minimum sample size of 40 patients/eyes (in order to avoid “small-study effects” due to publication bias) [22, 23] and outcome results reported after the first year of treatment.

Information Sources

Systematic searches were performed in May 2022 in electronic databases including MEDLINE, MEDLINE In-Process, Embase, the Cochrane Library database, and the ClinicalTrials.gov registry.

Screening and Extraction

Title/abstract and full-text screening were conducted independently by two reviewers to identify studies meeting the SLR eligibility criteria. The decisions from the two reviewers were combined, and discrepancies were resolved by consensus or by a third reviewer.

One reviewer extracted the data, and another reviewer validated the accuracy of the extracted data. The Cochrane risk-of-bias tool was used to assess the quality of the included studies [24]. This SLR adhered to the guidelines outlined by Cochrane [25], National Institute for Health and Care Excellence (NICE) [26], and the Centre for Reviews and Dissemination (CRD) [27].

Input Data Selection for Statistical Analysis

In order to minimize between-trial homogeneity and maximize the external validity, the statistical analysis was based on a subset of studies identified in the SLR, as presented in Table 1. Anti-VEGF therapies eligible for the analysis included approved monotherapies at clinically registered dosages by the EMA, with exclusion of combination therapies and treatment options not approved for the geographical scope (ex-US), such as ranibizumab 0.3 mg.

Table 1.

Studies identified within the SLR, meeting inclusion criteria for the base-case NMA

Study	Sample size	Follow-up [years]	Interventions and comparators	Baseline characteristics
				BCVA	Age	Male	CRT [µm]
				Mean (SD)	Mean (SD)	n (%)	Mean (SD)
ALTAIR	246	2	AFL T&E Q2	54.8 (13.1)	73 (7.9)	87 (70.7)	386.2 (159.2)
			AFL T&E Q4	55.3 (12.0)	75 (8.1)	91 (74.0)	370.3 (120.0)
ARIES	210	2	AFL Q8	61.3 (10.8)	76.6 (8.7)	46 (44.2)	448.3 (133.1)
			AFL T&E Q2	60.2 (12.1)	75.5 (9.0)	44 (41.5)	443.7 (120.0)
ARTIS	94	1	RBZ PRNnL	49.0 (16.2)	69.7 (8.6)	30 (66.7)	468 (202)
			RBZ PRN	52.0 (13.9)	70 (8.8)	30 (61.2)	428 (163)
CANTREAT	580	2	RBZ Q4	59.5	78.7 (8.0)	117 (39.9)	374.2 (111.9)
			RBZ T&E	58.9	78.9 (7.7)	113 (39.4)	382.5 (113.2)
CATT	599	2	RBZ Q4	60.1 (14.3)	79.2 (7.4)	118 (39.2)	251 (122)
			RBZ PRNnL	61.5 (13.2)	78.4 (7.8)	113 (37.9)	247 (122)
DRAGON	333	2	RBZ Q4	53.8 (13.46)	65.6 (8.4)	119 (71.3)	468.8 (178.18)a
			RBZ PRN	53.7 (13.36)	66.8 (8.3)	119 (71.7)	488.5 (195.98)a
Haga 2018	41	1	AFL Q8	63.1	78.5 (8.1)	13 (65.0)	353 (75)
			AFL T&E	56.9	75.5 (6.7)	15 (71.4)	374 (120)
HARBOR	550	2	RBZ Q4	54.2 (13.3)	78.8 (8.4)	113 (41.1)	348.3 (146.3)
			RBZ PRN	54.5 (11.7)	78.5 (8.3)	112 (40.7)	347.8 (143.8)
In-EYE	306	1	RBZ PRN	56.7 (14.7)	77.9 (7.6)c	NR	372.0 (103.4)a
			RBZ Q8	59.2 (13.3)	77.9 (7.6)c	NR	360.0 (91.9)a
			RBZ T&E	54.8 (14.1)	77.9 (7.6)c	NR	378.5 (100.9)a
IVANc	314	2	RBZ Q4	61.8 (15.0)	77.8 (7.6)	129 (41.0)	468 (187)
			RBZ PRN	61.8 (15.0)	77.8 (7.6)	129 (41.0)	468 (187)
LUCERNE	658	1	AFL Q8	58.9 (13.3)	76.1 (8.6)	139 (43)	359.0 (131.1)a
			FAR Q8/Q12/Q16	58.7 (14.0)	74.8 (8.4)	128 (39)	353.1 (120.1)a
Mori 2017	58	1	AFL PRN	67.43 (4.6)	76.5 (NR)	22 (73.3)	278.4 (33.79)
			AFL Q8	65.75 (4.5)	72.8 (NR)	20 (71.4)	314.9 (44.69)
PULSAR	1009	1	AFL Q8	58.9 (14.0)	74.2 (8.8)	148 (44.0)	367.1 (133.6)
			AFL 8 mg Q12	59.9 (13.4)	74.7 (7.9)	153 (45.7)	370.3 (123.7)
			AFL 8 mg Q16	60.0 (12.4)	74.5 (8.5)	158 (46.7)	370.7 (132.7)
RABIMO	40	1	RBZ PRN	60.5 (16.5)d	81 (NR)d	8 (40)	428 (183)d
			RBZ Q8	60.5 (17.5)d	79 (NR)d	6 (30)	370 (92)d
RIVAL	281	2	AFL T&E	65.2 (12.6)	78.7 (7.45)	63 (45.3)	483 (168)b
			RBZ T&E	65.0 (15.4)	76.6 (8.5)	70 (49.3)	468 (151)b
STAIRWAY	71	1	FAR Q12/Q16	60.4 (10.8)	77.7 (8.38)	13 (42)	382.2 (80.9)a
			FAR Q12	57.8 (10.5)	80.3 (7.23)	11 (46)	417.9 (84.3)a
			RBZ Q4	55.3 (12.1)	77.3 (10.29)	6 (37.5)	443.1 (125)a
TENAYA	671	1	AFL Q8	61.5 (12.9)	76.7 (8.8)	126 (37)	356.1 (107.0)a
			FAR Q8/Q12/Q16	61.3 (12.5)	75.9 (8.6)	143 (43)	360.5 (124.1)a
TREND	650	1	RBZ Q4	60.6 (13.92)	75.2 (8.13)	146 (44.6)	497.7 (187.23)a
			RBZ T&E	59.5 (13.21)	75.3 (8.61)	144 (44.6)	504.0 (189.94)a
TREX-AMD	60	3	RBZ Q4	60.3 (2.4)	77 (NR)	22 (37)	533 (45)
			RBZ T&E	59.9 (2.4)	77 (NR)	22 (37)	489 (28)
VIEW 1	909	2	AFL Q4	55.2 (13.2)	77.7 (7.9)	110 (36.2)	313.6 (103.4)
			AFL Q8	55.7 (12.8)	77.9 (8.4)	123 (40.9)	324.4 (111.2)
			RBZ Q4	54.0 (13.4)	78.2 (7.6)	132 (43.4)	315.3 (108.3)
VIEW 2	906	2	AFL Q4	52.8 (13.9)	74.1 (8.5)	133 (43.0)	334.6 (119.8)
			AFL Q8	51.6 (13.9)	73.8 (8.6)	131 (42.8)	342.6 (124.0)
			RBZ Q4	53.8 (13.5)	73 (9)	122 (41.9)	325.9 (110.9)

AFL aflibercept, BCVA best-corrected visual acuity, CRT central retinal thickness, FAR faricimab, n number of patients, NMA network meta-analysis, NR not reported, PRN pro re nata, PRNnL pro re nata (no loading phase), QX every X weeks, RBZ ranibizumab, SD standard deviation, SLR systematic literature review, T&E treat and extend

^aCentral subfield thickness (CST) in µm

^bCentral foveal thickness

^cAggregated for all groups

^dMedian (interquartile range [IQR])

Although brolucizumab was approved by the EMA for treatment of patients with nAMD, the product is not frequently used in clinical practice [28] due to safety concerns identified post-approval not reflected in RCT data [29]. In that sense, brolucizumab was not included in the base case, but was part of the scenario analysis. Bevacizumab also did not meet the inclusion criteria to be part of the base-case analysis (agents approved by EMA at the time of the analysis), but given its clinical relevance [30–32], it was selected to be part of the scenario analysis.

Efficacy outcomes of interest included change from baseline in best-corrected visual acuity (BCVA), the percentage of patients experiencing gains or losses of 15 Early Treatment Diabetic Retinopathy Study (ETDRS) letters, change from baseline in central retinal thickness (CRT) or central subfield thickness (CST). The safety outcomes of interest included the incidence of ocular and non-ocular adverse events, as reported in the published manuscripts of each included trial. Treatment burden was defined as the mean number of injections administered during the first year of treatment.

Statistical Analysis

The analysis for efficacy and safety outcomes was based on a Bayesian network meta-analysis (NMA) approach, which was conducted in accordance with the NICE guidelines [33, 34]. A random-effects model was selected as the base case [34]. As the first step, for each outcome of interest, a feasibility assessment was performed in order to create outcome-specific networks of studies. The feasibility assessment included verification of similarities in outcome definitions, study populations, and study designs across trials, a detailed investigation on the completeness and compatibility of the numerical data, and checking network connectivity.

Mean change from baseline in BCVA score and in CRT/CST were modelled as normally distributed data, using the study-arm-level mean change from baseline as the input. All other endpoints were modelled using a binomial likelihood with logit link except for the gain/loss of at least 15 EDTRS letters, for which the multinomial model was used.

The NMA model used uninformative prior distributions for all parameters of estimations. Estimates were calculated using Markov chain Monte Carlo (MCMC) sampling. Three parallel chains were run with 100,000 iterations and a burn-in of 50,000 iterations. The convergence was assessed by inspection of the trace plots as well as using the Brooks–Gelman–Rubin diagnostic tool.

For each pairwise comparison represented by more than one study, the between-study heterogeneity was assessed. To this end, the Cochran Q statistic with associated p value and I² statistic were calculated using a frequentist approach. The values of 0.1 and 50% were used as significance level and threshold value for the Q test and I², respectively [25].

The network consistency was assessed following the node-splitting approach [34]. For each comparison, if applicable, the direct and indirect evidence were contrasted using the significance level of 0.05.

A meta-analysis was performed to examine the number of injections across treatments; however, adjustments were made to account for the heterogeneity in the dosing regimens across clinical trials. Since fixed-interval regimens that have a pre-specified number of injections regardless of patient clinical status, and flexible regimens were both included in the analysis, the mean number of injections was estimated for each regimen separately. Specifically, data for each treatment regimen were pooled using the arm-based random-effects model within a frequentist framework. To ensure consistency across studies, an adjustment was made for studies reporting results at week 48 by aligning the follow-up duration to 52 weeks. This was achieved by incorporating a scaling factor of 52/48 for flexible regimens and ni₅₂/ni₄₈ for fixed regimens, where ‘ni_t’ indicates the theoretical number of injections up to the time point ‘t’ (i.e., 52 or 48 weeks), according to the injection schedule.

Two sensitivity analyses for the base case were considered to confirm that findings were consistently not dependent on specific assumptions of applied methodology.

The first sensitivity analysis was related to inclusion of the RIVAL study. In the base-case analysis, after discussion with clinical experts, the RIVAL study was excluded. The objective of the RIVAL study was to assess the incidence of geographic atrophy in patients treated with anti-VEGFs. The treatment protocol used aimed to maximize drug exposure, leading to a high number of injections in the aflibercept 2 mg T&E arm that is not observed in any other trial [35, 36]. To ensure thoroughness, sensitivity analyses were undertaken including this study.

Moreover, an in-depth examination of the medical literature and discussion with clinical experts uncovered a potential interference of baseline visual acuity with the observed treatment effects [37]. Therefore, in the second sensitivity analysis, the NMA models for the change in BCVA score were extended by incorporating BCVA at baseline as a covariate via a fixed interaction network meta-regression (NMR) model.

Outputs from the NMAs included estimated comparative efficacy and safety parameters with 95% credible intervals (95% CrI) and surface under the cumulative ranking curve (SUCRA) values. The results are considered statistically significant if the 95% CrI interval excludes zero for difference outcomes or 1 for ratio outcomes.

All analyses were executed utilizing the R computing environment (version 4.2.1, The R Project for Statistical Computing, R Core Team), incorporating the meta and gemtc packages, whereas simulations were generated using WinBUGS [38] and, for the NMR, JAGS (just another Gibbs sampler) software [39].

Ethical Approval

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Results

Systematic Literature Review

The electronic database search identified a total of 7742 citations. After the removal of duplicates, 6309 citations remained, and after title and abstract screening, 399 full-text publications were assessed for eligibility. Finally, 21 studies were considered appropriate for inclusion in the NMA [20, 21, 35, 36, 40–72], as outlined in Fig. 1. Details of the included studies can be found in Table 1.

Fig. 1.

Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) diagram of search results. RAP retinal angiomatous proliferation, FA feasibility analysis, n number of studies, NMA network meta-analysis

Characteristics and Comparability of Included Studies

Mean baseline visual acuity exhibited some heterogeneity between studies, varying from 49.0 to 52.0 letters in the ARTIS trial and 65.75 to 67.43 letters in Mori 2017 [43, 64]. Additionally, there was noticeable variation in the mean baseline CRT/CST measurements, ranging from 247 to 251 µm in CATT [47–52] and 489 to 533 µm in TREX-AMD [66, 67]. The sample size across studies ranged from 40 to 1009 patients. The baseline age ranged from 66 to 79 years. The studies exhibit considerable heterogeneity in gender representation, with male participation ranging from as low as 30% in some trials such as RABIMO [41] to as high as 74% in others such as ALTAIR [40, 42]. The network of studies reporting a change in BCVA score from baseline is displayed in Fig. 2. Networks for other outcomes included in the NMA are reported in Supplementary Figs. 2–5.

Fig. 2.

Network of evidence for BCVA change from baseline. *RIVAL study was included in the sensitivity analysis. AFL aflibercept, BCVA best-corrected visual acuity, FAR faricimab, PRN pro re nata, PRNnL pro re nata (no loading phase), QX every X weeks, RBZ ranibizumab, T&E treat and extend

An additional scenario of analysis, which included studies on brolucizumab and bevacizumab, was considered to estimate the effect of aflibercept 8 mg regimens compared with brolucizumab and bevacizumab and to verify whether the findings observed in comparison with other comparators would remain constant despite the inclusion of additional trials in the network.

Efficacy

No difference was identified when aflibercept 8 mg (Q12 and Q16) was compared with faricimab, ranibizumab, and aflibercept 2 mg in any treatment regimen for the outcomes of BCVA change from baseline and the proportion of patients gaining or losing ≥ 15 EDTRS letters at 1 year (Figs. 3, 4).

Fig. 3.

Forest plot for comparison of AFL 8 mg vs comparators with regard to BCVA change from baseline. AFL aflibercept, BCVA best-corrected visual acuity, CrI credible interval, FAR faricimab, FE fixed effects, HD high dose (8 mg), PRN pro re nata, PRNnL pro re nata (no loading phase), QX every X weeks, RBZ ranibizumab, RE random effects, T&E treat and extend

Fig. 4.

Forest plot for odds ratio for comparison of AFL 8 mg vs comparators regarding proportion of patients who gained or lost ≥ 15 ETDRS letters. AFL aflibercept, CrI credible interval, ETDRS Early Treatment Diabetic Retinopathy Study, FAR faricimab, FE fixed effects, HD high dose (8 mg), OR odds ratio, PRN pro re nata, PRNnL pro re nata (no loading phase), QX every X weeks, RBZ ranibizumab, RE random effects, T&E treat and extend

A statistically greater decrease from baseline in CRT/CST at 12 months was observed with aflibercept 8 mg versus ranibizumab Q4W and Q8W, ranibizumab PRN, and ranibizumab PRNnL (PRN with no loading phase). No difference was observed in change from baseline CRT/CST for aflibercept 8 mg Q12 and aflibercept 8 mg Q16 compared with aflibercept 2 mg, faricimab, and ranibizumab T&E regimens (Fig. 5). Additional results, including the probability of each treatment performing best and the SUCRA, are presented in Supplementary Tables 9 and 10 for all efficacy outcomes. The input data detailing the studies included in all efficacy outcomes are presented in Supplementary Tables 6, 7, and 8. Overall, there were no inconsistencies detected between direct and indirect evidence for any of the efficacy endpoints.

Fig. 5.

Forest plot for comparison of AFL 8 mg vs comparators regarding CRT/CST change from baseline. AFL aflibercept, CRT central retinal thickness, CrI credible interval, CST central subfield thickness, FAR faricimab, FE fixed effects, HD high dose (8 mg), PRN pro re nata, PRNnL pro re nata (no loading phase), QX every X weeks, RBZ ranibizumab, RE random effects, T&E treat and extend

Safety

There were no significant differences observed in the proportion of patients with ocular or non-ocular adverse event (AEs) between aflibercept 8 mg (Q12 and Q16) and the relevant comparators. The NMA results for odd ratios of ocular AEs are presented in Fig. 6, and the results for non-ocular AEs are presented in the supplementary material (Supplementary Fig. 6). Additional results including the probability of each treatment performing best and the SUCRA are contained in Supplementary Tables 13 and 14 for both safety outcomes. The input data detailing the studies included in both safety outcomes are presented in Supplementary Tables 11 and 12.

Fig. 6.

Forest plot for odds ratio for comparison of AFL 8 mg vs comparators with regard to proportion of patients with ocular adverse events. AFL aflibercept; CrI credible interval; FAR faricimab; FE fixed effects; HD high dose (8 mg); PRN pro re nata; PRNnL Pro re nata (no loading phase); QX every X weeks; RBZ ranibizumab; RE random effects; T&E treat and extend

Burden of Treatment (Number of Injections in First Year of Treatment)

The mean number of injections was reported in 19 RCTs [20, 40–59, 63–72]. The analysis indicates that patients treated with aflibercept 8 mg Q16 reported the lowest number of intravitreal injections, with 5.1 injections in the first year of treatment. Patients treated with aflibercept Q12 received 5.9 injections during the first year. These values, along with aflibercept 2 mg PRN, represented the lowest mean number of injections across all treatments studied. The mean number of injections with faricimab regimens ranged from 6.2 (in Q12–Q16) to 6.7 (in Q12) injections. This was comparable to the mean number of injections reported for aflibercept 2 mg T&E (Q4) and ranibizumab PRNnL. The regimen requiring the most frequent administration was ranibizumab Q4 (12.14 injections). Full analysis of the injection frequency is described in Table 2.

Table 2.

Estimated mean (95% CrI) number of injections up to week 52

Regimen	Aflibercept 8 mg	Aflibercept 2 mg	Faricimab	Ranibizumab
Fixed time intervals
Q4	–	–	–	12.14 (11.73, 12.55)
Q8	–	7.67 (7.45, 7.9)	–	7.62 (7.5, 7.74)
Q12	5.9 (5.81, 5.99)	–	6.7 (6.34, 7.06)	–
Q16	5.1 (5.01, 5.19)	–	–	–
Q12/Q16	–	–	6.2 (5.87, 6.53)	–
Q8/Q12/Q16	–	–	6.4 (6.2, 6.6)	–
Flexible time intervals
PRN		5.24 (4.62, 5.85)	–	7.43 (6.48, 8.38)
PRNnL		–	–	6.58 (5.74, 7.43)
T&E Q2		7.17 (7.04, 7.3)	–	–
T&E Q4		6.9 (6.72, 7.08)	–	–
T&E		–	–	9.3 (8.82, 9.78)

CrI credible interval, PRN pro re nata, PRNnL pro re nata (no loading phase), QX every X weeks, SD standard deviation, SLR systematic literature review, T&E treat and extend

Scenario Analysis and Sensitivity Analyses

In the scenario analysis, the network was expanded to include brolucizumab and bevacizumab. The analysis demonstrated that there was no significant difference in the efficacy and safety of aflibercept 8 mg (Q12 and Q16) when compared with brolucizumab and bevacizumab after 1 year of treatment. The analysis also confirmed that patients treated with aflibercept 8 mg (Q12 and Q16) required fewer intravitreal injections than brolucizumab (6.31 injections in Q8W regimen) and bevacizumab (5.9 injections in Q12W, 6.3 injections in PRN, 8.9 injections in T&E, 9.2 injections in Q6W, and 11.9 injections in Q4W regimens). Aflibercept 8 mg was found to be significantly better than bevacizumab at producing a drying effect on the retina.

In the scenario including brolucizumab and bevacizumab, the other findings were in line with the base-case analysis—no significant differences in visual gains or occurrence of AEs were detected between aflibercept 8 mg Q12 and Q16 and all the other treatments. The network plot and the results of the scenario analysis are presented in Supplementary Figs. 11–14.

The NMA results of the sensitivity analysis including the RIVAL study were in line with the base-case analysis, showing no significant differences in efficacy outcomes. As the RIVAL study did not report on ocular and non-ocular AEs, scenario analysis regarding the safety outcomes was not performed.

Furthermore, following the results of the second sensitivity analysis, no evidence was found indicating a relationship between baseline BCVA and these efficacy outcomes.

Discussion

Anti-VEGF drugs have been demonstrated to be efficacious and safe and have become the standard of care for treating patients with nAMD. As the treatment of nAMD has evolved, a new paradigm has arisen. The recently approved treatments, such as aflibercept 8 mg, are now focusing on optimizing treatment efficacy while minimizing treatment burden [73].

In the pivotal PULSAR trial, aflibercept 8 mg (Q12 or Q16) demonstrated non-inferiority versus the current standard of care, aflibercept 2 mg, while improving treatment durability and reducing the injection burden [20, 21].

Despite the positive results registered in clinical trial versus standard of care, however, there was still uncertainty as to how aflibercept 8 mg would compare to flexible regimens and recently approved molecules like faricimab and brolucizumab. The evidence generated with this analysis adds to the data from the pivotal trial and shows that treatment outcomes with aflibercept 8 mg (Q12 and Q16) are comparable to treatment outcomes with aflibercept 2 mg flexible regimens, faricimab, ranibizumab, brolucizumab, and bevacizumab. The results of the base-case analysis showed no significant differences in terms of the mean change in BCVA or the gain and loss of 15 ETDRS letters between aflibercept 8 mg (Q12 and Q16) and the most commonly used anti-VEGFs. The base-case and scenario analyses confirmed that the efficacy of aflibercept 8 mg was comparable to any of the currently used anti-VEGFs.

The analysis of the anatomical outcomes also provided interesting insights. Both aflibercept 8 mg regimens (Q12 and Q16) exhibited superior drying effects, reducing CRT/CST significantly better than ranibizumab. Additionally, the point estimates for this outcome favoured aflibercept 8 mg over other treatment regimens, despite not reaching statistical significance. This finding builds on the PULSAR study outcomes, where aflibercept 8 mg demonstrated superiority versus aflibercept 2 mg in reducing CRT/CST at week 16 [20]. Although further research is necessary, the existing direct and indirect evidence suggests that aflibercept 8 mg might have advantages over the other anti-VEGF in CRT/CST reduction in patients with nAMD after 1 year of treatment.

In PULSAR, aflibercept 8 mg (Q12 and Q16) was associated with a lower mean number of injections when compared to the aflibercept 2 mg fixed regimen. The analysis of the injection burden shows that patients treated with aflibercept 8 mg also reported a lower number of injections than those treated with the aflibercept 2 mg T&E flexible regimen. Patients treated with aflibercept 8 mg (Q12 and Q16) also reported a lower number of injections than patients treated with faricimab, ranibizumab T&E, and all the other anti-VEGF molecules.

The analysis also confirms that the safety profile of aflibercept 8 mg is comparable to the safety profile of all the other anti-VEGFs. The safety profiles of aflibercept 8 mg Q12 and Q16 treatments did not differ significantly from fixed and flexible regimens of aflibercept 2mg, faricimab, ranibizumab, and bevacizumab, demonstrating similar safety outcomes, both ocular and non-ocular. Aflibercept 8 mg also demonstrated a comparable safety profile to brolucizumab based on data from clinical trials, which do not reflect the AEs reported with brolucizumab post-marketing authorization. Similarly, safety risks of retinal vasculitis with or without occlusion have been raised post-authorization with the use of faricimab [6, 74]. However, as no data are currently available on the potential increased risks of these AEs with faricimab, the potential safety differences relative to aflibercept 8 mg are also not quantifiable. Therefore, despite the comparable safety profile of aflibercept 8 mg to brolucizumab and faricimab observed in a clinical trial setting, further research is recommended to understand whether it transposes to clinical practice. It is furthermore recommended that safety results from RCTs be interpreted with consideration of further evidence, such as the most recent real-world data.

In a randomized clinical trial setting, aflibercept 8 mg was shown to be non-inferior to the most commonly used anti-VEGF, and patients treated with aflibercept (Q12 and Q16) reported a lower injection treatment burden. Like all other anti-VEGFs, aflibercept 8 mg needs to demonstrate that the results achieved in the pivotal trials are reproduced in clinical practice. In that sense, the data generated by the multicountry prospective observational study, SPECTRUM [75], are of critical importance. Another aspect to consider is the long-term outcomes and long-term treatment burden. Patients with nAMD receive treatment with anti-VEGFs for several years; therefore, it is important to confirm whether the outcomes registered with aflibercept 8 mg after 1 year of treatment are sustained over the following years. It is also important to assess whether the reduction in injection treatment burden is sustained when moving to the second year of treatment and beyond. Preliminary data from the second year of the PULSAR study [21] suggest that this is the case for the comparison versus aflibercept 2 mg; however, additional data are required to assess whether this remains valid for the comparison with other anti-VEGFs.

From a methodological perspective, this NMA has both strengths and limitations. To reduce the risk of bias, the NMA analysed safety and efficacy outcomes of primary and secondary endpoints from the included clinical trials. The analysis was conducted on data after 1 year of treatment, as most trials were powered to collect data at this time point. Furthermore, the analysis included multiple treatment options in different regimens, with inputs selected in a systematic way, which allowed for the estimation of relative effects between aflibercept 8 mg and other approved treatment regimens for nAMD. Additionally, no inconsistency between direct and indirect evidence was identified for any of the analysed endpoints. The inclusion of the scenario analysis comparing aflibercept 8 mg to brolucizumab and bevacizumab provides useful information for decision-making. Although brolucizumab is not commonly used, it is included in the full treatment landscape for nAMD. While bevacizumab was not approved for the treatment of ophthalmic conditions at the time of the analysis, it has recently received approval from the EMA and therefore can be considered a relevant treatment option [76]. The robustness of findings was confirmed by a sensitivity analysis, including adjustments for baseline BCVA and the inclusion of data from the RIVAL study. These adjustments did not alter the primary findings of our base-case analysis. The SLR and NMA have some limitations. First, substantial between-trial heterogeneity was observed in BCVA change from baseline for some of the pairwise comparisons (AFL Q4 versus AFL Q8, I² = 86%; AFL Q4 versus RBZ Q4, I² = 88%, and RBZ Q4 versus RBZ T&E, I² = 79%). Additionally, variations in the collection of data on ocular and non-ocular AEs across different trials may introduce bias and heterogeneity. Heterogeneity was identified for the following comparisons in non-ocular AEs: AFL Q4 versus RBZ Q4 (I² = 90%) and AFL Q8 versus RBZ Q4 (I² = 74%). The observed heterogeneity may be attributed to differences between trials assessing the same comparisons, including variations in baseline characteristics, treatment protocols, or geographic scope. For instance, the pivotal trials VIEW 1 and VIEW 2, which compared fixed-interval regimens of AFL and RBZ, shared the same design; however, VIEW 1 was conducted exclusively in North America, whereas VIEW 2 included multiple international regions. Similarly, trials comparing RBZ Q4 with RBZ T&E varied in the number of initial injections (e.g., two injections in TREND versus three injections in TREX-AMD and CANTREAT) and in their geographic scope. Consequently, the base-case analysis using the random-effects model reflects the inherent uncertainty in the estimated effects. Nevertheless, the detected heterogeneity does not appear to compromise the robustness of the overall findings. Sensitivity analysis comparing the base-case random-effects model with fixed-effects models—which assume no heterogeneity—demonstrated consistent point estimates, with confidence intervals showing no significant differences between AFL Q8 and other comparators. This alignment between RE and FE models underscores the stability of the conclusions, suggesting that the results are not meaningfully influenced by the observed heterogeneity. Future research should aim to identify specific sources of variability to further enhance the precision and interpretability of these comparisons. Furthermore, the low occurrence of non-ocular AEs leads to poor precision in the NMA-based estimations of odds ratios for this endpoint. Thus, those results should be interpreted with caution.

Moreover, the reliance on the number of injections as the primary indicator of treatment burden, based on RCTs, may not accurately reflect clinical practice. The broader spectrum of treatment burden, including factors such as monitoring visits and diagnostic tests or patient travel and impact on quality of life remains unassessed within the confines of RCTs with protocol-established visits to the clinic. In the real world, however, individualized disease management patterns and localized treatment protocols may result in different numbers of prescribed visits than in the trials [77]. Recommended follow-up times for different anti-VEGF injections, including monitoring requirements, can vary based on efficacy and durability seen in clinical trials [78]. Furthermore, clinic service capacity constraints can lead to desynchronized timing of visits for tests, monitoring, and injections. The cost and time burden patients experience due to the frequent visits often result in treatment non-compliance [79]. Given that delays in diagnosis or (re-)treatment risk permanent vision loss [80], treatment options with demonstrated high durability and consequently reduced number of injections can therefore be extremely important in optimizing visit frequency.

Conclusion

Aflibercept 8 mg has shown comparable visual gains and safety profile compared with other anti-VEGF therapies used in clinical practice, such as aflibercept 2 mg, faricimab, ranibizumab, brolucizumab, and bevacizumab. Furthermore, aflibercept has shown a potentially better drying effect on the retina than some of the currently used treatment options. These results were achieved with fewer injections, which suggests that aflibercept 8 mg with extended-dosing intervals has the potential to improve the management of patients with nAMD.

Supplementary Information

Below is the link to the electronic supplementary material.

Declarations

Conflict of Interest

Paolo Lanzetta consults for Aerie, Allergan, Apellis, Bausch & Lomb, Bayer, Biogen, Boehringer Ingelheim, I-Care, Genentech, Novartis, Ocular Therapeutix, Outlook Therapeutics, and Roche. Jean-Francois Korobelnik consults for Abbvie, Apellis, Bayer, Eyepoint Pharma, Janssen, NanoRetina, Roche, Thea, Carl Zeiss Meditec. Jean-Francois Korobelnik is a member of DSMB for Alexion, Novonordisk, Opthea and Oxular. Olivia Wu has received consultancy fees from Bayer for contribution to this research. Izabella Lunk, Joao Carrasco and Xin Zhang are employees of Bayer. Piotr Wojciechowski, Marlena Wdowiak and Malgorzata Panek declare that they were employed by Putnam at the time the research was conducted, a company which received funds from Bayer to conduct this research.

Ethical Approval

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.