Incidence of New Choroidal Neovascularization in Fellow Eyes of Patients With Age-Related Macular Degeneration Treated With Intravitreal Aflibercept or Ranibizumab

Key Points

Question  What were the incidence of and factors influencing the conversion of the untreated fellow eye in patients who had had neovascular age-related macular degeneration (AMD) and were treated with intravitreal aflibercept or ranibizumab in the Vascular Endothelial Growth Factor Trap-Eye: Investigation of Efficacy and Safety in Wet AMD studies?

Findings  In this second analyses of randomized clinical trial data, the incidence of fellow eye conversion was 26.2% to 32.2% across treatment groups through week 96. Baseline characteristics associated with higher risk of fellow eye conversion were increasing age, female sex, baseline intraretinal fluid in the study eye, and increasing choroidal neovascularization lesion size in the study eye; no difference in the rates of conversion across treatment groups was identified.

Meaning  Per this analysis, patients with active neovascular AMD in 1 eye are at high risk for fellow eye conversion and should be monitored closely.

Importance  Incidence of conversion to neovascular age-related macular degeneration (nAMD) in untreated fellow eyes of patients who are treated for nAMD in 1 eye with anti–vascular endothelial growth factor agents provides important prognostic information to clinically manage patients.

Objective  To investigate the association of treatment assignment (intravitreal aflibercept vs ranibizumab) and baseline characteristics with fellow eye conversion to nAMD in the VEGF (Vascular Endothelial Growth Factor) Trap-Eye: Investigation of Efficacy and Safety in Wet AMD (VIEW) studies.

Design, Setting, and Participants  This post hoc analysis of the VIEW 1 and VIEW 2 studies (randomized, double-masked, active-controlled, multicenter, 96-week, phase 3 trials comparing the efficacy and safety of intravitreal aflibercept in 2457 patients with treatment-naive eyes with nAMD) analyzed a subgroup of participants treated for nAMD in 1 eye who had untreated fellow eyes without neovascularization at baseline. All participants in the VIEW studies were included in 1 of 4 groups: ranibizumab, 0.5 mg, every 4 weeks; aflibercept, 2 mg, every 4 weeks; aflibercept, 0.5 mg, every 4 weeks; or aflibercept, 2 mg, every 8 weeks after 3 injections at 4-week intervals. Data collection in the VIEW studies occurred from July 2007 to August 2011; the data analysis presented in this report took place from April 2016 to November 2018.

Interventions  Patients received no treatment in the fellow eyes unless after conversion to nAMD, when any treatment approved by heath authorities was given per the investigators’ discretion.

Main Outcomes and Measures  Incidence of conversion to nAMD in patients with untreated fellow eyes that had not had clinical signs of neovascularization at baseline.

Results  A total of 1561 participants were included in this analysis. At 96 weeks, 375 patients (24.0%) experienced cases of conversion to neovascular disease in the fellow eye, including 107 of the 399 individuals who received ranibizumab, 0.5 mg, every 4 weeks; 93 of the 387 individuals who received aflibercept, 2 mg, every 4 weeks; 84 of the 387 individuals who received aflibercept, 0.5 mg, every 4 weeks; and 91 of the 388 individuals who received aflibercept, 2 mg, every 8 weeks after 3 doses at 4-week intervals. The rates were 18.1, 16.2, 14.7, and 16.0 per 100 patient-years at risk at week 96, respectively. On multivariate analysis, fellow eye conversion was associated with increasing patient age (per 10 years) at baseline (hazard ratio [HR], 1.20 [95% CI, 1.05-1.36]), female sex (HR, 1.32 [95% CI, 1.06-1.63]), intraretinal fluid in the study eye at baseline (HR, 1.28 [95% CI, 1.02-1.61]), and increasing choroidal neovascularization lesion size (per 10 mm²) in the study eye at baseline (HR, 1.29 [95% CI, 1.06-1.57]). Rates of fellow eye conversion were similar with either of the treatments.

Conclusions and Relevance  In this secondary analysis of randomized clinical trial data, patients with active nAMD in 1 eye appeared to have a high risk for fellow eye conversion. Such patients should be monitored closely.

Patients with unilateral neovascular AMD (nAMD) have considerable risk for conversion to choroidal neovascularization (CNV) in their fellow eye, with reported conversion rates of 12% to 22% at 2 years and 22% to 38.7% over 5 years.^1-5 The largest study (to our knowledge) on the incidence of fellow eye conversion was the Age-Related Eye Disease Study (AREDS), in which 278 of 714 patients (38.9%) with unilateral nAMD experienced conversion to nAMD in their fellow eye over 6.3 years.⁶ While these prior reports provide important data for the general AMD population, data from large intravitreal anti–vascular endothelial growth factor (VEGF) treatment trials appear to show substantially higher nAMD conversion rates for patients, with unilateral nAMD ranging from 19% to 38% at 2 years.^1,7

Some have hypothesized that systemic outcomes of intravitreal anti-VEGF therapy could potentially reduce the risk for conversion in a fellow eye with nonneovascular AMD. Avery et al⁸ were, to our knowledge, the first to report fellow eye outcomes in 2 patients in whom intravitreal bevacizumab given for proliferative diabetic retinopathy in 1 eye appeared to reduce angiographic leakage from neovascularization in a fellow eye. Pharmacokinetic studies have demonstrated that after intravitreal injection, both bevacizumab and ranibizumab are present in the peripheral blood as well as the fellow eye in animals.^9,10 Case reports suggest that among patients with diabetic macular edema, intravitreal anti-VEGF treatment in 1 eye can improve edema in the fellow eye.^11-14 Despite individual cases with supporting animal model data, further clinical data from the Comparison of Age-Related Macular Degeneration Treatments Trials: Lucentis-Avastin Trial (CATT), Anti-VEGF Antibody for the Treatment of Predominantly Classic Choroidal Neovascularization in AMD (ANCHOR) trial, and Minimally Classic/Occult Trial of the Anti-VEGF Antibody Ranibizumab in the Treatment of Neovascular AMD (MARINA) trials demonstrated no significant association of either bevacizumab or ranibizumab with the development of nAMD among the fellow eyes of patients.^1,7 However, the Inhibition of VEGF in Age-Related Choroidal Neovascularisation (IVAN) trial demonstrated that bevacizumab has a significantly greater effect on decreasing circulating levels of VEGF than ranibizumab.¹⁵ Avery et al¹⁶ further demonstrated that, based on evaluation of total drug concentrations, systemic exposure to intravitreal aflibercept and bevacizumab is higher than that of ranibizumab, which was subsequently cleared from the systemic circulation faster than the other 2 medications. Pharmacokinetic information, individual cases among patients with diabetic macular edema, and 2 retrospective studies of bevacizumab improving diabetic macular edema in fellow eyes provide biological and clinical plausibility that systemic effects of intravitreal anti-VEGF medication could have protective outcomes on fellow eyes.^13,14

The apparent discrepancy between the incidence of conversion between the cohort studies of patients who have not yet received anti-VEGF and the clinical trials of the anti-VEGF era, as well as the possibility of systemic effects on fellow nonneovascular eyes, led to this post hoc analysis. We aimed to assess the association of the anti-VEGF medication and dosing regimen with the incidence of conversion to nAMD in the fellow eye of patients treated with ranibizumab or intravitreal aflibercept injection (IAI). Further, we sought to identify baseline characteristics influencing the incidence of conversion to nAMD in the fellow eye.

The VEGF Trap-Eye: Investigation of Efficacy and Safety in Wet AMD (VIEW) 1 and VIEW 2 studies were 2 similarly designed, randomized, double-masked, active-controlled, multicenter, 96-week, phase 3 trials comparing the efficacy and safety of IAI and ranibizumab in patients with treatment-naive eyes with nAMD. Patients were enrolled at 362 sites in the VIEW studies. Each institutional review board or ethics committee approved the study protocols. Both trials were registered with ClinicalTrials.gov (identifiers: NCT00509795 and NCT00637377, respectively), and all patients signed a written consent form before initiation of the study-specific procedures. The VIEW 1 and 2 studies were conducted in compliance with regulations of the Health Insurance Portability and Accountability Act and the tenets of the Declaration of Helsinki.

The design of VIEW studies has been described previously.¹⁷ Patients were randomized in a 1:1:1:1 ratio to 1 of 4 regimens for the first 52 weeks: intravitreal ranibizumab, 0.5 mg, every 4 weeks (Rq4); IAI, 2 mg, every 4 weeks (2q4); IAI, 0.5 mg, every 4 weeks (0.5q4); and IAI, 2 mg, every 8 weeks after 3 initial monthly injections (2q8). From week 52 to week 96, patients continued to receive the same dosage of study drugs as in the first 52 weeks but received injections at least every 12 weeks, with monthly evaluations for interim injections based on prespecified retreatment criteria.

Only patients whose fellow eyes were defined as lacking signs of nAMD at baseline were included in this analysis. The absence of nAMD in the fellow eye at baseline was determined by both an independent, masked central reading center, Digital Angiography Reading Center, and the investigators (K.B.F., R.L.A., N.S., and D.T.). Anatomic evidence of fellow-eye CNV was determined within the Digital Angiography Reading Center database based on fluorescein angiography (FA). The fellow eye had to show the absence of all of the following angiographic characteristics: CNV, vascularized pigment epithelial detachment, subretinal blood, and fibrovascular tissue. Patients whose FA results could not be graded at baseline were excluded from the analysis. Investigator determination was based on masked review of medical and treatment histories (eTable 1 in the Supplement), as noted in case report forms.

Fellow eyes were evaluated monthly and best-corrected visual acuity was obtained at all visits. Fluorescein angiography was performed at baseline and weeks 24, 52, 72, and 96. Optical coherence tomography (OCT) evaluations were not available for the fellow eyes. In addition to the ocular history for both eyes obtained at baseline, concomitant medications, concurrent ocular procedures, and adverse events were collected at every study visit.

Follow-up data from FA and investigator recordings in case report forms were used to determine conversion to nAMD. Patients were categorized as having new CNV after baseline if the reading center noted choroidal neovascularization in the fellow eye at any follow-up visit. Masked investigator documentation of adverse events, concomitant medications, and concurrent ocular procedures during the study provided the criteria for investigator determination. To ensure capture of all conversions, patients meeting either the FA or investigator determination for development of new CNV were included in the analysis.

The CNV events in the fellow eye were analyzed with a predefined approach including incidence and event rate per 100 patient-years at risk in each treatment group. Hazard ratios (HRs) were derived using a Cox proportional hazards model stratified by study (VIEW 1 and VIEW 2). The cumulative incidence of CNV events was determined using Kaplan-Meier methodology, adjusted by study. The patient-years at risk were computed from the time of the first treatment (IAI or ranibizumab) in the study eye to the earliest of 3 dates: conversion, discontinuation from the study, or end of the study. Stratified Cox proportional hazards models were used in both the univariate and multivariate analyses to evaluate the association between identified risk factors and incidence of CNV conversion. Baseline risk factors evaluated were age and sex, as well as key study eye features, such as central retinal thickness (CRT) measured by time-domain OCT, intraretinal fluid status, leakage area, lesion size, best-corrected visual acuity, lesion type, size of pigment epithelial detachment, and presence of drusen. All analyses were performed using SAS version 9.4 (SAS Institute Inc). All results with P values less than .05 (2-sided) were considered nominally significant. Data collection occurred from July 2007 to August 2011 for the original VIEW 1 and VIEW 2 studies; data analysis occurred from April 2016 to November 2018 for this study.

Of the 2457 eyes randomized into the VIEW studies, 2419 patients (Rq4, 595 patients; 2q4, 613 patients; 0.5q4, 601 patients; and 2q8, 610 patients) received study medication (safety analysis set) and were included in these analyses (Figure 1). At baseline, 856 patients were determined to have a history of nAMD in the fellow eye (eTable 2 in the Supplement). There were insufficient data to make an assessment in 2 patients. The records for the remaining 1561 fellow eyes provided sufficient evidence to conclude that nAMD was not present.

Baseline demographics and study eye characteristics were similar across all treatment groups for patients without history of nAMD in the fellow eyes (eTable 3 in the Supplement). In general, the mean (SD) age of these patients was 75 (8.6) years, with 857 of 1561 patients (54.9%) being women. At baseline, in the study eyes, the mean (SD) best-corrected visual acuity was 53.3 (13.8) Early Treatment Diabetic Retinopathy Study letters and mean (SD) CRT as measured by time-domain OCT was 302.7 (134.3) μm. Overall, at baseline, 568 of 1561 eyes (36.4%) had the occult CNV lesion type, 544 of 1561 eyes (34.8%) had the minimally classic CNV lesion type, 439 of 1561 eyes (28.1%) had the predominantly classic CNV lesion type, and 10 of 1561 eyes (0.6%) had the missing CNV lesion type. The lesion size in about 1242 of 1561 eyes (79.5%) was less than 10.4 mm². As per the study protocol, all study eyes had retinal fluid at baseline, of which 929 of 1561 eyes (59.5%) had intraretinal fluid present. The mean leakage amount on FA at baseline was 7.8 (4.9) mm², and retinal pigment epithelial elevation or detachment was present in 1152 of 1561 study eyes (73.8%).

Over the 96 weeks of follow-up, there were 375 cases of conversions to neovascular disease in the fellow eyes of patients with nAMD (24.0% of 1561 study eyes; Rq4, 107 of 399 eyes [26.8%]; 2q4, 93 of 387 eyes [24.0%]; 0.5q4, 84 of 387 eyes [21.7%]; and 2q8, 91 of 388 eyes [23.5%]). The event rates per 100 patient-years at risk at 96 weeks were 18.1 in the Rq4 group, 16.2 in the 2q4 group, 14.7 in the 0.5q4 group, and 16.0 in the 2q8 group (Figure 2). The cumulative incidence per Kaplan-Meier analyses was 16.6% in the Rq4 group, 18.2% in the 2q4 group, 15.6% in the 0.5q4 group, and 14.2% in the 2q8 group at week 52 and 32.2% in the Rq4 group, 28.2% in the 2q4 group, 26.2% in the 0.5q4 group, and 29% in the 2q8 group at week 96. The rates per 100 patient-years at risk were similar at weeks 52 and 96. The differences between Rq4 and any of the IAI treatment groups, as measured by the HR, were not significant at weeks 52 and 96 (Figure 2).

For each potential baseline covariate, we fitted the Cox proportional hazards model. The results are summarized in Table 1. The univariate analyses suggest that age (HR, 1.25 [95% CI, 1.10-1.41]; P = .01), sex (HR, 0.71 [95% CI, 0.58-0.87]; P = .01), CRT (HR, 1.08 [95% CI, 1.01-1.16]; P = .03), intraretinal fluid (HR, 1.45 [95% CI, 1.17-1.81]; P = .01), leakage area (HR, 1.35 [95% CI, 1.11-1.64]; P = .01), and CNV lesion size (HR, 1.34 [95% CI, 1.11-1.62]; P = .01) at baseline in the study eyes may have important associations with the incidence of CNV conversion in the fellow eye. Analysis of the Pearson correlation indicated that leakage area and CNV lesion size were highly correlated (r = 0.89).

Hence, a multivariate Cox proportional hazards model using age, sex, baseline CRT, intraretinal fluid, and CNV lesion size as factors was performed to confirm and quantify the extent of their associations with CNV conversion in the fellow eye (Figure 3). When the correlations among the covariates were taken into account, conversion rates increased with increasing age (per 10 years) (HR, 1.20 [95% CI, 1.05-1.36]) (Figure 3). Women were 1.32 (95% CI, 1.06-1.63) times more likely than men to experience a conversion to CNV in the fellow eye. Conversion rates were 1.28 (95% CI, 1.02-1.61) times higher in fellow eyes when the study eye had intraretinal fluid at baseline. Conversion rates were also higher with increasing CNV lesion size (per 10 mm²) in study eye (HR, 1.29 [95% CI, 1.06-1.57]) (Figure 3). Conversion rates were not affected by baseline CRT or treatment assignment. A model containing each IAI treatment group separately did not improve the fit when compared with the pooled IAI treatment groups; the difference in χ²₂ was 0.50 (P = .78).

The conversion rate was calculated by the number of risk factors initially identified at baseline. It was found that the incidence of fellow eye conversion increased with the number of baseline factors (across all groups: patients with 1 factor, 7.5 conversions per 100 person-years at risk; patients with ≥6 factors, 21.5 conversions per 100 person-years at risk; Table 2). This association was independent of treatment assignment.

The cohort of patients with eyes undergoing anti-VEGF treatment for unilateral nAMD experienced conversion of nearly one-third of their fellow eyes to nAMD within 24 months in this post hoc analysis of the VIEW studies. Post hoc analyses of the MARINA, ANCHOR, and CATT studies, and now the VIEW 1 and VIEW 2 studies, have demonstrated a higher incidence of fellow eye conversion than previously reported in the era before anti-VEGF was used. This may be a function of monthly monitoring as opposed to prior studies with follow-up of 3 to 6 months, longer periods of follow-up, and the specific inclusion criteria. For example, in the MARINA study, patients were required to have “presumed recent progression of disease, as evidenced by observable blood, recent vision loss, or a recent increase in the lesion’s greatest linear diameter,”^18(p1420) while the VIEW studies required active primary subfoveal CNV and ANCHOR required primary or recurrent subfoveal CNV. This study provides key evidence that patients with active disease of recent onset in 1 eye are at a much higher risk for fellow eye conversion than previously thought from prior cohort studies, and they should be monitored more closely, as they were in the MARINA, ANCHOR, and VIEW studies.

The AREDS group found that increasing age, female sex, and drusen severity increased the risk of progression to advanced AMD.¹⁹ Although the VIEW 1 and 2 studies did not have data on drusen severity available, both increasing age and female sex were associated with a higher risk for fellow eye conversion. Neither the drug or regimen assigned to the study eye appeared to have a significant association with the incidence of fellow eye conversion, but conversion was higher in patients whose study eyes had increasing CNV lesion size and/or intraretinal fluid at baseline. These clinical features appear to indicate more severe disease activity, and thus patients with these features may warrant more frequent monitoring. This study provided additional clinical imaging factors associated with conversion while other studies (eg, Complications of Age-Related Macular Degeneration Prevention Trial [CATT], which determined risk factors for conversion in eyes with large drusen) found that, along with age, focal hyperpigmentation of more than 250 μm and active cigarette-smoking status were associated with greater risk for conversion.²⁰ These factors along with acute progression should also be taken into consideration in determining follow-up schedules and the potential need for other means of at-home monitoring.

Prior studies have provided pharmacokinetic evidence of intravitreal anti-VEGF drugs, leading to systemic absorption and decreased plasma VEGF levels. Most notably, Avery et al¹⁶ evaluated total drug concentrations and found that systemic exposure to intravitreal aflibercept and bevacizumab was higher than ranibizumab, which was subsequently cleared from the systemic circulation faster than the other 2 medications. Another prospective randomized clinical trial demonstrated similar findings with patients receiving intravitreal aflibercept to have significantly lower plasma VEGF levels, even 4 weeks after intravitreal aflibercept administration, compared with ranibizumab.²¹ Although the logical presumption would be that intravitreal aflibercept would be more likely to have systemic effects, the conversion rates in patients with IAI-treated eyes (ranging from 26.2% to 29.0%) were not significantly different from the conversion rate in the ranibizumab-treated eyes (32.2%). It is conceivable that this tendency could become significant with larger numbers, but this study is larger than others that have assessed fellow eye outcomes to date. Further, the incidence of conversion was similar across the different dosage strengths and intervals of IAI (2q4, 28.2%; 0.5q4, 26.2%; 2q8, 29%). Further studies are warranted to reconcile the clinical data showing no difference between IAI and ranibizumab in terms of fellow eye conversion and the pharmacokinetic data suggesting higher systemic exposure with IAI compared with ranibizumab.

To our knowledge, this study is the first to demonstrate the association of IAI (compared with ranibizumab) with outcomes in fellow eyes and provide important information on the conversion of fellow eyes to nAMD, especially in patients experiencing acute disease. The strengths of this study include data from randomized clinical trials, use of independent masked reading centers, monthly follow-up, fixed dosing in study eyes and the comparison of 2 different anti-VEGF medications.

The limitations of this study are inherent to the post hoc analysis. The lack of data from OCT evaluations may also be a limiting factor in the current environment, because OCT is primarily used for diagnosis and follow-up of nAMD. Recent evidence suggests that OCT angiography may also be useful for detecting subclinical, nonexudative, type 1 neovascularization in eyes with AMD.^22-24 Neovascular flow detected with OCT angiography appears to correlate with plaques of indocyanine green angiographic staining that were previously shown to correspond to subclinical neovascular lesions.^25,26 Eyes with subclinical neovascular lesions appear to be at higher risk for exudation than eyes lacking this finding, and OCT angiography is currently being used in several ongoing treatment trials of for both neovascular and nonneovascular AMD.^23,26

These findings along with baseline risk factors should be used to identify and more closely monitor patients at higher risk with a combination of more frequent clinic visits, at-home self-monitoring techniques, or telemedicine modalities with mobile patient-operated or other fundus cameras or OCT. The substantial risk of conversion in the fellow eyes of patients with nAMD in 1 eye highlights the need for further studies examining the use of preventive measures in patients at risk, such as the Prophylaxis Against Conversion to Neovascular AMD study, which examines the efficacy of prophylactic IAI in patients whose eyes are at high risk of conversion to nAMD.²⁷

Accepted for Publication: March 25, 2019.

Corresponding Author: K. Bailey Freund, MD, Vitreous Retina Macula Consultants of New York, 460 Park Ave, New York, NY 10022 (kbfnyf@aol.com).

Published Online: July 11, 2019. doi:10.1001/jamaophthalmol.2019.1947

Author Contributions: Dr Freund had full access to all the data in study and takes responsibility for the integrity of the data and the accuracy of data analysis.

Concept and design: All authors.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Thompson.

Administrative, technical, or material support: Parikh, Saroj, Freund.

Supervision: Parikh, Avery, Freund.

Conflict of Interest: Dr Avery reported personal fees from Allergan, Alimera, Amgen, Bausch & Lomb, Ocular Therapeutix, Iridex, RegenXbio, Santen, Genentech, and Eyepoint; personal fees and stock from Novartis and Regeneron; and stock in Replenish outside the submitted work. In addition, Dr Avery has a patent on retinal drug delivery issued and licensed. Dr Saroj was an employee of Regeneron at the time of the study and is currently a paid consultant for Regeneron; Dr Saroj is also a consultant for Aerie, Adverum, Allegro, Apellis, RegenxBio, and SamaCare and is an equity owner of Allegro, Pr3vent, and SamaCare, outside the submitted work. Dr Thompson reports personal fees from Regeneron during the conduct of the study. Dr Freund reported personal fees from Zeiss, Novartis, Optovue, Allergan, and Heidelberg Engineering; grants from Regeneron; and grants and personal fees from Genentech/Roche during the conduct of the study. No other disclosures were reported.

Funding/Support: The VEGF (Vascular Endothelial Growth Factor) Trap-Eye: Investigation of Efficacy and Safety in Wet AMD (Age-Related Macular Degeneration) 1 and 2 studies were funded by Regeneron Pharmaceuticals Inc and Bayer HealthCare.

Role of the Funder/Sponsor: The sponsors participated in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.