Randomized Trial of Treat-and-Extend Intravitreal Aflibercept for Radiation Retinopathy: 1-Year Outcomes

Abstract

Purpose: Radiation retinopathy (RR)-related macular edema commonly causes poor visual acuity outcomes in patients previously treated with ocular radiation therapy. Current treatments are not US Food and Drug Administration (FDA)-approved and prior studies have variable outcomes. We performed a multicenter, prospective, randomized clinical trial to assess the safety and efficacy of 2 mg intravitreal aflibercept injections (IAIs) for the treatment of RR.

Methods: Thirty-nine eyes in 39 patients with RR-related macular edema causing vision loss were assigned randomly to cohorts (1:1 ratio) in which patients either did or did not receive a loading dose of 3 IAIs followed by a treat-and-extend regimen. The primary outcome measure was the mean Early Treatment Diabetic Retinopathy Study best-corrected visual acuity (BCVA) change from baseline.

Results: Of the 39 randomized patients, 30 (76.9%) completed the year 1 follow-up visit. The overall mean BCVA change from baseline was 4.3 letters (P = 0.087), with 1.57 letters and 6.69 letters gained in cohorts 1 and 2, respectively (P = 0.31). There was a significant difference in central retinal thickness (CRT) from baseline to week 52 overall (484.4 µm to 326.5 µm) and within cohorts 1 (441.2 µm to 311.1 µm) and 2 (522.3 µm to 339.9 µm), respectively (P < 0.001). A total of 96.7% of the patients had visual acuity of 20/200 or better, and 30.0% improved 10 or more letters.

Conclusions: Aflibercept may improve CRT and may prevent vision loss in patients with RR using a treat-and-extend regimen through 52 weeks of therapy. Larger, multicenter studies are needed to confirm these findings.

Radiation retinopathy (RR) is a chronic, vision-threatening vasculopathy that develops as a side effect of radiation therapy for the treatment of intraocular or orbital tumors.^1,2 The incidence of RR is dependent on irradiation type and dosage, tumor size and location, and the presence of systemic diseases like diabetes.^2–4 The clinical manifestations of RR, like those of diabetic retinopathy (DR), include vascular exudation, intraretinal and vitreous hemorrhages, capillary non-perfusion, neovascularization, and macular edema.^5–7 

RR is a progressive disease that, if left untreated, can result in vision loss when complicated by RR-related macular edema.⁵ In the Collaborative Ocular Melanoma Study (COMS), in which patients with choroidal melanoma underwent either iodine (I¹²⁵) brachytherapy or enucleation, the percentage of patients with visual acuity (VA) of 20/200 or worse increased from 10% to 45% at 3 years after irradiation.⁸ Currently, there is no standard treatment for RR. Treatment options commonly used for DR have been tested for the management of RR by various studies due to similar clinical manifestations.^1,9–22 Studies investigating the use of laser photocoagulation and intravitreal corticosteroids for the prevention of RR have shown significant improvements in VA and retinal edema; however, these results may only be visible in the short term.^9–14 

Intravitreal antivascular endothelial growth factor (VEGF) agents, including aflibercept, ranibizumab, and bevacizumab have been proposed as treatment options for RR. Patients treated with anti-VEGF agents at fixed intervals displayed significant improvement or stabilization in VA, mean central subfield thickness, retinal hemorrhages, and resolution of edema.^15–17 Anti-VEGF injections administered to patients in variable-interval regimens have also been shown to be effective in limiting vision loss and edema that can be attributed to RR.^19,20–22 

Although there is mounting evidence that intravitreal anti-VEGF pharmacotherapies, such as aflibercept, can be used to potentially ameliorate the effects of RR, more studies with larger sample sizes and differing regimens must be conducted to further examine its tolerability and efficacy. The purpose of this phase II, open-label, randomized clinical trial was to evaluate the effectiveness and safety of intravitreal aflibercept injections (IAIs) given in a treat-and-extend regimen for RR and optic neuropathy. 

Methods

The Intravitreal Aflibercept Injection for Radiation Retinopathy Trial (ARRT) is a phase II, open-label, randomized, controlled clinical trial (ClinicalTrials.gov identifier, NCT03085784; Sponsor, Retina Consultants of Texas; Collaborator, Regeneron Pharmaceuticals, Inc.). Institutional Review Board/Ethics Committee approval was obtained by Houston Methodist Hospital and informed consent was obtained from every subject. This study adhered to the tenets of the Declaration of Helsinki and was in accordance with Health Insurance Portability and Accountability Act regulations. Data were collected at the Retina Consultants of Texas (Houston, Katy, and The Woodlands, TX, USA) and the Retina Specialists of Michigan (Grand Rapids, MI, USA). Results were reported following the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline (Supplement S1). 

Subjects who met the following inclusion criteria were enrolled: 18 years or older, RR with evidence of fluid on spectral-domain optical coherence tomography (SD-OCT) causing vision loss in the study eye, and history of ocular radiation for any primary intraocular cancer. Exclusion criteria included but were not limited to the following: metastatic cancer or any active primary cancer at the time of enrollment, prior anti-VEGF treatment in the study eye within 60 days of screening, prior intravitreal or subconjunctival treatment with corticosteroids in the study eye within 90 days of screening, macular ischemia greater than 5 disc areas, media opacity obscuring a view of the fundus or any other reason for vision loss other than RR, and evidence of ocular infection in the study eye at the time of screening. To keep the sample more homogenous, three patients were excluded from the final analysis because they underwent radiation for an orbital tumor rather than an intraocular tumor. 

At all visits, subjects underwent visual function assessment using the 4-meter Early Treatment Diabetic Retinopathy Study (ETDRS) best-corrected visual acuity (BCVA) protocol, retinopathy level assessment via slit-lamp examination and indirect ophthalmoscopy, and SD-OCT imaging using Heidelberg Spectralis HRA + OCT (Heidelberg Engineering, Heidelberg, Germany). The Heidelberg OCT acquisition protocol featured a volume scan (20 × 20 degrees, 49 lines, and 768 A-scans per line) with nine times image averaging. The severity of macular edema was assessed via central retinal thickness (CRT), defined as the average macular thickness in the central 1-mm radius of the ETDRS grid. CRT was obtained from the automated macular topographic information in the Heidelberg Eye Explorer OCT software. The retinal vasculature was evaluated at baseline, week 26, and week 52 by funduscopic examination, fluorescein angiography, fundus photographs, and OCT angiography. The severity of intraretinal hemorrhages, hard exudates, and neovascularization were assessed through slit lamp and/or indirect ophthalmoscopy by the investigator. 

Thirty-nine subjects from 4 study sites in the United States were enrolled and randomized into cohorts 1 and 2 in a 1:1 ratio. A randomly selected envelope containing a cohort assignment was chosen for each subject by a masked study coordinator. Cohort 1 received a loading dose of 3 consecutive monthly IAIs followed by a treat-and-extend protocol. Cohort 2 followed the treat-and-extend protocol with no loading dose. In the treat-and-extend protocol, patients were extended in the absence of retinal fluid with less than 5 ETDRS letters lost from the previous visit. The absence of retinal fluid, otherwise referred to as a dry macula, was defined as the resolution of intraretinal and subretinal fluid or the presence of small intraretinal cysts that did not distort the foveal contour on SD-OCT. If more than five ETDRS letters were lost from the previous visit, the patients were extended if the letter loss was not due to new or persistent retinal edema. Each extension was 2 weeks beyond the initial 4-week interval. If the extension criteria were not met on a follow-up visit, the treatment interval was reduced by 2 weeks until the extension criteria were met or until a 4-week interval was reached. 

The primary outcome measure for safety and tolerability was mean change in ETDRS BCVA at week 52. ETDRS BCVA values were converted to Snellen equivalent and logarithm of the minimum angle of resolution (logMAR) values for numerical analyses. Secondary outcome measures included mean change in CRT at week 52; percentage of patients who achieved a dry macula by week 52; mean number of IAIs received through week 52; and the percentage of patients with resolution of retinal hemorrhages, retinal exudates, optic disc edema, and capillary non-perfusion. In a post hoc analysis, eyes were categorized by baseline VA into good (≥65 ETDRS letters and Snellen equivalent ≥20/50) and poor (<65 ETDRS letters and Snellen equivalent <20/50) vision groups. Statistical analysis and sample size estimation was performed using RStudio (version 1.4.1103; RStudio Inc., Boston, MA, USA). A priori power calculation revealed that a sample size of 40 was needed to achieve 80% power. Based on this result, the final study size was 39 (Fig. 1). Between-group comparisons were conducted using Student's t test for the comparison of continuous variables and Fisher's exact test for the comparison of categorical variables. A P value less than 0.05 was considered statistically significant. Continuous variables were expressed as the mean (median and range) unless otherwise indicated. 

Figure 1.

 

Schematic representation of the randomization and enrollment of ARRT patients.

Results

Demographics and Tumor Characteristics

Thirty-nine patients were enrolled between June 2017 and October 2020. Nine patients were excluded from analysis. Of these, 5 patients were lost to follow-up, 3 patients had pathology after orbital irradiation, and one patient underwent radiotherapy 48 years prior to the screen visit. After exclusion, 14 (46.7%) patients were in cohort 1 and 16 (53.3%) patients were in cohort 2 (see Fig. 1). Sixteen (53.3%) patients were women and 14 (46.7%) patients were men. All 30 (100%) patients were non-Hispanic White. At baseline, the mean patient age was 61 (63, 23–81) years (Table 1). The mean ETDRS BCVA of all subjects at enrollment was 60.7 letters (62.5, Snellen equivalent, 20/61; 21–82, Snellen equivalent, 20/381–20/23). The mean initial CRT was 484 µm (486 µm; 294–927 µm). There were no statistically significant differences in baseline CRT, age, laterality of the study eye, and sex among the two cohorts (P > 0.05), with the exception of mean baseline ETDRS BCVA (P = 0.021). 

Table 1.

 

Baseline Patient Demographics

 

Regarding ocular tumor characteristics, the mean proximity to the macula and optic nerve head (ONH) was 2.10 disc diameters (DD) and 2.35 DD, respectively, for patients with available data (24/30, 80%; Table 2). The mean tumor height was 3.60 mm and the largest basal diameter was 11.58 mm for patients with available data (24/30, 80%). There were no statistically significant differences in tumor characteristics between the two cohorts. Moreover, for patients with available data that were treated with iodine-125 brachytherapy (22/30, 73%), the mean maximum radiation dose to the macula and ONH was 136.64 Gy and 63.57 Gy, respectively. There were no statistically significant differences in radiation doses between the two cohorts. 

Table 2.

 

Tumor Characteristics and Radiation Dosage for Patients Treated With Iodine-125 Brachytherapy

 

Visual Acuity Outcomes

Mean ETDRS BCVA letters were 67.6 and 54.7 at baseline for cohorts 1 and 2, respectively (Snellen equivalent, 20/45 and 20/81) and were statistically significant (P = 0.021) due to chance randomization differences. Mean ETDRS BCVA letter gains from baseline to week 52 were +4.3 letters for the entire cohort (P = 0.087), +1.57 and +6.69 letters for cohorts 1 and 2, respectively (P = 0.29; Fig. 2A). Student's t test demonstrated no significant differences between baseline and year 1 mean absolute ETDRS BCVA values overall (P = 0.31) and within each cohort (cohort 1, P = 0.61 and cohort 2, P = 0.09). There was no significant difference in mean BCVA between the cohorts at week 52 (P = 0.20). 

Figure 2.

 

Best-corrected visual acuity (BCVA) and central retinal thickness (CRT) outcomes from baseline to year 1. (A) Mean BCVA letters from baseline to year 1. Mean ETDRS BCVA letter gain from baseline to week 52 was +4.3 letters for the entire cohort (P = 0.07), +1.57 and +6.69 letters for cohorts 1 and 2, respectively. There was no significant difference between or within cohorts in the mean BCVA change from baseline over the 1-year period (P > 0.05). Error bars indicate the standard error of the mean. (B) Mean CRT from baseline to year 1. Mean CRT changes from baseline to year 1 were −158 µm for the entire cohort (P < 0.001), and −130.1 µm and −182.3 µm for cohorts 1 and 2, respectively (P = 0.35). Error bars indicate the standard error of the mean.

ETDRS BCVA scores were converted into Snellen equivalents to compare these data with historical controls. Accounting for all subjects, at week 52, 96.7% (n = 29) of patients had vision of 20/200 or better; 53.3% (n = 16) of patients had vision of 20/40 or better; and 30% (n = 9) had improved 10 ETDRS letters or more from baseline. Only 3.3% (n = 1) of patients had 20/200 or worse vision at week 52. Patients in the current study presented with RR at a mean of 3 years after radiation therapy. Because this report analyzes visual outcomes 1 year after aflibercept therapy initiation following RR diagnosis, the visual outcomes at year 1 were compared to the visual outcomes at 3 years post-brachytherapy in the prospective COMS trial and the 4-year VA results from nonrandomized historical controls of a retrospective trial (Table 3).^8,17 

Table 3.

 

Visual and Anatomic Outcome Comparisons to Historical Controls and Other Anti-VEGF Pharmacotherapies for Radiation Retinopathy

 

Anatomic Outcomes

Mean CRT decreased in both cohorts, demonstrating macular edema improvement over 1 year. Mean CRT values were 441.2 µm and 522.3 µm at baseline for cohorts 1 and 2, respectively. Mean CRT changes from baseline to year 1 were –158 µm for the entire cohort (P < 0.001), and –130.1 µm and –182.3 µm for cohorts 1 and 2, respectively (Fig. 2B). There were significant differences between baseline and week 52 mean CRT values overall (P = 0.0002) and within each cohort (cohort 1, P = 0.003 and cohort 2, P = 0.003). There was no significant difference between cohorts in mean CRT values at week 52 (P = 0.60). 

At the baseline visit, of the 30 patients that were included in the analysis, a total of 30.0% (n = 9), 20.0% (n = 6), 13.3% (n = 4), 50.0% (n = 15), and 6.7% (n = 2) of patients had any degree of capillary nonperfusion, exudates, optic disc edema, retinal hemorrhages, and vitreous hemorrhage, respectively. No significant differences were found between groups at baseline. At year 1, a total of 33.3% (n = 10), 3.3% (n = 1), and 3.3% (n = 1) had any degree of capillary nonperfusion, exudates, and vitreous hemorrhage, respectively. No incidences of optic disc edema or retinal hemorrhages were reported at year 1 (Fig. 3). No significant differences were found between the groups at year 1. 

Figure 3.

 

Case example of ARRT subject. The subject, presenting with a history of radiation retinopathy from brachytherapy for choroidal melanoma 3 years prior to enrollment, was treated with a loading dose of three intravitreal aflibercept injections followed by a treat-and-extend regimen (cohort 1). (A) At baseline, the subject presented with optic disc edema, VA of 63 letters, and CRT of 413 letters. (B) At week 26, improvement was observed with, optic disc edema resolution, a VA of 69, and CRT of 267 µm. (C) At week 52, the subject improved to a VA of 75, whereas CRT was consistent at 267 µm after initial improvement at week 26. Overall, there was a VA gain of 12 letters and CRT improvement of 146 µm.

At year 1, there were 56.7% (n = 17) of subjects in the study who achieved a dry macula. This was observed in 71.4% (n = 10) and 43.8% (n = 7) of patients in cohorts 1 and 2, respectively. There was no significant difference between cohorts for the percentage of patients with a dry macula at week 52 (P = 0.16). Notably, no patients developed neovascularization during the first year of the study. The percentage of patients with retinal hemorrhages decreased over the course of the study, with 50.0% (n = 15) at baseline and 0.0% (n = 0) at week 52 (P < 0.001). There were no other significant differences between time points for the other morphological parameters. 

Injection Frequency

Through week 52, patients in cohorts 1 and 2 received a mean of 9.9 (9, 8–17) and 10.4 injections (10.5, 7–14), respectively, with a mean of 10.1 injections (9, 7–17) when accounting for all 30 patients. Student's t test indicated no significant differences in the number of injections received between cohorts throughout the study (P = 0.59). 

Adverse Events

There was one serious ocular event of optic nerve atrophy as a result of concurrent optic neuritis and autoimmune disease. There was one serious systemic adverse event of deep vein thrombosis as a result of a deep vein ablation. Both serious adverse events were deemed unrelated to treatment. There were no cases of endophthalmitis and only one case of transient anterior segment inflammation. 

Discussion

The main goal of this randomized, phase II, prospective clinical trial was to assess the tolerability and efficacy of 2 mg IAI administered in a treat-and-extend regimen for the treatment of RR. Through the first year of the study, both cohorts experienced nonsignificant improvements in mean ETDRS BCVA, with cohort 2 exhibiting a greater change in mean BCVA than cohort 1. CRT decreased significantly for both cohorts from baseline to year 1, and the mean changes in CRT were not significantly different between cohorts. The IAI treatment displayed an acceptable safety profile with low rates of ocular adverse events and no incidence of treatment-related serious adverse events. 

Although both cohorts experienced improvements in VA compared to baseline, the differences within or between cohorts were not statistically significant, likely due to the flexible inclusion criteria and lack of power. However, the natural history of patients with this condition is steady visual decline, rendering these results meaningful. Similar visual outcomes have been observed in other studies; for example, in the second year of the Ranibizumab for Radiation Retinopathy trial, there was a stabilization of vision among all cohorts upon transition to a treat-and-extend protocol.²³ Other studies that have administered intravitreal ranibizumab or aflibercept injections to subjects with RR at variable treatment intervals or on a pro re nata (PRN) basis resulted in maintenance or stabilization of mean BCVA.^5,20 These data show that administration of anti-VEGF therapy following an individualized treatment regimen may help prevent visual decline that is typically observed in patients with RR, whereas minimizing the treatment burden on patients compared to fixed-interval dosing. It should be noted that, in the current study, the mean number of injections over the 1-year study period was 10.1, approximating the need for monthly injections in patients with RR. Although expected, a second year of follow-up would be necessary to determine if more patients can meet the criteria for extension. 

Despite the lack of significant improvements in BCVA, the visual outcomes from this study compare favorably to historical controls and other studies investigating the use of anti-VEGF pharmacotherapies for RR treatment (see Table 3). The results from the COMS show that 3 years after I¹²⁵ brachytherapy, 34% of patients had a Snellen VA of 20/40 or better,⁸ whereas at the 1-year time point of the present study, that number was 53.3%. Similarly, the median VA observed in the COMS at 3 years was appreciably lower than that observed at the 1-year end point of ARRT (20/125 vs. 20/39, respectively).⁸ Moreover, in the current trial, 97% of subjects ended with a Snellen VA of 20/200 or better, compared to 95% reported by Murray et al. for subjects treated with IAI,²⁰ 89% reported by Kim et al. for subjects with uveal melanoma treated with proton beam irradiation and intravitreal ranibizumab every 2 months for a year,¹⁸ and 83% reported by Schefler et al. for subjects with RR-related macular edema treated with intravitreal ranibizumab on a monthly or PRN basis.²⁴ 

Chance randomization differences between cohorts in this study resulted in baseline BCVA differences between cohorts, limiting the ability to compare the effect of the loading dose on the ultimate BCVA. Thus, a post hoc analysis was conducted to compare the final VA of patients with good or poor vision at baseline (Fig. 4). Of 30 eyes, 14 eyes had good (≥20/50 Snellen) baseline VA and 16 eyes had poor (<20/50 Snellen) baseline VA. The mean baseline VA of the good vision group was 75 letters, whereas that of eyes with poor vision was 48 letters. The mean final VA was compared across both baseline groups. Eyes with good baseline vision had a significantly better final VA (76 ETDRS letters and Snellen equivalent 20/32) than the poor baseline vision group (56 ETDRS letters and Snellen equivalent 20/80) at year 1 (P < 0.001). These data suggest that treatment for patients with baseline vision greater than 65 letters (Snellen equivalent 20/50) can lead to better preservation of vision at least in the first year and that recovery of vision to 70 or more letters is rare once vision drops below 50 letters. 

Figure 4.

 

Comparison of final visual acuity (VA) between baseline vision groups. The final visual acuity of patients with good baseline vision (≥20/50 Snellen equivalent) was significantly higher than those with poor baseline vision (<20/50 Snellen equivalent), indicated by the asterisks. The dashed line represents the median VA at baseline (62.5 letters and Snellen equivalent = 20/56) for the entire cohort. The mean final visual acuity of patients with good and poor baseline vision was 76 letters (Snellen equivalent = 20/32) and 56 letters (Snellen equivalent = 20/80), respectively (P < 0.001).

Previous studies have similar anatomic outcomes to the present study. At the 1-year end point, CRT decreased significantly for the whole subject population and each of the cohorts. These results, compounded with the existing literature on RR, show that the administration of intravitreal anti-VEGF injections in a treat-and-extend regimen can help reduce radiation-induced macular edema.^{5,19,20,22,24} Considering all subjects that completed the study regardless of initial ETDRS BCVA, cohort 2 experienced a larger decrease in CRT from baseline to week 52, although the difference between the two cohorts was not significant (P = 0.41). This indicates that there may not be a considerable benefit to having an initial monthly loading dose of IAIs. 

The severity of retinal damage was assessed by the change in capillary nonperfusion, exudates, optic disc edema, retinal hemorrhages, and vitreous hemorrhage. Only the change in retinal hemorrhages was significantly lower at year 1 (0%) from 50.0% at baseline. The ranibizumab for radiation retinopathy (RRR) study similarly saw an improvement in retinal hemorrhages over the course of the study, with decreases from baseline to 0% to 25% from initial incidences ranging from 255 to 62.5%. Overall, these findings are reflective of the literature on DR which sustains that anti-VEGF injections can promote the regression of DR and diabetic macular edema (DME).²⁵ 

The main limitations of the current study were the significant difference in mean BCVA between the two cohorts at baseline due to random chance, the small sample size, lack of standardization of interval from exposure to radiation to enrollment in the study, and the lack of a true control group or fixed-interval group. A significant difference in mean BCVA at baseline could skew the mean BCVA change comparisons given that cohort 2 had a lower mean BCVA at baseline. The lack of a randomized control group and the relatively small size of the study limits the strength of the data. Further studies with randomized controls are needed to confirm the findings of the current study. 

In summary, intravitreal aflibercept given in a treat-and-extend regimen over 1 year stabilizes vision and improves macular anatomy for patients with RR. This cohort experienced a significant improvement in CRT overall and within groups, with meaningful, yet non-significant, improvements in mean ETDRS BCVA. Given the similar clinical manifestations of RR with DR, well-established DR anti-VEGF treatments seem to be helpful for the treatment of RR. Additional multicenter collaborative studies through the Diabetic Retinopathy Clinical Research Network (DRCR) are underway to resolve these important questions in this patient population. Because these studies are expensive and time-consuming, the data from this small pilot trial is provocative and informative until larger-scale data is available. 

Acknowledgments

Supported by Regeneron Pharmaceuticals, Inc (Tarrytown, NY, USA). The sponsor or funding organization had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and decision to submit the manuscript for publication. 

Previous Presentation: This data was presented at the Association for Research in Vision and Ophthalmology (ARVO), Denver, CO, May 2022; The 45th Macula Society Annual Meeting, Berlin, Germany, June 2022; The Congress of the International Society of Ocular Oncology, Leiden, the Netherlands, June 2022; The American Society of Retina Specialists (ASRS) 40th Annual Scientific Meeting, New York, NY, July 2022. 

Disclosure: S. Trejo Corona, None; C. Villanueva Boone, None; A.M. Ali, None; C. Moore, None; A. Brown, None; J. Munoz, None; T. Aaberg, None; A.C. Schefler, Regeneron (F)