Clinical Characteristics and Long-Term Outcomes of Eyes With Neovascular Age-Related Macular Degeneration Requiring Frequent Treatment

Abstract

Purpose: To compare long-term visual outcomes in neovascular age-related macular degeneration (AMD) patients who received frequent vs infrequent antivascular endothelial growth factor injections. Methods: This retrospective case series included treatment-naïve AMD patients receiving either frequent injections (≤ 6-week intervals) or infrequent injections (≥ 10-week intervals). Best-available visual acuity (VA) and anatomic outcomes were assessed at the initial visit, 52 weeks, 104 weeks, and final visit. Results: A total of 151 eyes were studied over a mean follow-up of 42.6 months: 81 eyes (54%) in the frequent group and 70 eyes (46%) in the infrequent group. Baseline central foveal thickness (CFT) was higher in the frequent group (320 µm vs 265 µm; P = .002). Though CFT improved in both groups, it remained higher in the frequent group at 52 weeks (230 µm vs 185 µm; P = .004), 104 weeks (203 µm vs 173 µm; P = .004), and final visit (197 µm vs 165 µm; P = .015). The frequent group showed more subretinal and intraretinal fluid at all times. Baseline logMAR VA was 0.63 and 0.84 in the frequent and infrequent groups, respectively (P = .10). Visual improvement was similar between groups at 52 weeks (0.21 vs 0.22 logMAR; P = .916), 104 weeks (0.21 vs 0.18 logMAR; P = .714), and final visit (0.1 vs 0.05 logMAR; P = .510). Poor baseline VA (P < .001) and geographic atrophy (P = .025) were associated with worse outcomes. Conclusions: Over 4 years, despite higher CFT and more fluid, frequent injections showed similar visual improvement to infrequent injections. Baseline vision was the strongest predictor of final outcomes, regardless of injection frequency.

Introduction

Neovascular age-related macular degeneration (AMD) is a leading cause of significant vision loss in older adults. ¹ Treatments with antivascular endothelial growth factor (anti-VEGF) agents maintain or improve vision by suppressing the exudation associated with macular neovascularization (MNV). ²

Despite continuous treatment, the long-term visual prognosis for some patients remains poor due to the development and progression of macular fibrosis and geographic atrophy (GA).^3,4 Data from the ANCHOR, ⁵ MARINA, ⁶ and HORIZON ⁷ studies revealed that 37% of eyes became legally blind (visual acuity [VA] of 20/200 or worse) at a mean follow-up of 7.3 years, primarily due to macular atrophy. ⁸

Similarly, findings from the CATT study suggested a potential link between repeated frequent injections and GA progression. ⁹ However, evidence is still lacking on how frequent injections influence long-term visual outcomes and GA progression in real-world settings.

Treatment strategies for anti-VEGF injections include fixed monthly injections, pro re nata regimens, and treat-and-extend regimens. According to the American Society of Retina Specialists (ASRS) 2019 Preferences and Trends Survey, treat-and-extend regimens are currently the most commonly adopted approach both within and outside the United States. ¹⁰

Randomized prospective clinical trials, such as TREX-AMD,^11,12 LUCAS, ¹³ RIVAL,^14,15 TREND, ¹⁶ and CANTREAT ¹⁷ have shown that treat-and-extend regimens with either bevacizumab, ranibizumab, or aflibercept are equivalently effective and similar to monthly injections, while reducing injection burden. However, in the real-world setting, the frequency of injections is highly variable, and some eyes cannot achieve any extension of the treatment interval, even after years, because of persistent fluid.

Injection frequency has been identified as an important factor associated with better visual outcomes in studies using pro re nata regimens.^18–21 Nevertheless, the long-term visual outcomes of treat-and-extend regimens in eyes requiring frequent vs infrequent injections have not been well studied.^22,23

This study aims to compare long-term visual outcomes in patients with neovascular AMD undergoing frequent (every 4 to 6 weeks) vs infrequent (once every 10 or more weeks) injections under treat-and-extend regimens. Additionally, we evaluate anatomic outcomes, such as the persistence of subretinal fluid (SRF), intraretinal fluid (IRF), GA progression, and MNV type, and their impact on vision between and within each group using optical coherence tomography (OCT).

As the mainstay for diagnosing and monitoring AMD, OCT enables retina specialists to precisely localize and delineate pathological changes, facilitating accurate disease characterization and management. ²⁴

Methods

This study was approved by the Institutional Review Board of the Wills Eye Hospital and adheres to the Declaration of Helsinki. Patients with a diagnosis of neovascular AMD seeking care at Mid Atlantic Retina, the Retina Service of Wills Eye Hospital between April 2016 and April 2018 were identified using diagnostic and procedural billing codes.

Inclusion and Exclusion Criteria

We included treatment-naïve patients with AMD pathology exclusively in 1 eye. These inclusion criteria reduce the bias introduced by treatment needs in the other eye, which would influence follow-up times and, potentially, treatment intervals. The diagnosis of AMD was primarily made using OCT, with fluorescein angiography (FA) used for additional confirmation in selected cases.

Patients were categorized by those requiring frequent injections, defined as being unable to extend the treatment interval past 6 weeks (determined by the provider to be failing at 8 weeks, requiring interval shortening). Those defined as requiring infrequent injections required anti-VEGF therapy only every 10 weeks or longer.

Excluded were those with neovascular AMD in the fellow eye, retinal pigment epithelium tears, significant epiretinal membrane, diabetic macular edema, and endophthalmitis in the eye affected with neovascular AMD. Patients with a follow-up period of less than 2 years, as well as those who were lost to follow-up, were excluded from the study.

Outcome Measures

Data were compiled using EMR documentation at the patients’ initial visit, 52 weeks and 104 weeks from initial visit (±3 weeks), and final visit within the study period.

The primary outcome was best-available VA (spectacle and/or pinhole corrected) in those receiving frequent vs infrequent injections. Snellen VA was converted to logMAR for analysis. Characteristics including presence of SRF and/or IRF, and MNV classification based on location were evaluated as secondary outcomes. For quality assurance, OCT imaging from every visit was reviewed, and the MNV classification, and presence of IRF and/or SRF was confirmed.

The central foveal thickness (CFT) was determined by measuring the distance from the inner boundary of the retina to the inner boundary of the hyperreflective retinal pigment epithelium layer. These measurements were manually conducted using the caliper function in the Heidelberg Eye Explorer software (Heidelberg Engineering, Inc). A single investigator (M.S.) carried out all measurements.

Statistical Analysis

Depending on the distribution of the data, either an independent t test or a Mann-Whitney U test was used to compare the differences in the mean number of injections, VA, and CFT measurements between the 2 groups at baseline, 52 weeks, 104 weeks, and final visit timepoints. The presence of fluid within the retina (IRF and SRF), and pigment epithelial detachments (PED) at these same intervals, was compared between groups using χ² analysis.

Sex, phakic status, and presence of hemorrhage at the initial visit were also compared using χ² analysis. Depending on the distribution of the data, either a paired t test or a Wilcoxon signed rank test was used to compare improvements in VA and CFT over the course of the study period in each treatment group.

The presence of SRF and IRF was compared over the course of the study period within each treatment group using the McNemar test. P values < .05 were considered statistically significant. Mean values are ± SD. Finally, linear regression analysis was performed to identify factors associated with poor visual outcomes.

Results

Patient Characteristics

This study had a mean follow-up time of 42.6 months (range, 25.2-57.6 months). Among 151 eyes in this study, 81 (54%) eyes underwent frequent injections, and 70 (46%) eyes underwent infrequent injections.

Baseline characteristics of each group are listed in Table 1. At the baseline, aflibercept was used in 2.5% vs 10%, ranibizumab in 79% vs 65.7%, and bevacizumab in 18.5% vs 24.3% of eyes in the frequent and infrequent groups, respectively (P = .09).

Table 1.

Demographic and Clinical Characteristics at Baseline, 52 Weeks, 104 Weeks, and Final Visit of Patients With Frequent vs Infrequent Anti-VEGF Injections.

Characteristics	Frequent Injections Group a (81 eyes)	Infrequent Injections Group b (70 eyes)	P Value
Mean age (y) ± SD	76.9 ± 8.8	79 ± 8.2	.14
Sex (male/female)	33/48	29/41	.93
Lens status (phakic/pseudophakic)	37/44	25/45	.21
Anti-VEGF type, n (%)			.09
Aflibercept	2 (2.5)	7 (10)
Ranibizumab	64 (79)	46 (65.7)
Bevacizumab	15 (18.5)	17 (24.3)
Presence of hemorrhage at first visit (%)	43.2	42.8	.97
Mean VA (logMAR) ± SD
Baseline	0.63 ± 0.54	0.84 ± 0.67	.10
52 weeks	0.42 ± 0.43	0.62 ± 0.63	.39
104 weeks	0.42 ± 0.43	0.61 ± 0.60	.21
Final	0.52 ± 0.50	0.78 ± 0.67	.04 c
Mean CFT (µm) ± SD
Baseline	320 ± 131	265 ± 121	.002 c
52 weeks	230 ± 96	185 ± 54	.004 c
104 weeks	203 ± 66	173 ± 54	.004 c
Final	197 ± 75	165 ± 56	.015 c
SRF (%)
Baseline	92.5	78.6	.013 c
52 weeks	53.1	21.4	<.001 c
104 weeks	43.2	11.3	<.001 c
Final	42.0	10.0	<.001 c
IRF (%)
Baseline	69.1	74.3	.48
52 weeks	50.6	27.1	.003 c
104 weeks	44.4	21.4	.003 c
Final	35.8	17.1	.01 c
PED (%)
Baseline	76.5	35.7	<.001 c
52 weeks	77.8	28.6	<.001 c
104 weeks	76.5	28.6	<.001 c
Final	76.5	30.0	<.001 c
Mean follow-up time (mo) ± SD	42.1 ± 7.3	43 ± 8.3	.48

^aUnable to extend treatment interval > 6 weeks.

^bTreatment interval ≥ 10 weeks.

^cIndicates statistical significance.

Abbreviations: anti-VEGF, antivascular endothelial growth factor; CFT, central foveal thickness; IRF, intraretinal fluid; ME, macular edema; PED, pigment epithelial detachment; SRF, subretinal fluid; VA, visual acuity.

At the final visit, aflibercept usage increased to 25.9% vs 14.3%, ranibizumab usage remained high at 74.1% vs 78.6%, and bevacizumab usage declined to 0% vs 7.1% in the frequent and infrequent groups, respectively (P = .015), with 40.7% of patients in the frequent group switching their anti-VEGF treatment by the final visit compared with 24.3% in the infrequent group (P = .038).

There was no significant difference in age, gender, laterality, phakic status, presence of hemorrhage at first visit, frequency of IRF, or follow-up lengths between the 2 groups. However, those who were frequently injected had an increased frequency of SRF (92.7% vs 78.6%; P = .013), mean CFT (320 µm vs 265 µm; P = .002), and PED (76.5% vs 35.7%; P < .001) compared with those infrequently injected.

Visual Outcomes

Frequent vs Infrequent Injections

VA data are depicted in Table 1. In the frequent injections group, mean logMAR VA at baseline was 0.63 ± 0.54 (Snellen equivalent, 20/85). Mean VA was 0.42 ± 0.43 (Snellen equivalent, 20/52) at week 52 and 0.42 ± 0.43 (Snellen equivalent, 20/53) at week 104 (P < .001 and P < .001, respectively).

No significant difference was found when comparing the baseline VA with the final VA (0.52 ± 0.50 [Snellen equivalent, 20/66]; P = .06). In the infrequent group, baseline VA changed from 0.84 ± 0.67 (Snellen equivalent, 20/138) to 0.62 ± 0.63 (Snellen equivalent, 20/83) at the end of week 52 (P < .001) and to 0.61 ± 0.60 (Snellen equivalent, 20/81) at the end of week 104 (P < .001). No significant change was found when comparing VA at the final visit (0.78 ± 0.67 [Snellen equivalent, 20/120]) with the baseline visit (P = .384).

When comparing the frequent injections group with the infrequent group, there was only a significant difference between the 2 groups’ VA at their final visit (0.52 ± 0.50 [Snellen equivalent, 20/66] vs 0.78 ± 0.67 [Snellen equivalent, 20/121], respectively; P = .04), with no evidence to support a difference at 1 year. The visual improvement in individuals receiving frequent injections compared with those receiving infrequent injections was 0.21 vs 0.22 logMAR (P = .916) at 52 weeks, 0.21 vs 0.18 logMAR (P = .714) at 104 weeks, and 0.1 vs 0.05 logMAR (P = .510) at the final visit.

Factors Associated With Visual Outcomes

The results of the linear regression analysis for factors associated with worse visual outcomes at 104 weeks and final visit are shown in Tables 2 and 3, respectively. Multivariable linear regression analysis identified poor baseline VA as a significant factor associated with worse visual outcomes at both 104 weeks (R² = 0.461; P < .001) and final visit (R² = 0.457; P < .001). Additionally, GA presence was found to be significant in the multivariable analysis at the final visit (R² = 0.457; P = .025).

Table 2.

Linear Regression Analysis for Factors Associated With Worse Visual Outcomes at 104 Weeks.

Parameters	Univariate Linear Regression			Multivariable Linear Regression
	β Coefficient	R 2	P Value	β Coefficient	R 2	P Value
Age	0.015	0.061	.003 a	0.007	0.461	.083
Male sex	0.132	0.015	.134
Lens status	0.152	0.020	.083
Baseline VA	0.573	0.419	<.001 a	0.516	0.461	<.001 a
Infrequent injection	0.194	0.034	.025 a	0.084	0.461	.229
Presence of GA at baseline	0.191	0.060	.003 a	0.088	0.461	.085
MNV type	0.094	0.008	.281
Presence of SRF
Baseline	−0.019	<0.000	.881
Week 104	−0.169	0.021	.077
Presence of IRF
Baseline	0.256	0.048	.007 a	0.042	0.461	.592
Week 104	0.185	0.028	.041 a	0.039	0.461	.600
CFT
Baseline	<0.000	0.004	.434
Week 104	−0.001	0.016	.131
PED
Baseline	−0.078	0.005	.378
Week 104	−0.151	0.021	.082
Presence of hemorrhage at baseline	0.067	0.004	.449

^aIndicates statistical significance.

Abbreviations: CFT, central foveal thickness; GA, geographic atrophy; IRF, intraretinal fluid; MNV, macular neovascularization; PED, pigment epithelial detachment; SRF, subretinal fluid; VA, visual acuity.

Table 3.

Linear Regression Analysis for Factors Associated With Worse Visual Outcomes at the Final Visit.

Parameters	Univariate Linear Regression			Multivariable Linear Regression
	β Coefficient	R 2	P Value	β Coefficient	R 2	P Value
Age	0.017	0.061	.002 a	0.008	0.457	.095
Male sex	0.162	0.018	.105
Lens status	0.187	0.024	.061
Baseline VA	0.634	0.416	<.001 a	0.533	0.457	<.001 a
Infrequent injection	0.277	0.053	.005 a	0.068	0.457	.427
Presence of GA at baseline	0.257	0.083	<.001 a	0.132	0.457	.025 a
MNV type	0.153	0.016	.123
Presence of SRF
Baseline	−0.120	0.005	.399
Final visit	−0.196	0.021	.076
Presence of IRF
Baseline	0.259	0.038	.017 a	-0.005	0.457	.956
Final visit	0.143	0.011	.197
CFT
Baseline	<0.001	0.001	.658
Final visit	−0.002	0.056	.004 a	−0.001	0.457	.165
PED
Baseline	−0.096	0.006	.338
Final visit	−0.194	0.026	.049 a	−0.033	0.457	.696
Presence of hemorrhage at baseline	0.140	0.013	.162

^aIndicates statistical significance.

Abbreviations: CFT, central foveal thickness; GA, geographic atrophy; IRF, intraretinal fluid; MNV, macular neovascularization; PED, pigment epithelial detachment; SRF, subretinal fluid; VA, visual acuity.

Anatomic Outcomes

Central Foveal Thickness

In the frequent group, mean for CFT was 320 ± 131 µm at the baseline visit. CFT changed to 230 ± 96 µm at the end of week 52 (P = .001) and to 203 ± 66 µm at the end of week 104 (P < .001). Final CFT was significantly thinner compared with baseline visit (P < .001). In the infrequent group, mean baseline CFT was 265 ± 121 µm, which decreased to 185 ± 54 µm at the end of week 52 (P < .001) and 173 ± 54 µm at the end of week 104 (P < .001). Final CFT was significantly thinner than baseline visit (P < .001). The average CFT in the frequently injected group remained consistently and significantly more elevated between the 2 groups, at 30 µm or larger at each data collection point.

Subretinal Fluid

The proportion of eyes with SRF in the frequent group changed from 75/81 (92.5%) at the baseline visit to 43/81 (53.1%) at week 52 and 35/81 (43.2%) at week 104 (P < .001, respectively). SRF was present in 34/81 eyes (42%) at the final visit (P < .001).

In the infrequent group, the proportion of eyes with SRF changed from 55/70 (78.6%) to 15/70 (21.4%) at week 52 (P < .001) and to 8/70 (11.4%) at week 104 (P < .001). SRF was present in 7/70 (10%) of the eyes at the final visit (P < .001). When comparing the 2 groups, there was a significantly greater frequency of SRF present at each timepoint in the frequent injections group than those receiving injections less frequently.

Intraretinal Fluid

In the frequent group, the proportion of eyes with IRF changed from 56/81 (69.1%) at the baseline visit to 41/81 (50.6%) at week 52 and 36/81 (44.4%) at week 104 (P = .002 and P < .001, respectively). IRF was present in 29/81 eyes (35.8%) at final visit (P < .001).

In the infrequent group, the proportion of eyes with IRF changed from 52/70 (74.3%) to 19/70 (27.1%) at week 52 (P < .001) and to 15/70 (21.4%) at week 104 (P < .001). IRF was present in 12/70 (17.1%) of the eyes at final visit (P < .001). When comparing the 2 groups, there was a significantly greater frequency of IRF present in the frequent injections group at each timepoint (except for the patient’s baseline visit).

Geographic Atrophy

At baseline visit, foveal and/or nonsubfoveal GA was present in 9/71 (12.9%) in the infrequent group compared with 0/81 (0%) in the frequent group (P = .001). At the final visit, GA was present in 15/81 (18.5%) in the frequent group compared with 32/70 (45.7%) in the infrequent group (P < .001). After excluding those with GA at baseline, the infrequent injection group still had a significantly greater frequency of GA than the frequent injection group at the final visit (23/61, 37.7% vs 9/71, 12.9%; P = .001).

MNV Type

At baseline visit, in the frequent group, the MNV type was type 1 in 51 eyes (63%) and type 2 in 30 eyes (37%). In the infrequent group, the MNV type was type 1 in 31 eyes (44%) and type 2 in 39 eyes (56%). The rate of type 1 MNV was significantly higher in the frequent group, compared with the infrequent group (P = .033).

Conclusions

In this retrospective study, patients who received frequent injections for AMD showed comparable visual improvement to those with infrequent injections over a 4-year follow-up period, despite having significantly higher mean CFT and more frequent occurrences of IRF and SRF. Baseline vision emerged as the most significant predictor of final visual outcomes, regardless of injection frequency in our analysis.

Retina specialists often adjust injection frequency aiming to resolve all retinal fluid, using protocols such as treat-and-extend or pro re nata. In the treat-and-extend protocol, intervals between office visits are extended if OCT is exudative free or stable and reduced if disease activity is detected. ²⁵ Various RCTs, such as HARBOR, ²⁶ CATT, ²⁰ and VIEW 1/2, ²⁷ have shown clinical gains with less than monthly injections over 24 months. Similarly, in our study, there was no significant difference in VA between the frequent and infrequent injection groups at 24 months (104 weeks).

However, by the final visit in this study (with the mean follow-up being approximately 4 years), those who were frequently injected had better VA than those injected infrequently. These patients likely received more anti-VEGF injections due to the presence of fluid in their retina, Other studies have supported better long-term visual outcomes in those more frequently injected, including CATT 5-year follow-up, ²⁸ HORIZON, ⁷ and an Australian retrospective review. ²²

Regression analysis revealed that the most important factor in predicting final vision is baseline vision, regardless of the number or frequency of injections. This finding is crucial for counseling patients about their long-term visual outcomes. Our findings indicate that baseline vision was better in eyes that were frequently injected compared with those in the infrequent injection group, though this difference was not statistically significant.

The frequently injected group received more injections due to greater residual IRF and SRF, signaling disease activity. Despite more severe exudation, this group achieved better final vision after an average of 4 years, and comparable visual improvement, highlighting the efficacy of repeated injections in managing such cases.

Importantly, our findings suggest that remnant fluid in the retina could, and perhaps should, be tolerated, as it may not impact long-term VA as much as was previously thought. It is salient to delineate which type of retinal fluid should be tolerated.

Prior studies of anti-VEGF treatment have found associations between IRF and worse VA outcomes and delayed responses to treatment, in comparison to SRF and greater improvements VA.^19,29–33 This notion is supported in the FLUID study, where those assigned to the ranibizumab treat-and-extend protocol achieved a VA comparable to those whose treatment plan aimed to resolve all SRF. ³⁴ Tolerating SRF led to significantly fewer injections at 12 and 24 months, ³⁴ though our study results favor more frequent administration. In our study, the group with the best final VA had a greater frequency of both IRF and SRF.

With more frequent anti-VEGF injections, there is a theoretical concern for increased GA risk.^28,30,35 Contrary to this, participants in this study had a significantly lower prevalence of GA in the frequent injectors group compared with the infrequent injectors group. The infrequent injection group had a greater frequency of patients with type 2 macular neovascularization (MNV), or angiographically classic (subretinal) MNV, compared with the frequent injections group. ³⁶ Eyes with type 2 MNV may be at a greater risk of atrophy compared with type 1 (angiographically occult) MNV, which is located in the subretinal pigment epithelium (sub-RPE) space. ³⁵ It has been suggested that type 1 MNV might offer a protective benefit to the RPE and the overlying neurosensory retina, potentially slowing the localized progression of atrophy. ³⁷

This study is retrospective and noncontrolled, which are limitations in exploring these relationships but better reflect real-world circumstances compared with RCTs. Patients are typically undertreated in real-world studies at baseline, and many are lost to follow-up due to time and resource constraints that prevent patients from making frequent visits. ³⁸ This could lead to more patients being included in the infrequent injections group who were recommended to get injections more frequently.

Although all biases in treatment timing could not be eliminated, we included only patients with AMD in one eye to eliminate needs in the other eye effecting treatment timing. As demonstrated in the Australian retrospective review, patients frequently stop injections after experiencing a loss in vision, but this has an equal likelihood in both frequent and infrequent injection groups and is not likely influencing our comparative results. ²²

Additionally, we did not stratify long-term for those receiving different anti-VEGF agents, which could be studied in the future. Different providers may have applied varying thresholds for injections, which could influence treatment frequency. However, this variability reflects real-world data, capturing diverse clinical practices.

A limitation of our study is the observed baseline differences between the frequent and infrequent injection groups, particularly in terms of baseline VA, CFT, and the presence of SRF and PED. These differences may reflect selection bias and reduce the comparability of the groups, potentially limiting the generalizability of our findings. Future studies may benefit from a matched-cohort design or propensity score matching to reduce baseline imbalances.

Another limitation of this study is the use of CFT measurements rather than central subfield thickness (CST), which is more commonly used in clinical trials. CFT values are generally lower than CST, which may partially account for the lower baseline thickness values observed, particularly in the infrequent treatment group. This difference in measurement approach should be considered when comparing our findings with studies that report CST.

In conclusion, over a long follow-up time frame of nearly 4 years, patients with neovascular AMD managed with treat-and-extend who received frequent injections exhibited similar visual improvement compared with those with infrequent injections, despite having higher CFT and more frequent occurrences of retinal fluid. Baseline vision was identified as the most significant predictor of final visual outcomes, regardless of injection frequency.