Age-related macular degeneration (AMD) is the leading cause of legal blindness in developed countries and is predicted to affect 288 million people worldwide by 2040 [1–3]. The disease is classified into early, intermediate, and advanced forms, with the latter including choroidal neovascularization (CNV) or central geographic atrophy (GA) [4]. Intravitreal injections of anti-vascular endothelial growth factor (anti-VEGF) agents are considered the gold standard in the treatment of neovascular AMD (nAMD), showing improvement or at least maintenance of visual acuity and anatomical restoration in the majority of patients with nAMD [5, 6].

Despite the proven efficacy of anti-VEGF agents in nAMD treatment, the Comparison of AMD Treatment Trials (CATT) study found that GA was developed in approximately 18% of patients with nAMD treated with anti-VEGF agents for 2 years, and reached 38% at year 5 [7, 8]. Several more recent studies have shown similar results, highlighting atrophy as a major player of visual impairment in eyes with nAMD even when the neovascular component is controlled [9–16].

Thus, the study of atrophy development and evolution has gained self-evident importance both in nAMD and non-nAMD eyes. Several baseline factors have been associated with GA development in eyes with nAMD under anti-VEGF treatment, including the absence of subretinal fluid (SRF), the presence of intraretinal cysts, reticular pseudodrusen, type III CNV, nascent GA, increased central foveal thickness and drusen volume, and the collapse of pigment epithelium detachment, as well as the presence of atrophy in the fellow eye [10–14], while the role of the number and frequency of injections remains controversial [11, 13, 15, 17]. Similarly, the progression of atrophy in dry AMD (dAMD) patients has been associated with such factors as the existence of non-subfoveal (vs subfoveal) MA, the multifocal (vs non-multifocal) pattern of atrophic lesions at baseline, fundus autofluorescence patterns, and hypofluorescent area surrounding by hyperfluorescent margins attributing to lipofuscin concentration and structural abnormalities at the junctional zone of macular atrophy including irregular retinal pigment epithelium elevations and increased inner nuclear layer thickness [18, 19].

In an effort to understand the dynamics of atrophy, some researchers have tried to compare its progression in eyes with nAMD versus eyes with dAMD [18, 20]. Typically, in these studies, the group of nAMD patients is different from this of dAMD patients. However, given the multifactorial nature of AMD, the proper matching of the groups is extremely hard. Until today, very few studies addressed a “within-subjects” approach for the comparison of nAMD and dAMD eyes which would eliminate several confusing factors [21].

In light of the above, the purpose of this study was to evaluate the evolution of macular atrophy in the eyes of patients with nAMD treated with anti-VEGF agents, in comparison to their fellow dAMD eyes.

In this study, we evaluated atrophy development and progression in nAMD eyes treated with anti-VEGF in comparison to their fellow eyes with dAMD. Our study was designed so that we followed the behavior of the atrophic component in patients that have developed neovascular AMD in one eye, while they remain in the dry AMD spectrum in their other eye. We have selected our group on the basis of this asymmetry, and our study focused on a specific time period for each patient that this asymmetry was present. Thus, out of the total number of our patients, 45% of nAMD eyes exhibited MA until the end of the follow-up, compared to only 16.5% of their fellow eyes with dAMD during the same time period. Our findings for nAMD eyes were similar to Grunwald et al. [8], who reported a MA cumulative incidence rate of 38% during the 5 years of the CATT study in eyes with nAMD, which was much lower than the 98% MA reported by the SEVEN-UP study, following participants for up to 8 years [28]. This large difference may be attributed to the variation in definitions and assessment of atrophy, as well as in the extended follow-up time in the latter study. MA first detection in treated eyes with nAMD increased significantly from year 2 to year 5 compared with fellow eyes. This observation suggests that year 2 of the follow-up period seems to be a crucial time point for the development of MA in eyes with nAMD treated with intravitreal anti-VEGF injections, although one should consider that the specific time points could not be precise especially in retrospective studies. When we compared the time course of MA in both eyes of patients that completed a 4-year follow-up using a synchronous paired comparison, nAMD eyes presented an overall change rate of MA which was significantly faster compared to their fellow dAMD eyes. This difference was more pronounced after the 1st year of follow-up indicating that the time effect is more pronounced in nAMD eyes than in their fellow dAMD eyes. These findings demonstrate that in patients whose eyes maintain asymmetry in regard to the neovascular component for a certain time period, this asymmetry tends to be present for the atrophic component of their disease too, during this period.

Several reasons may explain the difference in MA development and progression in the nAMD and dAMD eyes of the same patient during a particular time period. Since both conditions share common pathophysiological background, one could attribute the higher incidence and higher rate of change of atrophy we found in the group of nAMD eyes, to the fact that during our follow-up period, these eyes suffered a more advanced AMD stage compared to their fellow dAMD eyes. This underlines the possible role of local factors in the development and progress of AMD, which may explain the asymmetry in presentation and temporal evolution of both the neovascular and atrophic components of the disease in the two eyes [29]. Mann and coworkers [30] have noticed that asymmetry in AMD presentation tends to be more pronounced in the late stages of the disease and hypothesized that the later manifestations of AMD may be affected by as yet unknown stochastic factors affecting the eyes separately, acting at different times during the patient’s life.

Prolonged VEGF inhibition is one of the local factors that have been implicated in MA development in nAMD patients under anti-VEGF therapy [7, 30, 31]. In our study, there was no correlation between MA area and total number of injections in eyes with nAMD. Our results are close to IVAN and HARBOR studies that did not show an association between injection frequency and new-onset or worsening of MA over 24 months [32]. Accordingly, Sadda et al. [13] reported that monthly treatment was not found to be a risk factor for MA development over 24 months. Furthermore, Christakis et al. [33] evaluated the natural history of MA in untreated nAMD eyes of AREDS and found a relatively high incidence of MA, increasing linearly over time and affecting half of the eyes by 8 years. Based on their findings, these authors suggested that factors other than anti-VEGF therapy are involved in atrophy development, including natural progression to GA [33]. Other local factors such as the absence of SRF, the presence of intraretinal cysts, reticular pseudodrusen, type III CNV, nascent GA, increased central foveal thickness, and drusen volume, as well as the collapse of pigment epithelium detachment, have been associated with GA development in eyes with nAMD [10–14].

In our study, the absolute value of growth rate in nAMD eyes (starting from the first atrophy incidence) was found to be 0.176 mm/year for a 2-year follow-up. In the HARBOR study, the growth rate in nAMD eyes was found to be 0.39 mm/year [34], while Grunwald et al. [8] reported a progression of MA at 0.30 mm/year, which is higher than our findings. This large difference may be attributed to the variation in definitions and assessments of atrophy. Of note, in the nAMD eyes with MA at baseline, the MA growth rate was relatively stable during a 4-year period of observation, compared to those which did not have MA at baseline, where the MA synchronous growth rate was found to be much faster between year 1 and 4 of the follow-up period. This finding is consistent with previous studies in the literature, declaring that the MA progression rate decreases with increasing baseline MA [35, 36]. As for the absolute value of growth rate in dAMD eyes (starting from the first atrophy incidence), we found it to be 0.257 mm/year over a 2-year follow-up period. Differences are observed in the literature in eyes with dry AMD when compared with the dAMD eyes of our study. Proxima A reported a growth rate of 0.35 mm/year [18] over a period of 24 months, while Kim et al. [37] 0.27 mm/year, and Keenan et al. [38] 0.27 mm/year. This could be explained by differences in study design, as well as differences in number of patients and differences in imaging techniques used to grade MA.

Thus, when MA growth rate of our patients was measured from the first atrophy incidence in an asynchronous manner, it was found to be higher in dAMD eyes compared to nAMD eyes. Atrophy progressed significantly both in nAMD and dAMD eyes but raised more rapidly in dAMD eyes, especially from end of year 1 to end of year 2. Even when we used a paired asynchronous comparison in a smaller group of 17 patients that developed bilateral atrophy (both in their nAMD and the dAMD eye), we found again a higher growth rate in dAMD eyes compared to their fellow nAMD eyes, although this time the difference did not reach statistical significance. Our findings agree with the findings of Airaldi et al. [20] who compared enlargement rates of macular atrophy in eyes with nAMD vs. eyes with dAMD for a period of 5 years in two different groups of patients. They also found that the presence of a neovascular component was associated with a slower progression of MA over time [20]. The main difference in this work compared to ours is the use of two different groups for the study of nAMD and dAMD growth rates. We believe that the within-subjects methodology of our work makes this finding stronger. Finally, in their work about the natural history of geographic atrophy (GA) secondary to age-related macular degeneration, Holekamp et al. [18] analyzed the results from the Proxima A and B clinical trials and included a subgroup analysis where data of GA progression rate in fellow eyes with nAMD were presented. Although they did not provide details in methodology and they did not specifically compare study eyes to their fellow nAMD eyes, the authors present a graph showing that during the 24 months of follow-up, the mean change of GA surface in the subgroup of nAMD eyes was very similar to study eyes. It is important to remember that Proxima trials were designed to study GA and their study eyes were required to have advanced disease with a minimum surface of GA [18], and this is not in analogy to our study where most of the fellow dAMD eyes had intermediate AMD.

All these findings along with ours indicate that while patients with nAMD in one eye and dAMD in their fellow eye, whose eyes are followed for a specific time period in a synchronous manner, tend to show more atrophy in their treated nAMD eye, when MA does appear in their dAMD eye, it grows faster than in the treated nAMD eye. This finding may support the hypothesis of a possible protective role of neovascularization over MA growth [20]. Also, in combination with our findings of the absence of correlation between the number of injections and MA area, it provides additional indication against the role of anti-VEGF agents as accelerators of MA.

As far as we can say, our work is one of the very few that a methodology of within-subjects comparison of MA progression in nADM and fellow dAMD eyes is used. We believe that this approach offers a unique perspective by eliminating factors (genetic, environmental, etc.), which may interfere with the pathophysiology of such a complex disease as AMD [18, 21]. Searching the literature, we were able to locate only one work that was presented at the 2020 ARVO annual meeting in which Grobben et al. [21] adopted a similar “within-subjects’ comparison” between treated nAMD eyes and fellow non-nAMD eyes. They found that the risk of developing new MA in nAMD eyes during 5 years of therapy was almost twice greater than that of fellow dry AMD eyes, and this is in agreement with our findings [21]. It is important, however, to notice that the authors of this work based their MA progress measurements on color photos and fluorescein angiograms. Moreover, making a direct comparison with our methodology was not possible since our only reference was the ARVO abstract of this work.

Potential limitations of our study are related to its retrospective nature. In a prospective design, MA measurements could be assessed with more accuracy using a standardized OCT scanning protocol with dense scans of the entire atrophic area. In this study, SD-OCT B-scans that examined to confirm MA on NIR images were representative of the atrophic areas. However, it has been reported that with the use of the combination of the two modalities (NIR + SD-OCT images), MA can be assessed with more accuracy in measuring atrophic areas [23, 39]. In our study, this was justified by the high interdepartmental raters’ agreement. Denser and consistent scanning protocols could be implemented in future studies permitting the evaluation of the entire atrophic area against CAM criteria. Another limitation pertains to the fact that nAMD cases already treated with anti-VEGF and with co-existent MA at baseline were included in the study. The optimal design would focus on treatment-naïve patients with nAMD without MA at baseline. Finally, this study was designed to compare macular atrophy in eyes treated for wet AMD, with their fellow dry AMD eyes in a specific time period. Since AMD can present temporal asymmetry between the two eyes, some of our findings could differ if the follow-up was longer, and this should be addressed in future studies. All-in-all, the comparison of our findings regarding the incidence of macular atrophy in treated and fellow eyes with other studies should be interpreted considering all methodological differences of the works.

In conclusion, in this study, we provide important information on MA progression in AMD eyes comparing treated nAMD with fellow dry AMD eyes in a real-life data setting. When followed in a synchronous manner, treated nAMD eyes tend to show more atrophy compared to their fellow dAMD eyes. This indicates that in patients that, for a certain time period, maintain asymmetry between their two eyes in regard to neovascularization presence, this asymmetry tends to exist for the atrophic component of their disease too, during this period. However, when macular atrophy does appear in the dAMD eyes, it has a significantly higher growth rate compared to nAMD eyes. Future works could focus on the one hand in elucidating local or other factors that underlay the asymmetry between the two eyes, and on the other, in investigating the possible protective role of neovascularization over MA growth. Although our findings indicate the absence of a correlation between the number of injections and MA in treated eyes, this study was not specifically designed to answer this question, and as a result, more research may be necessary to clarify this matter.