Development of Macular Atrophy in Patients with Wet Age-Related Macular Degeneration Receiving Anti-VEGF Treatment

Age-related macular degeneration (AMD) is a leading cause of blindness in the developed world [1-4]. Progression from early and intermediate AMD to late AMD is more frequent in patients with bilateral AMD [5]. Late AMD can be divided into 2 forms: exudative neovascular AMD, commonly known as wet AMD (wAMD), and dry AMD. Both can progress to macular atrophy (MA), some forms of which are called geographic atrophy. Here, we use the term “wAMD” to refer to the exudative form of neovascular AMD and the term “macular neovascularization (MNV)” to encompass both exudative and non-exudative neovascular AMD.

wAMD is characterized by pathological neovascularization into the subretinal or retinal pigment epithelium (RPE) spaces [6]. Although there is heterogeneity in the presentation and underlying pathology of wAMD, the consensus is that there are 3 primary types of MNV. Based on the proposed definitions from the 2020 Consensus on Neovascular Age-Related Macular Degeneration Nomenclature study group, type I MNV is characterized by neovascularization from the choriocapillaris into the sub-RPE space and may result in the development of pigment epithelial detachments (PEDs) [7]. Type II MNV originates from the choroid, traversing Bruch’s membrane and the RPE to proliferate in the subretinal space [7]. Type III MNV, historically known as retinal angiomatous proliferation, typically originates as neovascularization from the deep retinal capillary plexus and grows toward the outer retina [7]. All 3 types of MNV may lead to exudation and fluid accumulation [7].

MA is characterized by photoreceptor death and vision loss, and typically follows progressive atrophy and thinning of the RPE and choriocapillaris [8-12]. Thinning of the Henle fibre layer, which contains the photoreceptor axons, may be used to identify photoreceptor loss [13]. Photoreceptor loss can also be defined by several optical coherence tomography (OCT) criteria, including the loss of the ellipsoid layer and external limiting membrane [14], and thinning of the outer nuclear layer, which appear together with the Henle fibre layer and photoreceptors as a single hyporeflective band on OCT images [13].

MA may occur at any point in AMD and arise by several distinct mechanisms (Table 1). Primary MA (without prior wAMD) that follows age-related choroidopathy is well demarcated and precedes atrophy and photoreceptor loss, tending to occur in patients aged >80 years and in association with reticular pseudodrusen and/or glaucoma [16, 17]. In addition, thinning of the choroid is linked to the development of outer retinal atrophy after regression of subretinal drusenoid deposits [18]. The diffuse-trickling subtype of MA (named after its appearance on fundus autofluorescence imaging [19]) can progress rapidly [20] and is associated with rarefaction of the large choroidal vessels [21]; a link to systemic cardiovascular disorder has also been suggested [20]. Other processes may lead to primary MA. For example, there is evidence that, in some eyes, marked photoreceptor loss can occur early and precede RPE loss [22]. Wu et al. [23] described this sequence in drusen-associated MA, wherein photoreceptor loss precedes choriocapillaris loss – with the first signs of atrophy being loss of RPE and inner-segment ellipsoid bands on OCT, a process that they termed “nascent geographic atrophy” [23].

The precise definition of MA remains under discussion, with some researchers continuing to use the older term “geographic atrophy” to describe primary MA. Geographic atrophy has historically been used to refer to areas of atrophy exclusive of wAMD and MNV [24]. This type of primary MA appears to have a 2-stage development process: initiation and progression. Once the process is initiated, the area of atrophy enlarges concentrically, giving rise to a distinct demarcation between the atrophic and non-atrophic retina that is “geographic” in profile. By contrast, MA that arises secondary to wAMD often lacks a sharp boundary [6, 25]. The term MA initially emerged to describe the development of this existing “in-lesion” atrophy in eyes with wAMD and MNV but is increasingly used as a more general term that also encompasses de novo “extra-lesion” atrophy [26]. The specific pathogenesis of MA is unclear, and the progression and development of in-lesion atrophy may be different from those of primary or extra-lesion atrophy [27]. Of the patients with dry AMD, 10–15% have notable vision loss due to MA [28]. There is currently no approved treatment to slow or reverse vision loss in MA [6, 29].

Anti-vascular endothelial growth factor (VEGF) agents are an established means of treating patients with wAMD [30]. In 2004, pegaptanib became the first US Food and Drug Administration-approved therapy for wAMD but has since been superseded by other VEGF-A inhibitors, such as ranibizumab, bevacizumab (off label), brolucizumab, and aflibercept [31-34].

There has been discussion of emerging evidence suggesting that anti-VEGF treatment may play a role in the development of MA in some patients [26, 30, 35-39]. As there are no effective treatments for MA [25], this can lead to irreversible vision loss. The development of MA and the extent of lesion size are associated with visual decline in patients with wAMD treated with anti-VEGF agents [40]. Establishing the risk factors that precipitate or accelerate the onset and progression of MA could facilitate the development of novel and improved treatments for late AMD, as well as aid the identification of patients with wAMD who could benefit from additional treatment.

Several clinical trials have included the development of MA as an outcome during treatment of wAMD with anti-VEGF agents (Table 2). The results from a 5-year follow-up study including 647 patients from the original Comparison of Age-Related Macular Degeneration Treatments Trial (CATT), which investigated the efficacy and safety of monthly versus “as needed” schedules of ranibizumab or bevacizumab [41], showed that MA was present in 41% of gradable eyes, with an average follow up of 5.5 years [42, 43]. Similarly, in the Inhibition of VEGF in the Age-Related Choroidal Neovascularization (IVAN) trial, which compared the efficacy of monthly or as needed intravitreal injections of bevacizumab with ranibizumab in 610 patients with untreated wAMD [44], 30% of patients developed extra-lesional MA. In a further one-third of cases, MA developed within the wAMD lesions. Although no association between incident MA and treatment group was found [45], significantly more patients developed MA in the monthly administration group than in the “as needed” group (34% vs. 26%, respectively; p = 0.03) [46].

Unlike the CATT and IVAN trials, the RIVAL study specifically compared the development of MA in patients with wAMD treated with different anti-VEGF agents. In total, 281 patients with untreated wAMD were enrolled and received either ranibizumab or aflibercept for 3 months followed by a “treat-and-extend” regimen, during which disease activity was monitored [47]. Although the choice of agent did not significantly affect the likelihood of developing MA, the proportion of patients who developed MA increased in both groups over the 24-month treatment period (ranibizumab 5–37%; aflibercept 6–32%) [48].

A subanalysis of the HARBOR study was also used to specifically examine the incidence of MA in patients with wAMD; the analysis included 1,095 evaluable patients with wAMD treated with either monthly or as needed intravitreal ranibizumab [27]. Incidence of MA was 29.4% at 24 months and the risk factors included intraretinal cysts and fellow eye atrophy. In addition, subretinal fluid (SRF) was associated with a lower risk of MA, while treatment with ranibizumab was not associated with MA development [27]. As patients with and without MA had mean best-corrected visual acuity gains from baseline over 24 months (+6.7 and +9.1 letters, respectively), the authors concluded that the benefits of ranibizumab for the treatment of wAMD outweighed the risks of MA development during this 2-year period [27]. However, long-term data are yet to be reported.

In a similar post hoc analysis of 60 patients with wAMD from the TREX-AMD trial, which examined monthly versus treat-and-extend regimens of ranibizumab, 10% of eyes with no MA at baseline had developed MA within 18 months of MNV. Of the 43.3% of eyes with MA and MNV at 18 months’ follow up, 84.6% had evidence of overlap between areas of MA and MNV [49], indicating a topographic correspondence between the development of MNV and MA.

Finally, the SEVEN-UP trial investigated longer outcomes (7–8 years) in 65 patients from 3 previous studies of ranibizumab in wAMD (MARINA, ANCHOR, and HORIZON). MA was detected in 98% of eyes and the area of atrophy was significantly correlated with poor vision outcomes (p < 0.0001) [50]. One-third of patients had poor vision outcomes, with visual acuity declining by ≥15 letters. This suggests that the current standard of care is not sufficient to prevent vision loss in the long term for many patients.

Overall, the results from previous clinical trials show that at least one-quarter of patients with wAMD treated with anti-VEGF agents develop MA within a follow-up period of 12–24 months [27, 42, 45, 48, 49]. Furthermore, the longer the follow-up period after a clinical trial, the greater the reported incidence of MA, and this incidence may reach 100% when the follow-up period is extended to 7–8 years [50]. However, it is unclear if this represents a relationship between treatment duration and MA development, or simply shows the natural disease progression. The outcomes of the IVAN trial suggest that there could be a link between the degree of anti-VEGF treatment and MA progression, although the results were inconclusive [45]. It should also be noted that patients enrolled in clinical trials may not be representative of real-world patients with wAMD receiving anti-VEGF treatment. A closer examination of the potential risk factors that could precipitate or accelerate the progression of MA, as summarized in Table 3 and discussed later, may provide a more complete picture.

“Soft” refers to both the size of the drusen (in this case, large, exceeding 125 μm in diameter) and their indistinct edges [51]. The results from the Age-Related Eye Disease Study (AREDS) showed a strong association between drusen and the development of MA. Drusen were found in 100% of eyes at sites where MA later developed [52, 53], and the presence of multiple large drusen increases the probability of developing MA (15-year odds ratio [OR] 14.5, 95% confidence interval 5.9–35.7; 10-year rate of 26% in patients aged 75–80 years) [9, 54]. Likewise, drusen-associated materials are linked to a higher risk of developing MA. One speculation is that debris may extrude from soft drusen into subretinal spaces, leading to excessive phagocytosis that in turn results in RPE death, which could hypothetically lead to the onset and progression of MA [55]. Ultimately, the presence of drusen is a marker of a system that is “under stress” due to a generalized insult, and therefore may precede cellular atrophy and death [9].

Pachydrusen are a recently described subtype of drusen that are large (>125 μm), few in number, and with distinct borders. Pachydrusen are associated with a thickened choroid [56] and appear to arise from a different process to soft drusen, thus representing a distinct entity [57]. At present, the relationship between pachydrusen and MA development is unclear.

Pseudodrusen, identified as hyper-reflective material anterior to the RPE [58, 59] and known as subretinal drusenoid deposits [60], have been independently associated with the onset of MA [54, 61, 62]. Areas of the eye with pseudodrusen are more likely to develop MA within 2 years than those without pseudodrusen (∼74% vs. 42%, respectively) [16, 63], and MA may progress more quickly in eyes with pseudodrusen than drusen alone [62]. Furthermore, the results from a retrospective study in patients with wAMD receiving long-term anti-VEGF treatment showed that reticular pseudodrusen were associated with a greater mean area of MA outside of the wAMD lesion (p = 0.018) [26]. Similarly to drusen, the presence of pseudodrusen may indicate a system “under stress,” thus preceding cellular atrophy [9].

Several demographic factors have been linked to MA. In particular, increasing age, smoking (historic or current), and hypertension are considered general risk factors for the development of MA [28, 64-68]. Smoking is associated with an increased risk of MNV; smoking >40 pack-years is associated with an OR of 3.43 for MA and 2.49 for MNV [64]. Ethnicity has also been explored as a risk factor for MA; Caucasians are more likely than people of other ethnicities to develop large drusen and progress from medium-to large-sized drusen, which may increase their risk of progressing to MA [68].

Drusen, which are associated with MA, as described earlier [52, 53], contain cholesterol [69]. Some research has shown that higher levels of total serum cholesterol are significantly associated with the incidence of MA (OR 1.08 per 10 mg/dL) [65], although other work has shown a relationship with serum high-density lipoprotein but not total serum cholesterol (OR 1.20) [70] or no relationship at all [71]. The results from 1 study in a small population (n = 26) showed that high-dose statin treatment (80 mg/day) in patients with AMD resulted in regression of drusen deposits and improvement of visual acuity [72]. Several studies exploring the effect of statins on the progression of AMD have produced inconclusive results, suggesting that any effect is, at best and small [73]. Although the relationship between cholesterol, AMD, and MA is unclear, elevated cholesterol may be a marker for patients with AMD at high risk of developing MA.

Recent research has identified 3 possible risk groups for MA based on a cluster analysis of genotypes and clinical features [74]. The first risk group had a “high complement genetic risk score” and were clinically characterized by the presence of foveal atrophy and large, soft drusen [74]. This group was associated with a high genetic risk score for genes involved in lipid metabolism. In this group, the 7-year risk of progression to MA from intermediate AMD (defined by the presence of soft drusen) was primarily driven by polymorphisms in APOE, LPL, and CFH (risk scores of 100.0, 87.5, and 76.1, respectively) [28]. It should be noted that environmental factors, such as smoking and body mass index, contributed more than genetic factors to the progression of MNV (as opposed to MA) [28].

The second risk group had a “low complement genetic risk score” and were characterized by foveal atrophy and few drusen [74]. The presence of few drusen suggests the presence of pachydrusen, which are associated with a thick choroid and do not have any clear genetic associations [75].

The third risk group had “high age-related maculopathy susceptibility” and were characterized by reticular pseudodrusen and extrafoveal atrophy [74]. Patients in this group had features of MA associated with a thin choroid and the diffuse-tickling phenotype on fundus autofluorescence, and would be expected to progress rapidly. This group exemplifies an increasingly well-defined subgroup that has genotype correlations both for MA, and for the associated features of this subgroup.

CFH and ARMS2 polymorphisms are associated with reticular pseudodrusen [21]. Furthermore, in the elderly population, CFH polymorphism is associated with choroidal thinning [74, 76], while the ARMS2 polymorphism is significantly associated with a higher incidence of MA (p = 0.01) [63]. Rapid progression of MA is more frequent among patients with CFH and ARMS2 polymorphisms [63].

It is unclear if VEGFA polymorphisms are indicative of response to anti-VEGF agents; evidence for and against an association between VEGFA polymorphisms and anti-VEGF response has been previously reported [77, 78]. Finally, not all research supports distinct genetic etiologies for MA; some work has found no significant genotype-phenotype associations. This suggests that MA phenotypes may exist along a spectrum, rather than as distinct subtypes [79].

Development of atrophy often follows collapse of PEDs [54, 80, 81]. There are 3 major types of PEDs that occur in AMD: serous, drusenoid, and PEDs secondary to subretinal neovascularization [82].

Serous PEDs appear as sharply demarcated domes in the RPE [81] and likely arise from obstructed movement of fluid between the RPE and choriocapillaris by the age-related formation of a hydrophobic barrier [83]. Focal RPE damage as a result of serous PEDs may result in atrophy [84].

Drusenoid (or solid) PEDs are formed by the conflux of large areas of soft drusen, typically located in the central macula [85]. The results from 1 study showed that 49% of eyes with drusenoid PEDs developed MA, while 13% developed MNV after a mean follow-up of 4.6 years [85]. Within 10 years, 75% of eyes with drusenoid PEDs developed MA and 25% developed MNV [85].

Pigment epithelial detachments may also occur secondary to subretinal neovascularization, termed “fibrovascular PEDs” [81, 86]. Hemorrhages may localize to the sub-RPE space of fibrovascular PEDs, causing cell damage and death; therefore, hemorrhagic PEDs are associated with poor visual acuity [87]. They may also lead to RPE tears [88].

Overall, the presence of PEDs may be a precursor to the onset of atrophy, and there is a higher rate of MA in eyes with complete anatomic resolution of PEDs via anti-VEGF therapy [80]; RPE tears subsequent to PEDs leave areas denuded of RPE [89], and it is these areas that develop overlying retinal atrophy. As such, the presence of PEDs may also be a risk factor for MA specifically in eyes treated with anti-VEGF agents.

There is evidence from clinical trials that the wAMD phenotype of a patient may affect their risk of developing MA. Some evidence indicates that patients with type III MNV are at a significantly greater risk of developing de novo MA (p = 0.001) [37, 53]. There is also a high probability of developing RPE atrophy in the fellow eye of patients with type III MNV (∼84%) [90]. As type III, MNV and MA share some common etiology, their association may be rooted in similar pathology rather than a causative relationship; for example, both are associated with reticular pseudodrusen, which are considered a potential risk factor for MA [91]. Furthermore, patients with type III MNV have reduced choriocapillaris flow [92], which may limit response to increased levels of VEGF and be a risk factor for onset or progression of MA.

The choroid supplies oxygen to the outer retina, and thins over time, from ∼200 μm at birth to ∼80 μm thickness by the age of 90 years [93, 94]. Choroidal thinning decreases retinal vascularization, compounding retinal ischemia, and leading to atrophy [95].

Treatment of wAMD with anti-VEGF agents can lead to thinning of the choroid [96, 97], and a thin choroid is associated with poor outcomes post anti-VEGF treatment. This has been further validated by a study showing that that sub-foveal choroidal thickness significantly correlated with visual outcome (p = 0.003) [98]. A thinner choroid also correlates with increased MA area [99]; OCT imaging showed that the choroid of eyes with MA is thinner than in unaffected eyes [19]. RPE-derived VEGF is essential to the development of the choroid and choriocapillaris [100]; thus, the use of anti-VEGF agents may compound choroidal thinning in patients with wAMD (who are typically aged ≥50 years), further increasing their risk of developing MA and worsening visual outcomes. In eyes with type III MNV, choroidal thinning is specifically associated with an increased risk of MA development subsequent to anti-VEGF treatment [101]. In these patients, the loss of choroidal circulation due to atrophy precedes pathological neovascularization [102]; therefore, treatment of neovascularization with anti-VEGF agents may revert the disease to its previous atrophic state. As such, multiple risk factors for MA in 1 eye, including MNV subtype, older age, and choroidal thickness, may compound one another.

Although iron is necessary for oxygen transport in the retina, a build-up of excess iron (e.g., from hemorrhages) is toxic and induces a fibrotic response [103, 104]. Patients with wAMD often develop subretinal hemorrhages; these may lead to fibrosis, which in some cases precipitates MA [105]. Excess iron in the retina can form damaging reactive oxygen species through the Fenton reaction. Iron will also induce activation of the NLRP3 inflammasome, a pathway implicated in AMD [106]. NLRP3 upregulation occurs in the RPE in MA and wAMD [107]. Post-mortem examinations have revealed that the retinas of patients with AMD have significantly increased iron levels (both chelatable and non-chelatable) compared with those from age-matched controls [108]. Importantly, iron accumulation was localized to areas with extensive photoreceptor loss and disorganization [108]. The development of MA subsequent to wAMD often has a close topographic relationship with the wAMD exudative lesion area, emphasizing a potential association between hemorrhage and atrophy [45, 109].

Pre-existing areas of atrophy in wAMD can expand over the course of treatment with anti-VEGF agents [36]. This may result from the mechanism of action of these agents, which counteract the role of VEGF in vascular maintenance [110]. In the IVAN trial, significantly more patients developed MA when receiving regular intravitreal injections of anti-VEGF than those who received it less frequently (34% vs. 26%, p = 0.03) [46]. Similarly, the results from the CATT study showed that monthly dosing of anti-VEGF led to a higher risk of MA development than dosing as needed [37]. A meta-analysis of 31 articles (including 4,609 study eyes) found a moderate positive linear correlation between the total number of anti-VEGF injections and incidence of MA in patients with wAMD (p = 0.01) [111]. Conversely, results from 1 retrospective study demonstrated that overall injection frequency was not associated with MA lesion size and that higher injection frequency led to decreased lesion growth [26]. Administration of anti-VEGF treatment has also been associated with an increased incidence of RPE tears [112, 113], which have been linked to onset of MA [113, 114].

To date, most comparative trials between anti-VEGF treatments have not shown that one specific anti-VEGF therapy is more likely to precipitate MA than another [41, 47, 115]. However, the results from the CATT study showed that patients treated with ranibizumab were more likely to develop MA than those treated with bevacizumab [37]. The growth rate of MA lesions was also significantly higher in patients receiving ranibizumab than those receiving bevacizumab (0.49 mm/year vs. 0.37 mm/year, respectively; p = 0.03) [37]. Ranibizumab is believed to have a greater ability to penetrate the retina than bevacizumab [116], meaning that the effective dose of ranibizumab may have been higher. Furthermore, the efficacy of bevacizumab in patients with wAMD is more variable than that of ranibizumab [116]. This may suggest that more intense anti-VEGF therapy could increase the risk of developing MA.

In addition to factors that increase the risk of MA, evidence has emerged of factors that may protect against atrophy, such as wAMD type and the presence of SRF. These protective factors could be markers for patients at a lower risk of developing MA subsequent to anti-VEGF treatment (Table 4).

Eyes with type I lesions have mature, tangled blood vessels that may be associated with a lower risk of MA development, shown on OCT imaging [38, 117]. Patients with type I MNV have a reduced risk of intralesional MA progression than those with other types of MNV (OR 0.31) [45, 118]. Type I MNV localizes beneath the RPE and therefore may also protect the RPE and photoreceptors from degeneration [118].

Treated chronic fibrovascular PEDs can develop layered hyper-reflective bands, termed “multi-layered PEDs” [119], and may confer a protective effect against atrophy of the RPE and outer retina, although the mechanism is unclear. In 1 study, 82% of eyes with multi-layered PEDs developed atrophy significantly eccentric to the area of PED (p = 0.0465) [120].

Outer retinal tubulations (ORTs) are a branching tubular structure located in the outer retinal layer, with hyper-reflective borders that enclose a hypo-reflective center; they are a common feature in advanced disease, present in about 30% of patients with wAMD [121, 122]. ORTs are an important predictor of poor visual outcomes [123] and are associated with neovascular fibrosis and vision loss [121, 124]. They are formed from photoreceptors and Müller cells, and may represent a forme fruste type of repair that occurs mostly in association with type II MNV [125]; notably, ORTs are not a feature of disease activity [126]. MA lesions in eyes with ORTs spreads at a significantly slower rate than lesions in eyes without ORTs (increase of 1.85 mm² vs. 2.67 mm² from baseline to 18 months, respectively; p = 0.001) [127]. However, although patients with ORTs may have slower MA progression, ORTs are also associated with poor vision due to their connection with fibrosis [121, 124].

Greater overall thickness of the subretinal tissue complex (>275 μm relative to ≤75 μm) has a significant protective effect against MA progression (p < 0.001) [37, 43]. MA progresses slowly and is preceded by RPE and choriocapillaris thinning [8-11]; therefore, relatively greater thickness of the subretinal tissue complex may delay the progression of atrophy.

The presence of SRF halves the probability of MA developing within a wAMD lesion [45] and in the HARBOR trial, SRF was associated with a lower risk of MA incidence over 24 months [27]. SRF thickness of >25 μm is significantly associated with a lower risk of MA development (p < 0.001) and its presence has generally been associated with slower progression of atrophy and better visual outcomes [37, 43, 128]. Patients with wAMD whose phenotype derives from SRF alone show low rates of MA development over a 5-year follow-up period, despite 96.2% of eyes showing drusen, a hallmark precursor of MA. SRF may be an indicator of milder or more benign wAMD [109], or may contain neuroprotective factors that promote RPE and outer retina survival [129]. Speculatively, anti-VEGF treatment may cause a reduction in SRF [130] lowering its anti-atrophic effects and precipitating MA. Alternatively, the presence of residual SRF may simply indicate an incomplete blockade of VEGF, allowing preservation of choriocapillaris.

Although the approval of anti-VEGF agents marked an important step forward in the treatment of late AMD, it is crucial that such treatments are used in a way that ensures the best long-term outcomes for patients. Intensive treatment with anti-VEGF agents could result in irreversible vision loss for patients with wAMD by converting the disease phenotype from pathological neovascularization to MA. In some trials, there has been a positive correlation between the number of anti-VEGF injections and the development of MA in the long term [131, 132]. However, in the short term, patients who receive monthly anti-VEGF treatment show greater improvements in visual acuity over 2 years than patients who are treated less frequently [133], despite the increased risk and incidence of MA [37]. As such, it is unlikely that overtreatment with anti-VEGF agents will lead to quantitatively worse vision outcomes for patients in the short term versus no treatment. In fact, vision loss in wAMD typically arises from secondary photoreceptor loss: bleeding, tears, or fibrosis [134, 135]. Secondary photoreceptor loss often results from undertreatment of patients with wAMD [136, 137] and real-world evidence supports that it is indeed undertreatment of wAMD that leads to vision loss for patients [138, 139]. There is a large unmet need for wAMD treatments that can either simultaneously address MA with MNV, or that do not increase the likelihood of developing MA during treatment, as this could lead to vision loss once MNV resolves.

New therapies that target both neovascularization and atrophy should be investigated. Although several anti-complement treatments are in late-stage development for MA, these treatments may increase the risk of MNV [25, 140]. A neuroprotective agent in combination with an anti-VEGF agent may protect against both photoreceptor death and pathological neovascularization (wAMD). Neuroprotective agents that protect photoreceptors from damage will increase the duration of time for which neural tissue can survive, improving vision outcomes for patients with wAMD and MA [141]. This type of therapy may be most effective for the treatment of in-lesion MA, as de novoatrophy tends to be extrafoveal in patients with wAMD and therefore less likely to cause vision loss.

MA encompasses several distinct processes, and its development is predisposed by several risk factors. These factors range from smoking, older age, and high cholesterol levels [28, 64, 66, 67] to the presence of drusen [52-54, 62], PEDs [54, 85], choroidal thinning [95], and specific genotypes or clinical features [28, 74]. Many of these risk factors such as older age, type III MNV, and choroidal thinning have been shown to compound one another to increase the overall risk of MA. Furthermore, treatment with anti-VEGF agents may lead to choroidal thinning, which is a risk factor for MA [96, 97]. At least one-quarter of all patients treated with anti-VEGF agents in clinical trials develop or experience MA onset or progression within 12–24 months’ follow-up. In some trials in which the follow-up period extended over several years, almost all patients with wAMD developed MA [42, 46, 48, 50, 61].

Further research is needed to delineate if and how certain risk factors interact to result in MA in patients treated with anti-VEGF agents; some identified factors may simply be the result of shared pathology resulting from ocular stress. It is important to find appropriate treatments for each patient, or to consider how such treatments might be developed if they are currently lacking. Using risk factors for MA development to identify patients who could benefit from additional treatment may be crucial to improving vision outcomes and reducing vision loss. An identified group of at-risk patients could form the basis for priority inclusion in trials examining novel therapies that target both wAMD and MA.

MA presents a major challenge to retinal health, both as a primary disease and as a secondary complication of wAMD and/or its treatment. Improving our understanding of the causes and risk factors underlying MA may accelerate the development of novel treatments to address MA and its various subtypes.

Medical writing support was provided by Imogen Allred, DPhil, and Tom Priddle, DPhil, of OPEN Health Communications (London, UK), funded by Boehringer Ingelheim.

A.F. declares no conflicts of interest. T.R. has received funding from Bayer, Novartis, and Thea Pharmaceuticals for sponsored talks and advisory boards. V.C. and T.E. are employees of Boehringer Ingelheim.

Funding for medical writing support was provided by Boehringer Ingelheim.

All authors were involved in the conceptual design of the manuscript, drafting and development, and agreement to publish. The views expressed are those of the authors and not necessarily those of the University of Nottingham Medical School, National and Kapodistrian University of Athens, or Boehringer Ingelheim.