Article Text

Download PDFPDF

Morphological characteristics of eyes with neovascular age-related macular degeneration and good long-term visual outcomes after anti-VEGF therapy
  1. Mengyuan Fang1,2,
  2. Karntida Chanwimol2,
  3. Jyotsna Maram2,
  4. Ghazala A Datoo O'Keefe3,
  5. Charles C Wykoff4,5,
  6. David Sarraf6,
  7. A’sha Brown3,
  8. Shaun Ian Retief Lampen4,5,
  9. Brenda Zhou4,5,
  10. Alexander M Rusakevich4,5,
  11. SriniVas Sadda2,6
  1. 1 Department of Ophthalmology, The First Affiliated Hospital of Zhengzhou University, Zhengzhou, Henan, China
  2. 2 Doheny Eye Institute, Doheny Image Reading Center, Los Angeles, California, USA
  3. 3 Department of Ophthalmology, Emory University School of Medicine, Atlanta, Georgia, USA
  4. 4 Retina Consultants of Houston, Houston, Texas, USA
  5. 5 Blanton Eye Institute, Houston Methodist Hospital & Weill Cornell Medical College, Houston, Texas, USA
  6. 6 Department of Ophthalmology, David Geffen School of Medicine at University of California-Los Angeles, Los Angeles, California, USA
  1. Correspondence to Dr SriniVas Sadda, Doheny Eye Institute Doheny Image Reading Center, Los Angeles, California, USA; ssadda{at}


Purpose To analyse the morphological characteristics of eyes with neovascular age-related macular degeneration (AMD) with good long-term visual acuity after anti-VEGF (vascular endothelial growth factor) therapy.

Methods Retrospective, observational study of 175 patients with neovascular AMD with >5 years of follow-up after initiating anti-VEGF therapy. Spectral-domain optical coherence tomography images were assessed for thickness of pigment epithelial detachment (PED), subretinal hyper-reflective material (SHRM), subretinal fluid and subfoveal choroidal, as well as the integrity of the outer retinal bands.

Results The final analysis cohort included 203 eyes (175 patients) followed for a mean of 7.84±1.70 years (range: 5–11). The maximum PED thickness in the foveal central subfield (FCS) was significantly lower (p<0.001) in the poor vision group (13.11 μm) compared with the intermediate (86.25 μm) or good (97.92 μm) vision groups, respectively. In contrast, the maximum thickness of SHRM in the FCS was significantly thicker (p<0.001) in eyes with poor vision (149.46 μm) compared with eyes with intermediate vision (64.37 μm) which in turn were significantly thicker (p<0.001) than eyes with good vision (9.35 μm). The good vision group also had better continuity of all outer retinal bands (external limiting membrane, ellipsoid zone, and retinal pigment epithelium) compared with the other two groups (all p<0.001).

Conclusion A thicker PED and thinner SHRM were correlated with better vision in eyes with neovascular AMD following long-term anti-VEGF therapy. If replicated in future prospective studies, these findings may have implications for design of optimal anatomic endpoints for neovascular AMD treatment.

  • imaging
  • retina
  • treatment other

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information. There is no additional information available.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Age-related macular degeneration (AMD) is the leading cause of severe vision loss in older adults in the developed world.1 With anti-VEGF (vascular endothelial growth factor) therapy, improvement in visual acuity (VA), at least in the short term, has become an expectation of many patients with neovascular AMD. In the long term (ie, >5 years), however, visual outcomes for patients treated with neovascular AMD is variable.2 3 Some studies have demonstrated a decline in VA with long-term anti-VEGF therapy.4–6 While under-treatment and progressive fibrosis is certainly a contributor to vision loss in some cases, development of atrophy with loss of photoreceptors and retinal pigment epithelium (RPE) may also be an important factor.

Several large retrospective studies have reported that baseline VA, choroidal neovascularization (CNV) lesion size and age are important predictors of VA outcomes.7 8 In the Comparison of AMD Treatments Trials, eyes with foveal intraretinal cystoid fluid (IRF), abnormally thin retina, greater thickness of subretinal tissue complex on optical coherence tomography (OCT) at the follow-up visit (week 104) had worse VA outcomes while eyes with subretinal fluid (SRF) at the follow-up visit (week 104) was associated with better VA.9 These morphological features identified through year 1 were maintained or strengthened at year 5.6

Most studies correlating morphology to visual outcomes have focused on the first 1–2 years following anti-VEGF therapy. Relatively few studies have evaluated morphological characteristics beyond 5 years, which may be of relevance to many patients, and may provide guidance as to the optimal morphological target to strive for with therapy. Thus, in this study, we performed a quantitative analysis of spectral domain OCT (SD-OCT) scans from neovascular AMD eyes obtained more than 5 years after the start of anti-VEGF therapy to define morphological features which correlated with good and poor long-term visual outcomes.


Participants and data collection

In this retrospective multicentre study, clinical and imaging data of patients with treatment-naïve neovascular AMD who were treated and then followed for at least 5 years were collected and reviewed. Patients were identified by reviewing the injection logs and clinical records between 2005 and 2018 at Retina Consultants of Houston, Emory Eye Center and UCLA (University of California, Los Angeles) Eye Centers.

For inclusion in the study, eyes were required to have a diagnosis of neovascular AMD and no other retinal disease, with at least 5 years of follow-up data after initiation of therapy. Clinical data (demographic data, best-corrected VA (BCVA)) from the initial (‘baseline’) visit with a diagnosis of active treatment-naïve neovascular AMD were collected. All the patients received anti-VEGF intravitreal injections of bevacizumab (Avastin, Roche/Genentech), ranibizumab (Lucentis, Roche/Genentech) or aflibercept (Eylea, Regeneron Pharmaceuticals) over the course of a minimum of 5 years of follow-up. The treatment regimen was at the discretion of the treating physician, though review of the clinical records suggested that most patients were treated using a treat-and-extend protocol. SD-OCT images and clinical data (BCVA, interval disease diagnosis, number of injections, timing of injections, number of months/years of follow-up) were obtained from over the follow-up period. The last visit with OCT and vision data was considered the final visit and was used for morphological analysis. Interval visits between baseline and final follow-up were evaluated to identify number and type of injections that were administered. Eyes were excluded from this analysis if information regarding injection/treatment at any visit was missing, alternative treatment (aside from intravitreal anti-VEGF therapy; eg, photodynamic therapy) was given, or OCT images at the final visit were of insufficient quality for grading. If both eyes for a subject met the eligibility criteria, they were both included, though intereye correlations were adjusted for in the statistical analysis.

SD-OCT acquisition protocols

All eyes were imaged with SD-OCT using either the Spectralis HD-OCT (Heidelberg Engineering, Heidelberg, Germany) or the Cirrus HD-OCT (Carl Zeiss Meditec, Dublin, California, USA). Across the clinical sites, the Heidelberg SD-OCT acquisition protocol of the collected scans consisted of a 6×6 mm macular cube (20×20 degrees, 49 lines, 512 A-scans per line). The Zeiss SD-OCT acquisition protocol for this study also consisted of a 6×6 mm macular cube with 512 A-scans per B-scan, but a denser volume composed of 128 B-scans (20×20, 128 lines, 512 A-scans per line). Only the final (last available) visit OCT was used for grading of morphological features.

Image analysis

Evaluation of all final visit OCT images was performed by two independent certified OCT graders (MF and JM) at the Doheny Image Reading Center who were masked to the BCVA or any other clinical data from the patients.

The presence or absence of pigment epithelial detachment (PED), subretinal hyper-reflective material (SHRM) and SRF was assessed using all scans of the OCT volume. In addition to this initial qualitative survey, the maximum thickness of PED, SHRM, and SRF were measured across the entire macula (ie, by scrutinising all B-scans) and within the fovea centre subfield (central circle of the Early Treatment Diabetic Retinopathy (ETDRS)-grid, diameter of 1 mm). Maximum thickness was measured using the calliper tools provided by the instrument software. PED was defined as a discrete or localised dome-shaped or irregular elevation of the RPE with separation from the presumed Bruch’s membrane (BM). The maximum PED thickness was measured at the point of greatest separation between the RPE and BM bands, and was measured from the external RPE border to the internal BM border. The type(s) of PED which were present were also assessed as being drusenoid (homogeneous medium internal reflectivity) (figure 1A), fibrovascular (heterogeneous internal reflectivity) (figure 1B), or serous (homogeneous internal hyporeflectivity) (figure 1C). Drusenoid PED was defined as confluent drusen at least 350 µm in the narrowest diameter on fundus photographs.10 Drusenoid elevations smaller than this were not considered in the PED assessment. SHRM was defined as hyper-reflective material above the RPE/Brush’s membrane complex but below the neurosensory retina (figure 1D) and was measured from the inner surface of the SHRM to the inner border of the RPE (when an RPE band could be discerned) or to the BM (when no definite RPE band could be identified). SRF was defined as a hyporeflective space below the neurosensory retina, but overlying SHRM and RPE. Intraretinal fluid (IRF) was defined by the presence of hyporeflective (cystoid) spaces within the neurosensory retinal layers.

Figure 1

(A) Drusenoid pigment epithelial detachment (PED) showing homogeneous medium internal reflectivity. (B) Fibrovascular PED with a shallow irregular retinal pigment epithelial (RPE) elevation and heterogeneous internal reflectivity with bright and dark areas. (C) Serous PED. The nasal portion likely contains some fibrovascular tissue, but the more elevated temporal portion is homogeneous and hyporeflective consistent with serous fluid. (D) Subretinal hyper-reflective material (SHRM). Hyper-reflective material is located between the neurosensory retina and the RPE/Bruch’s membrane complex.

In addition to measurements of these individual lesion subcomponents, subfoveal choroidal (SFC) thickness (measured at the foveal centre point) and the foveal central subfield (FCS) retinal thickness were also measured. SFC thickness was defined as the vertical perpendicular distance between BM and the choroidoscleral interface. The FCS retinal thickness was defined as the retinal thickness within the central circle of the ETDRS-grid (diameter of 1 mm) centred over the fovea. Before computing the thickness values, the graders evaluated all B-scans and manually corrected any segmentation or decentration errors.

Finally, a qualitative assessment of the integrity of the outer retinal bands was performed at the foveola (assessing a region within a 100μm radius from the fovea centre). Specifically, the appearance of the external limiting membrane (ELM), ellipsoid zone (EZ) and RPE bands were individually scrutinised for discontinuities. When an outer retinal band was continuous within this foveolar region it was classified as level 1(intact); when an outer retinal band could be detected but was disrupted within the foveola it was classified as level 2 (disruption); and when an outer retinal band could not be detected at all, it was classified as level 3 (absent).

Statistical analysis

VAs at baseline and at the final follow-up visit (>5 years) were measured with Snellen charts and then converted to logarithm of the minimal angle of resolution (LogMAR) equivalents. Measurement values among the groups were compared using one-way analysis of variance. Bivariate relationships were examined using Spearman correlation coefficients. General estimating equations were used to account for correlations between eyes for patients where both eyes were included in the analysis. Statistical analysis was performed using the IBM SPSS V.23 statistical analysis package. A p≤0.05 was considered to be statistically significant.


Patient enrolment and follow-up demographic information

A total of 175 patients (203 eyes) met the eligibility criteria and were included in this analysis. The mean interval from the baseline to the final follow-up visit was 7.84±1.70 years (with a range from 5 to 11 years, IQR, 6.42–9.42). Demographics and characteristics of the study population are shown in table 1.

Table 1

Follow-up demographics and characteristics of the study population

To assess the morphological characteristics associated with good long-term VA eyes after anti-VEGF therapy for neovascular AMD, eyes were divided into three groups based on the BCVA at the last available follow-up: (1) ‘good vision group,’ BCVA ≥20/40 (66 eyes, 32.5%); (2) ‘intermediate vision group,’ BCVA between 20/200 and 20/40 (91 eyes, 44.8%); (3) ‘poor vision group,’ BCVA ≤20/200 group (46 eyes, 22.7%). These cut-points were chosen as 20/40 is the minimum vision for driving in many regions, and 20/200 is the threshold for legal blindness.

There was no statistically significant difference in follow-up years among the three groups (p=0.19): 7.70±1.55 years in the poor vision group, 8.08±1.73 years in the intermediate vision group and 7.61±1.73 years in the good vision group.

There was no statistically significant difference in age among the three groups (p=0.059), although there was a trend for older age in the poor vision group (85.20±9.47 years), compared with the intermediate vision group (83.03±8.68 years) and the good vision group (81.12±8.74 years).

There was a significant difference in number of anti-VEGF injections given among the groups, both for the total number of injections (p=0.005; intermediate vision: 47.3; good vision: 38.2; poor vision group: 32.7) and mean number of injections per year (p=0.03; intermediate vision: 5.8; good vision: 5.1; poor vision group: 4.2). Overall, there was no significant relationship between total number of injections or mean injections/year with VA (p=0.544, R=−0.043; p=0.615, R=−0.036).

Morphological characteristics

Clinical and OCT morphological characteristics within the three groups at the last available follow-up are summarised in table 2. One hundred and seventy-two (84%) eyes showed evidence of a PED at the last available visit. Among these eyes with PED, 6 (3%) eyes showed a serous type only, 8 (5%) eyes demonstrated a drusenoid type only, and 158 (92%) eyes demonstrated fibrovascular type or mixed (fibrovascular plus another type). The maximum PED thickness at the final follow-up among these eyes ranged from 40 to 612 μm. Maximum PED thickness (figure 2A) was significantly lower in the poor vision group compared with each of the other two groups (p<0.001 for both). A similar overall trend was observed when considering the maximum PED thickness within the foveal centre subfield (figure 2B).

Figure 2

The mean maximum thickness of PED (A, B) and SHRM (C, D) in three groups based on final vision. Poor vision was defined as 20/200 or worse. Intermediate vision was defined as better than 20/200 but worse than 20/40. Good vision was defined as 20/40 or better. FCS, foveal central subfield PED, pigment epithelial detachment; SHRM, subretinal hyper-reflective material. *p<0.05 were considered statistically significant.

Table 2

Clinical and SD-OCT morphological characteristics at the available last follow-up

Maximum SHRM thickness was 218.89±156.63 μm in the poor vision group, 110.13±103.12 μm in the intermediate vision group and 53.52±102.08 μm in the good vision group, respectively (p<0.001) (figure 2C). Similarly, when considering the maximum SHRM (figure 2D) thickness within the foveal centre subfield, the poor vision group had thicker SHRM compared with the intermediate vision group which was in turn thicker than the good vision group (p<0.001).

Of note, there was no significant difference among the groups when considering the maximum thickness of SRF throughout the macula (p=0.348) or within the fovea centre subfield (p=0.087).

The retinal FCS thickness was not significantly different (p=0.548) among the groups. Cystoid macular oedema was observed in 15.3% of eyes, especially in the poor vision group. SFC thickness was numerically thinner in the poor vision group and the intermediate vision group compared with the good vision group, but the difference among groups was not significant (p=0.263). However, when considering the combined maximum PED thickness at the fovea centre subfield with the SFC thickness, this combined measure of all choroidal vascular (including the fibrovascular PED) tissue under the fovea was significantly lower in the poor vision group compared with the intermediate group (p<0.001) and the good vision group (p<0.001) (figure 3A).

Figure 3

(A) Combined PED and choroidal thickness in three groups based on final vision. (B) Long-term visual outcomes according to final lesion composition. The combined thickness is the maximum thickness of the PED in the foveal central subfield (FCS) plus the subfoveal choroidal thickness (SFC). Poor vision was defined as 20/200 or worse. Intermediate vision was defined as better than 20/200 but worse than 20/40. Good vision was defined as 20/40 or better. BCVA, best-corrected visual acuity at the final visit; PED, pigment epithelial detachment; SHRM, subretinal hyper-reflective material. *p < 0.05 were considered statistically significant.

When evaluating the relationship between the integrity of the outer retinal bands and the final logMAR VA, the status of the ELM (R=0.673), EZ (R=0.701) and RPE (R=0.711) in the foveal centre showed a moderately strong correlation (all p<0.001; Spearman correlation coefficients) with acuity, with the worst vision in eyes with an absence of the band.

Final LogMAR VA was also correlated with the foveal thickness of the morphological CNV lesions parameters of interest (table 3). Whereas SRF in the fovea was not associated with vision at this long-term follow-up visit, thicker SHRM was associated with worse vision and a thicker PED was associated with better vision. Although SFC thickness alone was not associated with VA, a thicker combined choroid and PED was associated with better vision. Of note there was no relationship between the neurosensory retinal thickness and VA.

Table 3

Relationship between foveal thickness of morphological parameters and logMAR visual acuity at final follow-up

To better define and illustrate the morphological configuration with the best visual outcome, we calculated the mean VA in eyes with only PED present, only SHRM present, and eyes with absence of both SHRM and PED. VA was worst in eyes with only SHRM present (1.25 LogMAR), followed by eyes with absent SHRM and PED (0.43 LogMAR, p<0.001), with the best acuity observed in eyes with a PED only (0.29 LogMAR, p<0.001) (figure 3B).


In this study, we report on long-term (mean of 7.75 years) visual and morphological outcomes in patients treated for neovascular AMD with anti-VEGF therapy, and define the CNV lesion and outer retinal morphology associated with the best visual outcomes. In our cohort, the frequency of eyes with vision at final follow-up in the legally blind range (VA of 20/200 or worse) was 21.6%, which is similar to the 20% reported by the CATT at 5 years.11 However, CATT reported a higher frequency (49.6%) of patients with driving-level vision (VA of 20/40 or better) at approximately 5 years compared with 32.4% in our study. However, it should be noted that the mean follow interval was 2.25 years longer in our study compared with CATT (mean follow-up time of 5.5 years). These findings highlight that a significant proportion of patients may still retain good vision despite years of neovascular AMD with chronic anti-VEGF treatment.

The patients in the current study received a mean of 5.22 injections per year, which is also similar to the 5-year CATT observations. It should be noted that the treatment regimen in our study differed from CATT in that the centres participating in our study predominantly used a treat-and-extend regimen. In addition, since injection frequency tends to diminish over time in all studies, the fact that our study showed a similar annual injection frequency compared with CATT despite longer follow-up suggests that in general patients were treated more aggressively in our study compared with CATT. In CATT, patients who switched at 1 year from a monthly to a PRN dosing regimen received 5 to 6 injections on average over the second year and received 4–5 injections per year on average during the 3 years after release from the CATT protocol. In contrast in the SEVEN-UP (7-year outcomes in ranibizumab-treated patients in ANCHOR, MARINA and HORIZON trials) study, participants only received a mean of 6.8 anti-VEGF injections during the entire mean 3.4-year interval between exit from the HORIZON study and the SEVEN-UP evaluation (ie, ~2 per year).5 Such low treatment frequencies may have contributed to the lower percentage of good vision and higher percentage of poor vision reported in SEVEN-UP (with only 23% of eyes achieving a VA of ≥20/40 compared with 37% of eyes with VA of ≤20/200 after 7 years). Thus, as many recent studies have shown, particularly real-world studies,12 more frequent treatment appears to be associated with better long-term outcomes, and many patients require treatment through 5 years and beyond.

Despite this emerging concept that eyes receiving more injections in the long-term achieve better visual outcomes, the relationship appears to be more complicated than this in the long term. Whereas eyes in our intermediate vision group received significantly more treatments than the poor vision group (5.75 /year vs 4.24 /year), the eyes with best visual outcomes received an intermediate number of injections (4.94 /year). Based on this, not surprisingly, we observed no correlation overall between number of injections and final vision. While it has been assumed that reduced numbers of injections given in real-world setting may be related to non-compliance with protocol and factors related to treatment burden, the underlying reasons may be more complex. In a post hoc review of cases with good outcomes and few injections over time, we observed that in many of these cases, these eyes developed a stable situation where a fibrovascular PED could be noted on OCT with intact overlying RPE and outer retinal bands and no evidence of recurrent exudation despite long (>16 week) treatment intervals. Our findings would appear to suggest that the optimal number of injections in the long-term may vary greatly among individuals.

A number of studies have specifically evaluated the relevance of PEDs with regards to visual outcomes.10 13–15 PEDs are an important marker of disease severity and progression in neovascular AMD, though their relevance to visual outcomes has yielded inconsistent findings. Multilayered PEDs, as defined by a fibrovascular and fibrocellular layer with SD-OCT, have been shown to be associated with a favourable morphological outcome associated with stable and functional VA despite chronic anti-VEGF therapy over many years. In other studies of neovascular AMD, eyes with PEDs have been associated with worse VA outcomes.16 17 One analysis of 125 patients with vascularised PED showed that anti-VEGF therapy did not improve vision, in part due to the development of RPE tears.18 In a prospective study of patients with vascularised PED and serous PED, Panos et al 19 observed that anti-VEGF therapy was beneficial in improving vision in both types of PED, but the actual anatomical response in the PED was not correlated with BCVA improvement at 12 months follow-up after the first injection. In the PrONTO study, there was also no correlation between PED at baseline or 3 months with VA at 12 months.20 21 However, in a post hoc analysis of the HARBOR trial, Sarraf et al 22 found that patients with PED exhibited greater BCVA outcomes than patients without PED at baseline.

Given these inconsistencies, there has been uncertainty regarding the optimal management of PEDs in the setting of neovascular AMD. For example, it is not clear whether continued aggressive treatment with the goal of achieving complete or maximal flattening of PEDs would yield the best visual outcomes for patients in the long term. Sarraf et al 22 and Khanani et al 23 showed that PED flattening was not an optimal morphological outcome and that the goal of anti-VEGF therapy should be aimed at reducing intraretinal and subretinal fluid not necessarily reducing PED height. Gaining further insight into this question was an important motivation for this study.

Overall, our findings suggest that complete resolution of the PED may not be necessary for VA gains or the best visual outcomes. However, we should acknowledge that our study was not designed to specifically address this issue given its retrospective design.

There has been considerable interest over the past decade in better understanding the relationship between morphological parameters and visual outcomes. We first reported that thicker hyper-reflective material in the subretinal space was the most important predictor of poor visual outcomes in patients with neovascular AMD.24 This finding regarding the importance of SHRM was subsequently confirmed by multiple investigators, including the CATT trial group.9 25 In our study 68% of eyes displayed SHRM at final follow-up which is similar to the frequency of 66% reported at 5.5 years by CATT. The fact that thicker SHRM is associated with worse visual outcomes makes sense as one would expect that such material, which may represent fibrovascular or fibrocellular tissue,26 separating (or replacing) the RPE from the photoreceptors would result in photoreceptor dysfunction and loss.27

Some investigators have suggested that a controlled (ie, treated or non-exudative) type 1 NV membrane with intact overlying RPE may provide a trophic effect to the overlying retina.28 For example, Christenbury et al 29 observed that when RPE and photoreceptor atrophy develops after long-term anti-VEGF therapy, it tends to develop in regions at the margin of the PED or away from the PED, but not over the PED itself. In an exploratory subanalysis from the HARBOR trial, Sarraf et al observed that patients with PED started with better vision than those without PED at baseline, and that the month 24 mean BCVA also was better in patients with PED at baseline compared with patients without PED at baseline. Furthermore, patients with complete resolution of PED did not necessarily see an additional vision benefit and were more likely to demonstrate macular atrophy at month 24.

Our might hypothesise that a type 1 CNV may be a compensatory response by the body to an underlying choroidal insufficiency. Previous studies have shown that AMD eyes with thinner choroids are more likely to develop atrophy of the photoreceptors and RPE. In our study, there was a trend for a thinner choroid in the poor vision group, but this did not reach statistical significance. Our study, however, may not have been sufficiently powered to detect a small numerical difference. In addition, more important than the overall choroidal thickness, the status of the choriocapillaris (CC) is likely the more critical factor. There is mounting evidence from OCT angiography with regards to the vital importance of the CC in the pathophysiology of AMD. CC flow deficits have been shown to predict the appearance of drusen and the progression of atrophy.30 CC flow deficits have also been shown to surround CNV and GA lesions.31 A working hypothesis for advanced AMD pathogenesis is that in some cases CNV may be a response (attempting to prevent photoreceptor/RPE loss) to this CC insufficiency, and that eyes which fail to mount an ‘adequate’ (undefined) CNV response go on to develop atrophy. Of note, though choroidal thickness was not associated with visual outcomes in our study, the combined thickness of the choroid and PED was associated with visual outcome, with a thicker combined measurement being associated with better vision. This would appear to further support the concept that the PED may be ‘rescuing’ a choroidal insufficiency.

Taken together, our observations have important implications for defining the optimal endpoint for anti-VEGF therapy. Most clinicians will metre their anti-VEGF therapy based on anatomic response on OCT regardless of whether a PRN or treat-and extend approach is used. It has been unclear, however, whether treatment should be continued (without extension) until the PED resolves despite absence of SRF and/or IRF. Our observations would suggest, however, that a residual PED under anti-VEGF therapy is compatible with good outcomes, and actually, may be preferable. In fact, we observed the best vision in eyes with the thickest PED under anti-VEGF treatment. At this stage, it is premature to suggest that active measures (eg, withholding therapy for short intervals to allow PED enlargement) should be taken to preserve or enlarge PEDs, and in indeed there is compelling evidence to show that leaving CNV un-treated can lead to devastating visual consequences. However, these findings do suggest that more precise longitudinal studies should be undertaken to compare visual outcomes between treatment regimens aimed at achieving different morphological targets.

Another interesting observation from our study relates to central retinal thickness (CRT). CRT has been used as a quantitative criterion to guide retreatment decisions in several previous clinical trials.32 33 However, Simader et al observed no association between retinal function and thickness beyond 3 months.34 Similarly, we observed no significant difference in retinal thickness in eyes with poor or good vision. This highlights the disconnect between retinal thickness and vision. There are several explanations for this. One might expect a thinner retina in eyes with poor vision due to loss of photoreceptors. However, such eyes may develop compensatory changes such as glial proliferation or development of degenerative cystoid spaces. In addition, some eyes with poor vision may have had persistent exudative activity which may have also been associated with worse vision and thicker retina. These inconsistencies and potential multiple mechanisms by which retinal thickness can be impacted by the course of neovascular AMD, highlight the challenges and pitfalls of using CRT as a measure of disease activity or an indicator for treatment decisions in eyes with neovascular AMD.

Our study is not without significant limitations which must be considered when assessing our results. First, as this was a retrospective analysis it was susceptible to selection and ascertainment bias. Moreover, although a thicker PED under anti-VEGF therapy was associated with better vision outcomes, the retrospective design does not allow us to make specific recommendations with regards to how a PED should be managed or to determine whether therapy should be modulated to specifically achieve a thicker PED. Another limitation was the variability is follow-up ranging from a minimum of 5 years to 7–8 years in the majority of cases and as long as 11 years in some cases. Third, for several features such as PED, SHRM and SRF, thickness was only measured at one location (maximum point). Three-dimensional volumetric assessments may have provided a more precise estimate of extent of these features.35 Fourth although most of the investigators contributing cases to this study followed a treat-and-extend protocol, there was no standardisation between centres and no assessment of adherence to protocol. On the other hand, as this is a real-world study, the findings may have more general applicability. Our study also has several strengths including the use of masked, independent, certified, experienced reading centre OCT graders; the use of a standardised grading protocol with definitions for the various morphological characteristics; and the use of dense volume SD-OCT.

In summary, we observed that in eyes with neovascular AMD undergoing anti-VEGF therapy, thicker SHRM is associated with worse visual outcomes whereas a thicker PED under anti-VEGF treatment is associated with better visual outcomes. Longitudinal studies are warranted to more precisely define the optimal anatomic target or goal for the management of these patients.

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information. There is no additional information available.

Ethics statements

Patient consent for publication

Ethics approval

Approval for data collection and analysis was obtained from the institutional review board of each participating centre. As this was a retrospective study, a waiver of informed consent was granted for all clinical sites. All research adhered to the tenets of the Declaration of Helsinki.



  • Contributors MF and SS had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: MF and SS. Acquisition, analysis or interpretation of data: All authors. Drafting of the manuscript: MF and SS. Critical revision of the manuscript for important intellectual content: MF, GDO’K, CW, DS and SS. Statistical analysis: MF, KC and JM. Administrative, technical or material support: GDO’K, CW, DS, AB, SIRL, BZ, AMR and SS. Study supervision: GDO’K, CW, DS and SS.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. CW reported receiving grants, personal fees or nonfinancial support from Adverum, Aerie Pharmaceuticals, Aldeyra, Alimera Sciences, Allergan, Apellis, Arctic Vision, Arrowhead, Bausch + Lomb, Bayer, Bionic Vision Technologies, Boehringer Ingelheim, Chengdu Kanghong Biotechnologies (KHB), Clearside Biomedical, EyePoint, Gemini Therapeutics, Genentech, Graybug Vision, Gyroscope, IONIS Pharmaceutical, IVERIC Bio, Kato DSRC, Kodiak Sciences, LMRI, Neurotech Pharmaceuticals, NGM Biopharmaceuticals, Novartis, OccuRx, Ocular Therapeutix, ONL Therapeutics, Opthea, Outlook Therapeutics, Oxurion (formerly Thrombogenics), Palatin, Pentavision, PolyPhotonix, RecensMedical, Regeneron, RegenXBio, Roche, SAI MedPartners, SamChunDang Pharm., Santen, Senju, Taiwan Liposome Company, Takeda, Thea Open Innovation, Verana Health, Visgenx, Xbrane BioPharma. DS reported receiving grants, personal fees, or nonfinancial support from Amgen, Genentech, Heidelberg, Optovue, Regeneron, Topcon, Bayer, Iveric Bio, Novartis. SS reported receiving grants, personal fees, or nonfinancial support from Carl Zeiss Meditec, Nidek, Topcon, Heidelberg, Optos, Centervue, Amgen, Allergan, Genentech/Roche, Oxurion, Novartis, Regeneron, Bayer, 4DMT, Merck, Apellis, Astellas.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles

  • Highlights from this issue
    Frank Larkin