The Progression of Stargardt Disease as Determined by Spectral-Domain Optical Coherence Tomography over a 24-Month Period (ProgStar Report No. 18)

Although there are currently no approved treatments for Stargardt macular dystrophy (STGD1; OMIM 248200), the most common juvenile macular dystrophy due to disease-causing sequence variants in the ABCA4 gene, several therapeutic approaches including pharmacotherapy, gene augmentation, and stem cell therapy are in early clinical phases [1, 2]. However, the success of future clinical trials will largely depend on how efficacy is demonstrated and therefore robust outcome measures are needed. Analysis of visual acuity (VA) in the “Progression of Stargardt disease” (ProgStar) cohorts revealed that VA may not be a potential outcome measure for trials or only of value in select subgroups [3‒5]. Incidence and progression rates of atrophic lesions as determined by fundus autofluorescence (FAF), the primary outcome measure in the ProgStar studies based on its acceptance by the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) as a surrogate endpoint [6‒8]. Changes in the size of lesions defined as “definitely decreased autofluorescence (DDAF)” are very likely to be robust measures, but this is frequently not the case for lesions of questionably decreased autofluorescence (QDAF), because they are often poorly demarcated and may also transform over time into DDAF [8‒10]. However, lesions of QDAF often represent earlier stages of STGD1 in which retinal tissue might be still amenable to rescue by pharmacotherapy or gene augmentation therapy, and hence other surrogate endpoints of the effectiveness of treatment are needed. There are suggestions that the integrity of photoreceptors as determined by spectral-domain optical coherence tomography (SD-OCT) can be affected earlier than changes in the retinal pigment epithelium (RPE) as detected by FAF, therefore SD-OCT changes were included as a secondary outcome measure [11]. Furthermore, there are several useful candidate SD-OCT quantitative measurements such as the means of retinal (layer) thickness(es) and volume(s) [12, 13]. In addition, SD-OCT-based quantification of photoreceptor integrity has been accepted as a surrogate endpoint measure of efficacy by the FDA for outer retinal degenerative diseases [14]. Herein, we present the estimated progression rates of atrophy derived from SD-OCT analysis in the prospective ProgStar study.

The study design, inclusion, and exclusion criteria of the prospective ProgStar study have been published in detail previously [6]. Briefly, this study was a multicenter, observational cohort study conducted at nine centers in the USA, Europe, and the United Kingdom (the ProgStar study group is listed in the online suppl. material; for all online suppl. material, see https://doi.org/10.1159/000540028). Key inclusion criteria relevant for this report were: (1) presence on FAF of at least 1 well-demarcated area of atrophy with a minimum diameter of 300 µm, and the total area of all atrophic lesions being 12 mm² or less (equivalent to no more than 5 standard disc areas in a least 1 eye) as certified by the site PI; (2) presence of at least 2 likely disease-causing variants in ABCA4 or 1 likely disease-causing variant with at least 1 eye with flecks at the level of the RPE typical for Stargardt disease; (3) clear ocular media and adequate pupillary dilation to permit good-quality FAF and SD-OCT imaging; (5) minimum age of at least 6 years.

Patients underwent a baseline visit and four follow-up visits at six month-intervals (time window for each visit ± 5 weeks). At each visit, in addition to a complete clinical examination, FAF, SD-OCT, and mesopic microperimetry were obtained. The focus of the present paper is to analyze SD-OCT imaging acquisition and analysis.

The detailed image acquisition protocols have been published previously [6]. Using a Heidelberg Spectralis SD-OCT device (Heidelberg Engineering, Heidelberg, Germany), an infrared reflectance image and a SD-OCT cube scan (20° × 20° volume scan consisting of 49 B-scans) were obtained; the last was to be centered onto the anatomical fovea and had an OCT ART mean of ≥9 frames. After acquisition, the baseline visit scans were set as reference scans. For follow-up visit scan acquisitions, the “follow-up” option was utilized to refer to corresponding baseline visit scans if possible. All clinical center staff (photographers) were certified by the central reading center (RC, located at Doheny Imaging Reading Center, Los Angeles, USA) prior to enrollment of the first patient.

De-identified images were sent from the participating sites to the reading center (RC); these included short-wavelength reduced-illuminance and conventional autofluorescence images, infrared-reflectance images and SD-OCT cube scans (20° × 20° volume scan consisting of 49 B-scans). At the RC, a dual, independent, masked grading was performed by two graders (with at least one being a senior grader) including adjudication processes by a senior investigator where necessary as previously described in detail [6]. Images with insufficient quality (ungradable images) were excluded from analysis. DIRC’s validated SD-OCT grading application, OCTOR 3.0, was used for viewing, annotating and quantifying OCT scans. The Doheny Image Analysis Laboratory’s (DIAL’s) segmentation algorithm was used to generate automated segmentation which served as the starting point for manual readjustments, to support quantitative assessment using OCTOR 3.0. Additionally, Heidelberg Engineering Heidelberg Eye Explorer (HEYEX) review software was used to generate certain linear measurement grading variables (i.e., choroidal thickness). Images of the baseline, 6-month, 12-month, and 24-month visits were graded. Results for thickness and intact area of the individual layers as described below were derived from grading for the inner subfield, inner ring (IR), and outer ring (OR) of the on Early Treatment of Diabetic Retinopathy Study (ETDRS) subfield grid [15] and sent to the central data coordinating center.

The presence of the following qualitative and potentially cofounding parameters was determined: (1) intraretinal cystoid spaces; (2) subretinal fluid (SRF); (3) epiretinal membrane (ERM); (4) vitreomacular traction (VMT).

At least 25 of 49 B-scans per cube scan were graded per eye/visit using an adopted grading method as previously described [16]. In each of the selected B-scans of the 20° × 20° high resolution volume scan, the following boundaries, presented in innermost (anterior) to outermost (posterior) order were segmented:

• Inner limiting membrane (ILM): inner was placed along the inner boundary of the ILM.

• (Dendritic) outer plexiform layer (OPL): outer was placed along the inner boundary of the OPL. In areas where the inner boundary of the OPL was invisible, the visible edge of the OPL boundary was to be merged posteriorly to the nearest intact boundary, e.g., ELM, ellipsoid zone (EZ), or photoreceptor layer (PRS).

• External limiting membrane: in areas where the ELM was absent/invisible, the edges of the visible ELM were merged posteriorly to the EZ line, if the EZ band was intact. In areas where both the ELM and the EZ band were absent, both the ELM and the EZ lines were merged with the PRS line posteriorly.

• Inner segment-outer segment (IS-OS) junction (now termed EZ); this boundary was placed on the outer (posterior) side of the IS-OS junction/EZ band. In areas where the EZ band was invisible, the EZ line was merged with the PRS line, which was already extended anteriorly, in such cases, to merge with the ELM (if intact), as described in the PRS section. In areas where both the EZ band and the ELM are invisible, the visible edges of the EZ band and the ELM were both merged unto the PRS line. In such cases, the PRS line itself remains in its original place on top of the inner border of the RPE.

• Photoreceptor segment layer (PRS) – outer: this boundary was placed on the inner (anterior) side of any SRF/hyperreflective material, etc., and along the inner boundary of any deposits, debris, or fluid (collectively termed as hyperreflective debris) at the level of the RPE (see Fig. 1). In areas where there was none of such material, the line was merged posteriorly to the RPE inner boundary line. If the RPE inner boundary was moved posteriorly and merged to the choroidal line (in areas or RPE atrophy), the PRS is to stay on top of the atrophied RPE remnant rather than following the RPE inner boundary line. In the areas where the EZ band was absent, the PRS line was extended anteriorly and merged into the ELM line. In such cases, the PRS line contributed to the information collected about both the IS and OS. In the areas where both the EZ band and the ELM were absent/disrupted/invisible, the PRS is to remain on top of the RPE and/or hyperreflective debris. In such cases, however, both the EZ and the ELM lines were moved posteriorly and merged to the PRS. In such cases, the PRS line contributes to the information collected about the IS, OS, and ONL.

• Retinal pigment epithelium (RPE) cell layer – inner: in the cases where the RPE was completely or almost completely atrophied, the graders were instructed to merge the inner RPE boundary to the choroidal boundary resulting in a 0-thickness in the intervening areas where merging occurred.

• Choroid – inner

Boundaries were generated from DIAL’s algorithm as a starting point, and any segmentation errors were manually corrected or other boundaries as needed added. Especially, the photoreceptor segment layer – outer was manually readjusted in all instances where SRF or material was visible. Hyperreflective debris was separately delineated and excluded from measurements.

In B-scans where a given layer was completely absent, the immediately adjacent (anterior/posterior) boundaries were snapped together resulting in a thickness value of 0. By applying these boundaries, the following layers were outlined and segmented:

• Mean inner retinal thickness (IR) and intact area: generated from the two boundaries: ILM and inner boundary of OPL

• Mean ONL thickness and intact area (ONL): generated from the two boundaries: inner boundary of OPL and external limiting membrane (ELM)

• Mean IS thickness and intact area (IS): generated from the two boundaries: external limiting membrane and IS-OS junction

• Mean OS thickness and intact area (OS): generated from the two boundaries: IS-OS Junction and inner boundary of RPE cell layer

• Mean RPE thickness and intact area: generated from the two boundaries (RPE): RPE cell layer inner boundary and inner choroid boundary

• Mean total retinal thickness (TR) and intact area: generated from the two boundaries: ILM and inner choroid boundary.

The graders then placed the “Early treatment diabetic retinopathy study” (ETDRS) [17] grid onto the anatomical fovea and results were calculated for the central subfield (CS; 0.5 mm radius), the IR (0.5–1.5 mm) and the OR (1.5–3 mm). The overlap of the ETDRS-grid with the imaged area was defined as the “scanned” area. However, in some cases, the imaged area was shifted, and therefore, the center of these cube scans was not coinciding with the center (CS) of the ETDRS-grid. As a consequence, there was no complete overlap of measurements of the OR of ETDRS grid with the imaged area in such cases, i.e., the scanned area was <21.2 mm², which is the normal size.

Longitudinally, linear mixed models, with time as the independent variable, were used to estimate the yearly change for each outcome. The models included random effects for the intercept and the slope for time which take into account the potential correlation between eyes and the correlation between repeated measurements of the same eye. The rate of thickness and intact area change associated with each variable was estimated in univariate analysis. In cases where the scanned area of the OR was <23.2 mm², the following formula was applied to interpolate the intact area: adjusted intact area = actual intact area × (scanned area/21.23) mm².

A total of 487 eyes with available SD-OCT cube scans at baseline visit were enrolled into the prospective ProgStar study. Cube scans were excluded from analysis if the following factors were present that might influence retinal thickness/intact area measurements: presence of SRF (14 eyes, 2.9%), presence of ERM (22 eyes, 4.5%), presence of vitreo-macular traction (3 eyes, 0.6%) and poor quality (not gradable, 2 eyes, 0.4%): reasons for the last were one with poor signal with the consequence of impossible delineation and segmentation of retinal sublayers and one with incomplete cube scans (i.e., not all 49 B-scans could be acquired). Furthermore, for the analysis of each different subretinal layer of the CS and the two different rings), a thickness >0 µm at baseline was required and enrolled patients had to have at least two visits. Table 1 summarizes the baseline characteristics of the entire ProgStar cohort and the eyes included in this analysis.

All eyes already exhibited significant damage at baseline due to the required presence of atrophy as an inclusion criterion of ProgStar. At first visit, 22 eyes (7.0%) had mean thickness of 0 for ONL, 273 eyes (87.5%) had a mean thickness of 0 for IS/OS, and 121 eyes (38.8%) had a mean thickness of 0 for RPE. Therefore, we report only mean total retinal thickness herein which was 129.7 (±36.3) µm at baseline. The estimated decline over 24 months was −3.1 (±0.3) µm/year (p < 0.0001).

The mean retinal thickness values of TR, IR, ONL, IS, OS, and RPE for the visits at baseline, 6 months, 12 months, and 24 months are summarized in Table 2. Inner retina showed a statistically significant thickening, while all other layers decreased significantly in thickness (p < 0.0001, respectively). Intact area of IR and total retina did not change, while all other segmented retinal layers showed a statistically significant decrease (p < 0.0001; Table 3).

Also in the OR, all segmented retinal layers except the IR and total retina showed a significant reduction in thickness (p < 0.0001; Table 2) and intact area (p < 0.0001; Table 3). The decline in intact area was largest in the inner and outer segments (OSs).

In patients younger than 18 years, mean retinal thickness did decline faster than in patients aged between 18 and 50 years in the CS, IR and OR; in the CS, it declined also faster than in those older than 50 years (Table 4). In the inner and OR, mean retinal thickness also declined faster in patients with duration of symptoms <2 years compared to those with duration of symptoms of longer than two, but less than 5 years (Table 4). Flecks beyond the arcades were associated with a faster decline of mean retinal thickness in the OR (Table 4).

The arrival of SD-OCT revolutionized ophthalmic care and research because of its ability to image the human retina in vivo, providing information on retinal pathology in situ and in real time, with resolutions approaching that of excisional biopsy and histopathology [18]. It has been utilized with a variety of complementary structural and functional tests such as FAF and adaptive optics scanning laser ophthalmoloscopy (AOSLO) to identify the sequence of events that ultimately result in RPE and photoreceptor cell death [19]. Khan and colleagues [19] described the earliest anatomic changes that occur in childhood-onset ABCA4-associated retinopathy as absence of clinically apparent outer retinal atrophy. They observed that one of the earliest observed structural changes were in the ONL, where the SD-OCT images revealed an obvious increase in reflectivity extending internally from the line thought to be the optical correlate of the ELM [20, 21]. The maximum of this thickness change was at the foveola and declined as a function of eccentricity. This uncommon SD-OCT abnormality has been previously described and termed ELM thickening [22, 23], presumably because of the increased reflectivity appeared to be continuous with the ELM. Khan et al. [19] speculated that the identified changes are the result of pathologic disruption within the outer lamella of the ONL, specifically in the cones, because within the ONL, cone photoreceptor nuclei have defined spatial distribution, being most numerous and tightly packed at the fovea and less densely packed, residing more closely approximated to the ELM, in the perifoveal retina. SD-OCT imaging in their case series confirmed that the EZ is still evident in early disease, but it seems to be reduced qualitatively in intensity, suggesting that in addition to the changes to the ONL, there are subtle OS pathologic features [19]. Finally, increased ONL reflectivity seems to be a transient phenomenon, sustained only as long as the ONL volume is preserved, whereas the line representing the ELM remains preserved at the same stage; Khan et al. [19] concluded based on their SD-OCT, FAF, and AOSLO data that cone photoreceptor dysfunction is intrinsic to, and not secondary to RPE failure. This is in agreement with previous studies in which on histopathology, photoreceptors were found to have shorter OSs even when the opposing RPE appears organized [24] AOSLO imaging revealed increased photoreceptor spacing in areas that otherwise appeared normal on FAF and OCT [25, 26]. However, depending on the imaging modality, it has been suggested that photoreceptor loss occurs before RPE loss [11, 27]; that RPE cell loss and/or dysfunction occur first, followed by secondary photoreceptor cell loss [28, 29]; or that changes in the PRS occur simultaneously with the development of abnormalities in the RPE layer [30].

Other retinal layer changes have been also investigated such as the status of intact EZ (EZ or the inner segment/OS [IS/OS] line, as described in our study). This zone likely correlates with mitochondria within photoreceptor ISs and has been associated with photoreceptor health and function [12, 20, 21, 31]. SD-OCT scans in STGD1 patients demonstrated EZ disorganization with a normal-appearing RPE on both SD-OCT and FAF, and when SD-OCT and fundus autofluorescence are compared, the edge of EZ loss was found beyond areas of hypofluorescence on FAF [11, 28, 29, 31]. It has also been observed that EZ loss extends beyond areas of RPE thinning in the nasal and temporal macula in SD-OCT [27, 32]. Evaluating the EZ could also have an advantage as a potential surrogate outcome measure of the efficacy of treatment because it has been reported that the degree of foveal EZ disorganization in STGD correlates with differences in visual acuity, microperimetry sensitivity and extent of fundus lesions, as well as multifocal ERG results [33, 34]. In another study, visual acuity was directly correlated with EZ/photoreceptor OS volume and inversely correlated with en face EZ loss/atrophy and attenuation of the photoreceptor OSs/EZ [35].

Cai et al. [31] recently estimated the mean rate of EZ loss in 31 eyes of 16 STGD1 patients was 0.31 ± 0.31 mm²/year. However, these investigators used an “indirect approach” by labeling the border of EZ loss on a infrared reflectance image and then determining the extent of EZ loss by measuring the bordered area. SD-OCT has already been utilized to determine the decline of total retinal thickness and volume in neurodegenerative diseases such as multiple sclerosis, and has been established as a monitoring tool in neurology [36].

Given these considerations, SD-OCT might offer potential outcome measures for upcoming clinical trials in STGD1 and was therefore chosen as a secondary outcome measure in the ProgStar study [6]. The grading of atrophy in STGD1 using SD-OCT, however, is challenging because the hyperreflective debris in the area of atrophy inevitably lead to a much higher failure rate of the applied software algorithm (e.g., in comparison to dry age-related macular degeneration), and therefore significant manual corrections were necessary [37]. Indeed, in a preliminary study we detected that in 20.2% of B-scans the outer retina was misidentified and more than 30% of B-scans revealed software errors [13]. Therefore, correcting these algorithm errors required a significant amount of time, funding, and personnel despite the application of a grading protocol. In a preliminary study, the adaptive approach described in the methods section yielded thickness and area measurements of retinal sublayers comparable to the reference ground truth derived from using all 49 B-scans in the volume [16]. The grading of the 18 month-visit had to be omitted due to the aforementioned reasons. Additional challenges in analyzing and quantifying SD-OCT images in STGD1 ABCA4-related STGD1 is a phenotypically very heterogeneous disease, especially depending on the stage of disease and the individual phenotype.

We also encountered problems with confounding pathologies such as ERMs that can influence thickness measurements in a degenerative disease. In addition, typical features such as flecks might affect different layers of the outer retina, and at least 5 different types of lesions associated with retinal flecks in STGD1 have been proposed [38]. Flecks appear or disappear over time with possible consequences on thickness and intact area measurements [39]. Also the true nature and precise cellular processes represented by the hyperreflective features (e.g., flecks, debris, compromised RPE) sub/intraretinal fluid (most likely originating “e vacuo” from degeneration) could not always be determined and the transition zones were often too gradual. In a previous effort, we conducted a preliminary study to test the repeatability of SD-OCT grading [40]. These data showed:

• That measurements of the RPE layer had considerable noise with large variability in the difference between gradings (as indicated by poor intra-class correlations (ICCs) ICCs and also by the high relative absolute differences (RADs) between gradings for thickness and intact area

• Measurements of thickness and intact area the inner and OSs in the inner and OR regions had good to excellent the ICCs

• Thickness measurements of ONL in the IR and OR had good ICCs and smaller RADs, but poor ICCs for intact area measurements although the median RAD was 0, respectively.

These observations must be taken into account when analyzing the progression data reported herein. We were initially hesitant to present the data on progressive changes in RPE thickness and intact area measurements because of significant noise. We also think that RPE measurements might be suitable only for distinct subtypes of atrophy. Also the ONL must be carefully reviewed in the context of intact area measurements. Inner segments and OSs (including the EZ) measurements, however, might serve as a potential outcome measure. In general, the change in mean intact area of outer retinal layers, especially IS and OS, seems to be more robust than the change in mean thickness of an individual retinal layer. However, it has also be taken into account that STGD1 is a slowly progressing disease and the observational period of 2 years might be rather short, although statistically significant changes could be observed.

A third difficulty in the application of SD-OCT as a monitoring tool can be the location of measurement. The imaging protocols of ProgStar required the photographers to center the 20° × 20° cube scan on the anatomical fovea and then use the follow-up and eye-tracking function of the Heidelberg Spectralis device. However, patients in the ProgStar study often had (due to the requirement of the presence of a distinct atrophy) eccentric fixation, and therefore the cube scan was not always perfectly centered on the anatomical fovea with the consequential missing of parts of the outer ETDRS ring that was placed anatomically correctly by the graders. Furthermore, some patients developed a new preferred retinal locus for fixation during the course of the study [41], and sometimes the follow-up function and eye tracking option could not be used, but rather a new “baseline” visit image had to be taken during the follow-up period with a possible impact on grading results, especially in the determination of the OR, which is especially prone to this inconsistency, leading us to exclude a large number of eyes/visits from this analysis.

Nevertheless, SD-OCT offers several advantages to the provision of possible surrogate outcome measures in upcoming clinical trials. First, it is a noninvasive procedure that is comfortable for patients; it requires a shorter time to preform than other imaging techniques such as Microperimetry, and has no or only minimal risk of potential light-toxicity (in contrast to FAF) [9]. The imaging methodology can be standardized across sites. SD-OCT provides the possibility of direct imaging of photoreceptors or parts thereof, and may be therefore acceptable as a surrogate endpoint outcome measure of efficacy by regulatory authorities [14]. Segmentation of individual layers showed a statistically significant decline in outer retinal layer thickness and intact area, in a time period as short as 6 months. The measurement of the intact area provides robust data that can serve as a clinical endpoint measure. In contrast, the IR showed an increase in thickness, at least in the IR; this in in agreement with previous observations that inner retinal laminar abnormalities in ABCA4-STGD1 are likely due to the retinal remodeling that accompanies photoreceptor loss [42]. However, this may also influence the measurement of total retinal thickness which may therefore not an ideal candidate as a potential outcome measure.

Additional considerations need to be taken when choosing SD-OCT photoreceptor integrity as an outcome measure. It appears that SD-OCT might be especially suitable for early stage STGD1 in which patients have a stable fixation; and in whom only the central macula is affected and can be easier be tracked. In these patients, the disease alterations appear to impact the EZ more than the RPE. Moreover, these patients may be better candidates for pharmaco- or gene therapy to rescue residual photoreceptors and/or RPE cells and/or slow down degeneration [1]. However, additional studies are needed to explore the correlation between structural changes in SD-OCT and retinal function as determined by microperimetry or scotopic microperiemtry [43, 44], especially in order to evaluate the impact of these structural changes on visual function; this may further confirm SD-OCT derived variables as potential surrogate endpoints.

The study was conducted according to the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) good clinical practice (GCP) Guidelines, the applicable regulatory requirements, and the current Declaration of Helsinki and was in compliance with the Health Insurance Portability and Accountability Act. Ethics Committee approval was granted by the Western Institutional Review Board (WIRB PRO NUM: 20130051), the Local Institutional Review Boards, and the Human Research Protection Office of the United States Army Medical Research and Materiel Command before enrollment of the first patient. This full list of participating site and ethics committees can be found at [6]. The studies were registered at www.clinicaltrials.gov (identifier, NCT01977846). All patients gave written informed consent before enrollment. If patients were of minor age, written informed consent from parents/legal guardians was obtained for all participants aged under 18 years.

The author(s) have made the following disclosure(s):

D.G.B. is a consultantof AGTC (Alachua, FL); Nacuity (Ft. Worth, TX); ONL Therapeutics (Ann Arbor, MI); and Bluerock Therapeutics (Boston, MA). Dr. Birch is supported by the Foundation Fighting Blindness. S.S. is consultant for 4DMT, Abbvie, Alexion, Allergan Inc., Alnylam Pharmaceuticals, Amgen Inc., Apellis Pharmaceuticals, Inc., Astellas, Bayer Healthcare Pharmaceuticals, Biogen MA Inc., Boehringer Ingelheim, Carl Zeiss Meditec, Catalyst Pharmaceuticals Inc., iCare Inc., GENENTECH, Gyroscope Therapeutics, Heidelberg Engineering, Hoffman La Roche, Ltd., Iveric Bio, Janssen Pharmaceuticals Inc., Nanoscope, Notal Vision Inc., Novartis Pharma AG, Optos Inc., Oxurion/Thrombogenics, Oyster Point Pharma, Regeneron Pharmaceuticals Inc., Samsung Bioepis, Topcon Medical Systems Inc.

Recipient of honoraria for symposia from Carl Zeiss Meditec, Heidelberg Engineering, Nidek Incorporated, Novartis Pharma AG, Topcon Medical Systems Inc., Research Instruments from Carl Zeiss Meditec, Heidelberg Engineering, Optos Inc., Nidek, Topcon, iCare. S.W. is a member of Scientific Technical Advisory Committee Alcon Research Institute (Fort Worth, TX); Research to Prevent Blindness, Inc, New York, New York. I.A. is a consultant for Janssen and Novartis Therapeutics. H.P.N.S. is supported by the Swiss National Science Foundation (Project funding: “Developing novel outcomes for clinical trials in Stargardt disease using structure/function relationship and deep learning” #310030_201165, and National Center of Competence in Research Molecular Systems Engineering: “NCCR MSE: Molecular Systems Engineering (phase II)” #51NF40-182895, the Wellcome Trust (PINNACLE study), and the Foundation Fighting Blindness (ProgStar study).

Dr. Scholl is member of the Scientific Advisory Board of: Boehringer Ingelheim Pharma GmbH & Co; Droia NV; Eluminex Biosciences; Gyroscope Therapeutics Ltd.; Janssen Research and Development, LLC (Johnson & Johnson); Okuvision GmbH; ReVision Therapeutics Inc.; and Saliogen Therapeutics Inc. Dr. Scholl is a consultant of Alnylam Pharmaceuticals Inc.; Gerson Lehrman Group Inc.; Guidepoint Global, LLC; and Tenpoint Therapeutics. Dr. Scholl is member of the Data Monitoring and Safety Board/Committee of Belite Bio (DRAGON trial, NCT05244304; LBS-008-CT02, NCT05266014), F. Hoffmann-La Roche Ltd (VELODROME trial, NCT04657289; DIAGRID trial, NCT05126966; HUTONG trial), ViGeneron (protocol No. VG901-2021-A) and member of the Steering Committee of Novo Nordisk (FOCUS trial; NCT03811561). M.I. is a consultant of Allergan, Thrombogenics, Omeros, Genentech, Quark, Alimera, and Boehringer Ingelheim. All other authors have no conflicts of interest to declare.

The ProgStar studies are supported by the Foundation Fighting Blindness (FFB) and a grant to FFB by the U.S. Department of Defense USAMRMC TATRC, Fort Meade, Maryland (Grant Nos. W81-XWH-07-1-0720 and W81XWH-09-2-0189);

Dr. Hendrik Scholl is supported by the Foundation Fighting Blindness; Shulsky Foundation, New York, NY; National Centre of Competence in Research (NCCR) Molecular Systems Engineering (University of Basel and ETH Zürich), Swiss National Science Foundation.

Rupert W. Strauss is supported by the Austrian Science Fund (FWF; Project No. J 3383-B23) and the Foundation Fighting Blindness.

R.W.S. is supported by Austrian Science Fund (Vienna, Austria; project No. J 3383-B23) and Foundation Fighting Blindness. A.V.C. is supported in part – National Institutes of Health, Bethesda, Maryland (Grant No. EY013203). M.M. is supported by the National Institute for Health Research Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, London, United Kingdom.

All authors fulfilled the ICMJE criteria for authorship: conception or design of work: R.W.S., A.H., D.G.B., A.V.C., M.M., E.I.T., E.Z., S.S., S.W., and H.P.N.S. Acquisition, analysis, and/or interpretation of data: L.L., A.J., M.I., P.S.B., D.G.B., A.V.C., M.M., I.A., J.S.S., E.I.T., E.Z., S.S., L.J.-K., S.W., X.K., and H.P.N.S. Drafting the work and/or reviewing the work: R.W.S. Final approval of the version to be published, reviewing the work, and Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: R.W.S., L.L., A.H., A.J., M.I., P.S.B., D.G.B., A.V.C., J.S.S., E.I.T., E.Z., S.S., L.J.-K., S.W., and H.P.N.S.

Xiangrong Kong and Hendrik P.N. Scholl share senior authorship.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author H.P.N.S. upon reasonable request.