Clinical Phenotype and Course of PDE6A-Associated Retinitis Pigmentosa Disease, Characterized in Preparation for a Gene Supplementation Trial

Key Points

Question  What are the clinical features and course of retinitis pigmentosa associated with biallelic sequence variations in the PDE6A gene?

Findings  In this longitudinal cohort study of 57 adults, 17 of the PDE6A variants appeared to be novel. Disease was highly symmetrical between right and left eyes, and visual impairment was mild or moderate in 90% of patients.

Meaning  These data suggest that PDE6A–retinitis pigmentosa may be amenable to gene therapy.

Importance  Treatment trials require sound knowledge on the natural course of disease.

Objective  To assess clinical features, genetic findings, and genotype-phenotype correlations in patients with retinitis pigmentosa (RP) associated with biallelic sequence variations in the PDE6A gene in preparation for a gene supplementation trial.

Design, Setting, and Participants  This prospective, longitudinal, observational cohort study was conducted from January 2001 to December 2019 in a single center (Centre for Ophthalmology of the University of Tübingen, Germany) with patients recruited multinationally from 12 collaborating European tertiary referral centers. Patients with retinitis pigmentosa, sequence variants in PDE6A, and the ability to provide informed consent were included.

Exposures  Comprehensive ophthalmological examinations; validation of compound heterozygosity and biallelism by familial segregation analysis, allelic cloning, or assessment of next-generation sequencing–read data, where possible.

Main Outcomes and Measures  Genetic findings and clinical features describing the entire cohort and comparing patients harboring the 2 most common disease-causing variants in a homozygous state (c.304C>A;p.(R102S) and c.998 + 1G>A;p.?).

Results  Fifty-seven patients (32 female patients [56%]; mean [SD], 40 [14] years) from 44 families were included. All patients completed the study. Thirty patients were homozygous for disease-causing alleles. Twenty-seven patients were heterozygous for 2 different PDE6A variants each. The most frequently observed alleles were c.304C>A;p.(R102S), c.998 + 1G>A;p.?, and c.2053G>A;p.(V685M). The mean (SD) best-corrected visual acuity was 0.43 (0.48) logMAR (Snellen equivalent, 20/50). The median visual field area with object III4e was 660 square degrees (5th and 95th percentiles, 76 and 11 019 square degrees; 25th and 75th percentiles, 255 and 3923 square degrees). Dark-adapted and light-adapted full-field electroretinography showed no responses in 88 of 108 eyes (81.5%). Sixty-nine of 108 eyes (62.9%) showed additional findings on optical coherence tomography imaging (eg, cystoid macular edema or macular atrophy). The variant c.998 + 1G>A;p.? led to a more severe phenotype when compared with the variant c.304C>A;p.(R102S).

Conclusions and Relevance  Seventeen of the PDE6A variants found in these patients appeared to be novel. Regarding the clinical findings, disease was highly symmetrical between the right and left eyes and visual impairment was mild or moderate in 90% of patients, providing a window of opportunity for gene therapy.

Retinitis pigmentosa (RP) is a hereditary, degenerative retinal disease that causes severe visual impairment and visual acuity loss because of progressive degeneration of primarily the rod and secondarily the cone photoreceptors. The disease manifests with early-onset nyctalopia, followed by daytime visual field defects progressing from the midperiphery to the periphery and the center. Best-corrected visual acuity typically remains relatively well preserved until macular involvement by macular edema and/or photoreceptor atrophy causes central visual acuity loss.

To date, sequence variations in 89 genes are known to be associated with RP.¹ In 1995, the gene encoding for the α subunit of the rod photoreceptor cyclic guanosine monophosphate (cGMP) phosphodiesterase (PDE6A) was identified as the seventh RP locus, on chromosome 5q31.2-q34 (OMIM, 180071 and 613810).^2,3 It includes 22 coding exons and encodes 859 amino acid residues. Rod photoreceptor cGMP phosphodiesterase is made of 4 subunits, the catalytic α subunit (PDE6A), the catalytic β subunit (PDE6B), and 2 inhibitory γ subunits (PDE6G). The enzyme functions to hydrolize the intracellular cytoplasmic cGMP level, which causes closure of cyclic nucleotide–gated channels, an essential step in vertebrate phototransduction.⁴ The PDE6A gene appears to account for disease in less than 4% of families with autosomal recessive RP in North America, approximately 2% of the cases in cohorts of French and Pakistani patients, and approximately 1% of families with inherited retinal diseases in Israel.^5-8

Sequence variations in PDE6A do not cause retinal dystrophy in only humans. Biallelic sequence variations in the PDE6A gene have been identified to cause autosomal recessive progressive retinal atrophy in the Cardigan Welsh Corgi dog.^9,10 Additionally, induced and natural mouse models have been studied for variations of the disease phenotype (eg, pace of photoreceptor degeneration) associated with different PDE6A sequence variations.^10,11 In preclinical gene supplementation trials, mice showed more effective photoreceptor cell rescue when the rod-specific transgene AAV2/8(Y733F)-Rho-Pde6α was delivered before the onset of disease. It was thus concluded that the success of therapeutic clinical trials will depend on identifying patients as early as possible to maximize the number of rods still viable and treatable with gene therapy.^12-14

In 2012, the RD-CURE Consortium started working on a collaborative project for the clinical translation of gene therapy for patients with inherited retinal dystrophies; PDE6A-associated RP is one of the diseases for which a therapeutic transgene is currently evaluated in a human gene therapy trial, using recombinant adeno-associated viral vectors. The aim of the present study was to assess clinical features and genetic findings of patients with RP associated with biallelic sequence variations in the PDE6A gene in preparation for this gene supplementation trial.

The study (ClinicalTrials.gov Identifier NCT02759952) was conducted in accordance with the Declaration of Helsinki, with approval from the ethics committee of the University of Tübingen. Written informed consent was obtained from all participants. Patients were recruited from the clinics for hereditary retinal degenerations at the Centre of Ophthalmology of the University of Tübingen and 12 collaborating European tertiary referral centers. Patient travel and accommodation costs were compensated.

All patients were examined at the Centre for Ophthalmology of the University of Tübingen, Germany, between January 2001 and December 2019. A comprehensive ophthalmological examination was performed, including best-corrected visual acuity (BCVA) with Early Treatment Diabetic Retinopathy Study charts, a semiautomated 90° kinetic visual field (VF) exam with objects lll4e and I4e (Octopus 900 [Haag-Streit]), full-field electroretinography (ff-ERG) and multifocal electroretinography testing according to International Society for Clinical Electrophysiology of Vision standards (Espion [Diagnosys]), spectral-domain optical coherence tomography imaging (Spectralis HRA+OCT [Heidelberg Engineering]), and slitlamp and dilated fundus examinations and photography. Results of BCVA testing were converted to logarithm of the minimum angle of resolution (logMAR) visual acuity values.¹⁵ Visual field parameters were assessed as total VF area in square degrees for both objects III4e and I4e, using the built-in software of the testing device.

The patients enrolled in this study had a confirmed genetic diagnosis of PDE6A-associated retinitis pigmentosa. If possible, validation of homozygosity, compound heterozygosity, and biallelism was performed by familial segregation analysis, allelic cloning, or assessment of next-generation sequencing–read data. Whole-exome sequencing to exclude the presence of pathogenic variants in other genes associated with inherited retinal degeneration was performed for 23 cases. Standards and guidelines provided by the American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology were applied to classify the identified variants.¹⁶ The potential pathogenicity of missense changes was assessed using 5 online prediction software tools, namely SIFT, PolyPhen-2, Mutation Taster, Mutation Assessor, and Provean.^17-21 Assessment of variants potentially affecting splicing was performed with the Alamut Genova software version 1.4 (Sophia Genetics) using default parameters. Variant designation is based on the National Center for Biotechnology Information reference sequence for PDE6A (NM_000440.3; GRCh38) involving 22 coding exons.

Statistical analyses were performed using SPSS Statistics for Windows, version 26.0 (IBM). Normal distribution was tested using the Shapiro-Wilk test. Differences between genetic subgroups were tested using regression analysis.

Fifty-seven patients from across Europe were included. Of the 57 patients (114 eyes) in this analysis, 25 (44%) were male and 32 (56%) were female. The mean (SD) age at baseline was 40 (14) years (range, 12-78 years). The mean (SD) follow-up time was 2.9 (2.1) years. Forty of the 57 patients (70%) were followed up as part of the study. The mean (SD) follow-up time was 2.9 (2.1) years.

All 57 affected individuals harbored rare and potentially disease-causing variants compatible with autosomal recessive inheritance in the PDE6A gene (Table 1). Thirty patients were homozygous for disease-causing alleles. Validation of true homozygosity by parental segregation analysis was possible in 9 cases. Twenty-seven patients were heterozygous for 2 different PDE6A variants each. In these, compound heterozygosity of variants could be validated in 19 cases by segregation analysis. In addition, in 2 cases, transconfiguration of variants was established by assessing independent next-generation sequencing–read data or allelic cloning, respectively.

The sequence variation spectrum included 33 different alleles, 17 of which have not been reported to date, to our knowledge. They included 12 missense variants, 8 canonical splice-site variants, 7 nonsense variants, 5 frameshift variants, and 1 in-frame deletion or insertion variants. Recurrent alleles were frequently observed: 16 patients carried the c.304C>A;p.(R102S) variant on 1 or both alleles, making it the most common allele in the cohort (23 alleles), followed by c.2053G>A;p./(V685M) (12 alleles). The splice-site variant c.998 + 1G>A;p.? accounted for 14 alleles in the cohort. However, this high number is associated with 6 siblings from the same family having homozygosity for this variant (Figure 1). All variants were classified according to their pathogenicity based on the ACMG guidelines (Table 2).¹⁶

The mean (SD) BCVA was 0.43 (0.48) logMAR (range, −0.10 to 2.30; 114 eyes; Snellen equivalent, 20/50; range, hand movement to 20/16). Visual impairment was mild in 15 of 114 eyes (13.2%), moderate in 19 of 114 eyes (16.7%), and severe in 7 of 114 eyes (6.1%), and 4 of 114 eyes (3.5%) met the criteria for legal blindness as defined by the World Health Organization with respect to BCVA. The BCVA findings were highly symmetrical in right and left eyes (R², 0.786).

Kinetic visual field testing with object III4e of Goldmann allowed for evaluation in 107 of 114 eyes (93.9%). The median visual field area in these 107 eyes was 660 (5th-95th percentiles, 76-11 019; 25th-75th percentiles, 255-3923) square degrees (Figure 2). In 3 eyes, object III4e was not recognized. Kinetic visual field testing obtained with object I4e allowed for evaluation in 92 of 114 eyes (80.7%). The median visual field area in these 92 eyes was 150 (range, 2-2833) square degrees (Figure 2). In 16 eyes, object I4e was not recognized. Kinetic visual field measurements were highly symmetrical in right and left eyes (R², 0.951). Progression of visual field defects with increasing age or disease duration is shown in Figure 2.

Optical coherence tomography (OCT) images were available in 108 of 114 eyes (94.7%). As is typical for RP, OCT imaging revealed thinning of the outer retinal layers from the periphery to the center of the retina (macula or fovea), with disruption or loss of the ellipsoid zone in advanced disease. In the cohort, 69 of 108 eyes (63.9%) showed additional findings on OCT: cystoid macular edema (CME) in 27 of 108 eyes (25.0%), CME and macular atrophy in 2 of 108 (1.9%), macular atrophy in 18 of 108 (16.7%), epiretinal membrane (ERM) in 8 of 108 (7.4%), ERM with traction in 6 of 108 (5.6%), ERM with a lamellar hole in 2 of 108 (1.9%), an isolated lamellar hole in 1 of 108 (1%), a full-thickness macular hole in 1 of 108 (0.9%), and posterior staphyloma in 4 of 108 (3.7%) (eFigure 1 in the Supplement).

Full-field ERG findings were available in 108 of 114 eyes (94.7%). In 4 of 108 eyes (3.7%), dark-adapted and light-adapted ff-ERG responses were within normal limits. In 8 of 108 (7.4%) each, light-adapted ff-ERG showed some measurable, albeit subnormal responses, whereas dark-adapted ff-ERG showed either residual or no responses. In 88 of 108 (81.5%), responses for both dark-adapted and light-adapted ff-ERGs were absent. Multifocal ERG findings were available in 90 of 114 eyes (78.9%). In 53 of 90 eyes (88%), multifocal electroretinography showed no responses.

We also evaluated whether the clinical findings differed between patients harboring 1 of the common disease-causing PDE6A variants in a homozygous state, comparing the c.304C>A;p.(R102S) (group 1), and c.998 + 1G>A;p.? (group 2). Seven patients each were homozygous for either the variant c.304C>A;p.R(102S) or the variant c.998 + 1G>A;p.?. Visual field maps, OCT, and fundus autofluorescence findings of right eyes are shown in eFigure 1 in the Supplement. The respective groups (1 and 2) differed in their mean (SD) ages at baseline (46 [17] vs 38 [6] years). The mean (SD) visual acuity was worse in group 2 when compared with group 1 (0.50 [0.37] vs 0.19 [0.27] logMAR; Snellen equivalent: 20/60 vs 20/30). Similarly, the mean (SD) VF area was worse in group 2 when compared with group 1 (1178 [1304] vs 5548 [5040] square degrees). When corrected for age, these differences were statistically significant. The BCVA mean difference was −0.31 (95% CI, −0.70 to 0.07; P = .01), and the VF area mean difference was 4370 (95% CI, −644 to 9384; P = .003).

In this study, we examined 57 patients from across Europe with RP associated with pathogenic, biallelic variants in the PDE6A gene. Given that PDE6A-associated RP is such a rare genetic subtype of RP, the size of this cohort is uniquely large.⁶ From comprehensive genetic analyses in association with this study of PDE6A in patients with a clinical diagnosis of RP, we conclude that there is a frequency of PDE6A-associated RP of 1.6% in Germany (1100 patients residing in Germany, of whom 17 were found to harbor biallelic variants in PDE6A [data not shown]).

All patients with PDE6A-associated RP in our study exhibited highly symmetrical findings typical for RP with respect to visual acuity, visual field, and other modalities. When comparing BCVA with VF area in terms of symmetry, the VF area was more symmetrical in right and left eyes. Disease symmetry presents a strong case for using the untreated eye as the control eye in an interventional study with small patient numbers. The untreated eye may serve as a control to assess possible negative adverse effects as well as positive outcomes of therapy (such as gene supplementation therapy; ie, the preservation of VF area or the deceleration of VF area loss).

The greatest (numberwise) loss in VF area was approximately −7900 square degrees in a 23-year-old patient with a VF area of 8900 square degrees at baseline 2 years earlier. This change reflects the natural course of disease, characterized by a slow progression of VF defects before young adulthood, typically followed by a sudden loss of large or larger VF areas in young adulthood and subsequent slow progression of VF defects the smaller the remaining VF area gets.

Kinetic visual field testing with object III4e allowed for evaluation in more than 90% of eyes. However, kinetic visual field testing is a set of psychophysical examinations with advantages but also limitations. The limitations are that the condition of the patient affects the results of the examination, in that a patient who is alert will achieve better results than a patient with fatigue, and the same is true for the technician performing the examination and encouraging the patient. Thus, VF findings can vary between visits, and comparisons between patients, as well as within the same patient at different points, may yield misleading results. However, given the remarkable symmetry of the VF area in right and left eyes that we observed in this cohort and the fact that VF area is a measure of visual function rather than an anatomical end point, we judge VF testing as a suitable marker and end point for a phase I or II safety and efficacy trial.

Macular OCT findings may well limit a patient’s eligibility to participate in a gene therapy trial, since the preferred injection site for the therapeutic vector to date is the central retina. Generally, macular findings in patients with RP include cystoid macular edema in 6% to 25%, ERM in 1% to 27%, and macular holes in 1% to 5% of patients with RP.^30-32 In our cohort, CME was present in 25%, ERM in 7%, and macular holes in 4%. Thus, the frequency of macular OCT findings in PDE6A-associated RP was similar to that in the general population of patients with RP. Mild CME may not be as much of a limitation for a patient to participate in a gene therapy trial, and the same applies to mild ERMs without traction. Yet, extensive CME, as well as marked ERM with traction and any type of macular hole, would be expected to increase the risk of complications attributable to the surgical procedure, irrespective of the investigational new drug, thus having a potential effect on safety and efficacy observations.

Among the most frequent variants found in our cohort, the variant c.998 + 1G>A;p.? led to a more severe phenotype with respect to VF and BCVA when present in a homozygous state and when compared with individuals homozygous for the c.304C>A;p.(R102S) variant. The third most frequent variant c.2053G>A;p.(V685M) was mostly present in a (compound) heterozygous state, making it difficult to compare this genotype with the 2 other most frequent variants mentioned with respect to the severity of the phenotype. Our clinical findings are thus in line with the fact that c.998 + 1G>A;p.? is considered a null allele, in contrast with c.304C>A;p.(R102S).

Because we were unable to validate biallelism in all of the patients, the pathogenicity of single variants rather than genotypes will be discussed. Of the 33 putatively pathogenic variants identified in this study, 7 are nonsense variants and 5 are frameshift variants. These are considered to constitute loss-of-function alleles. Furthermore, we identified 8 canonical splice variants, 7 of which affect invariable donor sites and 1 of which affects an invariable acceptor site.

Because nonsense, canonical splice site, and frameshift variants have a strong weight in the ACMG scoring system, these classes of variants are consequently classified either as pathogenic or likely pathogenic (Table 2). On the other hand, missense variants that lack segregation data and functional analyses to support a damaging outcome are always classified as variants of uncertain significance per the ACMG guidelines. The potential pathogenicity of the missense variants affecting codon 102 and 389 of the PDE6A gene (c.304C>A;p.R102S, c.305G>A;p.R102H, and c.1166C>T;p.P389L) is supported by the fact that corresponding sequence variations in PDE6C have been shown to have a causal association with PDE6C-associated autosomal recessive achromatopsia.³³ Their effect on PDE6 function has further been substantiated by functional in vitro studies performed on the catalytic subunit of cone photoreceptor phosphodiesterase PDE6C, the PDE6 paralog in cone photoreceptors.³³ Grau and coworkers³³ demonstrated that the corresponding PDE6C missense variants c.310C>T;p.R104W and c.1172C>T;p.P391L led to a highly significant reduction in phosphodiesterase activity to almost baseline levels. Since codons 104 and 391 of PDE6C are homologous to codons 102 and 389 of PDE6A, the results obtained from functional studies performed on PDE6C can be transferred to PDE6A. Evidence for the pathogenicity of the missense variants c.1684C>T;p.R562W and c.2053G>A;p.V685M is derived from the functional analysis of Pde6a mouse models: Sothilingam and coworkers¹¹ showed that homozygous (R562W/R562W and V685M/V685M) and compound heterozygous (R562W/V685M) Pde6a mutant animals expressed only residual amounts of Pde6a protein, resulting in photoreceptor cell death. Functional studies or animal models are not available for the remaining 7 missense variants identified in this study. Yet, the respective amino acid residues are located either between cGMP-specific phosphodiesterases, adenylyl cyclases and FhlA (GAF) domains (c.743T>A;p.V248D), within the second GAF domain (c.959A>G;p.D320G, c.1235T>C;p.F412S), or in the catalytic domain of PDE6A (c.1610T>C;p.I537T, c.1689C>A;p.H563Q, c.1705C>A;p.Q569K, c.1862T>G;p.L621R). Specifically, the variants c.1684C>T;p.R562W and c.1689C>A;p.H563Q affect the Zn²⁺-binding motif, which is invariant among phosphodiesterases. All affected amino acid residues are highly conserved among vertebrates, not only for PDE6A, but also within PDE6B and PDE6C (eFigure 2 in the Supplement). Consequently, in silico assessment using 5 different prediction tools predicted damaging outcomes on protein function for all 7 variants (Table 2). However, without functional data supportive of damaging outcomes, their pathogenicity cannot be unequivocally proven.

Last, we have identified a novel in frame deletion/insertion variant, c.63_68del;p.(K21_Y23delinsN). It localizes near the N-terminus of PDE6A, upstream of the first GAF domain, and affects 3 amino acid residues that are conserved among vertebrate PDE6A and partially conserved among human PDE6B and PDE6C polypeptides. It has been shown that dimerization of PDE6 is mediated by multiple regions in the N-terminal domain.³⁴ We therefore hypothesize that the c.63_68del;p.(K21_Y23delinsN) variant may impair dimerization, thereby leading to loss of function of PDE6A.

True homozygosity and the absence of pathogenic variants in other genes that are known to cause inherited retinal dystrophies are key eligibility criteria for inclusion in a gene therapy trial. One limiting factor with respect to the clinical aspect (and course) of PDE6A-associated RP and subretinal gene therapy in our cohort may be macular pathology, which would increase the risk of complications attributable to the surgical procedure if the injection site is the central retina. However, given that PDE6 is rod specific, from a mechanistical standpoint, it would be more promising to treat regions outside the macula with high rod density early in life.

In summary, we observed and described the genetic and ophthalmologic characteristics in 57 patients with RP associated with pathogenic biallelic variants in the PDE6A gene. Regarding the genetic findings, 17 of the PDE6A variants found in these patients appeared to be novel. Regarding the clinical findings, disease was highly symmetrical between right and left eyes and visual impairment was mild or moderate in 90% of patients, providing a window of opportunity for gene therapy. These data suggest that individuals with PDE6A-associated RP may be eligible for gene therapy.

Accepted for Publication: August 6, 2020.

Corresponding Author: Laura Kuehlewein, MD, Institute for Ophthalmic Research, Centre for Ophthalmology, Eberhard Karls University, Elfriede-Aulhorn-Str. 7, 72076 Tübingen, Germany (laura.kuehlewein@med.uni-tuebingen.de).

Published Online: October 15, 2020. doi:10.1001/jamaophthalmol.2020.4206

Author Contributions: Drs Kuehlewein and Weisschuh had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Kuehlewein, Zobor, Fischer, Holz, Wilhelm, Wissinger, Kohl, Weisschuh.

Acquisition, analysis, or interpretation of data: Kuehlewein, Zobor, Andreasson, Ayuso, Banfi, Bocquet, Bernd, Biskup, Boon, Downes, Fischer, Kellner, Leroy, Meunier, Nasser, Rosenberg, Rudolph, Stingl, Thiadens, Wilhelm, Zrenner, Kohl, Weisschuh.

Drafting of the manuscript: Kuehlewein, Kohl, Weisschuh.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Kuehlewein.

Obtained funding: Kuehlewein, Zobor, Fischer, Wissinger, Zrenner, Kohl, Weisschuh.

Administrative, technical, or material support: Zobor, Andreasson, Banfi, Bocquet, Biskup, Downes, Kellner, Thiadens, Wilhelm, Zrenner, Kohl, Weisschuh, Nasser.

Supervision: Kuehlewein, Zobor, Banfi, Boon, Fischer, Holz, Leroy, Wilhelm, Zrenner, Kohl, Weisschuh.

Conflict of Interest Disclosures: Dr Downes reported grants from Clinical Research Network NIHR Thames Valley during the conduct of the study; involvement with a commercial study on age-related macular degeneration with Allergan and Novartis; personal fees from Allergan, Circadian Therapeutics, and Boehringer Ingelheim; grants from Retina UK RP Genome outside the submitted work; and service as the chair of the medical advisory board for Retina UK. Dr Fischer reported being a director of Fischer Consulting Limited. Dr Kellner reported study support from Medspace/GTScope, Samsung, Bayer Vital, and Novartis and personal fees from Roche outside the submitted work. Dr Leroy reported grants and personal fees from GenSight Therapeutic, ProQR Therapeutics, and Novartis Pharma; personal fees from IVERIC Bio, Spark Therapeutics, REGENXBio, Biogen, and Vedere Bio; and grants from MeiraGTx outside the submitted work. Dr Stingl reported grants from Kerstan Foundation during the conduct of the study and grants from Retina Implant and personal fees from ProQR and Novartis outside the submitted work. Dr Leroy reports being a senior clinical investigator of the Research Foundation–Flanders (Belgium). Drs Wilhelm, Wissinger, Zrenner, Kohl, Kuehlewein, Weisschuh, and Zobor reported grants from Tistou and Charlotte Kerstan Foundation during the conduct of the study. No other disclosures were reported.

Funding/Support: This work was supported by funds of the Tistou and Charlotte Kerstan Foundation. This study was performed as part of a collaboration within the European Reference Network for Rare Eye Diseases, which is cofunded by the Health Program of the European Union (framework partnership agreement 739543–‘ERN-EYE’).

Role of the Funder/Sponsor: The Foundation had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Members of the RD-CURE Consortium (in institutional and alphabetical order): Centre for Ophthalmology, Eberhard Karls University, Tübingen, Germany: Karl Ulrich Bartz-Schmidt, MD; Sylvia Bolz; M. Dominik Fischer, MD, PhD; Susanne Kohl, PhD; Laura Kuehlewein, MD; Regine Mühlfriedel, PhD; Jonas Neubauer, MD; Alex Ochakovski, MD; Francois Paquet-Durand, PhD; Mathias Seeliger, MD; Vithiyanjali Sothilingam, PhD; Katarina Stingl, MD; Marius Ueffing, PhD; Nicole Weisschuh, PhD; Bernd Wissinger, PhD; Fabian Wozar, MD; Ahmad Zhour, MD; Ditta Zobor, MD; Eberhart Zrenner, MD. Department of Pharmacy–Center for Drug Research, Ludwig-Maximilians-University Munich, Munich, Germany: Martin Biel, PhD; Stylianos Michalakis, PhD; Christian Schön, PhD. STZ Eyetrial at the Centre for Ophthalmology, University of Tübingen, Tübingen, Germany: Nadine Kahle, PhD; Tobias Peters, MD; Barbara Wilhelm, MD. Department of Ophthalmology, Columbia University, New York, New York: Stephen H. Tsang, MD, PhD. Deutsches Zentrum für Neurodegenerative Erkrankungen, Tübingen, Germany: Christian Johannes Glöckner, PhD.

Additional Contributions: Almudena Avila, PhD, and Fiona Blanco-Kelly, MD, PhD, Hospital Universitario Fundación Jiménez Díaz, provided contacts for Spanish patients. They were not compensated for their contributions.