Vasily Smirnov

Background: Rare eye diseases are characterized by low prevalence, clinical heterogeneity, and fragmented data collection, which limit the reliability of analysis results and multicenter research. In France, the development of health data warehouses is strictly regulated by national data protection authorities. While international initiatives aim to harmonize rare disease registries, capturing hyperspecialized, multimodal clinical records and ophthalmology-specific imaging within a fully compliant and sustainable infrastructure remains a major operational challenge. Objective: This study aims to describe the methodological framework, regulatory implementation, and early operational outcomes of FREDD (French Rare Eye Disease Database), a national interoperable health data warehouse dedicated to rare eye diseases, and to analyze its key success factors, technical bottlenecks, and long-term financial sustainability models. Methods: Developed under the oversight of the French National Institute of Health and Medical Research (Institut national de la santé et de la recherche médicale; INSERM), the warehouse implements a centralized 3-layer architecture encompassing data collection, processing, and research reuse environments. The dataset extends the national minimum rare disease dataset with detailed ophthalmology variables structured through a dynamic electronic case report form, as well as a dedicated collection of ophthalmic images. Interoperability is achieved using international domain ontologies and a custom parsing tool (FREDDEX) designed to automatically prefill clinical data from the existing French national rare disease registry. Centralized monitoring, data curation, and cross-center quality tracking are driven by a dedicated in-house dashboard (FREDDIn [FREDD Insights]), while unstructured retinal images are processed through a standardized pseudonymization and human-verified validation pipeline. Implementation (Results): Official regulatory authorization was obtained in April 2024, and active data collection began in March 2025 across 5 pilot expert centers, focusing its initial phase on inherited retinal dystrophies, primarily retinitis pigmentosa. Over a 1-year period, the warehouse successfully integrated clinical data from 1649 patients—representing 20.3% of all retinitis pigmentosa cases registered nationally—and accumulated more than 10,000 ophthalmic images. The centralized dashboard maintained an overall data inconsistency rate of approximately 7%, chiefly reflecting logical dependencies. Among the included patients, 36.7% (605/1649) patients had associated imaging data, with fundus photography being the most widely available modality (768/1649, 46.6%). Conclusions: The implementation of this database demonstrates that a highly specialized, multimodal health data warehouse can be successfully deployed in a highly regulated environment when regulatory anticipation, governance formalization, and technical flexibility are addressed in parallel. While initial development was supported by institutional grants, long-term sustainability will require a structured cost-recovery model. Future milestones will focus on expanding center coverage nationwide, refining image annotation frameworks for AI readiness, and deploying automated export pathways toward European registries.

BACKGROUND Rare eye diseases are characterized by low prevalence, clinical heterogeneity, and fragmented data collection, which limit statistical power and multicenter research. In France, the development of health data warehouses (HDWs) is regulated under the French Health Data Warehouse framework, which imposes strict requirements for governance, security, and interoperability. While national and European initiatives aim to harmonize rare disease data, disease-specific operational implementation remains challenging. OBJECTIVE To describe the design, regulatory implementation, and early operational outcomes of FREDD (French Rare Eye Diseases Database), a national HDW dedicated to rare eye diseases, and to analyze key success factors, bottlenecks, and sustainability considerations associated with its deployment. METHODS FREDD was developed as a centralized HDW compliant with French regulations. Its dataset was structured around the minimum rare disease dataset, ensuring compatibility with the French rare disease databases, and aligned with the European datasets to maximize semantic interoperability. The infrastructure implements a three-layer architecture (data collection, processing, and research reuse), structured electronic case report forms with embedded validation rules, dedicated imaging harmonization pipelines, and centralized data quality monitoring tools. Governance was formalized through a steering committee and a scientific and ethical committee, with strictly controlled data access procedures. RESULTS Regulatory authorization was obtained in April 2024, and the full infrastructure became compliant with CNIL Health Data Warehouse requirements in November 2025. Data collection began earlier, in March 2025, enabling patient inclusion across five SENSGENE pilot centers. Within the first year, over 1600 patients and 10000 imaging datasets were integrated, demonstrating rapid adoption and scalability. Data collection combined retrospective imports and prospective entry, supported by FREDDEX, which enabled pre-filling of ~30% of clinical data from BaMaRa. High data quality was ensured through automated controls and centralized monitoring, with an overall inconsistency rate of ~7%, mainly reflecting logical dependencies. Approximately 40% of patients had associated imaging data, processed through a standardized and secure pipeline. CONCLUSIONS The FREDD initiative illustrates that a disease-specific HDW can be successfully implemented in a highly regulated environment when regulatory compliance, governance, and technical architecture are addressed in parallel. Establishing FREDD required substantial institutional investment, yet early operational outcomes indicate strong potential for scalability. Moving forward, sustainability will depend on actively leveraging the database through multicenter research, high impact studies, and reinforced interoperability with European infrastructures to unlock the full scientific and clinical value of rare eye disease data over the long term.

Identification et caractérisation des défauts géniques responsables des dystrophies rétiniennes héréditaires: Les dystrophies rétiniennes héréditaires (DRH) sont un groupe cliniquement et génétiquement hétérogène de maladies qui détériorent la vision. Environ 4.5 millions de personnes sont atteintes dans le monde entier. Le but de cette thèse était de déterminer les défauts géniques (connus et nouveaux) chez les patients atteints de DRH, suivis dans le Centre Hospitalier National d’Ophtalmologie des Quinze-Vingts en utilisant les technologies de séquençage de nouvelle génération (NGS), de séquençage d’exome (WES) et de séquençage de génome (WGS). Les études fonctionnelles ont été entreprises pour déchiffrer les mécanismes pathologiques associés aux variations génétiques. Les corrélations phénotype-génotype ont été réalisées. Une cohorte de 768 cas-index a été analysée sur le panel NGS comportant 351 gènes (connus et candidats) associés aux DRH. La Dystrophie rétinienne bâtonnet-cône (DBC) isolée était un phénotype prédominant dans la cohorte (n=670, 87%). Nous avons pu trouver les variants causaux chez 75% de patients. Les gènes majeurs impliqués dans DRH étaient USH2A, RP1 et EYS qui comptaient pour 12, 6.7 et 6.5% de cas respectivement. Par la technique de WES nous avons identifié les variants pathogènes dans les gènes TULP1, HGSNAT et IDUA. En utilisant la technique de WGS, nous avons identifié une nouvelle duplication qui ségrégait avec le phénotype dans une famille atteinte de Dystrophie Maculaire de Caroline du Nord. Nous avons rapporté les corrélations génotype-phénotype : chez les patients porteurs de variants connus et nouveaux dans le gène TTLL5 ; dans le groupe de 15 patients atteints de DBC non-syndromique, porteurs de variants bialléliques dans le gène CLN3 (gène surtout connu pour une maladie neuro-dégénérative syndromique) ; enfin, nous avons étudié un phénotype rétinien atypique, sensible au traitement par rétinol chez un patient porteur d’un nouveau variant de RBP4. De même, nous avons rapporté le phénotype rétinien de 4 patients atteints du syndrome de Usher de type 3, lié aux variants pathogènes dans le gène CLRN1. Enfin, nous avons décrit un nouveau phénotype complexe d’héméralopie essentielle associée aux défauts dans le gène VSX2 dans deux familles non reliées. Identification de nouveaux défauts géniques et les analyses fonctionnelles subséquentes permettent de mieux comprendre la physiologie de la rétine et constituent une base pour des développements thérapeutiques ultérieurs.

Introduction: We report a case series of pediatric patients with incidentally discovered atypical maculopathy, namely, benign foveal depigmentation of the retinal pigment epithelium (BFD). Longitudinal data, both functional and morphological, were analyzed. Materials and methods: Functional tests (perimetry, full-field, multifocal and pattern ERG) were performed using the Mon PackOne® unit (Métrovision, Pérenchies, France), and multimodal retinal imaging was carried out using the Heidelberg Retinal Tomograph 2 and Heidelberg Spectralis SD-OCT (Heidelberg Inc., Germany). Results: Four patients were examined, two of whom were followed longitudinally. Pattern ERG, multifocal ERG, and central static perimetry were all within normal limits. The reflective structure of the retinal layers in the foveolar zone appeared normal on SD-OCT, except for a cone of choroidal hyperreflectivity corresponding to the lesion site. The lesion was more evident on near-infrared autofluorescence imaging, presenting as a well-defined, round, hypoautofluorescent foveal area. No progression was observed in the two patients for whom follow-up data were available. Conclusion: BFD is a rare and sporadic condition that may affect both adults and children. It has no functional visual impact. However, long-term follow-up is advisable to confirm the benign and non-progressive nature of these depigmented macular lesions.

Importance: Congenital stationary night blindness (CSNB) is a clinically and genetically heterogeneous inherited retinal disorder (IRD), and in many complete CSNB (cCSNB) cases, the underlying genetic cause remains unknown. Uncovering the genetic defects of IRDs helps to refine diagnostic methods and supports the development of specific therapeutic approaches. Objective: To describe the phenotype and the underlying gene defect in patients with cCSNB from 2 unrelated families. Design, Setting and Participants: This retrospective case series was conducted from January 2023 to July 2025. Data for 3 patients from cohorts of genetically unsolved IRD cases in France (n = 140 for CSNB) and the Netherlands (n = 2730 for IRD) were analyzed clinically and genetically. Exposures: Complete ocular examination, including multimodal retinal imaging and full-field electroretinography (ffERG) incorporating the International Society for Clinical Electrophysiology of Vision standards and multimodal retinal imaging, were performed. Gene defects were identified by genome sequencing (GS) and exome sequencing (ES). Main Outcomes and Measures: The main outcome was a gene defect, EGFLAM , underlying cCSNB. Measures included phenotyping, GS, ES, Sanger sequencing, and cosegregation analysis. Results: The series included 3 patients from 2 unrelated families of Moroccan ancestry showing high myopia, reduced visual acuity, and night blindness. Retinal imaging depicted myopic changes. ffERG revealed electronegative Schubert-Bornschein configuration in keeping with cCSNB with ON-bipolar cell dysfunction. Patients were lacking pathogenic variants in known genes implicated in IRDs, including CSNB. Two different homozygous pathogenic variants, c.1563_1566del, p.(Val522Glufs*18) and c.1795C>T, p.(Arg599*) in EGFLAM were identified by ES and GS. The corresponding protein is localized in the outer plexiform layer and important for ON-bipolar cell signaling in the retina. Conclusion and Relevance: This case series reports on a gene defect in EGFLAM implicated in human cCSNB. Clinicians should be aware about this association and consider including EGFLAM in diagnostic gene panels for IRDs. This discovery may lead to faster and more accurate diagnosis of cCSNB and genetic counseling, as well as a pathway for developing therapies.

Introduction: Retinal dysfunction associated with CACNA2D4 gene defects is a rare disorder of photoreceptor to bipolar cell signaling. We report two affected siblings presenting a surprising disparity of retinal involvement. Materials and methods: Patients underwent complete ocular examination, multimodal fundus imaging, and full-field electroretinography (ffERG). Genetic testing was performed by a targeted Next Generation Sequencing panel. Results: Two siblings presented a reduced visual acuity and light sensitivity. ffERG was specific for CACNA2D4 -related retinal dysfunction, but amplitudes of responses were different in the two patients. Additionally, an x-wave to a dim red flash was well preserved, and there was a reduced b/a ratio to a high intensity (30cd/m2) dark-adapted stimulus. Both patients were homozygous for the variant c.2406C > A, p.(Tyr802*) in CACNA2D4. During 17 years of follow-up, vision remained stable in patient 1, with no evidence of retinal degeneration. Conclusion: Intrafamily clinical and electrophysiological expression of CACNA2D4- associated retinal dysfunction can be variable.

Recessive RPE65-related retinopathy is an inherited retinal disease (IRD) that is a well-established target for gene therapy. Dominant RPE65-related retinopathy, however, due to Irish founder variant p.(D477G), is extremely rare. Here, we report the discovery, replication and characterization of a novel dominant retinopathy caused by RPE65 variant p.(E519K), identified in 83 individuals of European ancestry across IRD registries (Belgian discovery cohort, n=2,873; replication cohort, n=18,796). Long-read sequencing-based haplotyping revealed a shared region of 464 kb, supporting a founder effect. Genotype-phenotype data support dominant inheritance and phenotypic variability respectively, characterized by late-onset macular dystrophy with two main subtypes, a pathognomonic mottled subtype and a pattern dystrophy subtype. Functional studies showed that the p.(E519K) variant affects RPE65 enzymatic activity, correlating with lower protein expression. Protein modelling and cellular thermal shift assays further supported a destabilizing effect on protein structure. Overall, our work provides strong genetic, clinical, molecular and functional evidence for a novel dominant RPE65 retinopathy in multiple families in Europe and North America due to a Belgian founder variant. This discovery reduces the diagnostic gap in dominant IRD, particularly in individuals of European ancestry. Finally, it lays the foundation for developing therapeutic strategies targeting dominant RPE65 retinopathy.

Albinism is a genetically heterogeneous disease in which 21 genes are known so far. Its inheritance mode is autosomal recessive except for one X-linked form. The molecular analysis of exonic sequences of these genes allows for about a 70% diagnostic rate. About half (15%) of the unsolved cases are heterozygous for one pathogenic or probably pathogenic variant. Assuming that the missing variant may be located in non-coding regions, we performed sequencing for 122 such heterozygous patients of either the whole genome (27 patients) or our NGS panel (95 patients) that includes, in addition to all exons of the 21 genes, the introns and flanking sequences of five genes, TYR, OCA2, SLC45A2, GPR143 and HPS1. Rare variants (MAF < 0.01) in trans to the first variant were tested by RT-PCR and/or minigene assay. Of the 14 variants tested, nine caused either exon skipping or the inclusion of a pseudoexon, allowing for the diagnosis of 11 patients. This represents 9.8% (12/122) supplementary diagnosis for formerly unsolved patients and 75% (12/16) of those in whom the candidate variant was in trans to the first variant. Of note, one missense variant was demonstrated to cause skipping of the exon in which it is located, thus shedding new light on its pathogenic mechanism. Searching for non-coding variants and testing them for an effect on RNA splicing is warranted in order to increase the diagnostic rate.

Introduction Infantile nystagmus and foveal hypoplasia associated with AHR gene defects is a newly recognized and rare disorder. Our aim was to present a patient with a novel biallelic AHR pathogenic variant with electrophysiological evidence of chiasmal misrouting. Materials and methods Complete ocular examination, fundus imaging, visual evoked potentials (VEP) and full-field electroretinography were performed at initial presentation. Genetic testing was performed by whole exome sequencing. Results Female patient of 6 years old presented a reduced best corrected visual acuity, an infantile nystagmus and a grade III typical foveal hypoplasia without ocular hypopigmentation. A crossed asymmetry was discovered on pattern onset/offset VEP. Genetic testing put in evidence a novel homozygous variant in AHR: c.2242del, p. (Gln748Lysfs^*5). During 11-years follow-up period, BCVA gradually improved. There was no evidence of retinal degeneration. Conclusion AHR gene defects could be associated with infantile nystagmus, foveal hypoplasia and chiasmal misrouting.