Carel B. Hoyng

Purpose: A significant proportion of cases with rare inherited retinal disease (IRD) remain genetically unresolved following short-read whole genome sequencing (WGS). This study aimed to increase the diagnostic yield of two cohorts in a research setting. One cohort was previously screened by short-read WGS (n=120 cases), while the second cohort (n=28 cases) was not included in any previous short-read WGS studies. In contrast to the larger cohort, these cases only underwent either exome sequencing (ES) or single-molecule molecular inversion probe (smMIPs) sequencing as pre-screening in earlier studies. For all cases, we performed a stepwise, case-by-case reanalysis in which short-read WGS was generated (n=28 cases) and (re)analyzed short-read WGS data using an optimized approach. For a subset of cases (n=20) that remained unresolved, this was followed by long-read WGS. Together, these steps aimed to improve the diagnostic yield in these cohorts. Methods: Short-read WGS data of 148 IRD-cases were (re)analyzed using updated allele frequency databases, new variant caller and variant predictor tools, as well as updated and extended gene-panels to identify causal single nucleotide variants (SNVs) and structural variants (SVs). For 20 genetically unresolved cases with sufficient high-quality DNA available, long-read WGS was performed. Results: The (re)analysis of short-read WGS data in the 120 previously WGS-screened unresolved cases resulted in a diagnostic yield of 15% (18/120). For the 28 individuals without prior WGS, performing WGS analysis within this study provided an additional yield of 32% (9/28). Furthermore, long-read WGS contributed two genetic diagnoses among the 20 long-read WGS-sequenced probands. Amongst other reasons, new genetic diagnoses were attributed to pathogenic variants in genes newly associated with IRDs and in genes that were not included in the gene panels applied in previous WGS studies. The Mobile Element Locator Tool facilitated the identification of two pathogenic Alu insertions in short-read WGS. Long-read WGS identified two additional pathogenic variants. Conclusion: These results support periodic reanalysis of existing short-read WGS data as a key component of IRD genetic diagnostics, with selective long-read WGS representing a valuable additional approach for unresolved cases. Together, these complementary strategies offer a practical framework for improving molecular diagnosis and narrowing the remaining diagnostic gap in IRDs.

Background: The availability of dedicated image analysis scripts for adaptive optics (AO)–flood illumination ophthalmoscopy (FIO) is limited, especially for large-scale measurements and analyses of nonhealthy images. This limitation highlights the need for alternative approaches to facilitate automated and scalable analysis. Large language models may help develop such scripts. Objective: This study aimed to generate an analysis script for AO-FIO images in the R programming language using a widely available generative AI (GenAI; specifically, GPT-4) as a proof of principle for generating a functional but nonvalidated script. Methods: GPT-4 was used to generate an R script for the analysis of AO-FIO images. The code generated by GPT-4 was fine-tuned iteratively based on trial and error, testing the script for image preprocessing and analysis using images from 4 participants, including 1 healthy individual and 3 patients with Stargardt disease. The script code was subsequently checked for errors by another researcher who was naive to previous coding, using a different test set of AO-FIO images from 4 other participants (n=1 healthy individual and n=3 patients with Stargardt disease). The cone counts from 5 AO image snippets were compared with the counts independently recorded by 2 human graders and those generated by pre-existing AO analysis software that was trained on healthy participants. Results: After 54 iterations of instructions, a functional R script for the analysis of AO-FIO images was generated. The script identified and quantified blobs. Conclusions: We developed a preliminary nonvalidated script for cone detection in retinal AO images with support from GPT-4. Before such a script can be used in clinical research, it should undergo more fine-tuning and extensive testing. Future work should enhance the image analysis capabilities of the script and validate its results to assess the potential of AO-based cone counts as biomarkers in clinical trials.

Age-related macular degeneration (AMD) shows substantial clinical heterogeneity that remains unexplained despite extensive genetic and clinical characterization. We evaluated whether proteomic stratification could provide insight beyond clinical phenotype and genetic risk. We performed 384-plex plasma proteomics in a cohort of 215 individuals, including patients with early and late neovascular AMD, other complement-associated retinal diseases, and age-matched controls. Proteome-based reclassification identified four disease-overarching clusters. Neovascular AMD cases were partitioned almost exclusively between two clusters (30/36). Early AMD cases were predominantly assigned to one of these clusters (10/18), whereas only two localized to the other (2/18). Both AMD-associated clusters shared elevated levels of a protein module enriched for lipoprotein-related functions compared to the other clusters. However, the cluster containing both early and neovascular AMD cases showed higher levels of additional protein modules enriched for complement pathways and cellular stress-response pathways compared with the other AMD-associated cluster. Importantly, this molecular divergence in neovascular AMD could not be explained by genetic predisposition (i.e., 52-variant AMD genetic risk score), signatures of biological ageing, nor by other clinical features. Together, these findings support two proteomic endotypes of neovascular AMD with distinct involvement of cellular stress pathways.

BACKGROUND The availability of dedicated image analysis scripts for adaptive optics (AO)–flood illumination ophthalmoscopy (FIO) is limited, especially for large-scale measurements and analyses of nonhealthy images. This limitation highlights the need for alternative approaches to facilitate automated and scalable analysis. Large language models may help develop such scripts. OBJECTIVE This study aimed to generate an analysis script for AO-FIO images in the R programming language using a widely available generative AI (GenAI; specifically, GPT-4) as a proof of principle for generating a functional but nonvalidated script. METHODS GPT-4 was used to generate an R script for the analysis of AO-FIO images. The code generated by GPT-4 was fine-tuned iteratively based on trial and error, testing the script for image preprocessing and analysis using images from 4 participants, including 1 healthy individual and 3 patients with Stargardt disease. The script code was subsequently checked for errors by another researcher who was naive to previous coding, using a different test set of AO-FIO images from 4 other participants (n=1 healthy individual and n=3 patients with Stargardt disease). The cone counts from 5 AO image snippets were compared with the counts independently recorded by 2 human graders and those generated by pre-existing AO analysis software that was trained on healthy participants. RESULTS After 54 iterations of instructions, a functional R script for the analysis of AO-FIO images was generated. The script identified and quantified blobs. CONCLUSIONS We developed a preliminary nonvalidated script for cone detection in retinal AO images with support from GPT-4. Before such a script can be used in clinical research, it should undergo more fine-tuning and extensive testing. Future work should enhance the image analysis capabilities of the script and validate its results to assess the potential of AO-based cone counts as biomarkers in clinical trials.

Objective: This study aims to analyze the progression of hearing loss using prospectively collected data and describe the vestibular phenotype of Usher syndrome type 2A (USH 2a ). Study Design: Longitudinal, prospective, observational natural history study. Patients: Patients with USH2a and USH2A -associated non-syndromic retinitis pigmentosa (nsRP). Main Outcome Measures: Hearing loss progression was measured by pure tone audiometry (PTA) and phoneme scores. Vestibular function was assessed by velocity step tests (VSTs), video head impulse tests (vHITs), caloric reflex tests, and vestibular evoked myogenic potentials (VEMPs). Patient-reported symptoms were assessed by questionnaires (Usher Lifestyle Survey, SF-12, PHQ-9, DHI, and SSQ). Results: Thirty-three patients with USH2a and 2 patients with nsRP were included. PTA _0.5-4kHz thresholds showed a significant decrease of 2.3 dB ( P =0.017) over 4 years. The progression of hearing loss was most pronounced in the mid-to-high frequencies. No significant differences were observed in phoneme scores as measured by the Speech Reception Threshold (SRT). The 2 patients with nsRP had normal hearing. Vestibular evaluation showed abnormal cervical VEMPs in 34% and abnormal ocular VEMPs in 75% of USH2a patients. VSTs, caloric reflex tests, and vHIT results were within normal limits in more than 90% of subjects. Patient questionnaires revealed no major balance problems as indicated by the DHI. Conclusions: This analysis provides robust evidence of measurable hearing loss progression in PTA _0.5-4kHz thresholds in USH2a patients beyond presbycusis. Speech understanding, as measured by SRT, remains relatively stable. Vestibular evaluation revealed no semicircular canal dysfunction, although VEMP results suggest a potential subclinical otolith organ impairment in USH2a patients.

To quantitatively investigate disease progression and genotype-phenotype correlations in Stargardt disease patients harboring the common ABCA4 variant c.5882G>A p.(Gly1961Glu), aiming to evaluate the contribution of second ABCA4 variants to disease severity. This multicenter study included 77 Stargardt disease patients uniformly carrying the ABCA4 c.5882G>A p.(Gly1961Glu) variant from Nijmegen, Bonn, and Basel. Ellipsoid zone (EZ) loss was measured using automated segmentation of optical coherence tomography images. Progression was assessed with linear mixed-effects models. Disease severity was quantified by the age-at-criterion EZ loss, defined as the estimated age when EZ atrophy reached 6.25 mm², allowing a comparison of the effects of different second alleles. A total of 52 patients were included in the study. The median age at first observation was 41 years (interquartile range, 27-52 years). The median age at criterion EZ loss across all patients was 43 years (interquartile range, 31-51 years). Substantial variability in the progression of EZ loss was observed among patients despite the shared c.5882G>A p.(Gly1961Glu) genotype. Differences in specific second ABCA4 variants were strongly associated with variation in disease progression, with some variants linked to markedly earlier progression (e.g., c.1957C>T, -50.7 years) and others associated with delayed progression (e.g., c.1648G>A, +19.8 years). The ABCA4 c.5882G>A p.(Gly1961Glu) variant is associated with substantial interpatient variability in disease severity, heavily influenced by the nature of the second ABCA4 allele. These findings highlight the importance of precise genotypic characterization in prognostic counseling, clinical trial stratification, and the design of targeted gene therapies.

Purpose: Mesopic microperimetry (mMP) is a promising functional endpoint in clinical trials for Stargardt disease type 1 (STGD1). This study evaluated the test–retest variability of mMP and influencing factors, which is essential for ensuring reliability in future STGD1 trials. Methods: One hundred and fifteen eyes from 68 patients enrolled in the prospective, tertiary, multicentre STArgardt Remofuscin Treatment Trial (STARTT) underwent mMP testing using the macular integrity assessment (MAIA) microperimeter (CenterVue, Padova, Italy) at both the screening (first) and baseline (second) visits of the trial. Test–retest variability was assessed using Bland–Altman analyses and coefficients of repeatability (CoR). Retinal sensitivity metrics included mean sensitivity ( MS ) and pointwise sensitivity ( PWS ). Other factors including fixation stability, exam duration and learning effect were analysed. Results: MS demonstrated the lowest variability (CoR: 3.53 dB, 95% CI: 3.07–3.99), while PWS exhibited the highest (CoR: 12.69 dB, 95% CI: 12.47–12.91). Variability decreased in sensitivity ranges from −1 to 3 dB and 16 to 32 dB and from central to peripheral regions. Test duration (Spearman's ρ = 0.609, p < 0.001) and fixation losses (Spearman's ρ = 0.284, p = 0.003) were significantly associated with increased variability. Other fixation stability metrics showed no correlation. No learning effect was observed. Conclusions: Given its high variability, PWS should be used cautiously. MS offers lower variability but may mask localised functional changes. A parafoveal ring strategy may improve reliability but requires validation. Limiting test duration to ≤450 seconds and comprehensive operator training are recommended to minimise potential bias.