Yannis Mantas Paulus

Objective: This study aims to assess the association between systemic hypoxia and progression from non-proliferative diabetic retinopathy (NPDR) to proliferative diabetic retinopathy (PDR) and vision-threatening outcomes. Methods and analysis: This retrospective cohort study used the TriNetX Network to identify adults with NPDR with no prior PDR. Systemic hypoxia was defined using International Classification of Diseases 10th revision (ICD-10) codes for anaemia, obstructive sleep apnoea, chronic respiratory failure, chronic obstructive pulmonary disease, pulmonary hypertension or interstitial lung disease. Both cohorts were propensity score-matched. The primary outcome was progression to PDR. Secondary outcomes included vitreous haemorrhage, tractional retinal detachment, legal blindness, diabetic macular oedema and therapeutic interventions. Time-to-event analyses were performed using Cox proportional hazards models, and 5-year cumulative incidence was assessed using Kaplan-Meier analysis. Results: After matching, 92 454 patients were included in each cohort with well-balanced baseline characteristics (all standardised mean differences <0.1). Systemic hypoxia was associated with increased risk of PDR progression (HR 1.24; 95% CI 1.18 to 1.31; p<0.001). 5-year cumulative PDR incidence was higher in the hypoxia cohort (4.2% vs 2.6%; risk ratio 1.63). Patients with systemic hypoxia also had higher risks of vitreous haemorrhage, tractional retinal detachment, blindness and increased utilisation of panretinal photocoagulation and pars plana vitrectomy. Subgroup analyses identified anaemia as the primary driver of PDR progression, whereas respiratory aetiologies were not independently associated after matching. Conclusions: Systemic hypoxia, a diagnosis-code-based composite in which anaemia-associated phenotypes showed the strongest association, is associated with accelerated PDR progression and near-doubled legal blindness risk. Anaemia screening may offer an opportunity to reduce vision-threatening complications in diabetic patients. These findings are hypothesis-generating and warrant prospective validation before informing screening pathways, given the retrospective design and absence of measured oxygen saturation, haemoglobin and glycaemic data.

Usher syndrome type 3A (USH3A), caused by CLRN1 mutations, leads to progressive deaf-blindness. Although CLRN1 expression has been localized to Müller glia (MG), the pathogenic mechanisms linking glial CLRN1 deficiency to photoreceptor death remain undefined. We utilized a novel large-animal model to elucidate cell-type-specific pathogenesis at single-nucleus resolution. CLRN1-/- rabbits were generated via CRISPR/Cas9. Retinal phenotyping included longitudinal optical coherence tomography (OCT), electroretinography (ERG), and histology for up to 3 years. Single-nucleus RNA sequencing (snRNA-seq) was performed at a pre-symptomatic stage (10 months, n = 3/genotype). Differentially expressed genes (DEGs) were identified using pseudobulk DESeq2 and validated via immunofluorescence. CLRN1-/- rabbits exhibited progressive outer nuclear layer thinning and delayed ERG responses starting after 20 months, mimicking human clinical kinetics. The snRNA-seq revealed that CLRN1 is exclusively expressed in a discrete subset (∼24%) of MG. Mutant MG exhibited transcriptomic downregulation of the adherens junction component CTNNA2, which was validated at the protein level by immunofluorescence staining. Photoreceptors exhibit extensive non-cell-autonomous dysregulation (232 rod and 68 cone DEGs) characterized by loss of the cone synaptic organizer TENM2 and rod-specific splicing dysregulation (CWF19L2). Concurrently, inner retinal neurons mounted a robust proteostatic response via HSP90 family upregulation (HSP90AB1, HSP90AA1, HSP90B1), a defensive signature absent in vulnerable photoreceptors. Our findings suggest an "anchor-shield" mechanism: MG CLRN1 loss compromises the outer limiting membrane (OLM) adhesion complex ("anchor"), and photoreceptors lack the proteostatic response ("shield") seen in resilient inner neurons. Restoring CLRN1 in Müller glia or implementing photoreceptor neuroprotection represents promising therapeutic strategies for USH3A.

How far can DeepSeek-R1 and OpenAI see in ophthalmology diagnosis and decision making?

Photoacoustic tomography (PAT) is a rapidly emerging biomedical imaging modality. To achieve high-performance imaging, signal acquisition in PAT should ideally meet conditions such as full-view detection and dense sampling. However, these circumstances are rarely met in reality, leading to the ill-posed problem of image reconstruction from incomplete projections. Here we propose a physics-driven deep learning-based filtered back projection (dFBP) framework to address this important challenge. The dFBP network is inspired and constructed based on the physical model of the analytical filtered back projection algorithm and consists of a filtering module, a back-projection module, and a fusion module connected in cascade. The dFBP network is driven by physics and is thus interpretable and easy to train while being highly robust and versatile. Numerical and experimental validation on animals and humans show that dFBP-based PAT can achieve direct signal-to-image transformation with enhanced accuracy and reconstruct high-quality, artifact-suppressed images from sparse-view, limited-view, and acoustic heterogeneity-contaminated projections. The proposed dFBP provides a practical solution for high-performance PAT imaging under non-ideal conditions.

Background: Diabetic Retinopathy (DR), a leading cause of preventable blindness, is conventionally evaluated using seven-field retinal fundus photography. Ultra-widefield fluorescein angiography (UWF FA) allows for detailed visualization and quantification of peripheral retinal non-perfusion (NP). This study aims to assess changes in NP and neovascularization (NV) following intravitreal anti-VEGF and steroid treatments using UWF FA. Methods: This retrospective, single-center cohort study included 65 eyes from 45 patients with Type 1 or 2 diabetes mellitus (DM) and UWF FA images acquired across two visits. Images were manually segmented by masked, trained graders, annotating foveal avascular zone (FAZ), NP, and NV surface areas. Patient demographics, clinical history, and treatment details were extracted via chart review. Stepwise multivariate linear regression identified predictors of changes in FAZ, NP, and NV areas, while causal modeling estimated effects of anti-VEGF and steroid treatments over time. Results: Eyes treated with steroid injections ( N = 13) had decreased total and mid-periphery retinal NP areas compared to untreated eyes or eyes that received anti-VEGF alone (total change in NP area: 4.32 ± 24.66 vs. 32.30 ± 65.40 vs. 27.21 ± 42.09 mm ² ; p < 0.05). Increased frequency of steroid injections predicted decreased FAZ area ( p = 0.0036). Longer duration between FAs was associated with larger NP areas by 0.87 mm ² per additional month (95% CI, 0.25 to 1.5; p = 0.0072), whereas more steroid injections ( p = 0.035), DME on initial clinical presentation ( p = 0.013), and vitreous hemorrhage development ( p = 0.0002) were associated with decreased NP areas. Compared to eyes in type 1 diabetics, eyes with type 2 DM were associated with − 56.45 mm ² smaller total NP areas over time (95% CI, -92.3 to -20.52; p = 0.0028). Causal analyses suggest each additional steroid injection was associated with a 31.38 mm ² decrease in total NP area (95% CI, -58.14 to -4.62; p = 0.022). Anti-VEGF treatment ( N = 28) did not have a significant effect on FAZ (95% CI, -0.153 to 0.060; p = 0.40), NP (95% CI, -17.20 to 3.91; p = 0.22), or NV areas (95% CI, -0.391 to 0.282; p = 0.75). Conclusions: Compared to anti-VEGF, intravitreal steroids produced significant improvement in retinal perfusion. Quantitative biomarkers from UWF FA may provide an objective approach to assessing treatment response and DR severity over time. Trial registration: Not applicable.

Choroidal neovascularization (CNV) represents a major cause of vision loss in various retinal diseases such as age-related macular degeneration (AMD). Current treatment involves frequent, often monthly, eye injections. The development of minimally invasive, long-term, painless, and effective ocular drug delivery systems is crucial for advancing the treatment of AMD. This study explores a novel method that integrates controllably bioresorbable silicon nanoneedles loaded with bevacizumab (Si NNs-Bev) on a tear-soluble subconjunctival patch for sustained, 1 year ocular drug delivery. The Si NNs-Bev embed into the sclera in a minimally invasive manner, undergoing controlled degradation over one year. This approach facilitates the sustained release of therapeutic agents, enhancing treatment efficacy and reducing treatment burden. Si NNs-Bev for the treatment of CNV are validated in a rabbit model of AMD. The SiNN-Bev patch achieved a sustained therapeutic effect on CNV regression, with a mean reduction of 82% by 4 months that is persistent for at least 1 year with minimal recurrence, which is consistent with the localized drug delivery mechanism facilitated by the transscleral microneedles. These preliminary findings underscore the potential of SiNNs as a platform technology for long-term, sustained ocular therapeutics.

We present a multimodal optical imaging platform for in vivo tracking of subretinally transplanted hiPSC-RPE stem cells in a rabbit model. The approach combines photoacoustic microscopy (PAM), optical coherence tomography (OCT), and fluorescence imaging, enhanced with gold nanochains, to monitor cell migration.

Retinal laser therapy has been a mainstay for treating proliferative diabetic retinopathy, retinal vascular disease, and retinal breaks since 1961. However, conventional millisecond photocoagulation can cause permanent scarring and procedure discomfort, motivating the development of damage-sparing approaches that preserve the neurosensory retina. Clinically, panretinal photocoagulation remains effective for proliferative disease but trades off peripheral visual field and night vision. This review synthesizes development, mechanisms, and clinical evidence for laser modalities, including short-pulse selective retinal therapy (SRT), subthreshold diode micropulse (SDM), and pattern-scanning photocoagulation. We conducted a targeted narrative search of PubMed/MEDLINE, Embase, Web of Science, and trial registries (1960–September 2025), supplemented by reference list screening. We prioritized randomized/prospective studies, large cohorts, systematic reviews, mechanistic modeling, and relevant preclinical work. Pulse duration is the primary determinant of laser–tissue interaction. In the microsecond regime, SRT yields retinal pigment epithelium (RPE)-selective photodisruption via microcavitation and uses real-time optoacoustic or OCT feedback. SDM 100–300 µs delivers nondamaging thermal stress with low duty cycles and titration-based dosing. Pattern-scanning platforms improve throughput and tolerance yet remain destructive photocoagulation. Feedback-controlled SRT shows anatomic/functional benefit in chronic central serous chorioretinopathy and feasibility in diabetic macular edema. SDM can match threshold macular laser for selected DME and may reduce anti-VEGF injection burden. Sub-nanosecond “rejuvenation” lasers show no overall benefit in intermediate AMD and may be harmful in specific phenotypes. Advances in delivery, dosimetry, and closed-loop feedback aim to minimize collateral damage while retaining therapeutic effect. Key gaps include head-to-head trials (SRT vs. PDT/SDM), standardized feedback thresholds across pigmentation and devices, and long-term macular safety to guide broader clinical adoption.