Moorfields Eye Hospital NHS Foundation Trust

Artificial intelligence (AI) is rapidly transforming numerous aspects of daily life, including clinical practice and biomedical research. In light of this rapid transformation, and in the context of medical genetics, we assembled a group of leaders in the field to respond to the question about how AI is affecting, and especially how AI will affect, medical genetics. The authors who contributed to this collection of essays intentionally represent different areas of expertise, career stages, and geographies, and include diverse types of clinicians, computer scientists, and researchers. The individual pieces cover a wide range of areas related to medical genetics; we expect that these pieces may provide helpful windows into the ways in which AI is being actively studied, used, and considered in medical genetics.

This review offers comprehensive insights into the presenting features, terminology, imaging techniques and management strategies associated with uveitis, specifically designed to help neurologists understand this complex condition. We have created a glossary of terms used in uveitis care and ocular imaging to help clarify terminology. We have ordered uveitis subtypes in an intuitive manner, focusing on those that neurologists are more likely to encounter. We have written the article from the perspective of uveitis specialists practising in the UK, while emphasising the global variability in clinical presentations and causes. By offering practical guidance on recognising uveitis features as well as treatment options, we aim for this to be a neurologists’ aide mémoire to help manage intraocular inflammation.

Purpose Central serous chorioretinopathy (CSC) is a prevalent maculopathy, but epidemiological studies are few. In this study, we determined the prevalence of CSC for the first time in a Scandinavian population. Methods This cross-sectional study was based on nationwide opportunistic retinal examination from 79 high street chain optician stores in Denmark. Retinal imaging was made using non-mydriatic colour fundus photography. Any abnormal result in the optometrist-facilitated retinal examination was referred to tele-ophthalmologic evaluation, which was performed by experienced consultant ophthalmologists who diagnosed CSC. Results During the 4-year study period, a total of 968 610 unique individuals underwent retinal examination, which corresponded to 16.3% of the entire population of Denmark. Of these, 113 individuals were diagnosed with CSC, which corresponded to a prevalence of 14 per 100 000. Individuals with CSC presented at a mean age of 48.2 ± 12.3 years; however, CSC was present in a large age range as both teenagers and the elderly with CSC were identified. Male biological sex was a statistically significant risk factor (odds ratio: 2.33; 95% confidence interval: 1.64–3.33, p < 0.0001). By extrapolating prevalence numbers to population statistics, we estimate that 219 females and 511 males had CSC in Denmark in December 2022. Conclusion We identified a prevalence of 14 per 100 000, confirmed male biological sex as a significant risk factor for CSC, and found that the disease most commonly occurs among individuals aged 30–60 years. Further studies with multimodal imaging including optical coherence tomography are warranted for better accuracy.

Purpose To evaluate the performance of a custom ChatGPT-based chatbot in triaging ophthalmic emergencies compared to trained ophthalmologists. Methods One hundred hypothetical ophthalmic cases were created based on actual patient data from an ophthalmic emergency department, including details such as age, symptoms and medical history. Three experienced ophthalmologists independently graded these cases using a four-tier severity scale, ranging from Grade 1 (immediate care required) to Grade 4 (non-urgent care). A customized version of ChatGPT was developed to perform the same grading task. Inter-rater agreement was measured between the chatbot and the ophthalmologists, as well as among all human graders. Results The chatbot demonstrated substantial agreement with the ophthalmologists, achieving Cohen's kappa scores of 0.737, 0.749 and 0.751, respectively. The highest agreement was between ophthalmologist 3 and the chatbot (κ = 0.751). Fleiss’ kappa for overall agreement among all graders was 0.79, indicating substantial agreement. The Kruskal–Wallis test showed no statistically significant differences in the distribution of grades assigned by the chatbot and the ophthalmologists (p = 0.967). Bootstrap analysis revealed no significant difference in kappa values between the chatbot and human graders (p = 0.572, 95% CI −0.163 to 0.072). Conclusions The study demonstrates that a customized chatbot can perform ophthalmic triage with a level of accuracy comparable to that of trained ophthalmologists. This suggests that AI-assisted triage could be a valuable tool in emergency departments, potentially enhancing clinical workflows and reducing waiting times while maintaining high standards of patient care.

Background Inherited retinal diseases (IRDs) are the leading cause of blindness in young people in the UK. Despite significant improvements in genomics medicine, the diagnosis of these conditions remains challenging, and around 40% do not receive a definite genetic diagnosis after extensive genetic testing. This survey aims to investigate the experience of individuals affected by IRDs, their relatives, friends and caregivers, focusing on their care and diagnostic journey. Additionally, it explores the potential acceptability of artificial intelligence (AI) technologies, such as Eye2Gene, that predict causative genes from retinal images of patients with IRDs. Methods This cross-sectional survey included Likert scale and open-ended questions and was distributed electronically using the Qualtrics platform between April and August 2024. The survey included questions on respondent demographics; their journey to receive specialist care and genetic testing; their information needs and their attitude towards AI-augmented diagnosis. Descriptive statistics and content analysis were used to interpret the survey responses. Results The survey had 247 responses, of which 242 were analysed after removing four duplicates and one without consent; 80.2% were patients and the remainder were relatives, friends or caregivers. There was substantial variability in patient diagnostic journeys in terms of waiting times to see a specialist (IQR, 1–4 years), commute required (IQR, 10–74 miles) and number of visits to reach a diagnosis (IQR, 2–4). A substantial proportion of patients (35.8%) had a change in diagnosis. The majority of respondents (>90%) were overwhelmingly in favour of the integration of AI into the IRD pathway to accelerate genetic diagnosis and improve care. Conclusion This survey identifies several key gaps and disparities in the IRD care pathway which may potentially be bridged with AI. The survey also reveals a favourable attitude towards incorporating AI into diagnostic testing of IRDs.

Purpose To present a case of non-uveal intraocular metastatic disease originating from relapsed breast carcinoma. Methods We report a 67-year-old female with a history of HER2-positive breast carcinoma in remission and recent squamous cell carcinoma of the esophagus presenting with ocular symptoms. Comprehensive ophthalmic examinations and multimodal imaging were performed. Diagnostic procedures included diagnostic vitrectomy and cytology analysis. Results The patient was found to have metastatic carcinoma to the vitreous and retina with an immunophenotype compatible with breast carcinoma. Neurological symptoms led to the discovery of metastatic brain lesions. The patient underwent palliative systemic treatment and stereotactic brain irradiation. Follow-up showed the progression of the ocular lesions. Conclusion This case informs on the clinical and pathological findings of a rare non-uveal intraocular metastatic disease originating from relapsed breast carcinoma. Vitreous biopsy and multimodal imaging were required for a proper diagnosis and management and would serve as a reference for further cases.

Primary vitreoretinal lymphoma (PVRL) is a high‐grade extranodal non‐Hodgkin lymphoma, with limited prospective data to inform practice. High rates of central nervous system (CNS) relapse contribute to its poor prognosis. This international multicentre retrospective cohort study aimed to characterise real‐world contemporary practice and outcomes in PVRL (2010–2022). Sixty patients were included from 11 centres across Australia, Singapore, Canada and the United Kingdom. Most patients had systemic therapy included in their initial management (63%) either alone or in combination with local therapies; 13% had upfront autologous stem cell transplantation (ASCT). The overall response rate was 78%. With a median follow‐up of 68 months, the median progression‐free survival (PFS) was 25 months, with a median overall survival (OS) of 73 months. Neither incorporation of systemic therapy into initial treatment nor upfront ASCT demonstrated a statistically significant impact on PFS or OS. The 5‐year cumulative incidence of CNS relapse was 33%, with front‐line systemic therapy being the only predictive factor for CNS relapse in a multivariate model, hazard ratio of 0.30 (95% CI 0.09–0.98, p = 0.05). Concerning heterogeneity in real‐world approaches to diagnosis, staging and management approaches were identified. Further international collaborative efforts are required to address the unmet need in this rare entity.

Purpose To evaluate the long-term outcomes of next-generation tyrosine kinase inhibitor (TKI) therapy in patients with choroidal metastases (CMs) secondary to non-small cell lung cancer (NSCLC). Major Findings This case series includes five eyes from three patients with stage IV NSCLC and CMs. All patients presented with unilateral visual loss, though two exhibited bilateral choroidal lesions on multimodal imaging, including fundus photography, dye-based angiography, optical coherence tomography (OCT), and ultrasonography. Systemic evaluation confirmed ALK-rearranged NSCLC (case 1) and EGFR-mutated NSCLC (cases 2 and 3), leading to first-line treatment with Alectinib and Osimertinib, respectively. Within months of therapy initiation, significant anatomical and visual improvements were noted, with complete lesion resolution and photoreceptor restoration by 6 months. At final follow-up (24–36 months), high-resolution OCT showed residual granularity in the outer retinal layers, while OCT angiography (OCTA) revealed persistent choriocapillaris flow voids. Microperimetry demonstrated distinct functional alterations based on lesion location. Conclusions This series highlights favorable long-term outcomes in NSCLC patients with CMs treated with next-generation TKIs. It underscores the value of multimodal imaging and functional assessments in monitoring disease progression and treatment response, emphasizing the importance of early targeted therapy in optimizing visual outcomes.

Purpose To report on the 1-year incidence and risk factors of visual loss in patients with diabetic macular ischemia. Methods Prospective, observational cohort. Patients had stable panretinal photocoagulation treated eyes for either proliferative diabetic retinopathy or severe non-proliferative diabetic retinopathy. Main outcomes measures: best corrected visual acuity (BCVA) and low luminance visual acuity (LLVA). Optical coherence tomography (OCT), OCT angiography (OCTA), and retinal sensitivity were recorded at baseline and week 52. Results The cohort included 53 patients and 66 eyes. The mean (SD) age was 57.9 (10.6) years, with a mean (SD) diabetes duration of 17.9 (11.0) years. 40.4% of eyes lost ≥5 BCVA and/or LLVA letters. Disorganization of retinal layers (DRIL) was observed in 71.4% of eyes with vision loss compared to 38.7% of eyes without vision loss, odds ratio=3.96 (95% CI, 1.25 to 13.84, P= 0.02). Baseline deep vascular density and superficial vascular density correlated positively with week 52 BCVA (rho=0.32, 95% CI: -0.02 to 0.59; P =0.05) and week 52 LLVA (rho=0.33, 95% CI -0.02 to 0.61; P =0.04) respectively. Conclusion DRIL was associated with vision loss, and baseline OCTA vascular parameters positively predicted VA outcomes.

Acanthamoeba keratitis (AK) was first identified in 1972 and the first patient cured with propamidine was reported in 1985. Treatment outcomes, before the advent of the first effective anti-amoebic treatment, were known to be poor and often required therapeutic keratoplasty (TK) but have not been evaluated in detail. Analysis of these outcomes has value for several reasons: it gives an historical perspective, describes the natural history of AK when the disease was minimally modified by the early treatments and provides a benchmark against which current treatments can be compared and how these have changed the therapeutic results. We conducted a systematic literature review for the period 1970–1995 using PRISMA guidelines. The population of interest comprised patients with AK treated without products having established anti-amoebic activity against both trophozoites and cysts (biguanides or diamidines). The outcomes of interest were medical cure, TK and enucleation. Proportions and 95% confidence intervals were estimated. Fifty-six case reports were eligible. Risk factors for AK were reported in 44/56 patients: contact lens wear in 30/44 (68.2%) and trauma in 14/44 (31.8%). The mean time from presentation to diagnosis was 7.3 weeks (standard deviation 9.3 weeks); 13/56 (23.2%) were diagnosed within 4 weeks. Topical treatments given to patients included corticosteroids (85.2%), antibiotics (85.2%), antivirals (72.2%) and antifungals (51.8%). Final visual acuity was ≥ 20/40 in 17/33 (51.5%) patients with no missing data. Medical cures were reported in 11/56 patients (19.6%), TK in 38/56 (67.9%), other surgery in 4/56 (7.1%) and enucleation in 3/56 (5.4%). This study suggests that, before the availability of propamidine as the first effective treatment for AK, the clinical outcome of these patients was poor with only a few patients cured without surgery. These findings should be interpreted with caution because they rely on case reports and series that are subject to inherent bias.

Purpose To study the distribution of lesions of diabetic retinopathy (DR), cotton wool spots (CWS) and scattered exudates on ultra-widefield images (UWFI) to better define DR phenotypes in Indians. Methods This multicenter cross-sectional observational study on UWFI in individuals with DR was conducted across four tertiary eye care centres in South India. The Early Treatment Diabetic Retinopathy Study (ETDRS) grid was superimposed on the UWFI and each image was analysed for the distribution of the retinal lesions. Results Out of a total of 1002 eyes of 1002 patients with gradable images, 459 eyes (46 %) had predominantly peripheral retinal lesions (PPL), of which haemorrhages /microaneurysms (86.1%) were the most frequent lesions. Only 5.5 % of eyes had retinal neovascularization limited to outside the 7-field. Strong agreement was demonstrated in the grading of retinopathy using ETDRS and UWFI (κ = 0.9). 15.3% eyes had a higher grade of DR when graded on UWFI compared to 7-field ETDRS grid. Scattered exudates and cotton wool spots on UWFI are prevalent in Indian eyes with DR. Scattered exudates on UWFI was more prevalent in eyes with PPL and was associated with increasing severity of diabetic retinopathy. Conclusion The presence of PPL and scattered exudates on UWFI are associated with increasing DR severity in Indians and these findings need to be considered in new classifications of DR.

Brachytherapy with an iodine-125 or ruthenium-106 plaque is the first choice of treatment for uveal melanomas in most centers. Local tumor control and prevention of avoidable ocular morbidity require meticulous preoperative assessment, expert radiation dosimetry, safe treatment planning, good surgical technique, long-term surveillance, and proper treatment of any side effects and complications. The most serious complication is local tumor relapse, which appears to be associated with increased metastatic mortality. It is useful to distinguish between collateral damage to healthy ocular tissues when these receive high-enough levels of radiation, and toxic tumor syndrome, caused by intra-tumoral radiation-induced vasculopathy, with exudation, retinal detachment, production of angiogenic factors, iris neovascularization, and neovascular glaucoma.

Research into the molecular genetics of uveal melanomas (UM) has identified several predictors of metastatic disease as well as signaling pathways and molecules that can be targeted using systemic therapies. Most UM have GNAQ/11 initiating mutations, which activate the MAPK signaling pathway, with dysregulation of the PI3K/AKT and others pathways. Some patients develop UM because they have the BAP1 tumor predisposition syndrome, which also causes malignant mesothelioma and renal cell carcinoma. Metastasis is associated with many genetic aberrations, particularly chromosome 3 loss, chromosome 8q gain, and somatic BAP1 mutation, whereas SF3B1 mutation predicts late metastasis, while chromosome 6p gain, and EIF1AX and CNKSR3 mutations are associated with a good prognosis. Data from The Cancer Genome Atlas (TCGA) indicate four prognostic groups of UM, categorized according to chromosome copy number variations, mutations, and DNA methylation profiles. Other abnormalities include PTEN inactivation, inhibition of the retinoblastoma (Rb) and p53 pathways, cyclin D1 dysregulation, CDKN2a promoter hypermethylation, and overexpression of preferentially expressed antigen in melanoma (PRAME). The latter, which occurs in 25% of UM, leads to genomic instability in these tumors. Several potential MAPK/MEK, PI3K/AKT targets for UM therapy have been identified, using mTOR inhibitors, tyrosine kinase inhibitors, c-Met pathway inhibitors, CXCR4 small molecule inhibitors, histone deacetylase inhibitors, and others. More recent therapies aim at harnessing the tumor microenvironment of metastatic uveal melanomas. For example, immunotherapies utilising bispecific T-cell receptor molecules that helps the body’s immune system identify and destroy uveal melanoma cells, is a strategy that is having some success with these aggressive tumors.

Many patients with uveal melanoma die as a result of metastatic spread despite successful treatment of the primary tumor. This raises important and unsettling questions about how and in whom ocular treatment influences survival, if at all. Such questions cannot be fully answered without a proper understanding of the natural history of uveal melanomas, particularly in relation to the time of onset of metastatic spread. The lack of firm evidence on which to base patient care complicates decision-making regarding matters such as the treatment of small melanocytic tumors of indeterminate malignancy and screening for metastatic disease.

Almost 50% of uveal melanoma patients develop metastatic disease despite successful ocular treatment. Metastases usually involve the liver and only rarely respond to treatment, so most patients die within a year of the onset of symptoms. There are hopes that systemic adjuvant therapy may delay or prevent metastatic disease. This requires targeting of high-risk patients according to anatomic, histologic, and genetic predictors. The metastatic disease occurs almost exclusively in patients whose tumor shows genetic aberrations such as chromosome 3 loss, chromosome 8q gains, a class 2 gene expression profile (GEP), increased preferentially expressed antigen in melanoma (PRAME) expression, and BAP1 mutation. The detection of lethal genetic aberrations has become more sensitive with the development of methods such as comparative genomic hybridization, microsatellite analysis, multiplex ligation-dependent probe amplification, single-nucleotide polymorphisms, gene expression profiling, next-generation sequencing, and immunohistochemical analysis of nuclear BAP1 expression. The time from ocular treatment to the onset of metastatic disease tends to be shorter in patients with a larger, more extensive, ocular tumor and a higher grade of malignancy as evidenced by epithelioid cytomorphology and higher mitotic count. Ideally, prognostication is performed by multivariable analysis using a tool such as the Liverpool Uveal Melanoma Prognosticator Online (LUMPO). Accurate prognostication enhances quality of life, especially in patients with a good prognosis. It also allows the targeting of systemic surveillance and systemic adjuvant therapy for high-risk patients and enhances opportunities for research.

The management of patients with posterior uveal melanoma involves detection, diagnosis, treatment, prognostication, counseling, surveillance, and, in some cases, treatment for metastatic disease (Box 13.1). Each of these activities is advancing rapidly; however, many aspects of care remain controversial.

Histologic examination of biopsy and enucleation specimens in patients with uveal melanoma can confirm the diagnosis when clinical investigation is inconclusive. Histologic predictors of metastasis enhance prognostication, especially with multivariable analysis including anatomic and genetic predictors. After enucleation, microscopic assessment of extraocular spread may help identify patients who require adjunctive external beam radiotherapy. Histologic investigation also provides insights into the pathophysiology of uveal melanomas as well as the morbidity that can be caused by the treatment of this disease. This chapter highlights the most significant histopathologic features of uveal melanoma.

Iris melanoma is the least frequent of all uveal melanomas. Because of anterior location, iris melanoma is diagnosed and treated when relatively small compared to tumors located in ciliary body and choroid. Melanoma located in the peripheral iris may represent anterior extension of a ciliary body tumor, and therefore it is of paramount importance to examine the ciliary body in cases of iris melanoma when all iris margins of the tumors are not visualized. Iris melanoma may be circumscribed or diffuse. Tapioca iris melanoma is a rare variant of diffuse iris melanoma. Slit-lamp examination, gonioscopy, and ultrasound biomicroscopy (UBM) allow staging of the tumor to guide the most appropriate treatment. Iris melanomas are managed by observation, incisional biopsy, excisional biopsy, or radiotherapy depending upon the size, extent, and visual potential. Iris melanoma tends to be less aggressive in comparison to melanomas in other uveal locations.