Prognostic modelling of number of patients with retinal vein occlusion in anti-VEGF therapy.

PURPOSE: The purpose of the study was to evaluate temporal changes in age- and sex-stratified incidence rates of retinal vein occlusion (RVO) commenced in anti-vascular endothelial growth factor (anti-VEGF) treatment, proportion of patients remaining in active anti-VEGF therapy over time, and to develop a forecasting model for future number of patients with RVO in active anti-VEGF therapy. METHODS: This was a registry-based study of patients with RVO in the Capital Region of Denmark from commenced in anti-VEGF therapy from 1 January 2007 to 30 June 2022. Census data were extracted from Statistics Denmark for incidence rate analyses and forecasting data of future demographics. RESULTS: A total of 2641 patients with RVO were commenced in anti-VEGF therapy, of which 2192 were later discontinued. Number of patients rose dramatically during the first years of introduction of anti-VEGF therapy, after which growth was slower and followed the demographic changes. Trend analyses revealed that the COVID-19 epidemics impacted with fewer referrals and more aggressive discontinuation practices. Annual incidence of RVO in 2012-2021 was 13.1 per 100 000 (95% CI: 12.6-13.6 per 100 000). Proportion of patients with RVO remaining in active anti-VEGF treatment was 55.0%, 40.1%, 30.8% and 12.1% after Years 1, 2, 3 and 8, respectively. According to our forecast, number of patients with RVO in active anti-VEGF therapy will grow slowly but continually at least until year 2035. CONCLUSION: Our study reports incidence rates and provides prognostic modelling of number of patients with RVO in anti-VEGF therapy.

Artificial Intelligence-Based ChatGPT Responses for Patient Questions on Optic Disc Drusen.

INTRODUCTION: Optic disc drusen (ODD) are acellular deposits in the optic nerve head, which are most often benign and asymptomatic. Patients may develop visual field defects and be at increased risk of ischemic co-morbidities. As ODD can be difficult to distinguish from papilledema, patients are at risk of unnecessary clinical workups. Patient information is a key aspect of ODD management. In this study, we explored the accuracy of ChatGPT responses for typical patient questions on ODD. METHODS: Two content experts reached consensus on 20 typical patient questions. We retrieved five separate responses for each question from ChatGPT, totaling 100 responses. Each response was evaluated on a 5-point Likert-scale on accuracy by each content expert in an individual fashion. RESULTS: The two experts were in fair/substantial agreement in the evaluation of responses (Cronbach's alpha: 0.64). Of the 100 responses, 17 were relevant and without any inaccuracies, 78 were relevant and with inaccuracies not being harmful, and five were relevant and with inaccuracies potentially harmful. The lowest accuracy scores were obtained for questions dealing with treatment and prognosis. CONCLUSIONS: ChatGPT often provides relevant answers for patient questions on ODD, but inaccuracies become potentially harmful when questions deal with treatment and prognosis.

Anti-VEGF Treatment of Diabetic Macular Edema in Denmark: Incidence, Burden of Therapy, and Forecasting Analyses.

BACKGROUND: The aim of this study was to analyze demographically stratified incidence rates of patients with diabetic macular edema (DME) commenced in anti-VEGF therapy, to study temporal trends, to report the proportion of patients in active therapy over time, and to develop a model to forecast the future number of patients in active treatment. METHODS: This was a retrospective registry-based study of all patients with DME who received at least one intravitreal anti-VEGF treatment from 1 January 2007 to 30 June 2022. Population data were extracted from Statistics Denmark. RESULTS: This study included 2220 patients with DME who were commenced in anti-VEGF therapy. Demographic analyses revealed higher incidence rates among males than females and among those aged 60-80 years. The number of patients in active treatment followed an exponential decay curve; hence, this was used to mathematically model the number of patients in active therapy. The number of patients in active treatment is expected to stay relatively stable with a minimal increase until the year 2023. CONCLUSIONS: This study provides insight into the practical aspects of the anti-VEGF treatment of DME that allow the planning of adequate health services.

Diagnostic Accuracy of the Amsler Grid Test for Detecting Neovascular Age-Related Macular Degeneration: A Systematic Review and Meta-analysis.

Importance: Patients with nonneovascular age-related macular degeneration (AMD) are encouraged to use the Amsler grid test for self-assessment to facilitate early diagnosis. The test is widely recommended, suggesting a belief that it signals worsening AMD, warranting its use in home monitoring. Objective: To systematically review studies of the diagnostic test accuracy of the Amsler grid in the diagnosis of neovascular AMD and to perform diagnostic test accuracy meta-analyses. Data Sources: A systematic literature search was conducted in 12 databases for relevant titles from database inception until May 7, 2022. Study Selection: Studies included those with groups defined as having (1) neovascular AMD and (2) either healthy eyes or eyes with nonneovascular AMD. The index test was the Amsler grid. The reference standard was ophthalmic examination. After removal of obviously irrelevant reports, 2 authors (J.B. and M.S.) independently screened the remaining references in full text for potential eligibility. Disagreements were resolved by a third author (Y.S.). Data Extraction and Synthesis: Two authors (J.B. and I.P.) independently extracted all data and evaluated quality and applicability of eligible studies using the Quality Assessment of Diagnostic Accuracy Studies 2. Disagreements were resolved by a third author (Y.S.). Main Outcomes and Measures: Sensitivity and specificity of the Amsler grid for detecting neovascular AMD with comparators being either healthy control participants or patients with nonneovascular AMD. Results: Of 523 records screened, 10 studies were included with a total of 1890 eyes (mean participant age ranging from 62 to 83 years). Sensitivity and specificity to diagnose neovascular AMD were 67% (95% CI, 51%-79%) and 99% (95% CI, 85%-100%), respectively, when comparators were healthy control participants and 71% (95% CI, 60%-80%) and 63% (95% CI, 49%-51%), respectively, when control participants were patients with nonneovascular AMD. Overall, potential sources of bias were low across studies. Conclusions and Relevance: Although the Amsler grid is easy and inexpensive to use for detection of metamorphopsia, its sensitivity may be at levels typically not recommended for monitoring. Coupling this lower sensitivity with only moderate specificity to identify neovascular AMD in a population at risk, these findings suggest that such patients typically should be encouraged to undergo ophthalmic examination regularly, regardless of any results of Amsler grid self-assessment.

Assessment of Anterior Uveitis Through Anterior-Segment Optical Coherence Tomography and Artificial Intelligence-Based Image Analyses.

Purpose: The purpose of this study was to develop an automated artificial intelligence (AI) based method to quantify inflammation in the anterior chamber (AC) using anterior-segment optical coherence tomography (AS-OCT) and to explore the correlation between AI assisted AS-OCT based inflammation analyses and clinical grading of anterior uveitis by Standardization of Uveitis Nomenclature (SUN). Methods: A prospective double blinded study of AS-OCT images of 32 eyes of 19 patients acquired by Tomey CASIA-II. OCT images were analyzed with proprietary AI-based software. Anatomic boundaries of the AC were segmented automatically by the AI software and Spearman's rank correlation between parameters related to AC cellular inflammation were calculated. Results: No significant (p = 0.6602) differences were found between the analyzed AC areas between samples of the different SUN grading, suggesting accurate and unbiased border detection/AC segmentation. Segmented AC areas were processed by the AI software and particles within the borders of AC were automatically counted by the software. Statistical analysis found significant (p < 0.001) correlation between clinical SUN grading and AI software detected particle count (Spearman ρ = 0.7077) and particle density (Spearman ρ = 0.7035). Significant (p < 0.001) correlation (Pearson's r = 0.9948) between manually and AI detected particles was found. No significant (p = 0.8080) difference was found between the sizes of the AI detected particles for all studies. Conclusions: AI-based image analysis of AS-OCT slides show significant and independent correlation with clinical SUN assessment. Translational Relevance: Automated AI-based AS-OCT image analysis suggests a noninvasive and quantitative assessment of AC inflammation with clear potential application in early detection and management of anterior uveitis.

Automated artificial intelligence-based system for clinical follow-up of patients with age-related macular degeneration.

PURPOSE: In this study, we investigate the potential of a novel artificial intelligence-based system for autonomous follow-up of patients treated for neovascular age-related macular degeneration (AMD). METHODS: A temporal deep learning model was trained on a data set of 84 489 optical coherence tomography scans from AMD patients to recognize disease activity, and its performance was compared with a published non-temporal model trained on the same data (Acta Ophthalmol, 2021). An autonomous follow-up system was created by augmenting the AI model with deterministic logic to suggest treatment according to the observe-and-plan regimen. To validate the AI-based system, a data set comprising clinical decisions and imaging data from 200 follow-up consultations was collected prospectively. In each case, both the autonomous AI decision and original clinical decision were compared with an expert panel consensus. RESULTS: The temporal AI model proved superior at detecting disease activity compared with the model without temporal input (area under the curve 0.900 (95% CI 0.894-0.906) and 0.857 (95% CI 0.846-0.867) respectively). The AI-based follow-up system could make an autonomous decision in 73% of the cases, 91.8% of which were in agreement with expert consensus. This was on par with the 87.7% agreement rate between decisions made in the clinic and expert consensus (p = 0.33). CONCLUSIONS: The proposed autonomous follow-up system was shown to be safe and compliant with expert consensus on par with clinical practice. The system could in the future ease the pressure on public ophthalmology services from an increasing number of AMD patients.

Detection of oedema on optical coherence tomography images using deep learning model trained on noisy clinical data.

PURPOSE: To meet the demands imposed by the continuing growth of the Age-related macular degeneration (AMD) patient population, automation of follow-ups by detecting retinal oedema using deep learning might be a viable approach. However, preparing and labelling data for training is time consuming. In this study, we investigate the feasibility of training a convolutional neural network (CNN) to accurately detect retinal oedema on optical coherence tomography (OCT) images of AMD patients with labels derived directly from clinical treatment decisions, without extensive preprocessing or relabelling. METHODS: A total of 50 439 OCT images with associated treatment information were retrieved from databases at the Department of Ophthalmology, Rigshospitalet, Copenhagen, Denmark between 01.06.2007 and 01.06.2018. A CNN was trained on the retrieved data with the recorded treatment decisions as labels and validated on a subset of the data relabelled by three ophthalmologists to denote presence of oedema. RESULTS: Moderate inter-grader agreement on presence of oedema in the relabelled data was found (76.4%). Despite different training and validation labels, the CNN performed on par with inter-grader agreement in detecting oedema on OCT images (AUC 0.97, accuracy 90.9%) and previously published models based on relabelled datasets. CONCLUSION: The level of performance shown by the current model might make it valuable in detecting disease activity in automated AMD patient follow-up systems. Our approach demonstrates that high accuracy is not necessarily constrained by incongruent training and validation labels. These results might encourage the use of existing clinical databases for development of deep learning based algorithms without labour-intensive preprocessing in the future.

Modelling and prognostication of growth in the number of patients treated for neovascular age-related macular degeneration.

PURPOSE: The number of patients receiving anti-VEGF therapy has increased rapidly since its introduction in Denmark in 2007, placing an enormous pressure on the public healthcare system. In this study, we attempt to describe this growth and identify the factors driving it. METHODS: Data on treatment of the entire population of neovascular AMD patients in the Capital Region of Denmark between 2007 and 2019 was retrieved. The age and sex standardized incidences of first time treatment and changes in duration of treatment were analysed. RESULTS: The number of patients in active treatment increased from 576 in 2007 to 3684 in 2019. The growth was initially driven by accumulation of patients continuing anti-VEGF therapy for extended periods of time (259 patients/year). As larger numbers of patients began to be discharged, the increase slowed to 181 patients/year in late 2010s, with demographic change becoming the main growth driving factor. The incidence of first treatment increased slightly during the study period, mainly for individuals over 85 years. For patients under 85 years, treatment incidences closely followed neovascular AMD incidences from population studies. The likelihood of remaining in anti-VEGF treatment after the initial injection followed exponential decay curve (t1/2 = 3.6 years). Based on this observation, a model was created to describe the number of patients in active treatment, which accurately described historical data (R2 = 0.999), and forecast a linear growth of 138 patients/year until 2030. CONCLUSION: Treatment incidences and modelling reported in this study might facilitate a more informed and accurate planning of future ophthalmology services.

Donor Endothelial Cell Count Does Not Correlate With Descemet Stripping Automated Endothelial Keratoplasty Transplant Survival After 2 Years of Follow-up.

PURPOSE: To analyze the influence of low endothelial cell density (ECD) of donor cornea tissue, donor age, and sex on the transplant survival rate after Descemet stripping automated endothelial keratoplasty (DSAEK). METHODS: Graft ECD, age, and sex of donors used for DSAEK (n = 1789) during 7 years (2007-2014) in 4 Scandinavian hospitals were assessed for potential association with transplant survival at 2 years of follow-up using a Cox regression model correcting for confounding factors. The data were obtained from The Swedish Cornea Transplant Registry. RESULTS: Transplant failure occurred in 196 patients, with 69 early failures during the first 3 postoperative months, and 127 late secondary failures. Twenty-five of the late secondary failures were due to rejection. Reversible rejections occurred in 67 patients. There was no significant impact of donor age [hazard ratio (HR) 1.0, 95% confidence interval (CI), 0.99-1.02, P = 0.32] or endothelial cell count (HR 1.00, 95% CI, 0.99-1.01, P = 0.3) on the survival rate of DSAEK transplants at 2 years of follow-up. The use of donor grafts with low ECD (<2300 cells/mm) did not influence the survival rate (HR 1.3, 95% CI, 0.76-2.35, P = 0.31). Male donor sex was associated with lower 2-year graft survival (HR 1.5, 95% CI, 1.04-2.28, P = 0.03), but not with rejection events (P = 0.26). CONCLUSIONS: Based on data from The Swedish Cornea Transplant Registry, low donor ECD was not detrimental to graft survival, whereas donor sex seemed to influence the outcome at the end of the 2-year follow-up.

Radiologic evaluation of lumps in the male breast.

BACKGROUND: Gynecomastia has a typical appearance on mammography, and occurs frequently in men. However, imaging is often performed on men with breast lumps to exclude breast cancer, which only comprises 1% of male breast masses. PURPOSE: To assess whether ultrasound and fine needle aspiration cytology (FNAC) are necessary investigations when mammograms show classical gynecomastia. MATERIAL AND METHODS: We have retrospectively collected data on male patients referred for mammography during the period 2011-2013 (a total of 539 patients). All radiological images were re-read, and descriptions of ultrasound images were reviewed. Clinical information supplied with the original referrals was assessed, along with pathology and cytology reports. RESULTS: Among the 539 male patients who underwent mammography, 483 were also examined with ultrasound, and 335 were further evaluated with FNAC. Mammograms showed gynecomastia in 350 patients, and among these subjects ultrasound was performed in 340 (97%), FNAC in 261 (75%), and core biopsies in four (1%) patients. The diagnosis gynecomastia was unchanged in all patients who underwent FNAC or biopsy. Malignant tumors were found in eight patients, six of which were invasive ductal carcinomas. CONCLUSION: In patients with a classical appearance of gynecomastia on mammography, supplemental ultrasound, FNAC, or biopsy is superfluous and contributes to unnecessary costs.

Glycan-related gene expression signatures in breast cancer subtypes; relation to survival.

Alterations in glycan structures are early signs of malignancy and have recently been proposed to be in part a driving force behind malignant transformation. Here, we explore whether differences in expression of genes related to the process of glycosylation exist between breast carcinoma subtypes - and look for their association to clinical parameters. Five expression datasets of 454 invasive breast carcinomas, 31 ductal carcinomas in situ (DCIS), and 79 non-malignant breast tissue samples were analysed. Results were validated in 1960 breast carcinomas. 419 genes encoding glycosylation-related proteins were selected. The DCIS samples appeared expression-wise similar to carcinomas, showing altered gene expression related to glycosaminoglycans (GAGs) and N-glycans when compared to non-malignant samples. In-situ lesions with different aggressiveness potentials demonstrated changes in glycosaminoglycan sulfation and adhesion proteins. Subtype-specific expression patterns revealed down-regulation of genes encoding glycan-binding proteins in the luminal A and B subtypes. Clustering basal-like samples using a consensus list of genes differentially expressed across discovery datasets produced two clusters with significantly differing prognosis in the validation dataset. Finally, our analyses suggest that glycolipids may play an important role in carcinogenesis of breast tumors - as demonstrated by association of B3GNT5 and UGCG genes to patient survival. In conclusion, most glycan-specific changes occur early in the carcinogenic process. We have identified glycan-related alterations specific to breast cancer subtypes including a prognostic signature for two basal-like subgroups. Future research in this area may potentially lead to markers for better prognostication and treatment stratification of breast cancer patients.

Methylation profiling with a panel of cancer related genes: association with estrogen receptor, TP53 mutation status and expression subtypes in sporadic breast cancer.

Breast cancer is a heterogeneous disease that can be divided in subtypes based on histology, gene expression profiles as well as differences in genomic aberrations. Distinct global DNA methylation profiles have been reported in normal breast epithelial cells as well as in breast tumors. However, the influence of the tumor methylome on the previously described subgroups of breast cancer is not fully understood. Here we report the DNA methylation profiles of 80 breast tumors using a panel of 807 cancer related genes interrogating 1505 CpG sites. We identified three major clusters based on the methylation profiles; one consisting of mainly tumors of myoepithelial origin and two other clusters with tumors of predominantly luminal epithelial origin. The clusters were different with respect to estrogen receptor status, TP53 status, ErbB2 status and grade. The most significantly differentially methylated genes including HDAC1, TFF1, OGG1, BMP3, FZD9 and HOXA11 were confirmed by pyrosequencing. Gene Ontology analysis revealed enrichment for genes involved in developmental processes including homeobox domain genes (HOXA9, HOXA11, PAX6, MYBL2, ISL1 and IPF1) and (ETS1, HDAC1, CREBBP, GAS7, SPI1 and TBX1). Extensive correlation to mRNA expression was observed. Pathway analyses identified a significant association with canonical (curated) pathways such as hepatic fibrosis including genes like EGF, NGFR and TNF, dendritic cell maturation and the NF-κB signaling pathway. Our results show that breast tumor expression subtypes harbor major epigenetic differences and tumors with similar gene expression profiles might belong to epigenetically different subtypes. Some of the transcription factors identified, with key roles in differentiation and development might play a role in inducing and maintaining the different phenotypes.

Glycan gene expression signatures in normal and malignant breast tissue; possible role in diagnosis and progression.

Glycosylation is the stepwise procedure of covalent attachment of oligosaccharide chains to proteins or lipids, and alterations in this process have been associated with malignant transformation. Simultaneous analysis of the expression of all glycan-related genes clearly gives the advantage of enabling a comprehensive view of the genetic background of the glycobiological changes in cancer cells. Studies focusing on the expression of the whole glycome have now become possible, which prompted us to review the present knowledge on glycosylation in relation to breast cancer diagnosis and progression, in the light of available expression data from tumors and breast tissue of healthy individuals. We used various data resources to select a set of 419 functionally relevant genes involved in synthesis, degradation and binding of N-linked and O-linked glycans, Lewis antigens, glycosaminoglycans (chondroitin, heparin and keratan sulfate in addition to hyaluronan) and glycosphingolipids. Such glycans are involved in a number of processes relevant to carcinogenesis, including regulation of growth factors/growth factor receptors, cell-cell adhesion and motility as well as immune system modulation. Expression analysis of these glycan-related genes revealed that mRNA levels for many of them differ significantly between normal and malignant breast tissue. An associative analysis of these genes in the context of current knowledge of their function in protein glycosylation and connection(s) to cancer indicated that synthesis, degradation and adhesion mediated by glycans may be altered drastically in mammary carcinomas. Although further analysis is needed to assess how changes in mRNA levels of glycan genes influence a cell's glycome and the precise role that such altered glycan structures play in the pathogenesis of the disease, lessons drawn from this study may help in determining directions for future research in the rapidly-developing field of glycobiology.