Abnormalities of retinal microvascular structure and risk of mortality from ischemic heart disease and stroke

Retinal Vascular Signs and Cerebrovascular Diseases

BACKGROUND

Cerebrovascular disease (CeVD), including stroke, is a leading cause of death globally. The retina is an extension of the cerebrum, sharing embryological and vascular pathways. The association between different retinal signs and CeVD has been extensively evaluated. In this review, we summarize recent studies which have examined this association.

EVIDENCE ACQUISITION

We searched 6 databases through July 2019 for studies evaluating the link between retinal vascular signs and diseases with CeVD. CeVD was classified into 2 groups: clinical CeVD (including clinical stroke, silent cerebral infarction, cerebral hemorrhage, and stroke mortality), and sub-clinical CeVD (including MRI-defined lacunar infarct and white matter lesions [WMLs]). Retinal vascular signs were classified into 3 groups: classic hypertensive retinopathy (including retinal microaneurysms, retinal microhemorrhage, focal/generalized arteriolar narrowing, cotton-wool spots, and arteriovenous nicking), clinical retinal diseases (including diabetic retinopathy [DR], age-related macular degeneration [AMD], retinal vein occlusion, retinal artery occlusion [RAO], and retinal emboli), and retinal vascular imaging measures (including retinal vessel diameter and geometry). We also examined emerging retinal vascular imaging measures and the use of artificial intelligence (AI) deep learning (DL) techniques.

RESULTS

Hypertensive retinopathy signs were consistently associated with clinical CeVD and subclinical CeVD subtypes including subclinical cerebral large artery infarction, lacunar infarction, and WMLs. Some clinical retinal diseases such as DR, retinal arterial and venous occlusion, and transient monocular vision loss are consistently associated with clinical CeVD. There is an increased risk of recurrent stroke immediately after RAO. Less consistent associations are seen with AMD. Retinal vascular imaging using computer assisted, semi-automated software to measure retinal vascular caliber and other parameters (tortuosity, fractal dimension, and branching angle) has shown strong associations to clinical and subclinical CeVD. Other new retinal vascular imaging techniques (dynamic retinal vessel analysis, adaptive optics, and optical coherence tomography angiography) are emerging technologies in this field. Application of AI-DL is expected to detect subclinical retinal changes and discrete retinal features in predicting systemic conditions including CeVD.

CONCLUSIONS

There is extensive and increasing evidence that a range of retinal vascular signs and disease are closely linked to CeVD, including subclinical and clinical CeVD. New technology including AI-DL will allow further translation to clinical utilization.

Cigarette Smoking Is Related to Endothelial Dysfunction of Resistance, but Not Conduit Arteries in the General Population-Results From the Gutenberg Health Study

Aims: Cigarette smoking is one of the most complex and least understood cardiovascular risk factors. Importantly, differences in the tobacco-related pathophysiology of endothelial dysfunction, an early event in atherogenesis, between circulatory beds remain elusive. Therefore, this study evaluated how smoking impacts endothelial function of conduit and resistance arteries in a large population-based cohort.

Methods and results: 15,010 participants (aged 35–74 years) of the Gutenberg Health Study were examined at baseline from 2007 to 2012. Smoking status, pack-years of smoking, and years since quitting smoking were assessed by a computer-assisted interview. Endothelial function of conduit and resistance arteries was determined by flow-mediated dilation (FMD) of the brachial artery, reactive hyperemia index (RHI) using peripheral arterial tonometry, as well as by reflection index (RI) derived from digital photoplethysmography, respectively. Among all subjects, 45.8% had never smoked, 34.7% were former smokers, and 19.4% were current smokers. Mean cumulative smoking exposure was 22.1 ± 18.1 pack-years in current smokers and mean years since quitting was 18.9 ± 12.7 in former smokers. In multivariable linear regression models adjusted for typical confounders, smoking status, pack-years of smoking, and years since quitting smoking were independently associated with RHI and RI, while no association was found for FMD. Overall, no clear dose-dependent associations were observed between variables, whereby higher exposure tended to be associated with pronounced resistance artery endothelial dysfunction.

Conclusions: Cigarette smoking is associated with altered endothelial function of resistance, but not conduit arteries. The present results suggest that smoking-induced endothelial dysfunction in different circulatory beds may exhibit a differential picture.

Retinal Vessel Phenotype in Patients with a History of Retinal Vein Occlusion

INTRODUCTION

The aim of the study was to estimate the phenotype of retinal vessels using central retinal artery equivalent (CRAE), central retinal vein equivalent (CRVE), tortuosity, and fractal analysis in the unaffected contralateral eye of patients with central or branch retinal vein occlusion (CRVO or BRVO).

METHODS

Thirty-four patients suffering from CRVO, 15 suffering from BRVO, and 49 controlled matched subjects had a fundus image analyzed using the VAMPIRE software. The intraclass correlation coefficient and a Bland-Altman plot were done for the reproducibility study.

RESULTS

There was a lack of evidence of difference between the control group and the CRVO group for CRAE (p = 0.06), CRVE (p = 0.3), and arterio-venule ratio (AVR, p = 0.6). Contralateral eyes of CRVO exhibited a significantly higher arterial and minimum arterial tortuosity values (p = 0.012), as compared with control eyes. Contralateral eyes of patients with a history of BRVO had a significantly higher CRAE (p = 0.02), AVR (p = 0.006), and minimal arterial tortuosity (p = 0.05). Fractal analysis showed that contralateral eyes of BRVO had higher values of fractal parameters (D0a, p = 0.005).

CONCLUSION

This study suggests that CVRO or BRVO is not triggered by the same retinal vascular phenotypes in the contralateral eye. The morphology of retinal vasculature may be associated with the occurrence of RVO, independently of known risk factors.

Capillary vascular density in the retina of hypertensive patients is associated with a non-dipping pattern independent of mean ambulatory blood pressure

BACKGROUND

A nocturnal non-dipping pattern has been associated with hypertension-mediated organ damage (HMOD), morbidity and mortality. Retinal imaging through application of modern technologies including optical coherence tomography angiography (OCT-A) can provide detailed insights into early vascular damage. In this observational study, we investigated the relationship of microscopic vascular density in the retina measured with OCT-A and nocturnal blood pressure (BP) dipping.

METHODS

Retinal OCT-A and ambulatory BP monitoring (ABPM) data prospectively obtained from 142 patients referred to a tertiary hypertension clinic were analysed with regression models for associations between BP night-time dipping and retinal capillary vascular density in three different zones around the fovea.

RESULTS

More pronounced nocturnal SBP and DBP dipping was significantly associated with increased vascular density in the central foveal area of the retina. These associations were robust to adjustment for other available risk factors including mean daytime BP. Parafoveal and whole image vascular density did not show equivalent significant associations with nocturnal BP dipping. The results were reproducible when assessed in a subgroup of patients who had concomitant type 2 diabetes.

CONCLUSION

Foveal vascular density was associated with the nocturnal BP dipping pattern in hypertensive patients. These associations were robust to adjustment of relevant factors such as daytime BP. Our findings highlight the importance of nocturnal BP features reflected in ambulatory BP monitoring in the assessment of HMOD. Whether routine assessment of retinal damage markers may improve risk management of hypertensive patients remains to be determined.

Hypertrophic remodelling of retinal arterioles in patients with congestive heart failure

AIMS

Analysis of microvascular parameters in the retinal circulation-known to reflect those in the systemic circulation-allows us to differentiate between eutrophic and hypertrophic remodelling of small arteries. This study aimed to examine microvascular changes in patients with congestive heart failure (CHF) and reduced as well as mid-range ejection fraction.

METHODS AND RESULTS

Forty subjects with CHF underwent measurement of retinal capillary flow (RCF), wall-to-lumen ratio (WLR), vessel and lumen diameter, wall thickness, and wall cross-sectional area (WCSA) of retinal arterioles of the right eye by scanning laser Doppler flowmetry (SLDF). Applying a matched pair approach, we compared this group with reference values of age-matched controls from a random sample in the population of Pilsen, Czech Republic. There was no significant difference in RCF and WLR between the groups (RCF: P = 0.513; WLR: P = 0.106). In contrast, wall thickness and WCSA, indicators of hypertrophic remodelling, were higher in CHF subjects (WT: 15.0 ± 4.2 vs. 12.7 ± 4.2 μm, P = 0.021; WCSA: 4437.6 ± 1314.5 vs. 3615.9 ± 1567.8 μm² , P = 0.014). Similarly, vessel (109.4 ± 11.1 vs. 100.5 ± 14.4 μm, P = 0.002) and lumen diameter (79.0 ± 7.9 vs. 75.2 ± 8.5 μm, P = 0.009) were increased in CHF.

CONCLUSIONS

In CHF subjects, we observed hypertrophic remodelling of retinal arterioles indicative of similar changes of small resistance arteries in the systemic circulation. Microvascular structure and function assessed by SLDF may thereby represent a useful, non-invasive method for monitoring of microvascular damage in patients with CHF and may offer innovative treatment targets for new CHF therapies.

Retinal microvascular parameters are not significantly associated with mild cognitive impairment in the Northern Ireland Cohort for the Longitudinal Study of Ageing

BACKGROUND

The retinal and cerebral microvasculature share similar embryological origins and physiological characteristics. Improved imaging technologies provide opportunistic non-invasive assessment of retinal microvascular parameters (RMPs) against cognitive outcomes. We evaluated baseline measures for associations between RMPs and mild cognitive impairment (MCI) from participants of the Northern Ireland Cohort for the Longitudinal Study of Ageing (NICOLA).

METHODS

RMPs (central retinal arteriolar / venular equivalents, arteriole to venular ratio, fractal dimension and tortuosity) were measured from optic disc centred fundus images and analysed using semi-automated software. Associations between RMPs and MCI were assessed by multivariable logistic regression with adjustment for potential confounders including age, sex, alcohol consumption, smoking status, educational attainment, physical activity, cardiovascular disease (CVD), hypertension, mean arterial blood pressure, triglycerides, diabetes, body mass index, and high density lipoprotein levels. P < 0.05 was considered statistically significant.

RESULTS

Data were available for 1431 participants, of which 156 (10.9%) were classified with MCI defined by a Montreal Cognitive Assessment (MoCA) score ≤ 26, with subjective cognitive decline, in the absence of depression or problems with activities of daily living. Participants had a mean age of 62.4 ± 8.5 yrs. and 52% were female. As expected, individuals with MCI had a lower MoCA score than those without (23.5 ± 2.6 versus 26.3 ± 2.7, respectively), were more likely to be female, have a lower level of educational attainment, be less physically active, more likely to have CVD, have higher levels of triglycerides and lower levels of high density lipoprotein. No significant associations between RMPs and MCI were detected in unadjusted, minimally adjusted or fully adjusted regression models or subsequent sensitivity analyses.

CONCLUSION

Previous studies have reported both increased retinal venular calibre and reduced fractal dimension in association with mild cognitive impairment. Our study failed to detect any associations between RMPs and those individuals at an early stage of cognitive loss in an older community-based cohort.

Differential artery-vein analysis in quantitative retinal imaging: a review

Quantitative retinal imaging is essential for eye disease detection, staging classification, and treatment assessment. It is known that different eye diseases or severity stages can affect the artery and vein systems in different ways. Therefore, differential artery-vein (AV) analysis can improve the performance of quantitative retinal imaging. In this article, we provide a brief summary of technical rationales and clinical applications of differential AV analysis in fundus photography, optical coherence tomography (OCT), and OCT angiography (OCTA).

Retinal Vascular Fractal Dimensions and Their Association with Macrovascular Cardiac Disease

INTRODUCTION

As the only part of the human vasculature, the retina is available for direct, noninvasive inspection. Retinal vascular fractal dimension (DF) is a method to measure the structure of the retinal vascular tree, with higher noninteger values between 1 and 2 representing a more complex and dense retinal vasculature. Retinal vascular structure has been associated with a variety of systemic diseases, and this study examined the association of DF and macrovascular cardiac disease in a case-control design.

METHODS

Retinal fundus photos were captured with Topcon TRC-50X in 38 persons that had coronary artery bypass grafting (CABG, cases) and 37 cardiovascular healthy controls. The semiautomatic software VAMPIRE was used to measure retinal DF.

RESULTS

Patients with CABG had lower DF of the retinal main venular vessels compared to the control group (1.15 vs. 1.18, p = 0.01). In a multivariable regression model adjusted for gender and age, eyes in the fourth quartile with higher DF were less likely to have CABG compared to patients in the first (OR, 7.20; 95% confidence interval: 1.63-31.86; p = 0.009) and second (OR, 8.25; 95% confidence interval: 1.70-40.01; p = 0.009) quartiles.

CONCLUSIONS

This study demonstrates that lower complexity of main venular vessels associates with higher risk of having CABG. The research supports the hypothesis that the retinal vascular structure can be used to assess nonocular macrovascular disease.

High-salt intake affects retinal vascular tortuosity in healthy males: an exploratory randomized cross-over trial

The retinal microcirculation is increasingly receiving credit as a relatively easily accessible microcirculatory bed that correlates closely with clinical cardiovascular outcomes. The effect of high salt (NaCl) intake on the retinal microcirculation is currently unknown. Therefore, we performed an exploratory randomized cross-over dietary intervention study in 18 healthy males. All subjects adhered to a two-week high-salt diet and low-salt diet, in randomized order, after which fundus photographs were taken and assessed using a semi-automated computer-assisted program (SIVA, version 4.0). Outcome parameters involved retinal venular and arteriolar tortuosity, vessel diameter, branching angle and fractal dimension. At baseline, participants had a mean (SD) age of 29.8 (4.4) years and blood pressure of 117 (9)/73 (5) mmHg. Overall, high-salt diet significantly increased venular tortuosity (12.2%, p = 0.001). Other retinal parameters were not significantly different between diets. Changes in arteriolar tortuosity correlated with changes in ambulatory systolic blood pressure (r = - 0.513; p = 0.04). In conclusion, high-salt diet increases retinal venular tortuosity, and salt-induced increases in ambulatory systolic blood pressure associate with decreases in retinal arteriolar tortuosity. Besides potential eye-specific consequences, both phenomena have previously been associated with hypertension and other cardiovascular risk factors, underlining the deleterious microcirculatory effects of high salt intake.

Retinal Microvascular Signs as Screening and Prognostic Factors for Cardiac Disease: A Systematic Review of Current Evidence

The substantial burden of heart disease promotes an interest in new ways of screening for early disease diagnosis, especially by means of noninvasive imaging. Increasing evidence for association between retinal microvascular signs and heart disease prompted us to systematically investigate the relevant current literature on the subject. We scrutinized the current literature by searching PubMed and Embase databases from 2000 to 2020 for clinical studies of the association between retinal microvascular signs and prevalent or incident heart disease in humans. Following exclusions, we extracted the relevant data from 42 publications (comprising 14 prospective, 26 cross-sectional, and 2 retrospective studies). Our search yielded significant associations between retinal vascular changes, including diameter, tortuosity, and branching, and various cardiac diseases, including acute coronary syndrome, coronary artery disease, heart failure, and conduction abnormalities. The findings of our research suggest that the retinal microvasculature can provide essential data about concurrent cardiac disease status and predict future risk of cardiac-related events.

Identifying Microvascular and Neural Parameters Related to the Severity of Diabetic Retinopathy Using Optical Coherence Tomography Angiography

Purpose

To identify microvascular and neural parameters related to the severity of diabetic retinopathy (DR) by using optical coherence tomography angiography in patients with type 2 diabetes.

Methods

This cross-sectional study included 110 eyes (63 patients) with no DR, 46 eyes (33 patients) with mild nonproliferative DR, 36 eyes (23 patients) with moderate nonproliferative DR, 36 eyes (22 patients) with severe nonproliferative DR, and 31 eyes (19 patients) with proliferative DR. The optical coherence tomography angiography images were processed to quantify the foveal avascular zone parameters, macular vessel density (VD), retinal thickness, peripapillary VD, retinal nerve fiber layer thickness, and ganglion cell complex thickness. A LASSO-based continuation ratio model was used to select the most clinically relevant parameters for predicting the stage of DR.

Results

The regression model identified a set of regional parameters for each scanning pattern that identified the DR severity, including foveal avascular zone perimeter; FD-300; temporal perifoveal superficial capillary plexus VD and retinal thickness; temporal and nasal parafoveal deep capillary plexus VD; peripapillary VD in the temporal superior, nasal inferior, and temporal inferior sectors; temporal superior and nasal inferior retinal nerve fiber layer thickness; ganglion cell complex thickness; and FLV, which changed significantly with the progression of DR. Furthermore, two combined blocks exhibited different sensitive parameters to differentiate between the groups based on DR severity. Similar results were obtained in eyes without diabetic macular edema.

Conclusions

We identified microvascular and neural parameters related to the severity of DR using optical coherence tomography angiography, suggesting their potential clinical application for better screening and staging of DR.

Microvascular changes of the retina in ankylosing spondylitis, and the association with cardiovascular disease - the eye for a heart study

OBJECTIVE

Ankylosing spondylitis (AS) is associated with an increased risk of cardiovascular disease (CVD). Microvasculature changes can precede overt CVD, but have been studied poorly in AS. The retinal vasculature is easily accessible and changes are associated with CVD (e.g. arteriolar narrowing, venular widening, loss of tortuosity). This proof of concept study compared the retinal microvasculature of AS patients with healthy controls, and the influence of gender.

METHODS

Cross-sectional case-control study comparing AS patients with healthy controls. Main inclusion criteria were: age 50-75 years, no diabetes mellitus and, for AS, fulfillment of the modified New York criteria. All subjects underwent fundus photography, analyzed with Singapore I Vessel Assessment software, and Optical Coherence Tomography Angiography (OCTA). Subjects were compared with generalized estimating equations (GEE). Multivariable analyses were adjusted for demographics and cardiovascular risk, and stratified for gender.

RESULTS

Fifty-nine AS patients and 105 controls were included (50% women). Controls were significantly older than patients (68 versus 60, p<0.01), but did not differ in cardiovascular profile. Patients had a lower retinal arteriolar tortuosity (β ̶-0.1, 95%CI [-0.2; -0.01], p = 0.02), and higher vessel density (β 0.5, 95% CI [0.1; 0.9], p = 0.02). In addition, male AS patients showed a lower arteriovenular ratio compared to male controls (β -0.03, p = 0.04, 95%CI [-0.05; -0.001]). There were no differences found between women with and without AS.

CONCLUSION

This study detected several retinal microvascular changes, in AS patients compared to controls, which have been associated with CVD. Retinal imaging might be an interesting tool for future CVD screening.

Impending central retinal vein occlusion in patient with idiopathic cutaneous leukocytoclastic vasculitis

Purpose

To report a case of impending central retinal vein occlusion (CRVO) associated with idiopathic cutaneous leukocytoclastic vasculitis (LCV) that demonstrated significant resolution following treatment with intravenous (IV) methylprednisolone.

Observations

A 27-year-old man presented to a tertiary Uveitis Clinic with a five-day history of blurry vision in the right eye (OD). He had a history of a purpuric rash and arthralgias five years ago and a biopsy-confirmed diagnosis of LCV controlled with colchicine two years ago in India. Recently, he presented with a recurrent rash and severe abdominal pain. After being evaluated by rheumatology and gastroenterology, he was placed on Helicobacter pylori treatment and high dose oral prednisone, which improved his skin and gastrointestinal symptoms. At the first ophthalmic exam, his systemic findings included lower extremity purpura. His best-corrected visual acuity (BCVA) was 20/20 in both eyes (OU). Slit-lamp examination revealed no cells or flare in OU. Dilated fundus exam showed mild enlarged, tortuous veins, optic nerve hemorrhage, and intraretinal hemorrhages temporal to the macula in OD. Spectral-domain optical coherence tomography (SD-OCT) demonstrated multiple hyper-reflective, plaque-like lesions involving the inner nuclear layer, consistent with paracentral acute middle maculopathy (PAMM). The patient was diagnosed with impending central retinal vein occlusion (CRVO) in OD. Laboratory evaluations were unremarkable. Aspirin was initially started for the patient but was later held due to the worsening of retinal hemorrhage and retinal vein tortuosity at the one-week follow-up. The patient then received three doses of intravenous methylprednisolone, followed by systemic oral prednisone and mycophenolate mofetil. One month later, retinal hemorrhages, venous stasis, and skin manifestations resolved.

Conclusion and importance

Ocular involvement in LCV is rare and may present with different manifestations. The index case is the first report of impending CRVO in a patient with idiopathic LCV and without any other known risk factors for CRVO. Our report not only describes the unique course of LCV-related ocular involvement, but also introduces and underscores a potentially effective therapeutic plan.

Association of Cardiovascular Mortality and Deep Learning-Funduscopic Atherosclerosis Score derived from Retinal Fundus Images

PURPOSE

The prediction of atherosclerosis using retinal fundus images and deep learning has not been shown possible. The purpose of this study was to develop a deep learning model which predicted atherosclerosis by using retinal fundus images and to verify its clinical implications by conducting a retrospective cohort analysis.

DESIGN

Retrospective cohort study.

METHODS

The database at the Health Promotion Center of Seoul National University Hospital (HPC-SNUH) was used. The deep learning model was trained using 15,408 images to predict carotid artery atherosclerosis, which was named the deep-learning funduscopic atherosclerosis score (DL-FAS). A retrospective cohort was constructed of participants 30-80 years old who had completed elective health examinations at HPC-SNUH. Using DL-FAS as the main exposure, participants were followed for the primary outcome of death due to CVD until Dec. 31, 2017.

RESULTS

For predicting carotid artery atherosclerosis among subjects, the model achieved an area under receiver operating curve (AUROC) and area under the precision-recall curve (AUPRC), accuracy, sensitivity, specificity, positive and negative predictive values of 0.713, 0.569, 0.583, 0.891, 0.404, 0.465, and 0.865 respectively. The cohort consisted of 32,227 participants, 78 cardiovascular disease (CVD) deaths, and 7.6-year median follow-up visits. Those with DL-FAS greater than 0.66 had an increased risk of CVD deaths compared to those with DL-FAS <0.33 (hazard ratio: 8.33; 95% confidence interval [CI], 3.16-24.7). Risk association was significant among intermediate and high Framingham risk score (FRS) subgroups. The DL-FAS improved the concordance by 0.0266 (95% CI, 0.0043-0.0489) over the FRS-only model. The relative integrated discrimination index was 20.45% and net reclassification index was 29.5%.

CONCLUSIONS

A deep learning model was developed which could predict atherosclerosis from retinal fundus images. The resulting DL-FAS was an independent predictor of CVD deaths when adjusted for FRS and added predictive value over FRS.

Comprehensive retinal vascular measurements: a novel association with renal function in type 2 diabetic patients in China

To examine the association between various retinal vascular measurements and microalbuminuria in patients with type 2 diabetes in a northwestern China study. Data from 911 patients with type 2 diabetes were analyzed. Novel retinal vascular measurements from the whole vascular tree were extracted using a validated fully automatic computer program. Retinal vascular measurements were analyzed continuously and categorically for associations with microalbuminuria using multiple logistic regressions, adjusted for related variables. In logistic regression adjusting for multiple variables, microalbuminuria was associated with smaller peripheral arteriolar caliber, larger peripheral venular caliber, larger arteriolar tortuosity, and smaller arteriolar fractal dimension (p = 0.028, p < 0.001, p = 0.038, p = 0.035, respectively). In further categorical analyses, microalbuminuria was related to smaller peripheral arteriolar caliber [T1 vs. T3: odds ratio (OR) 2.029; 95% confidence interval (CI) 1.186–3.473], larger peripheral venular caliber (T1 vs. T3: OR 0.609; 95% CI 0.362–1.024), and smaller arteriolar fractal dimension (T1 vs. T3: OR 1.659; 95% CI 1.028–2.675). Microalbuminuria in type 2 diabetes is associated with both retinal vascular caliber and geometry. These noninvasive vascular measurements serve as potential preclinical markers to identify populations at high risk of early kidney disease in the course of diabetes.

Morphological Analysis of Retinal Microvasculature to Improve Understanding of Retinal Hemorrhage Mechanics in Infants

Purpose

In this experimental study, we quantify retinal microvasculature morphological features with depth, region, and age in immature and mature ovine eyes. These data identify morphological vulnerabilities in young eyes to inform the mechanics of retinal hemorrhage in children.

Methods

Retinal specimens from the equator and posterior pole of preterm (n = 4) and adult (n = 9) sheep were imaged using confocal microscopy. Vessel segment length, diameter, angular asymmetry, tortuosity, and branch points were quantified using a custom image segmentation code. Significant differences were identified through two-way ANOVAs and correlation analyses.

Results

Vessel segment lengths were significantly shorter in immature eyes compared to adults (P < 0.003) and were significantly shorter at increasing depths in the immature retina (P < 0.04). Tortuosity significantly increased with depth, regardless of age (P < 0.05). These data suggest a potential vulnerability of vasculature in the deeper retinal layers, particularly in immature eyes. Preterm retina had significantly more branch points than adult retina in both the posterior pole and equator, and the number increased significantly with depth (P < 0.001).

Conclusions

The increased branch points and decreased segment lengths in immature microvasculature suggest that infants will experience greater stress and strain during traumatic loading compared to adults. The increased morphological vulnerability of the immature microvasculature in the deeper layers of the retina suggest that intraretinal hemorrhages have a greater likelihood of occurring from trauma compared to preretinal hemorrhages. The morphological features captured in this study lay the foundation to explore the mechanics of retinal hemorrhage in infants and identify vulnerabilities that explain patterns of retinal hemorrhage in infants.

The Discriminative Efficacy of Retinal Characteristics on Two Traditional Chinese Syndromes in Association with Ischemic Stroke

We aimed to investigate the efficacy of an objective method using AI-based retinal characteristic analysis to automatically differentiate between two traditional Chinese syndromes that are associated with ischemic stroke. Inpatient clinical and retinal data were retrospectively retrieved from the archive of our hospital. Patients diagnosed with cerebral infarction in the department of acupuncture and moxibustion between 2014 and 2018 were examined. Of these, the patients with Qi deficiency blood stasis syndrome (QDBS) and phlegm stasis in channels (PSIC) syndrome were selected. Those without retinal photos were excluded. To measure and analyze the patients' retinal vessel characteristics, we applied a patented AI-assisted automated retinal image analysis system developed by the Chinese University of Hong Kong. The demographic, clinical, and retinal information was compared between the QDBS and PSIC patients. The t-test and chi-squared test were used to analyze continuous data and categorical data, respectively. All the selected clinical information and retinal vessel measures were used to develop different discriminative models for QDBS and PSIC using logistic regression. Discriminative efficacy and model performances were evaluated by plotting a receiver operating characteristic curve. As compared to QDBS, the PSIC patients had a lower incidence of insomnia problems (46% versus 29% respectively, p=0.023) and a higher tortuosity index (0.45 ± 0.07 versus 0.47 ± 0.07, p=0.027). Moreover, the area under the curve of the logistic model showed that its discriminative efficacy based on both retinal and clinical characteristics was 86.7%, which was better than the model that employed retinal or clinical characteristics individually. Thus, the discriminative model using AI-assisted retinal characteristic analysis showed statistically significantly better performance in QDBS and PSIC syndrome differentiation among stroke patients. Therefore, we concluded that retinal characteristics added value to the clinical differentiation between QDBS and PSIC.

The Relationship Between Ambient Atmospheric Fine Particulate Matter (PM2.5) and Glaucoma in a Large Community Cohort

Purpose

Glaucoma is more common in urban populations than in others. Ninety percent of the world's population are exposed to air pollution above World Health Organization (WHO) recommended limits. Few studies have examined the association between air pollution and glaucoma.

Methods

Questionnaire data, ophthalmic measures, and ambient residential area air quality data for 111,370 UK Biobank participants were analyzed. Particulate matter with an aerodynamic diameter < 2.5 μm (PM2.5) was selected as the air quality exposure of interest. Eye measures included self-reported glaucoma, intraocular pressure (IOP), and average thickness of macular ganglion cell-inner plexiform layer (GCIPL) across nine Early Treatment Diabetic Retinopathy Study (ETDRS) retinal subfields as obtained from spectral-domain optical coherence tomography. We examined the associations of PM2.5 concentration with self-reported glaucoma, IOP, and GCIPL.

Results

Participants resident in areas with higher PM2.5 concentration were more likely to report a diagnosis of glaucoma (odds ratio = 1.06, 95% confidence interval [CI] = 1.01-1.12, per interquartile range [IQR] increase P = 0.02). Higher PM2.5 concentration was also associated with thinner GCIPL (β = -0.56 μm, 95% CI = -0.63 to -0.49, per IQR increase, P = 1.2 × 10-53). A dose-response relationship was observed between higher levels of PM2.5 and thinner GCIPL (P < 0.001). There was no clinically relevant relationship between PM2.5 concentration and IOP.

Conclusions

Greater exposure to PM2.5 is associated with both self-reported glaucoma and adverse structural characteristics of the disease. The absence of an association between PM2.5 and IOP suggests the relationship may occur through a non-pressure-dependent mechanism, possibly neurotoxic and/or vascular effects.

Comparison of Prediction Models based on Risk Factors and Retinal Characteristics Associated with Recurrence One Year after Ischemic Stroke

OBJECTIVES

To develop risk estimation models for 1-year ischemic stroke recurrence using clinical risk factors and retinal characteristics.

METHODS

From June 2017 to January 2019, 332 patients with first-ever ischemic stroke were enrolled and followed up in the Shenzhen Traditional Chinese Medicine hospital in China. The primary endpoint was defined as fatal or recurrent stroke after 1 year of the index stroke. Clinical risk factors and retinal characteristics were identified by multivariate logistic models.

RESULTS

The multivariate logistic model with only clinical risk factors showed that Cerebral Atherosclerosis (OR 1.68, 95%CI: 1.000-2.81), white matter lesions (OR 3.61, 95%CI: 2.18-5.98), and Cardiac disease (OR 1.88, 95%CI: 1.02-3.46) were statistically significantly associated with higher stroke recurrence risk. The sensitivity and specificity of this model were 69.1% and 68.4% respectively. The multivariate logistic model with only retinal characteristics showed that central retinal venule equivalent (OR .34, 95%CI: .14-.83), hemorrhage (OR .6, 95%CI: .41-.88), exudate (OR 1.64, 95%CI: 1.16-2.32), central retinal artery equivalent (OR 2.95, 95%CI: 1.23-7.08), and Aangle (OR 0.8, 95%CI: .61-1.004) were statistically significantly associated with stroke recurrence. The sensitivity and specificity of the model were 62.0% and 64.4% respectively. The multivariate logistic model with both clinical risk factors and retinal characteristics showed that cerebral atherosclerosis (OR 1.74, 95%CI: 1.020-2.981), white matter lesions (OR 3.65, 95%CI: 2.17-6.13), cardiac disease (OR 1.99, 95%CI: 1.06-3.74), hemorrhage (OR .68, 95%CI: .49-.96), exudate (OR 1.65, 95%CI: 1.16-2.36) were independent risk factors of stroke recurrence. The sensitivity and specificity of the model were 72.5% and 70.7% respectively.

CONCLUSIONS

Combining the traditional risk factors of stroke with the retinal vessels characteristics to establish the recurrent cerebral infarction prediction model may improve the accuracy of the prediction.

Optical coherence tomography angiography reveals progressive worsening of retinal vascular geometry in diabetic retinopathy and improved geometry after panretinal photocoagulation

Purpose

To quantify vessel tortuosity and fractal dimension of the superficial capillary plexus (SCP) of the macula in different stages of diabetic retinopathy (DR), and following panretinal photocoagulation (PRP) using optical coherence tomography angiography (OCTA).

Methods

75 eyes of 75 subjects were divided into five groups; healthy controls, diabetes with no clinical DR, non-proliferative diabetic retinopathy (NPDR), proliferative diabetic retinopathy (PDR) and patients who received PRP for PDR (PDR+PRP).For vessel tortuosity, SCP slabs from 3x3 mm macular OCTA scans were processed using imageJ (NIH, USA), where large perifoveal vessels were traced and their length was measured with tortuosity calculated as the ratio between the actual length and the straight Euclidean length. For fractal dimension, SCP slabs were processed and imported to Fractalyse (ThéMA, France), where box-counting analyses produced fractal dimension values.

Results

We found a significant difference in vessel tortuosity and fractal dimension between the five groups (one-way ANOVA, p < 0.001both). NPDR and PDR had significantly more tortuous vessels and lower fractal dimension compared to healthy controls (Tukey HSD: p = 0.02, 0.015,0.015 and <0.001, respectively). Fractal dimension was also significantly lower in NPDR and PDR compared to eyes with no clinical DR (p <0.001 both), and in PDR compared to NPDR (p = 0.014). Following PRP, vessel tortuosity was significantly lower and fractal dimension was higher in PDR+PRP compared to PDR (p = 0.001 and 0.031, respectively).

Conclusions

We used macular OCTA scans to demonstrate significantly higher perifoveal large vessel tortuosity, and lower fractal dimension in NPDR and PDR compared to healthy controls. Vessel tortuosity shows more dramatic normalization than fractal dimension and could be explored as a sensitive marker for successful PRP.

Computational assessment of the retinal vascular tortuosity integrating domain-related information

The retinal vascular tortuosity presents a valuable potential as a clinical biomarker of many relevant vascular and systemic diseases. Commonly, the existent approaches face the tortuosity quantification by means of fully mathematical representations of the vessel segments. However, the specialists, based on their diagnostic experience, commonly analyze additional domain-related information that is not represented in these mathematical metrics of reference. In this work, we propose a novel computational tortuosity metric that outperforms the mathematical metrics of reference also incorporating anatomical properties of the fundus image such as the distinction between arteries and veins, the distance to the optic disc, the distance to the fovea, and the vessel caliber. The evaluation of its prognostic performance shows that the integration of the anatomical factors provides an accurate tortuosity assessment that is more adjusted to the specialists’ perception.

Cardiovascular health and retinal microvascular geometry in Australian 11-12 year-olds

Traditional retinal microvascular parameters (smaller arteriolar and greater venular caliber) are associated with cardiovascular risk factors, pre-clinical vascular phenotypes and clinical cardiovascular events in adults. Although novel retinal microvascular geometric parameters showed analogous associations in adults, less is known whether these parameters are associated with cardiovascular health from childhood. In a population-based cross-sectional study in children (n = 1126, mean age 11.4 years, 50.3% girls), we examined associations of cardiovascular risk factors and pre-clinical arterial phenotypes with retinal geometric parameters. Cardiovascular parameters included body mass index (BMI), an inflammatory marker (GlycA), low-density lipoprotein and high-density lipoprotein (HDL) cholesterol, systolic (SBP) and diastolic blood pressure, large artery functional (pulse wave velocity, PWV and carotid arterial elasticity) and structural (carotid intima-media thickness) phenotypes. Retinal geometric parameters (fractal dimension (D_f) and tortuosity) were quantified from retinal images. Multivariable regression models were performed and adjusted for potential confounders. Higher values for BMI, SBP and PWV showed weak associations with lower (i.e. worse) arteriolar but not venular D_f (standardized mean difference (SMD) ranging from -0.07 to -0.09, 95% CIs -0.15 to -0.01). Higher HDL was associated with greater arteriolar D_f (SMD 0.07, 95% CI 0.01 to 0.13). Only higher SBP was associated with higher (i.e. worse) arteriolar but not venular tortuosity (SMD 0.09, 95% CI 0.02 to 0.16). In generally healthy children, some risk factors and pre-clinical arterial phenotypes show small associations with retinal geometric parameters. In childhood, emerging relationships between microvascular parameters and cardiometabolic risk may be better described by retinal vascular caliber than by geometric parameters.

Retinal vascular fractal dimension in bipolar disorder and schizophrenia

BACKGROUND

Bipolar disorder (BD) and schizophrenia (SCZ), are associated with greater vascular co-morbidities and adverse vascular events. Owing to shared developmental origins and morphology, retinal vasculature is a proxy assessment measure of the cerebral vasculature. Although retinal vascular fractal dimension (D_f), a measure of vascular geometry and complexity of branching, has been shown to be directly associated with cerebrovascular pathology, it has not been examined in SCZ and BD.

METHODS

We studied 277 participants (92 healthy volunteers, 98 SCZ, and 87 BD) from 18 to 50 years of age. Images were acquired by trained personnel using a non-mydriatic fundus camera and the retinal vascular D_f was calculated by the box-counting method using an automated algorithm. The average D_f across the left and right eyes were calculated.

RESULTS

Both SCZ and BD had significantly increased D_f compared to HV despite controlling for possible confounding factors. However, there was no significant difference between SCZ and BD. These findings suggest abnormal retinal vascular D_f in psychoses.

LIMITATIONS

The study design was cross-sectional, and patients were on medications. Confound of lifestyle factors such as diet and exercise, if any, was not controlled. Sub-group analysis between BD-I and BD-II was not performed in view of the small sample.

CONCLUSIONS

Considering the easy accessibility, affordability, and non-invasive nature of the examination, retinal vascular D_f could serve as a surrogate marker for cerebral vascular abnormality and could potentially identify BD and SCZ patients at risk of developing adverse vascular events.

Association between Coronary Artery Measurements and Retinal Microvasculature in Children with New Onset of Kawasaki Disease

About a quarter of children with new onset of Kawasaki disease (KD) encounter coronary arterial involvement. While KD is known to cause vasculitis of medium-sized vessels, few studies have been done to study the involvement of the microcirculation. We aimed to investigate the association between coronary arterial dilatation and retinal microvasculature in a pilot setting, in order to further study the pathophysiological mechanism of KD from the perspective of small vessels changes. We performed a cross-sectional, observational, hospital-based study on 11 children aged 2 years and above with new-onset KD. Cardiac imaging technicians performed the echocardiographic examinations and recorded right coronary artery (RCA), left coronary artery (LCA) and left anterior descending artery (LAD). Qualified retinal graders reviewed and graded standardised retinal photographs to assess retinal microvascular parameters. Among 11 participants, there were 7 boys and 4 girls. Median and interquartile range of participants’ age were 5.92 (3.08) years. After adjusting for age and sex, each unit increase in LAD (mm) was significantly associated with increment of retinal arteriolar tortuosity (4.25 × 10⁻⁵ units, 95% Confidence Interval: 1.19, 7.32). Retinal arteriolar geometric changes were associated with LAD dilatation in 11 children with new onset of KD. Our pilot provided proof-of-concept that retinal imaging might be useful for detecting coronary arterial involvement in young children with KD and it needs further investigation.

Impact of Ideal Cardiovascular Health in Childhood on the Retinal Microvasculature in Midadulthood: Cardiovascular Risk in Young Finns Study

Background

This study examined the association between ideal cardiovascular health (CVH) and the retinal microvasculature in midadulthood.

Methods and Results

The Cardiovascular Risk in Young Finns Study included children from 5 Finnish University cities, who were chosen randomly from the national population register. Participants ranged from 12 to 18 years in childhood (1986) and from 37 to 43 years in midadulthood (2011). Ideal CVH was defined according to the American Heart Association criteria. Retinal microvascular measures included diameters, lengths, length:diameter ratio, and tortuosity. From childhood to adulthood, fasting plasma glucose and blood pressure were significantly higher in those with impaired fasting glucose or diabetes mellitus. Childhood ideal CVH was negatively associated with adult arteriolar tortuosity (β=−0.008; 95% confidence interval [CI], −0.01 to −0.003; P=0.001). Improved ideal CVH from childhood to adulthood was positively associated with adult arteriolar diameter (β=0.122; 95% CI, 0.01–0.24; P=0.033) and negatively associated with adult length:diameter ratio (β=−0.666; 95% CI, −1.25 to −0.08; P=0.026). When stratified by glucose metabolism, among those with diabetes mellitus and impaired fasting glucose, there was a negative association between childhood ideal CVH and adult venular diameter (diabetes mellitus: β=−2.75; 95% CI, −5.46 to −0.04; P=0.047; impaired fasting glucose: β=−2.13; 95% CI, −4.18 to −0.08; P=0.042).

Conclusions

This study is the first to comprehensively examine the impact of CVH from childhood to midadulthood on quantitative measures of the retinal microvasculature. Ideal CVH in childhood and improvement in CVH from childhood to adulthood appears to have a protective effect on the retinal microvasculature in those with, without, and at risk of diabetes mellitus.

Associations of Retinal Microvascular Diameters and Tortuosity With Blood Pressure and Arterial Stiffness: United Kingdom Biobank

Supplemental Digital Content is available in the text.

To examine the baseline associations of retinal vessel morphometry with blood pressure (BP) and arterial stiffness in United Kingdom Biobank. The United Kingdom Biobank included 68 550 participants aged 40 to 69 years who underwent nonmydriatic retinal imaging, BP, and arterial stiffness index assessment. A fully automated image analysis program (QUARTZ [Quantitative Analysis of Retinal Vessel Topology and Size]) provided measures of retinal vessel diameter and tortuosity. The associations between retinal vessel morphology and cardiovascular disease risk factors/outcomes were examined using multilevel linear regression to provide absolute differences in vessel diameter and percentage differences in tortuosity (allowing within person clustering), adjusted for age, sex, ethnicity, clinic, body mass index, smoking, and deprivation index. Greater arteriolar tortuosity was associated with higher systolic BP (relative increase, 1.2%; 95% CI, 0.9; 1.4% per 10 mmHg), higher mean arterial pressure, 1.3%; 0.9, 1.7% per 10 mmHg, and higher pulse pressure (PP, 1.8%; 1.4; 2.2% per 10 mmHg). Narrower arterioles were associated with higher systolic BP (−0.9 µm; −0.94, −0.87 µm per 10 mmHg), mean arterial pressure (−1.5 µm; −1.5, −1.5 µm per 10 mmHg), PP (−0.7 µm; −0.8, −0.7 µm per 10 mmHg), and arterial stiffness index (−0.12 µm; −0.14, −0.09 µm per ms/m²). Associations were in the same direction but marginally weaker for venular tortuosity and diameter. This study assessing the retinal microvasculature at scale has shown clear associations between retinal vessel morphometry, BP, and arterial stiffness index. These observations further our understanding of the preclinical disease processes and interplay between microvascular and macrovascular disease.

Endothelial Function Assessed by Digital Volume Plethysmography Predicts the Development and Progression of Type 2 Diabetes Mellitus

Background

Endothelial dysfunction is a consequence of type 2 diabetes mellitus, but it is unclear whether endothelial dysfunction of conductance versus resistance vessels may also precede type 2 diabetes mellitus development.

Methods and Results

In a population‐based cohort of 15 010 individuals from the GHS (Gutenberg Health Study) (aged 35–74 years at enrollment in 2007–2012), we identified 1610 cases of incident pre–diabetes mellitus and 386 cases of incident type 2 diabetes mellitus by hemoglobin A_1c (HbA_1c) and/or medical history between 2012 and 2017. Endothelial function of conductance and resistance vessels was measured by flow‐mediated dilation and digital volume plethysmography–derived reactive hyperemia index, respectively. Multivariable regression modeling was used to estimate β coefficients of HbA_1c levels at follow‐up and relative risks of incident (pre–)diabetes mellitus. Reactive hyperemia index was independently associated with HbA_1c after multivariable adjustment for baseline HbA_1c, sex, age, socioeconomic status, arterial hypertension, waist/height ratio, pack‐years of smoking, non–high‐density lipoprotein/high‐density lipoprotein ratio, physical activity, family history of myocardial infarction/stroke, prevalent cardiovascular disease, medication use, and C‐reactive protein (β=−0.020; P=0.0029). The adjusted relative risk per SD decline in reactive hyperemia index was 1.08 (95% CI, 1.02–1.15; P=0.012) for incident pre–diabetes mellitus and 1.16 (95% CI, 1.01–1.34; P=0.041) for incident type 2 diabetes mellitus. Flow‐mediated dilation independently increased the relative risk for developing pre–diabetes mellitus by 8% (95% CI, 1.02–1.14; P=0.012), but it was not independently associated with incident type 2 diabetes mellitus (relative risk, 1.01; 95% CI, 0.86–1.19; P=0.92) and with HbA_1c (β=−0.003; P=0.59).

Conclusions

Endothelial dysfunction of resistance rather than conductance vessels may precede the development of (pre–)diabetes mellitus. Assessment of endothelial function by digital volume plethysmography may help to identify subjects at risk for development of type 2 diabetes mellitus.

Retinal vascular tortuosity in schizophrenia and bipolar disorder

The micro-vasculature of retina and brain share common morphological, physiological, and pathological properties. Retina being easily accessible, retinal vascular examination provides an indirect assessment of cerebral vasculature. Considering the high prevalence of vascular morbidity in SCZ and BD a few studies have examined retinal vascular caliber and have reported increased retinal venular caliber in schizophrenia (SCZ). Retinal vascular tortuosity could serve as a better structural measure than caliber as it is static and less susceptible to pulse period variations. However, to date, no study has examined retinal vascular tortuosity in SCZ and bipolar disorder (BD). Hence, we examined retinal vascular tortuosity in comparison with healthy volunteers (HV). We included 255 subjects (78 HV, 79 SCZ, and 86 BD) in the age range of 18 to 50 years. Trained personnel acquired images using a non-mydriatic fundus camera. To measure the average retinal arteriolar tortuosity index (RATI) and retinal venular tortuosity index (RVTI), we used a previously validated, semi-automatic algorithm. The results showed significant differences across the three groups in RATI but not in RVTI; both BD and SCZ had significantly increased RATI compared to HV. There was also a significant difference between SCZ and BD, with BD having higher RATI. If shown to be of predictive utility in future longitudinal studies, it has the potential to identify patients at risk of development of adverse vascular events. As retinal vascular imaging is non-invasive and inexpensive, it could serve as a proxy marker and window to cerebral vasculature.

Lumen narrowing and increased wall to lumen ratio of retinal microcirculation are valuable biomarkers of hypertension-mediated cardiac damage

Purpose: In the course of hypertension, left ventricular hypertrophy and diastolic dysfunction develop very often and may progress toward heart failure. The aim of the study was to analyze the relationship between abnormalities of retinal microcirculation and cardiac damage defined as left ventricular hypertrophy and/or diastolic dysfunction.Materials and methods: The study comprised 88 patients with essential hypertension. The group was divided into two subgroups: hypertensives without cardiac damage (n = 55) and with cardiac damage (n = 33). Control group comprised 32 normotensive subjects. Scanning laser Doppler flowmetry was used to evaluate retinal microcirculation. Echocardiography was used to assess cardiac damage.Results: Lumen diameter of retinal arterioles was significantly smaller in patients with cardiac damage vs. controls (77 vs. 84 µm, p = 0.02). Additionally, there was an evident trend with respect to lumen diameter (LD) across all three studied subgroups; i.e.: the smallest dimeters were present in cardiac damage patients, moderate size in hypertensives' without cardiac damage, and the largest diameters in healthy controls (pfor trend < 0.01). Lumen diameter was inversely correlated with cardiac intraventricular septum diameter (R = -0.25, p = 0.02), left ventricular mass (R = -0.24, p = 0.02), and left atrial volume (R = -0.22, p = 0.04). Wall to lumen ratio was associated with intraventricular septum diameter (R = 0.21, p = 0.044) and left atrial volume (R = 0.21, p = 0.045). In multivariable regression analysis, lumen diameter was independently associated with intraventricular septum diameter (β = -0.05, p = 0.03), left ventricular mass (β = -1.15, p = 0.04), and left atrial volume (β = -0.42, p = 0.047); wall to lumen ratio was independently associated with intraventricular septum diameter (β = 3.67, p = 0.02) and left atrial volume (β = 30.0, p = 0.04).Conclusions: In conclusion, retinal arterioles lumen diameter and wall to lumen ratio were independent biomarkers of cardiac damage. Retinal examination performed by means of scanning laser Doppler flowmetry might be a valuable tool to improve cardiovascular risk stratification of hypertensive patients.

Automatic optic disc detection in colour fundus images by means of multispectral analysis and information content

The optic disc (OD) in retinal fundus images is widely used as a reference in computer-based systems for the measurement of the severity of retinal disease. A number of algorithms have been published in the past 5 years to locate and measure the OD in digital fundus images. Our proposed algorithm, automatically: (i) uses the three channels (RGB) of the digital colour image to locate the region of interest (ROI) where the OD lies, (ii) measures the Shannon information content per channel in the ROI, to decide which channel is most appropriate for searching for the OD centre using the circular Hough transform. A series of evaluations were undertaken to test our hypothesis that using the three channels gives a better performance than a single channel. Three different databases were used for evaluation purposes with a total of 2,371 colour images giving a misdetection error of 3% in the localisation of the centre of the OD. We find that the area determined by our algorithm which assumes that the OD is circular, is similar to that found by other algorithms that detected the shape of the OD. Five metrics were measured for comparison with other recent studies. Combining the two databases where expert delineation of the OD is available (1,240 images), the average results for our multispectral algorithm are: TPR = 0.879, FPR = 0.003, Accuracy = 0.994, Overlap = 80.6% and Dice index = 0.878.

Detection of smoking status from retinal images; a Convolutional Neural Network study

Cardiovascular diseases are directly linked to smoking habits, which has both physiological and anatomical effects on the systemic and retinal circulations, and these changes can be detected with fundus photographs. Here, we aimed to 1- design a Convolutional Neural Network (CNN), using retinal photographs, to differentiate between smokers and non-smokers; and 2- use the attention maps to better understand the physiological changes that occur in the retina in smokers. 165,104 retinal images were obtained from a diabetes screening programme, labelled with self-reported "smoking" or "non-smoking" status. The images were pre-processed in one of two ways, either "contrast-enhanced" or "skeletonized". Experiments were run on an Intel Xeon Gold 6128 CPU @ 3.40 GHz with 16 GB of RAM memory and a NVIDIA GeForce TiTan V VOLTA 12 GB, for 20 epochs. The dataset was split 80/20 for training and testing sets, respectively. The overall validation outcomes for the contrast-enhanced model were accuracy 88.88%, specificity 93.87%. In contrast, the outcomes of the skeletonized model were accuracy 63.63%, specificity 65.60%. The "attention maps" that were generated of the contrast-enhanced model highlighted the retinal vasculature, perivascular region and the fovea most prominently. We trained a customized CNN to accurately determine smoking status. The retinal vasculature, the perivascular region and the fovea appear to be important predictive features in the determination of smoking status. Despite a high degree of accuracy, the sensitivity of our CNN was low. Further research is required to establish whether the frequency, duration, and dosage (quantity) of smoking would improve the sensitivity of the CNN.

Intravitreal dexamethasone implant for central retinal vein occlusion without macular edema

Background

To evaluate the efficacy of an intravitreal dexamethasone (IVD) implant (Ozurdex®) for the treatment of central retinal vein occlusion (CRVO) without macular edema (ME).

Methods

A retrospective cohort study was designed, and 20 eyes of 20 patients diagnosed with non-ischemic CRVO without ME were included. A total of 10 CRVO eyes were observed without treatment, and another 10 CRVO eyes received a single IVD injection at baseline. Mean changes in pathomorphologic parameters of fundus and optical coherence tomography parameters were measured at baseline and at 1, 3, 6, and 12 months.

Results

The decreases in venous tortuosity (p = 0.014 for superior; 0.036 for inferior arcades) and width (p = 0.024 for superior; 0.003 for inferior arcades) from baseline to 12 months after injection were significantly greater in the treated group than the observed group. The improvements in RNFL swelling (p = 0.010) and retinal hemorrhage (p = 0.006) were also significantly greater in the treated group. Visual symptom improvement was significantly faster in the treated group (p = 0.001). In two cases, IVD injection resulted in complete resolution of cilioretinal artery occlusion associated with the CRVO, leading to complete visual recovery in 1 week. None of the treated eyes showed signs of ME development, ischemia progression, or neovascularization.

Conclusions

IVD implant was significantly effective in improving venous engorgement, retinal hemorrhage, RNFL swelling, and visual symptoms by presumed alleviation of disc swelling and venous outflow. This treatment may be a considerable treatment option in CRVO patients with no ME.

Adaptive optics imaging of the human retina

Adaptive Optics (AO) retinal imaging has provided revolutionary tools to scientists and clinicians for studying retinal structure and function in the living eye. From animal models to clinical patients, AO imaging is changing the way scientists are approaching the study of the retina. By providing cellular and subcellular details without the need for histology, it is now possible to perform large scale studies as well as to understand how an individual retina changes over time. Because AO retinal imaging is non-invasive and when performed with near-IR wavelengths both safe and easily tolerated by patients, it holds promise for being incorporated into clinical trials providing cell specific approaches to monitoring diseases and therapeutic interventions. AO is being used to enhance the ability of OCT, fluorescence imaging, and reflectance imaging. By incorporating imaging that is sensitive to differences in the scattering properties of retinal tissue, it is especially sensitive to disease, which can drastically impact retinal tissue properties. This review examines human AO retinal imaging with a concentration on the use of the Adaptive Optics Scanning Laser Ophthalmoscope (AOSLO). It first covers the background and the overall approaches to human AO retinal imaging, and the technology involved, and then concentrates on using AO retinal imaging to study the structure and function of the retina.

Retinal vascular tortuosity assessment: inter-intra expert analysis and correlation with computational measurements

Background

The retinal vascular tortuosity can be a potential indicator of relevant vascular and non-vascular diseases. However, the lack of a precise and standard guide for the tortuosity evaluation hinders its use for diagnostic and treatment purposes. This work aims to advance in the standardization of the retinal vascular tortuosity as a clinical biomarker with diagnostic potential, allowing, thereby, the validation of objective computational measurements on the basis of the entire spectrum of the expert knowledge.

Methods

This paper describes a multi-expert validation process of the computational vascular tortuosity measurements of reference. A group of five experts, covering the different clinical profiles of an ophthalmological service, and a four-grade scale from non-tortuous to severe tortuosity as well as non-tortuous / tortuous and asymptomatic / symptomatic binary classifications are considered for the analysis of the the multi-expert validation procedure. The specialists rating process comprises two rounds involving all the experts and a joint round to establish consensual rates. The expert agreement is analyzed throughout the rating procedure and, then, the consensual rates are set as the reference to validate the prognostic performance of four computational tortuosity metrics of reference.

Results

The Kappa indexes for the intra-rater agreement analysis were obtained between 0.35 and 0.83 whereas for the inter-rater agreement in the asymptomatic / symptomatic classification were between 0.22 and 0.76. The Area Under the Curve (AUC) for each expert against the consensual rates were placed between 0.61 and 0.83 whereas the prognostic performance of the best objective tortuosity metric was 0.80.

Conclusions

There is a high inter and intra-rater variability, especially for the case of the four grade scale. The prognostic performance of the tortuosity measurements is close to the experts’ performance, especially for Grisan measurement. However, there is a gap between the automatic effectiveness and the expert perception given the lack of clinical criteria in the computational measurements.

Relationship of ideal cardiovascular health metrics with retinal vessel calibers and retinal nerve fiber layer thickness: a cross-sectional study

BACKGROUND

Ideal cardiovascular health (CVH) metrics have been found to be associated with subclinical vascular abnormalities. However, the relationship between ideal CVH metrics and retinal vessel calibers and retinal nerve fiber layer (RNFL) thickness in a Chinese population is unknown.

METHODS

We collected information on the seven ideal CVH metrics among 3376 participants aged 40 years or older from the Asymptomatic Polyvascular Abnormalities Community Study in 2012. Retinal vessel calibers and RNFL thickness were assessed by retinal photography and spectral-domain optical coherence tomography. Multivariable linear models were used to analyze the relationship between ideal CVH metrics and retinal parameters.

RESULTS

With the decreased number of ideal CVH metrics, central retinal arteriolar equivalents (CRAE) was significantly narrowed and arterio- venous ratio (AVR) significantly decreased (p < 0.0001). While the RNFL thickness and central retinal venous equivalents (CRVE) showed no significant changes with the decreased ideal CVH metrics. Linear regression showed that both CRAE and AVR was positively related with the number of ideal CVH metrics (regression coefficient beta: 0.806, 95% confidence interval (CI): 0.266-1.346 for CRAE (micron); and regression coefficient beta: 0.005, 95% CI: 0.002-0.009 for AVR) after adjusting for age (year), sex = male (n), education (n), average monthly income (¥) and other related risk factors.

CONCLUSIONS

These findings suggested a clear positive relationship between the number of ideal CVH metrics and CRAE and AVR in Chinese population, supporting the importance of ideal health behaviors and factors in subclinical vascular abnormalities prevention.

Association of serum lutein and zeaxanthin with quantitative measures of retinal vascular parameters

To evaluate the association between serum carotenoids and quantitative measures of retinal vasculature in elderly Singapore Chinese subjects. The following details were collected in 128 healthy subjects: sociodemographics, lifestyle information, medical and drug history, and anthropometric measurements. Serum concentrations of carotenoids were estimated in fasting venous blood using high performance liquid chromatography. Retinal vascular parameters were quantitatively measured from retinal photographs using a computer-assisted program (Singapore I Vessel Assessment). The mean age of the population was 54.1 years (range 40 to 81 years). In multiple linear regression analysis, per SD decrease in retinal arteriolar caliber [β = 0.045 (0.003 to 0.086), p = 0.036], per SD increase in retinal venular caliber [β = -0.045 (-0.086 to -0.003), p = 0.036] and per SD increase in arteriolar branching angle [β = -0.039 (-0.072 to -0.006), p = 0.021] were associated with decreased serum lutein. Per SD increase in retinal venular tortuosity [β = -0.0075 (-0.0145 to -0.0004), p = 0.039] and per SD increase in arteriolar branching angle (β = -0.0073 [-0.0142 to -0.0059], p = 0.041) were associated with decreased serum zeaxanthin. None of the other carotenoids demonstrated meaningful relationship with quantitative measures of retinal vasculature. Lower levels of lutein and zeaxanthin demonstrated significant relationship with adverse quantitative measures of retinal vasculature in elderly healthy subjects.

Retinal Vascular Geometry and the Prevalence of Atrial Fibrillation and Heart Failure in a Clinic-Based Sample

BACKGROUND

We aimed to examine the cross-sectional association between a range of retinal vascular geometric variables and the prevalence of atrial fibrillation (AF) and heart failure.

METHODS

The Australian Heart Eye Study (AHES) surveyed 1,680 participants presenting to a tertiary referral hospital for the evaluation of potential coronary artery disease by coronary angiography. Retinal vascular geometric variables (tortuosity, branching, and fractal dimension) were measured from retinal photographs using a computer-assisted program (Singapore I Vessel Assessment). Atrial fibrillation was determined based on a combination of: self-reported history of AF; self-reported use of rate-control and anti-arrhythmic medications; and/or screening electrocardiogram. Self-reported echocardiography-confirmed heart failure was also documented.

RESULTS

A total of 1,169 participants had complete information on retinal vascular geometric variables and AF and of these 104 (8.9%) had AF. Participants in the second tertile of fractal dimension (D_f) compared to those in the highest tertile (reference group), had 92% increased likelihood of having AF after multivariable adjustment. A threshold effect for D_f was identified, and participants below versus those above a D_f threshold value of 1.472, had greater odds of having AF: multivariable-adjusted OR 1.85 (95% CI 1.03-3.31). Measures of retinal tortuosity and branching were not associated with AF. Retinal vascular geometric variables were also not associated with prevalence of heart failure.

CONCLUSIONS

A sparser retinal microvascular network (lower D_f) was independently associated with greater likelihood of AF. Further studies are needed to investigate whether temporal changes to the retinal vascular geometry are predictive of AF in the longer term.

Periodontitis and multiple markers of cardiometabolic risk in the fourth decade: A cohort study

OBJECTIVES

To examine associations between periodontitis at ages 32 and 38 and a range of early cardiometabolic risk biomarkers at age 38.

METHODS

Periodontal probing depth and bleeding on probing data collected during the age-32 and age-38 assessments in the Dunedin Multidisciplinary Health and Development Study were used to quantify periodontal inflammatory load. Retinal microvascular abnormalities, endothelial dysfunction, and metabolic syndrome data were collected during the age-38 assessment. Regression models were used to examine associations between these cardiometabolic risk markers and (1) the inflammatory load at age 38 and (2) the change in inflammatory load between ages 32 and 38.

RESULTS

Periodontal inflammatory load was recorded for 890 Study members at age 32, 891 at age 38, and 856 at both ages. Retinal vessel data were available for 922, endothelial dysfunction data for 909 and metabolic syndrome data for 905 at age 38. Neither the inflammatory load of periodontitis at 38 nor the changes in inflammatory load 32-38 were found to be associated with any of the three cardiometabolic risk markers.

CONCLUSIONS

Periodontitis was not associated with markers of cardiometabolic risk at this relatively early stage in the life course. It is possible that any influence of periodontitis on cardiometabolic health develops later in life, or periodontitis is not involved in the putative causal chain comprising systemic inflammation, cardiometabolic risk markers, and subsequent cardiovascular risk.

Risk of retinal artery occlusion in patients with diabetes mellitus: A retrospective large-scale cohort study

There is a globally increasing prevalence and incidence of diabetes mellitus (DM). Prolonged hyperglycaemia could lead to both macrovascular damage, such as carotid artery atherosclerosis, and microvascular damage, such as retinal arteriolar narrowing, and might contribute to retinal artery occlusion (RAO). Accordingly, it is important to determine whether DM is a contrubuting factor of RAO. We conducted a retrospective cohort study that included 241,196 DM patients from the Longitudinal Cohort of Diabetes Patients Database who were recruited between 2003 and 2005. An age- and sex-matched non-DM control group included the same number of patients who were selected from the Taiwan Longitudinal Health Insurance Database of 2000. Relevant data of each patient were collected from the index date until December 2013. The incidence and risk of RAO were calculated and compared between the DM and non-DM groups. The hazard ratio for RAO was calculated using Cox proportional hazards regression analysis after adjusting for confounders. The cumulative incidence rate of RAO was calculated by Kaplan-Meier analysis. In total, 317 patients with DM and 144 controls developed RAO during the follow-up period, leading to an incidence rate of RAO in DM patients that was 2.30 times (95% confidence interval [CI] = 1.89-2.80) greater than that in controls. After adjustment for potential confounders, patients with DM were 2.11 times (95% CI, 1.71-2.59) more likely to develop RAO in the total study cohort. In conclusion, DM increases the risk of RAO, which is an interdisciplinary emergency. Close collaboration between endocrinologists and ophthalmologists is important in managing RAO following DM.

QUANTIFICATION OF RETINAL VESSEL TORTUOSITY IN DIABETIC RETINOPATHY USING OPTICAL COHERENCE TOMOGRAPHY ANGIOGRAPHY

PURPOSE

To investigate the association of vessel tortuosity with severity of diabetic retinopathy (DR) using optical coherence tomography angiography.

METHODS

We retrospectively analyzed 30 healthy eyes and 121 eyes of diabetic subjects with no DR, mild nonproliferative DR (NPDR), moderate to severe NPDR and proliferative DR (PDR). Binarized images were used to quantify the vessel tortuosity, vessel density, foveal avascular zone (FAZ) area, and FAZ acircularity. The vessels were divided vertically as superficial retinal layer and deep retinal layer, and horizontally as circular areas with 3 mm and 1.5 mm diameters. Analysis of variance was performed for multiple comparisons. Correlation analysis evaluated the association between the quantified parameters.

RESULTS

Compared with healthy eyes, vessel tortuosity increased as DR severity was more in NPDR, but decreased in PDR (P = 0.033). The decrease in vessel density and the increase in both FAZ area and FAZ acircularity were consistent, while DR approached PDR. Among all parameters, statistically significant difference between no DR and mild NPDR was observed only in vessel tortuosity, especially within the 1.5 mm area of superficial retinal layer (P = 0.011). Correlations of vessel tortuosity with FAZ area and acircularity were confined to the 3 mm and 1.5 mm areas of superficial retinal layer (r = -0.185, P = 0.023 for FAZ area; r = 0.268, P = 0.001 for FAZ acircularity), while vessel density strongly correlated with FAZ parameters in the superficial retinal layer and deep retinal layer.

CONCLUSION

Vessel tortuosity increased as the stage of NPDR was more severe, but decreased in PDR. The vessel tortuosity determined using optical coherence tomography angiography might be a useful parameter indicating the progression to PDR, circumventing the risk from invasive conventional angiography.

A Review on the Extraction of Quantitative Retinal Microvascular Image Feature

Digital image processing is one of the most widely used computer vision technologies in biomedical engineering. In the present modern ophthalmological practice, biomarkers analysis through digital fundus image processing analysis greatly contributes to vision science. This further facilitates developments in medical imaging, enabling this robust technology to attain extensive scopes in biomedical engineering platform. Various diagnostic techniques are used to analyze retinal microvasculature image to enable geometric features measurements such as vessel tortuosity, branching angles, branching coefficient, vessel diameter, and fractal dimension. These extracted markers or characterized fundus digital image features provide insights and relates quantitative retinal vascular topography abnormalities to various pathologies such as diabetic retinopathy, macular degeneration, hypertensive retinopathy, transient ischemic attack, neovascular glaucoma, and cardiovascular diseases. Apart from that, this noninvasive research tool is automated, allowing it to be used in large-scale screening programs, and all are described in this present review paper. This paper will also review recent research on the image processing-based extraction techniques of the quantitative retinal microvascular feature. It mainly focuses on features associated with the early symptom of transient ischemic attack or sharp stroke.

Longitudinal association between psychosocial stress and retinal microvasculature in children and adolescents

BACKGROUND

Retinal microvessels provides a window to assess the microcirculation of heart and brain, and might reflect cardio- or cerebrovascular disease risk. Limited information exist on the relation between psychosocial stress and the microcirculation, even though psychosocial stress might trigger vascular diseases. This study investigates whether childhood psychosocial stress is a predictor of retinal microvasculature.

METHODS

We followed-up 182 Belgian children, aged 5.7-11.3 years at baseline (53.3% boys). Information about psychosocial stress was obtained using emotional, behavioral and negative life events questionnaires and hair cortisol, an objective stress marker. Retinal photographs were used to calculate vessel diameters, bifurcation angles and optimality deviation with semiautomated software. Cross-sectional and longitudinal associations were explored using multivariable regression analysis with retinal parameters in 2015 as outcome, while adjusting for age, sex, socioeconomic status, cardiovascular parameters and lifestyle factors.

RESULTS

Feelings of happiness, sadness and negative life events were associated with retinal vascular diameter, but behavior and hair cortisol were not. High stress levels over a 4-year time period (less happy, sadder and higher total negative emotions) were associated with larger venules (β = 0.21-0.43) and children who experienced more negative life events had smaller arterioles (β = -0.15). No consistent patterns were seen with bifurcation angles and optimality deviation.

CONCLUSION

Based on the results, we conclude that high levels of childhood psychosocial stress unfavorably affect the retinal vascular diameters, potentially reflecting the microvasculature of the heart and brain. It seems this might even be independent of lifestyle and BMI, but further research on mechanisms is necessary.

Retinal capillary and arteriolar changes in patients with chronic kidney disease

RATIONALE

Premature cardiovascular disease is a leading cause of death in patients with chronic kidney disease (CKD). In animal models CKD has been shown to cause renal and extrarenal vascular remodeling and capillary rarefaction, but data in humans with CKD are sparse. Retinal arteriolar wall-to-lumen ratio (WLR) is an established marker of early end-organ damage and there is evidence that arteriolar and capillary changes in the retinal circulation mirror those in the general and in particular the cerebrovascular microcirculation.

OBJECTIVE

The aim of this study was to compare retinal capillary density and arteriolar structure between patients with CKD and healthy individuals.

METHODS

We compared 76 patients with CKD stage 3+ or proteinuria >500 mg/g creatinine in the presence of a normal GFR from the German Chronic Kidney Disease cohort to 53 healthy control subjects, who participated in clinical trials during 2007 and 2015 in our Clinical Research Center. Retinal vascular parameters were measured non-invasively in vivo by scanning laser Doppler Flowmetry (SLDF, Heidelberg Engineering, Germany). Capillary rarefaction was assessed by intercapillary distance.

RESULTS

Patients with CKD showed greater WLR (0.403 ± 0.11 vs 0.351 ± 0.11, p = 0.010) and greater wall thickness (WT) (15.1 ± 4.1 vs 13.5 ± 3.8, p = 0.026) compared to healthy individuals. Intercapillary distance (ICD) (22.4 ± 5.7 vs 20.2 ± 4.1, p = 0.008) was greater in the CKD group compared to the healthy control group. After adjustment for differences in clinical characteristics of the groups (age, gender, BMI, serum cholesterol) WLR (p = 0.046), WT (p = 0.025) and ICD (p = 0.003) remained significantly different between the two groups. There was a correlation between serum phosphate level and WLR in the CKD group (r = 0.288, p = 0.013).

CONCLUSION

Patients with moderately severe CKD show retinal signs of end-organ damage indicated by an increased wall-to-lumen ratio and capillary rarefaction.

Three-question set from Michigan Neuropathy Screening Instrument adds independent prognostic information on cardiovascular outcomes: analysis of ALTITUDE trial

AIMS/HYPOTHESIS

The self-administered Michigan Neuropathy Screening Instrument (MNSI) is used to diagnose diabetic peripheral neuropathy. We examined whether the MNSI might also provide information on risk of death and cardiovascular outcomes.

METHODS

In this post hoc analysis of the Aliskiren Trial in Type 2 Diabetes Using Cardio-Renal Endpoints (ALTITUDE) trial, we divided 8463 participants with type 2 diabetes and chronic kidney disease (CKD) and/or cardiovascular disease (CVD) into independent training (n = 3252) and validation (n = 5211) sets. In the training set, we identified specific questions that were independently associated with a cardiovascular composite outcome (cardiovascular death, resuscitated cardiac arrest, non-fatal myocardial infarction/stroke, heart failure hospitalisation). We then evaluated the performance of these questions in the validation set.

RESULTS

In the training set, three questions ('Are your legs numb?', 'Have you ever had an open sore on your foot?' and 'Do your legs hurt when you walk?') were significantly associated with the cardiovascular composite outcome. In the validation set, after multivariable adjustment for key covariates, one or more positive responses (n = 3079, 59.1%) was associated with a higher risk of the cardiovascular composite outcome (HR 1.54 [95% CI 1.28, 1.85], p < 0.001), heart failure hospitalisation (HR 1.74 [95% CI 1.29, 2.35], p < 0.001), myocardial infarction (HR 1.81 [95% CI 1.23, 2.69], p = 0.003), stroke (HR 1.75 [95% CI 1.20, 2.56], p = 0.003) and three-point major adverse cardiovascular events (MACE) (cardiovascular death, non-fatal myocardial infarction, non-fatal stroke) (HR 1.49 [95% CI 1.20, 1.85], p < 0.001) relative to no positive responses to all questions. Associations were stronger if participants answered positively to all three questions (n = 552, 11%). The addition of the total number of affirmative responses to existing models significantly improved Harrell's C statistic for the cardiovascular composite outcome (0.70 vs 0.71, p = 0.010), continuous net reclassification improvement (+22% [+10%, +31%], p = 0.027) and integrated discrimination improvement (+0.9% [+0.4%, +2.1%], p = 0.007).

CONCLUSIONS/INTERPRETATION

We identified three questions from the MNSI that provide additional prognostic information for individuals with type 2 diabetes and CKD and/or CVD. If externally validated, these questions may be integrated into the clinical history to augment prediction of CV events in high-risk individuals with type 2 diabetes.

Determinants of retinal microvascular features and their relationships in two European populations

Supplemental Digital Content is available in the text

Objectives:

To examine factors influencing retinal vasculature in two environmentally contrasted, cross-sectional studies of adult participants of European descent and to estimate the extent and specificity of genetic contributions to each retinal vasculature feature.

Methods:

Retinal images from 1088 participants in the Orkney Complex Disease Study and 387 in the CROATIA-Korčula study, taken using the same nonmydriatic camera system and graded by the same person, were evaluated. Using general linear models, we estimated the influence of an extensive range of systemic risk factors, calculated retinal traits heritabilities and genetic correlations.

Main results:

Systemic covariates explained little (<4%) of the variation in vessel tortuosity, substantially more (>10%, up to 31.7%) of the variation in vessel width and monofractal dimension. Suggestive not well trodden associations of biological interest included that of urate, tissue plasminogen activator and cardiac PR interval with arteriolar narrowing, that of carotid intima–media thickness with less-tortuous arterioles and of cardiac QT interval with more tortuous venules. The genetic underpinning of tortuosity is largely distinct from that of the other retinal vascular features, whereas that of fractal dimension and vessel width greatly overlaps. The previously recognized influence of ocular axial length on vessel widths was high and can be expected to lead to artefactual genetic associations [genetic correlation with central retinal arteriolar equivalent: −0.53 (standard error 0.11)]. The significant genetic correlation between SBP and central retinal arteriolar equivalent, −0.53 (standard error 0.22) (after adjusting for age, sex and axial length of the eye), augurs more favourably for the discovery of genetic variants relevant to vascular physiology.

Fractal dimension and lacunarity analysis of retinal microvascular morphology in hypertension and diabetes

Hypertension and diabetes mellitus represent modifiable risk factors for vascular disease. They cause microvascular remodeling, and ultimately result in end-organ damage. Therefore, development of methods for noninvasive quantification of the effects of hypertension and diabetes mellitus on microvasculature is of paramount importance. The two goals of the study were: 1) to characterize the geometric complexity and inhomogeneity of retinal vasculature in hypertensive retinopathy (HR) and in proliferative diabetic retinopathy (PDR) by using box counting fractal dimension and lacunarity analysis, and 2) to determine if the combination of these two parameters can be used to describe differences in the vascular tree geometry between HR and PDR. The extended set of retinal images from the publicly available STARE database was manually segmented by our expert, validated, and made available for other researchers to use. The healthy retinal vascular network has a higher complexity (fractal dimension) compared to that in HR and in PDR. However, there is no difference in microvascular complexity between HR and PDR. The inhomogeneity of the retinal microvascular tree (lacunarity) was higher in PDR compared to HR. Lacunarity and fractal dimension together quantitatively characterize microvascular geometry in the retina with higher specificity than fractal analysis alone.

Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning

Traditionally, medical discoveries are made by observing associations, making hypotheses from them and then designing and running experiments to test the hypotheses. However, with medical images, observing and quantifying associations can often be difficult because of the wide variety of features, patterns, colours, values and shapes that are present in real data. Here, we show that deep learning can extract new knowledge from retinal fundus images. Using deep-learning models trained on data from 284,335 patients and validated on two independent datasets of 12,026 and 999 patients, we predicted cardiovascular risk factors not previously thought to be present or quantifiable in retinal images, such as age (mean absolute error within 3.26 years), gender (area under the receiver operating characteristic curve (AUC) = 0.97), smoking status (AUC = 0.71), systolic blood pressure (mean absolute error within 11.23 mmHg) and major adverse cardiac events (AUC = 0.70). We also show that the trained deep-learning models used anatomical features, such as the optic disc or blood vessels, to generate each prediction.