Prediction of cardiovascular risk factors from retinal fundus photographs: Validation of a deep learning algorithm in a prospective non-interventional study in Kenya.

AIM: Hypertension and diabetes mellitus (DM) are major causes of morbidity and mortality, with growing burdens in low-income countries where they are underdiagnosed and undertreated. Advances in machine learning may provide opportunities to enhance diagnostics in settings with limited medical infrastructure. MATERIALS AND METHODS: A non-interventional study was conducted to develop and validate a machine learning algorithm to estimate cardiovascular clinical and laboratory parameters. At two sites in Kenya, digital retinal fundus photographs were collected alongside blood pressure (BP), laboratory measures and medical history. The performance of machine learning models, originally trained using data from the UK Biobank, were evaluated for their ability to estimate BP, glycated haemoglobin, estimated glomerular filtration rate and diagnoses from fundus images. RESULTS: In total, 301 participants were enrolled. Compared with the UK Biobank population used for algorithm development, participants from Kenya were younger and would probably report Black/African ethnicity, with a higher body mass index and prevalence of DM and hypertension. The mean absolute error was comparable or slightly greater for systolic BP, diastolic BP, glycated haemoglobin and estimated glomerular filtration rate. The model trained to identify DM had an area under the receiver operating curve of 0.762 (0.818 in the UK Biobank) and the hypertension model had an area under the receiver operating curve of 0.765 (0.738 in the UK Biobank). CONCLUSIONS: In a Kenyan population, machine learning models estimated cardiovascular parameters with comparable or slightly lower accuracy than in the population where they were trained, suggesting model recalibration may be appropriate. This study represents an incremental step toward leveraging machine learning to make early cardiovascular screening more accessible, particularly in resource-limited settings.

N-CAM exhibits a regulatory function in pathological angiogenesis in oxygen induced retinopathy.

BACKGROUND: Diabetic retinopathy and retinopathy of prematurity are diseases caused by pathological angiogenesis in the retina as a consequence of local hypoxia. The underlying mechanism for epiretinal neovascularization (tuft formation), which contributes to blindness, has yet to be identified. Neural cell adhesion molecule (N-CAM) is expressed by Müller cells and astrocytes, which are in close contact with the retinal vasculature, during normal developmental angiogenesis. METHODOLOGY/PRINCIPAL FINDINGS: Notably, during oxygen induced retinopathy (OIR) N-CAM accumulated on astrocytes surrounding the epiretinal tufts. Here, we show that N-CAM ablation results in reduced vascular tuft formation due to reduced endothelial cell proliferation despite an elevation in VEGFA mRNA expression, whereas retinal developmental angiogenesis was unaffected. CONCLUSION/SIGNIFICANCE: We conclude that N-CAM exhibits a regulatory function in pathological angiogenesis in OIR. This is a novel finding that can be of clinical relevance in diseases associated with proliferative vasculopathy.

A two-way communication between microglial cells and angiogenic sprouts regulates angiogenesis in aortic ring cultures.

BACKGROUND: Myeloid cells have been associated with physiological and pathological angiogenesis, but their exact functions in these processes remain poorly defined. Monocyte-derived tissue macrophages of the CNS, or microglial cells, invade the mammalian retina before it becomes vascularized. Recent studies correlate the presence of microglia in the developing CNS with vascular network formation, but it is not clear whether the effect is directly caused by microglia and their contact with the endothelium. METHODOLOGY/PRINCIPAL FINDINGS: We combined in vivo studies of the developing mouse retina with in vitro studies using the aortic ring model to address the role of microglia in developmental angiogenesis. Our in vivo analyses are consistent with previous findings that microglia are present at sites of endothelial tip-cell anastomosis, and genetic ablation of microglia caused a sparser vascular network associated with reduced number of filopodia-bearing sprouts. Addition of microglia in the aortic ring model was sufficient to stimulate vessel sprouting. The effect was independent of physical contact between microglia and endothelial cells, and could be partly mimicked using microglial cell-conditioned medium. Addition of VEGF-A promoted angiogenic sprouts of different morphology in comparison with the microglial cells, and inhibition of VEGF-A did not affect the microglia-induced angiogenic response, arguing that the proangiogenic factor(s) released by microglia is distinct from VEGF-A. Finally, microglia exhibited oriented migration towards the vessels in the aortic ring cultures. CONCLUSIONS/SIGNIFICANCE: Microglia stimulate vessel sprouting in the aortic ring cultures via a soluble microglial-derived product(s), rather than direct contact with endothelial cells. The observed migration of microglia towards the growing sprouts suggests that their position near endothelial tip-cells could result from attractive cues secreted by the vessels. Our data reveals a two-way communication between microglia and vessels that depends on soluble factors and should extend the understanding of how microglia promote vascular network formation.

Quantitative comparison of constitutive promoters in human ES cells.

BACKGROUND: Constitutive promoters that ensure sustained and high level gene expression are basic research tools that have a wide range of applications, including studies of human embryology and drug discovery in human embryonic stem cells (hESCs). Numerous cellular/viral promoters that ensure sustained gene expression in various cell types have been identified but systematic comparison of their activities in hESCs is still lacking. METHODOLOGY/PRINCIPAL FINDINGS: We have quantitatively compared promoter activities of five commonly used constitutive promoters, including the human ß-actin promoter (ACTB), cytomegalovirus (CMV), elongation factor-1α, (EF1α), phosphoglycerate kinase (PGK) and ubiquitinC (UbC) in hESCs. Lentiviral gene transfer was used to ensure stable integration of promoter-eGFP constructs into the hESCs genome. Promoter activities were quantitatively compared in long term culture of undifferentiated hESCs and in their differentiated progenies. CONCLUSION/SIGNIFICANCE: The ACTB, EF1α and PGK promoters showed stable activities during long term culture of undifferentiated hESCs. The ACTB promoter was superior by maintaining expression in 75-80% of the cells after 50 days in culture. During embryoid body (EB) differentiation, promoter activities of all five promoters decreased. Although the EF1α promoter was downregulated in approximately 50% of the cells, it was the most stable promoter during differentiation. Gene expression analysis of differentiated eGFP+ and eGFP- cells indicate that promoter activities might be restricted to specific cell lineages, suggesting the need to carefully select optimal promoters for constitutive gene expression in differentiated hESCs.