TL;DR
This study leverages Optical Coherence Tomography Angiography (OCTA) to analyze retinal microvascular changes as a potential early diagnostic marker for Coronary Artery Disease (CAD). Researchers found that individuals with CAD exhibit significantly reduced retinal vessel density, length, tortuosity, and bifurcation complexity, particularly in the left eye, suggesting OCTA's promise as a non-invasive tool for early CAD detection.
What Problem Does This Paper Address?
Coronary Artery Disease (CAD) is a major global killer, and its early detection is crucial for effective prevention and treatment. Traditional diagnostic methods like coronary angiography are invasive, costly, and not ideal for widespread screening. The paper addresses the need for a non-invasive, accessible, and accurate method for early CAD risk assessment. The authors highlight that the retina, with its similar vascular structure to the heart, can serve as a "window" into systemic vascular health. Previous research has shown correlations between retinal microvascular abnormalities and cardiovascular issues, but this study aims to specifically quantify these changes using OCTA and evaluate their diagnostic potential for CAD.
Methods
The study employed Optical Coherence Tomography Angiography (OCTA) to capture high-resolution images of the retinal microvasculature. A key methodological contribution is the use of a pre-trained adaptive feature fusion multi-task network (VAFF-Net) for simultaneously segmenting retinal vessels and the foveal avascular zone (FAZ). This network is designed to extract features from en-face OCTA projections across different retinal layers (superficial vascular complex - SVC, and deep vascular complex - DVC).
From these segmented images, seven quantitative parameters were extracted:
- Vascular-related: Vascular Area Density (VAD), Vascular Length Density (VLD), Vascular Fractal Dimension (VFD), Vascular Tortuosity (VT), and Vascular Bifurcation Number (VBN).
- FAZ-related: FAZ Area (FA) and FAZ Axis Diameter Ratio (FR).
These parameters were then compared between a cohort of CAD patients and healthy controls using independent t-tests. Statistical significance was set at p < 0.01 for most analyses, with a more lenient p < 0.05 used for ETDRS grid analysis.
Datasets
The study utilized OCTA images from a cohort of 747 participants, comprising 332 patients diagnosed with CAD and 415 healthy controls. These images were collected at the Department of Cardiovascular Medicine, First Affiliated Hospital of Ningbo University, between October 2022 and October 2023. The OCTA images were acquired using the AngioVue system on an RTVue XR Avanti SD-OCT system (Optovue, USA).
Key Insights and Ideas
- Retinal Microvasculature as a CAD Biomarker: The study strongly supports the idea that changes in the retinal microvasculature, observable via OCTA, are directly associated with the presence of CAD. This opens up a non-invasive avenue for early detection.
- Left Eye Dominance: A particularly striking finding is the greater sensitivity of the left eye (OS) to microvascular changes associated with CAD compared to the right eye (OD). This asymmetry is hypothesized to be due to shared hemodynamic and anatomical pathways with the left coronary artery.
- Specific Parameter Reductions in CAD: Patients with CAD consistently showed lower VAD, VLD, and VBN, indicating a less dense and less complex vascular network.
- VAFF-Net's Efficacy: The VAFF-Net model demonstrated robust performance in segmenting retinal vessels and FAZ, highlighting the potential of advanced AI for analyzing complex OCTA data.
Key Results
- Demographic Differences: CAD patients were significantly older and had a higher prevalence of hypertension and diabetes compared to controls. Smoking and alcohol consumption were also more prevalent in the CAD group.
- OCTA Parameter Differences (Global Analysis): In a global analysis (combining both eyes), VAD, VLD, VFD, VT, and FR showed statistically significant differences between CAD patients and controls (p < 0.01).
- OCTA Parameter Differences (Eye-Specific): The left eye (OS) analysis revealed significant differences in VAD, VLD, and VFD. While VLD and VFD were also significant in the right eye (OD), the overall sensitivity was lower.
- ETDRS Grid Analysis: This analysis showed significant spatial heterogeneity in parameter differences, with VAD, VLD, and VFD exhibiting extensive regions of significant difference (p < 0.05) across the global analysis, reinforcing the widespread impact of CAD on retinal vasculature.
Limitations and Future Work
The authors acknowledge several limitations:
- Sample Size and Data Collection: The participant numbers were relatively small, and data from only one eye were used for some participants.
- Longitudinal Data: The study did not track changes in retinal microvascular parameters over time or during disease progression, which would be valuable for understanding disease trajectory.
- Axial Length Data: The absence of axial length data could have introduced bias in the estimation of vascular parameters, particularly in myopia.
- Risk Factor Information: More comprehensive patient information is needed to fully determine the role of various risk factors.
Future work should focus on larger, longitudinal studies, incorporating axial length measurements, and collecting more detailed patient data to build robust, personalized cardiovascular risk models. The integration of OCTA biomarkers with clinical and genomic data is also a promising direction.
Summary updated