A recent study published in Scientific Reports highlighted how routine diabetic eye screenings enhanced with artificial intelligence (AI) and deep learning could serve as a non-invasive tool for detecting early signs of cardiovascular disease (CVD).
The potential: A "one-stop" method for risk assessment in diabetes patients.
Let’s start with some background.
CVD remains the leading cause of death globally, with early detection vital for effective management and prevention.
Meanwhile: Diabetic retinopathy (DR) is a common complication of diabetes that involves damage to the small blood vessels in the retina.
- As such: The retina's vascular network is a unique "window" into the body's overall microcirculatory health because it's one of the few places where blood vessels can be easily viewed non-invasively.
How does AI factor into this?
Recent advances in AI and deep learning have enabled automated and highly-precise grading of DR from retinal images, enabling researchers to explore whether retinal microvascular changes could serve as biomarkers for systemic vascular damage—including subclinical CVD.
As such: These advancements highlight the capability of newer learning models to detect subtle retinal changes that might elude healthcare professionals.
And in terms of this research?
Researchers sought to explore whether changes observed in the retina could serve as a biomarker for systemic vascular damage, including that which leads to CVD.
Why this is important: Diabetes patients face a heightened risk for both microvascular complications like DR and macrovascular conditions such as CVD.
Expand on that.
A crucial—yet often overlooked—link between ocular and systemic health is the significant global burden of CVD, which remains the leading cause of mortality.
As such, investigators in the aforementioned study offer a novel approach to early detection that leverages the routine diabetic eye exam—a cost-effective and non-invasive method—to assess CVD risk.
Interesting … tell me more about this study.
The research was a sub-analysis of the PREDICT (Prevalence and Determinants of Subclinical Cardiovascular Dysfunction in Adults with Type 2 Diabetes) study.
Its purpose: Was to determine if retinal changes could predict early signs of heart failure and coronary atherosclerosis.
And its setup?
The study included 255 asymptomatic adults with type 2 diabetes (T2D) who had no prior history of CVD.
- The participants underwent comprehensive cardiac imaging, including echocardiography and coronary computed tomography.
- Their retinal images were then analyzed using advanced deep learning (DL) tools to grade for DR.
And the findings?
Researchers identified a strong association between the presence of even mild DR and subclinical CVD.
Specifically: Individuals with DR were found to have a significantly higher burden of coronary atherosclerosis (as measured by a coronary artery calcium score ≥ 100; OR 2.63) and a greater risk of early heart failure markers, such as concentric left ventricular remodeling (OR 3.11).
- Notably: These associations were independent of traditional cardiovascular risk factors.
So what do these results mean?
Big picture: This data is particularly impactful for T2D patients, who are at a heightened risk for both microvascular and macrovascular complications
- In fact, the results suggest that integrating AI-powered analysis into existing ophthalmic screenings could flag high-risk individuals who might otherwise be missed, leading to earlier cardiac evaluation and intervention.
And, ultimately, the research promotes a more holistic and proactive approach to managing the health of people with diabetes—moving beyond traditional single-organ assessments.
Interesting … any limitations to be aware of?
While the findings are promising, the study did have a few notable limitations:
- The focus was specifically on T2D patients, so the results may not be generalizable to the broader population
- The deep learning tool didn't find direct associations between specific microvascular geometric characteristics and subclinical CVD, suggesting that the relationship is complex
- The use of AI in this context is still a developing field that requires further validation and standardization before widespread clinical adoption
And the expert opinion on this matter?
The results from this research could represent a significant step forward, according to A.S. Alatrany, MD, a lead author of the study.
"Routine diabetic eye screening may serve as a clinically relevant and accessible alternative method to currently advocated screening tools" for detecting CVD in T2D patients, he noted.
- As such: The data highlights the potential to leverage an existing and routine healthcare procedure to provide an additional layer of critical health information.
Anything else to take note of?
Just a few other key points …
This study supports the broader trend of using AI and advanced imaging to turn the eye into a diagnostic tool for various systemic conditions.
And—looking beyond CVD—oculomics is being explored for detecting early signs of Alzheimer's disease, chronic kidney disease, and other neurodegenerative and metabolic disorders.
- With that in mind: This approach could lead to new, non-invasive screening methods that revolutionize how we manage chronic diseases.
Now the take home.
DR screenings, especially when enhanced with AI, show significant potential for detecting early signs of cardiovascular disease in people with T2D.
The presence of retinopathy is a strong, independent predictor of subclinical CVD, offering a unique opportunity for earlier risk stratification and intervention during a routine eye exam.