For Doctors in a Hurry
- Researchers investigated whether a machine learning polygenic risk score and lifestyle factors independently or jointly influence diabetic retinopathy risk.
- This multicenter study analyzed 91,691 participants from the United Kingdom and 1,119 from the Guangzhou Diabetic Eye Study.
- High genetic risk combined with an unfavorable lifestyle doubled retinopathy risk (Hazard Ratio 2.09, 95% Confidence Interval 1.67 to 2.60).
- The authors concluded that genetic risk and modifiable behaviors exert independent and joint impacts on diabetic retinopathy development.
- Clinicians should recommend universal behavioral interventions for retinopathy prevention, as favorable lifestyles significantly reduce risk regardless of genetic background.
Mitigating Microvascular Risk in the Genomic Era
Diabetic retinopathy remains a primary cause of vision loss and serves as a critical indicator of diabetic panvascular disease, a clinical syndrome characterized by the coexistence of microvascular and macrovascular complications across multiple organ systems [1]. While standard management focuses on glycemic and blood pressure control, the marked heterogeneity in patient outcomes suggests that inherited susceptibility and lifestyle factors interact to drive disease progression [2, 3, 1]. Clinical guidelines have begun to evaluate the utility of genetic testing and polygenic risk scores (statistical tools that aggregate the effects of multiple genetic variants to estimate an individual's disease predisposition) to refine risk stratification [4, 5, 6]. Furthermore, the application of machine learning (computational algorithms that identify complex patterns in large datasets to predict clinical outcomes) offers a pathway for more precise prognostic assessments [7, 8, 9]. A recent study now quantifies the synergy between these genetic markers and modifiable behaviors, demonstrating how lifestyle interventions can directly offset inherited ocular risks.
Quantifying Genetic and Behavioral Risk Factors
To evaluate the extent to which lifestyle factors might mitigate genetic predisposition, researchers conducted a multicenter, multiethnic cohort study to develop a machine learning-driven polygenic risk score for diabetic retinopathy. The primary analysis utilized a large dataset of 91,691 participants from the UK Biobank who had prediabetes or diabetes but were initially free of retinopathy. To ensure the findings were applicable across different clinical populations, the authors performed an independent validation using 1,119 participants with diabetes and no retinopathy from the Guangzhou Diabetic Eye Study. The genetic foundation of the score was established through a systematic literature review of major medical databases up to May 1, 2025, to identify candidate loci associated with the condition.
To quantify inherited susceptibility, the researchers integrated 182 literature-derived single nucleotide polymorphisms (specific variations in the DNA sequence that can influence disease risk) into the predictive model. Alongside this genetic assessment, the study evaluated lifestyle adherence through a scoring system focused on four specific behavioral factors: no smoking, optimal weight control, regular physical activity, and healthy sleep. Based on these criteria, participants were stratified into three distinct categories: favorable (3 to 4 factors), intermediate (2 factors), and unfavorable (0 to 1 factor). This dual-pronged approach allowed the investigators to measure the independent and joint impacts of modifiable behaviors and fixed genetic risk on the development of incident retinopathy, providing a clearer picture of how daily habits influence long-term microvascular outcomes.
Synergistic Impact on Incident Retinopathy
The findings revealed that both genetic risk and modifiable lifestyles exert independent and joint impacts on diabetic retinopathy risk. When analyzing these factors separately, the researchers found that participants at high genetic risk had a 37% higher risk of developing incident diabetic retinopathy than those at low genetic risk, a finding that remained consistent regardless of the patient's lifestyle (hazard ratio [HR] = 1.37, 95% CI: 1.18 to 1.60; P < 0.001). Similarly, behavioral choices heavily influenced clinical outcomes. An unfavorable lifestyle was associated with a 49% higher risk of diabetic retinopathy than a favorable lifestyle, an effect observed independently of the patient's underlying genetic predisposition (HR = 1.49, 95% CI: 1.34 to 1.65; P < 0.001).
The most pronounced clinical risk occurred when high genetic susceptibility intersected with poor behavioral habits. The data showed that high genetic risk combined with an unfavorable lifestyle more than doubled the diabetic retinopathy risk (HR = 2.09, 95% CI: 1.67 to 2.60; P < 0.001). Conversely, the study highlighted a significant opportunity for risk modification through clinical intervention. Among participants at high genetic risk, adopting a favorable lifestyle was associated with a 44% lower risk of developing retinopathy (HR = 0.56, 95% CI: 0.42 to 0.75; P < 0.001). This behavioral shift resulted in a reduction of standardized retinopathy rates from 7.5% (95% CI: 7.4% to 7.6%) down to 4.6% (95% CI: 4.5% to 4.7%). These gene-lifestyle synergistic associations were confirmed by replication in the Guangzhou Diabetic Eye Study, reinforcing the validity of the findings across different populations and ethnic groups.
Clinical Implications for Prevention and Counseling
For clinicians managing patients with a strong family history or known genetic predisposition to microvascular complications, these data provide a quantifiable basis for lifestyle counseling. By demonstrating that among participants at high genetic risk, a favorable lifestyle was associated with a 44% lower risk of developing diabetic retinopathy (HR = 0.56, 95% CI: 0.42 to 0.75; P < 0.001), the study translates behavioral changes into meaningful absolute clinical outcomes. Specifically, in the high genetic risk group, adhering to a favorable lifestyle reduced standardized diabetic retinopathy rates from 7.5% (95% CI: 7.4% to 7.6%) to 4.6% (95% CI: 4.5% to 4.7%). These figures allow physicians to offer concrete prognostic expectations when discussing the benefits of smoking cessation, weight management, physical activity, and sleep hygiene with their patients, moving beyond generic advice to targeted risk mitigation.
While the risk reduction was greatest among those with high genetic risk, the clinical utility of behavioral modification extends across the entire patient population. The researchers noted that the protective effects of a healthy lifestyle were not limited to those with a high polygenic risk score (a numerical estimate of an individual's genetic liability to a disease based on multiple genetic variants). Consequently, the findings support universal behavioral interventions for diabetic retinopathy prevention, regardless of genetic background. For the practicing clinician, this underscores that while genetic testing may eventually refine risk stratification, the fundamental pillars of metabolic health remain the primary defense against vision loss for all patients with diabetes or prediabetes.
References
1. Li Y, Liu Y, Liu S, et al. Diabetic vascular diseases: molecular mechanisms and therapeutic strategies. Signal Transduction and Targeted Therapy. 2023. doi:10.1038/s41392-023-01400-z
2. Misra S, Wágner R, Özkan B, et al. Precision subclassification of type 2 diabetes: a systematic review. Communications Medicine. 2023. doi:10.1038/s43856-023-00360-3
3. Yang S, Xin Z, Liu B, et al. Decoding Gene-Lifestyle Synergy in Diabetic Retinopathy Development: A Transethnic Prospective Cohort Study.. Ophthalmology. 2026. doi:10.1016/j.ophtha.2026.04.017
4. Sacks DB, Arnold MA, Bakris GL, et al. Guidelines and Recommendations for Laboratory Analysis in the Diagnosis and Management of Diabetes Mellitus. Diabetes Care. 2023. doi:10.2337/dci23-0036
5. Luo S, Shi C, Wang F, Wu Z. Association between the Angiotensin-Converting Enzyme (ACE) Genetic Polymorphism and Diabetic Retinopathy—A Meta-Analysis Comprising 10,168 Subjects. International Journal of Environmental Research and Public Health. 2016. doi:10.3390/ijerph13111142
6. Sacks DB, Arnold MA, Bakris GL, et al. Guidelines and Recommendations for Laboratory Analysis in the Diagnosis and Management of Diabetes Mellitus. Diabetes Care. 2011. doi:10.2337/dc11-9998
7. Kee OT, Harun H, Mustafa N, et al. Cardiovascular complications in a diabetes prediction model using machine learning: a systematic review. Cardiovascular Diabetology. 2023. doi:10.1186/s12933-023-01741-7
8. Ahmad A, Lim L, Morieri ML, et al. Precision prognostics for cardiovascular disease in Type 2 diabetes: a systematic review and meta-analysis. Communications Medicine. 2024. doi:10.1038/s43856-023-00429-z
9. Fregoso-Aparicio L, Noguez J, Montesinos L, García-García JA. Machine learning and deep learning predictive models for type 2 diabetes: a systematic review. Diabetology & Metabolic Syndrome. 2021. doi:10.1186/s13098-021-00767-9
