ASN's Mission

To create a world without kidney diseases, the ASN Alliance for Kidney Health elevates care by educating and informing, driving breakthroughs and innovation, and advocating for policies that create transformative changes in kidney medicine throughout the world.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on X

Kidney Week

Abstract: TH-PO001

Retina and Deep Learning-Based CVD Biomarker in Patients with Varying eGFR Levels: Data from the UK Biobank

Session Information

Category: Augmented Intelligence, Digital Health, and Data Science

  • 300 Augmented Intelligence, Digital Health, and Data Science

Authors

  • Cho, Yunnie, Mediwhale Inc., Seoul, Korea (the Republic of)
  • Joo, Young Su, Yonsei University, Seoul, Korea (the Republic of)
  • Rim, Tyler Hyungtaek, Mediwhale Inc., Seoul, Korea (the Republic of)
  • Park, Jung Tak, Yonsei University, Seoul, Korea (the Republic of)
Background

Retinal imaging is a non-invasive method for assessing systemic disease risk. Reti-CVD, a deep learning-based retinal biomarker, predicts atherosclerotic cardiovascular disease (ASCVD) events with performance comparable to coronary artery calcium. Cardiovascular risk assessment is critical for patients with chronic kidney disease, given their higher CVD event risk. This study evaluates the prognostic value of Reti-CVD in patients with varying renal function, as measured by estimated glomerular filtration rate (eGFR).

Methods

A retrospective cohort study was conducted using the UK Biobank, including patients in three eGFR groups: G1 (eGFR ≥ 90; N=6600), G2 (eGFR 60-89; N=5130), and G3-G5 (eGFR < 60; N=183). Survival analysis focused on ASCVD events, with a mean follow-up duration of 9.9 years. Reti-CVD's predictive performance for each eGFR group was assessed using Cox proportional hazards models and concordance statistics. Hazard ratios (HR) of Reti-CVD were adjusted for eGFR groups, age, gender, hyperlipidemia, diabetes, and smoking status.

Results

In G1, Reti-CVD had a C-statistic of 0.715 [95% CI: 0.677 - 0.754], indicating strong predictive capability. The combined model of Reti-CVD, age, and gender improved the C-statistic to 0.727 [0.691 - 0.763]. In G2, Reti-CVD showed a C-statistic of 0.711 [0.679 - 0.744], with the combined model achieving 0.716 [0.684 - 0.748]. In G3-G5, the predictive performance of Reti-CVD was lower (0.605 [0.468 - 0.742]), but improved with age and gender (0.670 [0.521 - 0.819]). Multivariable HR analysis revealed the medium Reti-CVD category had more than double the risk compared to the low category (adjusted HR = 2.08 [1.52 - 2.94]), and the high Reti-CVD category had nearly four times the risk (3.78 [2.48 - 5.76]).

Conclusion

Reti-CVD, performing similarly to CAC in predicting cardiovascular risk, was effective in patients with kidney disease in the UK Biobank. This suggests potential for improved cardiovascular risk monitoring in chronic kidney disease patients.

Figure1. Kaplan-Meier curves in G1 and G2 eGFR group according to Reti-CVD risk categories