Nature | npj Digital Medicine, June 2023

연구 초록

Despite the importance of preventing chronic kidney disease (CKD), predicting high-risk patients who require active intervention is challenging, especially in people with preserved kidney function. In this study, a predictive risk score for CKD (Reti-CKD score) was derived from a deep learning algorithm using retinal photographs. The performance of the Reti-CKD score was verified using two longitudinal cohorts of the UK Biobank and Korean Diabetic Cohort. Validation was done in people with preserved kidney function, excluding individuals with eGFR <90 mL/min/1.73 m2 or proteinuria at baseline. In the UK Biobank, 720/30,477 (2.4%) participants had CKD events during the 10.8-year follow-up period. In the Korean Diabetic Cohort, 206/5014 (4.1%) had CKD events during the 6.1-year follow-up period. When the validation cohorts were divided into quartiles of Reti-CKD score, the hazard ratios for CKD development were 3.68 (95% Confidence Interval [CI], 2.88–4.41) in the UK Biobank and 9.36 (5.26–16.67) in the Korean Diabetic Cohort in the highest quartile compared to the lowest. The Reti-CKD score, compared to eGFR based methods, showed a superior concordance index for predicting CKD incidence, with a delta of 0.020 (95% CI, 0.011–0.029) in the UK Biobank and 0.024 (95% CI, 0.002–0.046) in the Korean Diabetic Cohort. In people with preserved kidney function, the Reti-CKD score effectively stratifies future CKD risk with greater performance than conventional eGFR-based methods.

Introduction

Chronic kidney disease (CKD) is a leading cause of cardiovascular disease and non-communicable disease mortality. CKD prevalence is growing rapidly due to an aging global population and increased prevalence of hypertension and diabetes, two major causes of CKD. Since CKD is an irreversible condition, prevention is a key factor in decreasing CKD-related morbidity and mortality.

The current approach to CKD screening is based on measuring the estimated glomerular filtration rate (eGFR, calculated from serum creatinine) or examining urine for proteinuria. However, recent evidence indicates that these biomarkers are suboptimal for kidney disease early detection. Predicting kidney damage is difficult, especially in people without blood or urine test abnormalities. In addition, risk stratification based on eGFR, which incorporates age and serum creatinine levels, can be misleading in younger, older, pregnant, overweight, or muscular individuals. Similarly, amount of urine proteinuria is also affected by various factors. Moreover, screening adherence tends to be low due to the invasive nature of collecting blood samples.

Retinal photography, a non-invasive and widely utilized diagnostic test, provides information on not only the eye but also the systemic vasculature. The kidney and eye are both highly vascularized organs and share common developmental, physiological, and pathogenic pathways. Damage of one organ often indicates damage to the other which is typically noticeable in hypertensive and diabetic conditions. Recently, artificial intelligence application was shown to be capable of providing biomarker estimates, including creatinine, which also led to effective detection of prevalent CKD.

In this study, we develop a non-invasive CKD risk stratification tool (“Reti-CKD” score) for people with preserved kidney function, hypothesizing that subtle retinal vasculature changes provide information for future CKD development risk. This is done by applying deep learning algorithms trained on 158,216 retinal photographs and incorporating clinical factors. Internal and external validation in the UK Biobank and Korean Diabetes Cohorts show that the Reti-CKD score effectively stratifies CKD development risk and its predictive performance is superior to traditional eGFR-based methods.

Results
Characteristics of the study population

The clinical characteristics of the participants are shown in Table and Supplementary Table. In the Korean health screening data (n = 79,108) used for the development of the deep-learning algorithm, mean age was 49.5 (standard deviation [SD], 11.8) years and mean eGFR was 100.3 (SD, 14.3) mL/min/1.73 m2.