Study on the prevalence of diabetic retinopathy in Pudong New Area and the predictive value of estimated glucose disposal rate: a cross-sectional study

Jia Gao, Kang Wu, Xiaonan Wang, Zhitao Li, Xiaolin Liu, Jiahui Song, Yang Liu, Hua Qiu, Wei Feng, Qian Xu, Shaotan Xiao, Yi Zhou, Xiaonan RuanFrontiers in Endocrinology 2025 · DOI

Abstract

Diabetic retinopathy (DR) remains a major cause of vision loss in patients with type 2 diabetes mellitus (T2DM). This study investigates the prevalence of DR in the Pudong New Area of Shanghai and evaluates the estimated glucose disposal rate (eGDR) as a non-invasive marker for insulin resistance and a predictor of DR risk. Our findings reveal a significant negative linear association between eGDR and DR risk, suggesting that eGDR could serve as a valuable target for early prediction and clinical intervention.

Introduction

Diabetic retinopathy (DR) is a leading cause of preventable blindness globally. While early detection is critical, the initial stages of DR are often asymptomatic. Insulin resistance (IR) is a known driver of the microvascular damage underlying DR. However, the gold standard for measuring IR—the hyperinsulinemic-euglycemic clamp—is impractical for large-scale clinical use due to its invasive and costly nature.

This research, published in Frontiers in Endocrinology (June 2025), explores the utility of the estimated glucose disposal rate (eGDR) as a simple, non-invasive surrogate for IR. By analyzing a large cohort in Shanghai’s Pudong New Area, the study aims to validate eGDR as an "early warning system" for diabetic eye complications.

Methods

The researchers conducted a cross-sectional study involving 2,327 patients with T2DM in Pudong, Shanghai, between May and September 2023. The study utilized a multi-stage stratified random cluster sampling approach.

Diagnostic Procedures

DR was staged using a fully automatic non-mydriatic fundus camera (Airdoc AI-FD16aF). This technology allows for rapid, AI-assisted screening of retinal health by capturing high-resolution images of the fundus.

Figure 1: Fundus photographs showing normal control (A), non-proliferative DR (B), and proliferative DR (C).

eGDR Calculation

The eGDR was calculated using a validated formula incorporating waist circumference, hypertension status, and glycemic control: $$eGDR = 21.158 - [0.09 \times \text{waist circumference (cm)}] - [3.407 \times \text{hypertension (yes/no)}] - [0.551 \times \text{HbA1c (\%)}]$$

Statistical Analysis

Restricted Cubic Splines (RCS) were employed to visualize the dose-response relationship between eGDR and DR. Logistic regression models, adjusted for confounders such as disease duration, sleep quality (PSQI), and cognitive function (MoCA-B), were used to determine odds ratios (OR).

Results

Prevalence of DR

The study found a DR prevalence of 7.39% in the Pudong T2DM population. Most cases were identified as non-proliferative DR. This prevalence is lower than the global average (~22%), which the authors attribute to effective community-based diabetes management programs in Shanghai that facilitate earlier detection.

eGDR as a Predictor

A significant negative linear relationship was observed between eGDR and the risk of DR. For every one standard deviation increase in eGDR (indicating improved insulin sensitivity), the risk of DR decreased by 40.4%.

Patients in the lowest eGDR tertile—those with the highest levels of insulin resistance—were nearly three times more likely to have DR compared to those in the highest tertile. The predictive accuracy of eGDR was high, with an Area Under the Curve (AUC) ranging from 0.777 to 0.781.

Discussion

The linear nature of the risk relationship suggests that any improvement in metabolic health and insulin sensitivity provides a proportional benefit in reducing DR risk. Because eGDR integrates obesity, cardiovascular health, and glycemic control, it serves as a holistic metabolic "snapshot" that outperforms individual markers in predicting microvascular damage.

Limitations

As a cross-sectional study, these findings demonstrate association rather than causation. Additionally, lifestyle data relied on self-reported questionnaires, which may introduce recall bias. While AI-assisted fundus imaging is highly effective, it may not capture the most subtle lesions detectable by dilated-pupil specialist exams.

Conclusion

The eGDR is a powerful, low-cost, and non-invasive tool for predicting diabetic retinopathy risk in urban T2DM populations. Its strong linear association with DR risk makes it an ideal target for clinical monitoring and early intervention strategies aimed at preserving vision in diabetic patients.

Summary updated

Abstract

There is a significant negative linear association between eGDR and the risk of DR, and eGDR is expected to serve as a target for DR prediction and intervention.
BibTeX
@inproceedings{gao2025study,
  author = {Jia Gao and Kang Wu and Xiaonan Wang and Zhitao Li and Xiaolin Liu and Jiahui Song and Yang Liu and Hua Qiu and Wei Feng and Qian Xu and Shaotan Xiao and Yi Zhou and Xiaonan Ruan},
  title = {{Study on the prevalence of diabetic retinopathy in Pudong New Area and the predictive value of estimated glucose disposal rate: a cross-sectional study}},
  booktitle = {Frontiers in Endocrinology},
  year = {2025},
  doi = {10.3389/fendo.2025.1571488},
}
Figure