Diabetes has become one of the most important chronic diseases threatening human health worldwide due to its high incidence, rapid growth, and serious harm (Saeedi et al., 2019). Approximately 30%–40% of diabetic patients will develop at least one complication within 10 years of disease onset (Liew et al., 2014). Diabetic retinopathy (DR) ranks first among retinal vascular diseases that lead to blindness in working-age populations (Jee et al., 2013). DR can be asymptomatic and presents no obvious clinical symptoms in early stages, a complex disease with unclear pathogenesis and a high risk for disability and blindness (Hu et al., 2012).

Although DR is a serious and irreversible cause of blindness, it is also preventable. In the early stages of DR, improving microcirculation and controlling blood glucose through treatment can help slow the progression of the disease, preserving the patient’s vision and reducing the risk of blindness (Li et al., 2021a). Among the known risk factors for DR, diabetes severity and blood glucose control are the most significant (Thomas et al., 2015; Wat et al., 2016). However, some patients may develop DR even if their blood glucose is under good control and the duration of diabetes is relatively short (Silva et al., 2018). In contrast, some patients with poor blood sugar control and a longer duration of diabetes may not develop DR. It is especially important to research factors related to the progression of DR (Li et al., 2021a). Therefore, exploring the factors related to the progression of DR in diabetic patients will enhance the understanding of the mechanism of DR.

The global prevalence of myopia is increasing, and it is projected to affect 50% of the global population by the year 2050 (Holden et al., 2016). Diabetes and myopia have indeed become global health issues, and they are expected to continue to increase in the future (Yuan et al., 2022). Both diseases can lead to vision impairment and can interact with each other (Su et al., 2023). Previous studies have revealed that diabetic patients with myopia have a lower probability of developing DR than those without myopia (Wang et al., 2016; Lin et al., 2020; Yuan et al., 2022). It is believed that myopia may serve as a protective factor against the onset and progression of DR (Man et al., 2019; Lin et al., 2020). an increased axial length (AL) plays a critical role in this protective effect (Dogru et al., 1998; Wang et al., 2016). However, this speculation is still controversial, and some studies suggest that there is no correlation between myopia and DR (Xie et al., 2008; Ganesan et al., 2012). This study hypothesizes that high myopia may modify the patterns of microvascular damage in diabetic patients through structural changes in the choroid. The primary objective was to evaluate retinal and choroidal perfusion changes in diabetic patients with high myopia using SS-OCTA and AI-driven analysis, aiming to identify early biomarkers for DR risk stratification.

In this study, we systematically evaluated retinal and choroidal microvascular alterations in diabetic patients with high myopia using UWF-SS-OCTA combined with artificial intelligence AI algorithms. Several studies have investigated the characteristics of retinal and choroidal changes in myopic diabetic patients (Chao et al., 2016; Man et al., 2019; Lin et al., 2020), but none have been comprehensive. Specifically, recent studies have reported that different phenotypes of macular ischemia might account for different severities of visual function loss (Usman, 2018; Tsai et al., 2023). However, some studies only studied the characteristics of retinal changes, while others only explored the characteristics of choroidal changes without studying the related risk factors. It is unclear whether myopia has a protective effect on DR, as well as its role. According to the present study, diabetic patients with high myopia had decreased retinal structure and choroid perfusion at the early stages, which was especially noticeable in the middle and large vessels of the choroid. Visual impairment was closely related to a decrease in ChV perfusion. The findings of the study will enable further understanding of the pathophysiology of diabetes with myopia and explore the morphological parameters of early visual function impairment.

The interaction between DM and macular microvascular reduction has received a great deal of attention in recent years (Barot et al., 2013; Shin et al., 2019). In our study, the retinal whole vascular density and the superficial vascular density showed a decreasing trend in DR and NDR with HM when compared with the diabetic group, while the deep vascular density was increased in the DR group. Specifically, a study found that vascular changes in eyes with no to early DR were present primarily in the deeper vascular layers, whereas in eyes with advanced DR, the opposite was observed (Ashraf et al., 2020). Another study found that diabetic patients exhibited significantly lower retinal VD in deep vascular complexes than controls. These findings suggest that the deep vascular density is affected differently than the superficial vascular density in DR (Li et al., 2021b). In the current study, patients with DR had a long course of disease, and some had advanced stages. We documented that increased DVD in eyes with NPDR might be a confounding factor for revealing the relationship between DR and degrees of severity. Future studies to explore how different stages of DR are associated with specific layers in the retina and consequently impact visual function are warranted. Indeed, the decrease in RVD was found to be associated with age and duration of diabetes (Thomas et al., 2015; Wat et al., 2016). This suggests that these factors may contribute to the development and progression of DR. These factors can interact with and contribute to changes in retinal vascular density. Regular eye examinations are essential for individuals with diabetes, especially as they age and their duration of diabetes increases. Monitoring both intraocular pressure and retinal vascular density can help healthcare professionals assess the health of the eye and make necessary recommendations for managing and preserving vision in diabetic individuals.

The choroid is a highly vascularized tissue that plays a multifunctional role in the eye (Nickla and Wallman, 2010). Its main function is to supply oxygen and nutrients to the outer retina (Hurley, 2021). The choroid also provides metabolic exchange between the blood and the outer retinal layers, which is essential for the proper functioning of the retina (Linsenmeier and Padnick-Silver, 2000). Previous studies have suggested that the choroid is implicated not only in the pathogenesis of many chorioretinal diseases but also in the progression of DR (Wang et al., 2020) and myopia (Wang Y. et al., 2023) development. Studies have demonstrated that choroidal thinning accompanies the development and progression of myopia (Read et al., 2013; Wang et al., 2020), and a close link was established between eye growth and choroidal thickness changes (Bartol-Puyal et al., 2019). In diabetic patients, choroidal structural changes have been found to correlate with the severity of DR (Wang H. et al., 2023). A study by Wang et al. (Wang H. et al., 2023) used SD-OCTA to show that diabetic patients had a lower choroidal perfusion density than healthy controls. In addition, Wang et al. (Wang et al., 2020) reported decreased choroidal perfusion in diabetic patients without DR using SS-OCTA. However, Dai et al. (Dai et al., 2020) reported that diabetic patients’ choriocapillaris blood FDs were significantly higher than those of controls. Our current study showed that both CT and the CVI decreased significantly with disease severity in individuals with diabetes, and visual impairment was closely related to choroidal perfusion among the groups. Interestingly, CPV did not change significantly in patients with high myopia. Zhao et al. (Zhao et al., 2023) also found no statistically significant difference in CPV between DM patients and a control group. Diabetic choroidopathy (DC) might preferentially affect larger choroidal veins (Haller’s layer) initially before medium (Sattler’s layer) and small (choriocapillaris) arterioles (Foo et al., 2020). It is possible that in diabetic patients with myopia, the occurrence of DR is delayed or reduced because of changes in the structure and perfusion of the retina and choroid, as well as a decrease in tissue oxygen demand.

Data on the relationship between risk factors and the CVI in diabetic populations are still scarce. A longer duration of diabetes has been associated with a decreased CVI of Sattler’s layers and subfoveal choroidal volume, indicating a decline in choroidal perfusion (Foo et al., 2020). Several studies have reported that CT is associated with age and AL (Read et al., 2013; Agrawal et al., 2016; Xiong et al., 2017), and the proportion of CVI has not been consistently demonstrated to change with age and AL. He et al. (Xuan et al., 2023) reported negative correlations between the CVI and age, whereas Zhou et al. (Zhou et al., 2020) found no significant association between the CVI and age in healthy adults by univariable analyses. Furthermore, the significant associations of the CVI with age and AL disappeared when CT was included as an independent variable in a healthy adult population study (Agrawal et al., 2016). The purpose of this study was to examine the choroidal changes in DR and high myopia patients, as well as to investigate the factors influencing choroidal changes. Our results indicated that both CT and the CVI were associated with age, AL and SE in diabetic patients. However, based on the multiple linear regression, after adjusting for all interaction factors, only age and SE were related to CT and CVI. The findings of this study contribute to our understanding of the factors influencing CT and CVI in diabetic patients with high myopia. Nevertheless, more research is needed to confirm these results and explore the underlying mechanisms of the observed relationships. The automated analytical workflow in this study demonstrates that AI-driven OCTA image interpretation significantly improves the detection efficiency of diabetic microvascular damage. By integrating deep learning-extracted features with traditional biometric parameters (e.g., CVI and CT), multimodal risk prediction models could be constructed for future applications. Recent studies indicate that Transformer-based vision models exhibit high sensitivity in identifying early-stage microvascular abnormalities in DR (Schmidt-Erfurth et al., 2018), offering a new direction for developing intelligent decision-making systems to personalize DR screening frequency.

We acknowledge several limitations of the present study. First, our study could not determine causality due to the inherent limitations of cross-sectional studies. The question of whether microvascular alterations in myopic eyes occur before or after DM remains uncertain. Longitudinal analysis should be performed to further understand the relationships. Second, due to the poor cooperation of the diabetic patients, we only selected the choroidal blood supply of the macular area and the peripheral area in the scan mode of 12 × 12 mm area and did not select a larger mode, which may have resulted in a loss of important peripheral lesion information. Third, the exclusion criterion in the current study was relatively stringent, aiming to explore the characteristics of choroidal changes in patients with simple diabetes and patients with myopia. Therefore, patients with other systemic diseases were excluded from the study. Nevertheless, the associations between these variables may still reflect the intrinsic links or underlying mechanisms.

This study evaluated and quantified diabetes-related retinal and choroid perfusion changes in individuals with and without high myopia. We found that the retina and choroid of diabetic patients with high myopia had a lower vessel density and thinner thickness, even lower than those with DR. The integration of deep learning algorithms for automated OCTA image segmentation and ResNet-50-based vascular analysis significantly enhanced the detection sensitivity of early microvascular alterations, particularly in CVI quantification. Diabetic eyes with high myopia had a lower macular CVI among the three groups, although there were no significant differences in CPV, indicating that AI-assisted monitoring of the CVI could facilitate precise risk stratification of the diabetic population and optimize personalized DR screening strategies. When analyzing choroidal alterations in diabetic eyes, age and spherical equivalent should be considered. Future studies should explore multimodal AI models combining structural and vascular parameters to predict disease progression and guide clinical interventions.