Evaluating the potential of retinal photography in chronic kidney disease detection: a review

Diabetic retinopathy as a marker for CKD

The exploration of DR as a potential marker for CKD has been a focal point in several studies, revealing a complex relationship. Yip et al. (2015) conducted a study that highlighted the association between retinopathy and both prevalent and incident end-stage renal disease (ESRD) in a multi-ethnic Asian population. Their findings suggested that retinopathy could serve as an early indicator of subclinical damage in the renal microvasculature, particularly among diabetic individuals. Similarly, Park et al. (2019) examined into the relationship between the severity of DR and the progression of CKD in patients with type 2 diabetes mellitus (T2DM). Their research indicated that the severity of DR, categorized into non-proliferative DR (NPDR) and proliferative DR (PDR), could act as a prognostic indicator for CKD progression. The study emphasized the importance of early DR severity evaluation and rigorous monitoring of renal function and albuminuria in patients with significant DR.

In addition to these studies, other research has also explored the link between retinal changes and renal function. For instance, Bao et al. (2015) examined the association of retinal vessel diameter with CKD in a rural Chinese cohort, observing that narrower retinal arterioles were linked to increased albuminuria, a marker for kidney damage. However, their study did not demonstrate a significant relationship between the presence of retinopathy and the incidence of CKD or albuminuria across the general population. Hwang et al. (2016) found that CKD patients with retinopathy had a heightened prevalence of vascular calcification in the abdominal aorta and iliofemoral artery, suggesting a connection between retinal and renal health.

Moreover, studies like those conducted by Zhao et al. (2021) and Nusinovici et al. (2021) further focused the quantitative relationships between retinal vascular characteristics and clinical indicators of renal function. Their findings disclose significant correlations, such as between DR severity and renal indicators like albumin to creatinine ratio, as well as between wider retinal vascular calibers and heightened risk of diabetic kidney disease, which could be instrumental for early detection and management of diabetes-related kidney complications. Collectively, these studies underline the potential of DR as an early marker for renal dysfunction in diabetic patients. They highlight the need for vigilant monitoring of retinal health to predict and potentially mitigate the progression of CKD in diabetic populations.

The contrasting findings

While numerous studies have highlighted a significant association between retinal changes and CKD, some research presents contrasting findings, highlighting the complexity of this relationship. Although Bao et al. (2015) found an association between retinal arteriovenous ratio (AVR) and albuminuria, suggesting a possible vascular component in renal impairment, however, the absence of a significant connection with CKD and retinopathy in a broader cohort indicates a complex connection rather than a straightforward relationship.

McKay et al. (2018) conducted a study that found retinal microvascular parameters, such as vascular spread, tortuosity, and branching patterns, were not significantly associated with reduced renal function in individuals with type 2 diabetes. Their findings suggested that retinal microvascular parameters did not forecast eGFR degradation over an average of 3 years, challenging the idea of a direct and consistent link between retinal and renal health. Similarly, Keel et al. (2017) aimed to evaluate the relationship between retinal vascular caliber and kidney function in a cohort of Australian children and adolescents with type I diabetes. Their findings indicated that retinal vascular caliber was not significantly associated with microalbuminuria in the sample, adding another layer of complexity to the understanding of the relationship between retinal and renal health.

Meanwhile Wang et al. (2023) provides a more disease-specific insight, linking lower eGFR levels to an increased risk of DR while dissociating eGFR and microalbuminuria levels from diabetic macular edema. This compilation of studies highlights the potential of retinal imaging in indicating systemic vascular health, but also cautions against a one-size-fits-all approach, emphasizing the need for further investigation in the relationship of DR with potential renal complications. These studies emphasize the complex nature of the relationship between retinal changes and CKD. They suggest that while there is a connection, it may be influenced by various factors such as age, diabetes type, and the specific retinal parameters studied, and may not be universally applicable across different populations and conditions.

The promising potential of retinal microvasculature changes in diabetic retinopathy for early detection of CKD

The exploration of retinal microvasculature changes in DR as a predictive tool for CKD has shown promising results. Studies such as those conducted by Wang et al. (2020) and Zhuang et al. (2020) have employed advanced imaging techniques like optical coherence tomography angiography (OCTA) to demonstrate a significant association between retinal vessel density and renal function in patients with type 2 diabetes. Similarly, Yan et al. (2023) observed a higher prevalence of DR in patients with diabetic nephropathy, pointing out the potential relationship between retinal changes and kidney function. The predictive potential of retinal evaluations is further highlighted by research from Wang et al. (2023), which found that low eGFR levels and increasing microalbuminuria levels corresponded with the progression of DR. The integration of AI has also shown promise in enhancing early detection capabilities. For instance, Zhang et al. (2021) developed an AI system that could diagnose CKD and predict its onset using retinal fundus images.

However, it is important to acknowledge certain complexities and challenges, as highlighted by previous works Bao et al. (2015), Keel et al. (2017), McKay et al. (2018) and Wang et al. (2023). These studies concluded that the relationship between retinal microvascular parameters and renal function was not always straightforward, suggesting that various factors may influence this connection (Bao et al., 2015; Keel et al., 2017; McKay et al., 2018; Wang et al., 2023). In conclusion, while there are challenges to be addressed, the collective findings from these studies suggest that the potential of utilizing retinal microvasculature changes in DR as a means of detecting CKD is promising. These findings promote further exploration and validation to fully harness retinal assessments in the early detection and management of CKD in diabetic patients.

To provide a concise overview of the findings and relationships identified in the literature, Table 1 summarizes key aspects of the studies, including the retinal vasculature features observed, the kidney pathologies identified, and the established relationships between these two.

Retinal vasculature

Micro-vessels, with a luminal diameter of less than 300 micrometers, are crucial in controlling tissue blood flow and influencing systemic vascular resistance, a function intrinsically related to endothelial performance. Multiple pathological events can both induce and result from endothelial malfunction, subsequently affecting the micro-vessels (Farrah et al., 2020). The inception and implications of microvascular disease include a range of pathological changes and risk factors. For instance, endothelial dysfunction is associated with a variety of risk factors including decreased kidney function, high blood pressure, smoking, elevated levels of glucose and insulin, and dyslipidemia. Meanwhile, micro-vessel dysfunction is characterized by a series of pathological alterations including decreased fibrinolysis, reduced ability to respond to vasodilators, heightened platelet reactivity, vascular remodeling, and rarefaction, all of which collectively contribute to an increase in vascular resistance and a reduction in blood supply.

Modifications in microvascular structure and function are significant factors in the initiation and progression of hypertension, diabetes, CKD, and cardiovascular disease (CVD) (Deanfield, Halcox & Rabelink, 2007; Houben, Martens & Stehouwer, 2017; Stehouwer, 2018). Remarkably, these alterations manifest before any perceptible damage to the end organs and seem to be correctable (Halcox et al., 2002; Remuzzi et al., 2016). Additionally, microvascular impairments in peripheral tissues reflect similar malfunctions in visceral organs (Anderson et al., 1995; Bonetti et al., 2004), justifying the examination of easily accessible micro-vessels, like those in the eye. The clarity of the eye’s media enables direct observation of the microvasculature, which can be compromised by systemic conditions such as hypertension, diabetes, and CKD.

The relationship between retinal vascular alterations and CKD has been a subject of multiple studies. Although there has been investigation into the connection between the caliber of retinal vessels and the onset of CKD, findings remain inconclusive. For instance, a study conducted by Yau et al. (2011) did not find any correlation between retinal microvascular caliber and the onset of stage 3 CKD in the general populace; however, it did find that among white individuals, narrower arterioles were correlated with an increased likelihood of developing stage 3 CKD. Similarly, research by Sabanayagam et al. (2011) did not establish any significant correlation between the diameter of retinal vessels and the likelihood of reduced kidney function. As such, both studies implied that although there might be shared mechanisms between retinal vessel diameters and CKD, they are not directly causally related. In a nutshell, while alterations in retinal vasculature may signal CKD in specific demographic groups, further studies are necessary to comprehensively understand the connection between changes in retinal vasculature and CKD.

Retinal imaging techniques and CKD detection

The exploration of retinal imaging as a diagnostic tool for CKD has attracted significant attention in recent research, revealing the promising potential of detailed analysis of retinal vasculature in the early detection and monitoring of CKD. Progressions in retinal imaging techniques, such as fundus photography and OCT, have equipped researchers and clinicians with powerful tools to visualize and assess the retinal vasculature with exceptional detail. These techniques facilitate the identification of potential markers that might indicate the presence or risk of developing CKD. A multitude of studies have explored the complex relationship between retinal and renal health, leveraging these advanced imaging techniques alongside machine learning algorithms.

Innovations in retinal imaging, particularly fundus imaging and OCTA, have been crucial in advancing our understanding of the link between retinal microvascular changes and CKD. As presented in Table 1, Zhuang et al. (2020) demonstrated that OCTA could detect preclinical microvascular impairments in DR, which are indicative of CKD. The study highlighted OCTA’s ability to measure retinal vessel density and thickness, revealing a significant correlation between reduced vessel density in the superficial vascular complex and impaired renal function. In addition, Xu et al. (2020) expanded the scope of retinal vascular analysis using digital fundus photography to include peripheral vascular calibers and arteriolar geometries, finding strong associations with microalbuminuria, an early marker of renal dysfunction in type 2 diabetes. Kang et al. (2020) leveraged deep learning models with retinal fundus imaging to detect early renal function impairment, achieving area under curve (AUC) of 0.81, which highlights the potential of retinal imaging in identifying systemic cardiovascular risks. Zhao et al. (2021) developed a deep learning methodology for retinal vessel segmentation, which, when combined with OCTA, revealed significant correlations between the intricacies of retinal vessel structure, such as fractal dimension, and various markers of renal function across various stages of DR.

Meanwhile, Frost et al. (2021) used OCTA to observe retinal capillary rarefaction, finding it to be associated with reduced eGFR in hypertensive patients, suggesting that retinal microvascular health reflects systemic vascular integrity. Paterson et al. (2021) employed the Vessel Assessment and Measurement Platform for Images of the Retina (VAMPIRE) software to quantify retinal microvascular parameters, discovering a strong link between reduced fractal dimension and increased odds of albuminuria, independent of diabetes and blood pressure. Peng et al. (2021) identified a correlation between retinal deep vascular plexus vessel density and early cognitive deficiencies in CKD patients, suggesting retinal neurovascular biomarkers as potential indicators of cognitive impairment. Yong et al. (2022) measured OCTA metrics across different CKD retinal characteristics and found significant differences in vascular density and perfusion density, which varied based on the underlying cause of CKD. Yan et al. (2023) documented a substantial relationship between diabetic nephropathy and retinal microvascular changes, such as DR and hard exudates, using a combination of color fundus imaging, OCT, and fluorescence angiography.

Additionally, Wang et al. (2023) conducted a three-year study using fundus photography and swept-source OCT to explore the relationship between renal function and DR, finding that lower eGFR and higher microalbuminuria levels were associated with the progression of DR, although no direct link to diabetic macular edema was established. These studies collectively enlighten the critical role of retinal imaging in detecting and monitoring CKD, offering a window into the systemic microvascular changes associated with the disease. An interesting technique has been employed by Mustafar et al. (2023) where they combined fundus photography for assessing retinal changes, OCT for macular volume measurements, and blood sample analyses for cardiac biomarkers, when examining the relationship between retinal vessel changes and cardiac biomarker levels in patients with various stages of CKD. The findings suggest a negative correlation between eGFR and retinal vessel tortuosity, a positive correlation between proteinuria and central retinal venular equivalent (CRVE), and various associations with cardiac biomarkers like high-sensitivity C-reactive protein. These findings indicate the potential of using retinal and cardiac biomarkers as non-invasive tools for assessing CKD.

Figure 3 illustrates the application of retinal imaging—fundus photography and OCT, as investigative tools for the detection of CKD, showcasing their potential utility in identifying biomarkers associated with renal pathology.

Promising potential of fundus imaging in early detection of CKD

The exploration of fundus images as a diagnostic tool for CKD has shown promising results. Fundus imaging, being a non-invasive technique, offers a unique advantage in facilitating early detection of CKD. A significant aspect of this research is the established correlation between retinal pathologies and key kidney function markers. Studies by Edwards et al. (2005), Baumann et al. (2010), and Sabanayagam et al. (2009) have demonstrated a direct association between conditions like retinopathy and arteriolar abnormalities with serum creatinine levels and eGFR. This suggests that retinal changes can serve as early indicators of kidney health. Furthermore, the impact of vascular changes in the retina, such as arteriolar narrowing, has been linked to microvascular damage associated with CKD. This is supported by the research of Yau et al. (2011) and Ooi et al. (2011) and is further evidenced in the work of Grunwald et al. (2010). In addition, Kang et al. (2020), who attempted to automate the detection achieved an impressive area under curve (AUC) of 0.81 in detecting renal function impairment using deep learning models applied to fundus images.

Peng et al. (2020), Peng et al. (2021) reveal a complex relationship between retinal vascular health and kidney function in CKD. Their previous work Peng et al. (2020) found that blood pressure variability in CKD patients negatively correlates with retinal vessel density, suggesting that retinal microvascular integrity may reflect broader vascular health. However, the latter showed no significant link between retinal biomarkers and cognitive impairment in CKD, indicating that while retinal changes mirror microvascular damage, they do not necessarily predict cognitive decline in these patients. This highlights the potential yet complicated role of retinal imaging as an indicator of systemic vascular conditions.

The versatility of fundus imaging in adapting to various imaging techniques and algorithms for tailored retinal vessel segmentation, as demonstrated in the work of Zhao et al. (2021) and Zhuang et al. (2020), further emphasizes its compatibility and potential for widespread clinical application. The evidence in Table 2 clearly shows that retinal imaging is becoming important for detecting and managing CKD early. This approach, particularly when combined with advanced analytical methods like deep learning, holds significant promise. However, the need for continued research and validation is critical to fully harness and refine these diagnostic capabilities in the clinical setting.