Atherosclerotic cardiovascular disease (ASCVD), including coronary atherosclerotic heart disease (CHD), ischemic stroke, and peripheral arterial disease, remains the leading global cause of mortality (1–3). CHD can result in myocardial infarction and sudden cardiac death, ischemic stroke can lead to aphasia, prolonged immobility, and secondary cerebral hemorrhage, and peripheral arterial disease in the lower extremities may cause impaired mobility or necessitate amputation (4–7). These conditions severely impair patients’ quality of life, impose significant economic burdens, and increase mortality rates. Accurate risk stratification and early identification of individuals at high risk are essential to prevent disease initiation, mitigate progression, and reduce associated mortality.

The pathophysiological hallmark of ASCVD is the formation and advancement of atherosclerotic plaques, with dyslipidemia, particularly elevated low-density lipoprotein cholesterol (LDL-C), recognized as a principal risk factor (1–3). Evaluating ASCVD risk in individuals with dyslipidemia, predominantly characterized by elevated LDL-C levels, is crucial for implementing targeted interventions and enhancing clinical outcomes.

Previous ASCVD risk assessment tools have been developed (8–11), but they do not specifically cater to the needs of high-risk individuals with dyslipidemia. The predictive variables used, such as blood pressure, lipid levels, and waist circumference, exhibit considerable fluctuation. This highlights the importance of identifying stable, reliable biomarkers for risk evaluation. The retina, recognized as a window into microvascular health, has gained attention as a potential indicator for coronary artery alterations. Recent studies reveal significant differences in retinal vascular features between CHD patients and controls, with these differences correlating with the extent of coronary artery stenosis (12, 13). This evidence suggests an association between retinal microvascular changes and macrovascular conditions. Thus, the current study investigates the feasibility of assessing ASCVD risk in individuals with dyslipidemia by analyzing retinal vascular characteristics.

Optical coherence tomography angiography (OCTA) represents an advanced, non-invasive imaging technique that enables rapid and quantitative evaluation of retinal vascularity (14). Compared to traditional methods, such as color fundus photography and fundus fluorescein angiography, OCTA offers clear advantages. The former lacks quantitative capabilities, while the latter necessitates the injection of a contrast agent, making it invasive. OCTA provides key metrics, including retinal vascular density (VD), retinal nerve fiber layer (RNFL) thickness, and foveal avascular zone (FAZ) area (15), which together allow for a more thorough examination of retinal microvasculature. Previous investigations utilizing OCTA to assess retinal vascular changes have demonstrated significant alterations in the fundus vessels of patients with hypertension (16), diabetes (17), and dyslipidemia (18) when compared to controls. These conditions are also recognized as risk factors for ASCVD. The aim of this study is to investigate the role of retinal vascular alterations assessed by OCTA as biomarkers for ASCVD risk assessment in dyslipidemia patients.

A total of 261 dyslipidemic patients were initially assessed for eligibility in this study, all of whom were of yellow races. Fourteen patients were excluded due to intolerance to mydriasis, preventing OCTA examination; nine were excluded for having severe cataracts or a history of fundus hemorrhage; four had malignant tumors; and three were in the active phase of autoimmune diseases. Consequently, the final cohort consisted of 231 participants (458 eyes in total), including 170 males (73.6%) with a mean age of 63.68 ± 11.61 years (Figure 2).

Participants were classified into two groups based on the presence or absence of ASCVD: the non-ASCVD group (n = 132) and the ASCVD group (n = 99). A comparison of baseline characteristics between the groups is provided in Table 1. The ASCVD group was older and had a higher incidence of hypertension, increased use of antihypertensive medications, hypoglycemic agents, statins or PCSK9 inhibitors, antiplatelet agents, and anticoagulants, alongside elevated MSBP and HbA1c levels. Conversely, the non-ASCVD group exhibited higher levels of TC, LDL-C, and eGFR. No significant differences were noted between the groups concerning sex, diabetes, hyperuricemia, OSAHS, smoking status, BMI, MDBP, HDL-C, TG, Lp(a), or hemoglobin.

In the optic disc region, the ASCVD group exhibited a reduced RNFL thickness compared to the non-ASCVD group [111.50[102.36;118.61] vs. 107.00[97.00;118.00], P = 0.040]. No significant differences were found in the VD of total or small vessels between the two groups in this region (Table 2).

In the macular region, VD of the parafovea [48.05[44.53;51.30] vs. 45.32[41.90;49.10], P = 0.001] and perifovea [47.40[43.53;49.92] vs. 45.20[42.20;47.70], P = 0.001] in the SCP was significantly lower in the ASCVD group. OCTA imaging revealed a decrease in retinal vasculature density in the ASCVD group, confirmed by processed color images showing more extensive blue areas and fewer green areas (Figure 3). No significant difference was found in the VD of the parafovea [50.25[47.43;53.20] vs. 49.90[47.40;53.10], P = 0.944] or perifovea (43.94 ± 5.51 vs. 43.83 ± 5.36, P = 0.885) within the DCP between the two groups. Retinal thickness in the parafovea (321.88[309.56;332.44] vs. 316.50[303.75;329.25], P = 0.028) and perifovea [281.25[272.88;286.69] vs. 275.50[265.00;286.25], P = 0.052] was reduced in the ASCVD group, with only the parafoveal thickness difference reaching statistical significance (Table 2).

In the FAZ, no significant differences were observed between the groups in area, perimeter, or retinal thickness. However, the FD was significantly lower in the ASCVD group compared to the non-ASCVD group [49.63[45.40;53.31] vs. 47.59[43.13;50.74], P = 0.003], which was confirmed by statistical analysis (Table 2). Imaging also showed a decrease in FD in the ASCVD group, further supporting this finding (Figure 3).

A comparison of OCTA results between the two groups identified RNFL thickness, SCP parafovea VD, SCP perifovea VD, macular parafovea thickness, and FD as key parameters for constructing an ROC curve to evaluate their predictive value for ASCVD in dyslipidemia patients (Table 3 and Figure 4A). The analysis confirmed that all five parameters were indicative of ASCVD risk in this population. The optimal thresholds, determined via the Youden index, were found to be 47.275% for SCP perifovea VD, 50.850% for SCP parafovea VD, and 48.045% for FD.

Anatomically, the macular superficial layer, extending from medial to lateral, comprises the FAZ, SCP parafovea, and SCP perifovea, all of which are interconnected vascular regions (Figure 1B). Given the relatively intact terminal vascular ring of the superficial FAZ, FD is defined as the VD within a 300 μm-wide ring surrounding the superficial FAZ. As such, FD, along with SCP parafovea VD and SCP perifovea VD, serves as an indicator of superficial retinal VD. These three parameters were consolidated into a composite measure—macular superficial vessel density (MSVD)—to assess ASCVD risk in dyslipidemia patients. ROC curve analysis revealed an AUC of 0.647 for MSVD (95% CI: 0.576–0.718), which surpassed the AUCs for FD, SCP parafovea VD, and SCP perifovea VD. The maximum Youden index value of 0.258 indicated an optimal threshold of 143.22% (Figure 4B). DeLong test results indicated no statistically significant differences in AUC values when comparing MSVD with FD, SCP parafovea VD, SCP perifovea VD, RNFL thickness, and macular parafovea thickness (Table 4).

ASCVD includes a range of diseases primarily characterized by atherosclerosis, including CHD, ischemic stroke, and peripheral arterial disease. Recognized as a major global health threat (4–7), various regions have developed tools to assess ASCVD risk. Notably, the ASCVD risk calculator (2013) (9), created by the American College of Cardiology/American Heart Association (ACC/AHA), represents the first predictive model based on comprehensive data from White and Black American populations. The key variables include race (African American, White), gender, age, TC, HDL-C, systolic blood pressure (SBP), antihypertensive medication use, diabetes status, and smoking history. While the model offers satisfactory calibration for populations similar to its development cohort, its accuracy appears limited in other demographic groups (25–28). Furthermore, its predictive endpoints are confined to myocardial infarction and stroke. In 2014, the Joint British Societies (JBS) introduced the JBS3 risk score, a broader cardiovascular risk tool (11). This model extends its endpoints beyond myocardial infarction and stroke to include angina pectoris, coronary revascularization, transient ischemic attack, intermittent claudication, and related conditions. However, external validation of this tool is still pending. In 2016, China developed the Prediction for ASCVD Risk in China (China-PAR) score, which incorporated additional factors such as regional differences (north vs. south) and waist circumference. Its assessment endpoints included CHD-related mortality, myocardial infarction, and stroke (8). In comparison to the ASCVD risk calculator (2013) by the ACC/AHA, China-PAR demonstrates higher predictive accuracy for Chinese populations, but it has yet to be validated for non-Chinese adult cohorts.

Existing risk assessment tools are limited in their applicability across diverse populations and ethnicities. Moreover, these models are tailored to the general population, failing to address the specific needs of individuals with dyslipidemia, a group at heightened risk for ASCVD (1–3). In this study, significant differences in age and SBP were observed between the ASCVD and non-ASCVD groups, consistent with prior research in ASCVD risk models. However, no notable differences were found in HDL-C levels, diastolic blood pressure, or related parameters. Interestingly, TC and LDL-C levels were lower in the ASCVD group compared to the non-ASCVD group, likely reflecting the higher usage of lipid-lowering medications among ASCVD patients. These findings suggest that traditional indicators, such as TC, HDL-C, LDL-C, and DBP, may not be the most reliable markers for assessing ASCVD risk.

The retina, the only tissue with microvessels directly visible in the human body, has long been an important indicator for evaluating the effects of systemic conditions such as hypertension and diabetes on microvascular health (14, 21). Traditionally, color fundus photography is used to assess retinal blood vessels, primarily identifying signs like hemorrhage or neovascularization. The introduction of OCTA has advanced this assessment, offering a faster, non-invasive, and more detailed examination of retinal vascular changes (14, 29–31). Recent studies have established a relationship between retinal vascular parameters and coronary artery diseases. OCTA findings indicate that retinal VD is lower in patients with CHD compared to healthy controls (12, 32). Moreover, in CHD patients, higher Gensini scores correlate with reduced retinal VD (32). Patients with coronary total occlusion (CTO) also show higher Gensini scores and decreased retinal VD compared to those without CTO (33). The Gensini score, a standard clinical tool, assesses the severity of coronary artery stenosis (34). These observations suggest that retinal vasculature offers a distinct view of macrovascular changes. This study supports previous findings, noting that patients with ASCVD exhibit reduced retinal VD, including in the SCP and FD areas, alongside thinner RNFL and macular parafoveal thickness, when compared to individuals without ASCVD.

Among OCTA parameters, a reduction in retinal VD is indicative of vascular stenosis or occlusion. Existing literature confirms that hypertension, diabetes, and dyslipidemia contribute to decreases in retinal VD (16–18, 23, 35), with all three factors being well-established risks for ASCVD. In addition to ischemia caused by retinal vascular changes, RNFL and macular thickness are also influenced by local glucose levels and osmotic pressure (23).

In OCTA imaging, the detection region is divided into the macula and optic disc. Among the evaluated parameters, SCP perifovea VD, SCP parafovea VD, and FD correspond to macular superficial capillary plexus vessel densities and are correlated due to their interconnected capillary network. Combining these parameters results in a new composite measure, MSVD. Although DeLong test results did not demonstrate the superiority of MSVD over other indicators in assessing ASCVD risk in dyslipidemic patients, MSVD streamlines OCTA interpretation and provides practical benefits by enabling risk stratification based on macular scans alone, thus reducing both the scope of examination and associated costs. Considering its economic and clinical potential, MSVD represents a viable method for ASCVD screening in asymptomatic dyslipidemia patients. Its inclusion in future ASCVD risk prediction models merits further exploration.

The study has several limitations: (1) A cross-sectional design is initially selected for feasibility, but its statistical robustness in assessing the value of OCTA findings for evaluating ASCVD in individuals with dyslipidemia is inherently weaker than that of prospective cohort studies. (2) A higher proportion of patients in the ASCVD group are receiving medications (antihypertensive, lipid-lowering drugs, etc.) compared to the non-ASCVD group. Due to the small sample size, the study does not further stratify medication usage. (3) As a single-center study, it lacks the diversity of populations across different regions and ethnic groups, which limits the generalizability of the findings. Future research should involve a multicenter, multiracial prospective cohort study to adjust for confounding factors such as medication use and further assess the predictive value of retinal vascular changes detected by OCTA for ASCVD development in patients with dyslipidemia.

This study employs OCTA to assess retinal changes in dyslipidemia patients with ASCVD, identifying significant reductions in SCP perifovea VD, SCP parafovea VD, FD, RNFL, and macular thickness compared to those without ASCVD. These parameters may serve as potential biomarkers for evaluating ASCVD risk in individuals with dyslipidemia. Furthermore, a novel composite indicator, MSVD, is derived by combining the first three vessel density measures, offering an effective tool for ASCVD risk assessment in dyslipidemia patients. The incorporation of MSVD streamlines OCTA interpretation and lowers examination costs. OCTA-derived indicators, particularly MSVD, demonstrate considerable potential for integration into future ASCVD risk assessment models.