Introduction: This study aimed to evaluate the association between macular optical coherence tomography angiography (OCT-A) metrics, characteristics of ultrawide field (UWF) imaging, and cerebrovascular disease in patients with diabetes mellitus (DM) with different stages of diabetic retinopathy (DR). Methods: 516 eyes of 258 DM patients were enrolled in two centers (Milan and Belfast). UWF color fundus photos (CFPs) were obtained with Optos California (Optos, PLC) and graded for both DR severity and predominantly peripheral lesions presence (>50% of CFP lesions) by two independent graders. OCT-A (3 × 3 mm), available in 252 eyes of 136 patients, was used to determine perimeter, area, and circularity index of the foveal avascular zone and vessel density (VD); perfusion density (PD); fractal dimension on superficial, intermediate (ICP), and deep capillary plexuses; flow voids (FVs) in the choriocapillaris. Results: Out of 516 eyes, 108 eyes (20.9%) had no DR, and 6 eyes were not gradable. The remaining 402 eyes were as follows: 10.3% (53) had mild nonproliferative DR (NPDR), 38.2% (197) had moderate NPDR, 11.8% (61) had severe NPDR, and 17.6% (91) had proliferative DR. A worse DR stage was associated with a history of stroke (p = 0.044). Logistic regression analysis after taking into account sex, type of DM, age, DM duration, and OCT-A variables found that PD and VD on ICP were significantly associated with presence of stroke and DR severity. Conclusion: OCT-A metrics show an association with the presence of cerebrovascular complications, providing potentially useful parameters to estimate vascular risk in patients with DM.

Diabetes mellitus (DM) is one of the fastest growing health emergencies of this century, with an expected increase to 783 million worldwide by 2045 [1]. Diabetic retinopathy (DR) is the most common microvascular complication of DM [2] and is widely recognized as a reliable marker for DM-related complications in an individual [3]. Evidence suggests that DM patients have a 2- to 4-fold increased risk of stroke and coronary artery disease, with cardiovascular disease being the leading cause of mortality [4‒9].

Increasing DR severity has been shown to be associated with an increased risk of stroke [10]. In fact, individuals with DM are more likely to develop small subcortical infarcts or lacunar strokes than those without [11]. A recent meta-analysis from 19 studies involving 45,495 DM patients reported a pooled hazard ratio of 1.62 (1.28–2.06) for the risk of DR and stroke, with a robust correlation only for type 2 DM [12].

The demonstrated relationship between DR and cerebrovascular and/or cardiovascular accident suggests a potential link between microvascular and macrovascular damage [13], where presence of DR indicates that circulation has already been damaged by the hyperglycemic state [3]. Assessing DR severity does not only provide a unique opportunity to directly visualize the morphology of systemic vascular damage in DM [2] but it also represents a non-invasive, repeatable technique that might lead to the eye being considered a novel biomarker of vascular disease risk in asymptomatic DM patients [2].

New technologies, such as ultrawide field (UWF) fundus imaging and optical coherence tomography angiography (OCT-A) allow to grade DR with greater precision. UWF fundus imaging is defined as an image that includes the far periphery of the retina, at a range of 110°–220° [14]. Studies show that lesions outside the standard 7 Early Treatment for Diabetic Retinopathy Study (ETDRS) fields seen on UWF lead to a more severe DR stage in 10–15% of eyes of DM patients [15]. The presence of predominantly peripheral lesions (PPLs) has been linked with an increased risk of DR progression over 4 years, independent of baseline DR severity and HbA1c levels [16]. However, little data are available on whether PPL’s presence is associated with cerebrovascular complications and an increased risk of developing stroke or other systemic vascular complications in DM patients.

OCT-A is a fast and non-invasive imaging technique which allows a three-dimensional mapping of the retinal and choroidal vasculature and detailed evaluation of the foveal avascular zone (FAZ) [17]. OCT-A has the advantage of visualizing microvasculature with depth resolution, generating high contrast well-defined images [18] allowing evaluation of the superficial (SCP), intermediate (ICP), and deep (DCP) capillary plexuses.

Previous studies have linked stroke, neurodegeneration, and cognitive dysfunction with retinal findings and OCT-A metrics [19‒23]. However, no studies are available on the association between stroke and OCT-A metrics in DM patients. Using retinal images to determine the risk of future systemic morbidity and mortality is an area of considerable interest [13].

The aim of this study was to investigate the association between a history of stroke and the characteristics of OCT-A metrics, DR severity, and the presence of PPL on UWF color fundus photos (CFPs). The former has been evaluated on each of the 3 retinal plexuses and the choriocapillaris (CC) in the macula to determine non-invasive retinal imaging parameters associated with systemic cerebrovascular comorbidity in patients with DM.

Altogether, 258 patients with type 1 and 2 DM were recruited at the Retina and Imaging Unit, IRCCS MultiMedica, Milan, Italy and Belfast Health and Social Care Trust, UK The exclusion criteria were as follows: age less than 18 years, presence of other vascular retinal diseases (such as central/branch retinal vein occlusion, central/branch artery occlusion, etc.), unavailable data on systemic history, poor quality OCT-A imaging (signal strength <40). The study followed the tenets of the Declaration of Helsinki with collected written informed consent from all study participants from DR institutional registry after approval by the Institutional Ethics Committee in Milan and the local audit committee in Belfast. For each patient, the following data were collected from medical records: age; duration of DM; type of DM; presence of comorbidities including history of stroke, both ischemic and hemorrhagic and transient ischemic attack; history of cardiovascular disease, including ischemic cardiopathy and myocardial infarction.

UWF Color Photos Grading

UWF-CFP obtained with Optos California (Optos plc, Dunfermline, UK) were graded by two graders independently (SV and RS) according to the International Clinical Diabetic Retinopathy and Diabetic Macular Edema Disease Severity Scales [24]. Before the grading process, the two graders harmonized the gradings, reaching complete agreement. A mask representing the ETDRS seven-field region was overlapped on the UWF-CFP using Optos Advanced option to help identify PPL. PPLs were defined as the presence of more than 50% of lesions outside the 7 standard ETDRS fields (shown in Fig. 1).

Fig. 1.

Ultrawide field color fundus image showing an example of moderate nonproliferative diabetic retinopathy with PPLs, defined as the presence of >50% of lesions outside the Early Treatment for Diabetic Retinopathy Study (ETDRS) fields. The ETDRS seven-field mask was overlaid using Optos Advance (blue circles). Blue arrows point to peripheral retinal lesions.

Fig. 1.

Ultrawide field color fundus image showing an example of moderate nonproliferative diabetic retinopathy with PPLs, defined as the presence of >50% of lesions outside the Early Treatment for Diabetic Retinopathy Study (ETDRS) fields. The ETDRS seven-field mask was overlaid using Optos Advance (blue circles). Blue arrows point to peripheral retinal lesions.

Close modal

OCT and OCT-A Image Analysis

Altogether, 252 eyes of 136 patients underwent OCT and OCT-A imaging with Heidelberg Spectralis (Heidelberg, Germany) in Milan, Italy, using 10 ° × 10 ° macular scan after pupil dilation. Macular volume scan covering 20 ° × 15 ° was used for central macular thickness evaluation and to confirm the presence or absence of diabetic macular edema (DME). The full thickness retinal scan was exported and converted in 320 × 320 pixel format. The converted image was then imported into Image J (version 1.53, provided in the public domain by the National Institutes of Health, Bethesda, MD, USA), where the FAZ was manually drawn to calculate area, perimeter, and circularity index. OCT-A volume scans (10 ° × 10 ° field; 512 B-scans at 6 μm spacing; automatic real-time 5 of SCP, ICP, and DCP were exported and converted in 320 × 320 pixel format for the automatic analysis of the following metrics: perfusion density (PD); vessel density (VD); fractal dimension (FD); and CC slab which was used to estimate flow voids (FVs). All OCT-A images were checked for the quality of signal and the presence of artifacts. Only images with good quality and without artifacts (segmentation, projection, etc.) were included in the analysis.

Projection artifacts due to the presence of cysts in all 3 retinal plexuses were removed according to a previously published image processing method [25]. Briefly, a Gaussian filter was applied to the corresponding en-face SCP, ICP, and DCP images of 3 × 3 mm angio-cubes to reduce the noise; the resulting images were then binarized and dilated. Finally, black pixels due to noise were excluded from the final computation.

The automatic OCT-A image analysis was performed using MATLAB image processing toolbox and custom scripts (version 2020a, MathWorks, Natick, MA, USA). These scripts estimate PD, VD, and FD on SCP, ICP, and DCP, respectively. The FVs were estimated on the CC (shown in Fig. 2).

Fig. 2.

a The original optical coherence tomography angiography (OCT-A) image with the FAZ area marked in red. b The original OCT-A image of the superficial capillary plexus (SCP). c The binarized SCP image. d The skeletonized SCP image. e The original OCT-A image of the intermediate capillary plexus (ICP). f The binarized ICP image. g The skeletonized ICP image. h The original OCT-A image of the deep capillary plexus (DCP). i The binarized DCP image. j The skeletonized DCP image.

Fig. 2.

a The original optical coherence tomography angiography (OCT-A) image with the FAZ area marked in red. b The original OCT-A image of the superficial capillary plexus (SCP). c The binarized SCP image. d The skeletonized SCP image. e The original OCT-A image of the intermediate capillary plexus (ICP). f The binarized ICP image. g The skeletonized ICP image. h The original OCT-A image of the deep capillary plexus (DCP). i The binarized DCP image. j The skeletonized DCP image.

Close modal

The PD was estimated on the binarized images. Image binarization was performed by the “imbinarize” MATLAB function. The adopted algorithm was based on an adaptive threshold, achieving a threshold based on the local mean intensity in the neighborhood of each pixel. Then, this threshold was applied to the considered pixel. The threshold computation depended on a parameter called sensitivity, which ranges from 0 to 1. Different values had been carefully considered by expert ophthalmologist who evaluated the binarized images. Thus, the best threshold value has been identified, and it was equal to 0.59.

The binarized image was used to directly evaluate the PD. The area related to the FAZ, manually outlined by an expert ophthalmologist, was excluded from the analysis. Thus, the domain was defined as the set of pixels not belonging to the FAZ area. The PD was computed as follows:
where AreaPerf was the number of white pixels in the binarized image, and AreaD was the total number of pixels belonging to D. VD was computed using the same binarized image as for PD, excluding FAZ here as well. The binarized image was processed by the bwskel MATLAB routine, performing the so-called skeletonization. VD was computed as the percentage of the ratio of the total vessel area (white pixels of the skeletonized image) and the total area belonging to D.
FD used the same skeletonized image and based FD estimation on the box-counting method. It counted the N bidimensional boxes of size R that contain all the white pixels [26]. Then, the FD was computed as follows:

Finally, the FV has been computed by adopting the compensation method proposed by Zhang et al. [27].First, a Gaussian filter of size 3 × 3 to the en-face DCP image was applied, then the filtered image was inverted and subtracted from the DCP one. The output of this phase was the compensation image. This image was then subtracted to the CC, and the mean (Imean) and the standard deviation (ISD) were computed. These two values were then used to binarize the image.

Ii,jFV=whiteifIi,jImean>ISDblackotherwise,where Ii,j was the compensated image.

FVs with a diameter of less than 20 microns were discarded by excluding the binarized image groups of black pixels with 4 or fewer elements. This was done to reflect the 5.7 microns spatial resolution of the instrument. The FV was computed as follows:
where AreaFV was the number of black pixels belonging to groups with more than 4 elements, and AreaD was the domain of the computation.

Statistical Analyses

For continuous variables, the arithmetic mean (±standard deviation) is reported. Categorical variables were reported as experimental frequencies and/or percentages. For all patient-level analysis, the eye with more severe DR diagnosis was used.

The association between DR severity (using the worse eye grade) and clinical variables such as sex, type of DM, presence of PPL, presence of DME, previous history of stroke was assessed in 258 patients using a χ2 test in 5 × 2 contingency tables. As DM duration and the age were not normally distributed, the association between the DR severity and DM duration and age was assessed using Kruskal-Wallis ANOVA. The presence of significant differences in DM duration for eyes with PPL or DME was assessed using the Mann-Whitney test. One-way ANOVA was used for univariate analysis of OCT-A variables categorized by DR severity. The mean OCT/OCT-A values were compared between eyes with or without PPL and between eyes having DME or not, using a two-sided unpaired t-test.

The association between DR severity and the clinical and OCT-A variables was assessed on a patient basis using an ordinal logistic regression model. The DR severity was considered as the dependent variable; sex and type of DM as categorical predictors; age, diabetes duration, and OCT-A variables as the independent continuous predictors.

The mean OCT-A values were compared between the groups of patients with or without a history of stroke in a univariate analysis using a two-sided unpaired t-test. Finally, the association between previous history of stroke and the clinical and OCT-A variables was assessed using a logistic regression. The history of stroke was considered as the dependent categorical variable; sex and DM type as categorical predictors; age, diabetes duration, and OCT-A variables as the independent continuous predictors.

The mean value of OCT-A variables between two eyes was used as the best indicator of OCT-A metric in a single patient to be associated with the presence of stroke. The statistical analyses were performed using Statistica software version 6.0 (StatSoft, Inc., Tulsa, OK, USA), using a two-sided type 1 error rate of p ≤ 0.05.

Table 1 describes demographic data and clinical features of the enrolled patients. A total of 516 eyes (258 patients) were included in the study of which 58.7% (152) were males and 41.3% (107) were females. The mean age was of 67.1 ± 13.7 years (range, 19–95). 81.9% (212) of patients had type 2 DM, while 18.1% (47) had type 1 DM. Mean duration of DM was 19.1 ± 10.6 years. Positive history of stroke was recorded in 10.4% of DM patients.

Table 1.

Demographic data, DR severity by UWF-CFP, and DME on OCT

n = 259 patientsValue±SD or n(%)
Female sex 107 (41.3) 
Age, years 67.1±13.7 
Duration of DM, years 19.1±10.6 
Type 2 DM 212 (81.9) 
Stroke 27 (10.4) 
DR severity on UWF (n = 516 eyes) 
 No DR 108 (20.9) 
 Mild NPDR 53 (10.3) 
 Moderate NPDR 197 (38.2) 
 Severe NPDR 61 (11.8) 
 PDR 91 (17.6) 
 Ungradable 6 (1.2) 
 Presence of PPL 183 (35.5) 
DME on OCT (n = 252 eyes) 
 DME present 37 (14.7) 
 DME absent 215 (85.3) 
n = 259 patientsValue±SD or n(%)
Female sex 107 (41.3) 
Age, years 67.1±13.7 
Duration of DM, years 19.1±10.6 
Type 2 DM 212 (81.9) 
Stroke 27 (10.4) 
DR severity on UWF (n = 516 eyes) 
 No DR 108 (20.9) 
 Mild NPDR 53 (10.3) 
 Moderate NPDR 197 (38.2) 
 Severe NPDR 61 (11.8) 
 PDR 91 (17.6) 
 Ungradable 6 (1.2) 
 Presence of PPL 183 (35.5) 
DME on OCT (n = 252 eyes) 
 DME present 37 (14.7) 
 DME absent 215 (85.3) 

DR, diabetic retinopathy; UWF-CFPs, ultrawide field color fundus photos; DME, diabetic macular edema; OCT, optical coherence tomography; DM, diabetes mellitus; NPDR, nonproliferative diabetic retinopathy; PDR, proliferative DR; PPLs, predominantly peripheral lesions.

On UWF-CFP, 108 eyes (20.9%) did not have signs of DR, 10.3% (53) had mild nonproliferative DR (NPDR), 38.2% (197) had moderate NPDR, 11.8% (61) had severe NPDR, and 17.6% (91) had proliferative DR. Six eyes were not gradable due to poor image quality.

More severe DR was found in patients with type 1 DM, presence of PPL, presence of DME, longer duration of DM, and in younger patients (p < 0.0001 for all). A worse stage of DR was associated with a history of stroke (p = 0.044) (Table 2). OCT and OCT-A metrics were available for 252 eyes in 136 patients.

Table 2.

Tabulation of DR severity (patient level) with history of stroke

History of stroke, n (%)No history of stroke, n (%)Total
No DR 2 (7.5) 45 (19.5) 47 
Mild NPDR 2 (7.5) 21 (9.1) 23 
Moderate NPDR 7 (25.9) 94 (40.7) 101 
Severe NPDR 5 (18.5) 28 (12.1) 33 
PDR 11 (40.7) 43 (18.6) 54 
Total 53 231 258 
History of stroke, n (%)No history of stroke, n (%)Total
No DR 2 (7.5) 45 (19.5) 47 
Mild NPDR 2 (7.5) 21 (9.1) 23 
Moderate NPDR 7 (25.9) 94 (40.7) 101 
Severe NPDR 5 (18.5) 28 (12.1) 33 
PDR 11 (40.7) 43 (18.6) 54 
Total 53 231 258 

DR, diabetic retinopathy; NPDR, nonproliferative diabetic retinopathy; PDR, proliferative DR.

In univariate analyses, PD on SCP, FAZ perimeter, area, and circularity index (full retina scan) and FV in CC resulted in being statistically significantly associated with DR severity (Table 3). PD, VD, FD in SCP and FAZ circularity index and FV in CC showed a significant variation in eyes with DME versus eyes without DME (Table 3). No significant differences in OCT-A parameters were found in eyes with and without PPL (p > 0.47, for all). A univariate analysis using a two-sided unpaired t-test, of OCT-A parameters found a statistically significant borderline decrease for PD (14.6 vs. 17.9, t test = 1.87, p = 0.06), VD (5.1 vs. 6.4, t test = 1.85, p = 0.06), and FD (1.35 vs. 1.39, t test = 1.87, p = 0.06) in SCP in patients with history of stroke versus those without.

Table 3.

Mean OCT and OCT-A measurements across different DR levels

Total n = 252No DR(n = 62)Mild NPDR(n = 34)Moderate NPDR(n = 117)Severe NPDR(n = 19)PDR (n = 20)p value
CMT 275.6 (SD 48.3) 283.1 (SD 39.4) 296.0 (SD 66.8) 315.7 (SD 91.0) 282.1 (SD 31.5) 0.064 
PD SCP 20.8 (SD 6.5) 18.5 (SD 5.8) 19.4 (SD 6.1) 21.3 (SD 7.4) 23.8 (SD 7.5) 0.023 
FV CC 76.4 (SD 9.4) 78.2 (SD 8.8) 75.7 (SD 10.0) 73.6 (SD13.4) 68.6 (SD 12.9) 0.015 
FAZ perimeter 2.6 (SD 0.6) 2.3 (SD 0.6) 3.0 (SD 1.1) 3.3 (SD 1.1) 3.6 (SD 1.4) <0.001 
FAZ area 0.4 (SD 0.2) 0.3 (SD 0.1) 0.4 (SD 0.2) 0.4 (SD 0.2) 0.5 (SD 0.2) <0.001 
FAZ circularity 0.6 (SD 0.1) 0.6 (SD 0.1) 0.6 (SD 0.1) 0.5 (SD 0.1) 0.5 (SD 0.2) <0.001 
Total n = 252No DR(n = 62)Mild NPDR(n = 34)Moderate NPDR(n = 117)Severe NPDR(n = 19)PDR (n = 20)p value
CMT 275.6 (SD 48.3) 283.1 (SD 39.4) 296.0 (SD 66.8) 315.7 (SD 91.0) 282.1 (SD 31.5) 0.064 
PD SCP 20.8 (SD 6.5) 18.5 (SD 5.8) 19.4 (SD 6.1) 21.3 (SD 7.4) 23.8 (SD 7.5) 0.023 
FV CC 76.4 (SD 9.4) 78.2 (SD 8.8) 75.7 (SD 10.0) 73.6 (SD13.4) 68.6 (SD 12.9) 0.015 
FAZ perimeter 2.6 (SD 0.6) 2.3 (SD 0.6) 3.0 (SD 1.1) 3.3 (SD 1.1) 3.6 (SD 1.4) <0.001 
FAZ area 0.4 (SD 0.2) 0.3 (SD 0.1) 0.4 (SD 0.2) 0.4 (SD 0.2) 0.5 (SD 0.2) <0.001 
FAZ circularity 0.6 (SD 0.1) 0.6 (SD 0.1) 0.6 (SD 0.1) 0.5 (SD 0.1) 0.5 (SD 0.2) <0.001 
DMEMeanT valueSDp value
DME absentDME presentDME absentDME present
CMT 281.73 334.62 −5.16 47.38 97.34 <0.001 
PD SCP 19.61 23.20 −3.17 6.60 4.71 0.0017 
VD SCP 6.96 7.87 −2.04 2.59 1.82 0.0419 
FD SCP 1.40 1.42 −2.67 0.06 0.04 0.0081 
FV CC 76.32 70.51 3.19 10.39 9.21 0.0016 
FAZ circularity 0.58 0.53 2.04 0.14 0.13 0.0425 
DMEMeanT valueSDp value
DME absentDME presentDME absentDME present
CMT 281.73 334.62 −5.16 47.38 97.34 <0.001 
PD SCP 19.61 23.20 −3.17 6.60 4.71 0.0017 
VD SCP 6.96 7.87 −2.04 2.59 1.82 0.0419 
FD SCP 1.40 1.42 −2.67 0.06 0.04 0.0081 
FV CC 76.32 70.51 3.19 10.39 9.21 0.0016 
FAZ circularity 0.58 0.53 2.04 0.14 0.13 0.0425 

OCT, optical coherence tomography; OCT-A, OCT angiography; DR, diabetic retinopathy; DME, diabetic macular edema; SD, standard deviation; CMT, central macular thickness; PD, perfusion density; SCP, superficial capillary plexus; FVs, flow voids; VD, vessel density; FD, fractal dimension.

The clinical and OCT-A variables were successively inserted as independent variables in generalized linear models to study the association with DR severity (ordinal logistic regression) and the history of stroke (logistic regression). Ordinal logistic regression showed that VD (Wald statistic 9.2; p = 0.002) and PD in ICP (Wald statistic 13.1; p < 0.001) were the only significant predictors of DR severity.

Logistic regression showed that VD (Wald statistic 6.6; p = 0.010) and PD (Wald statistic 5.7; p = 0.017) in ICP were the only variables significantly associated with a previous history of stroke. The two generalized linear models are reported in Tables 4 and 5.

Table 4.

Patient’s OCT-A parameters associated with DR severity

Ordinal logistic regressionDependent variable: DR severity
 Wald statistics p level 
Intercept 115.4 <0.001 
PD ICP 13.5 <0.001 
VD ICP 10.5 0.001 
Ordinal logistic regressionDependent variable: DR severity
 Wald statistics p level 
Intercept 115.4 <0.001 
PD ICP 13.5 <0.001 
VD ICP 10.5 0.001 

OCT-A, optical coherence tomography angiography; PD, perfusion density; VD, vessel density; FAZ, foveal avascular zone in the full retina; ICP, intermediate capillary plexus.

Table 5.

Patient’s OCT-A parameters associated with history of stroke

Logistic regressionDependent variable:history of stroke
 Wald statistics p level 
Intercept 0.01 0.91 
PD ICP 5.73 0.017 
VD ICP 6.58 0.010 
Logistic regressionDependent variable:history of stroke
 Wald statistics p level 
Intercept 0.01 0.91 
PD ICP 5.73 0.017 
VD ICP 6.58 0.010 

OCT-A, optical coherence tomography angiography; PD, perfusion density; VD, vessel density; ICP, intermediate capillary plexus.

The mean value of OCT-A variables between two eyes was selected as the best indicator of OCT-A metric in a single patient.

In the present bi-center study we report on OCT/OCT-A and UWF color fundus imaging data from the European cohort of patients with different stages of severity of DR and its association to stroke. A more severe stage of DR on UWF-CFP was associated with a history of stroke in the present study. This is in agreement with data reported in the large Australian population cohort, in which people with type 2 DM with moderate NPDR or worse at baseline had a more than 2-fold increase in risk of any subsequent stroke compared with mild NPDR or no DR [10].

The present study documented borderline significant decrease in PD, VD, and FD in SCP in patients with stroke versus those without stroke, using a univariate analysis (two-sided unpaired t-test). This may suggest that DM patients with a history of stroke have major modifications of retinal microcirculation. The retina and the brain share similar embryological origin, anatomic, and physiologic characteristics, thus retinal vascular lesions are likely to reflect the presence of similar pathological processes in the cerebral microcirculation [3, 7]. It is known that DR represents an independent risk factor for stroke, suggesting a role of microvascular pathology in the development of stroke in diabetic patients [6]. However, the reason for this correlation has not been defined yet; no studies to our knowledge have investigated the correlation between stroke and OCT-A metrics in DM patients.

Previous studies have found an association between stroke and pathologic vascular findings on fundus photography; however, parameters used were qualitative and therefore potentially imprecise. A recent study by Liu et al. [23] reported that VD in SCP and DCP in all macular sectors were decreased in patients with stroke, after adjusting for numerous systemic factors including levels of HbA1c. However, patients with DR were excluded, while the present study aimed to investigate the association between OCT-A metrics and stroke in DM patients with DR, a subpopulation with higher risk of cerebrovascular disease.

FD is the measure of density and complexity of the retinal vascular network [28]. A decreased FD suggests a reduction of the complexity and a rarefaction of the retinal microcirculation [29]. Previous studies have correlated a decreased FD with both stroke and other neurodegenerative diseases [19‒22]. Many of these studies, however, calculated FD on fundus photography using different methods. Kawasaki et al. [22] reported that a higher risk of stroke is associated with a reduction in FD on fundus photography. Doubal et al. [30] documented an association between decreased FD calculated on fundus photography and lacunar stoke. DM represents a known risk factor for the lacunar stroke which is considered a specific type of ischemic stroke that occurs when blood flow to one of the small arteries (<1.5 cm) deep within the brain becomes blocked [11].

The logistic regression analysis taking into account patients’ clinical and OCT-A variables found that PD and VD in ICP were the only parameters mostly associated with the history of stroke. The results of our study strengthen the hypothesis of an existing association between the entity of retinal damage and the damage of the cerebral circulation in patients with DM; moreover, the results demonstrate the presence of association between OCT-A quantitative metrics and a history of stroke, highlighting that specific OCT-A parameters could be helpful in identifying a subgroup of patients with higher risk of cerebrovascular events. Future studies characterizing the type of stroke with strict control of other risk factors could be helpful in defining more in detail the nature of the association.

The importance of UWF fundus imaging in assessing the severity of DR has emerged recently and has been extensively studied. In fact, UWF-CFP can detect more severe DR in a substantial proportion of patients due to the presence of PPL [15]. Detection of PPL is important due to the greater risk of progression of DR and development of proliferative DR [16]. In the present study, presence of PPL did not show association with stroke; however, more advanced stages of DR (graded on UWF-CFP) were associated with a history of stroke.

OCT-A has been increasingly used for the evaluation of microvascular modifications in patients with diabetes with or without DR. Modifications of the FAZ, including FAZ enlargement, presence of irregular FAZ contour and loss of circularity as well as decreased perfusion and/or VD, and decreased FD have been documented and proposed as signs of macular ischemia associated with DR progression [29‒34].

Even if retinal vascular non-perfusion in the macula detected with OCT-A was correlated with increasing severity of DR, evaluated in the periphery [35]; however, in the present study, OCT-A quantitative parameters could predict only approximately 32% (data not shown) of DR severity evaluated on UWF-CFP. These parameters included PD and VD evaluated in the ICP, in a logistic regression analysis. This may suggest that macular and peripheral modifications in DR may be independent to a certain level and that both UWF imaging and OCT-A may be necessary for better understanding of retinal microvascular modifications due to DM. It remains to be evaluated if the use of wide field OCT-A, evaluating more peripheral microvascular condition in the retina may be better associated with the presence of PPL and severity of DR.

In the present study, the presence of PPL on UWF-CFP was not associated with the presence of cerebrovascular comorbidity, whereas OCT-A quantitative parameters in the macula were associated with the presence of stroke. OCT-A is a new non-invasive tool to quantify retinal and choroidal capillary modifications due to DM and vascular complications in other organs. Thus, future and prospective studies should investigate the ability of OCT-A metrics to predict cerebrovascular events and other systemic comorbidities, such as ischemic cardiopathy in patients with DM. Moreover, development and validation of the automatic algorithms of evaluation of retinal images and implementation in the clinical practice may help in better management of patients with DM and in predicting the risk of systemic vascular comorbidities.

The major limitations of the present study include the limited number of evaluated patients from the cohort for the OCT-A analysis and the lack of longitudinal data. The small sample size did not allow to take into consideration other variables implicated in stroke risk such as diabetic nephropathy stage [36], systolic pressure values [37], BMI [38], and smoking [39]. However, detailed analysis of the available data and the use of sophisticated and already validated methods of OCT-A image analyses should enable to set pilot results that may serve for future larger studies.

Data from the present study should encourage future larger and prospective studies that could strengthen the importance of a multidisciplinary approach together with a multimodal imaging of retinal fundus, automatic evaluation, evaluating both the macula and the far periphery, crucial for better assessing the actual multiorgan disease load on a single patient, thus indicating the best way to approach the patient (reducing the burden and maximizing the beneficial effect).

This study protocol was reviewed and approved by Ethical Review Board IRCCS MultiMedica Ethics Committee, protocol No. 502.2021. Written informed consent was obtained from participants.

The authors have no conflicts of interest to declare.

This research received no external funding.

Conceptualization and data curation: S. V., R. S., and T. P. Statistical analysis: M.B. Retinal images analysis: S.V., E.T., T. P., and RS. Data collection: S.V., F.F., T.P, R.S., C.L, and L.C. Writing – draft preparation: S.V., F.F, T.P., R.S., P.N., C.L., and L.C. Writing – review and editing: S.V., R.S., T.P., and P.N.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author (S.V.) upon reasonable request.

1.
IDF atlas 2021 10th edition.
2.
Cheung
N
,
Wong
TY
.
Diabetic retinopathy and systemic vascular complications
.
Prog Retin Eye Res
.
2008
;
27
(
2
):
161
76
.
3.
Vujosevic
S
,
Aldington
SJ
,
Silva
P
,
Hernández
C
,
Scanlon
P
,
Peto
T
.
Screening for diabetic retinopathy: new perspectives and challenges
.
Lancet Diabetes Endocrinol
.
2020
;
8
(
4
):
337
47
.
4.
Beckman
JA
,
Creager
MA
,
Libby
P
.
Diabetes and atherosclerosis: epidemiology, pathophysiology, and management
.
JAMA
.
2002
;
287
(
19
):
2570
81
.
5.
Penlioglou
T
,
Stoian
AP
,
Papanas
N
.
Diabetes, vascular aging and stroke: old dogs, new tricks
.
J Clin Med
.
2021
;
10
(
19
):
4620
.
6.
Bloomgarden
Z
,
Chilton
R
.
Diabetes and stroke: an important complication
.
J Diabetes
.
2021
;
13
(
3
):
184
90
.
7.
Wong
TY
.
Is retinal photography useful in the measurement of stroke risk
.
Lancet Neurol
.
2004
;
3
(
3
):
179
83
.
8.
Wong
KH
,
Hu
K
,
Peterson
C
,
Sheibani
N
,
Tsivgoulis
G
,
Majersik
JJ
.
Diabetic retinopathy and risk of stroke: a secondary analysis of the ACCORD eye study
.
Stroke
.
2020
;
51
(
12
):
3733
6
.
9.
Cheung
N
,
Rogers
S
,
Couper
DJ
,
Klein
R
,
Sharrett
AR
,
Wong
TY
.
Is diabetic retinopathy an independent risk factor for ischemic stroke
.
Stroke
.
2007
;
38
(
2
):
398
401
.
10.
Drinkwater
JJ
,
Davis
TME
,
Hellbusch
V
,
Turner
AW
,
Bruce
DG
,
Davis
WA
.
Retinopathy predicts stroke but not myocardial infarction in type 2 diabetes: the Fremantle Diabetes Study Phase II
.
Cardiovasc Diabetol
.
2020
;
19
(
1
):
43
11
.
11.
Arboix
A
,
Martí-Vilalta
JL
.
Lacunar stroke
.
Expert Rev Neurother
.
2009
;
9
(
2
):
179
96
.
12.
Wang
Z
,
Cao
D
,
Zhuang
X
,
Yao
J
,
Chen
R
,
Chen
Y
.
Diabetic retinopathy may Be a predictor of stroke in patients with diabetes mellitus
.
J Endocr Soc
.
2022
6
8
bvac097
9
.
13.
Modjtahedi
BS
,
Wu
J
,
Luong
TQ
,
Gandhi
NK
,
Fong
DS
,
Chen
W
.
Severity of diabetic retinopathy and the risk of future cerebrovascular disease, cardiovascular disease, and all-cause mortality
.
Ophthalmology
.
2021
;
128
(
8
):
1169
79
.
14.
Choudhry
N
,
Duker
JS
,
Freund
KB
,
Kiss
S
,
Querques
G
,
Rosen
R
.
Classification and guidelines for widefield imaging: recommendations from the international widefield imaging study group
.
Ophthalmol Retin
.
2019
;
3
(
10
):
843
9
.
15.
Silva
PS
,
Cavallerano
JD
,
Sun
JK
,
Soliman
AZ
,
Aiello
LM
,
Aiello
LP
.
Peripheral lesions identified by mydriatic ultrawide field imaging: distribution and potential impact on diabetic retinopathy severity
.
Ophthalmology
.
2013
;
120
(
12
):
2587
95
.
16.
Silva
PS
,
Cavallerano
JD
,
Haddad
NMN
,
Kwak
H
,
Dyer
KH
,
Omar
AF
.
Peripheral lesions identified on ultrawide field imaging predict increased risk of diabetic retinopathy progression over 4 years
.
Ophthalmology
.
2015
;
122
(
5
):
949
56
.
17.
Vujosevic
S
,
Cunha-Vaz
J
,
Figueira
J
,
Löwenstein
A
,
Midena
E
,
Parravano
M
.
Standardization of optical coherence tomography angiography imaging biomarkers in diabetic retinal disease
.
Ophthalmic Res
.
2021
;
64
(
6
):
871
87
.
18.
Spaide
RF
,
Fujimoto
JG
,
Waheed
NK
,
Sadda
SR
,
Staurenghi
G
.
Optical coherence tomography angiography
.
Prog Retin Eye Res
.
2018
;
64
:
1
55
.
19.
Cheung
CYL
,
Ong
S
,
Ikram
MK
,
Ong
YT
,
Chen
CP
,
Venketasubramanian
N
.
Retinal vascular fractal dimension is associated with cognitive dysfunction
.
J Stroke Cerebrovasc Dis
.
2014
;
23
(
1
):
43
50
.
20.
Lemmens
S
,
Devulder
A
,
Van Keer
K
,
Bierkens
J
,
De Boever
P
,
Stalmans
I
.
Systematic review on fractal dimension of the retinal vasculature in neurodegeneration and stroke: assessment of a potential biomarker
.
Front Neurosci
.
2020
;
14
:
16
.
21.
Shi
C
,
Chen
Y
,
Kwapong
WR
,
Tong
Q
,
Wu
S
,
Zhou
Y
.
Characterization by fractal dimension analysis of the retinal capillary network in Parkinson disease
.
Retina
.
2020
;
40
(
8
):
1483
91
.
22.
Kawasaki
R
,
Che Azemin
MZ
,
Kumar
DK
,
Tan
AG
,
Liew
G
,
Wong
TY
.
Fractal dimension of the retinal vasculature and risk of stroke: a nested case-control study
.
Neurology
.
2011
;
76
(
20
):
1766
7
.
23.
Liu
B
,
Hu
Y
,
Ma
G
,
Xiao
Y
,
Zhang
B
,
Liang
Y
.
Reduced retinal microvascular perfusion in patients with stroke detected by optical coherence tomography angiography
.
Front Aging Neurosci
.
2021
;
13
:
628336
.
24.
Wilkinson
CP
,
Ferris
FL
,
Klein
RE
,
Lee
PP
,
Agardh
CD
,
Davis
M
.
Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales
.
Ophthalmology
.
2003
;
110
(
9
):
1677
82
.
25.
Vujosevic
S
,
Toma
C
,
Villani
E
,
Muraca
A
,
Torti
E
,
Florimbi
G
.
Diabetic macular edema with neuroretinal detachment: OCT and OCT-angiography biomarkers of treatment response to anti-VEGF and steroids
.
Acta Diabetol
.
2020
;
57
(
3
):
287
96
.
26.
Zahid
S
,
Dolz-Marco
R
,
Freund
KB
,
Balaratnasingam
C
,
Dansingani
K
,
Gilani
F
.
Fractal dimensional analysis of optical coherence tomography angiography in eyes with diabetic retinopathy
.
Invest Ophthalmol Vis Sci
.
2016
;
57
(
11
):
4940
7
.
27.
Zhang
Q
,
Zheng
F
,
Motulsky
EH
,
Gregori
G
,
Chu
Z
,
Chen
CL
.
A novel strategy for quantifying choriocapillaris flow voids using swept-source OCT angiography
.
Invest Ophthalmol Vis Sci
.
2018
;
59
(
1
):
203
11
.
28.
Huang
F
,
Dashtbozorg
B
,
Zhang
J
,
Bekkers
E
,
Abbasi-Sureshjani
S
,
Berendschot
TTJM
.
Reliability of using retinal vascular fractal dimension as a biomarker in the diabetic retinopathy detection
.
J Ophthalmol
.
2016
;
2016
:
6259047
.
29.
Kim
AY
,
Chu
Z
,
Shahidzadeh
A
,
Wang
RK
,
Puliafito
CA
,
Kashani
AH
.
Quantifying microvascular density and morphology in diabetic retinopathy using spectral-domain optical coherence tomography angiography
.
Invest Ophthalmol Vis Sci
.
2016
57
9
OCT362
70
.
30.
Doubal
FN
,
MacGillivray
TJ
,
Patton
N
,
Dhillon
B
,
Dennis
MS
,
Wardlaw
JM
.
Fractal analysis of retinal vessels suggests that a distinct vasculopathy causes lacunar stroke
.
Neurology
.
2010
;
74
(
14
):
1102
7
.
31.
Vujosevic
S
,
Toma
C
,
Villani
E
,
Gatti
V
,
Brambilla
M
,
Muraca
A
.
Early detection of microvascular changes in patients with diabetes mellitus without and with diabetic retinopathy: comparison between different swept-source OCT-A instruments
.
J Diabetes Res
.
2019
;
2019
:
2547216
.
32.
Rodrigues
TM
,
Marques
JP
,
Soares
M
,
Simão
S
,
Melo
P
,
Martins
A
.
Macular OCT-angiography parameters to predict the clinical stage of nonproliferative diabetic retinopathy: an exploratory analysis
.
Eye
.
2019
;
33
(
8
):
1240
7
.
33.
Vujosevic
S
,
Muraca
A
,
Gatti
V
,
Masoero
L
,
Brambilla
M
,
Cannillo
B
.
Peripapillary microvascular and neural changes in diabetes mellitus: an OCT-angiography study
.
Invest Ophthalmol Vis Sci
.
2018
;
59
(
12
):
5074
81
.
34.
Akil
H
,
Karst
S
,
Heisler
M
,
Etminan
M
,
Navajas
E
,
Maberley
D
.
Application of optical coherence tomography angiography in diabetic retinopathy: a comprehensive review
.
Can J Ophthalmol
.
2019
;
54
(
5
):
519
28
.
35.
Nesper
PL
,
Roberts
PK
,
Onishi
AC
,
Chai
H
,
Liu
L
,
Jampol
LM
.
Quantifying microvascular abnormalities with increasing severity of diabetic retinopathy using optical coherence tomography angiography
.
Invest Ophthalmol Vis Sci
.
2017
58
6
BIO307
15
.
36.
Kaze
AD
,
Jaar
BG
,
Fonarow
GC
,
Echouffo-Tcheugui
JB
.
Diabetic kidney disease and risk of incident stroke among adults with type 2 diabetes
.
BMC Med
.
2022
;
20
(
1
):
127
.
37.
Sethi
R
,
Hiremath
JS
,
Ganesh
V
,
Banerjee
S
,
Shah
M
,
Mehta
A
.
Correlation between stroke risk and systolic blood pressure in patients over 50 Years with uncontrolled hypertension: results from the SYSTUP-India study
.
Cardiovasc Ther
.
2021
;
2021
:
6622651
.
38.
Strazzullo
P
,
D’Elia
L
,
Cairella
G
,
Garbagnati
F
,
Cappuccio
FP
,
Scalfi
L
.
Excess body weight and incidence of stroke: meta-analysis of prospective studies with 2 million participants
.
Stroke
.
2010
;
41
(
5
):
e418
26
.
39.
Mons
U
,
Müezzinler
A
,
Gellert
C
,
Schöttker
B
,
Abnet
CC
,
Bobak
M
CHANCES Consortium
.
Impact of smoking and smoking cessation on cardiovascular events and mortality among older adults: meta-analysis of individual participant data from prospective cohort studies of the CHANCES consortium
.
BMJ
.
2015
350
h1551
.