Table 2.
Associations of neuronal and microvascular structures and function of the neurovascular coupling unit, as quantified by retinal imaging, with cerebral MRI features of structural brain abnormalities.
| N | MRI features of structural brain abnormalities | Confounders | Main results | Reference |
|---|---|---|---|---|
| RNFL thickness | ||||
| Prospective studies | ||||
| No studies available | ||||
| Cross-sectional studies | ||||
| N=2,124 | Grey and white matter brain volume, cerebral microbleeds (n=352), lacunar infarcts (n=145). | Age, sex, sub cohort, education, axial length of the eye, intracranial volume, mean arterial blood pressure, blood pressure lowering medication, body mass index, total cholesterol, diabetes mellitus, and smoking | Lower peripapillary RNFL thickness was significantly associated with lower gray and white matter volume (beta per SD decrease in peripapillary RNFL thickness for the standardized mean difference in grey and white volume [95% CI], respectively, -0.03 [-0.05; -0.01] and -0.03 [-0.06; -0.01]). However lower RNFL thickness was not associated with the presence of cerebral microbleeds or lacunar infarcts (per SD lower RNFL thickness, OR [95%CI] 1.02 [0.90–1.15] and 1.11 [0.94–1.31], respectively). | Mutlu, 2017 (44); |
| N=2,131 | Total brain volume | Age, age squared and sex, race, education, body mass index, mean arterial blood pressure and smoking | Lower macular RNFL thickness was non-significantly associated with lower total brain volume (beta, per SD decrease in macular RNFL thickness, standardized mean difference [95% CI], -0.03 [-0.07; 0.00]). | Chua, 2021 (45); |
| Retinal arteriolar diameter | ||||
| Prospective studies | ||||
| No studies available | ||||
| Cross-sectional studies | ||||
| N=1,211 | Presence of lacunar infarcts (=200); Presence of white matter lesions (n= 384). | Age, sex, study site, hypertension, hypercholesterolemia, diabetes status and cigarette-smoking status. | Individuals with arteriolar narrowing (no definition provided), versus individuals without arteriolar narrowing, had a higher odds of the presence of a lacunar infarct (odds ratio [95% CI] for with versus without narrowing, 1.61 [1.01-2.57], but did not have a higher odds of the presence of white matter lesions (odds ratio [95% CI] for with versus without narrowing 1.24 [0.82; 1.87]) or for the presence of both a lacunar infarct and white matter lesions (odds ratio [95% CI] for with versus without narrowing 1.40 [0.86; 2.28]). | Liew, 2012 (46); |
| N=1,185 | Presence of white matter hyperintensities (2% had none; 36% had one white matter hyperintensity; 31% had two white matter hyperintensities; 31% had three or more white matter hyperintensities). | Age, sex ethnicity, hypertension, body-mass index, glycated hemoglobin, diabetes, total cholesterol, high density lipoprotein cholesterol, log C-reactive protein, coronary artery disease, years of education and smoking habit, presence of retinopathy | Narrower arteriolar diameter was significantly associated with a lower odds of the presence of white matter hyperintensities (odds ratio [95% CI] per micrometer lower retinal arteriolar diameter 0.98 [0.97; 0.99]). | Hughes, 2016 (47); |
| N=1,065 | Gray and white matter volume. | Age, sex, cohort, and the other vessels, venular diameter, systolic blood pressure, diastolic blood pressure, body-mass index, total cholesterol, high-density lipoprotein cholesterol, diabetes mellitus, and smoking | Wider retinal arteriolar diameter was not associated with gray and white matter volume (beta per SD increase in retinal arteriolar diameter for the standardized mean difference in grey and white volume [95% CI], respectively, -0.02 [-0.09; 0.06] and 0.04 [-0.03; 0.11]). | Ikram, 2013 (48); |
| N=1,013 (acute stroke patients) |
Cortical and subcortical cerebral atrophy (n=424). | Age, sex, study site, hypertension, hypercholesterolemia, diabetes, and smoking status | Narrow retinal arteriolar diameter (defined as lowest quintile versus other quintiles combined as reference group) was not significantly associated with cortical and subcortical cerebral atrophy (narrow versus wide retinal arteriolar diameter, OR [95%CI] 1.3 [0.9-2.0]). | Baker, 2010 (49); |
| N=1,114 (N=557 stroke cases and N=557 controls) |
Lacunar stroke (n=261). | Age, sex, race, smoking status, hypertension status, diabetes mellitus status, and hypercholesterolemia status | Narrower retinal arteriolar diameter (per SD) was significantly associated with a greater odds for the presence of lacunar stroke (per SD narrower retinal arteriolar diameter OR [95%CI] 2.87 [2.19-3.76]). | Ong, 2013 (50); |
| Retinal venular diameter | ||||
| Prospective studies | ||||
| No studies available | ||||
| Cross-sectional studies | ||||
| N=1,211 | Presence of lacunar infarcts (=200); Presence of white matter lesions (n= 384). | Age, sex, study site, hypertension, hypercholesterolemia, diabetes status and cigarette-smoking status | Individuals with venular widening (no definition provided), versus individuals without venular widening, had a higher odds for both the presence of a lacunar infarct and white matter lesions (odds ratio [95% CI 2.27 [1.41; 3.654], though these associations were not statistically significant for either the presence of a lacunar infarct (odds ratio 1.39 [0.87-2.20], or the presence of white matter lesions (odds ratio [95%CI] 1.44 [0.95; 2.17]). | Liew, 2012 (46); |
| N=1,185 | Presence of white matter hyperintensities (2% had none; 36% had one white matter hyperintensity; 31% had two white matter hyperintensities; 31% had three or more white matter hyperintensities). | Age, sex ethnicity, hypertension, body-mass index, glycated hemoglobin, diabetes, total cholesterol, high density lipoprotein cholesterol, log C-reactive protein, coronary artery disease, years of education and smoking habit, presence of retinopathy | Wider venular diameter (per micrometer) was not associated with the odds of the presence of white matter hyperintensities (odds ratio [95% CI] 1.00 [0.99; 1.01]). | Hughes, 2016 (47); |
| N=1,065 | Gray and white matter volume | Age, sex, cohort, and the other vessels, arteriolar diameter, systolic blood pressure, diastolic blood pressure, body-mass index, total cholesterol, high-density lipoprotein cholesterol, diabetes mellitus, and smoking | Wider retinal venular diameter was borderline significantly associated with white matter volume, but not with gray matter volume (beta per SD increase in retinal venular diameter for the standardized mean difference in white and grey volume [95% CI], respectively, -0.07 [-0.14; 0.00] and 0.06 [-0.02; 0.13]). | Ikram, 2013 (48); |
| N=1,013 (acute stroke patients) |
Cortical and subcortical cerebral atrophy (n=424). | Age, sex, study site, hypertension, hypercholesterolemia, diabetes, and smoking status | Wide retinal venular diameter (defined as highest quintile versus other quintiles combined as reference group) was not significantly associated with cortical and subcortical cerebral atrophy (wide versus narrow retinal venular diameter, OR [95%CI] 1.3 [0.8-2.0]). | Baker, 2010 (49); |
| N=1,114 (N=557 stroke cases and N=557 controls) |
Lacunar stroke (n=261). | Age, sex, race, smoking status, hypertension status, diabetes mellitus status, and hypercholesterolemia status | Wider retinal venular diameter (per SD) was significantly associated with a greater odds for the presence of lacunar stroke (per SD wider retinal venular diameter OR [95%CI] 1.60 [1.25-2.06]). | Ong, 2013 (50); |
| Arteriole-to-venule ratio | ||||
| Prospective studies | ||||
| No studies | ||||
| Cross-sectional studies | ||||
| N=1,635 | Cerebral infarct (defined as a lesion ≥3 mm in maximum diameter in a vascular distribution with typical imaging characteristics). | Age, race, center, sex, 6-year mean arterial blood pressure, antihypertensive medication use, fasting blood glucose, diabetes, smoking status, plasma total and high-density lipoprotein cholesterol, plasma triglycerides, and serum fibrinogen | In individuals with hypertension a low arteriole-to-venule ratio (defined as lowest quintile), versus high arteriole-to-venule ratio (defined as highest quantile) was significantly associated with a higher odds of a cerebral infarct (odds ratio [95%CI] 3.74 [1.51; 9.24]; however this association was not significant in individuals without hypertension (odds ratio [95%CI] 0.64 [0.22; 1.87]). | Cooper, 2006 (51); |
| Flicker light-induced increase in retinal microvascular diameters | ||||
| Prospective studies | ||||
| No studies | ||||
| Cross-sectional studies | ||||
| Retinal blood flow | ||||
| Prospective studies | ||||
| No studies | ||||
| Cross-sectional studies | ||||
| OCT-A | ||||
| Prospective studies | ||||
| No studies | ||||
| Cross-sectional studies | ||||
| N=85; (N=46 individuals with a recent subcortical infarction; and N=39 healthy controls). | A cerebral small vessel disease composite score was calculated (based on the presence of lacunes, white matter hyperintensities, cerebral microbleeds, and enlarged perivascular spaces). | Age, sex, systolic and diastolic blood pressure, diabetes, dyslipidemia | Individuals with a recent subcortical infarction had a significantly lower length of micro vessels of the perfused inner retina per unit area in millimeters squared (mm2) than healthy controls (mean difference [95%CI] between groups was −0.087 [−0.145; -0.029] mm2). Greater inner retina length of micro vessels of the perfused inner retina per unit area in millimeters squared (mm2) was significantly associated with greater cerebral small vessel disease burden (beta [95%CI] per point for the cerebral small vessel disease burden score, 0.067 [0.013–0.121]). | Kwapong. 2022 (52); |
| N=77; (N=47 individuals with cerebral small vessel disease features; and N=39 healthy controls). | Cerebral small vessel disease was assessed based on the presence of lacunar infarcts, and white matter hyperintensities (white matter hyperintensities were assessed with the Fazekas score). | Age, hypertension, diabetes, and hyperlipidemia | Greater temporal superficial retinal capillary plexus vascular density was significantly associated with a lower periventricular white matter hyperintensities Fazekas score and a lower deep white matter hyperintensities Fazekas score (beta per mm greater vascular density, -1.772 [-3.048; -0.497]; and -1.427 [-2.679; -0.175]); however, there was no significant association with less lacunar infarcts -0.160 [-0.373; 0.054]). However, these associations were not found when the radial peripapillary capillary network was investigated. | Wang, 2021 (53); |
| N=74; all individuals without stroke or dementia. | White matter hyperintensities were scored according to the Fazekas score. | Age, sex, hypertension, and signal strength index | Participants with Fazekas score ≥2, versus participants with Fazekas score ≤ 1, had a significantly lower radial peripapillary capillary network density (P = 0.02) and deep retinal capillary plexus network density (P = 0.012), but did not have a significantly lower superficial retinal capillary plexus network. | Peng, 2020 (54); |
| N=69; (N=29 individuals with amyloid-positive AD-related cognitive impairment; N=25 individuals with subcortical vascular cognitive impairment; and N=15 amyloid-negative cognitively normal individuals). | The following cerebral small vessel disease features were assessed: presence of lacunes, white matter hyperintensities, and cerebral microbleeds. | Age, sex, hypertension, diabetes, and image quality score | In individuals with subcortical vascular cognitive impairment, versus normal controls, the capillary density in the inferior radial peripapillary capillary network was significantly lower (P=0.001). However, individuals with amyloid-positive AD-related cognitive impairment, versus normal controls, the capillary density in the inferior radial peripapillary capillary network was not significantly lower (P>0.05). | Lee, 2020 (55); |
| N=16; (N=7 individuals with MCI; and N=9 individuals with AD). | Forebrain parenchyma, cortical gray matter, inferolateral ventricle, lateral ventricle, and hippocampus volumes. | None | Inferolateral ventricle volume inversely correlated with the vascular density in the 6-mm circle (ρ = −0.565, P = .028) and 3-mm ring (ρ = −0.569, P = .027) and the perfusion density in the 3-mm ring (ρ = −0.605, P = .0169). Forebrain, cortical gray matter, later ventricle, and hippocampus volumes did not significantly correlate with either vascular density or perfusion density (P >.05). | Yoon, 2019 (56); |
Table 2 shows the associations of neuronal and microvascular structures and function of the neurovascular coupling unit, as quantified by retinal imaging, with cerebral MRI features of structural brain abnormalities.
The following MRI features were considered: total brain volume, white matter hyperintensities, lacunar infarcts, and cerebral microbleeds. Fully adjusted results are shown (where appropriate, i.e. when associations where adjusted for confounders).
RNFL, retinal nerve fiber layer thickness; NR, not reported; SD, standard deviation; CI, confidence interval; MCI, mild cognitive impairment; AD, Alzheimer’s disease.