Wall-to-Lumen Ratio of Retinal Arterioles and Arteriole-to-Venule Ratio of Retinal Vessels in Patients with Cerebrovascular Damage

Abstract

purpose. There is evidence that generalized retinal arteriolar narrowing, which can be measured by the arteriole-to-venule ratio (AVR) of retinal vessels, predicts cerebrovascular events. The wall-to-lumen ratio (WLR) and wall cross-sectional area (WCSA) of retinal arterioles reflect structural arteriolar parameters. The primary objective was to test the association between WLR and AVR in a distinct cohort of patients with cerebrovascular damage.

methods. In this cross-sectional study, 23 patients (57.5 ± 9.4 years) with acute transitory ischemic attack or lacunar cerebral infarct were compared with two age-matched control groups: 83 subjects with essential hypertension (53.7 ± 5.5 years) and 16 normotensive subjects (52.2 ± 8.3 years). Retinal arteriolar parameters (WLR, WTH, and WCSA) were assessed in vivo with scanning laser Doppler flowmetry (SLDF). AVR and a qualitative evaluation of retinal vessels were obtained from digital retinal color photographs. The intima–media thickness (IMT) of the carotid artery was measured.

results. WLR (0.44 ± 0.1 vs. 0.34 ± 0.1 vs. 0.30 ± 0.1, P < 0.001) and carotid IMT (P < 0.05) were significantly greater in the cerebrovascular event group compared with normotensive subjects. WLR and WCSA were significantly higher in the cerebrovascular event group compared with subjects with mild arterial hypertension. AVR was similar in all three study groups (0.75 ± 0.07 vs. 0.74 ± 0.07 vs. 0.78 ± 0.1, P = 0.18).

conclusions. The increase in WLR and WCSA of retinal arterioles, as well as in IMT in patients with cerebrovascular damage suggests vascular hypertrophy in the microvascular and macrovascular bed. The lack of association between AVR of retinal vessels and WLR of retinal arterioles may point to different stages of cerebrovascular disease and/or different pathophysiological changes in the arteriolar wall.

Retinal and cerebral vessels share similar anatomic, embryologic, and functional features. ¹ An autopsy study in patients with strokes showed a close correlation between cerebral and retinal vascular abnormalities. ² Retinal vascular abnormalities have been shown to be associated with an adverse cardiocerebrovascular prognosis. ³ Much research effort over the past years has been focused on the evaluation of new noninvasive in vivo approaches to assess retinal vascular abnormalities as an indicator for cardiocerebrovascular damage. Of these, especially a decrease in the arteriole-to-venule ratio (AVR) of retinal vessels was demonstrated to be associated with increased incidence of cardiocerebrovascular events. ^{4 5}  

AVR is sensitive to changes in both arterial and venule diameter, but the single assessment of retinal arteriolar structure is expected to better reflect vascular target organ damage than the assessment of AVR of retinal vessels. This notion is supported by the fact that established biomarkers of target organ damage, including the intima–media-thickness (IMT) of carotid arteries, ⁶ are based on the measurements of arterial vessels, whereas established biomarkers for cardiocerebrovascular target organ damage based on measurements of venous vessels are lacking. AVR reflects changes in inner vessel diameter and is significantly associated with generalized narrowing of retinal arterioles. 

Wall-to-lumen ratio (WLR) of retinal arterioles represents a structural retinal arteriolar parameter, which has been shown to reveal a pronounced increase in hypertensive subjects with a history of a cerebrovascular event. ⁷ An overall increase in WLR can be a consequence of wall thickening, narrowing of the lumen or a combination of both. 

Vascular changes in arterial hypertension include alterations of the vessel wall and the vessel diameter. Arteriolar eutrophic inward remodeling of small arteries, the preferred physiological mechanism in arterial hypertension by which wall stress can be normalized, ⁸ is characterized by a rearrangement of the preexisting cells in the vascular wall increasing the number of cell layers and reducing the internal lumen diameter. ^{9 10 11} When hypertension is severe, accelerated or rapidly increased eutrophic inward remodeling is accompanied or even replaced by vascular wall hypertrophy. Hypertrophy of the vascular wall means growth of smooth muscle cells, which is reflected by an increase in vessel wall thickness and increase in cross-sectional area of the vascular wall (WCSA). 

A direct comparison between WLR of retinal arterioles and AVR of retinal vessels as two different indicators of vascular damage has not been performed so far. The present study was conducted to fill this gap. The primary objective of this study was to assess WLR of retinal arterioles and AVR of retinal vessels in essential arterial hypertensive subjects with an acute cerebrovascular event, in patients with essential arterial hypertension without target-organ damage and in normotensive control subjects to evaluate the association of both parameters with cerebrovascular damage. The secondary objectives of our study were to elucidate whether changes in WLR of retinal arterioles and AVR of retinal vessels occur in parallel and to analyze whether structural alterations of retinal arterioles in hypertensive subjects with a history of a cerebrovascular event are due to vascular hypertrophy or eutrophic remodeling or a combination of both processes. 

Methods

Study Population and Study Design

This cross-sectional study was based on a collaboration between the Department of Ophthalmology and Neurology and Nephrology and Hypertension of the University of Erlangen-Nürnberg. The study participants were either patients from one of these collaborating Departments or were recruited via advertisement in local newspapers. Informed consent was obtained from each participant, before enrollment in the study. The study protocol was approved by the Clinical Investigations Ethics Committee of the University Erlangen-Nürnberg. The study was conducted in accordance with Good Clinical Practice guidelines and in compliance with the Declaration of Helsinki. 

The study population consisted of three different groups. First, hypertensive subjects with an acute cerebrovascular event (determined as transitory ischemic attack [TIA] or lacunar cerebral infarct diagnosed by a neurologist within 1 to 5 days before examination of retinal vasculature). The term lacunar infarct refers to a well-defined, subcortical ischemic lesion at the level of a single perforating artery with a diameter ranging between 100 and 400 μm (sizes similar to the diameter of retinal arterioles) generally originating at right angles directly from the main cerebral arteries. ¹²  

Second, subjects with essential hypertension stage 1 or 2 (defined as daytime systolic BP ≥ 135 mm Hg and diastolic BP ≥ 85 mm Hg ¹³ or on antihypertensive treatment with at least one antihypertensive drug according to the European Guidelines of Hypertension ¹⁴ ) without a history or a clinical sign of a cerebrovascular event. Exclusion criteria were renal impairment (defined by serum creatinine above 1,2 mg/dL), diabetes mellitus (defined by a fasting glucose ≥ 126 mg/dL), any form of secondary arterial hypertension, eye disease other than grade I or II hypertensive retinopathy, atrial fibrillation, atrioventricular block II or higher, dilatative cardiomyopathy, a history of vasculitis, seizure disorder, and treatment with any vasoconstrictive drug. 

The third group was the control, made up of healthy normotensive subjects without a history or a clinical sign of a cerebrovascular event or other vascular organ damage. 

In each study participant, both eyes were studied, but only data from the right eye were entered the statistical analyses. 

Measurement of the AVR

Retinal digital photographs were taken in both eyes with a digital fundus camera (non-myd α-d digital; Kowa, Tokyo, Japan) fundus camera that produced lossless compressed RGB photographs of size 1600 × 1216 pixels with image resolution of approximately 5 μm. The 45° retinal photograph was centered on the region of the optic disc. The grading procedure was assessed by trained ophthalmologists unaware of the participant’s identity in the central telemedical reading center. AVR of retinal vessels was calculated according to the scale-dependent formulas described by Parr and Spears ^{15 16} and Hubbard et al. ¹⁷ using the semiautomatic system used in the ARIC (Atherosclerosis Risk in Communities) Study (Retinal Analysis; Optimate, Madison, WI). The diameters of six arterioles and venules with a diameter more than 40 μm coursing through an area half to one disc diameter from the margin of the optic disc were measured in each eye. Qualitative evaluation of the photographs comprised the assessment of retinal microvascular abnormalities. Arteriolar alterations were defined as the presence of any of the following lesions: generalized arteriolar narrowing, focal arteriolar narrowing, and arteriovenous nicking. 

Measurement of WLR and Other Retinal Arteriolar Parameters

Scanning laser Doppler flowmetry (SLDF at 670 nm, HRF; Heidelberg Engineering, Germany) ^{18 19 20} was performed to examine the retinal vessel structure and assess the retinal capillary blood flow. The method is described in detail elsewhere. ²⁰ Here, a brief description: To analyze retinal arteriolar structure, a retinal temporal superior arteriole with a size between 80 and 140 μm in a retinal sample of 2.56 × 0.64 × 0.30 mm was scanned by a confocal scanning laser system (670 nm) within 2 seconds at a resolution of 256 points × 64 lines × 128 lines. A specific length of the arteriole reflecting arteriolar structure during one heart beat (one systole and one diastole) was used for analyses, and the diameters were assessed every 10 μm of this specific length of the arteriole; the mean of measured diameters were finally considered for analyses. The measurement was performed in both eyes within one papilla diameter (2–3 mm temporal superior to the optic nerve); the mean of 3 measurements was taken. Analyses were performed with automatic full-field perfusion imaging analysis software (AFFPIA; Software SLDF ver. 3.7). ²¹ The reflectivity image created from the reflected laser light of the nonmoving tissue and the flow image generated by the Doppler effect of all moving blood corpuscles allowed calculation of the outer arteriole diameter (AD) in reflection images and lumen diameter (LD) in perfusion images. WLR, WTH, and WCSA were calculated according the formulas (AD − LD)/LD, (AD−LD)/2, and (π/4)(AD² − LD²), respectively. A growth index, describing the percentage of difference between the average WCSA of hypertensive arterial vessels and normotensive arterial vessels, was calculated according to Heagerty et al., ¹¹ expanding a previous observation of Baumbach and Heistad, ²² according to the formulas:  

\[\mathrm{Growth\ index\ of\ hypertensive\ group}\ {=}\ (\mathrm{WCSA}_{\mathrm{hypertensive}}\ {-}\mathrm{WCSA}_{\mathrm{normotensive}})/\mathrm{WCSA}_{\mathrm{normotensive}}\]

 

\[\mathrm{Growth\ index\ of\ cerebrovascular\ group}\ {=}\ (\mathrm{WCSA}_{\mathrm{cerebrovascular\ event}}\ {-}\mathrm{WCSA}_{\mathrm{normotensive}})/\mathrm{WCSA}_{\mathrm{normotensive}}\]

The examination was performed in sitting position after 20 minutes of rest, at room temperature and in daylight conditions between 8 AM and 2 PM, but before lunch. The readers (Joanna Harazny and Christiane Köhler) were masked to the blood pressure (BP), metabolic, and cerebral disease status of the patients. 

Assessment of Carotid IMT, BP Measurements, and Laboratory Parameters

Sonography (B-mode, Adara Sonoline; Siemens, Erlangen, Germany) of the carotid arteries was performed for determining the IMT of the common carotid artery. Only the IMT of right carotid artery was considered for analysis. 

BP was measured by ambulatory 24-hour BP-monitoring in the hypertensive and normotensive control group or by inpatient 24-hour BP-monitoring in the stroke unit in the cerebrovascular group. Only daytime BPs were considered for diagnosing arterial hypertension, as many participants did not complete the night measurements. 

All patients had routine blood examinations including evaluation of LDL, HDL, total cholesterin, and C-reactive protein. 

Statistics

Commercial software (SPSS software release ver. 14.0; SPSS Inc. Chicago, IL) was applied for statistical analysis. Results are expressed as the mean ± SD in the text and tables and as the mean ± SEM in the figures. For comparison between groups Mann-Whitney U test was used. To analyze all three groups combined Kruskal-Wallis test was used. For comparison with respect to sex and the use of antihypertensive drugs across study groups, the χ² test was used. Univariate correlation analyses were performed using Pearson’s correlation coefficient for parametric data and Spearman’s rho for nonparametric data. To take into account the correction for multiple testing the Bonferroni correction ²³ was performed by dividing the level of significance (without correction) α = 0.05 by the number of comparisons (three groups). The Bonferroni-corrected level of significance was 0.0166. 

Results

Baseline Characteristics of the Study Population

Twenty-three subjects (57.5 ± 9.4 years) with acute cerebrovascular events (12 patients with TIA and 11 patients with lacunar stroke) were recruited from the stroke unit. All subjects had essential arterial hypertension. Thirteen (56.5%) of these were on antihypertensive medication before the acute cerebrovascular event. Twenty in the cerebrovascular event group received immediate treatment with antiplatelet agents; 10 were treated with one antiplatelet agent alone, 2 received a combined treatment of two antiplatelet agents (aspirin and clopidogrel); and 8 received a combination of an antiplatelet agent with low-molecular-weight heparin. Two patients were treated with intravenous heparin alone, and no patient in the study group underwent thrombolysis or oral anticoagulation with warfarin. 

The hypertensive, otherwise healthy group comprised 83 subjects (53.7 ± 5.5 years). Among those, 32 had newly diagnosed and untreated (38.6%) hypertension, and 51 (61.4%) were on antihypertension medication. The normotensive control group consisted of 16 healthy subjects (52.2 ± 8.3 years). The clinical baseline characteristics of the study participants are shown in Table 1 . 

Blood Pressure

Daytime systolic and diastolic BPs were lower in the normotensive group than in the cerebrovascular event and hypertensive groups (both statistically significant, P < 0.001). Daytime diastolic BP was slightly higher, but not statistically significant in the hypertensive group compared with the cerebrovascular event group (P = 0.052), and daytime systolic BP was similar between the hypertensive group and the cerebrovascular event group (P = 0.824). Concerning the patients treated for arterial hypertension we found no significant differences in the frequency of use of antihypertension drug classes (all P > 0.20; χ² test) between the treated hypertensive and the treated cerebrovascular group (Table 2) . 

Arteriolar Narrowing

A qualitative evaluation of retinal arteriolar abnormalities is shown in Table 3 . The hypertensive patients with a cerebrovascular event were more likely to have retinal abnormalities than were the normotensive control subjects or hypertensive patients, although this was statistically significant only for the qualitative evaluation of generalized arteriolar narrowing. 

Generalized narrowing of retinal arterioles analyzed by the AVR of retinal vessels was similar across the three study groups (0.78 ± 0.1 in the normotensive control group vs. 0.74 ± 0.07 in the hypertensive group vs. 0.75 ± 0.07 in the cerebrovascular group, P = 0.179, Kruskal-Wallis test; Fig. 1 ). A nonparametric intergroup comparison by Mann-Whitney U test showed no significant difference between any of the study groups with respect to AVR (P = 0.124–0.715). 

Scanning Laser Doppler Flowmetry

Table 4provides an overview of BP levels and quantitative retinal vascular parameters (raw data) measured by SLDF in the three study groups. 

Diameter of Retinal Arterioles

Arterial outer diameter of retinal arterioles determined by SLDF did not differ significantly among the three study groups (110.62 ± 13.4 μm vs. 107 ± 10.9 μm vs. 110.72 ± 13.5 μm, P = 0.331). However, lumen diameter of retinal arterioles determined by SLDF tended to be different when all three study groups were analyzed combined, but did not reach the Bonferroni-corrected level of significance (P = 0.028). When the cerebrovascular event group was compared with the normotensive group, lumen diameter was significantly lower in the cerebrovascular event group (76.52 ± 6.45 μm vs. 85 ± 10.9 μm, P = 0.01). The lumen diameter in the cerebrovascular group appeared to be somewhat lower than in the hypertensive group (76.52 ± 6.45 μm vs. 80 ± 9.2 μm, P = 0.112), but the difference was not statistically significant. 

WLR of Retinal Arterioles

WLR was highest in patients with a history of a cerebrovascular event compared with the hypertensive and normotensive groups (0.44 ± 0.10 vs. 0.34 ± 0.11 vs. 0.30 ± 0.10, P < 0.001; Fig. 2 ). WLR was significantly higher in the cerebrovascular event group when compared with the hypertensive group (0.44 ± 0.10 vs. 0.34 ± 0.11, P < 0.001) or with normotensive subjects (0.44 ± 0.10 vs. 0.30 ± 0.10, P < 0.001), but there was no difference between the hypertensive and normotensive group (0.34 ± 0.11 vs. 0.30 ± 0.10, P = 0.224). 

WCSA of Retinal Arterioles

WCSA was highest in patients with a history of a cerebrovascular event compared with the hypertensive and normotensive groups (5133 ± 1713 μm² vs. 3991 ± 1291 μm² vs. 3980 ± 1359 μm², P = 0.014, Kruskal-Wallis test; Fig. 3 ). WCSA was significantly higher in the cerebrovascular event group when compared with the hypertensive control subjects (P = 0.004) and tended to be higher when compared with normotensive subjects (P = 0.04). The difference in WCSA was also not significantly different between the hypertensive and normotensive subjects (P = 0.887). 

Growth Index of the Retinal Arteriole Vessel Walls

The growth index of retinal arterioles was 28.9% in the cerebrovascular event group and 0.27% in the hypertensive group. 

IMT of the Right Common Carotid Artery

IMT of the right carotid artery was significantly higher in the cerebrovascular event group (0.96 ± 0.17 mm) when compared with normotensive subjects (0.78 ± 0.19 mm, P = 0.004), but it did not reach the level of significance between hypertensive and normotensive subjects (0.87 ± 0.21 mm vs. 0.78 ± 0.19 mm, P = 0.053) or between hypertensive and cerebrovascular patients (0.87 ± 0.21 mm vs. 0.96 ± 0.17 mm, P = 0.034; Fig. 4 ). 

Correlations

Correlation analyses did not reveal any significant correlation between AVR and WLR (r = 0.004, P = 0.9) nor between AVR and carotid IMT (r = 0.000, P = 0.9). Furthermore, we did not find correlations between WLR of retinal arterioles and carotid IMT (r = 0.020, P = 0.8). A similar lack of relationship was observed for wall thickness of retinal arterioles and carotid IMT (r = 0.010, P = 0.9) as well as for WCSA of retinal arterioles and carotid IMT (r = 0.005, P = 0.9). 

Influence of Underlying BP Medication on Main Outcome Measures

Consideration of treatment for BP did not affect intergroup differences in our main outcome measures WLR and AVR. AVR did not differ either between the hypertensive untreated group (0.78 ± 0.10) and the normotensive group (0.78 ± 0.10, P = 0.9) or between the hypertensive untreated group (0.78 ± 0.10) and the cerebrovascular untreated group (0.77 ± 0.09, P = 0.7). Intergroup differences in WLR were also not influenced by the absence of BP medication, WLR remaining significantly higher in the cerebrovascular untreated group (0.49 ± 0.08) over the normotensive (0.31 ± 0.10, P < 0.001) and the hypertensive untreated group (0.34 ± 0.10, P < 0.001). 

When only untreated patients were included in the analysis, WCSA was significantly higher in the cerebrovascular group than in the normotensive group (6130 ± 1351 μm² vs. 3980 ± 1359 μm², P = 0.001), suggesting a positive influence of the underlying BP medication on vascular hypertrophy in the cerebrovascular group. 

Discussion

This is the first study to directly compare WLR of retinal arterioles and AVR of retinal vessels in a cohort of patients with cerebrovascular damage. In our cohort, WLR of retinal arterioles was highest in hypertensive subjects with an acute cerebrovascular event compared with hypertensive and normotensive subjects. AVR of retinal vessels did not differ among the three study groups. Therefore, our data indicate that WLR of retinal arterioles might be a promising indicator of cerebrovascular damage. 

Critical vascular parameters in calculating the AVR of retinal vessels are arteriolar and venular diameter obtained from color photographs, in which the column of the moving blood particles can be visualized. Thus, vessel diameter from color photographs is defined as the diameter of the column of moving blood particles (surrounded by a transparent plasma edge stream) and does not represent the actual diameter of the vessel. Plasma edge stream and vessel wall cannot be visualized by this method. A decreased AVR of retinal vessels is widely considered a parameter of retinal arteriolar narrowing. AVR of retinal vessels was observed to be decreased in several studies in circumstances of peripheral vasoconstriction as widely present in hypertensive subjects. ^{5 24 25 26} However, growth of arteriolar wall components (i.e., hypertrophy or hyperplasia of smooth-muscle cells) in response to BP or metabolic factors (i.e., elevated serum glucose) cannot be detected when considering AVR of retinal vessels as a single parameter. Moreover, concomitant changes in retinal arterial and venule diameter (i.e., vasodilation in the absence of vascular tone) may lead to normal AVRs although structural alterations of retinal arterioles are already present. Abnormalities in venule diameter may therefore mask abnormalities of arteriolar structure in some subjects when only AVR is used as an indicator of cerebrovascular damage. 

By SLDF, the outer and the lumen diameters are assessable. This allows the calculation of WLR. As new parameter, WLR of retinal arterioles is increased in circumstances of a decrease in lumen diameter (i.e., in terms of vasoconstriction) or growth of arteriolar wall components (i.e., smooth muscle cells) or a combination of both processes. It takes only arteriolar changes into account and is independent of changes in venule structure. 

An important finding in our study is that hypertensive subjects predominantly undergo eutrophic inward remodeling, suggested by slightly increased WLR, slightly decreased lumen diameter, and almost unchanged WCSA compared with normotensive subjects. There was almost no growth response of the arteriolar wall (growth index, 0.27%) in the hypertensive control group in our cohort. This finding is in contrast to previous studies examining small artery and arteriolar structure that found at least a mild growth response of the vessel wall in animals and humans with essential hypertension. ²⁷ Although differences are not statistically significant, there is a trend to vasoconstriction detected by a decreased AVR and decreased arteriolar lumen diameter measured by SLDF. In the literature, the growth index calculated in vessels from either animal or human small arteries ranges up to 28% in chronic essential hypertension. In spontaneously hypertensive rats it was reported that femoral small arteries had a growth index of 18%. ²⁸ Park and Schiffrin ²⁹ found a growth index of 28% in resistance arteries dissected from gluteal subcutaneous (SC) tissue in human subjects with mild hypertension. In another study conducted by our group an average growth index of 18% was observed in overweight male human subjects with never-treated, mild-to-moderate essential hypertension. It is difficult to explain why the hypertensive group in our present study showed no growth response at all. One explanation could be the underlying antihypertension medication. This hypothesis is supported by the finding that, WCSA was significantly higher in the cerebrovascular untreated group compared with the normotensive group and compared with the cerebrovascular treated group, suggesting a positive influence of the underlying BP medication on vascular hypertrophy in the cerebrovascular group. Similarly, when the hypertensive control group into was divided into two subgroups, subjects treated and not treated for arterial hypertension, WCSA was slightly higher in the untreated subgroup (4036 ± 1066 μm² vs. 3964 ± 1424 μm², P = 0.6), but the difference was not statistically significant. 

There are recent reports suggesting that antihypertensive agents correct vascular remodeling of SC small arteries. ³⁰ To distinguish between the normotensive and hypertensive controls we applied very strict diagnostic criteria based not only on a single casual BP measurement, but on ambulatory daytime BP obtained from at least 30 measurements. Consequently, our hypertensive group consisted of patients with mild arterial hypertension, of which 38.6% were not aware of suffering from arterial hypertension and probably had a very short previous duration of disease. Therefore changes in vessel wall might not have yet occurred. It could also explain why there was no significant difference in AVR: Patients in the hypertensive group differed in the stage of hypertensive disease when compared with the hypertensive groups whose results have been so far exclusively reported. Of note, a decrease in AVR of retinal vessels was also found in prehypertensive and normotensive subjects, reflecting the fact that vasoconstriction of peripheral vessels even precedes the onset of hypertension in some susceptible subjects. ³¹ The lack of differences in AVR between the normotensive and the hypertensive group could indicate that a part of the normotensive subjects could have actually been prehypertensive. This hypothesis is supported by the unexpectedly large amount of arteriolar narrowing and AV nicking in the normotensive group (Table 3) . This fact, and also that in many of the subjects in our hypertensive group the hypertension was newly diagnosed and probably of short duration, may have led to the lack of an association between arterial hypertension and AVR of retinal vessels in our cohort. 

Subjects with a cerebrovascular event revealed a pronounced growth response of the retinal vascular wall (i.e., hypertrophy or hyperplasia of vascular smooth muscle cells) compared with normotensive subjects. This group had a pronounced increase in arteriolar WLR and WCSA with a growth index of 28.9% (compared with the normotensive group). This finding is in line with a previous report examining arterial structure of small arteries obtained from SC tissue that found that subjects with a growth response of vascular wall components had a higher risk of cardiovascular events than subjects without a growth response. ³²  

We found that AVR of retinal vessels was slightly greater (not statistically significant) in the cerebrovascular event group compared with the hypertensive group, although lumen diameter measured by SLDF tended to be smaller. We did not assess changes in the venular vasculature, which may be different in an acute cerebrovascular event than in chronic diseases like arterial hypertension. AVR was shown to be a reliable prognostic factor for incident stroke, but it was never assessed as an acute indicator of cerebrovascular damage that had already occurred. Therefore, AVR may not be changed in the postacute phase after a cerebrovascular event. The immediate therapy with antiplatelet agents in the cerebrovascular group may be another reason for our finding regarding the AVR. Administration of antiplatelet agents remains the core of management for preventing recurrent stroke and other cardiovascular events in at-risk patients. ³³ The most important antiplatelet agent, aspirin, is well known to inhibit cyclooxygenase-mediated prostaglandin synthesis, leading to an upregulation of the concentration of prostaglandin receptors in retinovascular tissues. ³⁴ There is evidence that aspirin increases significantly retinal blood flow in patients with minimal diabetic retinopathy (type 1 diabetes mellitus), but not in patients without retinopathy. ³⁵ The Blue Mountains Eye Study further demonstrated that in individuals who were on antihypertensive medication regular use of aspirin was associated with increasing retinal arteriolar diameters obtained from digitized retinal photographs, this effect being more pronounced at 5-year follow-up. Neither antihypertensive medication use alone nor aspirin use alone was associated with wider retinal vessels. ³⁶ It is possible that the systemic antiplatelet therapy combined with the antihypertensive medication administered immediately after the cerebrovascular event had vasodilatatory effects on retinal arterioles, leading to a not significantly reduced AVR in this study group compared with normotensive subjects. Measurements of arteriolar outer and lumen diameter performed by SLDF may consequently also be influenced by the treatment. Magnusson et al. ³⁷ has shown that intravenous infusions of pentoxifylline, a hemorrheologic drug widely used for the treatment of intermittent claudication and occasionally in diseases affecting retinal blood flow, such as diabetic retinopathy, ^{38 39} influences retinal blood flow measurements performed by SLDF leading to an increase in retinal blood flow during diastole, but not during systole. ⁴⁰ This effect may be mediated by increasing deformability of both erythrocytes and leukocytes as well as by a possible direct vasodilatation. To our knowledge there is no published evidence of the short-term effect of antiplatelet drugs on retinal vessels measured from retinal photographs or by SLDF. 

WLR of retinal arterioles and IMT of carotid artery were both highest in hypertensive subjects with a cerebrovascular event and revealed a similar pattern among the three study groups. Thus, microvascular and macrovascular arteriolar changes occurred in parallel in our cohort. However, correlation analysis did not reveal a relation between retinal arteriolar changes and IMT of carotid artery. 

The process of vascular remodeling may differ between microvascular and macrovascular arterial vessels. This possibility is supported from a previous study examining small artery structure obtained from SC tissue that found that small artery remodeling might be the earliest form of target organ damage in hypertension occurring before other cardiovascular abnormalities are detectable. ²⁹ The present study provides further evidence that vascular remodeling is a characteristic feature in hypertension and that it is present in small arteries with a lumen diameter of 100 to 350 μm (sizes similar to the diameter of retinal arterioles). It has been shown that vascular remodeling is a reversible dynamic process. ³⁰ Moreover, the severity of vascular remodeling of SC small arteries has prognostic significance over a 10-year period: Subjects with a growth response of vascular wall components of SC small arteries reveal adverse cardiovascular prognosis. ³²  

Whereas assessment of small artery structure through biopsies of subcutaneous tissue is invasive, the assessment of retinal arteriolar structure using SLDF allows noninvasive assessment of the structural changes of retinal arterioles. We have recently shown that retinal arterioles undergo changes similar to those of SC small arteries in essential hypertension. ⁴¹  

Eutrophic inward remodeling reflects a functional vascular adaptation to prolonged vasoconstriction and is a pressure-induced response to arterial hypertension. Vascular hypertrophy is not an usual pressure-induced change, being a less efficient compensatory mechanism than eutrophic inward remodeling. 

We did not find any direct correlation between AVR and WLR. AVR represents quantitatively only a generalized narrowing of retinal arterioles in the condition when retinal venules do not change. It has been shown that venular diameters are not related to increasing BP. ⁴² AVR represents a functional change of retinal arterioles and can be used as a parameter of eutrophic inward remodeling, whereas AVR is not able to detect vascular wall hypertrophy. In contrast, WLR of retinal arterioles can be increased in case of eutrophic inward remodeling, or in case of hypertrophy, or in a combination of both. With information about the WLR and WCSA, the type of remodeling might become clear. Increased WLR and simultaneous increased WCSA suggest hypertrophy; increased WLR and unchanged WCSA suggest eutrophic remodeling. 

Limitations

Our study has several important limitations. Our patient group was characterized by mild cerebrovascular damage. We included only patients with TIA and cerebral lacunar infarct with mild cognitive and physical impairment. These criteria were necessary because the technique of our method required fixation of the eye in a sitting position, necessitating good cooperation of the patient. Our findings may thus not be transferable to other types of stroke. 

Our study is a cross-sectional study, and the data assessment was performed after the acute cerebrovascular event had happened, whereas AVR has been applied in longitudinal population-based studies and is used as a predictor of future cerebrovascular events. The AVR measured in our cohort of patients with cerebrovascular damage may rather reflect acute vascular pathologic changes that take place directly after an acute cerebrovascular event due to compensatory mechanisms and due to the therapeutic measures. 

Furthermore, we had no data from color photographs concerning the measurements of all arteriolar diameter and venule diameter in absolute values. We had only data concerning the ratio of arteriole and venule diameters. Use of a ratio protects against variable magnification of photographs. Because of several unknown parameters during the acquisition of the fundus photographs, such as refraction indices of the cornea and vitreous or refractive errors of the lens, it is very difficult to obtain the exact size of the retinal image. Approaches presented in the literature give more or less an approximation only. The equations for the calculation of the AVR are denoted as the Parr-Hubbard formula and are based on the hypothesis that retinal veins do not change due to hypertension. In the ARIC Study this hypothesis was thought to be confirmed, ⁴² but a more recent study has shown that retinal vein diameters are variable and may play their own independent role in predicting cardiovascular disorders. ⁴³  

The different location of retinal vessels’ measurements performed by two different procedures is another major limitation. With SLDF, only one major retinal artery was assessed, which is located temperosuperiorly near the optic disc. The outer retinal arterial and lumen diameters measured by SLDF reflect the mean of three measurements of only one vessel. AVR is the average of the diameters of six different retinal arterioles and venules located between two circular zones in an area half to one disc diameter from the margin of the optic disc. The primary location for retinal arteriolar narrowing are precapillary arterioles, which are arterioles located not directly near the optic disc. Arteries located immediately next to the optic disc are not much affected. ⁴⁴ This difference may explain the fact that no direct correlation was found between AVR and WLR. Furthermore, because of different angioarchitecture, SLDF measurements of retinal outer and lumen diameter may not be directly comparable in all individuals between groups. 

We had only a small number of study participants in the normotensive and in the cerebrovascular group. Obtained data concerning WLR and WCSA are statistically conclusive. Regarding the AVR, the small number of study participants may have not been sufficient to detect statistically significant changes between normotensive and hypertensive subjects with respect to AVR because of the very low numerical differences in the values obtained. A subsequent statistical power calculation to detect differences in AVR between the study groups showed very low statistical power for our sample sizes: statistical power was 33.4% for the normotensive and the hypertensive groups, 17.9% for the normotensive and the cerebrovascular groups, and 9.3% for the hypertensive and the cerebrovascular groups. 

Finally, the SLDF technique of measuring the inner vessel wall border was based on the assumption that the plasma edge stream within the vessel remains constant. The calculated WTH of retinal arterioles has to consider that the laminar flow in normal vessels builds a protective plasma layer, which shields the endothelial layer from contact with blood corpuscles. ⁴⁵ Therefore, it comprises also the plasma edge stream. In elderly subjects the plasma layer is reduced leading to shear stress to the endothelium. ⁴⁶ In consequence, the difference of absolute values in WTH of retinal arterioles between younger and elderly individuals may be underestimated. Because the study groups were not significantly different in age, we suppose that the distortion of the results may not be very large, and therefore between our study groups comparable. Besides, as the main outcome used is a ratio (WLR) and not the absolute wall thickness in micrometers, distortion of results by the plasma layer is diminished. 

Conclusions

The technique of SLDF allows measurement of WLR of retinal arterioles and assessment of retinal arteriolar structure. In our cohort, we found that WLR of retinal arterioles showed alterations of the vessel wall in patients with cerebrovascular damage compared with that in the control subjects and in hypertensive patients. The AVR of retinal vessels showed no significant difference between groups, and AVR was not directly related to WLR. The lack of association between AVR of retinal vessels and WLR of retinal arterioles indicates that changes in WLR and AVR do not occur in parallel and that the two parameters represent two different indicators of vascular damage, indicating disparate structural and/or functional changes of retinal arteries, and should not be compared directly. They may point to different stages of cerebrovascular disease and/or different pathophysiological changes in the arteriolar wall. 

In hypertensive patients without target-organ damage, alterations of retinal arterioles are due predominantly to eutrophic remodeling. In hypertensive patients with cerebrovascular damage the pronounced increase in WLR and WCSA of retinal arterioles suggests hypertrophy of the microvasculature. This finding is underscored by increased carotid IMT representing wall hypertrophy of conduit arteries. We conclude that a pronounced growth response of arterial wall (i.e., vascular hypertrophy), in both microvascular and macrovascular arterial vessels, was associated with cerebrovascular damage, which is in line with a previous finding in small arteries of SC tissue. 

Future research work has to be performed to clarify whether a combined use of screening by AVR and WLR might be useful independent parameters to identify individuals with early arterial hypertension and advanced risk for cardio and/or cerebrovascular involvement. Clinical longitudinal studies with a larger number of participants are needed to determine whether identification of at-risk patients with increased WLR and a goal-oriented early intervention in the identified subjects may reduce the rate of lacunar infarcts. Structural alterations of the retinal arteriolar wall measured by SLDF in vivo noninvasively might become of predictive value for stroke prevention and for BP therapy monitoring. 

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, April 2008; and the ISH/ESH/DHL (International Society of Hypertension/European Society of Hypertension/German Hypertension Society) Meeting, Berlin, Germany, June 2008.

 Supported by Grants SFB 539 and KFO 109-A4 from Deutsche Forschungsgemeinschaft, Bonn, Germany, and the German Hypertension League.

 Submitted for publication December 5, 2008; revised February 11 and March 9, 2009; accepted June 30, 2009.

 Disclosure: D. Baleanu, None; M. Ritt, None; J. Harazny, None; J. Heckmann, None; R.E. Schmieder, None; G. Michelson, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Clinical Baseline Characteristics of the Study Population

Characteristic	Group 1 Normotensive Control Group (n = 16)	Group 2 Hypertensive Group (n = 83)	Group 3 Cerebrovascular Event Group (n = 23)	P (Group 1 vs. Group 2)	P (Group 1 vs. Group 3)	P (Group 2 vs. Group 3)
Age, y	52.2 ± 8.3	53.7 ± 5.5	57.5 ± 9.4	0.223*	0.061*	0.075*
Sex, n (%)				0.17, †	0.18, †	0.6, †
Male	7 (43.8)	52 (62.7)	16 (69.6)
Female	9 (56.3)	31 (37.3)	7 (30.4)
Antihypertensive treatment, n (%)				<0.001 , †	<0.001 , †	0.8, †
Yes	0	51 (61.4)	13 (56.5)
No	16 (100)	32 (38.6)	10 (43.5)
Body mass index (BMI), kg/m2	24.7 ± 2.7	26.7 ± 3.4	26.3 ± 5.9	0.036*	0.338*	0.757*
Spherical equivalent	−0.98 ± 2.4	−1.11 ± 2.6	0.33 ± 0.96	0.894*	0.046*	0.006*

 Data are the mean ± SD. The Bonferroni-corrected α was 0.0166; significant probabilities are in bold.

*  Mann-Whitney U test.

†  χ² test.

Table 2.

 

Frequencies of Antihypertensive Drug Classes Used in the Hypertensive Control Group and the Cerebrovascular Group

Antihypertensive Drug Class	Hypertensive Group Treated for Arterial Hypertension			Cerebrovascular Group Treated for Arterial Hypertension			P *
		Yes	No	Yes	No
Yes/no	51 (62.2)/32 (38.6)			13 (56.5)/10 (43.5)
β-Blockers	20 (24.1)	27 (32.5)	8 (34.8)	15 (65.2)	0.375
ACE inhibitors	16 (19.3)	31 (37.3)	6 (26.1)	17 (73.9)	0.546
Calcium antagonists	10 (12.0)	37 (44.6)	1 (4.3)	22 (95.7)	0.264
ARBs (angiotensin-II antagonists)	11 (13.3)	36 (43.4)	1 (4.3)	22 (95.7)	0.483
Diuretics	13 (15.7)	34 (41.0)	4 (17.4)	19 (82.6)	0.409
Other	1 (1.2)	—	2 (8.7)	—	—
Unknown	4 (4.8)	—	—	—	—

 Data are number of subjects (%). ACE, angiotensin converting enzyme inhibitors.

*  χ² test comparing results in groups treated for arterial hypertension.

Table 3.

 

Distribution of Retinal Arteriolar Abnormalities

Qualitative Parameter	Group 1 Normotensive Control Group (n = 16)	Group 2 Hypertensive Group (n = 83)	Group 3 Cerebrovascular Event Group (n = 23)	P * (Group 1 vs. Group 2)	P * (Group 1 vs. Group 3)	P * (Group 2 vs. Group 3)
Generalized arteriolar narrowing (qualitative evaluation)
Present	5 (31.3)	8 (9.6)	14 (60.9)	0.019	0.069	<0.001
Absent	11 (68.8)	75 (90.4)	9 (39.1)
Focal arteriolar narrowing
Present	2 (12.5)	9 (10.8)	4 (17.4)	0.614	0.957	0.461
Absent	10 (83.3)	69 (88.5)	19 (82.6)
Arteriovenous nicking
Present	2 (12.5)	22 (26.5)	6 (26.1)	0.400	0.529	0.842
Absent	10 (83.3)	56 (71.8)	17 (73.9)

 Data are the number of subjects (%). The Bonferroni-corrected α was 0.0166; significant probabilities are in bold.

*  χ² test.

Figure 1.

 

The AVR did not differ significantly across the three study groups. The thick horizontal line within the box marks the median; the lower and the upper boundaries of the box indicate the 25th and the 75th percentiles, respectively. The whiskers below and above the box indicate the 2.5th and the 97.5th percentiles, respectively.

Table 4.

 

Data Description of Blood Pressure Levels and SLDF Parameters (Absolute Values) of the Study Population

Parameter	Group 1 Normotensive Control Group (n = 16)	Group 2 Hypertensive Group (n = 83)	Group 3 Cerebrovascular Event Group (n = 23)	P * (Group 1 vs. Group 2)	P * (Group 1 vs. Group 3)	P * (Group 2 vs. Group 3)
Daytime systolic BP, mm Hg	127 ± 7	145 ± 9	148 ± 21	<0.001	<0.001	0.824
Daytime diastolic BP, mm Hg	80 ± 3	94 ± 9	89 ± 13	<0.001	0.038	0.052
Daytime mean arterial pressure, mm Hg	95.3 ± 4.1	111.1 ± 8.7	108.5 ± 14	<0.001	0.002	0.338
Arterial outer diameter by SLDF, μm	110.62 ± 13.4	107 ± 10.9	110.72 ± 13.5	0.303	1.0	0.209
Lumen diameter by SLDF, μm	85 ± 10.9	80 ± 9.2	76.52 ± 6.45	0.061	0.010	0.112
Wall thickness of retinal arterioles, μm	12.81 ± 3.75	13.50 ± 3.88	17.1 ± 4.34	0.549	0.006	0.001

 Data are the mean ± SD. The Bonferroni-corrected α was 0.0166; significant probabilities are in bold.

*  Based on the Mann-Whitney U test

Figure 2.

 

WLRs between groups. A significant increase in WLR was observed in patients with acute cerebrovascular ischemia compared with each of both control groups. Data are expressed as described in Figure 1 .

Figure 3.

 

WCSA in the three study groups. WCSA was unchanged in the hypertensive control group compared with the normotensive subjects, but showed a pronounced increase in patients with cerebrovascular damage compared with both control groups, suggesting hypertrophy of retinal arteriolar wall. Data are expressed as described in Figure 1 .

Figure 4.

 

Differences in the IMT of the carotid artery between study groups. IMT of the carotid artery was highest in patients with cerebrovascular damage compared with both control groups, suggesting analogue macrovascular alterations of the vessel wall as those seen in the retinal microvessels. Data are expressed as described in Figure 1 .