Factors Determining Flicker-Induced Retinal Vasodilation in Healthy Subjects

Abstract

Purpose: The purpose of this study was to analyze factors determining retinal arterial and venous responses to stimulation with diffuse luminance flicker in healthy subjects.

Methods: We retrospectively analyzed results obtained in 374 healthy subjects who had previously participated in clinical studies in our department. A total of 153 subjects underwent a protocol in which flicker stimulation was delivered through the fundus camera at 8 Hz (protocol 1), separating measurement and stimulation light depending on the wavelength, and 221 subjects underwent a protocol in which diffuse luminance flicker was delivered at 12.5 Hz with high modulation depth (protocol 2). We investigated whether sex, systemic blood pressure, baseline vessel size, blood plasma concentration of fasting glucose and hematocrit, and serum concentration of cholesterol, triglycerides, creatinine and C-reactive protein influenced the retinal vascular response to flicker stimulation

Results: Flicker responses in arteries and veins were more pronounced in protocol 2 than in protocol 1 (P < 0.001, each). In both of the protocols the vascular response to stimulation with diffuse luminance flicker was larger in smaller vessels (P between 0.001 and 0.016). In protocol 2 the retinal arterial flicker response was negatively associated with cholesterol serum levels (P = 0.033); in protocol 1, only a tendency toward this effect was observed (P = 0.056).

Conclusions: The present analysis indicates that retinal arterial and venous responses to stimulation with diffuse luminance flicker depend on the way the stimulation is delivered through the fundus camera. In addition, the flicker response varied with vessel size, that is, the smaller the vessel width, the larger the flicker response. Finally, our data indicate that, even within the normal range, higher cholesterol serum levels are associated with lower hyperemic flicker responses.

Neurovascular coupling refers to the phenomenon in which local activity of neurons is associated with an increase in local blood flow and metabolism.¹ This appears to be required to fulfill the increased oxygen and glucose needs of neurons when they are active.² Originally established in the brain, the phenomenon also occurs in the retina.^2,3 When the retina is illuminated with diffuse luminance flicker, a hyperemic response can be observed in the retina as well as in the optic nerve head.^4–9 Some of the mechanisms underlying neurovascular coupling have been identified and involve direct vasodilator release from active neurons as well as neuron/astrocyte/blood vessel interactions.^1,3,10–12 

In recent years, accumulated evidence has shown that flicker-induced retinal vasodilation is altered in a variety of ocular and systemic diseases, including diabetes and diabetic retinopathy,^13–26 animal models of diabetes,^27,28 patients with macular edema secondary to retinal vein occlusion,²⁹ glaucoma,^30–33 obesity,³⁴ systemic hypertension,^17,35,36 hypercholesterolemia,³⁷ atherosclerosis,^38,39 and coronary artery disease.^39,40 

In the present study, we set out to analyze determinants of flicker-induced vasodilation in healthy subjects. This was done by analyzing data collected previously in studies in healthy, young volunteers. Tested parameters included age, sex, blood pressure and heart rate, baseline vessel diameter, refraction, and several factors measured in blood plasma and serum. 

Methods

Subjects

The present study analyzed data from previously published clinical studies. These studies used two different regimens for flicker stimulation. In some studies using flicker-induced vasodilation, we used a wavelength separation technique at 8 Hz for light stimulation.^8,13,41–48 In other studies, we used the commercially available approach, working at 12.5 Hz.^{16,17,23,49–51} Data from these studies were analyzed separately. All study protocols were approved by the Ethics Committee of the Medical University of Vienna and followed the guidelines set forth in the Declaration of Helsinki. All subjects signed written informed consent prior to inclusion and passed a screening examination. Only nonsmoking healthy subjects with normal systemic and ocular findings were included in this analysis. All subjects were free of systemic or topical drug intake within the previous 2 weeks of the study day. In placebo-controlled trials, only the study day under placebo conditions was used for this analysis. Some subjects contributed to more than one clinical study. In this case, only the first study was chosen for analysis. In addition, pregnancy or lactation and a patient who had epilepsy or a family history of epilepsy was excluded. Participants were asked to abstain from beverages containing alcohol or caffeine for 12 hours before the study. Blood plasma levels of fasting glucose and hematocrit as well as serum levels of cholesterol, triglycerides, creatinine, and C-reactive protein were analyzed using routine techniques. 

Retinal Vessel Analyzer

In the above-mentioned studies, flicker stimulation was used based on the commercially available Retinal Vessel Analyzer (RVA; Imedos GmbH, Jena, Germany). The RVA has been described previously and provides good reproducibility and sensitivity for measuring retinal vessel diameters and retinal vessel responses to diffuse luminance flicker.⁵² In all experiments, vessels were studied before any bifurcations between 1- and 2-disc diameters from the optic nerve head. Two types of flicker stimulation were used in our studies. In the first protocol (protocol 1), a custom-built device was used that produced stimulating light flashes at a frequency of 8 Hz. The modulation depth of the flicker was 100%. The flicker was centered on the macula with an illumination angle of approximately 30°. To spectrally separate the flicker light from the light used to illuminate the fundus, a wavelength separation technique was used by using a 550-nm low-pass cutoff filter. Hence, light with wavelengths below 550 nm was used for flicker stimulation, and light with a center wavelength of 577 nm and a bandwidth of 10 nm was used for fundus illumination. Measurements were made for 60 seconds at baseline (BL) and for 60 seconds during flicker (FL) stimulation. Data from the last 30 seconds of each of these periods were used for evaluation. In some studies, a slightly different regimen used three flicker periods with increasing length. From those studies, only the last flicker stimulation period with a duration of 64 seconds was included in the analysis. In the second protocol (protocol 2), retinal vessel diameters were measured with a commercially available solution for flicker simulation (Dynamic Vessel Analyzer [DVA]; Imedos GmbH). The DVA is essentially a modification of the initially commercialized RVA system and includes an option for continuously measuring vessel diameters before, during, and after stimulation with diffuse luminance flicker. Again BL measurements were performed for 1 minute. For the second minute, light with a frequency of 12.5 Hz was used for stimulation by square wave pattern modulation of the fundus camera illumination at a contrast ratio of 25:1. Flicker is generated with a shutter that interrupts the observation illumination to the fundus. Because the readout frequency of the camera is 25 Hz, the stimulation frequency is 12.5 Hz. Baseline values of vessel diameters were calculated as an average of the last 20 seconds before start of the flicker stimulation. Values during flicker stimulation were calculated as the average of the last 20 seconds of the stimulation period. Flicker responses in retinal vessel diameters were expressed as percent of change over baseline values [(FL − BL) × 100/BL]. Both the retinal arteries (d_art) and the retinal veins (d_vein) were measured, and responses to flicker light stimulation were calculated (FR_art, FR_vein, respectively). 

Data Analysis

Flicker responses were calculated for the two protocols separately and divided into two groups. The group with the higher response was used as reference. In the present multiple linear regression model, the effect of the following factors on flicker responses were tested: sex, systemic blood pressure, vessel size, blood plasma concentration of fasting glucose and blood serum concentrations of cholesterol, triglycerides, creatinine, C-reactive protein, and hematocrit. In the model, data were adjusted for age. In addition, linear correlation analysis was performed between baseline vessel diameters and flicker responses. Data obtained in protocol 1 was compared with that from protocol 2 by using unpaired t-tests. For all calculations, a P value of <0.05 was considered the level of significance. All statistics were analyzed using SPSS version 22 software (SPSS; IBM, Armonk, NY, USA). 

Results

Protocol 1

A total of 153 subjects were studied according to protocol 1. The characteristics of this group are provided in Table 1. The FR_vein was less pronounced than FR_art (P = 0.007). Fasting glucose (n = 136: 5.0 ± 0.5 mmol/L), cholesterol (n = 146: 6.2 ± 1.0 mmol/L), triglycerides (n = 145: 2.2 ± 0.5 mmol/L), creatinine (n = 143: 101 ± 23 μmol/L), C-reactive protein plasma levels (n = 143: 3.4 ± 1.0 mg/L), and hematocrit (n = 137: 0.404 ± 0.050) were not available in all subjects. Subjects with higher flicker responses were more likely to have smaller vessel diameters. This effect was significant for both FR_art and FR_vein (Table 2). Flicker responses were not dependent on the other studied variables, although the association between FR_art and cholesterol serum levels was borderline significant (P = 0.056). Linear correlation analysis between baseline vessel diameters and flicker-induced vasodilation is presented in Figure 1. The association was highly significant for both arteries and veins (P < 0.001 each). 

Table 1

 

Characteristics of Healthy Subject Studied According to Protocol 1

 

Table 2

 

Relationships Among Flicker-Induced Vasodilation, Retinal Vascular Caliber, and Cholesterol Plasma Levels in Healthy Subject Studied According to Protocol 1

 

Figure 1

 

Correlation between retinal vessel diameter and flicker-induced vasodilation for healthy subjects studied according to protocol 1. Data are shown separately for arteries (A) and veins (B).

Protocol 2

A total of 221 subjects were studied according to protocol 2. Table 3 presents the demographics and retinal data obtained in this group of subjects. Compared to the data of subjects studied according to protocol 1, the average age of the participants was slightly lower (P = 0.021). In addition, flicker responses were higher in protocol 2 than in protocol 1. This effect was highly significant for FR_art and for FR_vein (P < 0.001, each). Protocol 2 subjects' fasting glucose (n = 202: 5.0 ± 0.5 mmol/L), cholesterol (n = 199: 6.2 ± 1.0 mmol/L), triglycerides (n = 197: 2.2 ± 0.5 mmol/L), creatinine (n = 197: 101 ± 23 μmol/L), C-reactive protein (n = 201, 3.3 ± 0.9 mg/L) and hematocrit (n = 189: 0.404 ± 0.050) were not different from those of protocol 1 subjects. Again, results of laboratory analysis were not available in all participating subjects. As in protocol 1, flicker responses were dependent on vessel diameters, with higher responses seen in smaller vessels (Table 4). Again, this effect was significant for arteries and for veins. In addition, we found a significant effect of cholesterol serum levels on FR_art (P = 0.033). A tendency was seen for FR_vein, but this effect did not reach the level of significance (P = 0.062) (Table 4). A highly significant association was observed between baseline vessel diameter and flicker-induced vasodilation (P < 0.001, each) (Fig. 2). 

Table 3

 

Characteristics of Healthy Subject Studied According to Protocol 2

 

Table 4

 

Relationships Among Flicker-Induced Vasodilation, Retinal Vascular Caliber, and Cholesterol Plasma Levels in Healthy Subjects Studied According to Protocol 2

 

Figure 2

 

Correlation between retinal vessel diameter and flicker-induced vasodilation for healthy subjects studied according to protocol 2. Data are shown separately for arteries (A) and veins (B).

Discussion

The present study has revealed two major results. On the one hand, vessel diameter responses to stimulation with diffuse flicker stimulation is dependent on the protocol used for light delivery; and on the other hand, the degree of retinal vessel dilation is dependent on retinal vessel size. In addition, the present analysis provides evidence that cholesterol serum levels determine flicker-induced vasodilation in healthy subjects. 

The two protocols used for the measurements of retinal vasodilation differ in several ways. Protocol 1 uses a frequency of 8 Hz, whereas protocol 2 uses a frequency of 12.5 Hz. In humans only one study which evaluated the effect of flicker frequency on the retinal vessel response to diffuse luminance is available.⁶ In this study, no differences in flicker response were found between stimulation frequencies of 4 to 40 Hz. More information about the frequency dependence of flicker response is available for optic nerve head blood flow. In cats, the peak flicker response was between 5 and 20 Hz when retinal illumination levels were in the mesopic range and shifted to higher frequencies at higher retinal illumination levels.⁴ In monkeys, the maximum of this response was reached at approximately 15 Hz,² whereas the response in humans was reached at a maximum of between 10 and 20 Hz.⁵³ It is therefore unlikely that flicker frequency plays an important role in the differences in flicker response as seen with the two protocols. Important differences, however, do exist between the two types of flicker stimulations in terms of modulation depth and chromaticity. In protocol 1, the light used for flicker stimulation and the light used for measurement of vessel diameter are separated by wavelength. The former protocol has an irradiance of approximately 300 μWcm⁻² at wavelengths below 550 nm, whereas the latter has an irradiance of approximately 260 μWcm⁻² at wavelengths between 567 and 587 nm. This results in a stimulus with relatively low modulation depth, which however, is clearly perceived by the subject under study also due to change in chromaticity. In protocol 2, the irradiance is approximately 200 μWcm⁻² at wavelengths between 530 and 600 nm. As the measurement light is used for flicker stimulation, modulation depth is greater than 25:1 with no chromaticity. Data obtained in healthy subjects using laser Doppler flowmetry to measure optic nerve head blood flow indicated that the modulation depth may play a key role in the hyperemic response to stimulation with diffuse luminance flicker showing almost a linear dependence.⁵³ With regard to chromaticity, no data are available in humans, but experiments with cats indicate that the wavelength dependence closely follows the scotopic luminous sensitivity curve of the retina.⁵⁴ A thorough investigation of the effects of variations in the characteristics of diffuse luminance flicker on the hyperemic response in the human retina with regard to field of stimulation, flicker frequency, modulation depth, retinal irradiance, and chromaticity is lacking, however. 

In the present analysis, the diameter of response to stimulation with diffuse luminance flicker was larger in vessels with smaller diameters. This was previously observed in patients with diabetes only¹⁹ and has been interpreted as a sign of endothelial dysfunction.^22,55 The conclusion is based mainly on data indicating that the hyperemic response to flicker stimulation is reduced by nonspecific NO synthase inhibition.^42,56,57 In the present study, we observed that the degree of vasodilation depended on the absolute size of the vessel in healthy subjects. This is compatible with results indicating that the increase in retinal blood flow during stimulation with diffuse luminance flicker is as high as 40% to 60%,⁸ whereas the increase in vessel diameters in larger retinal vessels is only on the order of 2% to 7%. It clearly indicates that the main decrease in retinal vascular resistance occurs in the smaller retinal resistance vessels that cannot be measured using fundus camera-based technology. In addition, future studies using the DVA should account for the vessel size dependence of flicker-induced vasodilation particularly when data in healthy subjects are compared with those in patients. Although the present study was performed in healthy subjects, we cannot exclude the fact that smaller vessel were contracted and therefore under different vascular tone than larger vessels. Hence, these vessels may have a larger capacity for vasodilation during any type of stimulation. 

It has previously been shown that patients with hypercholesterolemia have reduced retinal vasodilator responses to flicker stimulation.³⁷ Moreover, single low-density lipoprotein apheresis, which decrease triglyceride and cholesterol fraction by 21% to 74%, improved retinal venous vasodilation in response to flicker stimulation. The present study extends these findings by indicating that the retinal vascular flicker response is dependent on cholesterol serum level even in healthy subjects with values in the normal range. This effect only reached the level of significance in participants who were studied according to protocol 2, most likely because of the higher number of subjects in that subgroup. The association between plasma cholesterol level and diabetic retinopathy is controversial. Whereas increased cholesterol plasma levels were linked to diabetic retinopathy in the Hoorn study,⁵⁸ a recent multicenter study did not confirm those results.⁵⁹ In the present study, flicker-induced retinal vasodilation was dependent on serum cholesterol levels potentially due to atherosclerosis. This idea is supported by recent study indicating an age-related decline of flicker-induced vasodilation in healthy subjects.⁶⁰ 

The present study has a number of limitations. Most importantly, it was a retrospective analysis of a population that participated in clinical studies at our department previously. As such, the groups participating in the two protocols were not matched for age or for sex. With regard to age, the sample was far from being normally distributed, because some of the subjects served as controls for case-controlled studies, and the age differed between subjects participating in the two protocols. Hence, we decided not to include age in our analysis but rather to correct the data for a potential influence of age. As such, the present study does not represent a random sample of a healthy population in Austria but was selected based on the specific inclusion/exclusion criteria for each study. In some subjects who were studied according to protocol 1, three flicker periods were used, and only the last period, lasting 64 seconds, was used for analysis. As such, we cannot entirely exclude the possibility that the first two periods modified flicker-induced vasodilation in the third period. In addition, not all parameters from blood samples were available for all subjects. Because the number of missing data, however, was small, we do not think this represents a major problem. In addition, we did not measure low-density and high-density cholesterol serum levels separately and therefore cannot answer whether either of the two separately influence retinal flicker responses. Evidence for larger flicker-induced vasodilation in retinal vessels with smaller diameters was in the present study gained from data in one vessel per subject only. We did not perform measurements in smaller vessels after bifurcations to proof more pronounced vasodilation along the vascular tree when vessels become smaller in diameter. Indeed a previous study indicates that flicker-induced vasodilation becomes larger when more peripheral vessels are studied.⁶¹ 

In conclusion, the present data indicate that the retinal hyperemic flicker response is dependent on the size of the vessel in healthy subjects. The smaller the caliber of retinal arteries and veins, the bigger is the increase in diameter during flicker stimulation. Another determinant of retinal flicker response is cholesterol serum level. Even in a healthy population, higher cholesterol serum levels are associated with lower flicker-induced vasodilation. Finally, the present analysis indicates that the response of retinal vessels to flicker stimulation is dependent on the type of stimulation. Use of a 12.5-Hz diffuse luminance flicker with high modulation depth produces larger flicker responses than 8-Hz flicker in which the light used for measurement and the light used for stimulation are separated by wavelength. 

Acknowledgments

Supported by Austrian Science Foundation (Fonds zur Förderung der Wissenschaftlichen Forschung) Grant KLIF250. 

Disclosure: M. Sharifizad, None; K.J. Witkowska, None; G.C. Aschinger, None; S. Sapeta, None; A. Rauch, None; D. Schmidl, None; R.M. Werkmeister, None; G. Garhöfer, None; L. Schmetterer, None