The Diameters of the Human Retinal Branch Vessels Do Not Change in Darkness

Abstract

purpose. To examine whether the diameters of retinal branch vessels of the human eye change during dark and light adaptation.

methods. Images (S-VHS recordings) were obtained of the peripapillary region in 11 eyes of 11 healthy young adults (seven women, four men; mean age, 26.4 years). The images were made under a sequence of different illumination conditions (light, 30 minutes of darkness, light) with a scanning laser ophthalmoscope (SLO), using near-infrared illumination (785 nm). The recordings were then analyzed with a retinal vessel analyzer (RVA), and the caliber changes of one branch artery and one vein were measured in each eye.

results. For arteries, the changes of diameter under different illumination conditions showed no clear trend, and comparisons between the different time sections revealed no statistically significant changes (P = 0.933; repeated measures ANOVA). There was a slight dilation (average, 0.9%; range, −3.9% to +5.1%) in darkness, and a return to baseline (range, −2.9% to + 2.9%) on restoring normal illumination. Veins during darkness showed a small but fairly consistent constriction (average, 1.5%; range −5.4% to +3.9%; significant P = 0.05), again returning to baseline (range, −2.1% to +2.6%) in normal light.

conclusions. The small changes of retinal branch vessel diameters under different light conditions probably have little influence on the possible changes of retinal blood flow in healthy subjects.

Animal studies have shown that the glucose metabolism and oxygen consumption of the photoreceptors in the outer retina is significantly increased in darkness. ^{1 2 3} Most of this increased amount of oxygen is provided by the choroid. ⁴ However, 10% of the oxygen used by the photoreceptors in darkness was, in cats, found to be supplied by the retinal circulation. ⁴ The change of the retinal circulation is most probably an autoregulatory response to the lower partial oxygen tension in dark conditions. ^{5 6 7} The resultant changes of ocular blood circulation during dark and light adaptation have been investigated, both in animal ⁸ and in human studies, ^{9 10 11 12 13 14} focusing principally on the flow velocities. Several methods have been used to measure the response of blood flow velocities in the vessels of the eye to dark and light conditions. Doppler ultrasonography has been used to measure velocities in the ophthalmic and the central retinal artery ¹² and laser Doppler velocimetry to gauge flow velocity in the retinal branch vessels. ^{9 10 11 14} In addition, confocal laser Doppler flowmetry has been used to measure the velocity and number of red blood cells in the choroid and thus to assess blood flow in this region. ¹³  

Another important factor determining the blood supply, besides the flow velocity, is the blood vessel’s caliber. However, this is difficult to measure reliably in any of the vessels we have mentioned. The spatial resolution of a 10-MHz ultrasound B-scan is not fine enough to image and measure the diameter of small orbital vessels, although the blood flow velocity can nevertheless be determined by color Doppler ultrasonography. In some studies, the changes in diameter of the retinal branch vessels have been followed, using measurements made on fundus photographs. ^{9 10} However, the disadvantage of photographs is that they can show only the state of the vessels at a particular moment, whereas in reality, the diameters of the vessels change with the heart cycle and also with slower blood pressure changes (e.g., Meyer-waves). The resultant uncertainty could in principle be reduced by synchronizing the photography with the heart cycle, but this refinement is not mentioned in the studies cited herein. 

A validated technique now available, the newly developed retinal vessel analyzer (RVA; Imedos GmbH, Weimar, Germany), enables us to follow the changes of retinal branch vessel diameters more precisely. With this technique it is possible to measure retinal vessel diameters 25 times per second (on- or off-line), thus enabling errors due to the periodic changes to be eliminated. 

The purpose of our study was to examine the changes in the retinal branch vessel diameters during dark and light adaptation. To achieve this, we used near-infrared illumination to allow measurements under dark conditions. 

Subjects and Methods

Imaging Technique

We applied a “hybrid” technique, using the RVA to analyze recordings of images made with a scanning laser ophthalmoscope (SLO). Near-infrared laser light (785 nm wavelength, with laser beam power always kept under 100 μW at the cornea) was used to image the peripapillary region of the retina, with an SLO (Model 101; Rodenstock GmbH, Ottobrun, Germany) with the C3 type aperture. The images were recorded on S-VHS videotape, and the recorded images were subsequently analyzed with the RVA apparatus (Imedos GmbH), instead of using images from the RVA’s original fundus camera. Because this appeared to be a new combination of techniques, additional tests were performed to verify its validity, confirming that there were no significant differences in vessel diameter measurements between the two methods. These tests are outlined in the Discussion section and are described more fully in the Appendix. 

Procedures

The Regional Committee of Science and Research Ethics of Semmelweis University approved the study protocol, and the study conformed to the tenets of the Declaration of Helsinki. Informed consent was obtained from all subjects. 

Eleven eyes of 11 healthy young adults (seven women and four men, age range, 23–33 years; mean 26.4) were examined. The subjects were nonsmokers, were taking no medications, and had no history or sign of circulatory or ocular disease; their intraocular pressure and visual acuity were both normal, and they had clear ocular media. The subjects did not receive any medication. Mydriatics were not used, because they would have decreased the near visual acuity and consequently would have led to worse fixation, whereas in darkness the pupils were physiologically dilated anyway. The subject’s blood pressure was measured with automatic sphygmomanometry on the upper arm, with the subject in a sitting position, before and after the SLO recordings. 

The time sequence of the test procedure is shown in Figure 1 . The subject, initially in daylight conditions, was asked at 5 minutes before the start of measurements to look continuously at a 60-W white (opal) light bulb from a distance of 60 cm (approximately 700 lux at the cornea). In the first segment of the recording, daylight conditions were maintained in the room, and the eye was still illuminated with the lamp, to get a baseline recording under light-adapted conditions (period L1). The fixation target under light conditions was the fixation LED of the SLO. 

After this, the room was darkened, and the instrument display screens were either switched off or covered, as were all other sources of light. The only light remaining was a very faint red glimmer in the SLO, which was visible when using the near-infrared illumination. This glimmer of light (which could not be switched off) was in fact used for proper fixation, because the fixation LED (switched off) was visible as a dark spot over the faint red background. 

The recording was performed continuously during the light–dark transition, and continued during the start of the dark period (D1, D2, D3). It was then interrupted for a rest-period of 6 minutes, during which the subjects closed their eyes. After this, another recording was made (D4), followed by a longer pause. with recording restarted after 27 minutes of darkness (D5). At the end of the dark period, the room was rapidly lit up and the test eye again illuminated with the 60-W lamp. However, the lamp illumination was increased gradually over a period of 30 seconds, to avoid blinking and tearing by the subject, which would have significantly degraded the quality of the images obtained. Recording was again continuous during the transition and was extended for 9 minutes in light conditions (L2, L3, L4). 

For analysis, the recording was divided into nine 3-minute periods: the averaging of the large number of individual readings in each period enabled us to eliminate the effects of pulse waves and similar periodic changes. Sections of either the superior or the inferior temporal branch vessels—as close to the optic disc as possible, always within 1 disc diameter—were chosen for analysis, the choice depending on where the measurements could be performed better. The criteria of selection were that the vessel section be without junctions or crossings and that any other vessel within the region of interest (ROI) be present only if its diameter differed significantly from that of the examined vessel. The length of the examined vessel within the ROI was approximately half of the disc diameter. The RVA measured the diameters of the chosen vessel sections 1500 to 2000 times during a 3-minute period of recording, and the mean values of the measurements were used to compare the results in the different sections of the experimental sequence. 

The differences between the sections were analyzed with repeated-measures analysis of variance (ANOVA). 

Results

Figures 2 and 3 show the means ± SE of the diameter changes of the arteries and veins during dark and light adaptation. The overall change of the arteries was a dilation of less than 1% in darkness, which returned to baseline in light. This dilation was not statistically significant (P = 0.933 with repeated measures ANOVA). The diameter changes of the individual arteries were inconsistent, ranging from −3.9% to +5.1% during dark adaptation. The nature of the diameter changes (increase, no change, or decrease) for the different subjects are summarized in Table 1 . The actual thickness of the vessels was in the range of 100 to 150 and of 130 to 180 units for arteries and veins, respectively (the units would be micrometers in a schematic eye). 

The caliber changes for veins were more consistent than those of the arteries (Table 1) , ranging between −5.4% and +3.8% during the dark-adaptation period. The veins showed a mean constriction of at most 1.5% in darkness, which returned to baseline when the eye was illuminated again. This change was significant at P = 0.05 when comparing the D4 and D5 periods with the L1, L2, L3, and L4 periods. 

The blood pressure of the subjects was within the normal range (<130/>70 mm Hg, systolic/diastolic) and did not change significantly during the examinations. 

Discussion

There is little information available in the literature concerning changes in diameter of retinal branch vessels under light and dark adaptation, and because of the different study protocols, the results of the few published studies are not really comparable with each other. 

In the retinal arterioles of 16 eyes of cats, Hill and Houseman ⁸ found diameter changes after dark adaptation of between −4.2% and +8.3%, and after light adaptation of between −6.8% and +2.4%. The mean changes were +1% after dark adaptation and −1.5% after light adaptation, and in neither case were the changes statistically significant. Feke et al. ⁹ found arterial dilation of 2% to 3% in three eyes of three healthy subjects, and venous dilation of 6% to 8% in one subject after dark adaptation. Riva et al. ¹⁰ found venous dilation of 5% to 8% in dark-adapted eyes. In each of these three studies the vessel diameter measurements were made on individual fundus photographs, requiring visible light for adjustment or to take the photograph, or both. However, this unavoidable transition from dark-to-light conditions during fundus photography seems in itself to cause rapid changes in the retinal blood flow. ^{11 14}  

A major advantage of our examination technique is that the fundus imaging was made using near-infrared light, eliminating the effects of rapid dark-light changes that occur during normal photography. Further, the analysis of the recordings made with the RVA allowed us to achieve accurate results by averaging the periodic changes of vessel diameter (e.g., pulse waves) using 1500 to 2000 measurements made over a period of 3 minutes. 

Because there have been no previous reports of using a combination of SLO data with the RVA software in a clinical setting, additional experiments were performed with which we confirmed the reliability and the precision of this new combined technique. Details on these additional tests are given in the Appendix; herein, we summarize the findings. The reproducibility error in our supplementary test was very small. The average coefficient of variation (CV) between separate recording sessions (CVs, calculated as CVs = SD/mean) was 1.22% for arteries and 1.11% for veins, which is in fact rather better than the reported CVs of the RVA system with its own fundus camera (CVs = 1.8%, overall average including arteries and veins ¹⁵ ). Bräuer-Burchardt et al. ¹⁵ used two to three measurements for each vessel to gauge the CVs in their study, whereas, in our confirmation study, the CVs was derived for 10 consecutive sessions (arranged similarly to the nine sessions in our main study). This may, at least partially, explain the difference in CVs. Our CVs are also better than those published as initial results by Nagel et al. ¹⁶ in 1992. They used sequential SLO images and the software of the RVA and reported a CV of 2.4% for arteries and 3.4% for veins. In this case one reason for the difference may be our use of average values from a series of a large number of measurements obtained over a 3-minute period, instead of from a sequence of individual measurements. 

To compare the sensitivity of our method to that of the original RVA system, a small stress-induced vessel alteration (a constriction of some 3%) was measured in one subject, using the original RVA setup and our “hybrid” method in turn. The results showed no relevant difference between the methods, and were comparable to the results using the RVA system obtained by Blum et al., ¹⁷ who examined 40 subjects stressed in a similar way. 

The measurement algorithm used by the RVA was developed and optimized for the digitized images of its own fundus camera; therefore, using this to process the SLO recordings made through different optics may induce error in the measurements. However, any induced error is probably systematic and therefore is present for all the measurements. In our study, the absolute diameter of the vessels was irrelevant, because the intent was to assess the relative changes (in percent). Also, as discussed earlier, the reproducibility of the technique is as good as that of the RVA system itself, and the sensitivity of the two methods in measuring small changes is similar. The combined method (analyzing SLO recordings with the RVA) appears therefore to be an adequate analysis tool for the purposes of our experimental protocol. 

The faint red glimmer of the SLO, mentioned in the Methods section, may be expected to have had some small effect on the adaptation process of the red-sensitive cone population. However, because there are 125 million rods and altogether 6.5 million rods in the human retina, ¹⁸ the proportion of red-sensitive cones is small, only approximately 1.7% of all photoreceptors. The remaining photoreceptors (comprising approximately 3.3% of green- and blue-sensitive cones, and 95% of rods) are not sensitive to the 785 nm wavelength. The fact that 1.7% of the photoreceptors are not completely dark-adapted therefore has negligible impact on the metabolic requirement of the whole retina, and so we may neglect this effect for the purposes of this study. 

A separate question arises as to whether the near-infrared laser beam of the SLO may heat the retina and so give rise to changes in the diameter of the retinal vessels. However previous studies, using higher laser energy than in our case, have shown that there is in fact no significant heating of the retina, because the choroid exerts a cooling effect. ¹⁹ Therefore, we are reasonably confident that the laser beam used in our measurements had no significant influence of this sort. 

In our study the changes in diameter of the arteries of individual subjects ranged between −3.9% and +5.1% during dark adaptation, and returned to baseline in light conditions. These changes were not consistent between different subjects (Table 1) , but the grouped mean value in each time interval did not vary more than ±1% from baseline. The veins showed more consistent changes (Table 1) with a magnitude between −5.4% and +3.9% in darkness and, in the same way as for arteries, the values returned to baseline when the eyes were illuminated again. In the majority (9/11) of cases the veins showed constriction, and the overall change of their diameters was significant at P = 0.050 during the D4 and D5 dark-adapted periods. However, one should note that even in that case the mean value of venous diameter changes was a maximum of −1.5%, implying a reduction of blood flow volume of only approximately 3% (assuming constant flow velocity); and this small effect has doubtful clinical relevance. 

The major finding of our study is that the diameter of the retinal branch vessels was virtually constant during dark and light adaptation. Therefore, if the blood flow in retinal branch vessels changes under different light conditions, this change is essentially proportional only to the change of the blood flow velocity in the vessels, and the effect of any change of vessel diameter can probably be negated. 

Previous studies on ocular circulatory changes resulting from dark and light adaptation have predominantly focused on flow velocities in several vessels of the eye. Unfortunately, it is not easy to compare the results of these studies, because they used different methods to measure the flow velocity, and the measurements were made in different vessels. Havelius et al. ¹² used 5 MHz color Doppler imaging and found that in darkness the mean systolic flow velocity in the orbital part of the central retinal artery of 12 subjects increased by 25% to 32%, whereas the diastolic flow velocity increased by 53% to 68%. In 1983 Feke et al. ⁹ measured a 43% increase of flow velocity in the retinal branch arteries of three subjects after darkness. In the same year Riva et al. ¹⁰ published their results showing that the flow velocity in the retinal veins of three subjects had increased by 65% in dark-adapted eyes, which change could be suppressed by the breathing of 100% oxygen. However, both of these studies used laser Doppler flowmetry with a helium-neon laser (630 nm). Thus, they could not actually measure in darkness, only before and after dark adaptation. Later, Riva et al. ^{11 14} adapted their method and used near-infrared laser Doppler flowmetry to measure the flow velocities in retinal branch vessels during dark adaptation. Using this new method they did not find significantly increased flow velocity in the dark-adapted retinal vessels. In fact the velocities decreased slightly (by 2.7%) in darkness, but there was subsequently a very quick and marked transient increase of flow velocity when the eyes were illuminated again after the dark period. There is one study concentrating on the changes of blood flow in the choroid in darkness. Longo et al. ¹³ examined the subfoveal choroidal blood flow using a near-infrared confocal laser Doppler flowmeter, and found a statistically significant decrease in flow of 15% during dark adaptation. This was caused by a 2.8% decrease of flow velocity and an 11% decrease of volume, suggesting local vasoconstriction in the choroid. 

The relative uniformity of the vessel diameters found under different illumination conditions is most probably maintained by the vascular autoregulation in retinal vessels that regulates blood flow through the change or uniformity of vessel caliber. 

Changes in blood flow velocity in different vessels of the eye as a function of dark and light adaptation are not well understood as yet. Our results showed that the magnitude of the changes of retinal branch vessel diameters under different light conditions was very small in the healthy study population. 

 Supported by Grant T034483 from the Hungarian Scientific Research Fund.

 Submitted for publication August 28, 2002; revised February 26, and March 21, 2003; accepted March 21, 2003.

 Disclosure: G. Barcsay, None; A. Seres, None; J. Németh, 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.

 

Figure 1.

 

The time sequence of the measurements. After an initial 5 minutes of illumination without recording, each period was 3 minutes long. L1: baseline recording in light. Dark adaptation lasted for 30 minutes, with recordings made at the beginning (D1, D2, and D3), middle (D4), and end (D5). Then the eye was reilluminated and recordings continued in light conditions (L2, L3, and L4).

Figure 2.

 

Retinal branch arteries, change of diameter (in percent, mean ± 1 SE) during dark and light adaptation (P = 0.933 with repeated-measures ANOVA). L1, 3 minutes; D1 to D5: 30 minutes; L2 to L4: 9 minutes.

Figure 3.

 

Retinal branch veins, change of diameter (in percent, mean ± 1 SE) during dark and light adaptation (P = 0.01 with repeated-measures ANOVA; probabilities of the significant differences are shown). The x-axis is as defined in Figure 2 .

Table 1.

 

Vessel Caliber Changes

	Light to Dark					Dark to Light
		Increase	Same	Decrease	Increase	Same	Decrease
Artery diameter	5	3	3	5	4	2
Vein diameter	2	0	9	9	1	1

Supplementary Materials

PDF Version - 

Reproducibility and sensitivity of vessel diameter measurements using SLO video recording and RVA analysis combined 

Introduction 

Build HTML