Retina
Microstructural Alterations of Retinal Arterial Blood
Column along the Vessel Axis in Systemic Hypertension
Konstantin Kotliar,1 Edgar Nagel,2 Walthard Vilser,3 Seid-Fatima Seidova,4 and
Ines Lanzl 1,4
PURPOSE. Image analysis by the retinal vessel analyzer (RVA)
observes retinal vessels in their dynamic state online noninvasively along a chosen vessel segment. It has been found that
high-frequency diameter changes in the retinal artery blood
column along the vessel increase significantly in anamnestically
healthy volunteers with increasing age and in patients with
glaucoma during vascular dilation. This study was undertaken
to investigate whether longitudinal sections of the retinal artery blood column are altered in systemic hypertension.
METHODS. Retinal arteries of 15 untreated patients with essential arterial hypertension (age, 50.9 ⫾ 11.9 years) and of 15
age-matched anamnestically healthy volunteers were examined
by RVA. After baseline assessment, a monochromatic luminance flicker (530 – 600 nm; 12.5 Hz; 20 s) was applied to
evoke retinal vasodilation. Differences in amplitude and frequency of spatial artery blood column diameter change along
segments (longitudinal arterial profiles) of 1 mm in length were
measured and analyzed using Fourier transformation.
RESULTS. In the control group, average reduced power spectra
(ARPS) of longitudinal arterial profiles did not differ when
arteries changed from constriction to dilation. In the systemic
hypertension group, ARPS during constriction, baseline, and
restoration were identical and differed from ARPS during dilation (P ⬍ 0.05). Longitudinal arterial profiles in both groups
showed significant dissimilitude at baseline and restoration
(P ⬍ 0.05).
CONCLUSIONS. The retinal artery blood column demonstrates
microstructural alterations in systemic hypertension and is less
irregular along the vessel axis during vessel dilation. These
microstructural changes may be an indication of alterations in
vessel wall rigidity, vascular endothelial function, and smooth
muscle cells in this disease, leading to impaired perfusion and
regulation. (Invest Ophthalmol Vis Sci. 2010;51:2165–2172)
DOI:10.1167/iovs.09-3649
From the 1Department of Ophthalmology, Munich University of
Technology, Munich, Germany; the 2Nagel Eye Clinic, Rudolstadt,
Germany; the 3Department of Biomedical Engineering, Ilmenau University of Technology, Ilmenau, Germany; and the 4Chiemsee Augen
Tagesklinik, Prien, Germany.
Supported by a research grant from the Clinical Science Foundation, University Clinic, Munich University of Technology. The work
received the Adalbert-Buding-Scientific Prize at the Annual Conference
of the German Hypertension League, November 2007, Bochum, Germany.
Submitted for publication March 2, 2009; revised August 4 and 11,
2009; accepted August 11, 2009.
Disclosure: K. Kotliar, None; E. Nagel, None; W. Vilser, Imedos,
Ltd. (E), P; S.-F. Seidova, None; I. Lanzl, None
Corresponding author: Konstantin Kotliar, Department of Ophthalmology (Augenklinik rechts der Isar), Munich University of Technology,
Ismaninger Strasse, 22, 81675 Munich, Germany; [email protected].
E
arly diagnosis and prevention of vascular endothelial dysfunction, as well as structural atherosclerotic and hypertensive changes in human vessels is gaining increased importance,
especially in the industrialized countries due to sedentary life
styles, longer life expectancy, and increasing monetary restrictions within the health systems.
The eye possesses a few unique features in the body. One of
them is transparency of the optic media, which allows for
direct observation of the microvascular bed in the retina. The
fact that retinal vessels are similar in structure, function, and
regulation to cerebral vasculature underlines the importance of
ophthalmic assessment of the retina for the diagnosis of systemic hypertension. Ophthalmoscopy was introduced a long
time ago to detect severe changes in retinal vasculature and to
diagnose systemic hypertension in its advanced stages.1 However, there is evidence that this method does not allow early
diagnosis of the disease. One of the earliest and most classic
signs of hypertensive retinopathy is arteriolar narrowing,2 the
ophthalmoscopic detection of which is sometimes not specific
in earlier stages, and its clinical value is controversial.3 A
correlation of retinal arterial diameter or the arteriole-to-venule
ratio with systemic blood pressure was found in several epidemiologic studies.4,5 The semiautomatic method, called static
vessel analysis, was developed to assess this parameter from
retinal photographs and to use it for early detection of ocular
and systemic diseases including systemic hypertension.4 Nagel
et al.6 introduced the retinal vessel analyzer (RVA; Imedos,
Ltd., Jena, Germany) to measure temporal reactions of retinal
arterioles after stimulation with flickering light. These functional reactions were significantly different in patients with
moderate systemic hypertension when compared with healthy
volunteers.
From clinical and experimental studies, it is known that
arteries of both the microcirculation and the macrocirculation
can be afflicted by pathologic changes that include local microirregularities that can interfere with blood flow.1,7,8 A pattern of alternating constrictions and dilations along the vessel
has been shown in the microcirculation by in vivo microscopy
and in larger vessels by arteriography.9 This structure of longitudinal vessel profiles has been termed stationary arterial
waves10 or corrugated arteries11 by different investigators. Primarily, the phenomenon was reported in femoral arteries.12
Subsequently, it has been observed in carotid, radial, splenic,
superior mesenteric, and renal arteries.11
Another peculiar structure of longitudinal arterial profiles,
the so-called sausage-string appearance, occurs when severe
experimental hypertension is provoked in a microcirculation
system.7,8 The sausage-string pattern in microcirculation is
related to the development of vascular damage.9 Thus, it seems
that vessels in the elderly and in some diseases possess a less
regular longitudinal profile, for several possible reasons: instability of thin arteriolar walls, which become more rigid with
age7; partial endothelial damage of retinal arterioles13; and
Investigative Ophthalmology & Visual Science, April 2010, Vol. 51, No. 4
Copyright © Association for Research in Vision and Ophthalmology
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933453/ on 06/17/2017
2165
2166
Kotliar et al.
IOVS, April 2010, Vol. 51, No. 4
FIGURE 1. Definition of a longitudinal vessel profile (spatial diameter
changes of retinal vessel blood column along its longitudinal section).
partial degradation of smooth musculature of retinal arteriolar
walls.8,14
Previously, we found that high-frequency diameter changes
in the retinal arterial blood column along the vessel increase
significantly in anamnestically healthy volunteers with increasing age15 and in patients with glaucoma16 during vascular
dilation. These changes may signal endothelial damage.
In the present study, we used an RVA to detect possible
structural changes in the retinal arterial vasculature in moderate systemic hypertension. Such alterations in peripheral arterial vessels have been reported earlier in severe systemic hypertension.8,11
METHODS
Subjects
Fifteen patients with essential untreated arterial hypertension (8 men,
7 women, age 50.5 ⫾ 12.2 years) and 15 age- and sex-matched anamnestically healthy volunteers (8 men, 7 women; age 50.9 ⫾ 11.9 years)
were entered into the prospective clinical study. In the study, all
patients with arterial hypertension and an age-matched corresponding
number of the healthy volunteers reported by Nagel et al.6 were
included. Healthy subjects had no systemic disease and took no medication, with the exception of hormonal contraception in two cases.
Patients and volunteers with any ocular disease or affections were
excluded6: clouding of the optic media, visual acuity less than 0.5,
astigmatism more than 2.0 D, myopia more than 7.0 D, hyperopia more
than 4.0 D, history of eye surgery or injury in the examined eye,
wearing of contact lenses within the previous 24 hours, acute infection, known diabetes mellitus, pregnancy, and nursing.
At the start of the study, a clinical ophthalmic examination was
performed, including measurement of visual acuity, objective refraction, slit lamp microscopy, applanation tonometry, and funduscopy.
Then, the pupil was dilated with tropicamide 0.5% eye drops. The
preparation for the examination took at least 30 minutes.
Informed consent was obtained from all the subjects after explanation of the nature and possible consequences of the study. The study
design was reviewed and approved by the Ethics Committee of the
Thuringia State Medical Board and was performed in accordance with
the ethical standards laid down in the 1964 Declaration of Helsinki.
Measurements
Retinal vessels were assessed dynamically with the RVA. Details of this
device and its process of vessel diameter measurement have been
described elsewhere.14,17 Briefly, the device allows noninvasive online
assessment of the retinal arterial and venous red blood cell column
diameter depending on time and location along the vessel before,
during, and after provocations. For that purpose, the RVA consists of a
retinal camera (model 450 FF; Carl Zeiss Meditec, Oberkochen, Germany), a CCD camera for electronic online imaging, and a computer
for system control and analysis and recording of the obtained data.18
Functional Stimulation
During examination, retinal vessels were stimulated with flickering
light relying on the principles of neurovascular coupling.17,19 This
vascular stimulation is described in detail elsewhere.6 An optoelec-
tronic shutter was inserted in the retinal camera in place of an additional optical filter. The shutter interrupted the observation light with
a frequency of 12.5 Hz.6,14 Measurement of the baseline vessel diameter for 100 seconds in continuous light was followed by five cycles of
20-second flicker provocation and an 80-second observation. An arterial segment approximately 1 mm in length was evaluated in each eye.
Selection criteria for the segment were location within a circular area
of 2 disc diameters, no crossing or bifurcation in the measured segment, curvature of not more than 30°, a distance from neighboring
vessels of at least 1 vessel diameter and sufficient contrast to the
surrounding fundus. The vessel segment was scanned 25 times per
second in the measurement window under optimal conditions. Since
the image scale of each eye was unknown, the measured values were
expressed in relative units (RU). These units correspond to micrometers if the examination eye has the dimensions of the normal Gullstrand
eye. For each measurement time, an interval of 30 seconds before
every flicker provocation was considered as the baseline. The statistical
mean of arterial diameters during the baseline was calculated, to which
the subsequent vessel diameter response was normalized. Relative
vessel diameters are reported in percentage of baseline mean.
Pulse and Blood Pressure
Automatic blood pressure (BP) measurement was obtained with the
intensive care monitor (Cardiocup II; Datex Ohmeda, Louisville, CO).
Each measurement took approximately 33 seconds and was repeated
at 1-minute intervals. From those data the mean systemic arterial blood
pressure (MAP) was calculated as: MAP ⫽ BPdiastole ⫹ 1⁄3(BPsystole ⫺
BPdiastole) mm Hg.
Off-line Remeasurement of Data
To provide higher precision and accuracy, we took the measurements
of arterial reactions later off-line from videotape recordings with a
novel version of the RVA, the dynamic vessel analyzer (DVA). The
peculiarities of the DVA and its improvements on the RVA have been
published.20 The most important improvement in our study is that the
DVA scans each image pixel of a measured vessel segment at a rate of
25 times per second under optimal conditions.
Assessment of Spatial Arterial Blood Column
Diameter Change: The Longitudinal Vessel Profile
Temporal assessment of retinal arterial behavior in response to stimuli
is the most common feature of the device already published in several
studies.6,21,22 Changes in mean vessel segment diameter over time are
traced.
The data assessment with DVA allows observation of spatial
changes in vessel blood column diameter along a chosen vessel segment at chosen time intervals (Fig. 1). Through this feature, it is
possible to assess in vivo dynamic variations noninvasively in the vessel
longitudinal structure in humans at different stages of a vessel reaction
(Fig. 2). Differences in diameter along the vessel segment during a
defined time period can be measured. The method of data acquisition
for local vessel analysis with RVA has been explained in detail elsewhere.15,16 For each position along a measured vessel segment, the
mean of all measurements in this location during the chosen time
interval was calculated. We termed the result the longitudinal vessel
profile (Figs. 1, 2). Profiles obtained at different time intervals can be
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933453/ on 06/17/2017
IOVS, April 2010, Vol. 51, No. 4
Longitudinal Retinal Artery Profiles in Hypertension
2167
FIGURE 2. Assessment of longitudinal vessel profiles. Longitudinal vessel profiles during baseline (green),
constriction (red), dilation (blue and
light blue), and restoration (brown),
as displayed in the DVA system (right
bottom). Top right: an average longitudinal vessel profile during the
whole examination. Assessed temporal cycle with flicker stimulation is
marked within the whole temporal
vessel diameter course (left top). Bottom left: assessed arterial (red) and
venous (blue) segments are shown
within the fundus. 1 RU ⫽ 0.08
MU ⫽ 1 ␮m in a normal Gullstrand
eye; 1 MU equals 12.5 ␮m in a normal Gullstrand eye.
compared. External data transfer of spatial curves obtained by RVA is
performed in measurement units (MU); 1 MU equals 12.5 ␮m in a
normal Gullstrand eye.
Data Evaluation and Statistical Methods
Definition of Observation Time Intervals. The time
course of vessel diameter change in both groups was plotted. This time
response has been partially reported,6 and it was not the subject of the
present study. Average vessel diameter over time demonstrated a
diameter increase during provocation. A reactive vessel constriction
was observed after cessation of the stimulus, mainly in the control
group with an ensuing return to initial baseline values (Fig. 3).
From one complete arterial examination one temporal cycle (from
five assessed) was chosen for the evaluation of each subject as basis for
our spatial analysis. It consisted of 30 seconds of baseline, 20 seconds
FIGURE 3. Example of temporal arterial vessel reaction to flicker provocation and set of time intervals of local vessel reaction assessment.
of flicker stimulation, and 80 seconds of observation period (Fig. 3).
Five time intervals for investigation were defined as shown in Figure 3.
Longitudinal vessel profiles during the selected time segments were
evaluated for each subject at the defined time intervals (Figs. 2, 3). The
start of time segment IV was assigned individually. For each subject,
the individual time interval included the point of maximum constriction.
Mathematical and Statistical Analysis. To characterize longitudinal vessel profiles, we obtained their power spectra by Fast
Fourier transformation. Each power spectrum was reduced by dividing
each value in the frequency distribution by the whole area of the
power spectrum, as described in detail elsewhere.23 For each type of
spatial curves and for each group, average power spectra were derived
from these reduced individual power spectra by calculation of the
median value in the group for each point of frequency distribution, as
suggested by others for the analysis of electroencephalograms.23 The
following parameters of the averaged reduced power spectrum (ARPS)
were evaluated: average frequencies of peaks and average area under
the spectrum within the chosen frequency bands.
A program was created (MatLab 6.1; MathWorks Inc., Natick, MA)
to plot power spectra and evaluate the chosen parameters. Longitudinal vessel profile segments of ⬃800 ␮m in length (which corresponds
to 26 ⫽ 64 MU) were analyzed for each subject.
A template with corresponding macros was created (MS Excel
2000; Microsoft Corp., Redmond, WA) for each subject to filter, process, and analyze the numerical data from DVA. Since it was impossible
to prove the normal distribution of most measurement data (using the
Kolmogorov-Smirnoff test), the Mann-Whitney test and the Wilcoxon
test were used to assess the significance of differences in the evaluated
characteristics. Nonparametric data are expressed as the median (1st
quartile; 3rd quartile) and parametric data as the mean ⫾ SD. For
comparison of the five different phases of vessel reaction regarding one
parameter, necessary adjustment for multiple comparisons was made
by the Dunnett method,24 with a coefficient of 5. Because of the small
number of subjects, the nonparametric tests were applied on the level
of significance of P ⫽ 0.05 for each evaluated parameter, and nonparametric statistics were calculated (SPSS ver. 11.0 for Windows; SPSS
Inc., Chicago, IL, and Primer of Biostatistics, ver. 4.03 by Glantz24).
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933453/ on 06/17/2017
2168
Kotliar et al.
IOVS, April 2010, Vol. 51, No. 4
FIGURE 4. Typical longitudinal arterial profiles in the examined groups.
Two characteristic examples are represented for each group: 1 MU ⫽
12.5 ␮m in a normal Gullstrand eye.
RESULTS
The mean systemic arterial blood pressure amounted to
110.3 ⫾ 9.7 mm Hg in the hypertensive group and to 95.6 ⫾
8.3 mm Hg in the control group.6 No significant changes in
blood pressure occurred during the examination. The blood
pressure was significantly higher in the hypertensive group
(P ⬍ 0.01).
There were no statistically significant differences between
initial (baseline) average diameters of the measured arterial
segments in both groups (P ⫽ 0.52). This parameter amounted
to 119.6 (109.2; 137.1) RU in the hypertensive group and 112.6
(109.3; 127.0) RU in the control group.
Typical longitudinal arterial profiles are demonstrated in
Figure 4. The calculated ARPS of longitudinal arterial profiles
for the five phases of arterial reaction in both groups are
represented in Figure 5. The sequence of phases was chosen
according to the increase in vessel tonus from left to right.
In the control group ARPS did not differ when arteries
changed from constriction to dilation. No statistically significant differences in primary or secondary peak frequencies,
peak values, or average areas under ARPS within any frequency
band were found between the different phases of arterial
reaction in the control group (P ⬎ 0.2). This confirms our
results reported previously in healthy subjects.15 Typical examples in Figure 4, right, display this finding. The longitudinal
arterial profile shifted almost parallel during vessel reaction
without changing its configuration. ARPS at all phases possessed three peaks (Fig. 5). The primary one varied from 0.016
to 0.022 Hz at different phases. This range corresponds to one
oscillation in 45.5 to 62.5 MU or in 568 to 781 ␮m—a lowfrequency oscillation of a high magnitude. This characteristic
frequency is superimposed with other oscillations in Figure 4,
right. The secondary peak of the power spectrum varied from
0.035 to 0.042 Hz at different phases and corresponded to one
FIGURE 5. Average reduced power spectra of spatial longitudinal arterial profiles for systemic hypertension study at the five phases of vessel
reaction during flicker stimulation assessment. The sequence of phases was chosen according to the increase in arterial diameter from the left to
the right.
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933453/ on 06/17/2017
IOVS, April 2010, Vol. 51, No. 4
Longitudinal Retinal Artery Profiles in Hypertension
2169
FIGURE 6. Mathematical analysis on
ARPS of arterial profiles. Top left: in
the systemic hypertension group the
area under the ARPS within the black
frame differs significantly at baseline
and dilation. Different peaks correspond to dominating frequencies in
spatial longitudinal arterial profiles.
Bottom left: set of characteristic harmonic functions in the top left.
Right: mathematical reconstruction
of characteristic spatial longitudinal
arterial profile from ARPS in the top
left. Examples of representative longitudinal arterial profiles. Top right:
RVA-analyzed vessel segment; bottom right: retinal arterial segment of
2-mm length; 1 MU ⫽ 12.5 ␮m in a
normal Gullstrand eye.
oscillation in 23.8 to 28.6 MU (298 –357 ␮m)—a midfrequency
oscillation (Fig. 4, right). There also was a third small peak at
0.07 to 0.08 Hz at all phases, which corresponds to one
oscillation in 12.5 to 14.2 MU (156 –179 ␮m)—a high-frequency oscillation of a low magnitude (Fig. 4, right).
In the systemic hypertension group, ARPSs during constriction, baseline, and restoration were identical and differed from
both ARPSs during dilation (Fig. 5, bottom). This difference
between dilation and other phases is shown in Figure 4, bottom left. The average area under the reduced power spectra
within the frequency band of 0.03 to 0.06 Hz differs significantly in dilation 1 (0.226 [0.15; 0.292]) from constriction
(0.317 [0.221; 0.396]), baseline (0.371 [0.241; 0.425]), and
restoration (0.352 [0.243; 0.412]; P ⬍ 0.05; Fig. 5). This parameter differs significantly in dilation 2 (0.226 [0.15; 0.364])
compared with baseline (P ⬍ 0.05).
ARPS during constriction and baseline possessed three main
peaks: one primary and two secondary ones of almost the same
height. At restoration, two secondary peaks merged into one
(Fig. 5). The primary peak varied from 0.015 to 0.018 Hz at
different phases corresponding to one oscillation in 55.5 to
66.7 MU (694 – 833 ␮m). In Figure 4, left, the frequency is
superimposed on other oscillations. The secondary peak of the
power spectrum (0.038 – 0.042 Hz at different phases) corresponds to one oscillation in 23.8 to 26.3 MU (298 –328 ␮m)—a
midfrequency oscillation (Fig. 4, left). A third peak at 0.052 to
0.055 Hz at baseline and constriction corresponds to one
oscillation in 18.1 to 19.2 MU (227–240 ␮m)—a low-magnitude, midfrequency oscillation (Fig. 4, left). A merged secondary peak at restoration corresponds to 0.05 Hz (20 MU; 250
␮m, Fig. 5). A small additional high-frequency peak was found
at 0.092 Hz during baseline (10.9 MU, 136 ␮m; Fig. 4, left).
ARPS at dilation possessed two main peaks. The primary
one had the same frequency range as found during other the
phases of vessel reaction, described earlier, but it is significantly larger (P ⬍ 0.05). The secondary peak of ARPS, 0.035 to
0.042 Hz (23.8 –28.6 MU, 298 –357 ␮m; Fig. 4, left), was a
midfrequency oscillation similar to the control group. There
also was a small high-frequency peak during dilation 2.
The primary peak frequency at baseline longitudinal arterial
profiles differed significantly in the systemic hypertension
group (0.031 [0.016; 0.047] Hz) from the control group (0.016
[0.016; 0.031] Hz, P ⬍ 0.05). The average area under the ARPS
within the frequency band of 0.045 to 0.06 Hz differed significantly during restoration between the hypertension (0.172
[0.053; 0.195]) and the control group (0.05 [0.038; 0.069]; P ⬍
0.05). Both of these findings show significant dissimilitude
between both groups in longitudinal vessel profiles in low and
middle spatial frequency ranges.
To explain our findings in the frequency domain, we plotted the average representative longitudinal profiles for the
groups, which corresponds to ARPS in Figure 5. A mathematical reconstruction was performed. We took a set of characteristic oscillations, included in an average power spectrum and
shifted them in relation to each other to display the characteristic oscillations within the resultant curve. The sum of these
shifted oscillations formed a resultant longitudinal vessel profile, which conformed to the corresponding average power
spectrum and possessed characteristic peculiarities of longitudinal vessel profiles in the group. An example of such a reconstruction for baseline and dilation 1 in the systemic hypertension group is depicted in Figure 6.
DISCUSSION
Dynamic functional retinal vessel behavior was examined in a
group of patients with untreated systemic hypertension and an
age-matched control group by using the DVA including methods of mathematical signal analysis. In our present study, longitudinal retinal arterial profiles in untreated systemic hypertension differed from those in the age-matched healthy
volunteers at different stages of the dynamic vessel reaction
(Fig. 5). The profiles in systemic hypertension became less
irregular during vessel dilation than during other phases of the
vessel reaction to the stimulus of flicker light (Fig. 6, right). In
contrast to that finding, the microstructure of the longitudinal
arterial profiles in the healthy volunteers did not change during
the different phases of the functional vessel reaction (Fig. 5).
We distinguish between functional and structural alterations of longitudinal arterial profiles. The former are those
alterations that change during functional vessel reaction. The
latter are the alterations that are persistent and do not change
during functional vessel reaction. The longitudinal arterial profiles in the systemic hypertension group do not differ from
those of the control group at both dilation phases, although
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933453/ on 06/17/2017
2170
Kotliar et al.
they show differences compared with the control group. There
are also differences in the dilation phases of hypertensive
subjects compared with their baseline and restoration. We
therefore postulate that we are able to demonstrate functional
changes in the retinal arterial blood column in patients with
systemic hypertension. Previously, we demonstrated that
healthy subjects display age-related structural changes in their
retinal arterial blood column: its microstructure changes with
age. The change, however, is independent of the phase of
vessel reaction.15 Functional changes in retinal arterial behavior were also revealed in the patients with primary open-angle
glaucoma: Their microstructure of longitudinal arterial profiles
did not differ during baseline configuration from that of agematched healthy volunteers but it changed during vessel dilatory response.16 A discussion of the possible reasons for our
findings in the patients with systemic hypertension follows.
The lumen diameter of a vessel represents its primary functional characteristic. The diameter of a vessel determines its
resistance to blood flow. Vessel diameter is determined by its
functional and structural properties. The effectors for displaying active functional properties of a vessel are smooth muscle
cells, their number, their arrangement, and their state of contraction. The latter is massively influenced by the endothelial
cells and their mediators.
Fuhrmann (unpublished data, 2002) and Vilser et al.14 studied forced retinal arterial constriction during oxygen breathing.
They reported that retinal vessels of a patient with hypertension lose their constriction ability in contrast to those in a
healthy person. Some segments along a vessel can still react,
whereas other segments do not react any longer (Fig. 7).
These results, together with those in the present work, may
be explained by the effects of endothelial dysfunction in systemic hypertension, which presumably has focal distribution
along the vessel in primary stages of the disease and would lead
to nonuniform vessel reaction. Yamada25 described the irregularity of endothelial nuclei arrangement and their elongation
in retinal arteries of renal hypertensive rats.
We suppose irregularities in smooth muscle cell status and
segmental smooth muscle loss in vessels in systemic hypertension to be a more plausible explanation of the effects on
longitudinal vessel structure in systemic hypertension reported
in the present work and previously by Vilser et al.14 (Fig. 7,
top). This hypothesis is supported by light and electron microscopic studies. Kimura et al.26 reported that in the walls of
sclerotic blood vessels the smooth muscle cells have been
replaced by collagen fibers, proteoglycan filaments, and ruthenium red–positive materials. Such a pathologic effect had also
been described in the human diabetic retina (Rungger-Brändle
E, et al. IOVS 1997;38:ARVO Abstract 3563). During vessel
dilation in a patient with systemic hypertension, the unaffected
smooth muscles relax more than neighboring segments with
reduced smooth muscles. Consequently, the vessel wall becomes flatter during its dilation (Fig. 6, right). When transferring these considerations to vessel constriction, we can also
explain the findings of Vilser et al.14: Vessel segments with loss
of smooth muscles cannot constrict properly, but unaffected
segments can. This phenomenon can result in an uneven longitudinal profile of a constricted hypertensive vessel (e.g., Fig.
7, bottom).
The impaired normal segmental dilation behavior due to the
biomechanical alterations of the arterial wall status like vessel
narrowing or increase in vessel wall rigidity may represent an
alternative reason for our obtained results on smoothing of
longitudinal arterial profiles in systemic hypertension during
vessel dilation (Fig. 8). A completely relaxed or a completely
constricted normal retinal vessel would presumably represent
a tube with almost straight parallel inner walls. A normal
baseline longitudinal vessel configuration represents a config-
IOVS, April 2010, Vol. 51, No. 4
FIGURE 7. Top: Segmental loss of smooth muscle cells along a vessel
(Rungger-Brändle E, et al. IOVS 1997;38:ARVO Abstract 3563) (Modified with kind permission of Elisabeth Rungger-Brändle). Bottom: longitudinal arterial profiles (changes in vessel diameter along the vessel
segment) of a patient with systemic hypertension measured with RVA
before and after 100% oxygen breathing. Segmental vessel constriction. (Modified with permission from Vilser W, Nagel E, Lanzl I. Retinal
vessel analysis: new possibilities. Biomed Tech (Berlin). 2002;47(suppl
1):682– 685.) The top and bottom are adjusted to each other according
to the position along the vessel. The loss of smooth muscle cells is
assumed in unreactive arterial segments (grey arrows); 1 RU ⫽ 1 ␮m
in a normal Gullstrand eye.
uration between maximum vessel dilation and its maximum
constriction. Smooth muscles of the vessel wall are partially
relaxed, partially constricted at this stage, resulting in a wavy
vessel profile. A normal vessel has large reserves for reactions
in either direction. It never reaches its maximum dilation or
constriction during physiological blood flow regulation. Its
walls, when considering their shape in a longitudinal section,
shift up (dilation) or down (constriction) almost parallel during
vessel reaction (Fig. 8, top) in contrast to a diseased vessel with
exhausted dilation reserves because of vessel narrowing or
alterations in the vessel collagen structure. Such a vessel can
reach its maximum dilation ability during its response to flicker
stimulus. Consequently, the wavy baseline longitudinal vessel
profile becomes flatter (Fig. 8, bottom).
Using methods of computational fluid dynamics (CFD) simulation, we showed in a prior study that an irregular rough
structure of the internal vessel wall leads to increased resistance to flow (Kotliar KE, et al. IOVS 2006;47:ARVO E-Abstract
469): the rougher the inner wall of a retinal vessel, the higher
its resistance to blood flow. Consequently, since the arteries in
systemic hypertension become smoother during dilation, our
morphologic finding may represent a natural mechanism of
retinal blood flow regulation in systemic hypertension. Thus,
the retinal blood flow in systemic hypertension can be increased during vessel dilation, not only because of internal
diameter increase but also through the smoothening of internal
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933453/ on 06/17/2017
IOVS, April 2010, Vol. 51, No. 4
Longitudinal Retinal Artery Profiles in Hypertension
FIGURE 8. Vessel dilation shown in a longitudinal section. Normal and
reduced segmental dilation behavior. Dashed lines: dilation reserves.
arterial walls at the dilation. This sophisticated mechanism
would allow the blood flow to increase in dilated hypertensive
vessels to a certain extent, despite the vessel’s reduced dilatory
ability. A reduced reactive retinal arterial dilation in systemic
hypertension has been reported.6 Our CFD results show that a
smoothened, low-frequency, longitudinal arterial profile provides more blood flow in comparison to a more wavy midfrequency profile, even if the inflow vessel diameter remains
constant (Kotliar KE, et al. IOVS 2006;47:ARVO E-Abstract
469).
Characterization of microirregularities of the arterial inner
wall by mathematical methods including frequency analysis is
now widely performed in clinical medicine. Several groups
have analyzed intima media configuration in the carotid artery.27–29 However, those reported attempts have been limited
mostly to large vessels. Vascular characteristics of microcirculation so far are difficult to examine noninvasively in vivo.
Retinal vasculature resembles that of the central nervous system and is easily accessible by noninvasive optical methods.
Our findings in retinal arteries represent part of the central
microcirculation and complement the knowledge about microirregularities in large vessels.
Alstrom et al.7 and Jacobsen et al.8 reported that a uniform
increase in vessel wall tension induces a pattern characterized
by periodic constrictions and dilations along the vessel. Such
patterns have been observed in large and small blood vessels.
In macrocirculation, the phenomenon is called “corrugated
arteries” and has been described in detail elsewhere.10,11 We
want to emphasize that we could see periodic constrictions
and dilations along the vessel both in healthy volunteers and in
patients with systemic hypertension and were able to differentiate this pattern by using functional provocation.
In severe hypertension, a so-called sausage-string pattern
may occur under certain conditions during vessel constriction,
which is explained through the instability of the vessel wall
rather than by a mechanical failure of the vessel wall due to
high blood pressure or to standing pressure waves caused by
the heart beat.7,8 The phenomenon was shown in rats mostly
in large arterioles of 25 to 50 ␮m in diameter. The probability
of observing the phenomenon in larger or smaller vessels was
2171
low.8 The vessels in our study ranged from 90 to 210 ␮m.
According to Jacobsen et al.8, the length of a “sausage” can be
estimated by multiplication of the diameter to a coefficient of
6.4. In our case, it would range from 576 to 1344 ␮m and
would presumably occur during vessel constriction. The reasons that we did not observe such a pattern in our systemic
hypertension group are: large vessel diameters, insufficiency of
flicker-induced vasoconstriction, and no severe form of systemic hypertension in our patient group. However, we cannot
exclude mechanical instability of the vessel wall as a possible
reason for the main phenomenon observed in the present
study: although a relaxed dilated artery in systemic hypertension has a relatively smooth profile, it may switch to the
uneven form with periodic constrictions and dilations along
the vessel when vessel wall tension increases uniformly (Fig. 6,
right) because of the mentioned instability phenomenon.
It could be questioned whether the higher blood pressure
in the hypertensive group produces a structural alteration, per
se, in the arterioles by increased intraluminal pressure. We
think that it does not, since the blood pressure does not reach
very high levels that would necessitate a pressure-related passive distension of the vessel. Such a vessel would also not be
able to dilate after a stimulus. The vessels we examined did
dilate after flicker light application.
Before the conclusion, some considerations of the technical
limitations must be elucidated to enable better interpretation
of our reported results.
Frequencies in all power-spectrum charts were calculated
in reciprocal local measurement units (MU⫺1). The term frequency, measured in hertz, is commonly related to the temporal changes. For our purposes in the present work, we introduced the term spatial frequency, which shows that we are
dealing with spatial, not with temporal, curves. The measure of
spatial frequency is calculated in hertz. Like the temporal
meaning of hertz (one oscillation per second) the latter means
one oscillation per local measurement unit. For example, the
spatial frequency of 0.1 Hz corresponds to one whole oscillation of a longitudinal vessel profile in 10 MU or in 125 ␮m in
a Gullstrand eye. Thus, in the present work, the following
remains valid: spatial frequency ⫽ MU⫺1 ⫽ Hz. Consequently,
since analyzed spatial curves of longitudinal arterial profiles are
represented as a percentage of the baseline mean (Fig. 4), the
area under a reduced power spectrum and the area under ARPS
are dimensionless values: ARPS area ⫽ Hz 䡠 Hz⫺1 ⫽ 1.
Because of the definition of ARPS, statistical parameters
(median, quartiles) of peak or band frequencies calculated
from individual power spectra do not always coincide with the
peak frequencies seen in ARPS (e.g., Fig. 6, top left). During the
averaging of power spectra, individual primary peaks compensate for each other, building a peak that is not necessarily at the
site of the median peak frequency.
Although for each curve, one can build a reduced power
spectrum, there is an infinite set of curves corresponding to
the certain power spectrum chart. They all possess the frequency spectrum with different phases shifted to each other. It
is impossible to build an average representative profile of the
group from individual ARPS. A simple mean (median) of all
profiles in the group would contain a confusing superposition
of all oscillations in the different positions along individual
profiles. The attempt to perform Fast Fourier transformation of
all individual profiles to reduce them, to build an averaged
reduced Fourier transformation, and to perform finally the
inverse Fourier transformation would obviously lead to the
same confusing superposition. Therefore, we propose an alternative way to show average representative longitudinal profiles
for groups (Fig. 6).
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933453/ on 06/17/2017
2172
Kotliar et al.
CONCLUSIONS
Longitudinal sections of the retinal artery blood column do not
change for different phases of the vessel reaction in healthy
volunteers and undergo microstructural alterations in systemic
hypertension, although less irregularly during vessel dilation.
These microstructural changes in longitudinal profiles of retinal arteries in systemic hypertension may be an indication for
alterations in the vessel wall rigidity, vascular endothelium, and
smooth muscle cells in this disease, leading to impaired perfusion and regulation. Changes in vessel microstructure can alter
its hemodynamic parameters. Thus, our morphologic finding
may represent a natural mechanism that regulates retinal blood
flow in systemic hypertension.
References
1. Alm A. Ocular circulation. In: Hart WM, ed. Adler’s Physiology of
the Eye. St. Louis: Mosby; 1992:198 –227.
2. Tso MO, Jampol LM. Pathophysiology of hypertensive retinopathy.
Ophthalmology. 1982;89:1132–1145.
3. Hayreh SS. Duke-elder lecture: systemic arterial blood pressure and
the eye. Eye. 1996;10:5–28.
4. Hubbard LD, Brothers RJ, King WN, et al. Methods for evaluation
of retinal microvascular abnormalities associated with hypertension/sclerosis in the Atherosclerosis Risk in Communities Study.
Ophthalmology. 1999;106:2269 –2280.
5. Leung H, Wang JJ, Rochtchina E, et al. Relationship between age,
blood pressure, and retinal vessel diameters in an older population. Invest Ophthalmol Vis Sci. 2003;44:2900 –2904.
6. Nagel E, Vilser W, Lanzl I. Age, blood pressure, and vessel diameter
as factors influencing the arterial retinal flicker response. Invest
Ophthalmol Vis Sci. 2004;45:1486 –1492.
7. Alstrøm P, Equiluz VM, Colding-Jørgensen M, Gustafsson F, Holstein-Rathlou N-H. Instability and “sausage-string” appearance in
blood vessels during high blood pressure. Phys Rev Lett. 1999;82:
1995–1998.
8. Jacobsen JC, Beierholm U, Mikkelsen R, Gustafsson F, Alstrom P,
Holstein-Rathlou NH. “Sausage-string” appearance of arteries and
arterioles can be caused by an instability of the blood vessel wall.
Am J Physiol Regul Integr Comp Physiol. 2002;283:R1118 –
R1130.
9. Giese J. Acute hypertensive vascular disease, 2: studies on vascular
reaction patterns and permeability changes by means of vital
microscopy and colloidal tracer technique. Acta Pathol Microbiol
Scand. 1964;62:497–515.
10. New PF. Arterial stationary waves. Am J Roentgenol Radium Ther
Nucl Med. 1966;97:488 – 499.
11. Foster JH, Killen DA, Klatte E. Corrugated arteries: an enigma. Am
Surg. 1968;34:398 – 412.
IOVS, April 2010, Vol. 51, No. 4
12. Theander G. Arteriographic demonstration of stationary arterial
waves. Acta Radiol. 1960;53:417– 425.
13. Haefliger IO, Meyer P, Flammer J, Luscher TF. The vascular endothelium as a regulator of the ocular circulation: a new concept in
ophthalmology? Surv Ophthalmol. 1994;39:123–132.
14. Vilser W, Nagel E, Lanzl I. Retinal vessel analysis: new possibilities.
Biomed Tech (Berl). 2002;47(suppl 1):682– 685.
15. Kotliar KE, Mucke B, Vilser W, Schilling R, Lanzl IM. Effect of aging
on retinal artery blood column diameter measured along the vessel
axis. Invest Ophthalmol Vis Sci. 2008;49:2094 –2102.
16. Kotliar KE, Nagel E, Vilser W, Lanzl IM. Functional in vivo assessment of retinal artery microirregularities in glaucoma. Acta Ophthalmol Scand. 2008;86:424 – 433.
17. Polak K, Schmetterer L, Riva CE. Influence of flicker frequency on
flicker-induced changes of retinal vessel diameter. Invest Ophthalmol Vis Sci. 2002;43:2721–2726.
18. Seifert BU, Vilser W. Retinal vessel analyzer (RVA): design and
function. Biomed Tech (Berl). 2002;47(suppl 1):678 – 681.
19. Falsini B, Riva CE, Logean E. Neuro-vascular coupling in the human
inner retina. Ophthalmic Res. 2002;32(suppl 1):S133.
20. Gugleta K, Zawinka C, Rickenbacher I, et al. Analysis of retinal
vasodilation after flicker light stimulation in relation to vasospastic
propensity. Invest Ophthalmol Vis Sci. 2006;47:4034 – 4041.
21. Nagel E, Vilser W. Autoregulative behavior of retinal arteries and
veins during changes of perfusion pressure: a clinical study.
Graefes Arch Clin Exp Ophthalmol. 2004;242:13–17.
22. Polak K, Wimpissinger B, Berisha F, Georgopoulos M, Schmetterer
L. Effects of sildenafil on retinal blood flow and flicker-induced
retinal vasodilatation in healthy subjects. Invest Ophthalmol Vis
Sci. 2003;44:4872– 4876.
23. Bronzino JD, Siok CJ, Austin K, Austin-Lafrance RJ, Morgane PJ.
Spectral analysis of the electroencephalogram in the developing
rat. Brain Res. 1987;432:257–267.
24. Glantz SA. Primer of Biostatistics. New York: McGraw-Hill; 1999:
460.
25. Yamada A. Observation of retinal vasculature in experimental renal
hypertension rat using bis-benzimide (in Japanese). Nippon
Ganka Gakkai Zasshi. 1995;99:154 –160.
26. Kimura T, Mizota A, Fujimoto N, Tsuyama Y. Light and electron
microscopic studies on human retinal blood vessels of patients
with sclerosis and hypertension. Int Ophthalmol. 2005;26:151–
158.
27. Labropoulos N, Leon LR Jr, Brewster LP, et al. Are your arteries
older than your age? Eur J Vasc Endovasc Surg. 2005;30:588 –596.
28. Ishizu T, Ishimitsu T, Kamiya H, et al. The correlation of irregularities in carotid arterial intima-media thickness with coronary artery
disease. Heart Vessels. 2002;17:1– 6.
29. Schmidt-Trucksass A, Sandrock M, Cheng DC, et al. Quantitative
measurement of carotid intima-media roughness: effect of age and
manifest coronary artery disease. Atherosclerosis. 2003;166:57–
65.
Downloaded From: http://jov.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933453/ on 06/17/2017