Abstract
purpose. To measure the optical density of the crystalline lens and macular
pigments in a group of patients with diabetes mellitus and compare the
results with those in a group of control subjects.
methods. Color matches were performed using a Wright tristimulus
colorimeter. The reference wavelength used was 490 nm, desaturated with
650 nm. Lens optical density was measured by mixing spectral primaries
of wavelengths 420, 515, and 650 nm to match the reference. Wavelengths
420 and 515 nm were chosen, because they are absorbed equally by the
macular pigment. To measure macular pigment density, two color matches
were performed, one foveal and one 5° extrafoveal. The reference
stimulus was matched by mixing spectral primaries of 460, 530, and 650
nm. The ratio of the foveal to extrafoveal color match gives the
optical density of the macular pigment. Thirty-four diabetic patients
and 34 control subjects performed the lens density color match, and of
these, 26 diabetic patients and 30 control subjects performed the
macular pigment density color matches.
results. There is a significant increase in the optical density of the lens in
diabetes with age in comparison to the control subjects
(P < 0.001), with a duration dependence of 0.02
log units/year. The mean macular pigment density in the diabetic
patients was 0.13 ± 0.20 log units and in the control subjects
0.32 ± 0.24 log units (P = 0.0015). Patients
with grade 2 maculopathy had significantly lower pigment density than
those with no maculopathy (P = 0.016).
conclusions. The ocular media of diabetic persons are abnormal, with increased lens
and reduced macular pigment optical density. The relationship between
reduced macular pigment levels with increasing severity of maculopathy
may implicate oxidative stress as a causative
factor.
The absorption spectrum of the human crystalline
lens has been studied in vivo and after death.
1 2 3 4 5 There
is a degree of interindividual variation among people of the same age
and an increase in the relative absorption of short wavelengths with
age.
3 The optical changes that occur in the diabetic lens
before the onset of cataract have been studied previously, by using
lens autofluorescence
6 7 8 9 10 and psychophysical
thresholds.
11 Autofluorescence of the lens can be induced
by illumination with 420-nm light—the autofluorescence being at 530
nm. The degree of autofluorescence is calculated from digitized images
and the result expressed in arbitrary units of pixel gray scale
difference
8 or equivalent fluorescein
concentration.
9 These studies show a significant increase
in the autofluorescence of the diabetic lens compared with that of
age-matched normal lens. Autofluorescence is further increased in
diabetic persons with nephropathy compared with those
without.
7
There is only one published study in which a psychophysical method was
used to assess relative light loss in the ocular media in diabetic
persons.
11 Absolute thresholds to flashes of 420- and
550-nm light presented at 15° eccentricity were determined after 40
minutes of dark adaptation. These wavelengths were chosen to have an
equal absorption by rhodopsin. The difference between the thresholds
gives the relative light loss in the ocular media in units of optical
density. The normal group showed a small increase with age less than 60
years, increasing thereafter in agreement with the model developed by
Pokorny and Smith.
5 The diabetic group had an accelerated
increase in differential lens absorption that was parallel to the
increase in the normal group after the age of 60 years. The data were
analyzed to determine a separate effect of disease duration on the
diabetic lens, giving a gradient of 0.018 log units/year. Moreland used
the data to derive equations to estimate the age of lens-matched normal
persons.
12 These have been used to estimate the
contribution of the lens to the results of psychophysical
tests.
13 14
Lutein and zeaxanthin are concentrated in the photoreceptor axon layer
and the inner plexiform layer of the macula,
15 16 forming
a prereceptoral optical filter.
17 18 The absorption
spectra of these pigments have been well characterized, peaking at 460
nm.
18 Interobserver optical density of the macular pigment
can vary over 1 log unit, and this variation causes differences in
color perception.
15 19 20 Some studies have shown that
macular pigment density appears to be age-independent in normal
eyes
21 22 23 24 but a small age-dependent decline has been
reported recently.
25 26
Initially the macular pigment was thought to have an optical function
in the reduction of chromatic aberration.
27 More recently,
a potentially more important role as an antioxidant has been
described
28 —the carotenoids protecting the macula from
short wavelength light or against light-induced oxidative damage. This
has been of particular interest in age-related macular
degeneration,
29 where study has shown an association
between high levels/intake of antioxidants and reduced risk of
development of the disease.
30
In this study, color matching was used to investigate the differential
optical density of both the crystalline lens and macular pigments in a
group of patients with diabetes mellitus, and the results were compared
with those of a group of normal volunteers. The color of two adjacent
fields of view appears the same if the underlying cone excitations are
the same, even if the spectral distribution of the light in the
adjacent fields is different. Color matching is insensitive to the
absolute numbers of receptors stimulated,
31 unlike flicker
photometry,
32 but is sensitive to the absorption spectra
of the cones and of the ocular media. The cone excitation ratio to
monochromatic light is, however, independent of absorption in the
ocular media, because light incident on each cone type has been
affected equally by prereceptoral filtering. A recent study of the
effects of laser photocoagulation in diabetic persons showed complete
loss of photoreceptors in treated areas, but receptor morphology
appeared normal in nonphotocoagulated areas,
33 an
important prerequisite for the comparison of color match results to
assess the ocular media of diabetic persons and control subjects.
All patients gave written consent to be involved in the study.
The tenets of the Declaration of Helsinki were observed, and the study
had the approval of the Research and Ethics Committee of St. Mary’s
Hospital, Imperial College of Science, Technology and Medicine, United
Kingdom. Thirty-four diabetic persons and 34 control subjects
participated in the study to measure the lens optical density, and of
these, 26 diabetic persons and 30 control subjects performed further
color matches to assess the optical density of the macular pigments.
The age range of the patients was 25 to 72 years, and the duration of
disease ranged from 2 to 38 years (mean duration, 13.8 ± 10.3).
The diabetic group consisted of 24 persons with type II and 10 with I
diabetes. Twenty-two had had no laser treatment, four had undergone
panretinal photocoagulation, five had had macular photocoagulation, and
three had had both forms of treatment. Visual acuity was assessed using
the Early-Treatment Diabetic Retinopathy Study (ETDRS) log minimum
angle of resolution (logMAR) acuity chart; the mean acuity of the
patient group was 0.04 ± 0.08 log units. None of the diabetic
persons had cortical, posterior subcapsular, or significant nuclear
cataract. The level of retinopathy was assessed using the modified
Airlie House classification by dilated fundoscopy using a 90-D
biomicroscope lens and slit lamp and a 60-D lens to grade maculopathy.
The mean level of retinopathy was 3.0 ± 1.6 and the mean
maculopathy grade was 0.63 ± 0.80.
The 26 patients performing the macular pigment color match consisted of
19 who had type II diabetes, and the remaining 7 had type I diabetes.
Two of these patients had undergone laser treatment to the macula, two
had received panretinal photocoagulation only, and two had undergone
both macula laser and panretinal photocoagulation. The mean level of
retinopathy was 2.8 ± 1.6 and the mean maculopathy grade was
0.54 ± 0.76. The mean logMAR visual acuity was 0.03 ± 0.08.
The control group were selected on the basis of normal visual acuity
and color vision, the age range being 20 to 67 years. Previous studies
have shown an association of macular pigment density with iris
color,
34 smoking,
35 and
gender.
36 Comparison of these parameters across the two
groups showed no statistical differences (
P = 0.59 for
smoking,
P = 0.75 for iris color, and
P = 0.46 for sex distribution).
All color matches were performed using a tristimulus colorimeter
that produces a 1°, 20-minute bipartite field.
37 The
lower hemifield consisted of a reference wavelength and a desaturation
wavelength, and the upper hemifield contained two matching wavelengths
(different for the two experiments) and the desaturation wavelength.
When the relative radiance of the wavelengths in the upper hemifield
are adjusted such that the color and brightness of the two fields
appears the same, the cone excitations in the two retinal areas
illuminated are the same. For the reference wavelength, the ratio of
cone excitation is independent of prereceptoral filtering. The optical
density of the crystalline lens can be estimated using matching
wavelengths that are equally absorbed by the macular
pigments,
21 and the macular pigment density can be derived
from the ratio of two color matches, one foveal and the other
extrafoveal, made with matching wavelengths that are absorbed by the
pigment.
19 Color matching has been used to derive a
spectral absorption curve for the macular pigment
19 that
is in agreement with the spectral absorption characteristics found
using other methods.
18 38 Color matches performed using
different field sizes have been used to generate a spatial profile,
showing an approximately exponential decay of pigment density with
eccentricity.
39
Throughout the study, each color match was performed 10 times by each
observer to allow an estimate of the experimental error. Steady head
position was maintained using a dental bite bar, and each subject was
dark adapted for 10 minutes to allow cone adaptation.
The subjects underwent a period of training on the colorimeter,
performing the Rayleigh match, in which primaries of 530 and 650 nm
were matched to a reference wavelength of 590 nm. The color match to
assess lens optical density consisted of the reference wavelength 490
nm, desaturated with 650 nm, matched by mixing spectral primaries of
420, 515, and 650 nm. Wavelengths of 420 and 515 nm for the lens
density measure were chosen after a series of control experiments in
which different wavelengths were used on a small group of young control
subjects.
40 Comparison of postmortem transmission between
lens alone and whole media shows very little difference at the
wavelengths used.
3 The cornea absorbs mainly in the
ultraviolet,
41 the absorption being age
independent
42 ; the aqueous and vitreous can be considered
as having the absorption spectrum of water.
43 The value in
the color match gives a measure of light loss in the lens by both
absorption and scattering. Any forward-scattered light that uniformly
illuminates the bipartite fields cancels at the point of color match,
and scattering of light in the eye is largely wavelength independent in
normal eyes.
44 In this study we assumed that the effects
of absorption dominate the measurement. The relative optical density
for the lens is obtained by
\[d_{420}-d_{515}{=}\mathrm{log}\ E_{420}-\mathrm{log}\ E_{515}-\mathrm{log}\ K\]
where
E is the radiance of the matching wavelength,
K is a constant that depends on the short- (S) and medium-
(M) wavelength cone absorption spectra and the optical density
(
d ) is determined by
d = −log
T, where
T is the transmission of the ocular
media at the wavelengths used.
To obtain the macular pigment density, each subject performed two color
matches: The first was performed with foveal fixation, and the second
5° extrafoveally. The matching wavelengths used were 460, 530, and
650 nm. The reference stimulus (lower hemifield) was identical with
that used in the lens optical density experiment. Eccentric
presentation was achieved using a small (<0.1°) red fixation marker.
To avoid Troxler’s phenomenon, a rotating sector disc was used to
introduce a slow flicker in the bipartite field (5 Hz). Of the matching
wavelengths, 460-nm light is strongly absorbed by the macular pigment,
whereas the absorption of 530 nm is slight. The decrease in macular
pigment density with increasing eccentricity has been assessed, and by
5° eccentricity the pigment density is approximately 5% of its peak
value.
23 39 45 Another advantage of a 5° extrafoveal
match is that the light path is not significantly longer than for
foveal fixation, and consequently there is no confounding factor of
different path lengths through the remaining ocular media. The macular
pigment density is calculated from the ratio of the foveal to
extrafoveal color-match ratios as
\[d_{460}-d_{530}{=}\mathrm{log}\ \left(\frac{E_{460}}{E_{530}}\right)\ -\mathrm{log}\ \left(\frac{E^{{^\prime}}_{460}}{E^{{^\prime}}_{530}}\right)\]
where
E is the matching radiance for foveal fixation,
E′ is the matching radiance for the extrafoveal match, and
d is the optical density. Results are expressed as mean ± SD.
The mean age of the diabetic group was 48.1 ± 11.6 years and
of the control group, 36.7 ± 15.1 years (
P =
0.001). Given the known increase of lens optical density with
age
3 4 11 and the suggested decrease of macular pigment
optical density with age,
25 26 a comparison of means of
these measures between the two groups would not be appropriate.
Excluding control subjects less than 30 years of age provided an
age-similar control group of 16 subjects of mean age 47.1 ± 12.6
years (
P = 0.38 in comparison with the diabetic group).
For the purposes of establishing age-dependence relationships in the
data, all control subjects were used, but for comparison of means with
the diabetic persons, the age-similar control group was used.
The results for the Rayleigh color match showed no age dependence in
control subjects or diabetic persons and no statistically significant
difference between the diabetic persons in comparison with the
age-similar control group or the whole control group. The mean
color-match ratio for the control group was −0.26 ± 0.06 log
units and for the diabetic group, −0.24 ± 0.08 log units
(P = 0.17). This finding is important for the
interindividual comparison of results.
The optical density of the lens in the diabetic group was significantly
greater than that of the age-similar control group (
P < 0.001, Mann-Whitney test). The results for the lens optical density
are plotted as a function of age in
Figure 1a for the control group and in
Figure 1b for the diabetic patients. There
is a significant correlation with age in both groups, the rate of
increase being 0.0085 log units/year (
R = 0.81,
P < 0.0001) in the control subjects and 0.017 log
units/year (
R = 0.66,
P < 0.001) in the
diabetic patients.
Multivariate analysis of the data for the diabetic patients shows
significant correlation with subject age (P = 0.0012),
but not with duration of disease (P = 0.14), level of
retinopathy (P = 0.56), grade of maculopathy
(P = 0.13), or previous laser treatment
(P = 0.61). The diabetic group contained a small
subgroup of four patients of the same age (44 years). On the assumption
that these individuals have similar lens optical densities owing to
their age, linear regression gives a duration dependence of 0.021 log
units/year of diabetes (R = 0.86). The duration dependence
was also estimated for the 10 patients with type I diabetes in the
study, because the age of onset was known. The lens density at the
onset of diabetes was estimated from the line of best fit to the normal
subjects. The gradient of increase from this value to the present
measurement was calculated, and the gradient of the line of best fit to
the normal subjects was then subtracted, providing an estimate of the
duration effect. The mean value obtained for the 10 patients was
0.019 ± 0.009 log units/year.
An equation for the estimation of the age of a lens-matched normal
subject was also derived (see Appendix), which yields a linear
relationship of
\[A_{\mathrm{n}}{=}A_{\mathrm{d}}{+}2.364D\]
where
A n is the age of a
lens-matched normal subject, with the restriction
A n less than 60 years,
A d is the age of the diabetic subject,
and
D is duration of diabetes.
Six diabetic patients and three control subjects had measured macular
pigment densities of less than zero, although in all cases the
experimental error included zero. The negative values represent
sampling error around zero, and the macular pigment densities for these
subjects were taken as zero. Analysis of the data with these points as
zero, in comparison with analysis of the raw data obtained (including
the negative values), showed no effect on the overall results.
The mean macular pigment density in the diabetic group was 0.13 ±
0.20 log units and 0.36 ± 0.18 log units in the age-similar
control group (
P = 0.0015;
Figs. 1c 1d ). Linear
regression did not show a significant age dependence in either the
control group or the diabetic group (
R = −0.17,
P = 0.36 and
R = −0.16,
P =
0.43, respectively). In the absence of an age-dependence relationship
in the data, comparison of the mean of the entire control group and the
diabetic group can be performed, and this also showed a statistically
significant reduction in pigment density in the diabetic group
(
P = 0.0005). There was no correlation between pigment
density and duration of diabetes (
P = 0.20) or previous
laser treatment (
P = 0.37). Patients with more severe
grades of maculopathy had less macular pigment optical density
(Fig. 2) . Multiple nonparametric tests between the different grades of
maculopathy were performed with the probability required for
significance reduced to 0.017 (three tests) using the Bonferroni
method. At this level of significance the diabetic subjects with grade
2 maculopathy had a significantly reduced pigment density in comparison
with the diabetic patients with no maculopathy (
P =
0.016). A similar analysis was performed for the level of retinopathy
with the probability corrected for the number of tests performed, but
no significant relationships were found.
A plot of the measured lens optical density against the macular pigment
density is shown in
Figure 3 . The results of the lens optical density are independent of the macular
pigment density in both control and diabetic groups (
R = 0.25,
P = 0.17 and
R = 0.10,
P = 0.62 respectively).
The results of the Rayleigh match in this study showed no
difference between the diabetic patients and the control group, a
finding in agreement with Elsner et al.,
46 who found that
at low illuminance levels the Rayleigh match was normal in diabetes and
that the optical density of the M and L cone photopigments was normal.
At higher levels of illumination, the diabetic patients exhibited
bleaching abnormalities, a finding that was not investigated in this
study.
The experiments designed to measure the lens and macular pigment
densities relied on the underlying assumption that the cone absorption
spectra are the same in both the diabetic and control groups. If the
optical density of a cone photopigment changes, there is a
corresponding change in the absorption spectrum—the bandwidth
broadening for increasing optical density and narrowing for decreasing
optical density.
47 In a disease state leading to cone
dysfunction, it is possible that the photopigment density could be
reduced, because of absence of pigment production or of disc
replacement in the diseased receptor outer segment. This could lead to
a change in the color-match ratios that in turn would give an apparent
change in the optical density of the prereceptoral filters in the
diabetic patients. Although M-cone photopigment density is probably
normal in diabetes, there is evidence of S cone pathway
dysfunction.
48 49 Later studies suggest that this
localizes to the postreceptoral pathway
50 51 and the
dysfunction thus may not reflect an S-cone change. However, a model was
developed (see Appendix) to investigate the effect of cone dysfunction
on the color-match ratios and the derived media optical densities,
based on a consideration of the cone excitations and the effect of
reduced photopigment concentration and decreased outer segment length.
The results are shown in
Figure 4 which shows that, overall, the effect of changes in the cones on the
Rayleigh match, apparent lens density, and apparent macular pigment
density is small, even for a 2 log unit (100-fold) decrease in the
product of cone photopigment concentration and outer segment length.
The one-tailed 95% confidence intervals for Rayleigh match in the
control subjects and the diabetic patients are plotted in
Figure 4 (top), indicating that the spread of data in this experiment is well
within the bounds of the change predicted by the model (i.e., M and L
cone optical densities are not significantly affected by diabetes).
The estimated lens optical density would be decreased in the case of S
cone photopigment loss (
Fig. 4 , middle), which is opposite to the
effect seen in the results of the 420- and 515-nm color match.
The model was also used to investigate apparent changes in macular
pigment optical density that may occur due to cone dysfunction in
foveal and extrafoveal locations. A 2-log-unit reduction in S cone
photopigment and outer segment length product in the fovea, maintaining
normal S cones in the extrafoveal location, would result in an
underestimation of the macular pigment density by 0.05 log units (
Fig. 4 , bottom). This does not explain the reduction measured in the
diabetic patients. An apparent reduction in macular pigment density
caused by S-cone change would also be accompanied by an underestimation
of the optical density of the crystalline lens, as all the patients who
performed the macular pigment color matches had also performed the lens
density match. Our results for the optical density of the lens agree
closely with those obtained by Lutze and Bresnick.
11 The
two methods used in the present study to derive a diabetes
duration-dependence relationship gave values of 0.019 log units/year
and 0.021 log units/year, which agree with the result of 0.018 log
units/year derived by Lutze and Bresnick. Estimating the age of a
lens-matched normal subject using
equation 3 agrees well with the
equations from Moreland,
12 who combined the Lutze and
Bresnick data with the two-component model of Pokorny et
al.
5
The similarity of the results in the two studies is particularly
important, considering the different experimental methods used. Lutze
and Bresnick
11 measured dark-adapted absolute thresholds
to two wavelengths equally absorbed by rhodopsin. They conducted
further experiments to verify that rods were the receptors involved in
detection rather than cones. Our method uses a color threshold at low
photopic luminance, which relies on cones. The similarity of the
results obtained in the present work in comparison with those obtained
previously suggests that they represent a true ocular media change
rather than an apparent change caused by receptoral abnormality in the
foveal cones of the diabetic patients.
The results of autofluorescence studies have shown an increase in the
lens absorption in diabetes.
6 7 8 9 10 The origin of the
increased light loss in the short-wavelength end of the spectrum is not
entirely clear, but it has been suggested that it is due to an
accumulation of advanced glycosylated end products
(AGEs).
8 52 Autofluorescence studies have shown that
glycosylated collagen absorbs at 370 nm and emits at 440
nm,
53 and that autofluorescence occurs at these
wavelengths in “browned” lenses.
54 55 Browned lenses
also autofluoresce at other wavelengths,
56 with a
significant emission at 520 nm.
9 10 In diabetes, the
long-term exposure of the delicate lenticular environment to
hyperglycemia is likely to lead to an increased accumulation of AGEs,
with accompanying optical effects. AGEs have been implicated in the
pathogenesis of complications of diabetes
57 58 and
specifically in the formation of cataract.
59 60 61 62 Glycosylation of lens α-crystallin has been measured in excised
diabetic and normal lenses.
63 The diabetic lenses had a
threefold increase in α-crystallin glycosylation in comparison with
normal lenses, although there was no significant difference in the
degree of lens browning between the two groups in this study. The
authors suggest that differences in lens browning between diabetic
patients and control subjects may be due to glycosylation of proteins
other than crystallins. Studies of bovine lens have shown that
nonenzymatic glycosylation occurs both in the
crystallins
64 and the membrane proteins.
65 Further study may be able to identify the relative contribution to
overall lens browning resulting from the glycosylation of different
proteins.
Lutein and zeaxanthin are the only carotenoids present in the
lens
66 and are concentrated in the lens epithelium and
cortex.
67 A high dietary intake of carotenoids has been
linked with a reduced incidence of nuclear cataract
68 (although this study provided only weak support for the association),
and a reduced need for cataract extraction is seen in
women
69 and men
70 in the United States with
high carotenoid intake. The use of vitamin supplements containing
vitamin C and E for longer than 10 years may also lower the risk of
cataract,
71 suggesting a protective role of these
antioxidants. In our study there was no association between lens
optical density and macular pigment optical density for either the
control group or the diabetic patients
(Fig. 3) . The wavelengths used
to measure lens density are equally absorbed by lutein and zeaxanthin,
and consequently the result of this color match would be unaffected by
both macular and lenticular carotenoid concentrations.
Macular pigment optical density showed no dependence on age in our
study, a finding that is in agreement with some previous studies in
normal subjects,
21 22 23 although a small age-dependent
effect has also been reported.
25 26 The possibility of an
age-related decline in macular pigment density is not resolved at
present.
There are several mechanisms by which macular pigment levels could be
reduced in diabetes. First, there may be a genetic influence. There is
a wide variation in macular pigment density in the patients with no
maculopathy
(Fig. 2) , which makes a strong genetic influence on macular
pigment density in diabetic patients unlikely. Macular pigment density
has been measured in monozygotic twins,
72 with results
suggesting that pigment levels are not entirely genetically determined.
Second, the diabetic diet could be deficient in lutein and zeaxanthin
or absorption from the gut could be reduced. Granado et
al.
73 studied the serum levels of antioxidants in a group
of European insulin-dependent diabetic patients, first-degree
relatives, and control subjects. They found no significant difference
in serum levels of lutein and zeaxanthin between groups, although the
diabetic group had lower levels of retinol and higher levels ofβ
-carotene, α-carotene, and β-cryptoxanthin than did first-degree
relatives without diabetes. However, Ford et al.
74 found a
significant reduction in the serum levels of macular carotenoids in
patients with newly diagnosed and established diabetes patients in the
United States in comparison with normal subjects. These findings may
relate to different diets in the two study populations. Study of the
absorption of carotenoids in diabetic persons would help to resolve
this issue.
Finally, the pigment density could become low as a result of a reduced
rate of incorporation into retinal tissue or an increased rate of
removal from the retina. Thickening of basement membranes of retinal
capillaries in diabetes,
75 76 77 78 the increased affinity of
oxygen for glycosylated hemoglobin,
79 the existence of a
redox shift due to the effects of hyperglycemia on glycolysis and
sorbitol metabolism,
80 and the presence of abnormal
vasculature in the parafovea of diabetic persons
81 imply
that diabetic retinas are under continuous oxidative stress. An
analysis of retinal tissues from primate and human eyes for oxidation
products of lutein and zeaxanthin showed that, indeed, these pigments
appear to play a role as antioxidants.
28
Analysis of the data in this study has shown that the only factor with
a statistically significant correlation with lower levels of macular
pigment among the diabetic patients was grade of maculopathy. Because
the grade of maculopathy provides an indication of the severity of
microvascular disease in the macula, this may imply that lower pigment
levels are found in diabetic maculae that are under greater oxidative
stress. The macular pigment density in an individual is likely to
represent an equilibrium value of rate of incorporation into retinal
tissue combined with rate of removal from tissue, including conversion
to other compounds. A low macular pigment level could result from
either reduced incorporation into retinal tissue or from increased rate
of removal. It is possible that increasing antioxidant protection to
the diabetic retina may reduce the probability of development of
microvascular complications. In the San Luis Valley Diabetes study of
antioxidants in diabetes, the effect of dietary and supplement intakes
of vitamin C, vitamin E, and β-carotene on progression of diabetic
retinopathy was examined in patients with type II
diabetes.
82 There was no observed protective effect of
these nutrients against diabetic retinopathy, and indeed among those
patients not taking insulin, increased vitamin E intake was associated
with an increased risk for severity of retinopathy, as was increased
intake of β-carotene in patients taking insulin. This study
highlights the complexity of the relationship between antioxidants and
diabetic retinopathy, but the effect of dietary intake of lutein and
zeaxanthin on grade of maculopathy was not specifically assessed nor
were serum levels of any antioxidant measured.
Hammond et al.,
83 showed that macular pigment levels could
be modulated by diet in some but not all subjects. Of 11 participants,
8 showed increased serum levels and macular pigment density, 2 showed
an increase in serum level but not in macular pigment density, and 1
showed no response in either parameter. The increase in macular pigment
optical density persisted to the posttest time point (range, 1–6
months after cessation of the diet).
Our study showed that diabetic patients have increased differential
lens optical density and reduced levels of macular pigment. The
lenticular changes probably result from an accumulation of AGEs. Lens
density may also be affected by oxidative stress in diabetic persons,
although the color match used in the study cannot provide any
information regarding the concentration of carotenoids in the lens. The
results of a study of antioxidant levels in the diabetic lens,
including lutein and zeaxanthin, would be very interesting indeed.
Reduced macular pigment density may result from increased oxidative
stress in the diabetic macula. Further study is indicated to pinpoint
the cause more precisely. A controlled trial of dietary modification
with serial measurement of serum lutein and zeaxanthin and macular
pigment optical density would be of interest in patients with diabetes.
If pigment density does not increase after a dietary supplement, the
incorporation of the carotenoids into retinal tissues would be at
fault. Conversely, demonstration of an increase in pigment optical
density would indicate that an increased rate of elimination of pigment
from the tissues is the likely reason for the observed low value.
Color matching is dependent on the transmission coefficient of the
ocular media and also on the cone absorption spectra. The fraction of
light absorbed by the photoreceptors is determined by the
equation
47 \[f{=}1-\mathrm{e}^{\mathrm{-k}{\alpha}}\]
where
k is the product of photopigment concentration
and length of the outer segment and α is absorption spectrum of the
photopigment. This is known as the correction for self-screening. In a
pathologic state causing damage to the photoreceptors,
k is
likely to be reduced. Although receptor morphology appears to be normal
in nonphotocoagulated retina in diabetes,
33 the product
k could still be reduced.
When a color match has been achieved, a consideration of the S and M
cone excitation ratio and transmission through the ocular media leads
to the expression (equivalent to
equation 1 )
\[\frac{E_{1}}{E_{2}}{=}\ \frac{T_{2}}{T_{1}}\ \left(\frac{\frac{m_{2}s_{3}}{m_{3}}-s_{2}}{s_{1}-\ \frac{m_{1}s_{3}}{m_{3}}}\right)\]
where
E is the matching radiance,
T is the
media transmission, and
m and
s are the spectral
absorption coefficients, with subscripts 1 and 2 indicating the shorter
and longer matching wavelengths, respectively, and subscript 3 the
reference wavelength.
The color-match relationship is more accurately described by replacing
the cone absorption spectra with the more precise term after correction
for self-screening. This gives
\[\frac{E_{1}}{E_{2}}{=}\ \frac{T_{2}}{T_{1}}\ \left[\frac{(1-\mathrm{e}^{\mathrm{-k}_{\mathrm{m}}\mathrm{m}_{2}})(1-\mathrm{e}^{\mathrm{-k}_{\mathrm{s}}\mathrm{s}_{3}})/(1-\mathrm{e}^{\mathrm{-k}_{\mathrm{m}}\mathrm{m}_{3}})-(1-\mathrm{e}^{\mathrm{-k}_{\mathrm{s}}\mathrm{s}_{2}})}{(1-\mathrm{e}^{\mathrm{-k}_{\mathrm{s}}\mathrm{s}_{1}})-(1-\mathrm{e}^{\mathrm{-k}_{\mathrm{m}}\mathrm{m}_{1}})(1-\mathrm{e}^{\mathrm{-k}_{\mathrm{s}}\mathrm{s}_{3}})/(1-\mathrm{e}^{\mathrm{-k}_{\mathrm{m}}\mathrm{m}_{3}})}\right]\]
can be used as a model to investigate the effect of
cone dysfunction on apparent lesser density by altering the values of
the parameters
k s and
k m.
k for a healthy cone is
taken as 0.4,
84 and the Pokorny and Smith
85 fundamentals are used for the cone absorption spectra. A similar
equation considering the M and L cone absorptions can be used as a
model of the Rayleigh match.
To investigate the effect of cone dysfunction on the macular pigment
density, two equations are used, incorporating the macular pigment and
lens transmissions in a foveal model and solely the lens transmission
in the extrafoveal model. The log ratio of the matching radiances then
gives the measure of macular pigment density and altering the parameter k for changes in M or S cone self-screening at either or
both locations models the effect of differential cone dysfunction on
the apparent macular pigment density.
A diabetic person of age
A d and
duration of diabetes
D has a lens density
L d of
\[L_{\mathrm{d}}{=}m_{\mathrm{n}}A_{\mathrm{d}}{+}m_{\mathrm{d}}D{+}c\]
where
m n is the normal increase
in lens density per year,
m d is the
increase due to diabetes per year, and
c is lens density at
birth. A nondiabetic subject of age
A n has a lens
density of
L n determined by
\[L_{\mathrm{n}}{=}m_{\mathrm{n}}A_{\mathrm{n}}{+}c{^\prime}\]
Assuming there is no difference in the lens densities at birth
between those patients who are destined to become diabetic and those
who are not, the age of a normal person with a lens density matched to
a diabetic person is thus
\[A_{\mathrm{n}}{=}A_{\mathrm{d}}{+}\ \frac{m_{\mathrm{d}}}{m_{\mathrm{n}}}\ D\]
In this study
m d = 0.020 and
m n = 0.00846 log units/year, which
yields the result
\[A_{\mathrm{n}}{=}A_{\mathrm{d}}{+}2.364D\]
The authors thank The Wellcome Trust, London, United Kingdom, for
grant support and Nicholas Lee, Consultant Ophthalmologist, the Western
Eye Hospital, London, United, Kingdom, consultant in charge of the care
of the patients involved in the study.