| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Diabetes (F.M.E.E., M.W.J.S., B.M.F.), Royal Infirmary of Edinburgh, Edinburgh EH3 9YW and Department of Psychology (I.J.D.), University of Edinburgh, Edinburgh, EH8 9JZ Scotland United Kingdom
 |
Abstract |
|---|
 Top Abstract I. IntroductionII. Electroretinography (ERG)III. P100 Latency StudiesIV. P300 StudiesV. Color VisionVI. Contrast SensitivityVII. ConclusionsReferences |
|---|
|
I. Introduction |
|---|
|
TopAbstractI. IntroductionII. Electroretinography (ERG)III. P100 Latency StudiesIV. P300 StudiesV. Color VisionVI. Contrast SensitivityVII. ConclusionsReferences |
|---|
Diabetic retinopathy is usually considered to be a disease of retinal
blood vessels but is rarely thought of, in a wider sense, as a
neurosensory disorder (3). Although abnormalities within the peripheral
nervous system are well documented in diabetes, changes within the
central nervous system, and particularly their relationship to visual
function, have received much less attention. The concept of ‘diabetic
encephalopathy’ was introduced in a case report in 1950 by De Jong
(4), who observed diffuse histological abnormalities throughout the
central nervous system. The significance of these changes proved
difficult to investigate as, for many years, electroencephalography
(EEG) was the only technique available to study the
electrophysiological activity of the brain; however, the information
provided by this method is limited, particularly in the assessment
of deeper brain structures. In the last two decades, the advent of
newer neurophysiological techniques to assess retinal and cerebral
function, such as electroretinography and the measurement of brain
electrical-evoked potentials, has increased our understanding of normal
visual function and the possible effects that diabetes may exert (Table 1
). In addition, the development of
neuroimaging techniques, including magnetic resonance imaging, has
provided evidence for structural changes in the brain associated with
diabetes, suggesting that the central nervous system is affected as one
of the long-term complications of diabetes (5, 6). If one adds
psychophysical tests of visual function, such as contrast sensitivity
and hue discrimination, to the above techniques (Table 1) it becomes
clear that the examination of visual function provides a case study in
integrative neuroscience. In this spirit, the present report surveys
aspects of visual function in diabetes and is designed to provide an
overview of the subject for the generalist with an interest in diabetes
and its complications.
|
|
II. Electroretinography (ERG) |
|---|
|
TopAbstractI. IntroductionII. Electroretinography (ERG)III. P100 Latency StudiesIV. P300 StudiesV. Color VisionVI. Contrast SensitivityVII. ConclusionsReferences |
|---|
). Of these, the most commonly reported
are ‘a’ waves (a measure of photoreceptor activity), ‘b’ waves
(corresponding to the inner nuclear layer), and the oscillatory
potential (OP) (which reflects changes in the inner retina including
the retinal circulation). The pattern ERG (PERG) is the averaged
retinal field potential evoked by stimulating the retina with a
patterned stimulus such as the reversing checkerboard and is a focal
response reflecting ganglion cell activity specifically in the macular
area.
|
B. ERG and diabetes
A number of studies have assessed either flash ERG, PERG, or both
in patients with diabetes (Table 2). Good
evidence exists that early ERG changes occur in diabetes before the
development of retinopathy. Of the studies assessing only flash ERG,
the majority report abnormalities, mainly reduced OP amplitude, in
diabetic subjects with no evidence of retinopathy compared with
nondiabetic controls (13, 14, 15, 16, 17, 18, 19). Fewer studies have assessed PERG
(20, 21, 22, 23, 24, 25, 26), and the results are less conclusive, with some groups
reporting abnormalities in the diabetic subjects (20, 24, 25) in
contrast to those who found no difference (21, 22, 23, 26). Three groups
assessed both techniques in the same patient groups (21, 22, 23) with
variable results. Coupland (23) and Arden et al. (21) showed
that PERG was unaffected in aretinopathic diabetic subjects despite
changes in OP amplitudes. Wanger and Persson (22), however, did not
detect any changes in either PERG or OP, concluding that ERG was of
limited use in the assessment of early retinal dysfunction in diabetes.
|
C. Type of diabetes
Most ERG studies have examined only subjects with IDDM. There are
few data available, therefore, pertaining to the ERG changes in
non-insulin-dependent diabetes (NIDDM). Bresnick and Palta (29)
included a small number (7 patients) of NIDDM subjects with retinopathy
in their study but did not comment on any differences with the larger
group of IDDM subjects they also studied (78 patients). Boschi et
al. (26) observed that PERG amplitudes were lower in their IDDM
patients compared with their NIDDM subjects, but no statistical
comparison was reported. In their study of pregnant diabetic women,
Vingolo et al. (30) reported different flash ERG responses,
with a greater decrease in b2/b1 ratio in the
group with NIDDM or gestational diabetes mellitus, compared with IDDM
subjects or nondiabetic controls.
D. Duration of diabetes
No flash ERG changes were found in a study of recently diagnosed
patients with IDDM (27). Papakostopoulos et al. (28),
however, detected significantly lower ‘b’ wave amplitudes in a group
of patients with IDDM of less than 6 yr duration. In five studies the
data were specifically analyzed for any correlation between ERG changes
and duration of diabetes (17, 24, 25, 26, 31). The results are conflicting,
as only two groups have reported significant reductions in ERG
amplitude with longer duration of diabetes (24, 25).
E. Glycemic control
One study has assessed the effect of short-term strict glycemic
control on OP amplitude (13). They reported that OP amplitudes, which
were initially abnormal in a group of aretinopathic subjects with IDDM,
were normalized after 11 days of strict glycemic control. Greco
et al. (31), in a study of children with diabetes, found no
significant correlation between PERG abnormalities and longer-term
glycemic control as assessed by glycated hemoglobin (HbA1c). None of
the other studies commented on any relationship between HbA1c and ERG
abnormalities, although this association may not have been sought in an
era before the Diabetes Control and Complications Trial (DCCT) that
emphasized the importance of strict glycemic control in preventing
complications associated with diabetes (32).
F. Age of subject
Prepubertal diabetic children have traditionally been considered
to be at low risk of developing microvascular complications, including
retinopathy (33). The possible existence of early electrophysiological
changes affecting the retina, before the development of overt
retinopathy, has been investigated by ERG (15, 31, 34). The changes in
‘b’ wave amplitude demonstrated in adults with IDDM (28) have been
observed in adolescents (age range, 12–20 yr) with IDDM, but although
both eyes were examined using blue flash ERG, changes were found only
in the left eye (34). Mean OP amplitude is also affected in young
patients with IDDM (15). Focal ERG was employed to assess the macular
function of 20 prepubertal children with IDDM who had no evidence of
retinopathy using fluoroscein angiography, and a significant reduction
in ERG amplitude was found in either the 2F or 2P component,
abnormalities that are thought to correlate with macular neuronal loss.
In the diabetic group, 45% exhibited at least one abnormal component
of neuroretinal function.
G. Retinopathy status
The association between ERG abnormalities and presence of diabetic
retinopathy has been examined in a number of studies (Table 2
). Most of
the patients studied had evidence of background retinopathy on
ophthalmoscopy or fundal photography, although several studies also
included patients with proliferative changes (29, 35). The evidence for
changes in both PERG and OP amplitude in subjects with background
retinopathy is conflicting. Three groups have reported significant
changes in PERG amplitude compared with aretinopathic controls (21, 23, 25). These alterations, however, were not found by other investigators
(22, 26). Several studies showed significant reductions in OP amplitude
in subjects with background retinopathy (23, 28, 29). Perhaps the most
compelling evidence for any association is provided by the prospective
studies by Bresnick and Palta, which followed a group of young IDDM
patients over a period of 15 yr (17, 29). In addition to showing
changes in OP amplitude with early retinopathy, they also demonstrated
that over time those with initially altered amplitudes were at greater
risk of developing proliferative changes (29). Of the subjects who were
classified as having ‘hypernormal’ amplitudes, i.e.,
greater than normal or normal amplitudes at the start of the study,
only 7% and 26%, respectively, went on to develop proliferative
retinopathy (PDR) within the next 15 yr. By contrast, over the same
time period, 62% with initially reduced amplitudes developed PDR,
providing evidence for the predictive potential of such a test. By the
end of the study, those with PDR had very low or absent OP amplitudes
particularly if the proliferative lesions were central rather than
peripheral, a finding that was replicated in another study by
Gjötterberg (35). As might be expected, photocoagulation therapy
had a further negative effect with a reduction in ERG amplitudes that
correlated closely with the number of laser burns, especially those
close to the macula (16, 36, 37). No improvement was observed 3 months
later (36).
H. ERG and hypoglycemia
The effect of altered blood glucose levels on flash and PERG has
been examined in nondiabetic adults (38). Skrandies and Heinrich (38)
showed that during hyper- and hypoglycemia, flash ERG amplitudes
increased significantly (P < 0.05) when compared with
the results obtained at euglycemia. By contrast, the PERG results
showed a significant decrease in amplitude during hyper- and
hypoglycemia compared with euglycemia. This difference may be explained
by the different retinal structures examined by each test and the
possible regional alterations in retinal blood flow that occur with
changes in blood glucose outside the normal range.
I. Summary
Changes, both in flash and in PERG, have been demonstrated in
adults and children with IDDM, but the relative merits of each test
have yet to be established. There appears to be good evidence, however,
that ERG abnormalities occur in patients with diabetes before the
development of retinopathy. In those with established diabetic
retinopathy, the evidence, although not entirely conclusive, is
suggestive of further ERG changes that are most apparent in subjects
with PDR and laser scarring. ERG has been shown to be of value in
assessing those most at risk of developing significant proliferative
eye disease in the future and, as such, may have a place in screening
patients who require more intensive ophthalmological review.
|
III. P100 Latency Studies |
|---|
|
TopAbstractI. IntroductionII. Electroretinography (ERG)III. P100 Latency StudiesIV. P300 StudiesV. Color VisionVI. Contrast SensitivityVII. ConclusionsReferences |
|---|
). It derives its name from the fact
that it occurs approximately 100 msec after the stimulus onset and is a
highly consistent and reproducible waveform (39, 40, 41), which is
generated in the striate and parastriate visual cortex in response to a
visual stimulus (Fig. 3). Current
methods of clinical assessment using this waveform require the subject
to look at a checkerboard. The position of its maximal amplitude is
variable, occurring at the inion (the most prominent protuberance on
the posterior surface of the skull) in some individuals and
corresponding to electrode position Oz in the standard 10–20 electrode
placement system. In other subjects the amplitude is greatest in the
midline parietal region at the point corresponding to Pz. Therefore, in
standard clinical measurement, waveforms are recorded both at Oz and at
Pz. The majority of the P100 waveform is generated by the lower half of
the visual field and a number of variables will influence the
resultant waveform, such as the visual angle subtended by the
stimulus and the size of the checks (squares) on the checkerboard. The
P100 amplitude is greatest in children during the first decade of life,
declining thereafter to remain at a lower level that is stable
throughout adult life. A term widely used in the literature is the
latency, defined as the time interval between the stimulus
onset and a specific point (usually the peak) on the waveform, and it
has been suggested that the latency increases in people over 60 yr of
age (39).
|
|
). Most
studies used pattern-reversal checkerboards as the visual stimulus with
variable check sizes and subtended angles, often relying on
single-channel recordings of VEP. In many studies no distinction was
made between the type or duration of diabetes, the presence of
complications, the prevailing visual acuity, or the quality of glycemic
control in determining the inclusion of subjects, and the different
methodologies employed makes direct comparison between studies
difficult, if not impossible.
|
C. Type of diabetes
The type of diabetes is either not specified (42, 43) or, if it
is, data on subjects with IDDM and NIDDM have not been analyzed
separately (44, 45). Where it is possible to discern the type of
diabetes studied, the P100 latencies were increased significantly in
IDDM patients by between 9% and 54% (Table 3). Fewer studies have
examined the P100 changes in subjects with NIDDM (46, 47, 48). Moreo
et al. (46) reported that 39% of their NIDDM group had P100
latencies more than 2 SD greater than nondiabetic controls.
Pozzessere et al. (47) directly compared IDDM and NIDDM
subjects and found P100 latency changes in a similar proportion of each
group (21% in NIDDM and 18% in IDDM). Although the durations of
diabetes were equivalent, the NIDDM subjects were older. Algan et
al. (48) also found no significant difference between subjects
with IDDM and NIDDM, demonstrating increased P100 latencies in
approximately 28% of each group.
D. Duration of diabetes
Alterations in P100 latencies have been demonstrated within a few
weeks of diagnosis of IDDM (27). However, further investigation,
including flash ERG and OP, found no differences compared with
nondiabetic controls, suggesting that the abnormalities were functional
rather than structural and related to the metabolic disturbance in the
absence of pathological changes in the retina. This theory is supported
by a separate study (44) assessing the effects of improving glycemic
control on P100 latency (vide infra). By contrast, Ziegler
et al. (49) did not detect any changes in either P100
latencies or brainstem auditory evoked potentials (BAEP) in newly
diagnosed IDDM patients, nor did they find any alterations in P100 in
those with IDDM of longer duration, with or without peripheral
neuropathy, which is at variance with other published data (Table 3).
Prolonged P100 latencies have been found within 4 yr of diagnosis in
patients with IDDM and NIDDM (47) and, in a different study, within 6
yr of the onset of IDDM (28).
The evidence for an association between duration of diabetes and VEP
abnormalities is limited, in keeping with the presence of P100
abnormalities in people with recently diagnosed diabetes. Of the 10
studies that have specifically examined this association, a significant
positive correlation with the duration of diabetes was described in
only three reports (42, 45, 50) (Table 3). However, several studies are
prone to type 2 statistical error, because of the small number of
subjects studied.
E. Glycemic control
Improving glycemic control has been shown to alleviate the
symptoms of peripheral neuropathy (51). Measurement of VEPs (P100) in a
group of poorly controlled patients (including IDDM and NIDDM) showed
that, after 3 days of near-normoglycemia achieved by continuous
subcutaneous insulin infusion, a significant shortening of P100
latencies had occurred, although these still remained significantly
prolonged in comparison with nondiabetic values (44). In view of the
rapid rate of improvement, the authors proposed that the changes in
P100 latency resulted from metabolic effects. By contrast, short-term
hyperglycemia (of <3 h duration) does not significantly affect
monocular VEPs in IDDM subjects, irrespective of the duration of
diabetes or the presence of diabetic retinopathy (52). Many of the
studies sought a correlation between longer term glycemic control
(estimated by glycated hemoglobin) and P100 latency alterations. The
evidence for an association with glycemic control (Table 4) is less compelling than that for
duration of diabetes, with only 1 (45) of the 10 studies demonstrating
a significant correlation, and that study had examined patients with
NIDDM.
|
). No association was identified
either with duration of IDDM or with glycemic control. However, in a
different study (34), no changes in P100 latency were observed in a
cohort of young subjects with IDDM.
G. Retinopathy status
Some of the study cohorts included patients who had evidence of
diabetic retinopathy on ophthalmological assessment (28, 44, 48, 49, 55) (Table 3). In three studies (28, 48, 55) no significant correlation
was identified between P100 changes and the presence of retinopathy. In
a further study, no significant changes in P100 latency were identified
for any of the groups (49). Possible associations with other diabetic
complications, including nephropathy and peripheral or autonomic
neuropathy, failed to achieve significance in two studies (48, 55).
H. Peripheral neuropathy
The evidence for an association between changes in peripheral and
central neurophysiological function is conflicting. Two studies have
reported positive correlations between abnormalities in peripheral
nerve conduction and changes in P100 latency (43, 45). By contrast,
Yaltkaya et al. (42) did not find a significant correlation
with sural nerve conduction velocity. Ziegler et al. (49)
assessed cerebral glucose metabolism in addition to multimodal evoked
potentials (including P100, P300, BAEP, and peripheral nerve conduction
velocities) and, although no significant P100 changes were found,
cerebral glucose metabolism was significantly lower in the neuropathic
group, suggesting a possible association between peripheral diabetic
neuropathy and cerebral glucose consumption. Pozzessere et
al. (47) identified significant abnormalities both in peripheral
and central nerve conduction in IDDM and NIDDM subjects, but no
correlations between these phenomena were performed. The association
between short-latency evoked potentials (visual P100, BAEP, and
somatosensory evoked potentials) and auditory P300 (a cognitive rather
than sensory event-related potential) was investigated in IDDM subjects
(56). Although both latencies were increased significantly in the
diabetic group, no significant correlation was observed between the two
parameters.
I. P100 and hypoglycemia
Many of the visual function parameters that were discussed above
with respect to diabetes per se have also been examined
during acute hypoglycemia. Several studies have shown no change in
corrected visual acuity during acute hypoglycemia (57, 58) despite a
fall in intraocular pressure (59) and a reduction in volume and depth
of the anterior chamber of the eye (60). Measurement of VEPs during
acute hypoglycemia (venous blood glucose levels below 2.5 mmol/liter)
has demonstrated prolongation of the P100 latency, which did not differ
significantly between diabetic and nondiabetic subjects (57). This
conflicts with the results of another study, which failed to show any
effect of moderate hypoglycemia (venous blood glucose 2.4 mmol/liter)
on VEP latency in nondiabetic subjects (61). Concurrent EEG changes
were observed in both studies (57, 61) with an increase in slow waves
(
- and 
-bands) at the expense of fast 
-waves, particularly in
the frontal regions of the brain. This finding was replicated by
Tallroth et al. (62) but not by Lindgren et al.
(63) in their respective studies of the effects of acute hypoglycemia
in nondiabetic subjects. [It is of interest that acute hypoxia
provokes similar frontal lobe changes in patients with epilepsy (64)].
The EEG changes recovered more rapidly than those associated with the
VEPs, with rapid normalization after resolution of the hypoglycemia.
These EEG findings suggest that the frontal lobes are more sensitive to
acute hypoglycemia than other parts of the brain; this supposition is
supported by neurophysiological, cognitive, and histological
studies (65, 66, 67, 68). However, a case report of temporary acute cortical
blindness (69), induced by hypoglycemia in a child with Von Gierke’s
syndrome, suggests an alternative explanation of localized cerebral
anoxia induced by hypoglycemia (which alters regional cerebral blood
flow), implying that the occipital cortex may also be vulnerable. In a
separate report, Gold and Marshall (70) describe a case of cortical
blindness and cerebral infarction occurring as a result of a severe
episode of hypoglycemia.
Kern et al. (71) investigated the effect of human vs. porcine insulin on sensory processing by assessing VEPs (P100) during euglycemia and hypoglycemia in a double-blind cross-over study in nondiabetic subjects. This showed reduced amplitude and increased latency of the VEP components during hypoglycemia, which were significantly stronger with porcine insulin compared with human insulin, suggesting a direct modulation of visual processing during hypoglycemia that was affected by the species of insulin.
J. Summary
In conclusion, good evidence exists for abnormalities occurring in
the P100 response in people with diabetes before the development of
overt retinopathy, ranging from the newly diagnosed patient with IDDM
to those with diabetes of long duration. Short-term improvements in
blood glucose may normalize the P100 latency. The correlations between
retinopathy, P100, glycemic control, and duration of diabetes are
generally nonsignificant. Although this diminishes the value of the
test in diabetes, the measurement of P100 has been shown to be an
important specialized diagnostic tool in assessing the neural
pathways relating to visual function. There is some evidence to suggest
that P100 latency is prolonged during acute hypoglycemia in diabetic
subjects, but the findings from different studies are conflicting and
no significant difference between P100 latency changes during
hypoglycemia in diabetic and nondiabetic subjects has been
demonstrated.
|
IV. P300 Studies |
|---|
|
TopAbstractI. IntroductionII. Electroretinography (ERG)III. P100 Latency StudiesIV. P300 StudiesV. Color VisionVI. Contrast SensitivityVII. ConclusionsReferences |
|---|
B. P300 and diabetes
In contrast with the numerous studies examining P100 latencies in
people with diabetes, relatively few similar studies exist that have
investigated the effect of diabetes on visual P300 latency. Some
investigators have used P300 generated by an auditory stimulus and
studied the association with hypoglycemia, peripheral nerve conduction
velocities, or cerebral glucose metabolism (49, 56, 62, 75). Only four
studies (63, 76, 77, 78) have used visual P300 waveforms generated by a
checkerboard stimulus and have examined the specific effects in
diabetic (77, 78) and nondiabetic (63, 76) subjects during conditions
of euglycemia (76, 77, 78) and hypoglycemia (63, 76, 77).
C. Type of diabetes
Mooradian et al. (78) studied 43 people with NIDDM and
compared their P300 characteristics with 41 age-matched nondiabetic
controls. It is not apparent whether the subjects were matched on
grounds of cognitive ability, a variable known to affect P300 latency
(79). P100 and P300 latencies were elicited using a checkerboard as
visual stimulus, and, in addition, an EEG and a number of psychological
tests of cognitive function and memory were performed. Recordings were
made at Fz, Cz, and Pz electrode placements, and no significant
differences were detected between the two groups for either the P100 or
the P300 latencies. The authors indicated that their electrode
placement was designed specifically to measure P300 latency; the P100
latency was derived indirectly. A trend was observed for longer P300
latencies at both Fz and Cz in the diabetic group but did not achieve
statistical significance. The induction of transient hyperglycemia in
the nondiabetic group by an intravenous infusion of dextrose did not
significantly alter the P300 latency. A single study has examined
visual P300 parameters in patients with IDDM (77). The baseline P300
results did not differ significantly from the age-matched nondiabetic
controls in the IDDM group in whom glycemic control was poor.
D. Glycemic control and duration of diabetes
Although no data are available pertaining to the relationship
between visual P300 latency and either duration of diabetes or level of
glycemic control, the studies investigating auditory P300 have not
demonstrated any association with either duration of diabetes or
glycated hemoglobin (49, 56).
E. Retinopathy status
No studies have been done examining the effect of diabetic
retinopathy on visual P300 amplitude or latency.
F. P300 and hypoglycemia
Acute hypoglycemia has been shown to affect visual P300 waveforms
significantly in several studies (63, 76, 77). In nondiabetic subjects,
lowering the arterialized blood glucose to 2.6 mmol/liter significantly
increased the P300 latency, with full recovery occurring within 45–75
min after normoglycemia was restored (76). In a similar study, Lindgren
et al. (63) demonstrated that hypoglycemia (arterialized
blood glucose of 2.5 mmol/liter) maintained for 1 h, significantly
reduced the visual P300 amplitude during complex tasks, and recovery
was not complete until 40 min after restoration of normoglycemia.
Blackman et al. (77) also assessed visual P300 during
hypoglycemia in IDDM patients. Baseline P300 values did not differ from
nondiabetic controls, but when the arterialized blood glucose was
lowered to 2.5 mmol/liter, latencies increased significantly in the
diabetic group and eventually returned to normal after normoglycemia
was restored.
Evidence regarding the relationship between changes in visual and auditory P300 latencies during acute hypoglycemia is conflicting. Auditory P300 is a simpler test to administer, and a number of studies have measured this during hypoglycemia (62, 75), showing an increased latency in nondiabetic subjects and reduced amplitude both in diabetic and nondiabetic individuals at arterialized blood glucose concentrations below 3.0 mmol/liter (75) and between 1.6–2.3 mmol/liter (62). In one study (75) these changes were correlated with BAEPs and were preceded by the secretion of counterregulatory hormones. Several studies have compared visual and auditory P300 responses during acute hypoglycemia, and conflicting results have been reported (63, 76). Blackman et al. (76) observed increments of a similar magnitude both in auditory and visual P300 latencies at an arterialized blood glucose of 2.6 mmol/liter, whereas Lindgren et al. (63) failed to demonstrate any alterations in either auditory P300 latency or amplitude during moderate hypoglycemia (arterialized blood glucose 2.5 mmol/liter) in contrast to the changes they observed in visual P300.
G. Summary
Visual P300 and its relationship to diabetes remains relatively
unknown. From the small amount of published work, no significant
abnormality is evident in patients with either IDDM or NIDDM, although
more studies are required to confirm this. Visual P300 during
hypoglycemia (arterialized blood glucose, 2.6 mmol/liter), although
prolonged both in diabetic and nondiabetic subjects, is affected more
in subjects with diabetes. Evidence also exists to suggest that visual
P300 is more sensitive to hypoglycemia than auditory P300.
|
V. Color Vision |
|---|
|
TopAbstractI. IntroductionII. Electroretinography (ERG)III. P100 Latency StudiesIV. P300 StudiesV. Color VisionVI. Contrast SensitivityVII. ConclusionsReferences |
|---|
B. Color vision and diabetes
After initial reports of altered color vision in patients with
diabetes (81), a number of experimental studies have been conducted in
the past 25 yr to assess this association formally (80, 83, 85, 86, 87, 88, 89, 90).
With the exception of one study (88), the results all confirm a
significant deterioration in color vision (mainly assessed by the
100-Hue Test) in diabetic subjects without retinopathy compared with
nondiabetic controls. More specifically, spectral losses of yellow-blue
discrimination (tritanopia) have been observed in people with diabetes
with varying frequency; 30% (87), 70% (52, 80) and 80% (83). As a
congenital finding, tritanopia is very rare but has been reported in
several different pathological conditions including diabetes, (52) all
of which may be linked by neuronal hypoxia. Tritanopia is distinct from
the red-green spectral losses that can be seen in lens changes,
particularly opacification, providing further evidence that these
reported yellow-blue color defects relate to alterations at the retinal
level and not within the lens (80). The frequency of this abnormality
has important implications for the interpretation of blood glucose
concentrations when reading a visual strip (91, 92). Lombrail et
al. (87) showed that diabetic patients misread blood glucose
strips twice as frequently as nondiabetic controls, prompting the
authors to propose that tests of color discrimination should be
employed more widely in diabetic patients who perform self-monitoring
of blood glucose. Alternatively, diabetic patients should be encouraged
to use glucose meters wherever possible to avoid the effect of altered
color vision on matching color changes.
C. Type of diabetes
Most studies have been conducted exclusively in patients with
IDDM. Several studies have included a small number of subjects with
NIDDM (80, 89, 90), but no comparative analyses of the results were
performed.
D. Duration of diabetes
Some studies have performed correlational analysis between color
vision indices and duration of diabetes with conflicting results. Four
groups did not find a significant association (52, 83, 86, 88). By
contrast, two studies (80, 93) noted that poorer performance in the
100-Hue Test was associated with longer duration of diabetes. As they
included patients both with and without retinopathy, however, this may
reflect the fact that those with longer diabetes duration were more
likely to have retinopathy that affected their color vision
independently (vide infra).
E. Glycemic control
Poor glycemic control (as measured by HbA1c) was shown to
correlate significantly (r = 0.24; r =
0.3) with deterioration in color vision in two studies (52, 94). This
finding, however, was not replicated in four other studies that
examined this association (80, 83, 86, 88). Hardy et al.
(82) studied the effect of short-term increases in blood glucose (14
mmol/liter) but did not find any changes in color vision compared with
the results obtained during euglycemia.
F. Age of subject
Although one study included children within its cohort of subjects
(52), no studies have been specifically directed at assessing color
vision in children with diabetes. Some studies have included adults
with a wide age range (80, 90), and Lakowski et al. (80)
showed that increasing age was associated with deteriorating color
vision in both the diabetic and nondiabetic groups. In their 7-yr
prospective study, Aspinall et al. (90) reported that, in
the subjects under 40 yr of age, diabetes duration was found to be the
best predictive indicator of retinopathy development. By contrast, in
those over 40, yellow-blue color discrimination was the best single
predictive parameter.
G. Retinopathy status
Several studies have examined the association between abnormal
color discrimination and diabetic retinopathy (80, 83, 87, 88, 89, 90, 93, 95).
Lombrail et al. (87), although reporting that one-third of
their diabetic subjects had significant dyschromatopsia, did not
find any difference between those with and without diabetic
retinopathy. In a separate study, the same group (95) detected a
deterioration in 100-Hue Test scores in subjects with IDDM who had
retinopathy but did not find any correlation with early retinopathic
changes. Two other groups reported a significant deterioration in
performance on color testing in subjects with retinopathy, particularly
in the yellow-blue spectral region (80, 90). The degree of retinopathy
also appears to be important. Bresnick et al. (89) found a
positive correlation between color discrimination and extent of
retinopathy. Sixty-five percent of those with proliferative changes had
abnormal 100-Hue Test results and those with central changes,
especially macular edema, were most affected. In a separate study Roy
et al. (88) also found that the subjects who had PDR
performed most poorly.
H. Color vision and hypoglycemia
Lakowski et al. (80) made the first report of color
changes during hypoglycemia when one of their subjects accidentally
experienced an episode of hypoglycemia during one of the study
sessions. Although the subject was unaware of any deterioration in
vision during the episode, formal color vision examination confirmed a
deterioration in performance that corrected on retesting once
euglycemia had been restored. Harrad et al. (57) conducted
the first experimental assessment of color vision during hypoglycemia.
Using the 100-Hue Test, they reported a deterioration in performance
during acute hypoglycemia both in diabetic and nondiabetic subjects. It
should be noted that the degree of hypoglycemia induced by intravenous
injection of insulin was variable and often profound (mean venous blood
glucose 1.9 ± 0.4 mmol/liter). Any alteration in color vision
reported may be due to changes in cognitive function that would affect
performance in the 100-Hue Test, which takes time to complete and
requires concentration and decision making. In a more recent study,
Hardy et al. (82) studied the effect of more constant
hypoglycemia induced by a glucose clamp method (arterialized blood
glucose, 2.5 mmol/liter) in aretinopathic diabetic subjects but did not
find any significant changes in 100-Hue Test scores when compared with
the results obtained at euglycemia (5.0 mmol/liter).
I. Summary
The assessment of color vision may be more sensitive than ERG in
the detection of early visual dysfunction. Specific spectral losses,
especially yellow-blue discrimination, are widespread in patients with
diabetes, irrespective of the presence of retinopathy and duration of
diabetes. The results from subjects with background retinopathy are
conflicting, although, as expected, the presence of more advanced
retinopathy or maculopathy has a greater effect on color vision. The
data for color vision’s changes during acute changes in blood glucose
(hyper- and hypoglycemia) and its correlation with longer term glycemic
control are inconclusive.
|
VI. Contrast Sensitivity |
|---|
|
TopAbstractI. IntroductionII. Electroretinography (ERG)III. P100 Latency StudiesIV. P300 StudiesV. Color VisionVI. Contrast SensitivityVII. ConclusionsReferences |
|---|
Although both hue discrimination and contrast sensitivity reflect macular function, their exact physiological relationship has not been fully explained. Trick et al. (94) directly compared the ability of these two parameters to detect early visual dysfunction in diabetic patients. They found that 37.8% of aretinopathic subjects had evidence of abnormalities on either test, this figure rising to 60% in those with background retinopathy. Contrast sensitivity was abnormal more frequently than color discrimination (measured by the 100-Hue Test), and it appeared to be uncommon for individual subjects to exhibit simultaneous deficits both in contrast sensitivity and color vision. Brinchmann Hansen et al. (99) also found that contrast sensitivity correlated more closely to the grade of retinopathy than either color vision or macular recovery (another parameter of psychophysical function).
B. Contrast sensitivity and diabetes
Significant losses of contrast sensitivity have been observed in
patients with IDDM who had no evidence of retinopathy when compared
with nondiabetic controls (94, 96, 98, 100, 101), particularly in
spatial frequencies in the mid to high range (94, 96, 101). Di Leo
et al. (96) also demonstrated that these changes occur both
in dynamic and static modes, the latter appearing to be more sensitive
to early changes. By contrast, Sokol and colleagues (97) reported that
contrast sensitivity was normal in patients with IDDM who had no
retinopathy.
C. Type of diabetes
Some studies have included patients with NIDDM as well as IDDM in
their patient groups (94, 97). Sokol et al. (97) found
significant changes in contrast sensitivity at 22.8 cycles/degree in
the NIDDM group whereas the IDDM group results were normal
(P < 0.01). Trick et al. (94), however, did
not find any differences between the IDDM and NIDDM patients they
studied.
D. Duration of diabetes
IDDM of short duration was shown to be associated with significant
contrast sensitivity losses at all but the highest spatial frequencies
(96). Two other studies also demonstrated a significant negative
correlation between contrast sensitivity and duration of diabetes in
IDDM and NIDDM of longer duration (94, 97), although this relationship
was not found in a separate study by Buckingham and Young (102).
E. Glycemic control
Two studies have demonstrated a positive correlation between poor
glycemic control, as assessed by HbA1c, and deteriorating contrast
sensitivity (96, 100). Banford et al. (100) found a
significant correlation (r = -0.142) at 6 and 12
cycles/degree, whereas Di Leo et al. (96) reported positive
correlations (r = 0.34–0.51) at multiple spatial
frequencies. By contrast, one study failed to show any significant
relationship between contrast sensitivity changes and HbA1c (94).
F. Age of subject
Most of the studies reported in the literature examined contrast
sensitivity in adults with diabetes. One study also included children
and analyzed their results separately for this group (97). They showed
small, but significant, differences in contrast sensitivity scores
between children with diabetes and nondiabetic controls at two spatial
frequencies. This contrasts with their findings in adults with IDDM who
did not differ from their nondiabetic counterparts in tests of contrast
sensitivity.
G. Retinopathy status
The evidence regarding the association between abnormal contrast
sensitivity and the presence of retinopathy is conflicting. Ghafour
et al. (101) demonstrated a significant increase in the
contrast sensitivity threshold, which was most marked in a diabetic
group who had PDR, but was also elevated significantly in the diabetic
group with background retinopathy when compared with patients with no
retinopathy. Sokol et al. (97) also reported abnormal
contrast sensitivity at all spatial frequencies in a group of NIDDM
patients with background retinopathy. Howes et al. (103) and
Hyvarinen et al. (104) both reported changes in contrast
sensitivity that related to the degree of retinopathy. Brinchmann
Hansen and colleagues also found an association between contrast
sensitivity and grade of diabetic retinopathy, but only at 6
cycles/degree (99). By contrast, Collier et al. (105) and
Banford et al. (100), in separate studies, did not find any
significant difference in the performances of patients with IDDM and
retinopathy compared with those who had no retinopathy, although the
numbers involved in one study (100) were very small (only 8.3% of
patients had retinopathy). Bangstad et al. (106) used a
novel approach to assess the relationship between contrast sensitivity
changes and microvascular diabetic complications by studying subjects
who had established microalbuminuria and comparing their performance in
tests of contrast sensitivity to diabetic controls who did not have
microalbuminuria. They showed that, independent of background
retinopathy, contrast sensitivity was impaired in the subjects with
microalbuminuria.
H. Contrast sensitivity and hypoglycemia
Di Leo and colleagues (96) have postulated that recurrent episodes
of minor hypoglycemia may be responsible for the underlying
physiological changes to the optic nerve that result in the
nonselective loss of contrast sensitivity that this group has detected
in early IDDM. Hypoglycemia has been shown to impair contrast
sensitivity in nondiabetic subjects (58), and a recent study (107) in
diabetic subjects demonstrated a trend toward impaired performance in
this test, although this was less pronounced than the nondiabetic
subjects, under similar test conditions. Hyperglycemia has also been
shown to affect contrast sensitivity significantly in people with IDDM
who do not have retinopathy while visual acuity remained unchanged
(108).
I. Summary
Contrast sensitivity testing, in common with color vision (another
test of psychophysical function), demonstrates significant changes in
diabetic subjects compared with nondiabetic controls, and there is some
evidence for a relationship with grade of retinopathy. Changes in
contrast sensitivity have been demonstrated in children and adults with
diabetes of short duration, and some evidence exists for a correlation
with poor glycemic control, although prospective studies are required
to assess this relationship over a longer time period. Although both
color vision and contrast sensitivity demonstrate similar patterns,
studies that directly compare the two tests suggest that measurement of
contrast sensitivity is the more sensitive and specific.
|
VII. Conclusions |
|---|
|
TopAbstractI. IntroductionII. Electroretinography (ERG)III. P100 Latency StudiesIV. P300 StudiesV. Color VisionVI. Contrast SensitivityVII. ConclusionsReferences |
|---|
ERG, as an index of retinal electrical activity, is able to detect abnormalities at the retinal level before overt pathology is visible. Changes both in flash and PERG have been clearly demonstrated in patients with diabetes, although the relative merits of each test are still unclear. The association between retinopathy and ERG changes is an important one. One of the few long-term prospective studies in this field (17, 29) has shown the potential predictive use of this technique in identifying those patients who are at greatest risk of developing retinopathy in the future. Although present technical limitations make this a test that is available only through specialist-referral centers, it provides an important research tool, the use of which may be extended eventually to a wider number of diabetic patients in the clinical setting to identify those who require the most intensive ophthalmological follow-up.
P100 assesses the visual pathway from retina to visual cortex and, as such, provides important information about neural pathways within the brain. There is strong evidence of widespread P100 abnormalities both in IDDM and NIDDM, indicating that this may be a useful test to detect early visual dysfunction as part of the syndrome of ‘diabetic encephalopathy.’ The lack of correlation with retinopathy suggests that this parameter is most useful in assessing the postretinal visual pathway and providing information about previously unsuspected visual dysfunction. Although short-term improvements in glycemic control may partially reverse these changes, the evidence for any correlation with HbA1c is very limited. The DCCT (32) demonstrated that improvement in glycemic control, as measured by HbAlc, correlated with reduced rates of diabetic retinopathy. Such a relationship, however, has not been irrefutably shown in the small number of electrophysiological studies that assess any relationship between longer term glycemic control and changes in vision. Although this raises the question of the significance of any changes demonstrated in this area to clinicians working in diabetes, the fact that electrophysiological changes have been shown to occur early in the disease process provides evidence for aspects of visual dysfunction distinct from retinopathy whose clinical significance is yet to be fully demonstrated. Longer term prospective studies may also show the predictive value of these tests in assessing patients most at risk of developing retinopathy, as has been shown in the field of ERG (17, 29). The conflict in data concerning any association with changes in peripheral nerve conduction requires further research.
P300, as a marker of cognitive function in vision and other sensory modalities, provides useful information as it links the visual pathway to higher brain centers. Few studies are currently available relating to visual P300, and most of the research in this area relates to the changes induced by short-term hypoglycemia. More research is required to evaluate the wider implications for visual cognition, especially in its relationship to P100.
Both P100 and P300 require a degree of technical expertise to obtain reliable and reproducible results. Like ERG, therefore, both tests are likely to remain restricted to specialist centers, although they provide an invaluable tool in assessing brain structures that are inaccessible by other techniques. With increasing interest in the effects of diabetes on the brain, they complement other techniques to enhance our understanding of the central nervous system.
Practicality and patient acceptability are important aspects of any widely used screening test. The psychophysical tests of both contrast sensitivity and color vision meet these criteria and have the potential to be of use in primary care settings as well as the hospital diabetes clinic. There is good evidence that abnormalities are present in diabetic patients without overt retinopathy. These tests, which are simple and quick to perform, could complement the existing screening tests for retinopathy, providing additional information about visual function, especially its change over time. Any role these tests have to play in routine screening for early visual dysfunction remains to be evaluated, but an easily performed test of clinical value with high sensitivity and specificity is obviously desirable.
Diabetic retinopathy, although the major cause of visual loss in the diabetic population, is not the sole aspect of visual dysfunction in this group, and it is important to consider other aspects of visual dysfunction occurring in people with diabetes that remain undetected by present methods of routine clinical assessment.
|
Footnotes |
|---|
1 Supported by Grant R80650 from Novo Nordisk Pharmaceuticals
Ltd. 
2 Supported by Grant R80773 from Lilly Industries Ltd.
|
References |
|---|
|
TopAbstractI. IntroductionII. Electroretinography (ERG)III. P100 Latency StudiesIV. P300 StudiesV. Color VisionVI. Contrast SensitivityVII. ConclusionsReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. A. Phipps, J. L. Wilkinson-Berka, and E. L. Fletcher Retinal Dysfunction in Diabetic Ren-2 Rats Is Ameliorated by Treatment with Valsartan but Not Atenolol Invest. Ophthalmol. Vis. Sci., February 1, 2007; 48(2): 927 - 934. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. E. de Gooyer, K. A. Stevenson, P. Humphries, D. A. C. Simpson, T. A. Gardiner, and A. W. Stitt Retinopathy Is Reduced during Experimental Diabetes in a Mouse Model of Outer Retinal Degeneration Invest. Ophthalmol. Vis. Sci., December 1, 2006; 47(12): 5561 - 5568. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. J. Ramsey, H. Ripps, and H. Qian An Electrophysiological Study of Retinal Function in the Diabetic Female Rat Invest. Ophthalmol. Vis. Sci., November 1, 2006; 47(11): 5116 - 5124. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Gastinger, R. S. J. Singh, and A. J. Barber Loss of Cholinergic and Dopaminergic Amacrine Cells in Streptozotocin-Diabetic Rat and Ins2Akita-Diabetic Mouse Retinas. Invest. Ophthalmol. Vis. Sci., July 1, 2006; 47(7): 3143 - 3150. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. T. Elia, D. Daneman, J. Rovet, M. Abdolell, W.-C. Lam, C. Till, V. Erraguntla, S. Rubab, N. Lodha, and C. A. Westall Color Visual Evoked Potentials in Children with Type 1 Diabetes: Relationship to Metabolic Control Invest. Ophthalmol. Vis. Sci., November 1, 2005; 46(11): 4107 - 4113. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Bearse Jr, Y. Han, M. E. Schneck, and A. J. Adams Retinal Function in Normal and Diabetic Eyes Mapped with the Slow Flash Multifocal Electroretinogram Invest. Ophthalmol. Vis. Sci., January 1, 2004; 45(1): 296 - 304. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
