Dr Susan J Leat School of Optometry University of Waterloo 200 University Avenue West Waterloo Ontario N2L 3G1 CANADA E-mail: firstname.lastname@example.org
Background: It is of interest to examine paracentral functional deficits in early age-related maculopathy (ARM), as histopathological studies indicate that this is where the earliest changes occur. The purpose of this study is to detect the sensory functional deficits at chosen retinal areas around the fovea and at the fovea itself in patients with early age-related maculopathy and to determine the type of functional losses that are more pronounced in early ARM.
Methods: Ten participants with early ARM and 10 age-matched controls took part. Crowded and uncrowded visual acuity and static and transient contrast sensitivity were measured in the same selected eye of each participant at eight predetermined retinal locations plus the fovea in patients with early ARM and controls. All measurements were made using computer-generated targets.
Results: A significant difference between the controls and subjects with ARM was found in low spatial frequency static contrast sensitivity (p = 0.05) but not for transient contrast sensitivity (p = 0.586). Visual acuity (uncrowded VA and crowded VA) showed a borderline difference (p = 0.072 and p = 0.084, respectively). Compared to controls, there was no evidence of increased contour interaction effects in early ARM (p = 0.595).
Conclusion: The subjects with very early ARM showed significant loss of low spatial frequency static contrast sensitivity before the loss of high contrast VA, indicating that static contrast sensitivity may be one of the earliest functional losses in early ARM and this loss was found to extend across the central 10 degrees of the retina.
Age-related maculopathy (ARM) is the leading cause of severe visual impairment and irreversible visual loss in the elderly in developed countries.1–5 It is a degenerative retinal disease which is estimated to affect approximately 25 to 30 million people over the age of 50 globally.5,6 It is important to understand the functional changes that people with ARM experience. Visual acuity and contrast sensitivity have been studied extensively in this population, patients with ARM usually experiencing gradual and significant loss of VA and CS.7–12 These studies measured VA and CS with natural viewing, that is, at the fovea or the preferred retinal locus (PRL) in the case of people with central scotomata. There is evidence that histopathalogical13 and fundus changes in ARM, such as drusen and RPE pigmentary changes, start eccentrically. Knudston and colleagues14 found that drusen larger than 125 µm, soft indistinct drusen, reticular drusen and RPE depigmentation were most likely to develop outside the central 500 µm (approximately 1.5 degrees). In particular, drusen greater than 125 µm, soft indistinct and reticular drusen between 1.5 and 10 degrees had greatest cumulative incidence14 and these types of drusen are most predictive of progression to age-related macular degeneration (ARMD).15 Wang and associates16 found that the total percentage of area covered by drusen was greatest within 500 µm of the fovea. Knudston and colleagues14 found that eyes that developed neovascular or geographic ARMD were more likely to have early lesions between 1.5 and 5 degrees, although Wang and co-workers15 found that total drusen within the central 1.5 degrees followed by the 1.5 to 5 degrees ring were greater risk factors for progression. There are several studies that give evidence of early paracentral functional loss. These include forms of perimetry17–21 or sensitivity to distortion, such as the clinically-used Amsler chart, the Macular Computerized Psychophysical Test22 and shape discrimination.23 Brown and Lovie-Kitchin showed paracentral deficits of contrast sensitivity24 and measured temporal modulation sensitivity.25 Multi-focal ERG has also been used to study paracentral function.26–28
Some studies have specifically investigated cone or rod function. Brown and colleagues29 showed elevated thresholds for cone dark adaptation in paracentral retina, while Phipps, Guymer and Vingrys30 showed delayed cone recovery. Owsley and associates18 found more frequent losses for dark-adapted, rod mediated sensitivity than cone losses and these were most severe at paracentral locations and in a later study, found disturbances in rod-mediated rather than cone-mediated adaptation in the parafovea.31 Feigle and co-workers28 used mfERG to demonstrate delayed rod, but not cone, responses in early dry ARM.
The transient/sustained dichotomy starts in the bipolar cells of the retina32 and there is evidence that photoreceptor degeneration can lead to changes to synaptic connections with the bipolar cells.33 Therefore, it is of value to compare losses for static versus temporally modulated stimuli. Some studies have claimed that static CS is significantly lower in early ARM than age-matched controls.24,28,34–36 Others showed that there is a loss of alternating or transient CS in early ARM.10,30,35,37,38 Yet other studies have demonstrated that subtle deficits of CS due to widespread disease can be detected in patients with ARM before VA loss appears.35,36,38–40 CS measurement is believed to be useful for the evaluation of visual function and functional change in patients with ARM20,41 and the low contrast targets are likely to be more sensitive to visual loss than high contrast VA targets. Therefore, we examined whether changes in paracentral CS (and VA) can be demonstrated before there is evidence of central VA loss, using both transient and static gratings.
The first purpose of this study was to examine the extent of functional losses in very early ARM and to determine whether functional losses can be demonstrated before loss of high contrast acuity. The second purpose was to measure functional aspects of vision at and around the fovea, to determine at which retinal regions the earliest losses occur and whether these mirror the fact that the earliest receptor losses often do not occur at the fovea but paracentrally. Third, as it has been suggested that contour interaction may be increased in ARM,11 we wanted to measure the effect of contour interaction (crowding) at paracentral locations. This will give us further understanding of the functional impairments experienced by people with early ARM.
Ten subjects with early ARM (aged 55 years and older) and 10 age-matched controls were recruited from the Low Vision Clinic or the Primary Care Clinic at the School of Optometry, University of Waterloo. All participants in this study had good general health and no ocular disorders except for early ARM for the ARM group. The study was undertaken monocularly. ARM was diagnosed according to the Wisconsin age-related maculopathy grading system (WARMGS)42 and for the control group, visual acuity was 6/7.5 or better.43 All participants signed a consent form at their first visit, when they were informed about the nature and purpose of the study and questions were answered. The study was approved by the Office of Research Ethics at the University of Waterloo. All selected eyes underwent a subjective refraction using a Bailey-Lovie visual acuity chart and the updated prescription was used for the study. Habitual VA was measured with a second Bailey-Lovie chart and calculated using by-letter scoring.
Retinal drusen of each selected eye were graded using the WARMGS classification.42 The digital fundus photograph (Canon CR6-45NM non-mydriatic fundus camera) taken for each participant was used as a reference and an average horizontal optic disc diameter of 1.5 mm44,45 was assumed for this grading (Figure 1). A grader, who was not aware of the group (ARM or control) to which each subject belonged, undertook the grading. A grading grid was placed over the fundus and scaled to be equal to the horizontal disc diameter, when each unit of the grid would be equal to 0.124 mm. The largest drusen on each fundus were compared to this grid and graded from 0 to 6 according to Table 1.
Table 1. Fundus grades according to WARMGS classification42
Indistinct drusen, questionable drusen or stippling
<0.063 mm (63 µm)
0.063–0.124 mm (63–124 µm)
Hard distinct or soft drusen
0.125–0.249 mm (125–249 µm)
≥0.25 mm (250 µm)
Reticular drusen = drusen that form ill-defined networks of broad interlacing ribbons
Targets for visual acuity measurements
All measurements were taken from eight predetermined retinal areas (five degrees and 10 degrees from fovea in four quadrants along the diagonals) plus the fovea. A computer-generated black square Landolt C (Figure 2A) was used as a non-crowded visual acuity test target. The target was presented on the computer screen with a white surround. The luminance of the computer screen and the white surround was 69 cd/m2 and the letter luminance was 6.4 cd/m2 measured using a Minolta Chroma Meter CS-100. The Weber contrast of the Landolt C was 90.72 per cent (Contrast = (Lmax-Lmin)/Lmax, where Lmax was the luminance of the screen background and Lmin was the luminance of the letter). The pixel size of the screen was 0.27 mm. The target presentation duration was 100 ms. This value was chosen so that there was insufficient time for an eye movement towards the target triggered by the target presentation during paracentral visual acuity measurement. A variety of test distances was used to provide a range covering the anticipated best and worst visual acuity foveally and paracentrally.46 The test distances were selected as 3.8 m, giving a range of acuity of 1.1 logMAR to -0.1 logMAR (Snellen 6/75 to 6/4.8) or 1.5 m, giving 1.5 logMAR to 0.3 logMAR (Snellen 6/190 to 6/12). The interval size was 0.1 logMAR units. The testing locations on the retina were the eight pre-determined paracentral retinal areas plus the fovea. For the paracentral positions, subjects were asked to fixate a moveable fixation cross, which was placed so that the target was centred at the required eccentricity.
A computer-generated black Landolt C with surrounding bars was the target used for the crowded visual acuity test (Figure 2B). The gap width between the Landolt C and surrounding bars was four times the minimum angle of resolution (MAR). This value was chosen to give definite crowding47 but not so much crowding that the VA target might be too large to display on the screen for the retinal areas of poorest VA. The testing distance was 1.5 m (1.5 logMAR to 0.3 logMAR [Snellen 6/190 to 6/12]) or 0.38 m (2.1 logMAR to 0.9 logMAR [Snellen 6/750 to 6/48]). Otherwise the targets for crowded VA were the same as for uncrowded VA measurements. The closer working distance was used for the crowded targets presented at 10 degrees eccentricity. At this distance each pixel subtended 2.4 minutes of arc (equivalent to 6/14.4 Snellen acuity). As the measured acuities were all 6/48 or poorer, it is unlikely that the size of the pixels would have been resolved and therefore, would not have affected the measurements.
Procedures for uncrowded and crowded VA
All participants were tested with their updated refraction, made with full aperture trial lenses for the tested eye. An occluder was placed in front of the untested eye and a correction for each viewing distance was added. The position of the fixation cross for each eccentricity was randomly changed by the examiner to one of the predetermined positions and the participant was reminded to stare at the fixation cross before each presentation. Thus, the order of retinal location, upper nasal (UN), lower nasal (LN), lower temporal (LT) or upper temporal (UT) was randomised. Participants responded orally and their answers were entered by the examiner. At each retinal location, a modified psychophysical descending method of limits with a running average47 was used and there were two phases of target presentation. In the first phase, the target size was reduced in 0.1 logMAR steps and presented once at each step. A four-option forced-choice procedure was used. The subject was asked to respond to the position of the gap of the Landolt C, which was randomly presented at the top, bottom, left or right. Once an incorrect response was made, the target size jumped back two levels (for example, from the level of 0.3 logMAR to the level of 0.5 logMAR) and the second phase of target presentation started using a modified method of limits with four presentations at each level. The test terminated automatically once the running average was below the required threshold value of 62.5 per cent (halfway between the lower 25 per cent guessing rate and the upper 100 per cent correct level).48 The threshold was calculated by extrapolating between the running average above and below the 62.5 per cent point.47 The crowding effect was calculated as the ratio of the crowded to the uncrowded VA at each retinal location, that is, logMAR crowded VA - logMAR uncrowded VA.
Stimuli for SCS and TCS measurements
Morphonome 3.2 software49 was used to present vertical sine-wave gratings, which were the targets for both static CS and transient CS measurements. The luminance of screen and surround was 63.6 cd/m2 and 64.57 cd/m2, respectively, (measured with a Minolta Chroma Meter CS-100) and the screen had a refresh rate of 75 Hz. The grating had a three-degree diameter field with a vignette envelope and a cosine onset and cosine offset. The grating was symmetrically positioned within the field with a light bar at the centre. These patterns were chosen to prevent unwanted spatial or temporal transients from affecting the CS measurements.50 Spatial frequency for both SCS and TCS measurements was 0.4 cpd (cycle per degree). The flickering rate of 7.5 Hz (cycles per second or cps) was used for transient CS measurement. This flickering rate was chosen to gain the maximum CS at the low spatial frequency of 0.4 cpd50,51 and to preferentially stimulate transient mechanisms.51,52 The presentation durations for the static and transient gratings were 247 and 200 msecs, respectively. These two values were considered to be short enough so that the participant would not have time to make an eye movement towards the stimulus during the paracentral measurements.53
Procedures for SCS and TCS measurements
A viewing distance of 57 cm was used for both static and transient CS measurements and subjects wore their updated refraction plus a correction for the viewing distance. The sine-wave grating stimulus was presented in a fixed position on the computer screen and the participant's fixation was controlled in the same way as for VA measurements. CS data were collected from the same locations as the visual acuity measurements and the order of retinal locations was randomised. A psychophysical descending and ascending method of limits was used at each retinal location for both static and transient contrast sensitivity measurements and the step size was 0.1 log units. The participant was allowed one ascending and one descending trial for practice at the task and then the threshold was taken as the mean of four trials.
The data were analysed mainly using repeated-measures ANOVA with two groups (ARM and control) and three retinal eccentricities (fovea, the mean of the five degrees and the mean of the 10 degree paracentral measurements). These means were calculated to apply ANOVA across three eccentricities. Geiser-Greenhouse and Hyundh-Feldt corrections were considered. The data from eight predetermined paracentral retinal locations were also normalised by the data from fovea and analysed using repeated-measures ANOVA with two groups (ARM and control group) and eight retinal eccentricities (four at five degrees and four at 10 degrees). Overall the pattern of the results from this analysis gave results similar to the results of the ANOVA as described here.
The mean age of the ARM group was 68 ± 5.80 (SD) years and of the control group was 67 ± 5.54 years. There was no significant difference in habitual Snellen VA between ARM and controls. The mean logMAR for the controls was -0.09 ± 0.06 and for the participants with ARM was -0.066 ± 0.07 (two sample t-test, p = 0.445). All ARM subjects had very early ARM (Snellen VA better than 0.1 logMAR or 6/7.5). There was a significant difference of the fundus grades. Figure 3 shows a histogram and indicates that the controls and ARM subjects were completely separated according to the overall ARM fundus grade based on drusen size. Table 2 shows age, VA and overall fundus grading of ARM group.
Table 2. Age, VA and overall fundus grading (according to the largest drusen size) of participants with early age-related maculopathy (ARM) according to the WARMGS42
Habitual VA (logMAR)
Overall fundus changes
Overall fundus grade
Extensive hard and soft drusen and pigment changes
Hard and soft drusen
Extensive soft drusen
Extensive hard drusen
Soft and hard drusen
Soft and hard drusen
Soft and hard drusen
Soft and hard drusen
Soft and hard drusen
Soft and hard drusen
VA measurements and crowding effect
For the measurements of uncrowded VA and crowded VA, repeated-measures ANOVA (two groups and three retinal eccentricities) showed no overall significant difference between controls and ARM subjects (p = 0.072 and 0.084, respectively) although this was borderline (Figures 4A and 4B). There were no interactions between retinal eccentricity and group (p = 0.915, p = 0.161), which were also not significant with either the Geiser-Greenhouse or Hyundh-Feldt correction. As expected, there was an effect of eccentricity (p < 0.001 for both). ARM uncrowded VA and crowded VA measured at predetermined retinal areas were compared with the 95 per cent upper and lower range obtained from the control group in this study, which were obtained by 1.96 times the SD. This comparison showed that a few measurements of the ARM group fell outside of the normal range, while all the data points of the controls were within that range. The crowding effect (repeated-measures ANOVA with two groups and three retinal eccentricities) showed no overall significant difference between controls and ARM subjects (p = 0.595) and there was no interaction between retinal eccentricity and group (p = 0.343) (Figure 5). There was an effect of eccentricity (p < 0.001).
SCS and TCS measurements
For the SCS measurements, repeated-measures ANOVA (two groups and three retinal eccentricities) showed that there was a just significant difference between controls and ARM subjects (p = 0.05) (Figure 6A) but there was no interaction between retinal eccentricity and group (p = 0.92). The data also showed an effect of eccentricity (p < 0.001). ARM static CS measured at predetermined retinal areas compared with the 95 per cent upper and lower range obtained from the control group showed that 16 data points of ARM subjects fell outside of the normal range, while one data point of the controls fell outside of the normal range. For TCS measurements, there was no overall significant difference between controls and ARM subjects (p = 0.568) (Figure 6B) and there was no interaction between retinal eccentricity and group (p = 0.994). The data also showed an eccentricity effect (p < 0.001). Compared with the 95 per cent upper and lower range obtained from the control group, transient CS showed only three data points of ARM subjects fell outside the normal range.
Most previous studies of visual function in ARM have concentrated either on exudative ARM or on detecting central functional deficits. The present study was interested in whether functional losses would be demonstrated in people with very early ARM and collected data not only at the fovea but also at paracentral locations (five and 10 degrees). The subjects in the present study were defined as early ARM according to the WARMGS grading,42 as there was no significant loss of habitual VA and no advanced retinal changes based on the grading of the fundus photograph appearance. We have demonstrated that there are subtle losses of static CS before clinically measurable loss of VA and that these losses occur not only at the fovea but extend across the central area of the retina. These findings may be dependent on the spatial frequency chosen; we used a low spatial frequency for the SCS measurements. If we had chosen a higher spatial frequency, we may have found less effect for SCS, as VA correlates with CS at higher frequencies. Thus, it appears function for form with static targets may be the first affected by ARM and that loss of static CS (rather than transient) is the first functional loss of the measures that we included.
Although there are many studies that have shown reduced CS in moderate and advanced ARM,7–9,12 there are few studies in early ARM. Losses of static CS are present in subjects with early ARM and slight VA loss.24,28,30 More relevant are the studies that have shown subtle deficits of CS before VA loss for high and medium spatial frequencies,36 high spatial frequencies34 or all spatial frequencies.35 We used a fairly low spatial frequency target, which was chosen to ensure that paracentral CS measurement was not limited by the normally lower paracentral VA. It is possible that the reduced static CS was found because of fewer micro-ocular movements in people with ARM, which has been shown to improve static CS for people with normal vision.54 Poorer fixation stability has been documented for patients with ARMD, which would include visual acuity loss55,56 but there is little documentation of micro-saccades in people with ARM. Although we did not measure eye movements for these subjects, we think this explanation is unlikely. All subjects with ARM had good central VA with very minimal paracentral functional loss.
In the present study, there was no significant difference between the groups in contrast sensitivity for transient gratings. This is in agreement with findings obtained by some previous studies for flicker sensitivity with cone-mediated mfERG,28 for low temporal frequency flicker (1 to 8Hz)39 and for temporal CS and flicker sensitivity25 for pre-ARM eyes (defined as VA 6/6 or better but with macular drusen and/or pigmentary changes, that is, similar to the early ARM in the present study). In contrast, other studies have shown deficits for temporal functions in early ARM for flicker,30,38 high temporal frequency TCS25,35,39 and focal ERG flickering at 41 Hz.40 Therefore, there is some disagreement in the literature, some studies finding deficits of TCS or flicker sensitivity and others not. The reason for this difference is not clear but may be related to the severity of the early ARM of the subjects. Enoch57 demonstrated that sustained responses appear to be affected earlier in some ARM patients than transient responses, that is, there were some subjects in whom only sustained responses were reduced and in others, both sustained and transient. In no subject was transient reduced when the sustained was normal. Compared with some of these studies, we included participants with a low grade of ARM fundus changes within the particular predetermined retinal areas. In many cases, the drusen were not located at the predetermined retinal areas.
Most previous studies used centrally presented targets whereas the present findings also show a loss with paracentral targets out to 10 degrees. Histopathologically, there is evidence for the parafovea (1.5 to 10 degrees around the fovea) being most susceptible to degeneration of both rods and cones,13 the rods being affected before cones. There are a few studies that have shown paracentral functional deficits in early ARM for VA,58 CS,24,25 dark-adapted visual fields18 and static perimetry (plus ERG and EOG),19 while Swann and Lovie-Kitchin17 documented paracentral scotomata but the subjects in these studies included those with central VA reductions. Cheng and Vingrys20 found parafoveal field defects at five and 10 degrees in some subjects with pre-ARM (VA 6/7.5 or better with drusen or pigmentary changes), while Brown and Lovie-Kitchin25 found no loss of temporal CS at 10 or 20 degrees eccentricity in people who had early ARM but no central VA loss. There appear to be no other studies that have shown paracentral losses of CS before VA loss in early ARM. The present study is in agreement with the histopathological studies and a number of studies that showed functional losses paracentrally, but we have extended this to show that paracentral functional defects can be detected with contrast sensitivity targets and that these functional defects exist in very early ARM.
VA and the crowding effect
Visual acuity loss is probably the most obvious functional impairment in ARM but the WARMGS definition of ARM does not include VA loss for a diagnosis of ARM and is based solely on the fundus appearance. The present study did not find significant VA losses in this group of subjects with early ARM centrally or paracentrally yet it is clear from this study and others20,35,36,38,39 that there are measurable functional losses even in these cases, that is, the loss of high contrast VA appears later than some other functional deficits. This could imply that classifying the presence of drusen of more than grade 3 (0.063 mm or larger) as ARM is correct, that is, this line between absence and presence of ARM is correctly placed as functional impairments can be demonstrated using this definition of ARM, despite the fact that high contrast acuity is still intact.
With regard to contour interaction, the present study did not find evidence of increased crowding when comparing similar paracentral locations in subjects with normal vision and ARM. Both groups showed the expected increase of crowding with increased eccentricity. Recently, Cacho and co-worker59 found that subjects with ARMD exhibited only a small crowding effect, which was not greater than that documented in subjects with normal vision. We can conclude that early ARM does not increase contour interaction and that the increased crowding that has been documented in ARMD11,60 may be because an eccentric retinal location is used for fixation, at which there is more crowding than at the fovea.43,61,62 Further study is required to completely resolve this question.
To conclude, of the functional measures applied in this study, static CS appears to be the earliest to be affected in ARM and therefore, it might be inferred that sustained retinal mechanisms are affected before transient pathways. There was no difference from normal in the majority of the functional measures that were used. This finding suggests that SCS could be used as a sensitive test for detecting functional changes in ARM at a very early stage. This is the first study that has shown losses of CS using grating targets paracentrally to 10 degrees prior to VA loss in early ARM and is in agreement with histological studies and some studies of the distribution of hyperpigmentation and drusen, particularly those drusen types that are more likely to be precursors of advanced disease.14,15 Increased contour interaction was not found to be present in early ARM.
Although we determined a statistically significant CS loss, it is questionable whether this difference is large enough to have a functional impact. It would appear that these subjects do not have a subjectively noticeable change in their visual function due to their ARM. Indeed, this appears to be the case, as many of these subjects were not aware of their ARM and were certainly not experiencing any visual difficulties.
Thanks to Melanie Campbell for encouragement and Trefford Simpson for help with statistical analysis.
GRANTS AND OTHER FUNDING
This study was supported by the JP Bickell Foundation, a University of Waterloo-Canadian Institutes of Health Research Seed Grant and the Canadian Optometric Education Trust Fund.