Abstract
Purpose.:
The visual system rapidly adapts to contrast changes, often with each fixation. One key anatomical site underpinning contrast adaptation is the retinal ganglion cell dendrites, where degenerative changes occur in glaucoma. This study investigated the effects of early glaucoma and aging on rapid contrast adaptation.
Methods.:
Contrast detection and discrimination thresholds were measured in central vision for briefly presented (94 ms) Gabor patches with and without adaptation to 50% contrast Gabor patches (1000 ms). Fourteen people with glaucoma (aged 58–77 years), 17 age-similar controls (aged 50–72 years), and 19 younger adults (aged 20–31 years) participated. Detection thresholds were measured at various time points (47, 106, 200, 400, 600, and 1000 ms) post adaptation. Discrimination thresholds were measured post adaptation relative to a reference contrast below (30%), equivalent to (50%), or above (70%) the adaptor.
Results.:
The glaucoma group demonstrated elevated unadapted detection (P < 0.0001) and discrimination (P = 0.01) thresholds relative to age-similar controls. In normal observers, aging elevated unadapted thresholds (detection: P < 0.0001; discrimination: P < 0.0001). At 47 ms post adaptation, the glaucoma group demonstrated reduced effects of adaptation relative to controls (P = 0.009). Adaptation was also reduced when the reference contrast (50%) was equivalent to the adaptor (P = 0.02). Aging did not alter adaptation of normal observers.
Conclusions.:
Glaucoma alters rapid contrast adaptation while aging does not. Contrast adaptation is key to visual processing in natural visual environments. Our results imply that glaucoma produces abnormalities in natural visual experiences in central vision.
Fourteen people with glaucoma (12 with primary open-angle glaucoma, 2 with normal-tension glaucoma) aged 58 to 77 years (mean ± standard deviation [SD]: 69 ± 6 years) and 17 approximately age-matched control individuals aged 50 to 72 years (mean ± SD: 65 ± 7 years) participated. A third group of 19 young individuals aged 20 to 31 years (mean ± SD: 24 ± 3 years) were also included. Those with glaucoma were recruited from the Melbourne Optometry Clinic (Australian College of Optometry) or the Royal Victorian Eye and Ear Hospital or through advertisements via Glaucoma Australia (a patient support group). Those with healthy vision for age were recruited via responding to written advertisements placed in electronic newsletters associated with the University of Melbourne and in community newspapers.
All participants had visual acuity of 6/7.5 or better in the tested eye, no more than ±6.00 diopters (D) of sphere or −2.00 D of cylindrical distance refractive error, and no ophthalmological (other than glaucoma for the glaucoma participants) or systemic conditions, or medications known to affect visual performance. As migraine has been shown to affect contrast processing,
34 participants were required to have a self-reported migraine-free history. Participants with healthy vision had normal findings in a comprehensive eye examination (slit-lamp biomicroscopy, applanation tonometry [<21 mm Hg], ophthalmoscopy, optic nerve head imaging, and visual fields). Age-related lens changes in all older observers (glaucoma and control) were required to be ≤NC1.5 as classified using the Lens Opacity Classification System III.
35 All participants with healthy vision were classified as within normal limits (within the one-tailed 95% range of the normative database) using Moorfields Regression Analysis (MRA) or Glaucoma Probability Score (GPS) tool of the Heidelberg Retinal Tomograph II (HRT; Heidelberg Engineering, Heidelberg, Germany).
Visual fields were measured using the Medmont Central 30-2 test (Medmont Pty. Ltd.; Camberwell, VIC, Australia), which measures thresholds using a rapid ZEST procedure for size III luminance increment targets at 104 locations within the central 30°, including four points at 1° from fixation. All participants, including those with glaucoma, were required to have normal visual fields within the central 2° where the psychophysical testing was performed. Average defect (AD) and pattern defect (PD) scores for participants with healthy vision were required to fall within the one-tailed 95% range of normal, as defined by the normative database of the Medmont perimeter. All observers fulfilled the reliability indices (fixation loss, false positive, and false negative) criteria of ≤20%.
All glaucoma participants had an ophthalmological diagnosis of primary open-angle glaucoma or normal-tension glaucoma and were treated at the time of testing. Current ophthalmological records were supplied to the study. At the first study visit, a further examination was conducted including assessment of the anterior chamber, optic nerve head, and retinal nerve fiber layer. Quantitative optic nerve head imaging (Heidelberg Retinal Tomograph II) and visual field examination (Medmont Central 30-2 test) were also conducted. The results of these examinations were compared to the supplied ophthalmological records, and only those participants with an unambiguous diagnosis of primary open-angle or normal-tension glaucoma were included. All glaucoma participants were flagged as having failed either the structural (MRA or GPS being ≥95% probability of falling outside the normative database) or functional (AD or PD being ≥95% probability of falling outside the normative database) tests. The levels of structural (HRT) and/or functional (Medmont Central 30-2 test) abnormalities are shown in the
Table. Our patients have early glaucoma as indicated by AD values less than −6.10 dB.
Table Glaucoma Observer Characteristics: ID Number, Age, Diagnosis (POAG, NTG), MRA, GPS Tool of the HRT, and AD, PD of the Medmont Perimeter
Table Glaucoma Observer Characteristics: ID Number, Age, Diagnosis (POAG, NTG), MRA, GPS Tool of the HRT, and AD, PD of the Medmont Perimeter
ID | Age | Diagnosis | MRA | GPS | AD | PD |
G1 | 77 | POAG | Within normal limits | Within normal limits | −3.83* | 8.24† |
G2 | 67 | POAG | Outside normal limits | Outside normal limits | −2.95 | 12.45‡ |
G3 | 73 | POAG | Outside normal limits | Outside normal limits | −0.84 | 11.9‡ |
G4 | 77 | POAG | Outside normal limits | Outside normal limits | −3.96* | 18.31‡ |
G5 | 63 | POAG | Outside normal limits | Borderline | −0.46 | 16.42‡ |
G6 | 73 | POAG | Outside normal limits | Outside normal limits | −1.72 | 12.13‡ |
G7 | 71 | POAG | Within normal limits | Within normal limits | −1.7 | 10.46† |
G8 | 62 | POAG | Outside normal limits | Outside normal limits | −6.08† | 0.65 |
G9 | 63 | POAG | Outside normal limits | Outside normal limits | −3.58* | 17.43‡ |
G10 | 71 | POAG | Outside normal limits | Outside normal limits | −2.39 | 9.03† |
G11 | 61 | NTG | Outside normal limits | Borderline | −0.98 | 5.62* |
G12 | 69 | NTG | Outside normal limits | Outside normal limits | −0.65 | 4.58 |
G13 | 61 | POAG | Outside normal limits | Borderline | −0.39 | 0 |
G14 | 58 | POAG | Borderline | Within normal limits | −3.17* | 1.07 |
All participants provided written informed consent before participation in the study, in accordance with a protocol approved by our institutional human ethics committee and with the tenets of the Declaration of Helsinki.
Adapting (adaptor) and target patterns comprised Gabor patches (example shown in
Fig. 1). Each Gabor patch was truncated at 1° × 1° with a spatial frequency of 2 cyc/deg. Contrast of the Gabor was specified using Michelson contrast ([maximum luminance − minimum luminance]/average luminance).
Figure 1A illustrates the experimental design. All patterns were presented to central vision, with four diagonal lines to aid fixation. The beginning of each trial was marked by a tone, after which the adaptor (2 × 2 grid of 50% contrast Gabor patches) was shown for 1000 ms. After the offset of the adaptor, a time delay was introduced to assess threshold using a 94-ms target contrast pattern (single Gabor patch). The time points used to assess threshold post adaptation were 47, 106, 200, 400, 600, and 1000 ms, with each condition recorded on a separate experimental run. Each run began with the target pattern contrast clearly suprathreshold. The observer's task was to identify the location of the target Gabor patch out of four possible locations (a four-alternate, spatial forced choice; example shown in
Fig. 1A) using a CB6 (Cambridge Research Systems) response box. After the response, a blank screen of mean luminance was shown for 1000 ms, and the process started again until threshold was established. Observers received auditory feedback when the response was incorrect. A modified three-down, one-up staircase procedure was used to determine threshold. Initially, the contrast of the target stimulus was decreased by 20% with every correct response or increased by 20% for every incorrect response. After the first incorrect response, three sequential correct responses were required to decrease the contrast. Each experimental run comprised two interleaved staircases that each terminated after six reversals. The final threshold estimate from each staircase was calculated as the mean contrast at the last four reversals. The results from at least two experimental runs (four staircases) were averaged to give the final contrast detection threshold for each time point.
Within a given trial (adaptor, time delay, target pattern), the orientation (horizontal or vertical) and phase (chosen at random) of the Gabor elements were kept constant; however, they were randomized from trial to trial within a run.
Baseline contrast detection thresholds were measured using the same protocol but without the adaptor.
The method used here was similar to that reported by Wolfson and Graham.
28 The stimuli and methods are illustrated in
Figure 1B. The 50% contrast adaptor was presented for 1000 ms, immediately followed by the contrast discrimination target (94 ms), then followed by the 50% adaptor again, and finally a blank screen of mean luminance (1000 ms). The contrast discrimination target comprised four Gabor patches with the contrast of three of the four Gabor patches fixed (either 30%, 50%, or 70%). One of the four Gabor patches of the target had a different contrast, which was incremented for the 50% and 70% target contrasts and decremented for the 30% target contrast. Observers were required to indicate the patch location that was the “odd one out.”
Contrast discrimination thresholds were measured in the presence and absence of a 50% contrast adaptor for the three target contrasts (30%, 50%, 70%).
Contrast discrimination thresholds were collected with the staircase method previously described, with the mean threshold returned from the last two of four reversals.
All participants received substantive training prior to the collection of formal data. The order of tasks was randomized within each group and counterbalanced among the three study groups to avoid order-dependent effects of learning or fatigue between groups.
All contrast detection and discrimination data were log base 10 transformed prior to analysis using SPSS, version 20 (SPSS, Inc., Chicago, IL, USA). Statistical analysis tested the following hypotheses:
-
That unadapted contrast detection and discrimination thresholds are elevated in aging and in glaucoma;
-
That the between-group difference in contrast detection threshold post adaptation depends on the postadaptation time delay;
-
That the between-group difference in contrast discrimination threshold depends on contrast and adaptation state; and
-
That the magnitude of change in contrast detection and discrimination thresholds post adaptation is reduced in aging and in glaucoma.
In comparing group means for a single condition, independent t-test or Mann-Whitney U test was used as appropriate. To examine whether group means differed across time for the contrast detection task, or with contrast for the discrimination task, repeated-measures ANOVA was used. In testing of these hypotheses, results were considered statistically significant if P < 0.05.
Contrast detection thresholds pre and post adaptation are shown in
Figure 2. In the absence of adaptation, the glaucoma mean threshold (mean ± 95% confidence interval [CI] = 1.11 ± 0.11) was elevated relative to that in old observers (mean ± 95% CI = 0.83 ± 0.04) (
t(16.82) = −4.46,
P < 0.0001;
Fig. 2A, filled symbols). The mean threshold of older observers was elevated relative to younger adults (mean ± 95% CI = 0.69 ± 0.03) (
t(34) = −5.48,
P < 0.0001;
Fig. 2B, filled symbols). For all groups, thresholds were elevated when the target was shown in close succession to the adaptor, and recovered to baseline by 400 ms.
A repeated-measures ANOVA revealed a significant interaction between group (old and glaucoma) and time (
F 3.12,90.39 = 5.60,
P = 0.001), indicating that difference between groups varied with time post adaptation. This interaction can be visualized in
Figure 2A, where thresholds in the glaucoma group return to baseline earlier (∼200 ms) as compared to thresholds of the old group (∼400 ms). On examining the first three time points (47, 106, 200 ms), no interaction effect was found (
F 2,58 = 1.25,
P = 0.29), implying a common time course but that the threshold elevation due to the presence of the adaptor was reduced in the glaucoma group (
Fig. 2C).
Baseline contrast detection threshold (before adaptation) was subtracted from the threshold after adaptation for each individual to emphasize the effect of the adaptor (
Fig. 2C) at time point 47 ms, where maximal threshold elevation was expected. Adaptation in people with glaucoma (mean ± 95% CI = 0.12 ± 0.07) was significantly reduced relative to that in old controls (mean ± 95% CI = 0.22 ± 0.04) (
t(29) = 2.79,
P = 0.009).
When the older and younger normal observers were compared, no significant interaction was found for group and time (47, 106, 200, 400, 600, 1000 ms) (
F 3.35,113.78 = 2.17,
P = 0.09) implying that thresholds were elevated similarly relative to baseline in both groups post adaptation. Consistent with
Figure 2B, threshold change over the first three time points (47, 106, 200 ms) was also similar (
F 2,68 = 2.62,
P = 0.08) for the young and old groups, with thresholds returning to baseline at ∼400 ms. Despite having significantly different contrast detection thresholds before adaptation,
Figure 2C shows that young (mean ± 95% CI = 0.24 ± 0.04) and old (mean ± 95% CI = 0.22 ± 0.04) observers did not adapt differently (
t(34) = 0.58,
P = 0.57).
Although our stimuli were presented centrally in areas of normal standard automated perimetry outcomes, differences in baseline contrast detection thresholds were evident between groups of observers. Because adaptation effects are stronger for higher-contrast stimuli
24,37,38 and the 50% adaptor is closer to the contrast detection threshold of observers with poorer contrast sensitivity, there is a possibility that contrast sensitivity differences may, at least partially, explain differing adaptation effects between groups of observers. It is worth noting that despite the higher baseline contrast detection thresholds for older observers, no significant difference in adaptation was found when compared to that of the younger observers. This suggests that contrast sensitivity differences are not the dominant factor in establishing the magnitude of adaptation within observers with normal vision within the age range studied.
To explore the potential magnitude of effect of differences in contrast sensitivity on adaptation strength in the glaucoma group, we measured threshold 47 ms after the offset of the brief (1000 ms) adaptor pattern that varied over a range of contrast (10%, 20%, 40%, 50%, 75%, or 98%) on a subset of four old observers (ages 67, 71, 71, and 76 years).
Incremental thresholds measured against background contrast have been shown to exhibit a dipper function,
39,40 where thresholds are facilitated near the observer's detection threshold and rise steadily following a positive linear slope at suprathreshold contrasts. Here, we measured detection thresholds 47 ms after the offset of the adaptor. Previously, adaptation studies
24,37,38 demonstrated a linear increase of detection thresholds with increasing adapting contrasts. In considering previous literature
24,37–40 and examining the trend in our data (
Fig. 3A), we fit a linear function to the average thresholds (
r 2 = 0.97).
A contrast detection threshold (47 ms post adaptation, adaptor contrast = 50%) can be predicted using the old group linear model (slope = 0.004) for each glaucoma individual, with the y-intercept being each individual's contrast detection threshold before adaptation.
The measured and predicted contrast detection thresholds (47 ms post adaptation) for the glaucoma group are shown in
Figure 3B.
Predicted mean contrast detection threshold was significantly higher (difference ± 95% CI = +0.12 ± 0.08) than the measured mean contrast detection threshold (
t(13) = +2.26,
P = 0.04).
Figure 3B indicates that on average, the glaucoma group adapted less than predicted by their reduced contrast thresholds.
Predicting thresholds post adaptation (47 ms) of old observers and young observers (data not shown) from
Figure 3A did not reveal any statistically significant difference between the measured and predicted values (old:
t(16) = 0.62,
P = 0.52, young:
t(18) = 1.74,
P = 0.10).
These results suggest that the differences in measured adaptation effects between groups cannot be fully explained by the reduced contrast sensitivity of glaucoma observers.